Liver
What research found here. What was shown to help comes first, then the risks reported, most common first.
Liver
Strong several reviews agree · Moderate one review or consistent trials · Limited one or a few small studies · Untested no study in people, listed with what it rests on
Shown to help25
Death from any cause
Linked to more deaths from any cause, about 7% higher risk for each 5 U/L in a meta-analysis of general population cohorts.
The numbers and sources
- Effect
- In a meta-analysis of 19 cohort studies, the relative risk of death from any cause was 1.60 (95% CI 1.42 to 1.80) comparing the top and bottom thirds of baseline GGT, and 1.07 (1.04 to 1.10) per 5 U/L higher, consistent with a linear relationship [7]. A second meta-analysis of 35 prospective studies found GGT was associated with increased all-cause and cardiovascular mortality even at levels inside the reference interval [8]. In 14,950 US adults without viral hepatitis followed for 12 years, GGT above 51 U/L in men or 33 U/L in women went with more deaths from any cause (HR 1.5, 95% CI 1.2 to 1.8), from liver disease (HR 13.0, 95% CI 2.4 to 71.5), from neoplasms (HR 1.5, 95% CI 1.01 to 2.2) and from diabetes (HR 3.3, 95% CI 1.4 to 7.6) [4].
- Population
- Over 9.24 million participants and 242,953 deaths in 19 cohorts, which also covered other liver enzymes [7]. 571,511 participants and 72,196 deaths in 35 studies [8]. 14,950 adults in the third US National Health and Nutrition Examination Survey [4].
- Sources
- [7], [8], [4]
- Caveats
- Observational. Many of the pooled studies did not exclude people with liver disease or heavy alcohol use, which the authors of one meta-analysis say may have obscured the results [8]. Liver disease deaths were 0.13% of the US cohort, so the interval around that estimate is wide [4].
Liver enzymes in fatty liver disease
Shown to lower liver enzymes in fatty liver disease, ALT by about 9 U/L. Liver fat and scarring were not measured.
The numbers and sources
- Effect
- In people with metabolic dysfunction-associated steatotic liver disease, orlistat lowered AST by 4.01 U/L (95% CI 1.97 to 6.05), ALT by 8.70 U/L (95% CI 0.90 to 16.51) and GGT by 7.74 U/L (95% CI 0.59 to 14.88) against control. LDL cholesterol fell 5.61 mg/dL, fasting glucose 4.05 mg/dL, HOMA-IR 0.70 and BMI 0.99 kg/m². Blood pressure did not change [6]. A second meta-analysis also found lower ALT, AST, HOMA-IR, BMI, cholesterol, insulin and waist circumference [7].
- Population
- 6 RCTs [6]; 11 studies with 582 participants [7].
- Sources
- [6], [7]
- Caveats
- Both reviews measured blood markers, not liver fat on imaging or scarring on biopsy. The trials were small and likely overlap between the reviews. The abstract of [7] does not say whether all 11 studies were randomized.
Death from liver disease
Linked to more deaths from liver disease, with risk rising from values of 20 U/L upward in one large Korean cohort.
The numbers and sources
- Effect
- In 142,055 Korean adults followed for 8 years, death from liver disease rose with ALT even inside the usual reference range. Compared with ALT below 20 IU/L, the adjusted relative risk in men was 2.9 (95% CI 2.4 to 3.5) at 20 to 29 IU/L and 9.5 (7.9 to 11.5) at 30 to 39 IU/L. In women it was 3.8 (1.9 to 7.7) and 6.6 (1.5 to 25.6). The cut-off that predicted liver disease in men with the greatest accuracy was 30 IU/L [10]. In 14,950 US adults without viral hepatitis followed for 12 years, ALT above 30 U/L in men or 19 U/L in women went with more deaths from liver disease (HR 8.2, 95% CI 2.1 to 31.9) [11].
- Population
- 94,533 men and 47,522 women aged 35 to 59 in the Korea Medical Insurance Corporation study [10]. 14,950 adults in the third US National Health and Nutrition Examination Survey, negative for hepatitis B and C markers [11].
- Sources
- [10], [11]
- Caveats
- Observational. Deaths were taken from death certificates. Liver disease deaths were 0.13% of the US cohort over 12 years, so the interval around that estimate is wide [11]. The authors of a US study note that results from South Korea would not necessarily apply to countries where different causes of liver disease predominate [6].
Type 2 diabetes
Linked to more type 2 diabetes, 16% higher risk per 5 U/L in pooled cohorts, a result the authors said publication bias may explain.
The numbers and sources
- Effect
- In a meta-analysis of 17 prospective studies, the relative risk of type 2 diabetes was 1.16 (95% CI 1.08 to 1.25) per 5 IU/L higher ALT and 1.26 (1.14 to 1.41) per standard deviation of log ALT, with high heterogeneity (I2 88%). The association became nonsignificant after correction for publication bias [12]. In a Mendelian randomization study, genetically predicted higher ALT was related to more type 2 diabetes (odds ratio 1.45 per standard unit, 95% CI 1.10 to 1.92) [13].
- Population
- 60,359 participants and 3,890 new cases of type 2 diabetes in 17 studies [12]. Up to 64,094 people with type 2 diabetes and 607,012 controls in the genetic study [13].
- Sources
- [12], [13]
- Caveats
- The cohort studies are observational and their pooled result did not survive correction for publication bias [12]. The genetic study's authors read their findings as support for insulin resistance leading to fatty liver, which in turn raises diabetes risk, so ALT may mark that process and not drive it [13].
Cirrhosis and liver cancer
Linked to 44% lower cirrhosis and 35% lower liver cancer for each extra 2 cups a day, in pooled studies rated very low in quality.
Compared with 2 fewer cups a day, in a dose-response model [9][10]
The numbers and sources
- Effect
- For each extra 2 cups a day, the relative risk of cirrhosis was 0.56 (95% CI 0.44 to 0.68), of alcoholic cirrhosis 0.62 (95% CI 0.51 to 0.73) and of death from cirrhosis 0.55 (95% CI 0.35 to 0.74) [9]. For hepatocellular carcinoma it was 0.65 (95% CI 0.59 to 0.72), 0.71 in cohorts and 0.53 in case-control studies. Caffeinated coffee gave 0.73 (95% CI 0.63 to 0.85) and decaffeinated 0.86 (95% CI 0.74 to 1.00). The link held with or without heavy drinking, obesity, diabetes, smoking or hepatitis B and C [10].
- Population
- 432,133 participants and 1,990 cirrhosis cases in 9 studies [9]. 2,272,642 participants in 18 cohorts and 1,825 cases in 8 case-control studies [10].
- Sources
- [9], [10], [2]
- Caveats
- Heterogeneity was high for cirrhosis (I² 83.3%). The liver cancer review rated its evidence very low under GRADE, citing no randomized trials, possible publication bias and no accepted definition of a cup [10]. The decaffeinated estimate rests on 3 cohorts.
Organ damage in hereditary hemochromatosis
Linked to liver damage in people with the hemochromatosis genotype, where ferritin of 1,000 ng/mL or more marks those at higher risk.
The numbers and sources
- Effect
- Among people homozygous for the HFE C282Y variant followed for an average of 12 years, documented disease related to iron overload developed in 28.4% of men (95% CI 18.8 to 40.2) and 1.2% of women (95% CI 0.03 to 6.5). Men with the genotype and ferritin of 1,000 ng/mL or more were more likely to report fatigue, use of arthritis medicine and a history of liver disease than men without the variant [11]. A liver society guideline reports that fewer than 2% of people with HFE hemochromatosis and ferritin below 1,000 ng/mL at diagnosis have bridging fibrosis or cirrhosis in the absence of other conditions, a 94% negative predictive value. It also cites a review in which 64% of patients referred with ferritin over 1,000 ng/mL did not have iron overload on biopsy [12].
- Population
- 203 C282Y homozygotes among 31,192 people of northern European descent aged 40 to 69 [11]. Studies summarized in an Asian Pacific guideline [12].
- Sources
- [11], [12]
- Caveats
- Applies to people who carry the genotype. In people referred for a high ferritin without it, most did not have iron overload [12].
Liver enzymes and lipids in fatty liver disease (quercetin alone)
Shown to lower liver enzymes, LDL and triglycerides with quercetin alone in people with fatty liver disease, at very low to moderate certainty.
The numbers and sources
- Effect
- A 2025 meta-analysis of 7 RCTs (540 patients) with metabolic dysfunction-associated steatotic liver disease found that quercetin lowered ALT, AST, GGT, direct bilirubin and C-reactive protein, lowered total cholesterol, LDL and triglycerides, and raised HDL. BMI, body fat, kidney function and TNF-α did not change [10].
- Population
- 540 patients with fatty liver disease in 7 RCTs [10].
- Sources
- [10]
- Caveats
- GRADE certainty ranged from very low to moderate. Dose, baseline BMI and age may explain heterogeneity. The abstract gives no pooled effect sizes [10]. None of the trials measured senescent cells.
Alcohol use disorder
Shown to reduce the percentage of heavy drinking days on gabapentin in adults with alcohol use disorder, with other drinking outcomes not clearly different from placebo.
The numbers and sources
- How common
- Very common: more than 1 in 10 people.
- Effect
- In a meta-analysis of 7 trials, every drinking outcome leaned toward gabapentin, but only the percentage of heavy drinking days showed clear benefit (g -0.64, 95% CI -1.22 to -0.06) [15]. In a randomized trial, 27% on gabapentin had no heavy drinking days against 9% on placebo (number needed to treat 5.4), and 18% against 4% were abstinent (number needed to treat 6.2). The benefit appeared in people with more alcohol withdrawal symptoms before the trial and not in those with fewer [16].
- Population
- 7 randomized trials [15]; 90 evaluable adults with alcohol use disorder and a history of withdrawal symptoms, mean age 49.6 [16].
- Sources
- [15], [16]
- Caveats
- Not an approved use. The evidence is for gabapentin. See the misuse block below.
Fatty liver disease
Shown to lower liver enzymes in fatty liver disease, ALT by about 7 to 12 U/L, with low certainty and no data on deaths.
The numbers and sources
- Effect
- In 26 trials (2,375 patients), silymarin lowered ALT (pooled difference -12.39, 95% CI -19.69 to -5.08) and AST (-10.97, 95% CI -15.51 to -6.43), lowered total cholesterol, triglycerides and LDL, and improved steatosis on histology (OR 3.25, 95% CI 1.80 to 5.87) [5]. In a 48-week biopsy trial of 99 adults with NASH, a 30% or greater fall in disease activity score occurred in 32.7% on 700 mg three times a day and 26.0% on placebo (p 0.467). Fibrosis fell by 1 point or more in 22.4% against 6.0% (p 0.023) [6]. A 2025 Cochrane review of 17 RCTs (2,069 adults) found silymarin alone against placebo or no treatment may lower ALT by 7.21 U/L (95% CI -10.62 to -3.80, 6 trials, very low certainty). Silymarin complexes did not lower ALT, and no trial reported deaths or quality of life [7]. In a 2025 meta-analysis of 25 studies (2,283 patients), silymarin alone lowered LDL cholesterol (SMD -0.25), while ALT fell when silymarin was combined with diet or lifestyle change [8].
- Population
- Adults with fatty liver disease [5][7][8]. 99 adults with biopsy-proven NASH in Malaysia [6].
- Sources
- [5], [6], [7], [8]
- Caveats
- The meta-analysis authors state that the effects need confirmation [5]. The biopsy trial missed its primary outcome, and its fibrosis finding was secondary [6]. The Cochrane authors rated the evidence low to very low certainty and called benefits and harms unclear [7].
Liver fat
Shown to lower liver fat about 3 to 4 points more than placebo in adults with HIV. Liver scarring and outcomes were not measured.
The numbers and sources
- Effect
- Pooled across RCTs, liver fat percentage fell by 4.28 points against placebo (95% CI -6.31 to -2.24) [5]. Over 6 months, liver fat changed by a median -2.0% (lipid to water) on tesamorelin and +0.9% on placebo, a net effect of -2.9%. Visceral fat fell by 42 cm² more than placebo (95% CI -71 to -14) [7]. In people with HIV and fatty liver disease, liver fat fraction fell 4.1 points more than placebo at 12 months (95% CI -7.6 to -0.7), a 37% relative reduction. Liver fat under 5% was reached by 35% on tesamorelin and 4% on placebo [8].
- Population
- RCTs in adults with HIV [5]; 50 adults with HIV and abdominal fat accumulation [7]; 61 adults with HIV and a liver fat fraction of 5% or more [8].
- Sources
- [5], [7], [8]
- Caveats
- One meta-analysis built on small trials. They measured liver fat, not liver scarring or liver outcomes. The 2019 trial was funded by the NIH [8].
Drinking in alcohol use disorder
Shown to lower total alcohol consumption against placebo in adults with alcohol dependence, with no trial powered to show health benefits.
Compared with placebo, and indirectly nalmefene, naltrexone and acamprosate [10]
The numbers and sources
- Effect
- Topiramate reduced total alcohol consumption against placebo (SMD -0.77, 95% CI -1.12 to -0.42, I² 0%). Indirect comparisons suggested larger effects than nalmefene, naltrexone and acamprosate. No treatment showed harm reduction [10].
- Population
- 6,036 adults with alcohol dependence or alcohol use disorder who were still drinking, in 32 trials across five drugs [10].
- Sources
- [10]
- Caveats
- 81% of the trials had a risk of incomplete outcome data, and none was powered for health outcomes [10].
Fatty liver and liver enzymes
Shown to lower the liver enzymes ALT and AST slightly in adults with fatty liver disease, but not in people with diabetes or with piperine added.
The numbers and sources
- Effect
- In 15 RCTs (905 patients with nonalcoholic fatty liver disease), curcumin lowered ALT by 4.10 (95% CI -7.16 to -1.04) and AST by 3.27 (95% CI -5.16 to -1.39), with no change in ALP. Curcumin with piperine did not change ALT or AST significantly [36]. In 10 RCTs in prediabetes and diabetes, ALT and AST did not change significantly and GGT fell 2.35 IU/L [37]. Across 18 RCTs of curcuminoids with piperine, AST fell 2.29 IU and the ALT change was not significant (p 0.053) [18].
- Population
- Adults with nonalcoholic fatty liver disease [36]; adults with prediabetes or diabetes [37]; adults in curcumin with piperine trials [18].
- Sources
- [36], [37], [18]
- Caveats
- These are blood enzymes, not measures of liver fat or fibrosis. Results differed by population and formulation [36][37].
13 more
Fatty liver inflammation (NASH)
Shown to improve liver inflammation in adults with NASH without diabetes, 43% against 19% on placebo in one trial. Results on fibrosis disagree.
The numbers and sources
- Effect
- After 96 weeks, nonalcoholic steatohepatitis improved in 43% on 800 IU a day and 19% on placebo. Liver enzymes, fat and inflammation fell. Fibrosis scores did not improve [6]. Across 3 placebo-controlled RCTs in steatohepatitis, vitamin E lowered ALT (MD -12.27, 95% CI -16.66 to -7.89) and raised the odds of fibrosis improvement (OR 1.96, 95% CI 1.25 to 3.09). Resolution of steatohepatitis did not differ significantly (OR 1.71, 95% CI 0.69 to 4.27) [7]. Across 16 RCTs in fatty liver disease, vitamin E lowered AST (WMD -5.94 IU/L) and ALT (WMD -7.33 IU/L) [8]. In a Chinese trial, liver histology improved in 29.3% on 300 mg a day and 14.1% on placebo, with gains in steatosis, inflammation and fibrosis stage [9].
- Population
- 247 adults with nonalcoholic steatohepatitis and without diabetes [6]; adults with steatohepatitis in 2 US trials and 1 Chinese trial [7]; adults with metabolic dysfunction-associated steatotic liver disease [8]; 124 adults without diabetes with biopsy-proven steatohepatitis in China [9].
- Sources
- [6], [7], [8], [9]
- Caveats
- The 3-trial review is small and its resolution result was heterogeneous [7]. The first large trial found no fibrosis improvement [6], while the pooled analysis and the Chinese trial did [7][9]. The enzyme review could not judge fibrosis or long-term liver outcomes [8].
Liver detox
One small trial found a single dose of artichoke extract raised bile flow over placebo. No study in people measured whether it clears toxins from the liver.
The numbers and sources
- Effect
- In a randomized, placebo-controlled, double-blind crossover pilot study, 1.92 g of extract given into the duodenum raised bile secretion by 127.3% at 30 minutes, 151.5% at 60 minutes and 94.3% at 120 minutes over baseline. Placebo peaked at 139.5% at 30 minutes and fell below baseline after 3 hours. The authors reported the differences from placebo as significant (p < 0.01), with more bile on extract at 120 and 150 minutes (p < 0.05). No side effects or lab changes were seen [11]. In rats, the extract raised bile flow and bile acid output [12]. In mice given a toxic dose of acetaminophen, the extract lowered ALT, AST and ALP and reduced liver damage on histology [13].
- Population
- 20 people in a crossover pilot [11]. Wistar rats [12]. 40 male mice [13].
- Sources
- [11], [12], [13]
- Caveats
- One single-dose pilot, with the extract given by tube, not by mouth [11]. Bile flow is not a measure of toxin removal. The liver protection findings come from animals given a poison [13].
Death from liver disease
One large Korean cohort found deaths from liver disease rose with AST from 20 U/L upward, inside the usual reference range.
The numbers and sources
- Effect
- In 142,055 Korean adults followed for 8 years, death from liver disease rose with AST even inside the usual reference range. Compared with AST below 20 IU/L, the adjusted relative risk in men was 2.5 (95% CI 2.0 to 3.0) at 20 to 29 IU/L and 8.0 (6.6 to 9.8) at 30 to 39 IU/L. In women it was 3.3 (1.7 to 6.4) and 18.2 (8.1 to 40.4). The cut-off that predicted liver disease in men with the greatest accuracy was 31 IU/L [5].
- Population
- 94,533 men and 47,522 women aged 35 to 59 in the Korea Medical Insurance Corporation study [5].
- Sources
- [5]
- Caveats
- Observational, one cohort in one country, with deaths taken from death certificates. The abstract does not say which side of the cut-off a result of exactly 31 falls on, so it is not placed as a range on this page [5].
Death in people with Gilbert syndrome
One UK cohort found people with Gilbert syndrome had about half the death rate of people with normal bilirubin.
The numbers and sources
- Effect
- Mortality was 24 per 10,000 person-years in people with a diagnosis of Gilbert syndrome and a raised bilirubin, and 50 per 10,000 in a comparison cohort with normal bilirubin. After accounting for sociodemographic factors and general health indicators, the mortality rate ratio was 0.5 (95% CI 0.4 to 0.7) [10].
- Population
- 4,266 patients with Gilbert syndrome and 21,968 comparison patients in a UK primary care database, with 1,174 deaths over 350,000 person-years [10].
- Sources
- [10]
- Caveats
- One cohort, observational. It counted people whose Gilbert syndrome had been diagnosed and recorded, who may differ from people with the same trait and no diagnosis.
Fatty liver and muscle damage when choline intake is low
Limited evidence: most adults on very low choline diets developed fatty liver or muscle damage, which reversed when choline was added back.
The numbers and sources
- Effect
- On a diet with under 50 mg of choline a day, 77% of men and 80% of postmenopausal women developed fatty liver or muscle damage, against 44% of premenopausal women. Organ function returned when choline was added back. Six men developed signs on 550 mg a day [8]. In people on long-term IV nutrition with low choline, 1 to 4 g a day of IV choline chloride resolved fatty liver within 6 weeks, and it returned in one patient after choline stopped [9].
- Population
- 57 adults in a depletion study [8]; 4 adults on IV nutrition for about 10 years [9].
- Sources
- [8], [9]
- Caveats
- These are depletion and repletion studies, not trials of supplements in people eating normal diets.
Liver fibrosis
One small trial found liver fibrosis improved in 47% on dasatinib plus quercetin against 7% on placebo in adults with fibrotic MASH.
The numbers and sources
- Effect
- Fibrosis improved by at least one stage without MASH worsening in 47% on dasatinib plus quercetin and 7% on placebo (p = 0.02). MASH resolved in 53% against 7% (p = 0.02) [2].
- Population
- 31 adults with fibrotic MASH, median age 56, 58% with type 2 diabetes. Dasatinib 100 mg and quercetin 1,000 mg a day for 3 days a week, over three 7-week cycles [2].
- Sources
- [2]
- Caveats
- One small phase 2 trial with paired liver biopsies.
Heavy exercise
One small study found GGT was almost unaltered for 7 days after a weightlifting session that raised AST and ALT.
The numbers and sources
- Compared with
- The same men's results before exercise, and the laboratory's upper reference limit
- Effect
- After a single 1-hour weightlifting session in healthy men unused to lifting, GGT, bilirubin and alkaline phosphatase were almost unaltered over 7 days, and no man's GGT went above the upper reference limit. AST, ALT, lactate dehydrogenase, creatine kinase and myoglobin all rose significantly over the same days [16].
- Population
- 15 healthy men, mean age 24.5, 14 analyzed [16].
- Sources
- [16]
- Caveats
- One small study of one kind of exercise in men not accustomed to it [16].
Liver function in chronic hepatitis B
One trial without placebo found lower ALT when glutathione was added to entecavir for 3 months in chronic hepatitis B.
The numbers and sources
- Effect
- In one trial, reduced glutathione added to entecavir for 3 months lowered ALT to 31.11 U/L against 56.90 U/L on entecavir alone, from about 348 U/L in both groups. Fibrosis markers fell more, and HBV DNA was undetectable at 48 weeks in 68.89% against 42.22% [10].
- Population
- 90 patients with chronic hepatitis B [10].
- Sources
- [10]
- Caveats
- One trial without placebo. The abstract does not state the glutathione dose or route [10].
Liver fat and insulin sensitivity
One trial found that at equal 10% weight loss, liver insulin sensitivity rose two to three times more than on Mediterranean or very-low-fat diets.
Compared with diets that did not restrict carbohydrates to the level of ketosis
The numbers and sources
- Effect
- Calories were adjusted so every group lost about 10% of body weight. Muscle insulin sensitivity rose about 50% on all three diets. Liver insulin sensitivity rose two to three times more on the very-low-carbohydrate ketogenic diet (p < 0.001), and liver fat, liver fat production, HbA1c and 24-hour glucose and insulin fell most on it. LDL cholesterol, apolipoprotein B and 24-hour triglycerides did not differ between diets [12].
- Population
- Adults with obesity, prediabetes and fatty liver. The abstract does not give the number randomized [12].
- Sources
- [12]
- Caveats
- One small trial. Food was supplied during weight loss, so it measures the diets eaten as written. It does not apply to people without prediabetes and fatty liver.
Conditions found in people with a high MCV
Limited evidence: in one hospital series of 300 adults with an MCV of 100 fL or above, drug therapy, alcohol and liver disease were the most common explanations.
The numbers and sources
- Effect
- The most common cause was drug therapy, followed by alcohol, liver disease and a raised count of young red cells. Megaloblastic blood formation, the pattern seen in vitamin B12 and folate deficiency, accounted for less than 10% of cases. MCV values above 120 fL were usually caused by B12 deficiency [9].
- Population
- 300 consecutive hospitalized adults with an MCV of 100 fL or above at a New York City teaching hospital [9].
- Sources
- [9]
- Caveats
- Hospitalized patients at one center, so the mix of causes is not that of people tested as outpatients. The abstract gives the order of causes without a percentage for each [9].
Liver fat
One small trial found liver fat fell up to 82% over 24 weeks in adults with fatty liver disease. Liver scarring and outcomes were not measured.
The numbers and sources
- Effect
- At 24 weeks, liver fat changed by -42.9% on 1 mg, -57.0% on 4 mg, -81.4% on 8 mg and -82.4% on 12 mg, against +0.3% on placebo. Normal liver fat (under 5%) was reached by 86% on 12 mg and 0% on placebo [10].
- Population
- 98 adults with metabolic dysfunction-associated steatotic liver disease and at least 10% liver fat, a substudy of the phase 2 trial [10].
- Sources
- [10]
- Caveats
- A small phase 2 substudy. It measured liver fat, not liver scarring or outcomes.
Fatty liver inflammation and scarring (MASH)
One trial found liver inflammation resolved in 63% on semaglutide against 34% on placebo in adults with MASH. Outcomes were biopsy findings, not cirrhosis or liver failure.
The numbers and sources
- How common
- Very common: more than 1 in 10 people.
- Effect
- At 72 weeks, steatohepatitis resolved without worsening of fibrosis in 62.9% on semaglutide and 34.3% on placebo (difference 28.7 points, 95% CI 21.1 to 36.2). Fibrosis improved without worsening of steatohepatitis in 36.8% and 22.4% (difference 14.4 points, 95% CI 7.5 to 21.3). Both improved together in 32.7% and 16.1%. Body weight changed by -10.5% and -2.0% [7].
- Population
- The first 800 of 1,197 adults with biopsy-confirmed MASH and fibrosis stage 2 or 3, randomized 2 to 1 to 2.4 mg weekly or placebo (534 and 266) [7].
- Sources
- [7]
- Caveats
- An interim analysis of an ongoing 240-week trial. The outcomes are liver biopsy findings, not cirrhosis, liver failure or death. Funded by Novo Nordisk [7].
- Why limited
- This rests on one trial, of 1,197 people. Strength counts independent studies that agree, so one trial is limited at any size.
Liver health
Small trials without placebo found liver enzymes fell at 2 to 8 g a day in adults with fatty liver disease, but not at 1 g.
The numbers and sources
- Effect
- In three pilot trials, ALT, AST, cytokeratin 18, pro-C3 and a combined liver injury score fell from baseline at 8, 4 and 2 g a day over 12 to 24 weeks, but not at 1 g a day [6].
- Population
- 70 adults with metabolic dysfunction-associated steatotic liver disease and ALT of 50 U/L or more: 44 randomized to 4 or 8 g a day for 12 weeks, 10 on 1 g and 16 on 2 g a day for 24 weeks [6].
- Sources
- [6]
- Caveats
- No placebo group. Results compare each person's levels before and after treatment, and the outcomes are blood markers rather than liver biopsy [6].
Reported risks115
Raised liver enzymes
Shown to raise liver enzymes mildly to moderately in about 15% of people in trials. The label also reports clinical hepatitis.
The numbers and sources
- How common
- Very common: more than 1 in 10 people.
- Effect
- Mild to moderate rises in liver enzymes were seen in about 15% of people in clinical trials, some returning to normal with a lower dose or continued treatment. ALT, AST, GGT and alkaline phosphatase are among the tests the label lists as increased [3]. A meta-analysis found statistically significant changes in hepatic panel means, with ALT averaging 21.77 U/L and AST 22.67 U/L during treatment, and few results outside the normal range [9]. Elevated liver enzymes led to withdrawal from trials in some [1].
- Population
- Isotretinoin capsule trials [3]; 1,574 patients in 26 studies [9]; 760 people in 11 trials [1].
- Sources
- [3], [9], [1]
- Caveats
- The meta-analysis authors concluded that monthly laboratory testing is not supported for standard doses in the standard patient [9].
Raised ALT liver enzyme from unfiltered coffee
Small trials found unfiltered coffee raised the liver enzyme ALT. In one, 8 of 22 people drinking 5 to 6 strong cups of French press coffee exceeded the normal limit.
The numbers and sources
- How common
- Very common: more than 1 in 10 people.
- Effect
- Against filtered coffee, cafetière coffee raised alanine aminotransferase by up to 80%, and by 45% (9 U/L, 95% CI 3 to 15) after 24 weeks. ALT exceeded the upper limit of normal in 8 of 22 people on cafetière coffee, twice the limit in 3. The abstract gives no count for the filtered coffee group. All rises reversed after stopping [18]. Eating 8 g of fine coffee grounds a day for 3 weeks raised ALT by 18 U/L (95% CI 4 to 32) against controls [21].
- Population
- 46 adults, 22 on cafetière coffee and 24 on filtered coffee [18]. 14 volunteers in a parallel study and 15 in a crossover [21].
- Sources
- [18], [21]
- Caveats
- Small trials of high intakes of unfiltered coffee. Cohorts of all coffee types link coffee to lower liver disease (see Outcomes).
Liver enzyme rises in a 6-month trial
One trial found AST rose in 9 of 35 men with type 1 diabetes on a controlled-release GABA for 6 months, mostly briefly.
The numbers and sources
- How common
- Very common: more than 1 in 10 people.
- Effect
- In 35 men with long-standing type 1 diabetes given a controlled-release GABA (200 or 600 mg a day, one arm with alprazolam for 3 months) for 6 months, AST rose in 9 men and 2 withdrew. Most rises were transient, with no difference by dose. The authors report that side effects were common [10].
- Population
- 35 adult men in Sweden with type 1 diabetes for 5 years or more [10].
- Sources
- [10]
- Caveats
- All three arms received GABA, so there was no placebo comparison. It tested a pharmaceutical formulation (Remygen) in people with diabetes [10].
Liver enzymes at high doses
The label reports raised liver enzymes in 19% at 300 mg a day against 0% on placebo, a dose more than 60 times a low dose.
The numbers and sources
- How common
- Very common: more than 1 in 10 people.
- Effect
- The label reports that at 300 mg a day, 19% (5 of 26) of naltrexone recipients and 0% (0 of 24) on placebo developed raised liver enzymes after 3 to 8 weeks. It lists cases of hepatitis and liver dysfunction from clinical development and postmarketing reports [14].
- Population
- Placebo-controlled studies at up to 300 mg a day, summarized in the label [14].
- Sources
- [14]
- Caveats
- 300 mg is more than 60 times a 4.5 mg low dose. The label notes other causes were often present when liver enzymes rose [14].
Liver toxicity
One small trial found liver toxicity in 52% on sustained-release niacin and none on immediate-release, at doses rising to 3,000 mg a day.
The numbers and sources
- How common
- Very common: more than 1 in 10 people.
- Effect
- Liver toxicity developed in 12 of 23 people (52%) on sustained-release niacin and none of 23 on immediate-release niacin. 78% on sustained-release withdrew before reaching 3,000 mg a day, often for rising liver enzymes [7].
- Population
- 46 adults with high LDL cholesterol, dose rising from 500 to 3,000 mg a day in 6-week steps [7].
- Sources
- [7]
- Caveats
- One small trial of sustained-release products from 1994.
High ammonia, and low body temperature with valproate
The label reports raised blood ammonia, in 26% of teenagers on 100 mg against 9% on placebo, and more often with valproate, sometimes with confusion or low body temperature.
The numbers and sources
- How common
- Very common: more than 1 in 10 people.
- Effect
- In adolescents in migraine trials, hyperammonemia occurred in 26% on 100 mg, 14% on 50 mg and 9% on placebo. Hyperammonemia with or without encephalopathy was reported more often with valproic acid, including in people who tolerated either drug alone. Hypothermia below 35 °C has been reported with the combination, with or without high ammonia [3].
- Population
- Adolescents aged 12 to 17 in migraine trials, and postmarketing reports [3].
- Sources
- [3]
- Caveats
- The trial rate is from adolescents. Some cases had no symptoms [3].
Liver enzyme elevation
Shown to raise liver enzymes, about 6 times the odds on placebo, mostly at high doses or with antiepileptic drugs. No cases appeared under 300 mg a day in adults.
The numbers and sources
- How common
- Common: between 1 in 100 and 1 in 10 people.
- Effect
- Across 12 placebo-controlled trials (1,229 participants), CBD raised the odds of liver enzyme elevation (OR 5.85, 95% CI 3.84 to 8.92) and drug-induced liver injury (OR 4.82, 95% CI 2.46 to 9.45). In 28 trials, the pooled rate of enzyme elevation was 7.4% and of liver injury 3.0%. High dose (1,000 mg a day or more, or 20 mg/kg or more) and use with antiepileptic drugs were risk factors. No cases were reported in adults under 300 mg a day, and no severe injury was reported [12]. In an FDA-run trial of 201 healthy adults, 5 mg/kg a day for 28 days raised ALT or AST above 3 times normal in 5.6% on CBD and 0% on placebo, with increased eosinophilia [11]. In the Lennox-Gastaut trial, 9% of CBD patients had raised aminotransferases [2].
- Population
- Trial participants with epilepsy and healthy adults [2][11][12].
- Sources
- [2], [11], [12]
- Caveats
- In the healthy-adult trial, elevations meeting withdrawal criteria were found at day 21 in 2 people and day 28 in 5 [11]. Long-term effects at lower consumer doses were not measured [11].
Raised liver enzymes at 4 g a day
Shown to raise liver enzymes, with abnormal liver tests in 7.0% on paracetamol against 1.8% on placebo in osteoarthritis trials.
The numbers and sources
- How common
- Common: between 1 in 100 and 1 in 10 people.
- Effect
- In the osteoarthritis trials, abnormal liver function tests occurred in 70 per 1,000 on paracetamol and 18 per 1,000 on placebo (RR 3.79, 95% CI 1.94 to 7.39) [8]. In healthy adults given 4 g a day in a controlled inpatient setting, ALT rose above 3 times the upper limit of normal in 31% to 44% of the 4 acetaminophen groups and in none of 39 on placebo. Median peak ALT was 2.78 times that on placebo (95% CI 1.47 to 4.09) [20].
- Population
- 3,541 people with osteoarthritis in 10 trials [8]; 145 healthy adults, 106 on acetaminophen alone or with an opioid and 39 on placebo, for an intended 14 days [20].
- Sources
- [8], [20]
- Caveats
- The reviewers called the clinical importance of the enzyme rises uncertain [8]. The band here is taken from the pooled 7.0%. The healthy-volunteer trial was single-blind [20].
FDA boxed warning: weaker effect of clopidogrel in CYP2C19 poor metabolizers
FDA states clopidogrel has a reduced effect on platelets in people with two loss-of-function CYP2C19 alleles. Studies disagree on how far outcomes differ.
The numbers and sources
- How common
- Common: between 1 in 100 and 1 in 10 people.
- Effect
- The boxed warning states that clopidogrel depends on conversion to an active metabolite, mainly by CYP2C19, that it forms less of this metabolite in people with two nonfunctional alleles, and that tests can identify them [22]. In a meta-analysis of people treated mostly for stenting, 2.2% carried two reduced-function alleles and 26.3% one. Carriers of one or two had more cardiovascular death, heart attack or stroke (HR 1.55 and 1.76) and more stent thrombosis (HR 2.67 and 3.97) than noncarriers [25]. A second meta-analysis found a higher risk of cardiovascular events in carriers (RR 1.18, 95% CI 1.09 to 1.28) that disappeared in studies with 200 or more events (RR 0.97, 95% CI 0.86 to 1.09), with evidence of small-study bias [26]. In a trial after stenting, carriers given genotype-guided treatment had events in 4.0% against 5.9% on clopidogrel (HR 0.66, 95% CI 0.43 to 1.02) [27].
- Population
- 9,685 patients, 91.3% of whom had a stent procedure [25]; 42,016 patients in 32 studies [26]; 5,302 patients, 1,849 of them carriers [27].
- Sources
- [22], [25], [26], [27]
- Caveats
- The band above is the 2.2% who carried two reduced-function alleles in one pooled population [25]. The trial result did not reach statistical significance [27].
Raised liver enzymes on quetiapine
Shown to raise the liver enzyme ALT in 5% on quetiapine against 1% on placebo in schizophrenia and mania trials.
The numbers and sources
- How common
- Common: between 1 in 100 and 1 in 10 people.
- Effect
- In trials of 3 to 12 weeks, increased ALT was reported in 5% on quetiapine and 1% on placebo, and increased AST in 3% against 1% [3].
- Population
- 719 people on quetiapine and 404 on placebo [3].
- Sources
- [3]
- Caveats
- Reported as adverse reactions in short trials, with no data here on how long the rises lasted. The evidence is for quetiapine.
Liver toxicity
Shown to raise blood test signs of liver toxicity in 6% of women in one trial. French case reports include one liver transplant and two likely deaths.
The numbers and sources
- How common
- Common: between 1 in 100 and 1 in 10 people.
- Effect
- In the US trial, 6% of women had blood test evidence of liver toxicity on coumarin [3]. A French pharmacovigilance survey reviewed 34 cases of raised liver enzymes up to June 1996 and judged 15 likely or probably caused by coumarin. ALT exceeded 10 times normal in 68%, 41% were hospitalized, one needed a liver transplant and two others likely died of liver failure [5]. A woman in Sri Lanka developed abnormal liver tests on Lympedim that led to stopping the drug [6]. A risk assessment found that clinical data reveal a subgroup of people more susceptible to coumarin's liver toxicity than the animals studied, and derived a tolerable daily intake of 0.1 mg per kg of body weight [7].
- Population
- 140 women in an RCT [3]. Case reports to French regional centers and the manufacturer [5]. A single case [6].
- Sources
- [3], [5], [6], [7]
- Caveats
- Most French cases occurred 2 to 6 months into treatment at about 90 mg a day [5]. The cause of the higher susceptibility in some people is not known [7].
Ezetimibe: liver enzymes
Mixed results: the label reports more liver enzyme rises with ezetimibe plus a statin, 1.3% against 0.4%, while a Cochrane review found little or no difference in liver problems.
The numbers and sources
- How common
- Common: between 1 in 100 and 1 in 10 people.
- Effect
- Consecutive transaminase rises of 3 times the upper limit or more occurred in 1.3% on ezetimibe with a statin and 0.4% on a statin alone [2]. Adding ezetimibe to a statin may make little or no difference to hepatopathy (RR 1.14, 95% CI 0.96 to 1.35) [3].
- Population
- 11,308 patients in 28 label trials, median 8 weeks [2]; 20,687 people in 4 trials [3].
- Sources
- [2], [3]
- Caveats
- Low-quality evidence in [3].
103 more
Fenofibrate: liver enzymes and liver injury
Shown to raise ALT above 3 times normal in 5.3% against 1.1% on placebo. The label warns of serious liver injury after approval, including transplant and death.
The numbers and sources
- How common
- Common: between 1 in 100 and 1 in 10 people.
- Effect
- In 10 placebo-controlled trials, ALT rose above 3 times the upper limit in 5.3% on fenofibrate and 1.1% on placebo. In label trials, increased AST was 3.4% against 0.5% and abnormal liver tests 7.5% against 1.4%, and raised liver tests caused 1.6% to stop. In an 8-week study, ALT or AST rises of 3 times or more reached 13% at the higher doses. The label reports serious drug-induced liver injury after approval [7].
- Population
- 439 people on fenofibrate and 365 on placebo, and a pooled 10-trial analysis [7].
- Sources
- [7]
- Caveats
- Postmarketing liver injury reports have no denominator.
Liver enzyme rises
Shown to raise ALT or AST to 3 times normal or more in about 1% of women on Oriahnn and under 1% on Myfembree, against 0% to 0.4% on placebo.
The numbers and sources
- How common
- Common: between 1 in 100 and 1 in 10 people.
- Effect
- ALT rose to 3 times the upper limit or more in 1.1% (4 of 379) on Oriahnn and AST in 1.3% (5 of 379), against none on placebo, peaking at 8 and 6 times the limit, and returning to baseline within 4 months [6]. On Myfembree in fibroid trials, ALT rose in 0.4% (1 of 254) and AST in 0.8% (2 of 254), against 0% and 0.4% on placebo [2]. On leuprolide, 3% of women with fibroids had post-treatment transaminases at least twice baseline and above normal [10].
- Population
- Premenopausal women with fibroids in the placebo-controlled trials [2][6][10].
- Sources
- [6], [2], [10]
- Caveats
- Few events. Both labels contraindicate use in liver disease [2][6].
Raised liver enzymes
Shown to raise the liver enzyme ALT to three times the normal limit or more in 2.0% against 0.3% on placebo in short trials.
The numbers and sources
- How common
- Common: between 1 in 100 and 1 in 10 people.
- Effect
- ALT rose to at least three times the upper limit of normal in 2.0% (8 of 424) on mirtazapine and 0.3% (1 of 328) on placebo. Some people stopped the drug, and in others levels returned to normal while it was continued [7].
- Population
- 424 adults on mirtazapine and 328 on placebo in short-term US trials [7].
- Sources
- [7]
- Caveats
- Short trials, with no rate of liver injury given.
Liver enzyme changes
Shown to raise mean GGT and alkaline phosphatase in placebo-controlled trials, mostly within the normal range. ALT and AST did not differ.
The numbers and sources
- How common
- Common: between 1 in 100 and 1 in 10 people.
- Effect
- Mean GGT and alkaline phosphatase were higher after modafinil but not placebo, and rose over time, with few values outside the normal range. ALT, AST, albumin and total bilirubin did not differ. "Abnormal liver function" was reported as an adverse reaction in 2% vs 1% on placebo [3].
- Population
- Adults in placebo-controlled label trials [3].
- Sources
- [3]
- Caveats
- The label gives no size for the mean change. The band above is the adverse-reaction rate.
Tizanidine: liver test abnormalities
Shown to raise liver test abnormalities, reported in 6% on tizanidine against 2% on placebo. The label warns of liver cell injury.
The numbers and sources
- How common
- Common: between 1 in 100 and 1 in 10 people.
- Effect
- In the label's multiple-dose controlled studies, a liver test abnormality was reported in 6% on tizanidine and 2% on placebo. The label states that tizanidine may cause hepatocellular liver injury and calls for aminotransferase monitoring at baseline and 1 month after the maximum dose is reached [12].
- Population
- 264 people on tizanidine and 261 on placebo in the label's multiple-dose studies [12].
- Sources
- [12]
- Caveats
- The label table does not say which liver tests were abnormal or by how much.
Liver injury
Shown to raise liver enzymes on duloxetine, with ALT above 3 times the upper limit in 1.25% against 0.45% on placebo. Liver failure has been reported.
The numbers and sources
- How common
- Common: between 1 in 100 and 1 in 10 people.
- Effect
- In adult placebo-controlled duloxetine trials, ALT rose above 3 times the upper limit of normal in 1.25% (144 of 11,496) on duloxetine and 0.45% (39 of 8,716) on placebo, with a dose-response relationship in fixed-dose studies. Liver enzyme rises led 0.3% (92 of 34,756) to stop the drug. The label also reports cases of hepatic failure, some fatal, and cholestatic jaundice [18].
- Population
- Adults in duloxetine clinical trials for any indication, and postmarketing reports [18].
- Sources
- [18]
- Caveats
- The evidence in this block is for duloxetine only. The serious cases come from reports with no denominator.
Liver enzyme rises and liver injury
Shown to raise the liver enzyme ALT above twice the reference limit in 1.9% on glimepiride against 0.8% on placebo. Liver failure appears in postmarketing reports.
The numbers and sources
- How common
- Common: between 1 in 100 and 1 in 10 people.
- Effect
- In 11 pooled placebo-controlled trials, 1.9% on glimepiride and 0.8% on placebo developed serum ALT above 2 times the upper reference limit. Postmarketing reports include impaired liver function with cholestasis and jaundice, and hepatitis that may progress to liver failure [3]. The glyburide label states that liver enzyme elevations were more frequent when glyburide was taken with bosentan, and lists the combination as contraindicated [14].
- Population
- 745 people on glimepiride and 294 on placebo in 11 trials [3].
- Sources
- [3], [14]
- Caveats
- Postmarketing reports have no denominator.
Liver enzymes and liver injury
The label reports rises in AST, ALT and alkaline phosphatase in 1% or more of people, and cases of reversible liver injury.
The numbers and sources
- How common
- Common: between 1 in 100 and 1 in 10 people.
- Effect
- The label lists increases in AST, ALT and alkaline phosphatase among reactions occurring in 1% or more, and states that cases of reversible liver injury have occurred, with hepatic necrosis and granulomatous hepatitis reported in under 1% [5]. In a dose-escalation trial, mild rises in liver tests were frequent in both the escalation and control groups [3].
- Population
- People taking allopurinol, as described in the label [5]; 183 people with gout [3].
- Sources
- [5], [3]
- Caveats
- The label gives no placebo rate and no size for the enzyme rises.
Raised liver enzymes
The label reports ALT or AST above twice the upper limit in 1% to 2% of patients on ondansetron with chemotherapy, transient, with chemotherapy's role unclear.
The numbers and sources
- How common
- Common: between 1 in 100 and 1 in 10 people.
- Effect
- AST or ALT exceeded twice the upper limit of normal in about 1% to 2% of 723 patients receiving ondansetron with cyclophosphamide-based chemotherapy. The rises were transient and not related to dose or duration, and symptomatic liver disease did not occur. Liver failure and death have been reported in cancer patients also taking liver-toxic drugs [3].
- Population
- 723 patients in US chemotherapy trials [3].
- Sources
- [3]
- Caveats
- No placebo rate is given, and the label states the role of chemotherapy is unclear [3].
Gallbladder disease
Shown to raise gallbladder and bile duct disease about 37% across GLP-1 drug trials. In REWIND, serious acute cholecystitis occurred in 0.5% on dulaglutide against 0.3%.
The numbers and sources
- How common
- Uncommon: between 1 in 1,000 and 1 in 100 people.
- Effect
- In REWIND, gallstones occurred at 0.62 per 100 patient-years on dulaglutide and 0.56 on placebo, and serious acute cholecystitis in 0.5% against 0.3% [1]. Across GLP-1 receptor agonist trials, the relative risk of gallbladder or biliary disease was 1.37 (95% CI 1.23 to 1.52), higher at higher doses and with longer use [11].
- Population
- 9,901 adults in REWIND, as reported in the label [1]; 103,371 patients in 76 trials [11].
- Sources
- [1], [11]
- Caveats
- The meta-analysis pools all GLP-1 receptor agonists [11].
Liver enzyme rises and severe muscle injury
Shown to raise liver enzymes in about 0.2% on high doses. Severe muscle injury rose with dose, and myopathy was more frequent on simvastatin 80 mg.
The numbers and sources
- How common
- Uncommon: between 1 in 1,000 and 1 in 100 people.
- Effect
- Across 41 RCTs, transaminase elevation on high-dose statins occurred at a rate of 0.2% (95% CI 0.004 to 0.4%), and creatine kinase rises under three times the upper limit at 0.43% (95% CI 0.11 to 0.75%). Rhabdomyolysis risk rose with dose, and myopathy was more frequent on simvastatin 80 mg than on other statins [9].
- Population
- 64,728 participants in 41 RCTs across several diseases [9].
- Sources
- [9]
- Caveats
- Rates varied by disease group, and the abstract does not give the rhabdomyolysis rate [9].
Drug reaction with organ involvement (DRESS) and liver failure
The label reports multiorgan hypersensitivity reactions, some fatal, and deaths from organ or liver failure in 2 of 3,796 adults and 4 of 2,435 children in trials.
The numbers and sources
- How common
- Rare: fewer than 1 in 1,000 people.
- Effect
- Multiorgan hypersensitivity (DRESS) has occurred, presenting with fever, rash or swollen lymph nodes alongside hepatitis, kidney inflammation, blood abnormalities or heart or muscle inflammation. Deaths from acute multiorgan failure and liver failure occurred in 2 of 3,796 adults and 4 of 2,435 children in epilepsy trials. Isolated liver failure without rash has also been reported [4].
- Population
- Adults and children in epilepsy trials, and postmarketing reports [4].
- Sources
- [4]
- Caveats
- Trial deaths are counted without a placebo comparison.
Serious liver injury
Case reports: serious liver injury occurred on the daily fibroid dose, with five liver transplants reported in Europe. One review estimated severe injury at 13.5 per 100,000 users.
The numbers and sources
- How common
- Rare: fewer than 1 in 1,000 people.
- Effect
- In 2018 the European Medicines Agency reviewed serious liver injury, including four cases that led to liver transplant, among about 765,000 women treated, and judged a role for Esmya possible in 8 serious cases [11]. A US FDA review of postmarketing reports found 9 serious non-US cases, 5 with liver transplant or death judged probably caused by ulipristal [12]. A 2025 review counted five transplant cases across the 2018 and 2020 European reviews, a transplant reporting rate of about 0.52 per 100,000 users, and estimated severe drug-induced liver injury at 13.5 per 100,000 [13]. In a Korean cohort of 11,445 matched women per group, mild liver disease was slightly more common on ulipristal than on GnRH agonists (RR 1.094), while severe and toxic liver disease did not differ and no transplants occurred [14].
- Population
- Spontaneous reports to EMA and FDA [11][12]; regulator documents reviewed in [13]; 20,926 Korean women on ulipristal and 17,207 on GnRH agonists before matching [14].
- Sources
- [11], [12], [13], [14]
- Caveats
- Spontaneous reports cannot prove cause, and the rate is an estimate from exposure figures [13]. The 2025 review's authors argue the regulatory measures were overly cautious for the size of the risk [13]. The EMA states no concern about liver injury has been raised with the single-dose emergency contraceptive [15].
Liver cirrhosis
Linked to sharply higher cirrhosis risk with heavy drinking, 12 to 25 times in women at 5 or more drinks a day. Women had higher risk at the same intake.
The numbers and sources
- How common
- How often is not known: the sources give no rate.
- Effect
- Against long-term abstainers, 1 drink a day was linked to higher cirrhosis risk in women and not in men. At 5 to 6 drinks a day the relative risk was 12.44 (95% CI 6.65 to 23.27) in women and 3.80 (95% CI 0.85 to 17.02) in men, and at 7 or more 24.58 in women and 6.93 in men. Occasional drinkers had no higher risk [16]. An earlier review of 17 studies found the same amount carried higher risk in women, and a larger effect on cirrhosis death than on cirrhosis illness [17].
- Population
- 2,629,272 participants with 5,505 cirrhosis cases in 9 studies [16]. 17 cohort and case-control studies [17].
- Sources
- [16], [17]
- Caveats
- Observational. Heterogeneity pointed to other factors, such as genes, weight and drinking pattern [16].
Liver injury
Case reports: people developed jaundice and liver injury after 2 to 12 weeks, most recovering within months. Three with existing liver disease died.
The numbers and sources
- How common
- How often is not known: the sources give no rate.
- Effect
- Five cases from Iceland and the US Drug-Induced Liver Injury Network developed jaundice after 2 to 12 weeks, with cholestatic or mixed injury and itching lasting 5 to 20 weeks. None had liver failure, and liver tests normalized in 1 to 5 months [16]. In India, 8 patients had liver injury from single-ingredient ashwagandha. Three with existing chronic liver disease presented with acute-on-chronic liver failure and all three died. Chemical analysis found no adulteration [17]. Two further Serbian cases were rated probable by the updated RUCAM, and a review of published cases notes one liver transplant [15].
- Population
- Case series and case reports [16][17][15].
- Sources
- [16], [17], [15], [18], [7]
- Caveats
- Case reports cannot give a rate. In one Icelandic case, rhodiola was a possible co-cause [16]. In an 8-week trial of 1,002 adults, liver tests did not differ from placebo [18], and a review of 23 root extract trials found liver markers within normal ranges [7].
German regulator advice
Germany's BfR advises children, pregnant and breastfeeding women, and people with liver disease not to take ashwagandha, citing case reports of liver damage.
The numbers and sources
- How common
- How often is not known: the sources give no rate.
- Effect
- In September 2024 Germany's Federal Institute for Risk Assessment (BfR) advised children, pregnant and breastfeeding women, and people with current or past liver disease not to take ashwagandha preparations, and advised restraint for others. It cited case reports of liver damage including acute liver failure, possible effects on cortisol, blood sugar, thyroid and sex hormones, and possible interactions with antidiabetic, blood pressure and immunosuppressive drugs. It found that no guidance value for a harmless intake could be derived [19].
- Population
- German consumers [19].
- Sources
- [19]
- Caveats
- The BfR notes that human trials did not systematically record adverse effects. It names risk reassessments by Denmark's DTU, the Dutch RIVM, France's ANSES and Australia's TGA [19].
Drug-metabolizing enzyme inhibition
Shown to raise levels of drugs cleared by CYP2D6, CYP2C9 and CYP3A4 in healthy men, and cyclosporine levels by about 29% in kidney transplant recipients.
The numbers and sources
- How common
- How often is not known: the sources give no rate.
- Effect
- In healthy men, 300 mg three times a day for 2 weeks raised the dextromethorphan ratio ninefold (CYP2D6), doubled the losartan ratio (CYP2C9), and raised midazolam AUC by 40% with 27% lower clearance (CYP3A4) [11]. In 52 kidney transplant recipients, cyclosporine trough levels ended 29.3% higher with berberine than without. In 6 patients, cyclosporine AUC rose 34.5% after 12 days [12].
- Population
- Healthy male volunteers [11]. Kidney transplant recipients [12].
- Sources
- [11], [12]
- Caveats
- Probe-drug and transplant studies. Other enzymes tested (CYP2C19, CYP1A2) did not change [11].
Liver and kidney markers in trials
Liver and kidney markers did not worsen in most trials. Creatinine rose slightly in 3 trials, and doses above 2,000 mg a day were linked to a BUN change.
The numbers and sources
- How common
- How often is not known: the sources give no rate.
- Effect
- In 19 trials (1,295 participants), Nigella sativa lowered ALP by 10.8 U/L and BUN by 1.016. Creatinine rose by 0.070 in 3 trials with covariate-adjusted analyses, and doses above 2,000 mg a day were linked to a BUN change in 2 trials [11]. In 70 healthy adults given 200 mg a day of a thymoquinone-rich oil for 90 days, liver and kidney tests did not change and no serious adverse effects were reported [12].
- Population
- Adults in RCTs [11]. 70 healthy adults [12].
- Sources
- [11], [12]
- Caveats
- The safety trial's corresponding author is at Akay Natural Ingredients, the maker of the tested oil [12]. Heterogeneity for ALP and BUN was high [11]. In 16 RCTs in type 2 diabetes, AST, ALT and creatinine did not change significantly [7].
Drug interactions
Shown to raise clobazam levels by about 60% in children with epilepsy, and to change levels of several other antiepileptic drugs.
The numbers and sources
- How common
- How often is not known: the sources give no rate.
- Effect
- In 13 children taking clobazam, CBD raised clobazam levels by a mean 60% and its active metabolite by 500% at 4 weeks, and side effects in 10 of 13 eased when clobazam was reduced [13]. In 81 adults and children, rising CBD doses raised serum topiramate, rufinamide and N-desmethylclobazam and lowered clobazam. AST and ALT were higher in those also taking valproate [14].
- Population
- People with refractory epilepsy [13][14].
- Sources
- [13], [14]
- Caveats
- Open-label studies at prescription doses. Except for clobazam and its metabolite, the level changes stayed within accepted therapeutic ranges [14].
FDA position
FDA states it has not found adequate evidence on how much CBD can be consumed, and for how long, before causing harm, naming liver harm and drug interactions.
The numbers and sources
- How common
- How often is not known: the sources give no rate.
- Effect
- On January 26, 2023, FDA said it was not apparent how CBD products could meet safety standards for dietary supplements or food additives, and that it had not found adequate evidence to determine how much CBD can be consumed, and for how long, before causing harm. It named potential harm to the liver, interactions with certain medications, possible harm to the male reproductive system, and concern for children and those who are pregnant [20].
- Population
- US consumers [20].
- Sources
- [20]
- Caveats
- The FDA-run 2025 trial found no change in testosterone, inhibin B or thyroid hormones after 28 days [11].
Liver injury
Linked to liver injury in 22 US registry cases, mostly in products with green tea. 91% were hospitalized, one needed a transplant and one died.
The numbers and sources
- How common
- How often is not known: the sources give no rate.
- Effect
- Among 1,418 patients enrolled in the US Drug-Induced Liver Injury Network from 2004 to 2018, 22 cases were adjudicated with high confidence as liver injury from garcinia alone (5), with green tea (16) or with ashwagandha (1). Patients were 17 to 54 years old and injury began 13 to 223 days (median 51) after starting. 91% were hospitalized, one needed a liver transplant and one died. The HLA-B*35:01 allele was present in 55% of garcinia cases against 19% of other supplement cases [7]. Separate reports describe fulminant liver failure needing transplant [8] and acute liver failure in 4 women taking garcinia for weight loss [9].
- Population
- US registry cases [7]; case reports and series [8][9].
- Sources
- [7], [8], [9]
- Caveats
- Most network cases involved multi-ingredient products, and 16 of 22 also contained green tea. Liver injury from garcinia and from green tea could not be told apart clinically [7].
Liver enzymes and side effects in trials
Liver enzymes did not change in short trials. In one trial, gut side effects were twice as common as on placebo.
The numbers and sources
- How common
- How often is not known: the sources give no rate.
- Effect
- In pooled trials, garcinia did not change ALT (Hedges' g 0.27, 95% CI -0.20 to 0.73) or AST (Hedges' g -0.08, 95% CI -0.43 to 0.26) [6]. In one trial included in [1], gastrointestinal adverse events were twice as common with hydroxycitric acid as with placebo [1].
- Population
- 444 participants in 9 publications [6]; RCTs [1].
- Sources
- [1], [6]
- Caveats
- Trials were short and too small to detect rare liver injury. The registry and case findings [7][8][9][10] and the trial enzyme data describe different things.
Higher blood levels of some statins
The FDA states grapefruit juice can interact with some statins. One glass daily raised simvastatin levels 3.6-fold in a small trial, and pravastatin was unaffected.
The numbers and sources
- How common
- How often is not known: the sources give no rate.
- Effect
- The FDA names simvastatin and atorvastatin among drugs grapefruit juice can interact with, and states that too much statin in the body raises the risk of liver and muscle damage [1]. Double-strength juice three times a day raised simvastatin exposure (AUC) 16-fold [6]. One 200 mL glass a day for 3 days raised simvastatin AUC 3.6-fold and simvastatin acid 3.3-fold [8]. Double-strength juice raised atorvastatin acid AUC 2.5-fold, and did not change pravastatin [7]. In 60 patients who kept their atorvastatin dose and added 300 mL a day for 90 days, serum atorvastatin rose 19% to 26%, lipids changed negligibly, and liver tests and creatine phosphokinase showed no adverse effects [9]. A modeling review estimated a daily glass raises simvastatin and lovastatin blood levels about 260% when taken together and atorvastatin about 80%, and judged the added rhabdomyolysis risk minimal [18].
- Population
- 10 healthy volunteers [6]. 10 healthy volunteers [8]. 12 and 11 healthy volunteers [7]. 130 patients on stable atorvastatin [9]. A modeling analysis [18].
- Sources
- [1], [6], [8], [7], [9], [18]
- Caveats
- The pharmacokinetic studies were small and in healthy volunteers, and none measured muscle injury. The size of the interaction depended on the amount of juice and the statin [6][8][7].
Raised liver enzymes at high EGCG doses
EFSA's panel concluded that trials show 800 mg or more of EGCG a day as a supplement raises liver enzymes compared with control.
The numbers and sources
- How common
- How often is not known: the sources give no rate.
- Effect
- EFSA's panel concluded that interventional trials show 800 mg or more of EGCG a day taken as a supplement produced a statistically significant rise in serum transaminases compared with control [7]. A systematic review of 34 RCTs found liver-related adverse events in 7 people on green tea extract and 1 on control (OR 2.1, 95% CI 0.5 to 9.8), mostly mild enzyme rises [9].
- Population
- Adults in green tea extract trials [7][9].
- Sources
- [7], [9]
- Caveats
- Trial-based rates are small and imprecise. The trial review searched to December 2013 [9].
Liver injury
Linked to 3% of US registry liver injury cases, 35% of them severe, with 3 needing transplants. USP linked injury to 140 to about 1,000 mg EGCG a day.
The numbers and sources
- How common
- How often is not known: the sources give no rate.
- Effect
- Of 1,414 adjudicated cases in the US Drug-Induced Liver Injury Network, 40 (3%) were attributed to green tea. Injury was hepatocellular in 95%, 83% were jaundiced, 35% were severe, and 3 (8%) needed a liver transplant. Injury recurred in 3 people on re-exposure. HLA-B*35:01 was found in 72% of green tea cases versus 11% of population controls [10]. A USP review found hepatotoxicity associated with 140 to about 1,000 mg of EGCG a day, higher catechin bioavailability with bolus doses taken fasting, and no evidence that contaminants caused the injury [8].
- Population
- US registry cases [10]; systematic review of case reports, animal and human data [8].
- Sources
- [8], [10]
- Caveats
- Green tea is a frequent component of implicated multi-ingredient products [10]. EFSA also notes rare liver injury after traditional green tea infusions [7].
Liver failure and transplants
Case reports: liver failure and transplants in 11 kava users by 2002, including a 14-year-old girl. Most likely cases exceeded dose or duration limits.
The numbers and sources
- How common
- How often is not known: the sources give no rate.
- Effect
- By 2002, 11 people in Germany, Switzerland and the US who used kava products had liver failure and liver transplants. The two US cases were a 45-year-old woman after about 8 weeks of use and a 14-year-old girl, both previously healthy [5]. A causality analysis of 26 cases suspected by regulators found 8 with some degree of causality for kava with or without other drugs. In 6 of these, kava was taken above the regulatory dose or for longer than the regulatory duration. One patient who followed the dose and duration limits had probable kava liver injury [6].
- Population
- Case reports to US and European regulators [5][6].
- Sources
- [5], [6]
- Caveats
- Case reports cannot give a rate. The CDC report notes FDA research suggesting fewer than 1% of severe supplement adverse events are reported to the FDA [5]. Most European cases involved ethanol or acetone extracts [5].
Regulator actions
FDA issued a 2002 advisory on the potential risk of severe liver injury with kava, and five other countries restricted kava sales.
The numbers and sources
- How common
- How often is not known: the sources give no rate.
- Effect
- On March 25, 2002, the FDA issued a consumer advisory on the potential risk of severe liver injury with kava products, in response to five case reports. Germany, Switzerland, Canada, Australia and France restricted the sale of kava products, citing liver adverse events and documented liver toxicity after re-exposure [5].
- Population
- Consumers in the named countries [5].
- Sources
- [5]
- Caveats
- These actions date from 2002, as reported by the CDC [5]. Later changes to these restrictions were not checked for this page.
Gallbladder and bile duct disease
Across GLP-1 drug trials, shown to raise gallbladder and bile duct disease about 37%, and more in weight loss trials, at higher doses and with longer use.
The numbers and sources
- How common
- How often is not known: the sources give no rate.
- Effect
- Across GLP-1 receptor agonist trials, the relative risk of gallbladder or biliary disease was 1.37 (95% CI 1.23 to 1.52), and 2.29 (95% CI 1.64 to 3.18) in weight loss trials. Risk was higher at higher doses and with longer use [12]. Gallstones were observed more often with liraglutide alone than with liraglutide plus exercise [6].
- Population
- 76 RCTs with 103,371 patients [12]; 195 adults [6].
- Sources
- [12], [6]
- Caveats
- [12] pools all GLP-1 receptor agonists, not only liraglutide.
Gallbladder and bile duct disease
Shown to raise gallbladder and bile duct disease about 37% across GLP-1 drug trials, more at higher doses and longer use. The analysis pooled all GLP-1 drugs.
The numbers and sources
- How common
- How often is not known: the sources give no rate.
- Effect
- Across GLP-1 receptor agonist trials, the relative risk of gallbladder or biliary disease was 1.37 (95% CI 1.23 to 1.52), including gallstones (RR 1.27) and cholecystitis (RR 1.36). In weight loss trials it was 2.29 (95% CI 1.64 to 3.18). Risk was higher at higher doses and longer use [8].
- Population
- 76 RCTs with 103,371 patients, mean age 57.8 [8].
- Sources
- [8]
- Caveats
- The meta-analysis pools all GLP-1 receptor agonists, not only semaglutide.
Lower levels of HIV drugs and other medicines
Shown to lower levels of the HIV drug indinavir about 57% in healthy volunteers. Reports show lower levels or effect of warfarin, cyclosporine, digoxin and birth control pills.
The numbers and sources
- How common
- How often is not known: the sources give no rate.
- Effect
- In healthy volunteers, St. John's wort cut exposure to the HIV protease inhibitor indinavir by a mean 57% and its trough level by 81% [10]. A systematic review of literature and European regulator reports found clinically significant interactions with warfarin, phenprocoumon, cyclosporine, HIV protease inhibitors, theophylline, digoxin and oral contraceptives, each lowering the drug's level or effect. Sweden and the UK sent warnings to health professionals and patients [11].
- Population
- Healthy volunteers [10]. Published reports and spontaneous reports to European regulators [11].
- Sources
- [10], [11]
- Caveats
- The review notes the degree of enzyme induction varies with product content [11].
Liver injury
Case reports: 10 US cases of turmeric liver injury, one fatal, plus 7 Italian cases with high-dose, high-bioavailability products, most improving after stopping.
The numbers and sources
- How common
- How often is not known: the sources give no rate.
- Effect
- The US Drug-Induced Liver Injury Network found 10 turmeric cases, all since 2011 and 6 since 2017. Injury was hepatocellular in 9, latency was 1 to 4 months, 5 were hospitalized and 1 died of acute liver failure. Seven carried HLA-B*35:01 (allele frequency 0.450 versus 0.056 to 0.069 in population controls), and 3 of 7 tested products also contained piperine [49]. In Tuscany, 7 cases of acute hepatitis were linked to high-bioavailability, high-dose curcumin products, with most improving after stopping. A systematic review found 23 more published cases [48].
- Population
- US case series [49]; Italian phytovigilance reports and published cases [48].
- Sources
- [49], [48]
- Caveats
- Case series give no rate. In the 23 published cases, most people were also taking at least one other medicine [48].
Acetaminophen
Shown to raise the INR by a mean 0.62 against placebo in people on vitamin K antagonists, more at higher daily doses. No major bleeding was reported.
The numbers and sources
- How common
- How often is not known: the sources give no rate.
- Effect
- Across seven randomized trials, acetaminophen raised the INR by a mean 0.62 (95% CI 0.46 to 0.78) against placebo in people on vitamin K antagonists, with a mean rise of 0.17 for each daily gram (95% CI 0.004 to 0.33). The studies did not report any major bleeding event [25].
- Population
- 225 patients in seven randomized trials [25].
- Sources
- [25]
- Caveats
- Small trials measuring a lab value, not bleeding.
Liver failure
The label warns of a syndrome that starts with jaundice or hepatitis and can progress to liver failure and death. It gives no rate.
The numbers and sources
- How common
- How often is not known: the sources give no rate.
- Effect
- The lisinopril label states that ACE inhibitors have been associated with a syndrome that starts with cholestatic jaundice or hepatitis and progresses to fulminant hepatic necrosis and sometimes death. The mechanism is not understood [3].
- Population
- People taking ACE inhibitors, as described in the label [3].
- Sources
- [3]
- Caveats
- No rate and no comparison group are given.
Boxed warning on prescription combinations: liver failure
FDA states acetaminophen has been associated with acute liver failure, at times resulting in liver transplant and death, mostly at doses above 4,000 mg a day.
The numbers and sources
- How common
- How often is not known: the sources give no rate.
- Effect
- The boxed warning on hydrocodone and acetaminophen tablets states that acetaminophen has been associated with cases of acute liver failure, at times resulting in liver transplant and death. Most cases involved doses above 4,000 mg a day and often more than one product containing acetaminophen. The label adds that the risk is higher with underlying liver disease and with alcohol [22]. The over-the-counter label warns that severe liver damage may occur with more than 4,000 mg in 24 hours, with other drugs containing acetaminophen, or with 3 or more alcoholic drinks every day [19]. Of 662 consecutive patients with acute liver failure at 22 US centers over 6 years, 275 (42%) were attributed to acetaminophen. The median dose taken was 24 g. Overdoses were unintentional in 48% and intentional in 44%. Among unintentional cases, 38% took two or more acetaminophen products at once and 63% used combinations containing a narcotic. Overall 65% survived, 27% died without a transplant and 8% received a transplant [23].
- Population
- People taking acetaminophen, as described in the labels [22][19]; 662 patients with acute liver failure at 22 US tertiary centers, 1998 to 2003 [23].
- Sources
- [22], [19], [23]
- Caveats
- The registry counts people who already had liver failure, so it gives no rate among users [23].
With alcohol
The label warns of severe liver damage with 3 or more alcoholic drinks a day. One short trial in alcoholic patients found no rise in AST.
The numbers and sources
- How common
- How often is not known: the sources give no rate.
- Effect
- The over-the-counter label warns that severe liver damage may occur with 3 or more alcoholic drinks every day while using the product [19]. The prescription combination label states that the risk of acute liver failure is higher in people who drink alcohol while taking acetaminophen [22]. In patients entering an alcohol detoxification center, 1,000 mg four times a day for 2 days did not change liver tests against placebo. Mean AST on day 4 was 38.0 against 37.5 U/L, and AST rose above 120 U/L in 4 patients on acetaminophen and 5 on placebo [29].
- Population
- 201 patients in alcohol detoxification, 102 on acetaminophen and 99 on placebo, with baseline AST and ALT of 120 U/L or below [29].
- Sources
- [19], [22], [29]
- Caveats
- The trial gave 2 days of treatment after alcohol had been eliminated, and excluded people with raised enzymes at entry [29].
Liver fat
One trial found a liter of regular cola daily for 6 months raised liver fat more than milk, diet cola or water. Cross-sectional studies link sugary drinks to fatty liver.
The numbers and sources
- How common
- How often is not known: the sources give no rate.
- Effect
- Relative changes in liver fat, muscle fat (117% to 221%) and visceral fat (24% to 31%) were larger with regular cola than with the other three drinks, while total fat mass did not differ [12]. In 4 cross-sectional studies, the highest intake of sugar-sweetened drinks was linked to 40% higher odds of non-alcoholic fatty liver disease (OR 1.40, 95% CI 1.07 to 1.82) [15]. An umbrella review rated the link between added sugars and fat stored in organs as moderate quality [14].
- Population
- 47 overweight adults [12]. 6,326 participants with 1,361 cases [15]. 73 meta-analyses [14].
- Sources
- [12], [15], [14]
- Caveats
- One small trial. Cross-sectional studies cannot show which came first.
Other drug interactions
The label states that alkalinizing agents raise amphetamine blood levels, acidifying agents lower them, and tricyclic antidepressants can strengthen cardiovascular effects. It gives no rate.
The numbers and sources
- How common
- How often is not known: the sources give no rate.
- Effect
- The amphetamine label states that gastrointestinal or urinary alkalinizing agents increase blood levels and potentiate the action of amphetamine, acidifying agents lower blood levels, CYP2D6 inhibitors may increase exposure, tricyclic antidepressants may potentiate cardiovascular effects, and proton pump inhibitors shorten the time to peak level [17]. The methylphenidate label lists blood pressure drugs as an interaction that calls for closer blood pressure monitoring and dose adjustment of the blood pressure drug, and states that the risk of movement side effects from risperidone may rise when either dose is changed [21].
- Population
- Labeling statements [17][21].
- Sources
- [17], [21]
- Caveats
- Changes in drug levels or effects, with no rate of harm.
Interaction with acetaminophen and NSAID pain relievers
FDA requires over-the-counter acetaminophen labels to warn of severe liver damage, and NSAID labels of stomach bleeding, with 3 or more drinks every day.
The numbers and sources
- How common
- How often is not known: the sources give no rate.
- Effect
- Federal labeling rules require acetaminophen products to warn that severe liver damage may occur with 3 or more alcoholic drinks every day while using the product, and NSAID products to warn that the chance of severe stomach bleeding is higher with 3 or more drinks a day [23]. A review states that alcohol enhances the breakdown of acetaminophen into a toxic product [22].
- Population
- US labeling regulation [23]. Narrative review [22].
- Sources
- [23], [22]
- Caveats
- The regulation states the warning and gives no rate.
Interaction with strong CYP3A4 inhibitors
The labels warn against tamsulosin, alfuzosin or silodosin with strong CYP3A4 inhibitors such as ketoconazole or clarithromycin, which raise drug levels. Ketoconazole nearly tripled tamsulosin exposure.
The numbers and sources
- How common
- How often is not known: the sources give no rate.
- Effect
- Ketoconazole raised tamsulosin peak levels 2.2-fold and exposure 2.8-fold, and the label advises against combining tamsulosin 0.4 mg with strong CYP3A4 inhibitors [4]. Alfuzosin is contraindicated with ketoconazole, itraconazole and ritonavir [6]. Silodosin is contraindicated with strong CYP3A4 inhibitors including ketoconazole, clarithromycin, itraconazole and ritonavir [5].
- Population
- Pharmacokinetic studies in the labels [4][6][5].
- Sources
- [4], [6], [5]
- Caveats
- Drug level studies, not outcome trials.
Drugs that raise amitriptyline levels
The label states that SSRIs, quinidine, cimetidine and other drugs that block the liver enzyme CYP2D6 can raise tricyclic levels, and topiramate raised them sharply in some people.
The numbers and sources
- How common
- How often is not known: the sources give no rate.
- Effect
- CYP2D6 inhibitors include quinidine, cimetidine, many antidepressants, phenothiazines and the heart rhythm drugs propafenone and flecainide, and all SSRIs inhibit it to varying degrees. People who metabolize poorly through this enzyme, about 7% to 10% of Caucasians, can have up to 8-fold higher tricyclic exposure. Cimetidine raised tricyclic levels and side effects. Some people had a large rise in amitriptyline concentration with topiramate [13].
- Population
- Labeling statements drawn from drug interaction studies [13].
- Sources
- [13]
- Caveats
- Changes in blood levels, with no rate of harm given.
Liver injury
The labels warn of liver injury, including cholestatic jaundice with amoxicillin-clavulanate and severe, sometimes fatal hepatotoxicity with azithromycin.
The numbers and sources
- How common
- How often is not known: the sources give no rate.
- Effect
- Hepatitis and cholestatic jaundice have been associated with amoxicillin-clavulanate, usually reversible, with deaths reported mainly in people with serious underlying disease or other medicines. It is contraindicated after past liver problems on the drug [20]. Severe and sometimes fatal liver toxicity has been reported with azithromycin [18].
- Population
- Label statements and reports [20][18].
- Sources
- [20], [18]
- Caveats
- No rate given.
Clopidogrel raising levels of repaglinide
The label states that a clopidogrel metabolite strongly inhibits the enzyme CYP2C8 and can raise blood levels of the diabetes drug repaglinide.
The numbers and sources
- How common
- How often is not known: the sources give no rate.
- Effect
- The clopidogrel label states that its acyl-beta-glucuronide metabolite is a strong inhibitor of CYP2C8, and that clopidogrel can increase exposure to drugs cleared mainly by that enzyme, naming repaglinide [22].
- Population
- Interaction data summarized in labeling [22].
- Sources
- [22]
- Caveats
- The label section read here gives no rate of low blood sugar.
Liver reactions and Reye's syndrome on the aspirin label
The aspirin label lists indigestion, raised liver enzymes, hepatitis and Reye's syndrome among adverse reactions. It gives no rates.
The numbers and sources
- How common
- How often is not known: the sources give no rate.
- Effect
- The prescription aspirin label lists dyspepsia, hepatic enzyme elevation, hepatitis and Reye's syndrome among gastrointestinal adverse reactions. It also states that salicylate can reduce kidney clearance of methotrexate, leading to bone marrow toxicity, and can change blood levels of phenytoin and valproic acid [23].
- Population
- Reports summarized in labeling [23].
- Sources
- [23]
- Caveats
- The label gives no frequencies and covers aspirin at all doses.
Liver injury, atomoxetine
Case reports: the label describes rare severe liver injury after approval, including liver failure and one transplant. None was found in trials of about 6,000 patients.
The numbers and sources
- How common
- How often is not known: the sources give no rate.
- Effect
- No liver injury was detected in trials of about 6,000 patients. After approval, rare cases of clinically significant liver injury were reported, mostly within 120 days of starting, some with liver enzymes over 20 times normal and jaundice. In one patient, injury recurred on restarting, which the label calls evidence the drug likely caused it [9].
- Population
- About 6,000 trial patients and postmarketing reports [9].
- Sources
- [9]
- Caveats
- Reports have no denominator.
Drug interactions
The labels contraindicate both drugs with MAO inhibitors. Fluoxetine and paroxetine raise atomoxetine levels 3 to 8 times, and viloxazine raises duloxetine, ramelteon, tizanidine and theophylline levels.
The numbers and sources
- How common
- How often is not known: the sources give no rate.
- Effect
- Atomoxetine is contraindicated with MAO inhibitors, including linezolid and intravenous methylene blue, or within 14 days, after reports of serious, sometimes fatal reactions. Fluoxetine raised atomoxetine exposure 6 to 8 times, and paroxetine or fluoxetine raised peak levels 3 to 4 times. The label advises caution with blood pressure drugs and pressor drugs, and notes oral albuterol's effects on heart rate and blood pressure can be increased [9]. Viloxazine is contraindicated with MAO inhibitors because of hypertensive crisis risk, and with alosetron, duloxetine, ramelteon, tasimelteon, tizanidine or theophylline, whose levels it raises by blocking CYP1A2 [6].
- Population
- Labeling statements and interaction studies [9][6].
- Sources
- [9], [6]
- Caveats
- Changes in drug levels, not measured harm.
Drugs that change quetiapine or aripiprazole levels
The label states that strong blockers of two liver enzymes raise aripiprazole exposure, including fluoxetine and paroxetine, and that enzyme blockers and inducers change quetiapine exposure.
The numbers and sources
- How common
- How often is not known: the sources give no rate.
- Effect
- The aripiprazole label states that strong CYP3A4 inhibitors such as itraconazole and clarithromycin, and strong CYP2D6 inhibitors such as quinidine, fluoxetine and paroxetine, increased aripiprazole exposure, and that carbamazepine decreased it [10]. The quetiapine label states that exposure is increased by CYP3A4 inhibitors such as ketoconazole, itraconazole, indinavir, ritonavir and nefazodone, and decreased by CYP3A4 inducers. It also states that quetiapine may oppose the effects of levodopa and dopamine agonists [3].
- Population
- Drug interaction studies summarized in labeling [10][3].
- Sources
- [10], [3]
- Caveats
- Changes in drug levels, with no rate of harm.
Raised alprazolam levels with other medicines
The label reports alprazolam blood levels rising about 4 times with ketoconazole and 2.7 times with itraconazole, and lists such strong enzyme blockers as contraindicated.
The numbers and sources
- How common
- How often is not known: the sources give no rate.
- Effect
- Ketoconazole raised plasma alprazolam concentrations 3.98-fold and itraconazole 2.70-fold. The label lists strong CYP3A inhibitors, with ketoconazole, itraconazole and the antibiotic clarithromycin as examples, as contraindicated. It names the antidepressants nefazodone and fluvoxamine, cimetidine and the antibiotic erythromycin as weaker inhibitors that may raise levels, and carbamazepine and phenytoin as drugs that lower them [18].
- Population
- Drug interaction studies described in labeling [18].
- Sources
- [18]
- Caveats
- The block is for alprazolam. Changes in drug levels, with no rate of harm.
Liver enzyme and muscle taurine changes
Limited evidence: ALT rose slightly but stayed within reference ranges, and muscle taurine did not change in humans.
The numbers and sources
- How common
- How often is not known: the sources give no rate.
- Effect
- Beta-alanine raised circulating alanine aminotransferase slightly (effect size 0.274, 95% CrI 0.04 to 0.527), with mean values within clinical reference ranges. Muscle taurine (0.156, 95% CrI -0.38 to 0.72) and histidine did not change in humans. Taurine fell in mice only at 3% or more beta-alanine in drinking water [9].
- Population
- 101 human and 50 animal studies [9].
- Sources
- [9]
- Caveats
- The abstract does not report trial durations. No trials of use over years were located.
Kidney injury and muscle breakdown
Case reports: one person had acute kidney injury, muscle breakdown and liver toxicity after drinking black seed oil.
Raised levels of other drugs
The label states bupropion slows a liver enzyme and can raise levels of many antidepressants, antipsychotics, beta blockers and heart rhythm drugs.
The numbers and sources
- How common
- How often is not known: the sources give no rate.
- Effect
- The label states that bupropion and its metabolites inhibit the enzyme CYP2D6 and can raise exposure to drugs it clears. Its examples are the antidepressants venlafaxine, nortriptyline, imipramine, desipramine, paroxetine, fluoxetine and sertraline, the antipsychotics haloperidol, risperidone and thioridazine, the beta blocker metoprolol and the heart rhythm drugs propafenone and flecainide. It states that tamoxifen, which needs this enzyme to become active, could in theory have a reduced effect. In the other direction, ritonavir, lopinavir, efavirenz, carbamazepine, phenobarbital and phenytoin can lower bupropion levels, and ticlopidine and clopidogrel can raise them [8].
- Population
- Labeling statements drawn from drug interaction studies [8].
- Sources
- [8]
- Caveats
- These are changes in drug levels, not rates of harm. The tamoxifen statement is theoretical.
Raised buspirone levels with other medicines and grapefruit juice
The label reports buspirone blood levels rising 3 to 50 times with certain antibiotics, antifungals, heart drugs, nefazodone and large amounts of grapefruit juice.
The numbers and sources
- How common
- How often is not known: the sources give no rate.
- Effect
- In studies in healthy volunteers, total exposure to buspirone rose 6-fold with the antibiotic erythromycin, 19-fold with itraconazole, 5.5-fold with diltiazem, 3.4-fold with verapamil, up to 50-fold with the antidepressant nefazodone and 9.2-fold with double-strength grapefruit juice. The erythromycin and itraconazole studies also recorded more buspirone side effects. The antibiotic rifampin lowered exposure by 89.6%. Buspirone raised blood levels of haloperidol in one study [4].
- Population
- Healthy volunteers in drug interaction studies, 9 in the diltiazem and verapamil study [4].
- Sources
- [4]
- Caveats
- Small single-dose studies of drug levels, not rates of harm.
Raised liver enzymes with trazodone
Mixed results: one report linked buspirone taken with trazodone to liver enzyme rises of 3 to 6 times in a few patients, and a repeat study did not.
The numbers and sources
- How common
- How often is not known: the sources give no rate.
- Effect
- The label describes one report suggesting that taking trazodone and buspirone together may have caused 3-fold to 6-fold elevations in the liver enzyme ALT in a few patients. A similar study attempting to replicate the finding identified no effect on liver enzymes [4].
- Population
- A few patients in one report, and a follow-up study, both summarized in labeling [4].
- Sources
- [4]
- Caveats
- One unreplicated report. The label gives no numbers.
Gepant levels with other drugs
The labels state strong CYP3A4 inhibitors such as clarithromycin raise gepant levels up to almost tenfold, and inducers such as rifampin cut them about 80%.
The numbers and sources
- How common
- How often is not known: the sources give no rate.
- Effect
- Ketoconazole raised ubrogepant exposure 9.7-fold, and the label contraindicates ubrogepant with strong CYP3A4 inhibitors such as ketoconazole, itraconazole and clarithromycin. Verapamil raised exposure about 3.5-fold, and dose changes apply with moderate inhibitors such as cyclosporine, ciprofloxacin, fluconazole, fluvoxamine and grapefruit juice. Rifampin cut exposure 80%, and the label advises against use with phenytoin, barbiturates, rifampin or St. John's Wort [15]. Itraconazole raised rimegepant exposure 4-fold and rifampin cut it 80% [6]. The atogepant label lowers the dose with strong CYP3A4 inhibitors [16].
- Population
- Drug interaction studies described in labeling [15][6][16].
- Sources
- [15], [6], [16]
- Caveats
- Changes in drug levels, not measured harm.
Kidney and liver failure
Case reports: a woman taking 1,200 to 2,400 µg a day for months developed anemia, liver injury and kidney failure needing dialysis, and recovered.
The numbers and sources
- How common
- How often is not known: the sources give no rate.
- Effect
- A 33-year-old woman taking 1,200 to 2,400 µg a day of chromium picolinate for 4 to 5 months for weight loss developed anemia, liver injury with enzymes 15 to 20 times normal, and kidney failure needing dialysis. She recovered within a year [13].
- Population
- One case report [13].
- Sources
- [13]
- Caveats
- A single case. Plasma chromium was 2 to 3 times normal [13].
Other interactions
The labels state tricyclic antidepressants can blunt clonidine's blood pressure effect, and drugs that block or speed the liver enzyme CYP3A4 change guanfacine levels.
The numbers and sources
- How common
- How often is not known: the sources give no rate.
- Effect
- With tricyclic antidepressants, clonidine's blood pressure lowering may be reduced. With neuroleptics, low blood pressure on standing, dizziness and fatigue may occur or worsen [13]. Strong and moderate CYP3A4 inhibitors such as ketoconazole and fluconazole raise guanfacine exposure, and inducers such as rifampin and efavirenz lower it [7].
- Population
- Labeling statements drawn from interaction studies [13][7].
- Sources
- [13], [7]
- Caveats
- No rates of harm are given.
Air embolism
Case reports: a fatal case of air in the liver's portal vein and systemic air embolism after colon hydrotherapy.
The numbers and sources
- How common
- How often is not known: the sources give no rate.
- Effect
- A published case report describes fatal aeroportia, air in the portal venous system, with systemic air embolism after colon hydrotherapy [15].
- Population
- One patient [15].
- Sources
- [15]
- Caveats
- One case. No abstract was available, and the details here come from the report's title.
Medicines that raise or lower DOAC blood levels
The labels warn that ketoconazole, itraconazole and ritonavir raise blood levels of these drugs, and that rifampin, carbamazepine, phenytoin and St. John's wort lower them.
The numbers and sources
- How common
- How often is not known: the sources give no rate.
- Effect
- The apixaban label states that combined P-glycoprotein and strong CYP3A4 inhibitors such as ketoconazole, itraconazole and ritonavir increase apixaban blood levels, and that combined inducers reduce them [22]. The rivaroxaban label advises against use with ketoconazole and ritonavir, and with inducers such as carbamazepine, phenytoin, the antibiotic rifampin and St. John's wort [23]. The dabigatran label states that rifampin reduces dabigatran exposure, and that dronedarone or systemic ketoconazole with moderate kidney impairment raises exposure to the level seen in severe kidney impairment [20].
- Population
- Interaction studies summarized in labeling [22][23][20].
- Sources
- [22], [23], [20]
- Caveats
- Changes in blood levels, with no rate of bleeding or clots given.
Severe illness reported after kombucha
Case reports: a US public health investigation of two severe illnesses, one fatal, in people drinking kombucha daily, and a report of severe liver injury.
The numbers and sources
- How common
- How often is not known: the sources give no rate.
- Effect
- In 1995, two people in Iowa who had drunk kombucha daily for about 2 months developed unexplained severe illness, and one died. State health officials, the CDC and the FDA investigated [14]. A 2023 report describes severe acute liver injury in a previously healthy woman after drinking kombucha [15].
- Population
- Two people in Iowa [14]. One woman [15].
- Sources
- [14], [15]
- Caveats
- Case reports cannot prove cause or give a rate. The 1995 report describes the link as possible [14].
Hydroxycut liver injury
Case reports: liver injury in 17 people after Hydroxycut, a multi-ingredient product containing hydroxycitric acid. Three needed transplants and one died.
The numbers and sources
- How common
- How often is not known: the sources give no rate.
- Effect
- Hydroxycut was recalled by its manufacturer in May 2009 after reports of liver injury. A case series identified 8 patients treated for liver injury after Hydroxycut, all hospitalized, 3 needing liver transplant. Nine more cases came from FDA MedWatch, including one fatal acute liver failure. Causality was adjudicated as definite in 8, highly likely in 5, probable in 2 and possible in 2 [10]. Hydroxycitric acid was an active ingredient in the Hydroxycut products withdrawn in 2009 [8].
- Population
- 17 patients [10].
- Sources
- [8], [10]
- Caveats
- Hydroxycut was a multi-ingredient product. The series does not isolate which ingredient caused injury [10].
Pancreatitis with suspected liver injury
Case reports: a woman taking only water and green tea catechin supplements for 2 weeks developed severe pancreatitis and suspected toxic hepatitis.
The numbers and sources
- How common
- How often is not known: the sources give no rate.
- Effect
- A 36-year-old woman who took only water and green tea catechin supplements for 2 weeks developed severe acute pancreatitis with highly raised liver and pancreatic enzymes and suspected toxic hepatitis. She recovered and left hospital after 15 days [11].
- Population
- One case report [11].
- Sources
- [11]
- Caveats
- One case. She had eaten no food for the 2 weeks, and the abstract does not name the product or dose [11].
Labeling requirement
USP's monograph label warns not to take it on an empty stomach or with a liver problem, and to stop with symptoms such as dark urine or jaundice.
The numbers and sources
- How common
- How often is not known: the sources give no rate.
- Effect
- Based on its review, USP added a cautionary label to its powdered decaffeinated green tea extract monograph: do not take on an empty stomach, take with food, do not use with a liver problem, and stop and consult a practitioner with symptoms such as abdominal pain, dark urine or jaundice [8].
- Population
- Products that follow the USP monograph [8].
- Sources
- [8]
- Caveats
- USP is a standards body, not a regulator.
Raised levels of other drugs (cimetidine)
The cimetidine label states it slows the liver's clearance of warfarin, phenytoin, theophylline, lidocaine and other drugs, with clinically significant effects reported for warfarin.
The numbers and sources
- How common
- How often is not known: the sources give no rate.
- Effect
- The label lists reduced liver metabolism of warfarin-type anticoagulants, phenytoin, propranolol, nifedipine, chlordiazepoxide, diazepam, certain tricyclic antidepressants, lidocaine, theophylline and metronidazole, raising their blood levels. It reports clinically significant effects with warfarin and adverse effects with phenytoin, lidocaine and theophylline [14].
- Population
- Drug interaction reports and studies summarized in labeling [14].
- Sources
- [14]
- Caveats
- No rate of harm is given.
Liver test changes in trials and traditional use
Limited evidence: liver test abnormalities were more frequent on kava in one trial, and community kava use was linked to higher liver enzymes.
The numbers and sources
- How common
- How often is not known: the sources give no rate.
- Effect
- In the 16-week trial, liver function test abnormalities were more frequent on kava than placebo, though no participant met criteria for herb-induced liver injury [4]. In 98 Indigenous Australians, recent use of aqueous kava, averaging 118 g of powder a week, was linked to higher GGT and alkaline phosphatase, which began returning to baseline after 1 to 2 weeks without kava. ALT and bilirubin were not elevated [7]. In the 6-week trial, liver tests did not differ from placebo [3].
- Population
- 171 trial participants [4]. A cross-sectional study of 98 people in one Arnhem Land community [7]. 58 trial participants [3].
- Sources
- [4], [7], [3]
- Caveats
- The community study was cross-sectional and could not follow individuals over time [7].
Liver enzyme and blood pressure rise
One trial found a moderate rise in the liver enzyme AST and in diastolic blood pressure at 0.6 g a day in people with metabolic syndrome.
The numbers and sources
- How common
- How often is not known: the sources give no rate.
- Effect
- In a 90-day randomized placebo-controlled trial in patients with metabolic syndrome, volunteers using maca at 0.6 g a day showed a moderate rise in AST and diastolic blood pressure. No adverse effects were found in the groups taking silymarin, with or without maca or yacon, and the authors report the effect could be reversed with silymarin [11].
- Population
- Patients with metabolic syndrome [11].
- Sources
- [11]
- Caveats
- One trial. The abstract does not give the size of the rise [11].
Liver enzyme rises at 300 mg a day
One small trial at 300 mg a day was stopped when 3 of 4 patients had liver enzyme rises, which resolved after stopping.
The numbers and sources
- How common
- How often is not known: the sources give no rate.
- Effect
- A trial of 300 mg a day as add-on therapy in primary progressive multiple sclerosis was stopped after 3 of 4 patients on melatonin developed grade 1 to 2 liver enzyme rises, which resolved after stopping. All 3 took other drugs that share liver metabolism with melatonin [19].
- Population
- 8 patients randomized before the trial was halted [19].
- Sources
- [19]
- Caveats
- Very small numbers. The dose is far above sleep doses. A drug interaction is the authors' post hoc explanation.
Gallbladder disease and incontinence
Limited evidence: one trial reported more gallbladder disease and urinary incontinence during estrogen plus progestin, with most risks fading after stopping.
The numbers and sources
- How common
- How often is not known: the sources give no rate.
- Effect
- During estrogen plus progestin treatment, the WHI reported more gallbladder disease and more urinary incontinence. Most risks faded after stopping [5].
- Population
- WHI trial participants aged 50 to 79 [5].
- Sources
- [5]
- Caveats
- The abstract lists these without event counts.
Drugs that change mirtazapine levels
The label states that strong inducers of a liver enzyme lower mirtazapine blood levels, and that strong inhibitors and cimetidine may raise them.
The numbers and sources
- How common
- How often is not known: the sources give no rate.
- Effect
- The label states that strong CYP3A inducers, with phenytoin, carbamazepine and rifampin as examples, decrease the plasma concentration of mirtazapine. Strong CYP3A inhibitors, with itraconazole, ritonavir and nefazodone as examples, and the acid reducer cimetidine may increase it [7].
- Population
- Labeling statements drawn from drug interaction studies [7].
- Sources
- [7]
- Caveats
- Changes in drug levels, with no rate of harm.
Other drug interactions
The label states modafinil can lower cyclosporine levels and raise levels of drugs cleared by CYP2C19, such as diazepam, propranolol, phenytoin and omeprazole.
The numbers and sources
- How common
- How often is not known: the sources give no rate.
- Effect
- By inducing CYP3A4/5, modafinil may lower levels of cyclosporine, midazolam and triazolam. By inhibiting CYP2C19, it may raise levels of phenytoin, diazepam, propranolol, omeprazole and clomipramine, and, in people lacking CYP2D6, tricyclic antidepressants and SSRIs. The label advises more frequent INR checks with warfarin and caution with MAO inhibitors [3].
- Population
- Labeling statement based on drug interaction studies [3].
- Sources
- [3]
- Caveats
- The label describes changes in drug levels and gives no rate of harm.
Cancer
Mixed results: IARC found limited evidence that aspartame causes liver cancer and classed it possibly carcinogenic, while JECFA and the FDA found the evidence not convincing.
The numbers and sources
- How common
- How often is not known: the sources give no rate.
- Effect
- IARC classified aspartame in Group 2B on limited evidence for cancer in humans, specifically hepatocellular carcinoma, with limited evidence in animals and for mechanism. JECFA found the evidence of a link between aspartame and cancer in humans not convincing [2]. The FDA disagrees with IARC's classification and cites shortcomings in the studies relied on [3]. In a French cohort, higher consumers of artificial sweeteners had more cancer than non-consumers (HR 1.13, 95% CI 1.03 to 1.25), including aspartame (HR 1.15) and acesulfame potassium (HR 1.13), and breast and obesity-related cancers [13]. A review of trials and observational studies found no differences for cancer between users and non-users [8].
- Population
- 102,865 French adults followed a median 7.8 years, 3,358 cancers [13]. Evidence reviewed by IARC and JECFA [2].
- Sources
- [2], [3], [13], [8]
- Caveats
- An IARC class reflects the strength of evidence that something can cause cancer, not the risk at a given intake [2]. The cohort authors list selection bias, residual confounding and reverse causation as limits [13].
Label warnings: thyroid tumors, pancreatitis and gallbladder
The label warns of thyroid C-cell tumors seen with related drugs in rodents, with human relevance not determined. Pancreatitis and gallstones were slightly more common.
The numbers and sources
- How common
- How often is not known: the sources give no rate.
- Effect
- The label carries a boxed warning about thyroid C-cell tumors. Such tumors occurred in rodents with GLP-1 drugs that are active in rats and mice, while orforglipron is not active in rats or mice and did not produce tumors in them. The label states that the human relevance has not been determined, and the drug is contraindicated with a personal or family history of medullary thyroid carcinoma or MEN 2. In the weight trials, acute pancreatitis was confirmed in 6 people on orforglipron (0.14 per 100 years of exposure) against 1 on placebo (0.04), gallstones in 1% against 0.7%, acute kidney injury in 0.2% against 0.05%, low blood pressure in 2% against 0.5%, and hair loss in 4 to 5% against 2% [6].
- Population
- Label pool of the two weight trials and nonclinical studies [6].
- Sources
- [6]
- Caveats
- The thyroid warning rests on animal findings for the drug class. Pancreatitis events were few.
Liver injury
The label lists reports of severe liver injury, some ending in transplant or death. A UK study found injury rose before starting, which its authors tie to prior health changes.
The numbers and sources
- How common
- How often is not known: the sources give no rate.
- Effect
- The Xenical label lists postmarketing reports of severe liver injury with hepatocellular necrosis or acute liver failure, some ending in transplant or death [9]. The Alli label notes rare reports of liver injury [8]. In a self-controlled case series, liver injury incidence was raised in the 90 days before starting (IRR 1.50) and the first 30 days after (IRR 2.21), then returned to baseline. The first 90 days on treatment did not differ from the 90 days before (IRR 1.02, 95% CI 0.67 to 1.56) [12].
- Population
- Postmarketing reports [8][9]; 94,695 orlistat users in UK primary care, 988 liver injury cases [12].
- Sources
- [9], [8], [12]
- Caveats
- The authors of [12] state the timing points to the health changes that led to starting treatment rather than a drug effect. Case reports cannot give a rate.
Liver injury
Case reports: two men developed liver problems on reishi products, which resolved after stopping.
The numbers and sources
- How common
- How often is not known: the sources give no rate.
- Effect
- A 47-year-old man developed acute hepatitis with marked transaminase elevation after taking reishi powder with alcohol. Tests normalized within two weeks [8]. A 61-year-old man taking a commercial reishi product developed hypereosinophilia, multiple liver nodules of 1 to 3 cm and slight liver enzyme elevation. All resolved after he stopped it [9].
- Population
- Two case reports [8][9].
- Sources
- [8], [9]
- Caveats
- Case reports cannot give a rate. Alcohol was a co-factor in [8].
Liver injury
Case reports: severe liver injury with jaundice after SARM products, usually reversible after stopping. Products were rarely tested, so hidden steroids cannot be ruled out.
The numbers and sources
- How common
- How often is not known: the sources give no rate.
- Effect
- A review of published reports found 20 reports of adverse events since 2020, most described as drug-induced liver injury, with cholestatic or hepatocellular injury and jaundice [5]. A 32-year-old man developed severe liver injury after using LGD-4033, with cholestatic hepatitis and mild fibrosis on biopsy [6]. A 52-year-old man developed liver injury after products containing RAD-140 and LGD-4033, and his liver enzymes returned to normal about 3 months after stopping [7]. The 2026 review describes cholestatic liver injury as the main serious harm, prolonged but usually reversible [4].
- Population
- Published case reports of people using SARM products [4][5][6][7].
- Sources
- [4], [5], [6], [7]
- Caveats
- Case reports cannot give a rate. In almost all cases the product was not analyzed, so undeclared steroids cannot be ruled out [4]. LGD-4033 did not change liver enzymes over 21 days in the controlled trial [1].
Heart and tendon harms in case reports
Case reports: 15 men had liver injury, myocarditis or tendon rupture after SARMs. Most were athletes taking doses far above those in clinical studies.
The numbers and sources
- How common
- How often is not known: the sources give no rate.
- Effect
- A systematic review of 72 articles from 2003 to 2022 found 13 case reports describing 15 people harmed after using SARMs. All were men, median age 32 (range 19 to 52), more than half were athletes, and all took SARMs by mouth for a mean of 8 weeks. Reported harms included drug-induced liver injury, myocarditis and tendon rupture. Athletes reported doses much higher than those used in clinical studies [8].
- Population
- 15 published cases, plus preclinical and clinical studies summarized in the review [8].
- Sources
- [8]
- Caveats
- Case reports cannot give a rate, and the review rates its own evidence as level 4. The review estimates SARM use among athletes at 1% to 3% [8].
FDA warnings
FDA states life-threatening reactions, including liver injury needing hospitalization, have occurred with SARM products, and lists potential heart attack, stroke, psychosis and infertility.
The numbers and sources
- How common
- How often is not known: the sources give no rate.
- Effect
- FDA states that life-threatening reactions, including liver injuries that required hospitalization, have occurred in people taking products containing SARMs. It lists potential heart attack or stroke, psychosis or hallucinations, sleep disturbance, sexual dysfunction, liver failure, infertility, miscarriage and testicular shrinkage. FDA testing of one product labeled as containing YK-11 found an undeclared anabolic steroid, after an adverse event report of a stroke [10]. Warning letters in 2023 and December 2025 told sellers that SARM products were unapproved new drugs, despite labels saying they were for research only [11][12].
- Population
- US consumers and sellers [10][11][12].
- Sources
- [10], [11], [12]
- Caveats
- Regulator statements of potential harm, without rates.
Alcohol
Mixed results: one study found duloxetine did not add to impairment from alcohol, while three trial patients with liver injury were heavy drinkers.
The numbers and sources
- How common
- How often is not known: the sources give no rate.
- Effect
- When duloxetine and alcohol were given so that their peak levels coincided, duloxetine did not increase the impairment of mental and motor skills caused by alcohol. In the clinical trials database, three people on duloxetine had liver injury with raised ALT and bilirubin, each with substantial alcohol use. The label states duloxetine and alcohol may interact to cause liver injury [18].
- Population
- Volunteers in an interaction study and three cases from the duloxetine trials database [18].
- Sources
- [18]
- Caveats
- Three cases cannot show cause. The label does not give the size of the interaction study.
Changed drug levels with other medicines
The duloxetine label reports a 6-fold rise in its blood levels with fluvoxamine, and a 3-fold rise in desipramine levels when taken with duloxetine.
The numbers and sources
- How common
- How often is not known: the sources give no rate.
- Effect
- With fluvoxamine, a strong inhibitor of the enzyme CYP1A2, duloxetine exposure rose about 6-fold in 14 men. The label names cimetidine and the quinolone antibiotics ciprofloxacin and enoxacin as other inhibitors of that enzyme. Paroxetine raised duloxetine exposure by about 60%. Duloxetine itself is a moderate inhibitor of CYP2D6 and raised exposure to the antidepressant desipramine 3-fold [18].
- Population
- Volunteers in drug interaction studies described in labeling [18].
- Sources
- [18]
- Caveats
- These are changes in drug levels in small studies, not rates of harm. The block covers duloxetine only.
Raised levels of other drugs, and tamoxifen
The labels state that sertraline and paroxetine slow a liver enzyme and can raise levels of drugs it clears. The paroxetine label warns of reduced tamoxifen benefit.
The numbers and sources
- How common
- How often is not known: the sources give no rate.
- Effect
- The sertraline label states that it inhibits the enzyme CYP2D6 and may raise exposure to drugs cleared by it. Its examples include the antidepressants desipramine and venlafaxine, the beta blockers metoprolol and nebivolol, the heart rhythm drugs propafenone and flecainide, atomoxetine and dextromethorphan. It also states sertraline may raise phenytoin levels [24]. The paroxetine label states that some studies found tamoxifen less able to prevent breast cancer relapse and death when taken with paroxetine, and that other studies did not find this [28].
- Population
- Labeling statements drawn from drug interaction studies [24][28].
- Sources
- [24], [28]
- Caveats
- The strength of enzyme inhibition differs between SSRIs. The labels describe changes in drug levels and give no rate of harm.
Autoimmune hepatitis
Limited evidence: statins appeared in 9% of FDA autoimmune hepatitis reports, strongest for fluvastatin. Spontaneous reports cannot give a rate or prove cause.
The numbers and sources
- How common
- How often is not known: the sources give no rate.
- Effect
- In the FDA adverse event database from 2004 to 2024, 337 of 3,581 autoimmune hepatitis reports (9.41%) involved statins. Fluvastatin had the strongest signal (reporting odds ratio 54.85, 95% CI 32.32 to 93.10), and signals were stronger in people 65 and older and in women. The review also summarized 30 published cases [10].
- Population
- Spontaneous reports and 30 published cases [10].
- Sources
- [10]
- Caveats
- Spontaneous reports cannot give a rate or prove cause.
Liver injury
Case reports: one man developed jaundice and raised liver enzymes shortly after starting tongkat ali, improving after he stopped.
The numbers and sources
- How common
- How often is not known: the sources give no rate.
- Effect
- A 47-year-old man developed scleral icterus, raised liver enzymes and jaundice shortly after starting tongkat ali, and improved after stopping it. The authors describe it as the first reported case [9].
- Population
- One case report [9].
- Sources
- [9]
- Caveats
- A single case.
Drinking after alcohol detox
One trial found people given trazodone for sleep after alcohol detox improved less in days abstinent and drank more per drinking day after stopping it.
The numbers and sources
- How common
- How often is not known: the sources give no rate.
- Effect
- Sleep quality improved on trazodone while it was taken and matched placebo once it was stopped. The trazodone group improved less in the proportion of days abstinent during treatment, and had an increase of 4.6 drinks per drinking day (95% CI 2.1 to 7.1) after the medication ended [20].
- Population
- 173 adults with sleep disturbance after alcohol detoxification, 88 on trazodone and 85 on placebo, followed 6 months [20].
- Sources
- [20]
- Caveats
- One trial in a specific group. It reports changes in drinking, not a rate of harm.
Changed drug levels with other medicines
The label states that strong blockers of one liver enzyme raise trazodone levels, inducers lower them, and trazodone can raise digoxin and phenytoin levels.
The numbers and sources
- How common
- How often is not known: the sources give no rate.
- Effect
- The label states that strong CYP3A4 inhibitors raise trazodone exposure, giving itraconazole, ketoconazole, clarithromycin and indinavir as examples, and that this may increase the risk of adverse reactions including cardiac arrhythmias. Strong inducers lower it, with rifampin, carbamazepine, phenytoin and St. John's wort as examples. Trazodone can increase blood levels of digoxin and phenytoin [12].
- Population
- Labeling statements drawn from drug interaction studies [12].
- Sources
- [12]
- Caveats
- Changes in drug levels, with no rate of harm.
Kidney and liver injury
Case reports: three men developed serious liver or kidney injury after taking tribulus. Case reports cannot give a rate or prove cause.
The numbers and sources
- How common
- How often is not known: the sources give no rate.
- Effect
- A young man drinking a tribulus water preparation for 2 days to prevent kidney stones developed seizure, very high aminotransferases and raised creatinine, which normalized after stopping [11]. A young healthy man developed severe hyperbilirubinemia followed by acute kidney failure with bile casts in the kidney tubules [12]. A 46-year-old man taking tribulus daily for 2 months had bilirubin peak at 48 mg/dL and creatinine at 7.1, with a liver biopsy consistent with drug-induced injury. He was treated with plasmapheresis and recovered [13].
- Population
- Three case reports [11][12][13].
- Sources
- [11], [12], [13]
- Caveats
- Case reports cannot give a rate or prove cause.
Medicines that raise or lower warfarin levels
The label lists dozens of drugs that change warfarin's effect through liver enzymes, including amiodarone, fluconazole, metronidazole, some statins, oral contraceptives, rifampin, carbamazepine and phenytoin.
The numbers and sources
- How common
- How often is not known: the sources give no rate.
- Effect
- The label states that inhibitors of CYP2C9, 1A2 or 3A4 can raise the INR and inducers can lower it. Its examples of inhibitors include amiodarone, cotrimoxazole, fluconazole, metronidazole, fluvastatin, atorvastatin, fluoxetine, fluvoxamine, clarithromycin, ciprofloxacin, diltiazem, verapamil, cimetidine and oral contraceptives. Its examples of inducers include rifampin, carbamazepine, phenobarbital, phenytoin, bosentan and cigarette smoking [11]. A systematic overview found consistent reports of interaction for azole antifungals, macrolides, quinolones, omeprazole, lipid-lowering agents, amiodarone and fluorouracil. It rated 84% of all reports as poor quality, most of them single case reports [21].
- Population
- Enzyme interaction lists in labeling [11]; 181 articles on 120 drugs or foods [21].
- Sources
- [11], [21]
- Caveats
- The label states its list is not complete. Most published reports were single cases.
Liver, skin and stomach reactions on the label
The label lists hepatitis and raised liver enzymes, allergic reactions, rash, hair loss, nausea and diarrhea among other adverse reactions. It gives no rates.
The numbers and sources
- How common
- How often is not known: the sources give no rate.
- Effect
- Other adverse reactions on the label include hypersensitivity reactions including hives and anaphylaxis, vasculitis, hepatitis and elevated liver enzymes, nausea, vomiting, diarrhea, abdominal pain, rash, itching, hair loss and calcification of the windpipe [11].
- Population
- Reports summarized in labeling [11].
- Sources
- [11]
- Caveats
- The label gives no frequencies or comparator rates for these reactions.
Drugs that change zolpidem levels
The label states ketoconazole raised zolpidem exposure and effects, while rifampin lowered them and St. John's wort may lower blood levels. It gives no rate.
The numbers and sources
- How common
- How often is not known: the sources give no rate.
- Effect
- The label states that ketoconazole, a potent inhibitor of the liver enzyme CYP3A4, increased exposure to zolpidem and its effects. Rifampin, which induces the same enzyme, significantly reduced exposure and effects, and St. John's wort may decrease blood levels [13]. The eszopiclone label gives a ceiling of 2 mg for people taking potent CYP3A4 inhibitors [7].
- Population
- Drug interaction studies summarized in labeling [13].
- Sources
- [13], [7]
- Caveats
- Changes in drug levels, with no rate of harm. The label says the effect of drugs acting on other liver enzymes is not known.
Liver disease and liver injury
The label warns that drugs acting on GABA receptors have been associated with hepatic encephalopathy in people with liver failure, and lists postmarketing reports of liver injury.
The numbers and sources
- How common
- How often is not known: the sources give no rate.
- Effect
- The label states that people with liver impairment clear zolpidem more slowly, and that drugs affecting GABA receptors have been associated with precipitating hepatic encephalopathy in people with hepatic insufficiency. Postmarketing reports include acute hepatocellular, cholestatic or mixed liver injury with or without jaundice. In the trial database, abnormal liver function and raised ALT were listed as infrequent, between 1 in 100 and 1 in 1,000 [13].
- Population
- Postmarketing reports and 3,660 people in preapproval trials [13].
- Sources
- [13]
- Caveats
- The label states that events in the trial database were not necessarily caused by zolpidem, and gives no placebo rate.
Side effects across studies
Adverse events occurred in about 27% to 31% of users across hundreds of studies, most often digestive, reproductive and skin effects, with 4.7% severe.
The numbers and sources
- Effect
- Across 594 studies of 23,256 TwHF users, 26.7% had adverse events (95% CI 24.8% to 28.8%). Digestive symptoms occurred in 13.3%, reproductive adverse outcomes in 11.7%, skin reactions in 7.8%, blood count events in 6.5% and cardiovascular events in 4.9% [11]. Across 46 studies (2,437 people on Tripterygium wilfordii polyglycoside), 30.75% had adverse events and 4.68% had severe events. The authors list intestinal, reproductive, liver, kidney, blood and skin toxicity, related to dose, course and combined drugs [12].
- Population
- Users in trials and observational studies [11]. Users in RCTs, controlled trials and single-arm trials [12].
- Sources
- [11], [12]
- Caveats
- Rates varied widely by study type [11]. Heterogeneity was high (I² 97%) [12].
Liver injury
Case reports: hepatitis, jaundice and liver failure needing transplant were reported, mostly within 3 months. Experts rated the cases possibly caused by black cohosh.
The numbers and sources
- Effect
- A US Pharmacopeia expert committee reviewed 30 non-duplicate reports of liver damage and rated all as possibly caused by black cohosh, none as probable or certain [8]. The UK regulator received 21 reports of liver reactions by March 2006. Of 39 reports from the UK, Germany, Sweden, the US and the literature, 21 people were admitted to hospital. One UK patient needed a liver transplant. Reported reactions included hepatitis, jaundice and liver failure requiring transplant, mostly within the first 3 months [9].
- Population
- Spontaneous reports and published cases [8][9].
- Sources
- [8], [9]
- Caveats
- Case reports cannot give a rate. Some cases were confounded by other products or medicines [9].
Regulator label warnings
The Australian label warns black cohosh may harm the liver in some individuals. UK, EU and US Pharmacopeia bodies also require liver warnings.
The numbers and sources
- Effect
- In 2006, after assessing 47 cases, Australia required labels to state: "Black cohosh may harm the liver in some individuals. Use under the supervision of a healthcare professional" [9]. UK expert committees concluded the evidence supports an association with liver toxicity and had warnings about rare liver reactions added to all black cohosh product information [9]. The EU herbal committee lists liver problems as a side effect, and its summary tells users to stop and see a doctor if signs of liver problems appear [10]. The US Pharmacopeia decided black cohosh products are to be labeled with a cautionary statement [8].
- Population
- Products sold in Australia, the UK, the EU and the US [8][9][10].
- Sources
- [9], [10], [8]
- Caveats
- The Australian wording is as quoted by the UK regulator in 2006 [9]. The current Australian wording was not checked for this page.
Liver injury
Shown to raise adverse reactions, mainly liver damage, added to blood pressure drugs in diabetic kidney disease. In lupus nephritis, liver problems matched steroids alone.
The numbers and sources
- Effect
- In diabetic nephropathy trials, adverse reactions were more common with tripterygium glycosides added to ACE inhibitors or angiotensin receptor blockers (RR 1.96, 95% CI 1.43 to 2.68), and the authors name liver function damage as the main one [7]. A synthesis of 46 studies lists hepatotoxicity among the organ toxicities [12]. In lupus nephritis trials, liver dysfunction did not differ from glucocorticoids alone [9].
- Population
- 3,537 diabetic nephropathy patients [7]. 2,437 TWP users [12]. Lupus nephritis trial patients [9].
- Sources
- [7], [12], [9]
- Caveats
- The abstracts do not give a separate liver injury rate.
Liver injury, oral
Case reports: four women in Korea developed toxic hepatitis after weeks to months of oral aloe. In three, liver enzymes returned to normal after stopping.
The numbers and sources
- Effect
- Three women aged 55 to 62 developed acute toxic hepatitis after taking oral aloe preparations for months. All met RUCAM criteria, and liver enzymes returned to normal after stopping [19]. A 21-year-old woman developed acute toxic hepatitis after 4 weeks of an aloe vera preparation [20].
- Population
- Case reports from Korea [19][20].
- Sources
- [19], [20]
- Caveats
- Case reports give no rate, and the abstracts do not name the aloe product type.
Rise in liver enzymes in people with obesity
One pilot trial found liver enzymes, mainly AST, rose over 6 weeks on extract in 40 adults with obesity awaiting bariatric surgery.
Liver injury
Case reports: liver injury was reported with Petadolex, including nine severe cases and two transplants. Many patients also took other liver-toxic medicines.
The numbers and sources
- Effect
- Spontaneous reports describe herb-induced liver injury with Petadolex. Most patients had mild changes below 3 times the upper limit of normal, at 50 to 225 mg a day for 4 to 730 days. Nine had severe injury, with onset at an average of 103 days. Two needed liver transplants. Causality was rated likely in 3 cases, possible in 13, unlikely in 8 and unclassifiable in 24. 22 patients reported other liver-toxic medicines [9].
- Population
- Spontaneous case reports [9].
- Sources
- [9]
- Caveats
- Case reports give no rate, and many cases were confounded by other medicines [9].
Pyrrolizidine alkaloids
The EMA herbal committee lists Petasites among plants with pyrrolizidine alkaloids, linked to liver vein blockage and lung hypertension, and sets an adult limit of 1 µg a day.
The numbers and sources
- Effect
- The European Medicines Agency's herbal committee lists Petasites among plant genera with long-known pyrrolizidine alkaloid content. It describes these alkaloids causing hepatic veno-occlusive disease and pulmonary arterial hypertension, and notes animal carcinogenicity findings for Petasites japonicum extracts. It set an acceptable intake of 1.0 µg a day for adults from herbal medicinal products [10].
- Population
- Regulatory review of human poisonings and animal studies [10].
- Sources
- [10]
- Caveats
- The statement concerns the plant genus and the alkaloid class. It did not test specific butterbur products. The carcinogenicity findings are from animals [10].
Liver injury
Case reports: three women developed hepatitis with jaundice after 20 to 60 days of centella, improving after stopping. In one, it returned when she took it again.
The numbers and sources
- Effect
- Three women aged 61, 52 and 49 developed jaundice after taking Centella asiatica for 30, 20 and 60 days. ALT reached 1,193, 1,694 and 324 U/L. Biopsies showed granulomatous hepatitis in two and chronic hepatitis with cirrhotic change in one. All improved after stopping, and the first woman's liver injury returned when she took it again [7].
- Population
- Three case reports [7].
- Sources
- [7]
- Caveats
- Case reports cannot give a rate. The women also received ursodeoxycholic acid [7].
Blood test changes after one ice-cold immersion
One small study found liver and muscle enzymes rose after a single immersion below 4°C in 13 young men, all staying within non-pathological limits.
The numbers and sources
- Effect
- After the immersion, aspartate transaminase, alanine transaminase, lactate dehydrogenase, creatine kinase, its CK-MB form and total bilirubin were higher, and urea, creatinine and uric acid were lower. The authors report all changes stayed within non-pathological limits [27].
- Population
- 13 healthy young men [27].
- Sources
- [27]
- Caveats
- One immersion, no control group, healthy young men only.
Liver injury
Case reports: a 76-year-old woman taking Lyprinol developed acute hepatitis reported as caused by the extract.
Liver injury
Case reports: liver injury in a man taking a horse chestnut supplement and in a man given an injected extract.
The numbers and sources
- Effect
- A 70-year-old man developed idiosyncratic drug-induced liver injury attributed to a horse chestnut dietary supplement [7]. A 37-year-old man developed jaundice 60 days after a 65 mg injection of Venoplant, a horse chestnut extract. His liver biopsy showed marked cholestasis with zonal necrosis, and a lymphocyte stimulation test was positive [8]. A 46-year-old man who ate a horse chestnut seed, mistaking it for an edible chestnut, had slightly raised liver enzymes and amylase and a brief episode of atrial fibrillation, all normal the next day [9].
- Population
- Single case reports [7][8][9].
- Sources
- [7], [8], [9]
- Caveats
- Case reports cannot give a rate. One case followed an injection and one followed eating the raw seed, not taking the extract.
FDA regulatory position
FDA states kratom is not appropriate for use as a dietary supplement, and lists reported liver toxicity, seizures and substance use disorder.
The numbers and sources
- Effect
- FDA states there are no FDA-approved kratom drugs, that kratom is not appropriate for use as a dietary supplement, and that kratom cannot be lawfully added to conventional foods. It lists reported serious adverse events including liver toxicity, seizures and substance use disorder, and states that a well-designed human abuse potential study has not been conducted [3].
- Population
- Kratom products sold in the US [3].
- Sources
- [3]
- Caveats
- A regulatory position, not a study of harm.
Adverse events in controlled trials
Small trials found dizziness, nausea and headache rising with dose, and no serious adverse events. Higher doses raised drug liking ratings.
The numbers and sources
- Effect
- In a randomized, double-blind, placebo-controlled trial of dried kratom leaf powder in healthy adults who had not used kratom, no serious adverse events or deaths were reported. Treatment-emergent adverse events generally rose with dose. Dizziness, nausea and a feeling of relaxation were most common after single doses, and headache, feeling hot, raised alanine aminotransferase (a liver enzyme) and nausea after 15 daily doses [6]. In the same trial, abuse-related adverse events were more frequent at higher doses, drug-effect ratings exceeded placebo only at the highest single dose, and no participant met thresholds for clinically meaningful opioid withdrawal [7]. In a randomized, double-blind trial in recreational drug users with opioid experience, somnolence, vomiting and nausea were the most common adverse events. Kratom constricted the pupils at every dose above the lowest, and the highest dose raised ratings of drug liking and feeling high [8].
- Population
- 116 adults aged 18 to 55, 49 on kratom across four dose levels and 67 on placebo, given single doses and then 15 daily doses [6][7]. 40 recreational polydrug users, 6 on kratom and 2 on placebo in each of five cohorts, given single doses [8].
- Sources
- [6], [7], [8]
- Caveats
- [6] and [7] report the same trial, of one product, in healthy adults screened to exclude certain liver enzyme gene variants [6]. The authors of [8] note that its results may not represent other kratom products. Trials of a few weeks cannot show the dependence, liver injury or deaths reported in long-term users.
Liver injury
Case reports: liver injury with jaundice after kratom, mostly improving after stopping, with four cases progressing to liver transplant.
The numbers and sources
- Effect
- The US Drug-Induced Liver Injury Network identified 11 cases of liver injury attributed to kratom, with a recent increase. All had jaundice, at a median of 14 days after starting. Most were hospitalized, and all recovered [15]. A 2026 systematic review of 31 studies found 32 cases of kratom-associated liver injury, mostly single-patient US case reports in adult men, often with other substances. Injury was mostly cholestatic and in most cases improved after kratom was stopped. Four progressed to liver transplantation [16].
- Population
- Prospective US liver injury registry [15]. Published case reports [16].
- Sources
- [15], [16]
- Caveats
- A registry of referred cases and published case reports, not a rate. Only 7 of the 32 published cases gave complete exposure details [16].
Liver injury
Case reports: rhodiola was named a possible co-cause of liver injury in one patient, with ashwagandha the main suspect.
Mixed results15
Finding fatty liver disease
Shown to miss part of fatty liver disease: in pooled studies 25% of people with nonalcoholic fatty liver disease had a normal ALT.
The numbers and sources
- Compared with
- Nonalcoholic fatty liver disease or steatohepatitis diagnosed by imaging or biopsy
- Effect
- In a meta-analysis of 11 studies, 25% of people with nonalcoholic fatty liver disease had a normal ALT (95% CI 20 to 31), as did 19% of people with nonalcoholic steatohepatitis (95% CI 13 to 27) [16]. How many are missed depends on the limit used. In Italian blood donors, lowering the limit from 40 to 30 U/L for men and from 30 to 19 U/L for women raised sensitivity for hepatitis C viremia from 55% to 76.3% and lowered specificity from 97.4% to 88.5% [5].
- Population
- 4,084 patients in 11 studies [16]. 6,835 blood donors and 209 people with hepatitis C antibodies [5].
- Sources
- [16], [5]
- Caveats
- The pooled studies used different definitions of a normal ALT, and the authors state no threshold for acceptable performance, so the result is given as numbers only [16].
Variation from one test to the next
Shown to vary between tests: 31% of US adults with a raised ALT had a normal result when retested about 17 days later.
The numbers and sources
- Compared with
- A repeat measurement in the same person
- Effect
- Among 1,864 US adults measured twice a mean of 17.5 days apart, 31% of those with an initially elevated ALT had a normal result the second time, and more than 95% of initially normal results stayed normal. The within-person coefficient of variation was 20.4% [17]. In 91 healthy Europeans sampled weekly for 10 weeks under a tightly controlled protocol, within-person biological variation was 9.3% and variation between people was 28.2% [18].
- Population
- 1,864 adults in the third US National Health and Nutrition Examination Survey [17]. 91 healthy adults aged 21 to 69 at 6 European laboratories [18].
- Sources
- [17], [18]
- Caveats
- The US study had only 2 measurements per person and lacked a full liver disease history [17]. The two studies differ in design, which may explain the gap between 20.4% and 9.3%. Neither states a threshold for agreement.
Variation from one test to the next
Shown to vary less than other liver tests: 17% of US adults with a raised result were normal about 17 days later, and within-person variation was 5% to 7%.
The numbers and sources
- Compared with
- A repeat measurement in the same person
- Effect
- Among 1,864 US adults measured twice a mean of 17.5 days apart, 17% of those with an initially elevated alkaline phosphatase had a normal result the second time, against 31% for ALT and 38% for bilirubin. More than 95% of initially normal results stayed normal. The within-person coefficient of variation was 6.7% [12]. In 91 European adults sampled weekly for 10 weeks under a tightly controlled protocol, within-person biological variation was 5.3% and variation between people was 24.9% [13].
- Population
- 1,864 adults in the third US National Health and Nutrition Examination Survey [12]. 91 adults aged 21 to 69 at 6 European laboratories [13].
- Sources
- [12], [13]
- Caveats
- The US study had only 2 measurements per person and lacked a full liver disease history [12]. Neither study states a threshold for agreement. One survey manual notes that hemoglobin from a hemolyzed sample lowers the result [4].
Variation from one test to the next
Shown to vary between tests: 36% of US adults with a raised AST had a normal result when retested about 17 days later.
The numbers and sources
- Compared with
- A repeat measurement in the same person
- Effect
- Among 1,864 US adults measured twice a mean of 17.5 days apart, 36% of those with an initially elevated AST had a normal result the second time, and more than 95% of initially normal results stayed normal. The within-person coefficient of variation was 13.9% [12]. In 91 healthy Europeans sampled weekly for 10 weeks under a tightly controlled protocol, within-person biological variation was 9.5% and variation between people was 20.3% [13].
- Population
- 1,864 adults in the third US National Health and Nutrition Examination Survey [12]. 91 healthy adults aged 21 to 69 at 6 European laboratories [13].
- Sources
- [12], [13]
- Caveats
- The US study had only 2 measurements per person and lacked a full liver disease history [12]. Neither study states a threshold for agreement, so the result is given as numbers only.
Type 2 diabetes
Mixed results: 9 pooled cohort studies found no increase in type 2 diabetes with higher AST, while a genetic study found a 25% higher risk.
The numbers and sources
- Effect
- In a meta-analysis, 9 prospective studies that measured AST gave a relative risk of type 2 diabetes of 1.02 (95% CI 0.99 to 1.04) per standard deviation of log AST, which the authors read as no evidence of increased risk [8]. In a Mendelian randomization study, genetically predicted higher AST was related to more type 2 diabetes (odds ratio 1.25 per standard unit, 95% CI 1.14 to 1.38) [9].
- Population
- 17 studies with 60,359 participants in the meta-analysis, 9 of which measured AST [8]. Up to 64,094 people with type 2 diabetes and 607,012 controls in the genetic study [9].
- Sources
- [8], [9]
- Caveats
- The cohort studies are observational. The genetic study's authors read their findings as support for insulin resistance leading to fatty liver, which in turn raises diabetes risk, so AST may mark that process and not drive it [9].
Variation from one test to the next
Shown to vary less between tests than other liver enzymes: 12% of US adults with a raised GGT had a normal result on retest.
The numbers and sources
- Compared with
- A repeat measurement in the same person
- Effect
- Among 1,864 US adults measured twice a mean of 17.5 days apart, 12% of those with an initially elevated GGT had a normal result the second time, against 36% for AST and 31% for ALT. The within-person coefficient of variation was 13.8% [14]. In 91 healthy Europeans sampled weekly for 10 weeks, within-person biological variation was 8.9% and variation between people was 41.7%, the widest spread between people of the 9 enzymes studied [15].
- Population
- 1,864 adults in the third US National Health and Nutrition Examination Survey [14]. 91 healthy adults aged 21 to 69 at 6 European laboratories [15].
- Sources
- [14], [15]
- Caveats
- The US study had only 2 measurements per person [14]. Neither study states a threshold for agreement, so the result is given as numbers only.
Liver detox
Mixed results: one meta-analysis found silymarin cut liver injury from tuberculosis drugs by two thirds at 4 weeks, while another found no enzyme fall in drug-induced liver injury.
Compared with placebo [12]; control groups the source does not describe [4]
The numbers and sources
- Effect
- Across 5 placebo-controlled trials (1,198 patients starting tuberculosis drugs), silymarin lowered drug-induced liver injury at week 4 (RR 0.33, 95% CI 0.15 to 0.75) and improved liver function (SMD -0.15, 95% CI -0.24 to -0.07) [12]. In a meta-analysis of 55 trials of liver injury from various causes, silymarin had no significant effect on AST or ALT in the drug-induced liver injury subgroup [4].
- Population
- 1,198 adults taking tuberculosis drugs in 5 RCTs [12]. Patients with drug-induced liver injury within 55 trials (3,545 patients) [4].
- Sources
- [12], [4]
- Caveats
- "Detox" is not a measured outcome. These trials tested protection of the liver from medicines. The protective effect was reported at week 4 [12]. The two reviews asked different questions, prevention in [12] and treatment of existing injury in [4].
Differences between laboratories
One statewide survey found the upper limit printed for ALT varied widely between laboratories, mostly because analyzer makers supply different reference intervals.
The numbers and sources
- Compared with
- The upper limits and analyzer results of other laboratories in the same state
- Effect
- In a survey of Indiana laboratories, the upper limit of normal for ALT varied widely for both men and women. Results for the same 5 proficiency samples differed between analyzer brands by an amount that was statistically significant and that the authors judged not clinically significant. Most laboratories took their limit from the analyzer maker, and 17% tested their own volunteers [19]. An international study that standardized its methods found one reference interval could be shared across 4 regions [8].
- Population
- 108 eligible laboratories in Indiana, of which 62% responded [19].
- Sources
- [19], [8]
- Caveats
- One US state, published in 2009. The abstract gives no figures for the spread of limits [19].
As a marker of alcohol use
One multicenter study found AST detected high-risk drinking slightly less well than GGT or carbohydrate deficient transferrin.
The numbers and sources
- Compared with
- Alcohol consumption recorded by structured interview
- Effect
- In 1,863 people recruited in five countries and stratified by alcohol use, age and sex, carbohydrate deficient transferrin and GGT performed comparably on receiver operating characteristic analysis, with AST performing slightly less well. All three detected high-risk drinking better than intermediate-risk drinking [14].
- Population
- 1,863 people from Australia, Brazil, Canada, Finland and Japan, mostly community based [14].
- Sources
- [14]
- Caveats
- The abstract gives no sensitivity or specificity figures for AST. Drinking was self-reported [14].
Variation from one test to the next
One US study found 38% of adults with a raised bilirubin had a normal result about 17 days later, with within-person variation of 23%.
The numbers and sources
- Compared with
- A repeat measurement in the same person
- Effect
- Among 1,864 US adults measured twice a mean of 17.5 days apart, 38% of those with an initially elevated bilirubin had a normal result the second time, the largest share of the 5 liver tests studied. More than 95% of initially normal results stayed normal. The within-person coefficient of variation was 23.4% [16].
- Population
- 1,864 adults in the third US National Health and Nutrition Examination Survey [16].
- Sources
- [16]
- Caveats
- Only 2 measurements per person, and the study lacked a full liver disease history. It states no threshold for agreement, so the result is given as numbers only [16].
Direct bilirubin against the total
A US survey laboratory manual states the raised fraction differs by cause: direct in blocked bile flow, indirect in red cell breakdown, and both in hepatitis.
The numbers and sources
- Compared with
- Direct (conjugated) bilirubin measured on the same sample
- Effect
- The manual states that in obstructive jaundice the rise in total bilirubin comes mainly from direct bilirubin, in hemolytic jaundice mainly from the indirect (unconjugated) fraction, and in hepatitis from both. It adds that a direct result can read slightly above the total when nearly all the bilirubin is in the direct form [2]. In Gilbert syndrome the raised fraction is the unconjugated one [11]. In pregnancy, total and free bilirubin were significantly lower in all three trimesters than in matched controls, and conjugated bilirubin was lower in the second and third [18]. At 11 to 14 weeks of pregnancy one Croatian study found total bilirubin of 2 to 16 µmol/L and an upper limit for conjugated bilirubin of 5 µmol/L [15].
- Population
- 103 pregnant women and 103 matched controls [18]. 299 pregnant women [15].
- Sources
- [2], [11], [18], [15]
- Caveats
- Direct bilirubin is a different quantity from the total, so no direct bilirubin range is placed as a band on this page. The pregnancy figures are printed in µmol/L and are described here, not placed as bands [15].
Ceruloplasmin as a test for Wilson disease
One study found a ceruloplasmin below 20 mg/dL was a true Wilson disease result 48% of the time, against 100% below 14 mg/dL.
The numbers and sources
- Compared with
- ATP7B genotype by DNA sequencing
- Effect
- Ceruloplasmin is a separate blood test, reported in mg/dL or g/L, and its results are not placed against the copper ranges on this page. In a study that confirmed every diagnosis by sequencing the Wilson disease gene, ceruloplasmin below 0.20 g/L (20 mg/dL) had a positive predictive value of 48.3% and a negative predictive value of 98.7%. Below 0.14 g/L (14 mg/dL) the values were 100% and 97.1%, and below 0.10 g/L (10 mg/dL) they were 100% and 91.9%. Forty of 690 people without the disease had ceruloplasmin below 0.20 g/L, and none had a value below 0.14 g/L [15]. A US report cites a normal ceruloplasmin range of 180 to 400 mg/L, which is 18 to 40 mg/dL [3].
- Population
- 57 people with Wilson disease, 71 family members, 25 people with suspected disease and 690 people without it, in Hong Kong [15].
- Sources
- [15], [3]
- Caveats
- One center and one nephelometric assay. The authors suggest that laboratories set their own decision thresholds from local populations [15]. Serum copper itself was not the test evaluated.
3 more
As a marker of alcohol use
Mixed results: one multicenter study found GGT detected high-risk drinking about as well as carbohydrate deficient transferrin, with results also shifted by weight, sex, age and smoking.
The numbers and sources
- Compared with
- Alcohol consumption recorded by structured interview
- Effect
- In 1,863 people recruited in five countries and stratified by alcohol use, age and sex, GGT and carbohydrate deficient transferrin performed comparably on receiver operating characteristic analysis. Both detected high-risk drinking better than intermediate-risk drinking. GGT was influenced by body mass index, sex, age and smoking status [13]. In a second study, GGT in people averaging 1 to 40 g of ethanol a day was higher than in abstainers, and higher again above 40 g a day [6].
- Population
- 1,863 people from Australia, Brazil, Canada, Finland and Japan, mostly community based [13]. 103 heavy drinkers and 92 healthy volunteers, with reference data from 2,485 healthy volunteers [6].
- Sources
- [13], [6]
- Caveats
- The first abstract gives no sensitivity or specificity figures. Drinking was self-reported in both studies [13][6].
Toxin removal
Limited evidence: a review found a handful of small, flawed studies of commercial detox diets and no randomized trial. None measured toxin removal on a juice cleanse.
Compared with no randomized comparison exists, per the review [2]
The numbers and sources
- Effect
- A 2015 critical review found that a handful of clinical studies reported that commercial detox diets enhanced liver detoxification and removed persistent organic pollutants, but described them as hampered by flawed methods and small samples. It found no randomized controlled trial of a commercial detox diet in people, and noted that evidence for detoxifying foods such as coriander and nori came mostly from animals [2].
- Population
- A review of clinical and animal studies of detox diets [2].
- Sources
- [2]
- Caveats
- The studies reviewed were of detox programs in general, not juice-only cleanses [2]. No study measuring toxins in the blood or urine before and after a juice cleanse was located.
Liver protection with alcohol, oral steroids or other drugs
Case reports: one person developed drug-induced liver injury while taking TUDCA for protection. In mice, TUDCA eased acetaminophen liver injury, less than standard treatment.
The numbers and sources
- Effect
- A 2026 case report describes liver injury from fenbendazole in a person also taking TUDCA as an intended liver protectant, the first published report of that combination [8]. In mice given an acetaminophen overdose, TUDCA lowered transaminases, liver cell death and oxidative stress markers, to a lesser extent than N-acetylcysteine, and the two together did better than N-acetylcysteine alone [9].
- Population
- One person [8]. Mice [9].
- Sources
- [8], [9]
- Caveats
- A single case cannot show whether TUDCA changed the course of the injury [8]. The mouse study tested acetaminophen only, with treatment given after the overdose [9]. No study of TUDCA with alcohol or oral steroids in people was located.
Not shown to help15
Raising albumin by infusion
Not shown to lower deaths: two trials that raised albumin to 3.0 g/dL or more by infusion, in sepsis and in cirrhosis, found no benefit.
The numbers and sources
- Effect
- In severe sepsis, albumin infusion to a target of 30 g/L (3.0 g/dL) or more did not change death at 28 days (31.8% against 32.0% with crystalloid alone, RR 1.00, 95% CI 0.87 to 1.14) or at 90 days (41.1% against 43.6%, RR 0.94, 0.85 to 1.05) [11]. In hospitalized patients with decompensated cirrhosis and albumin below 30 g/L, infusions that raised albumin to 30 g/L or more did not change a composite of new infection, kidney dysfunction or death (29.7% against 30.2% with standard care, adjusted OR 0.98, 95% CI 0.71 to 1.33), and more severe or life-threatening serious adverse events occurred in the albumin group [12]. An earlier meta-analysis of 9 controlled trials of correcting low albumin was inconclusive [8].
- Population
- 1,818 patients with severe sepsis in 100 intensive care units [11]. 777 hospitalized patients with decompensated cirrhosis [12]. 535 patients in 9 trials [8].
- Sources
- [11], [12], [8]
- Caveats
- Both large trials were open-label and enrolled people who were seriously ill, so they do not address albumin that is low for other reasons [11][12].
Alcohol-related and viral liver disease
Not shown to lower deaths or liver complications in alcohol-related or viral liver disease. Results on liver enzymes disagreed.
The numbers and sources
- Effect
- Across 18 trials (1,088 patients), milk thistle did not significantly change all-cause mortality (RR 0.78, 95% CI 0.53 to 1.15), complications of liver disease (RR 0.95, 95% CI 0.83 to 1.09) or liver histology. Liver-related mortality was lower across all trials (RR 0.50, 95% CI 0.29 to 0.88) but not in the high-quality trials (RR 0.57, 95% CI 0.28 to 1.19) [1]. In 154 people with hepatitis C, 420 or 700 mg three times a day for 24 weeks did not lower ALT more than placebo. Two people in each group met the primary goal, and viral load and quality of life did not differ [2]. Reviews of liver enzymes in alcohol-related disease disagree. Across 15 RCTs (1,221 patients), silibinin capsules lowered ALT (SMD -1.16) and AST (SMD -1.56) [3]. A meta-analysis of 55 trials (3,545 patients) found no significant effect on AST or ALT in the alcohol-related liver disease subgroup [4].
- Population
- Adults with alcoholic liver disease or hepatitis B or C [1]. 154 adults in the US with hepatitis C not cleared by interferon [2]. Adults with alcohol-related liver disease [3]; adults with liver injury of several causes [4].
- Sources
- [1], [2], [3], [4]
- Caveats
- Only 28.6% of the Cochrane trials reported high methodological quality characteristics [1]. Enzyme changes are surrogate measures, and [3] did not report deaths.
Alcoholic liver disease
Not shown to lower death or liver transplant in adults with alcoholic liver disease. One trial's after-the-fact subgroup, excluding the most advanced cases, found fewer.
The numbers and sources
- Effect
- Across 9 trials (434 patients), SAMe did not significantly change all-cause mortality (RR 0.62, 95% CI 0.30 to 1.26), liver-related mortality (RR 0.68) or death or transplant (RR 0.55, 95% CI 0.27 to 1.09) [7]. In the largest trial, 123 people with alcoholic cirrhosis on 1,200 mg a day for 2 years had death or transplant in 16% against 30% on placebo (p 0.077). Excluding the most advanced cases, rates were 12% against 29% (p 0.025) [8].
- Population
- Adults with alcoholic liver disease [7]. 123 adults with cirrhosis in Spain [8].
- Sources
- [7], [8]
- Caveats
- The pooled analysis rests mostly on one trial [7]. The subgroup result in [8] excluded Child class C patients after the fact.
Heavy exercise and muscle injury
Small studies found ALT rose for at least 7 days after one weightlifting session, and rose in muscle injury with no liver disease.
The numbers and sources
- Compared with
- The same men's results before exercise, and the laboratory's upper reference limit
- Effect
- After a single 1-hour weightlifting session in healthy men unused to lifting, ALT rose significantly and was still raised 7 days later. The highest value was 4.1 µkat/L against an upper reference limit of 0.7 µkat/L, and the median time to peak was 120 hours. At the follow-up visit 10 to 12 days after exercise, mean ALT was higher than mean AST [20]. In 16 patients with muscle necrosis from extreme exercise, polymyositis or seizures and no evidence of liver disease, ALT and AST were both elevated alongside markedly raised creatine kinase [21].
- Population
- 15 healthy men, mean age 24.5, 14 analyzed [20]. 16 patients with muscle injury [21].
- Sources
- [20], [21]
- Caveats
- Both studies are small. The exercise study used men not accustomed to weightlifting, and people who train regularly may respond differently [20]. The muscle injury study also reports the ratio of AST to ALT, a separate quantity not placed on this page [21].
Differences between dye methods and laboratories
Limited evidence: one comparison found the bromcresol green method read 0.35 g/dL higher than electrophoresis, most at low albumin, and laboratories in England differed widely.
The numbers and sources
- Compared with
- Capillary zone electrophoresis on the same samples, and results from other laboratories
- Effect
- On 165 serum samples, the bromcresol green method read a mean 3.54 g/L (0.35 g/dL) higher than capillary zone electrophoresis, the bromcresol purple method agreed with electrophoresis to within 1 g/L, and the gap between the two dye methods was larger at lower albumin. The excess with bromcresol green rose with alpha-1-globulin and C-reactive protein [13]. Across 1,079,193 primary care results in England, the variation between laboratories was large and potentially clinically significant. Mean albumin peaked at around age 20 and then fell with age, faster in women until about 60 [14]. Both US survey manuals above used bromcresol purple, and the Croatian pregnancy study used bromcresol green [2][4][5].
- Population
- 165 serum samples [13]. 1,079,193 albumin reports sent to primary care in England over 3 weeks [14].
- Sources
- [13], [14], [2], [4], [5]
- Caveats
- One method comparison at one laboratory. The authors of the English study conclude that reference intervals for albumin are to be stratified by age and sex and set by each laboratory until methods are standardized [14]. A laboratory report does not usually say which dye was used.
Heavy exercise
One small study found AST rose above the laboratory's upper limit in every man 3 to 5 days after one weightlifting session.
The numbers and sources
- Compared with
- The same men's results before exercise, and the laboratory's upper reference limit
- Effect
- Healthy men used to moderate activity, not including weightlifting, did a single 1-hour weightlifting session. Six had AST above the upper reference limit the next day, and all 14 did on days 3, 4 and 5. The highest value was 16.0 µkat/L against an upper reference limit of 0.7 µkat/L. AST was still significantly higher than before exercise on day 7. Creatine kinase and myoglobin, which come from muscle, rose steeply over the same days, while bilirubin, GGT and alkaline phosphatase were almost unaltered [10].
- Population
- 15 healthy men, mean age 24.5, 14 analyzed [10].
- Sources
- [10]
- Caveats
- One small study in men not accustomed to weightlifting. People who train regularly may respond differently, and the study did not test them [10]. The study also reports the ratio of AST to ALT, a separate quantity not placed on this page.
Muscle injury and disease
Case series: 16 patients with muscle injury and no liver disease had raised AST, falling faster than ALT over the following days.
The numbers and sources
- Compared with
- Creatine kinase, lactate dehydrogenase and clinical review showing muscle, not liver, as the source
- Effect
- In a review of 16 patients with muscle necrosis from extreme exercise, polymyositis or seizures and no evidence of liver disease, AST and ALT were both elevated alongside markedly raised creatine kinase and lactate dehydrogenase. In acute cases AST was more than 3 times ALT. The two came closer within a few days because AST declined faster, and they were comparable in chronic muscle injury [11].
- Population
- 16 patients with muscle injury, in three groups: extreme exercise, polymyositis and seizures [11].
- Sources
- [11]
- Caveats
- A retrospective review of 16 records. The abstract gives no AST values [11].
A raised result without liver disease: Gilbert syndrome
Limited evidence: about 1 in 10 people in a Czech sample had bilirubin above 1 mg/dL with normal liver enzymes, the pattern of the inherited trait Gilbert syndrome.
The numbers and sources
- Compared with
- Liver disease or red blood cell breakdown as the explanation for a raised result
- Effect
- People with Gilbert syndrome have mild, chronic unconjugated hyperbilirubinemia in the absence of liver disease or overt hemolysis, with the liver's capacity to process bilirubin reduced to about 30% of normal. All 10 patients in the study that found the genetic basis carried two copies of a longer promoter in the UGT1A1 gene. The same variant was common in people without the syndrome, at an allele frequency of 40%, and the authors concluded that it is necessary but not sufficient for the syndrome [11]. In a Czech general population sample, 9.6% had bilirubin above 17 µmol/L (1 mg/dL) with normal liver enzymes, 12.6% of men and 6.7% of women [12].
- Population
- 10 unrelated patients with Gilbert syndrome and 55 unaffected subjects [11]. 2,594 people from the Czech general population, compared with 536 elite athletes [12].
- Sources
- [11], [12]
- Caveats
- The genetic study was small. The Czech study defined the syndrome by blood results alone, and its prevalence among elite athletes was 22% [12]. A raised total bilirubin has other causes, and total bilirubin alone does not show which fraction is raised [2].
Fasting
Small studies found fasting raised bilirubin: a 24-hour fast more than doubled it in 37 volunteers, with a rise already present at 12 hours.
The numbers and sources
- Compared with
- The same people when fed
- Effect
- In a crossover study, a 24-hour fast more than doubled total bilirubin from baseline, both in volunteers with normal levels and in those with consistently elevated levels. At 12 hours bilirubin was 17.3 µmol/L when fasting and 14.0 µmol/L when fed [13]. In volunteers screened for early-phase drug trials, fasting men had higher bilirubin than fed men and levels rose with fasting time, while no relation was found in women [14]. Among women in early pregnancy, those who were fasting had higher total bilirubin than those who had eaten (median 6 against 5 µmol/L) [15].
- Population
- 37 volunteers, 19 of them with consistently elevated bilirubin [13]. 105 men and 30 women aged 18 to 48 [14]. 299 pregnant women at 11 to 14 weeks [15].
- Sources
- [13], [14], [15]
- Caveats
- Small studies. The study in trial volunteers was retrospective and included only 30 women [14]. The authors of the pregnancy study judged the difference they found not clinically significant [15]. One US survey manual states that fasting is not required for the test [4].
Light and hemolysis in the sample
Limited evidence: light breaks down bilirubin in a blood sample, and hemolysis shifted results in one experiment on blood from 16 volunteers.
The numbers and sources
- Compared with
- Samples protected from light and free of hemolysis
- Effect
- One survey manual states that bilirubin in a specimen degrades when exposed to light for prolonged periods and instructs the laboratory to protect specimens from light [2]. The other reports an increase of less than 0.24 mg/dL at the lowest grade of visible hemolysis [4]. In blood from 16 volunteers hemolyzed in the laboratory, clinically meaningful changes in total bilirubin appeared at moderate hemolysis, with free hemoglobin above 1 g/L [17].
- Population
- 16 volunteers [17].
- Sources
- [2], [4], [17]
- Caveats
- The manuals describe their own assays and give no figure for how much light exposure lowers a result. The later manual reports no interference from hemolysis on its assay up to a high hemolysis index, so the effect differs between methods [2].
Antioxidant levels after coffee enemas
One small crossover trial in healthy men found coffee enemas did not raise blood glutathione or lower a marker of oxidative damage over 12 days.
Compared with drinking coffee, and baseline levels in the same men [8]
The numbers and sources
- Effect
- Neither single nor repeated coffee enemas raised serum glutathione or lowered serum malondialdehyde over 12 days, and the same held for drinking coffee. Total antioxidant capacity fell from baseline by day 12 after coffee enemas [8]. A companion study found caffeine reached about 3.5 times lower blood levels after a 500 mL coffee enema than after drinking 180 mL of coffee with the same caffeine content [9].
- Population
- 11 healthy Thai men in each crossover study [8][9].
- Sources
- [8], [9]
- Caveats
- Small studies in healthy men, not in people with the illnesses coffee enemas are used for.
Clearing alcohol from the blood
One small study found a liter of IV saline did not change how fast alcohol cleared from the blood.
Compared with the same volunteers given the same dose of alcohol without IV fluid [7]
The numbers and sources
- Effect
- Breath alcohol clearance did not differ with or without a liter of IV saline given rapidly after drinking. Clearance in both conditions was 15 mg/dL per hour (95% CI 12 to 18) [7].
- Population
- Volunteers given a set dose of alcohol on two occasions, number not given in the abstract [7].
- Sources
- [7]
- Caveats
- One small study measuring alcohol clearance, not hangover symptoms.
3 more
Illness and other states that shift it apart from iron stores
Limited evidence: US recommendations state that inflammation, infection, cancer, liver disease, nephrotic syndrome and malnutrition can lower TIBC, and pregnancy and oral contraceptives can raise it.
The numbers and sources
- Compared with
- Body iron stores
- Effect
- The 1998 US recommendations state that TIBC rises when serum iron and stored iron are low and falls when they are high, and that factors other than iron status affect it. They name inflammation, chronic infection, malignancies, liver disease, nephrotic syndrome and malnutrition as lowering the reading, and oral contraceptive use and pregnancy as raising it [3]. The 2017 to 2018 survey manual describes a low TIBC with low serum iron as the pattern of the anemia of chronic disorders, malignant tumors and infections [2]. In 807 people on hemodialysis, a lower TIBC went with lower body mass index and higher inflammatory markers [7].
- Population
- Sources cited by the 1998 recommendations [3]. 807 people on hemodialysis [7].
- Sources
- [3], [2], [7]
- Caveats
- The recommendations and the manual give no sizes for these shifts. The statements are quoted from them, not from the studies behind them [3][2].
Detecting the hemochromatosis genotype
Limited evidence: in one large screening study, saturation cut points of 45% for women and 50% for men missed 33% of people with the hemochromatosis genotype.
The numbers and sources
- Compared with
- Gene testing for two copies of the HFE C282Y variant
- Effect
- Among 209 people found to carry two copies of the variant, 68 (33%) had a transferrin saturation below the cut point at the first screening test, at a second test several months later, or both. Their median ferritin was 170 ng/mL. Fasting samples gave no advantage. The authors conclude that variation within a person limits the test as a first screen for this genotype [15]. A US liver society guideline states that a 45% cutoff has high sensitivity for the genotype and lower specificity than higher cutoffs [7].
- Population
- 101,168 primary care participants screened in the Hemochromatosis and Iron Overload Screening Study, 209 with previously undiagnosed genotype and repeat testing [15].
- Sources
- [15], [7]
- Caveats
- One screening study. A person was counted as missed if either of two tests fell under the cut point, so the share missed by a single test is lower than 33% [15].
Serum retinol against liver stores
Limited evidence: the liver holds serum retinol steady across a wide range of stores, so a result above the cut-off does not show that stores are adequate.
The numbers and sources
- Compared with
- Liver vitamin A stores, estimated by retinol isotope dilution or measured in liver tissue
- Effect
- WHO states that serum retinol reflects liver vitamin A stores only when they are severely depleted, below 0.07 µmol per gram of liver, or extremely high, above 1.05 µmol per gram. Between these extremes it is homeostatically controlled and not always correlated with intake or clinical signs [2]. An expert review in the Biomarkers of Nutrition for Development series reports that Thai children with low rates of inflammation kept serum retinol above 0.7 µmol/L although their mean liver reserves, by isotope dilution, were 0.09 µmol per gram, below the 0.1 µmol per gram the review uses to define deficiency [13]. A European guideline describes serum retinol as highly specific after correction for inflammation and of questionable sensitivity against liver stores from isotope dilution [12].
- Population
- Studies summarized in an expert review and two official documents [13][2][12].
- Sources
- [2], [13], [12]
- Caveats
- Summaries by expert bodies, without a pooled estimate of sensitivity. The same control by the liver is why WHO, and the review, describe serum retinol as a measure for populations [2][13].
Checked, no increase found7
Liver tests
Raised liver tests were about as common on orforglipron as on comparators across phase 3 trials, and other causes accounted for every serious case.
The numbers and sources
- How common
- Uncommon: between 1 in 1,000 and 1 in 100 people.
- Effect
- Across seven phase 3 trials, mean ALT and AST fell on orforglipron, and raised levels occurred at similar rates on orforglipron and comparators. Six people (0.1%) in each group had ALT or AST at 3 or more times the upper limit with bilirubin at 2 or more times, and other causes accounted for every orforglipron case [13].
- Population
- 11,220 participants, 6,920 on orforglipron, for up to 104 weeks [13].
- Sources
- [13]
- Caveats
- A pooled analysis by the manufacturer's investigators. The trials excluded people with markedly raised liver enzymes at screening [13].
Liver enzymes
ALT and AST did not change with CLA against control across 22 trials.
Liver enzymes
Liver enzymes did not change significantly across 16 controlled trials. Alkaline phosphatase fell at doses above 1 g a day.
The numbers and sources
- How common
- How often is not known: the sources give no rate.
- Effect
- A 2026 meta-analysis of 16 controlled trials (1,171 participants) found no significant change in ALT (-1.41 IU/L, 95% CI -3.07 to 0.26), AST (0.03 IU/L, 95% CI -0.12 to 0.18) or alkaline phosphatase (SMD -0.21, 95% CI -0.44 to 0.03). Alkaline phosphatase fell at doses above 1 g a day [10].
- Population
- Participants in 16 controlled clinical trials [10].
- Sources
- [10]
- Caveats
- Liver enzymes are a marker, not a count of liver injury. The authors call for trials in more uniform populations [10].
Adverse effects in trials
Adverse events were about as common on NMN as on placebo in trials of up to 24 weeks, with no rise in liver enzymes.
The numbers and sources
- How common
- How often is not known: the sources give no rate.
- Effect
- Across 10 trials with safety data, NMN did not increase overall, serious or withdrawal-related adverse events, or system-specific events. ALT and AST did not rise [8]. The dose trial reported no safety issues from adverse events, laboratory or clinical measures up to 900 mg a day [1].
- Population
- Trials of 250 to 2,000 mg a day for 14 days to 24 weeks [8]; 80 adults up to 900 mg a day [1].
- Sources
- [8], [1]
- Caveats
- All trials were 24 weeks or shorter. Rare or long-term effects cannot be seen in trials this size.
Drug metabolism interactions
Not shown to change most drug-clearing enzymes in healthy volunteers. It raised peak alprazolam levels slightly in one small study.
The numbers and sources
- How common
- How often is not known: the sources give no rate.
- Effect
- In 12 healthy volunteers, 1,000 mg nightly for 14 days raised peak alprazolam concentration from 25 to 31 ng/ml (p < 0.05), with no significant change in alprazolam AUC or half-life and no effect on CYP2D6 [7]. In a 28-day study of 12 volunteers, valerian did not change CYP1A2, CYP2D6, CYP2E1 or CYP3A4/5 activity [8].
- Population
- Healthy volunteers [7][8].
- Sources
- [7], [8]
- Caveats
- Probe-drug studies in small healthy groups. Effects with sedatives taken together were not measured.
Use in adolescents
Limited evidence: no consistent short-term changes in kidney, liver or heart markers in adolescents.
The numbers and sources
- How common
- How often is not known: the sources give no rate.
- Effect
- A systematic review of 5 studies, RCTs and cohorts, found no consistent short-term signals in kidney function, liver enzymes or cardiometabolic markers, and no serious adverse events attributed to creatine [21].
- Population
- Adolescent athletes and physically active youth, including some with medical conditions, in studies published 2015 to 2025 [21].
- Sources
- [21]
- Caveats
- Five small studies with short follow-up, found through a PubMed-only search [21].
Safety in controlled trials
Small trials in postmenopausal women found no significant adverse symptoms or lab changes. Osteoporosis trials reported mild gut or skin reactions.
The numbers and sources
- How common
- How often is not known: the sources give no rate.
- Effect
- In 58 postmenopausal women over 6 weeks, Epimedium prenylflavonoids caused no significant adverse symptoms and no changes in liver, blood or kidney tests [3]. Over 24 months, the icariin blend showed no detectable endometrial hyperplasia [1]. In 10 osteoporosis RCTs, adverse reactions were described as low, mainly mild gastrointestinal reactions or skin allergies [4].
- Population
- Postmenopausal women [1][3]; patients with osteoporosis [4].
- Sources
- [1], [3], [4]
- Caveats
- Trials in women for bone outcomes. No safety data in men using it for sexual function were located.
Not studied in people7
Gallstones
Not studied in people. Phyllanthus amarus, a related species, is used in folk medicine for gallbladder stones.
No study in people. This rests on documented traditional use.
The numbers and sources
- Effect
- A 2025 lab study describes Phyllanthus amarus as used in folk medicine to treat gallbladder stones, kidney stones and chronic liver disease. The same study found that the extract and its lignans activated drug-sensing receptors and changed the activity of the liver enzymes CYP3A4 and CYP1A2 in human liver cells [6].
- Population
- A statement of traditional use and lab tests in human liver and gut cells [6].
- Sources
- [6]
- Caveats
- The source names P. amarus, not P. niruri, and records use, not an effect. No study of gallstones in people or animals given either species was located.
Coffee enemas for detox and cancer
Not studied in people for effectiveness. A systematic review found no study reporting any benefit, and coffee enemas are used by naturopathic practitioners for detox and cancer.
No study in people. This rests on documented traditional use, what users report.
The numbers and sources
- Effect
- A systematic review searched Western and East Asian databases for any study of self-administered coffee enema effectiveness and found none. People use them for self-care and detoxification [10]. A case report notes their use by naturopathic practitioners for many diseases, including cancer [11].
- Population
- No study population. A systematic review [10] and a case report [11].
- Sources
- [10], [11]
- Caveats
- Use is not evidence of effect.
Liver health
Not studied in people. Dandelion has long been used for liver disorders, and in mice on a high-fat diet a leaf extract cut fat build-up in the liver.
No study in people. This rests on animal studies, documented traditional use.
The numbers and sources
- Effect
- A 2025 review notes that dandelion has historically been used for liver disorders, and found only preclinical studies of its liver effects [4]. In mice fed a high-fat diet, dandelion leaf extract reduced fat in the liver, liver weight, blood triglycerides and cholesterol compared with the high-fat diet alone [5].
- Population
- A narrative review of animal studies [4]. C57BL/6 mice [5].
- Sources
- [4], [5]
- Caveats
- The review states that clinical trials are needed to confirm any effect in people [4]. Fatty liver in diet-fed mice is a model, not a human outcome.
Liver detox
Not studied in people. In animals given liver poisons such as alcohol, carbon tetrachloride or paracetamol, dandelion extracts reduced liver damage.
No study in people. This rests on animal studies.
The numbers and sources
- Effect
- A 2025 review of animal studies reports that dandelion extracts protected against liver damage caused by alcohol, carbon tetrachloride and paracetamol, and points to taraxasterol, which acts on inflammation and oxidative stress pathways [4].
- Population
- Animal models of toxic liver injury [4].
- Sources
- [4]
- Caveats
- Protecting an animal from a large dose of a poison is not the same as clearing everyday toxins in people. No human study was located [4].
Liver health
Not studied in people. In 54 animal studies of liver injury, Schisandra compounds lowered the liver enzymes ALT and AST and markers of inflammation.
No study in people. This rests on animal studies, documented traditional use.
The numbers and sources
- Effect
- A 2025 meta-analysis of 54 animal studies of liver injury found Schisandra chinensis compounds lowered ALT (SMD -4.74, 95% CI -5.42 to -4.06), AST (SMD -5.10, 95% CI -5.84 to -4.37) and alkaline phosphatase, lowered TNF-α, IL-6 and IL-1β, and raised the antioxidants SOD and glutathione. The review notes more than 2,000 years of use in traditional Chinese medicine [6].
- Population
- Animal models of liver injury in 54 studies [6].
- Sources
- [6]
- Caveats
- Animal studies with very high heterogeneity (I² above 90%), and the effects varied with species, compound, injury model and dose [6]. No trial of Schisandra alone for liver health in people was located.
Fatty liver
Not studied in people. In mouse models of fatty liver, TUDCA reduced weight gain and liver fat and changed bile acid handling and gut bacteria.
No study in people. This rests on animal studies, how it acts in the body.
The numbers and sources
- Effect
- In mouse models of fatty liver disease, TUDCA reduced obesity and liver fat buildup, improved gut barrier function, raised liver bile acids and lowered fatty acid uptake through FXR and FATP5 [6]. A 2026 review describes TUDCA as easing endoplasmic reticulum stress, inflammation, oxidative stress and liver cell death, and states that high-quality clinical evidence is still limited [7].
- Population
- Mice with fatty liver disease [6]. Preclinical studies, reviewed [7].
- Sources
- [6], [7]
- Caveats
- Animal data. The one human metabolic trial on this page measured insulin sensitivity over 4 weeks, not liver fat [3]. No trial of TUDCA for fatty liver in people was located.
Glucuronidation (UGT) inhibition in lab studies
Lab studies only: hyperforin and hypericin blocked 11 drug-clearing enzymes. Modeled gut risk fell sharply once poor dissolution was included.
No study in people. This rests on lab studies.
The numbers and sources
- How common
- How often is not known: the sources give no rate.
- Effect
- In vitro, hyperforin and hypericin inhibited 11 UGT enzymes, which clear many drugs. Hyperforin was the most potent (UGT1A1 Ki 1.46 μM). A model using FDA thresholds first flagged a high gut interaction risk (R1,gut 69.49 against a threshold of 11), which fell to about 2.0 to 2.9 once hyperforin's poor dissolution was included [15].
- Population
- In vitro, recombinant human enzymes and computer modeling [15].
- Sources
- [15]
- Caveats
- No people were studied. The authors describe the interaction as potential, with a size that depends on the product [15].