Abstract
Antipsychotics (APs) and antidepressants (ADs) are the primary therapies for psychiatric disorders. However, drug-induced liver injury and liver biochemical abnormalities during treatment with APs and/or ADs remain significant safety concerns in clinical practice. Identifying factors associated with liver function abnormalities at an early stage may help support appropriate monitoring and preventive measures. A retrospective unmatched case-control study was conducted, including 708 discharged patients (354 cases and 354 controls) from Xuzhou Oriental Hospital Affiliated to Xuzhou Medical University between January and December 2025. Univariate and multivariable logistic regression analyses were performed to identify factors associated with liver function abnormalities. Olanzapine was the most frequently recorded AP/AD exposure among patients with liver function abnormalities (57.34%). Multivariable logistic regression showed that male sex (odds ratio [OR] = 1.51, 95% confidence interval [CI], 1.07–2.11), age ≥60 years (OR = 5.02, 95% CI, 2.48–10.16), obesity (OR = 3.60, 95% CI, 1.24–10.43), dysglycemia (OR = 2.59, 95% CI, 1.37–4.91), dyslipidemia (OR = 2.18, 95% CI, 1.55–3.05), length of stay (OR = 1.014, 95% CI, 1.003–1.025), and medication history ≤1 year (OR = 2.98, 95% CI, 1.87–4.74) were independently associated with liver function abnormalities. Liver function abnormalities occurring in psychiatric patients after AP and/or AD treatment mainly consisted of liver biochemical abnormalities. Olanzapine was the most frequently recorded AP/AD exposure among affected patients, but this descriptive finding should not be interpreted as indicating a comparative hepatotoxicity risk. These findings may help clinical pharmacists identify patients who require closer liver function monitoring, but multicenter validation is needed.
Keywords: antidepressants, antipsychotics, case-control study, liver function abnormalities, risk factors
1. Introduction
Liver function abnormality, a type of hepatic adverse drug reaction, is a key drug safety issue in clinical practice. In the past 2 decades, the widespread use of antipsychotics (APs) and antidepressants (ADs) in the treatment of schizophrenia and depression has garnered great interest and concern among researchers and clinicians regarding the risk of liver toxicity.[1] The liver is the primary organ involved in the metabolism of APs and ADs, which is mediated mainly by cytochrome P450 (CYP450) enzymes, including CYP450 family 1 subfamily A member 2 and CYP450 family 2 subfamily D member 6, and uridine diphosphate-glucuronosyltransferase.[2] Differences in individual metabolic capacity and preexisting liver damage may interfere with these steps, thus creating an imbalance favoring the accumulation of hepatotoxic metabolites. These are the basic mechanisms underlying the widespread occurrence of these unwanted events. Drug-induced liver injury (DILI) presents a broad clinical spectrum of varying severity. In most cases, DILI is characterized by mild to moderate increases in serum alanine aminotransferase (ALT) or alkaline phosphatase (ALP) levels, with or without jaundice, and is self-limiting within 6 months after drug withdrawal. In other cases, rapidly progressive acute liver failure or progression to chronic liver injury may occur.[3] Specifically, in the “Chinese Guidelines for Diagnosis and Treatment of Drug induced Liver Injury (2023 Edition),”[4] “drug-induced liver biochemical abnormalities” (short for “liver biochemical abnormalities”) refers to persistent increases in liver biochemical markers, such as ALT and γ-glutamyl transferase, that do not meet the diagnostic criteria for DILI. From a medical point of view, when a patient’s liver function abnormalities meet the formal criteria for DILI, the patient may already have sustained severe liver damage and missed the optimal window for treatment. Therefore, this study focused on liver function abnormalities after AP and/or AD exposure, including early liver biochemical abnormalities and, where applicable, events meeting the formal DILI criteria; the primary outcome was not intended to represent confirmed DILI in all cases. The retrospective analysis of risk factors for liver function abnormalities in patients taking APs and ADs aims to support early identification, monitoring, and prevention of clinically relevant liver biochemical abnormalities related to these psychiatric drugs.
2. Materials and methods
2.1. Study population
This retrospective unmatched case-control study screened discharged patients from Xuzhou Oriental Hospital Affiliated to Xuzhou Medical University, a Grade III Class A psychiatric hospital, between January 1 and December 31, 2025. A total of 354 cases and 354 controls were enrolled from the hospital information system.
The case group inclusion criteria were as follows: patients were diagnosed with a psychiatric disorder (such as schizophrenia, depression, or bipolar disorder) based on the International Classification of Diseases, 10th Revision criteria; had no previous history of hepatitis or liver surgery; normal liver function (ALT, aspartate aminotransferase, ALP, gamma-glutamyl transferase, and total bilirubin) was observed at admission; and after receiving APs and/or ADs for at least 14 days, liver function tests showed at least 1 parameter exceeding 1.5 × upper limit of normal (ULN). The threshold of >1.5 × ULN was used as a sensitive operational cutoff to identify early post-exposure liver biochemical abnormalities of potential clinical relevance and was not intended to define confirmed DILI. The Common Terminology Criteria for Adverse Events v5.0 was cited only as a general ULN-based adverse event grading reference, not as a psychiatric-specific diagnostic criterion.[5] Formal DILI generally requires higher biochemical thresholds and exclusion of alternative causes.[4,6] Cases were further classified into 2 subgroups: liver biochemical abnormalities without confirmed DILI and confirmed DILI, according to whether formal DILI criteria were met after clinical assessment and Roussel Uclaf Causality Assessment Method (RUCAM) evaluation.
The control group inclusion criteria were as follows: 354 patients discharged during the same study period were randomly selected at a 1:1 ratio from the same source population without individual matching. They were diagnosed with a psychiatric disorder (such as schizophrenia, depression, or bipolar disorder) based on the International Classification of Diseases, 10th Revision criteria; had no previous history of hepatitis or liver surgery; normal liver function (ALT, aspartate aminotransferase, ALP, gamma-glutamyl transferase, and total bilirubin) was observed at admission; and they received APs and/or ADs for at least 14 days during hospitalization, and all available liver function tests remained within the normal range until discharge.
Both cohorts shared the following exclusion criteria: liver injury or liver disease from nondrug-related causes, including viral hepatitis, autoimmune liver disease, alcoholic liver disease, fatty liver disease, or prior liver surgery; documented smoking or alcohol use in the admission medical records or personal history; and concurrent use of medications known to be hepatotoxic during admission.
Because this was a retrospective unmatched case-control study based on hospital records, no formal a priori sample size calculation was performed. The study aimed to identify factors associated with liver function abnormalities after AP and/or AD exposure rather than estimate incidence rates or time-to-event risks. Controls were randomly selected from the same source population and calendar period at a 1:1 ratio. Individual matching was not performed because age, sex, diagnosis, and medication exposure were variables of interest, and propensity score methods were not used because no single treatment exposure was predefined. With 354 outcome events and prespecified clinically relevant candidate variables, the available number of events was considered adequate for exploratory multivariable logistic regression.
2.2. Drug-relatedness assessment
In this retrospective study, RUCAM was used to assess the drug-relatedness of liver function abnormalities in a standardized manner.[6] Two trained raters independently reviewed medical records and assessed the RUCAM score, which includes 7 factors such as the latency period, disease progression, risk factors, and others. Disagreements were resolved by discussion and consensus, with the consultation of a senior reviewer when necessary. Drug-relatedness categories were classified according to the total score: >8 points indicated “highly probable,” 6 to 8 points indicated “probable,” and 3 to 5 points indicated “possible.”
Cases with RUCAM scores of 3 or higher were included to ensure at least possible drug-relatedness.
2.3. Data collection and statistical analysis
We collected baseline characteristics and medication details. Baseline data included sex, age, body mass index (BMI), dysglycemia (defined as fasting plasma glucose ≥6.1 mmol/L), dyslipidemia (defined as having at least 1 abnormal lipid parameter), hypertension, length of stay, and diagnosis. Medication details encompassed the generic names of APs and ADs, medication types, medication history, and the number of AP/AD agents prescribed during hospitalization. The number of AP/AD agents prescribed was used as a proxy indicator of psychiatric medication polypharmacy and reflected the medication exposure burden rather than causal attribution to each individual drug. Patients with incomplete baseline liver function tests or medication records were excluded, and variables included in the final regression model had no missing values.
Statistical analyses were performed using SPSS version 27.0 (IBM Corp., Armonk). Non-normally distributed quantitative data were expressed as median (P25, P75), with intergroup comparisons conducted using the Mann–Whitney U test. Count data were shown as n (%), and intergroup comparisons were made using the chi-square test. Because this was an unmatched case-control study, unconditional multivariable logistic regression was used. Candidate variables for entry into the binary logistic regression model were selected based on clinical relevance, previous literature, and univariate analysis results, rather than univariate statistical significance alone. Available clinically relevant variables, including sex, age, BMI, dysglycemia, dyslipidemia, hypertension, length of stay, diagnosis, medication history, and the number of AP/AD agents prescribed, were considered during model development. This approach aimed to minimize the omission of potentially important confounders and to identify factors associated with liver function abnormalities. Multicollinearity among candidate independent variables was assessed using variance inflation factors. All variance inflation factor values were <5, indicating no serious multicollinearity among the variables included in the multivariable model. Model calibration was evaluated using the Hosmer–Lemeshow goodness-of-fit test, discrimination was assessed using the receiver operating characteristic curve and area under the curve, and Nagelkerke R2 was reported as an additional measure of apparent model performance. Sensitivity analyses were performed by treating age and BMI as continuous variables. To address potential outcome heterogeneity, an additional sensitivity analysis was performed by restricting the case group to patients with liver biochemical abnormalities without confirmed DILI and comparing them with controls. Confirmed DILI cases were described separately, but a separate multivariable analysis was not performed because of the limited number of events. Exploratory interaction analyses were performed for clinically plausible interaction terms with adequate cell counts, including medication history ≤1 year × dyslipidemia and male sex × dyslipidemia. Influential observations were assessed using standardized residuals, Cook’s distance, leverage values, and difference in beta coefficient. Statistical significance was defined as P < .05.
3. Results
3.1. Case characteristics
In the case group, 155 patients (43.79%) were male, and 199 (56.21%) were female, resulting in a male-to-female ratio of 1:1.28. The median age (P25, P75) was 36 (20, 57) years. Of these patients, 238 (67.24%) had a medication history of 1 year or less. In addition, 46 patients (12.99%) had dysglycemia, 165 (46.61%) had dyslipidemia, and 37 (10.45%) had hypertension. In the case group, the most frequently recorded AP/AD exposures among patients with liver function abnormalities, with usage rates exceeding 10%, were olanzapine in 203 patients (57.34%), quetiapine in 75 patients (21.19%), sertraline in 57 patients (16.10%), duloxetine in 51 patients (14.41%), paroxetine in 45 patients (12.71%), and risperidone in 38 patients (10.73%).
3.2. Severity of liver injury and outcomes in patients
In the case group, 306 patients (86.44%) had liver biochemical abnormalities without confirmed DILI, and 48 patients (13.56%) met the criteria for confirmed DILI, including 47 with mild hepatic injury, one with moderate hepatic injury, and no cases of severe hepatic injury. Among the 48 patients with confirmed DILI, 33 (68.75%) were treated with olanzapine. After treatment with hepatoprotective and liver enzyme-lowering agents, liver function returned to normal in 213 patients (60.17%), improved in 106 patients (29.94%), did not improve in 11 patients (3.11%), and was unknown or unassessed in 24 patients (6.78%).
3.3. Univariate analysis of factors associated with liver function abnormalities
There were no statistically significant differences between the case group and the control group in terms of diagnosis (P = .076), presence of hypertension (P = .241), and the number of AP/AD agents prescribed (P = .055). However, statistically significant differences were observed between the case group and the control group in terms of sex, age, BMI, dysglycemia, dyslipidemia, length of stay, and medication history (P < .05), as shown in Table 1.
Table 1.
Baseline characteristics and medication exposure of patients with and without liver function abnormalities.
| Characteristics | Control group (n = 354) | Case group (n = 354) | χ2/Z | P value |
|---|---|---|---|---|
| Sex | 7.716 | .005 | ||
| Male | 119 (33.6) | 155 (43.8) | ||
| Female | 235 (66.4) | 199 (56.2) | ||
| Age (yr) | 39.838 | <.001 | ||
| <18 | 77 (21.8) | 73 (20.6) | ||
| 18–<30 | 80 (22.6) | 59 (16.7) | ||
| 30–<45 | 103 (29.1) | 78 (22.0) | ||
| 45–<60 | 78 (22.0) | 75 (21.2) | ||
| ≥60 | 16 (4.5) | 69 (19.5) | ||
| BMI (kg/m2) | 21.396 | <.001 | ||
| Underweight, BMI < 18.5 | 63 (17.8) | 61 (17.2) | ||
| Normal, 18.5–<24 | 232 (65.5) | 189 (53.4) | ||
| Overweight, 24–<28 | 54 (15.3) | 81 (22.9) | ||
| Obese, BMI ≥ 28 | 5 (1.4) | 23 (6.5) | ||
| Dysglycemia | 14.653 | <.001 | ||
| Yes | 17 (4.8) | 46 (13.0) | ||
| No | 337 (95.2) | 308 (87.0) | ||
| Dyslipidemia | 25.481 | <.001 | ||
| Yes | 100 (28.3) | 165 (46.6) | ||
| No | 254 (71.7) | 189 (53.4) | ||
| Hypertension | 1.372 | .241 | ||
| Yes | 28 (7.9) | 37 (10.5) | ||
| No | 326 (92.1) | 317 (89.5) | ||
| Length of stay (d) | 28.00 (21.0–39.3) | 29.00 (23.0–42.0) | −2.350 | .019 |
| Medication history (yr) | 14.998 | <.001 | ||
| >5 yr | 95 (26.8) | 70 (19.8) | ||
| >1–≤5 yr | 71 (20.1) | 46 (13.0) | ||
| ≤1 yr | 188 (53.1) | 238 (67.2) | ||
| Number of AP/AD agents prescribed | 2.00 (1.0–2.0) | 2.00 (1.0–2.0) | −1.917 | .055 |
| Cumulative olanzapine dose during hospitalization, 100-mg units | 0.00 (0.00–1.11) | 0.68 (0.00–2.30) | −5.480 | <.001 |
| Diagnosis | 8.471 | .076 | ||
| Schizophrenia | 123 (34.8) | 99 (28.0) | ||
| Depressive disorder | 138 (39.0) | 148 (41.8) | ||
| Bipolar disorder | 22 (6.2) | 13 (3.7) | ||
| Anxiety disorder | 48 (13.6) | 63 (17.8) | ||
| Others | 23 (6.5) | 31 (8.8) | ||
Data are presented as n (%) or median (interquartile range), as appropriate. Test statistics are χ2 values for categorical variables and Z values for continuous variables. Statistical significance was set at P < .05. Number of AP/AD agents prescribed refers to the number of different antipsychotic and/or antidepressant agents used during hospitalization and was used as a proxy indicator of psychiatric medication polypharmacy. This variable reflects medication exposure burden and does not indicate that each drug was implicated in liver function abnormalities. Cumulative olanzapine dose was analyzed in 100-mg units; patients not receiving olanzapine were assigned a cumulative dose of 0.
ADs = antidepressants, APs = antipsychotics, BMI = body mass index.
3.4. Multivariable analysis of factors associated with liver function abnormalities
Candidate variables were selected based on clinical relevance, previous literature, and univariate analysis results, rather than univariate statistical significance alone, and were entered into a multivariable logistic regression model. The analysis indicated that male sex, age ≥60 years, obesity, dysglycemia, dyslipidemia, length of stay, and medication history ≤1 year were independently associated with liver function abnormalities in psychiatric patients receiving APs and/or ADs, as shown in Table 2.
Table 2.
Multivariable logistic regression analysis of factors associated with liver function abnormalities.
| Characteristics | Variable | β | SE | Wald | P value | OR (95% CI) |
|---|---|---|---|---|---|---|
| Sex | Male vs female | 0.409 | 0.173 | 5.604 | .018 | 1.505 (1.073–2.111) |
| Age, yr | ≥60 vs 18–<30 | 1.613 | 0.360 | 20.109 | <.001 | 5.018 (2.480–10.157) |
| BMI, kg/m2 | Obese vs normal BMI | 1.281 | 0.543 | 5.573 | .018 | 3.600 (1.243–10.425) |
| Dysglycemia | Yes vs no | 0.951 | 0.326 | 8.502 | .004 | 2.588 (1.366–4.905) |
| Dyslipidemia | Yes vs no | 0.777 | 0.172 | 20.336 | <.001 | 2.175 (1.552–3.050) |
| Length of stay, d | Per 1-d increase | 0.013 | 0.005 | 6.002 | .014 | 1.014 (1.003–1.025) |
| Medication history, yr | ≤1 yr vs >5 yr | 1.091 | 0.237 | 21.131 | <.001 | 2.977 (1.870–4.741) |
Reference categories were female sex, age 18 to <30 years, normal BMI (18.5–<24 kg/m2), no dysglycemia, no dyslipidemia, and medication history >5 years. Length of stay was entered as a continuous variable. Only statistically significant variables or categories are shown; nonsignificant categories were retained in the model but are not displayed.
Model performance: Nagelkerke R2 = 0.198; Hosmer–Lemeshow goodness-of-fit test, χ2 = 5.926, df = 8, P = .656; area under the receiver operating characteristic curve = 0.717 (95% CI, 0.679–0.754; P < .001).
BMI = body mass index, CI = confidence interval, df = degrees of freedom, OR = odds ratio, SE = standard error.
The multivariable logistic regression model showed modest discrimination, with an area under the curve of 0.717 (95% confidence interval [CI], 0.679–0.754; P < .001), acceptable calibration according to the Hosmer–Lemeshow goodness-of-fit test (χ2 = 5.926, df = 8, P = .656), and a Nagelkerke R2 of 0.198.
To assess the robustness of our findings, we repeated the multivariable logistic regression analysis treating age and BMI as continuous variables. The results remained consistent: age (odds ratio [OR] = 1.017, 95% CI, 1.006–1.028, P = .002), BMI (OR = 1.053, 95% CI, 1.003–1.106, P = .036), male sex (OR = 1.533, 95% CI, 1.103–2.132; P = .011), dysglycemia (OR = 2.643, 95% CI, 1.417–4.931; P = .002), dyslipidemia (OR = 2.257, 95% CI, 1.620–3.144; P < .001), length of stay (OR = 1.013, 95% CI, 1.002–1.024; P = .016), and medication history ≤1 year (OR = 3.471, 95% CI, 2.214–5.444; P < .001) remained significant. The Hosmer–Lemeshow goodness-of-fit test indicated acceptable model calibration (χ2 = 6.958, df = 8, P = .541). These sensitivity analyses support the robustness of our primary categorical analysis for age and BMI.
To address potential outcome heterogeneity, we performed an additional sensitivity analysis restricted to patients with liver biochemical abnormalities without confirmed DILI (n = 306) and controls (n = 354). The results were generally consistent with the primary analysis. Male sex (OR = 1.520, 95% CI, 1.069–2.161; P = .020), age ≥60 years (OR = 5.901, 95% CI, 2.851–12.216; P < .001), obesity (OR = 4.326, 95% CI, 1.495–12.519; P = .007), medication history ≤1 year (OR = 2.608, 95% CI, 1.620–4.199; P < .001), dysglycemia (OR = 2.560, 95% CI, 1.326–4.942; P = .005), dyslipidemia (OR = 2.002, 95% CI, 1.404–2.854; P < .001), and length of stay (OR = 1.013, 95% CI, 1.002–1.024; P = .022) remained associated with liver biochemical abnormalities. The Hosmer–Lemeshow goodness-of-fit test indicated acceptable model calibration (χ2 = 5.332, df = 8, P = .722), as shown in Table S1, Supplemental Digital Content 1.
Additional exploratory analyses and model diagnostics were performed for the primary multivariable model. Exploratory interaction analyses showed no statistically significant interactions for medication history ≤1 year × dyslipidemia (OR = 0.574, 95% CI, 0.295–1.120; P = .104) or male sex × dyslipidemia (OR = 1.636, 95% CI, 0.811–3.297; P = .169). Assessment of influential observations did not identify any observation exerting substantial influence on the multivariable logistic regression estimates, as indicated by a maximum Cook’s distance of 0.256 and standardized residuals ranging from −2.815 to 2.077.
3.5. Exploratory analysis of cumulative olanzapine dose
To further characterize drug exposure, we performed an exploratory multivariable logistic regression model including cumulative olanzapine dose during hospitalization. Because cumulative olanzapine dose partly reflects exposure duration, length of stay was not included in this exploratory model. After adjustment for sex, age, BMI, medication history, dysglycemia, and dyslipidemia, cumulative olanzapine dose was associated with liver function abnormalities (OR = 1.109 per 100-mg increase, 95% CI, 1.035–1.188; P = .003). The Hosmer–Lemeshow goodness-of-fit test indicated acceptable model calibration (χ2 = 6.858, df = 8, P = .552), as shown in Table 3.
Table 3.
Exploratory multivariable logistic regression analysis including cumulative olanzapine dose.
| Characteristics | Variable | β | SE | Wald | P value | OR (95% CI) |
|---|---|---|---|---|---|---|
| Sex | Male vs female | 0.420 | 0.173 | 5.935 | .015 | 1.523 (1.086–2.135) |
| Age, yr | ≥60 vs 18–<30 | 1.640 | 0.361 | 20.630 | <.001 | 5.157 (2.541–10.466) |
| BMI, kg/m2 | Obese vs normal BMI (18.5–<24 kg/m2) | 1.337 | 0.540 | 6.130 | .013 | 3.808 (1.321–10.973) |
| Dysglycemia | Yes vs no | 0.956 | 0.326 | 8.578 | .003 | 2.602 (1.372–4.933) |
| Dyslipidemia | Yes vs no | 0.776 | 0.173 | 20.258 | <.001 | 2.174 (1.550–3.048) |
| Medication history, yr | ≤1 yr vs >5 yr | 0.937 | 0.228 | 16.963 | <.001 | 2.552 (1.634–3.987) |
| Cumulative olanzapine dose | Per 100-mg increase | 0.103 | 0.035 | 8.637 | .003 | 1.109 (1.035–1.188) |
Reference categories were female sex, age 18 to <30 years, normal BMI (18.5–<24 kg/m2), no dysglycemia, no dyslipidemia, and medication history >5 years. Cumulative olanzapine dose was analyzed per 100-mg increase. Length of stay was not included in this exploratory model because cumulative olanzapine dose partly reflects exposure duration. Only statistically significant variables or categories are shown; nonsignificant categories were retained in the model but are not displayed.
Model performance: Hosmer–Lemeshow goodness-of-fit test, χ2 = 6.858, df = 8, P = .552; area under the receiver operating characteristic curve = 0.717 (95% CI, 0.680–0.754; P < .001).
BMI = body mass index, CI = confidence interval, df = degrees of freedom, OR = odds ratio, SE = standard error.
4. Discussion
Psychiatric disorders are often chronic and persistent. They have a high recurrence rate and require long-term medication, either monotherapy or adjuvant combination drug therapy. Previous studies have found that the prevalence and incidence rates of chronic liver disease in patients with psychiatric disorders are 1.27 times and 1.15 times higher than those in the general population, respectively.[7] Similarly, the rates of hepatocellular carcinoma and cirrhotic complications are higher.[8] In this study, we performed univariate and multivariable logistic regression analyses to identify factors associated with liver function abnormalities in psychiatric patients receiving APs and/or ADs. The factors independently associated with liver function abnormalities in the multivariable analysis were male sex, age ≥60 years, obesity, dysglycemia, dyslipidemia, length of stay, and medication history ≤1 year. These findings may help inform preliminary risk assessment and liver function monitoring in psychiatric patients receiving APs and/or ADs. The sensitivity analysis restricted to liver biochemical abnormalities without confirmed DILI yielded generally consistent findings, suggesting that the main associations were not driven solely by the inclusion of confirmed DILI cases.
Olanzapine was the most frequently recorded AP/AD exposure among patients with liver function abnormalities in this study. This is consistent with the findings of Chen et al.[9] In the confirmed-DILI subgroup of the case group in this study, the proportion of patients using olanzapine was 68.75%, which was higher than that for the other drugs and suggests a potential signal requiring cautious interpretation. However, because complete drug-specific exposure denominators were unavailable, this finding may partly reflect local prescribing patterns, treatment indications, disease complexity, and residual confounding, and therefore should not be interpreted as evidence that olanzapine carries the highest hepatotoxic risk. In the additional exploratory exposure-related analysis, cumulative olanzapine dose during hospitalization was associated with liver function abnormalities. Specifically, each 100-mg increase in cumulative olanzapine dose was associated with higher odds of liver function abnormalities. This finding suggests that olanzapine exposure burden may be a clinically relevant signal requiring closer monitoring. However, because this was a retrospective observational study and complete dose, duration, dose escalation, and combination treatment data were not available for all APs and ADs, this result should be interpreted cautiously and should not be considered definitive evidence of a causal dose-response relationship. Biologically, olanzapine-related liver abnormalities may involve direct hepatocellular injury through oxidative stress and mitochondrial/lysosomal dysfunction, as well as indirect effects on hepatic lipid metabolism and steatosis.[10–13] Therefore, in patients receiving olanzapine, especially those with metabolic abnormalities or emerging liver dysfunction, clinicians should monitor liver function and metabolic parameters and consider individualized dose adjustment, drug switching, or hepatoprotective therapy when clinically indicated.
Age ≥60 years was found to be a risk factor for liver function abnormalities in psychiatric patients on both APs and ADs, in accordance with the previously published findings of a higher propensity for liver injury in elderly patients.[14] Another study by Han et al showed that the likelihood of a hepatic adverse drug reaction increases by 33% per decade of age.[15] The pharmacokinetic profile changes significantly with age. It has been established that elderly patients have a greater proportion of body fat (20%–40%) and a smaller total body water content (10%–15%). APs and ADs are mainly lipophilic compounds, and because of this, when used in elderly patients, their volume of distribution increases and their elimination half-life is prolonged, contributing to excessive drug accumulation. Elderly patients have reduced CYP enzyme activity, reduced hepatic blood flow, smaller liver mass, and reduced renal function, leading to lower first-pass metabolism and systemic clearance of drugs.[16] Individualized dosing principles of “low initial dose, slow titration” should be instituted for elderly psychiatric patients, in addition to closer monitoring of liver function and plasma drug levels to avoid toxic accumulation of lipophilic medications. Elderly patients with many comorbidities who use multiple and potentially conflicting medications should, in principle, have drug reviews performed by a clinical pharmacist on an ongoing basis to prevent further liver damage.
Metabolic disorders (including obesity, dysglycemia, and dyslipidemia) were independently associated with liver function abnormalities in psychiatric patients on APs or ADs. First, the association with obesity was strong. The result is consistent with a large prospective cohort study, suggesting that a high BMI in late adolescence is an independent risk factor predicting severe liver disease in adulthood, in which obesity was associated with a 117% increased risk.[17] The mechanism may be associated with obesity-mediated alterations in drug pharmacokinetics that promote a higher hepatotoxic potential of several drugs.[18] Obesity-related upregulation of CYP450 activity may also increase liver susceptibility to drug-induced hepatotoxicity, thereby contributing synergistically to an increased risk of DILI.[19] In addition, some weight-loss drugs have been shown to ameliorate liver steatosis and tissue inflammation, support weight management, and protect liver histology.[20] Second, in the case of maladaptive glucose metabolism, psychiatric drugs may affect glycemic control via various mechanisms, resulting in metabolic vicious cycles and distorted pharmacokinetics of the drug. Diabetes markedly affects the expression and activity of hepatic drug transporters and CYP450 enzymes and alters the communication between these enzymes to cause pharmacokinetic abnormalities.[21] However, at the same time, a hyperglycemic state triggers a hepatic inflammatory response, aggravating hepatocyte injury through mitochondrial oxidative stress, endoplasmic reticulum stress, and inhibition of lysosomal autophagy, which in turn inhibits liver activity.[22] Finally, dyslipidemia is strongly correlated with pathological changes in hepatocytes. Hyperlipidemia is a metabolic disease characterized by abnormal plasma or serum lipid levels. The liver is the primary organ responsible for lipid metabolism, which is involved in lipoprotein synthesis, β-oxidation of fatty acids, production of ketone bodies, synthesis of cholesterol, generation of bile, and lipid storage and mobilization.[23] Hyperlipidemia has been shown to produce oxidative stress, resulting in direct damage to intracellular components as well as injury to the hepatocytes and hepatic lipid peroxidation.[24] Given the role of metabolic dysregulation in liver biochemical abnormalities during AP and/or AD treatment, regular monitoring of weight, glucose, lipid profiles, and liver function should be considered in patients with metabolic abnormalities.
Length of stay was also associated with liver function abnormalities in this study, consistent with a previous case-control study.[25] This association may reflect clinical and exposure-related factors: longer hospital stays often indicate greater illness complexity and may be associated with a greater medication exposure burden and polypharmacy, which could contribute to liver biochemical abnormalities.[26,27] Therefore, length of stay should be interpreted as a clinical risk indicator rather than as direct evidence of a drug-specific hepatotoxic mechanism. Moreover, our results showed that patients with a medication history of ≤1 year were more likely to develop liver function abnormalities than those with a longer medication history (>5 years). Although this finding is compatible with previous observations that DILI and metabolic adverse events often emerge during early treatment,[6,28] it may also reflect illness severity, healthcare engagement, monitoring intensity, or other unmeasured factors rather than a purely pharmacological mechanism. Therefore, a medication history of ≤1 year should be interpreted as a clinical risk marker for closer monitoring, not as evidence of a causal treatment-duration effect.
Male sex was independently associated with liver function abnormalities following AP and/or AD treatment in patients with psychiatric disorders. This finding is consistent with the outcomes documented by Zhang et al.[29] Chao et al further elucidated that males, especially those between 21 and 30 years old, are at high risk for DILI.[30] Sex differences in the pathophysiology and clinical course of chronic liver diseases have been well documented, with women generally showing better outcomes than men.[31] Men with liver function abnormalities were more severely ill and had worse outcomes over a period, according to a prospective study, which may be partly related to the higher prevalence of smoking and alcohol drinking among these males.[32] Although smoking and alcohol histories were reviewed from the admission medical records and personal history, underreporting could not be completely excluded in this retrospective medical-record-based study, and residual confounding may remain. Accordingly, male sex should be interpreted as a marker for closer monitoring rather than as evidence of a direct biological effect. In clinical practice, liver function monitoring should be individualized, particularly when smoking, alcohol use, metabolic abnormalities, or other liver disease risk factors are present or incompletely documented.
5. Conclusion
In this study, liver biochemical abnormalities occurring after AP and/or AD exposure were evaluated as early warning signals of potential liver injury. Liver function abnormalities were associated with male sex, age ≥60 years, metabolic disorders, longer hospital stay, and a medication history of ≤1 year. Olanzapine was the most frequently recorded AP/AD exposure among affected patients, but this descriptive finding should be interpreted cautiously in the absence of complete dose-response data. These findings may support preliminary risk assessment and closer liver function monitoring in psychiatric inpatients receiving AP and/or AD treatment, but multicenter validation is needed.
6. Limitations
This study has some limitations. First, the >1.5 × ULN threshold was a sensitive operational cutoff for early liver biochemical abnormalities rather than a psychiatric-specific or formal DILI diagnostic criterion; therefore, mild or transient abnormalities may have been included, and outcome heterogeneity may have been introduced. Although we performed a sensitivity analysis restricted to biochemical abnormalities without confirmed DILI, the DILI subgroup was too small for a separate multivariable analysis. Second, because this was a single-center retrospective unmatched case-control study based on hospital records, selection bias, survivor-control bias, residual confounding, and limited external validity could not be excluded. Although length of stay was included as a continuous covariate to partly account for differential observation time, this adjustment could not fully eliminate potential survivor-control bias. Controls were required to remain free of liver biochemical abnormalities during hospitalization, and the 1:1 design did not involve individual matching or propensity score adjustment; therefore, the findings should be interpreted as associations rather than causal effects. Third, although patients with documented smoking, alcohol use, viral hepatitis, and fatty liver disease were excluded by design, and individual metabolic parameters (BMI, dysglycemia, dyslipidemia) were adjusted for, underreporting of smoking or alcohol use and incomplete screening for viral hepatitis or liver imaging could not be excluded. Comprehensive metabolic syndrome assessment, disease severity or healthcare-engagement indicators, herbal medicine use, over-the-counter drugs, and nutritional status were not available. Fourth, drug exposure assessment was incomplete: drug-specific exposure denominators, detailed daily dose, treatment duration, dose escalation, and combination patterns for most APs/ADs were unavailable, precluding reliable estimation of drug-specific incidence, adjusted comparative ORs, or diagnosis-/regimen-stratified analyses; only cumulative olanzapine dose was analyzed as an exploratory measure. Fifth, internal validation (bootstrap/cross-validation) was not performed because the model aimed to identify associated factors rather than to serve as a validated prediction tool. These limitations should be considered when interpreting the findings, and multicenter prospective validation is needed.
Author contributions
Conceptualization: Yaoyao Xiu.
Data curation: Jing Zhu.
Formal analysis: Jing Zhu.
Software: Jing Zhu.
Project administration: Qin Zhou.
Resources: Qin Zhou.
Supervision: Qin Zhou, Yaoyao Xiu.
Writing – original draft: Jing Zhu.
Writing – review & editing: Jing Zhu, Yaoyao Xiu.
Abbreviations:
- ADs
- antidepressants
- ALP
- alkaline phosphatase
- ALT
- alanine aminotransferase
- APs
- antipsychotics
- BMI
- body mass index
- CI
- confidence interval
- DILI
- drug-induced liver injury
- OR
- odds ratio
- RUCAM
- Roussel Uclaf Causality Assessment Method
- ULN
- upper limit of normal
This retrospective study was reviewed and approved by the Ethics Committee of Xuzhou Oriental People’s Hospital (Xuzhou Oriental Hospital Affiliated to Xuzhou Medical University; approval No. 20260609001). Individual written informed consent was not required for this retrospective analysis and was therefore not obtained because the study used only deidentified retrospective electronic medical record data, involved no additional examinations or interventions, involved no direct contact with patients, and posed no more than minimal risk to participants. All personal identifiers were removed before analysis to protect patient privacy and confidentiality. The study was conducted in accordance with the principles of the Declaration of Helsinki and relevant institutional requirements.
The authors have no funding and conflicts of interest to declare.
The datasets generated and/or analyzed during the current study are not publicly available but are available from the corresponding author on reasonable request.
Supplemental Digital Content is available in the online version of this article (http://dx.doi.org/10.1097/MD.0000000000050232).
How to cite this article: Zhu J, Zhou Q, Xiu Y. Risk factors for antipsychotic- and antidepressant-associated liver function abnormalities: A retrospective unmatched case-control study. Medicine 2026;105:33(e50232).
Contributor Information
Jing Zhu, Email: 921990398@qq.com.
Qin Zhou, Email: employ123@163.com.
References
- [1].Todorović Vukotić N, Đorđević J, Pejić S, Đorđević N, Pajović SB. Antidepressants- and antipsychotics-induced hepatotoxicity. Arch Toxicol. 2021;95:767–89. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [2].Hiemke C, Bergemann N, Clement HW, et al. Consensus guidelines for therapeutic drug monitoring in neuropsychopharmacology: update 2017. Pharmacopsychiatry. 2018;51:e1. [DOI] [PubMed] [Google Scholar]
- [3].Hassan A, Fontana RJ. The diagnosis and management of idiosyncratic drug-induced liver injury. Liver Int. 2019;39:31–41. [DOI] [PubMed] [Google Scholar]
- [4].Ma SW, Liu CH, Liu XY, et al. Chinese guidelines for the diagnosis and treatment of drug-induced liver injury (2023 edition) [in Chinese]. Wei Chang Bing Xue. 2023;28:397–431. [Google Scholar]
- [5].National Cancer Institute. Common Terminology Criteria for Adverse Events (CTCAE) version 5.0. U.S. Department of Health and Human Services, National Institutes of Health, National Cancer Institute; 2017. [Google Scholar]
- [6].Chalasani NP, Maddur H, Russo MW, Wong RJ, Reddy KR; Practice Parameters Committee of the American College of Gastroenterology. ACG clinical guideline: diagnosis and management of idiosyncratic drug-induced liver injury. Am J Gastroenterol. 2021;116:878–98. [DOI] [PubMed] [Google Scholar]
- [7].Hsu JH, Chien IC, Lin CH, Chou YJ, Chou P. Increased risk of chronic liver disease in patients with schizophrenia: a population-based cohort study. Psychosomatics. 2014;55:163–71. [DOI] [PubMed] [Google Scholar]
- [8].Yip TC, Wong GL, Tse YK, et al. High incidence of hepatocellular carcinoma and cirrhotic complications in patients with psychiatric illness: a territory-wide cohort study. BMC Gastroenterol. 2020;20:128. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [9].Chen H, Liang J, Zhao W, Xia QR. Clinical characteristics and risk factors of liver injury induced by atypical antipsychotic drugs. Zhongguo Yaowu Jingjie. 2025;22:410–4. [Google Scholar]
- [10].Eftekhari A, Azarmi Y, Parvizpur A, Eghbal MA. Involvement of oxidative stress and mitochondrial/lysosomal cross-talk in olanzapine cytotoxicity in freshly isolated rat hepatocytes. Xenobiotica. 2016;46:369–78. [DOI] [PubMed] [Google Scholar]
- [11].El-Shoura EAM, Abdelzaher LA, Mahmoud NI, et al. Combined sulforaphane and β-sitosterol mitigate olanzapine-induced metabolic disorders in rats: insights on FOXO, PI3K/AKT, JAK/STAT3, and MAPK signaling pathways. Int Immunopharmacol. 2024;140:112904. [DOI] [PubMed] [Google Scholar]
- [12].Buron N, Porceddu M, Loyant R, et al. Drug-induced impairment of mitochondrial fatty acid oxidation and steatosis: assessment of causal relationship with 45 pharmaceuticals. Toxicol Sci. 2024;200:369–81. [DOI] [PubMed] [Google Scholar]
- [13].Li R, Zhu W, Huang P, et al. Olanzapine leads to nonalcoholic fatty liver disease through the apolipoprotein A5 pathway. Biomed Pharmacother. 2021;141:111803. [DOI] [PubMed] [Google Scholar]
- [14].Yu S, Li J, He T, et al. Age-related differences in drug-induced liver injury: a retrospective single-center study from a large liver disease specialty hospital in China, 2002–2022. Hepatol Int. 2024;18:1202–13. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [15].Han YZ, Guo YM, Xiong P, et al. Age-associated risk of liver-related adverse drug reactions. Front Med (Lausanne). 2022;9:832557. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [16].Ngcobo NN. Influence of ageing on the pharmacodynamics and pharmacokinetics of chronically administered medicines in geriatric patients: a review. Clin Pharmacokinet. 2025;64:335–67. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [17].Hagström H, Tynelius P, Rasmussen F. High BMI in late adolescence predicts future severe liver disease and hepatocellular carcinoma: a national, population-based cohort study in 1.2 million men. Gut. 2018;67:1536–42. [DOI] [PubMed] [Google Scholar]
- [18].Allard J, Le Guillou D, Begriche K, Fromenty B. Drug-induced liver injury in obesity and nonalcoholic fatty liver disease. Adv Pharmacol. 2019;85:75–107. [DOI] [PubMed] [Google Scholar]
- [19].Lucena MI, Villanueva-Paz M, Alvarez-Alvarez I, et al. Roadmap to DILI research in Europe. A proposal from COST action ProEuroDILINet. Pharmacol Res. 2024;200:107046. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [20].Polyzos SA, Goulis DG, Giouleme O, Germanidis GS, Goulas A. Anti-obesity medications for the management of nonalcoholic fatty liver disease. Curr Obes Rep. 2022;11:166–79. [DOI] [PubMed] [Google Scholar]
- [21].Yang Y, Liu X. Imbalance of drug transporter-CYP450s interplay by diabetes and its clinical significance. Pharmaceutics. 2020;12:348. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [22].Ye SY, Qin Y. Research progress on the mechanisms of diabetes-induced liver injury. Chin Hepatol. 2023;28:737–9. [Google Scholar]
- [23].Hlušička J, Žák A. Dyslipidaemia in liver diseases. Folia Biol (Praha). 2024;70:239–47. [DOI] [PubMed] [Google Scholar]
- [24].Si L, Chen L, Abdulla R, Aisa HA. Anti-hyperlipidemic effects of Apocynum venetum L. leaves extract on high-fat diet-induced hyperlipidemia: modulation of lipid metabolism and oxidative stress. Food Res Int. 2025;221:117326. [DOI] [PubMed] [Google Scholar]
- [25].Liu Y, Yang M, Ding Y, et al. Clinical significance of potential drug-drug interactions in older adults with psychiatric disorders: a retrospective study. BMC Psychiatry. 2022;22:563. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [26].Hefner G, Wolff J, Hahn M, et al. Prevalence and sort of pharmacokinetic drug-drug interactions in hospitalized psychiatric patients. J Neural Transm (Vienna). 2020;127:1185–98. [DOI] [PubMed] [Google Scholar]
- [27].Vitale G, Mattiaccio A, Conti A, et al. Molecular and clinical links between drug-induced cholestasis and familial intrahepatic cholestasis. Int J Mol Sci. 2023;24:5823. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [28].Feng Y, Wong KC, Leung PB, et al. Longitudinal impact of different treatment sequences of second-generation antipsychotics on metabolic outcomes: a study using targeted maximum likelihood estimation. Psychol Med. 2025;55:e123. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [29].Zhang YX, Chen CM, Zhu YW, et al. Study on abnormal liver function and its related factors in patients with severe mental disorders in Shanghai communities. Shanghai Yufang Yixue. 2024;36:1018–25. [Google Scholar]
- [30].Chao N, Chen Y, Wang Y, et al. Retrospective analysis of 198 cases of liver damage caused by antipsychotic drugs [in Chinese]. Jiefangjun Yaoxue Xuebao. 2023;36:57–59. [Google Scholar]
- [31].Sayaf K, Gabbia D, Russo FP, De Martin S. The role of sex in acute and chronic liver damage. Int J Mol Sci. 2022;23:10654. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [32].Pedraza L, Laosa O, Rodríguez-Mañas L, et al. Drug induced liver injury in geriatric patients detected by a two-hospital prospective pharmacovigilance program: a comprehensive analysis using the Roussel Uclaf Causality Assessment Method. Front Pharmacol. 2020;11:600255. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.