A retrospective study of clinical characteristics and steroid therapy in immune checkpoint inhibitor-mediated hepatitis

Abstract

Background:

With the widespread clinical application of immune checkpoint inhibitors (ICIs), immune-mediated hepatitis (IMH) has become increasingly prevalent.

Objectives:

This study aims to analyze the clinical characteristics, steroid treatment, and prognosis of IMH patients, providing further evidence to enhance the safety of ICIs in clinical practice.

Design:

A retrospective cohort study included tumor patients who received ICI therapy. Comprehensive analyses were conducted to explore the factors influencing the occurrence, clinical characteristics, and prognosis of IMH.

Methods:

Tumor patients treated with PD-1/PD-L1 inhibitors were enrolled in our study. Patients were stratified based on the occurrence and severity of IMH, as well as the administration of glucocorticoid therapy, to investigate the risk factors for IMH development, critical factors influencing IMH progression, and treatment-specific outcomes.

Results:

Of 744 patients, 341 (45.8%) developed IMH. IMH was significantly more frequent in females (p = 0.001), younger patients (p < 0.001), those receiving ICIs with targeted therapy (p = 0.009), and patients with hepatocellular carcinoma (HCC; p < 0.001) or gastric cancer (p < 0.001). Coexisting hepatitis B (p < 0.001), cirrhosis (p = 0.005), and fatty liver disease (p = 0.028) were also associated with higher IMH risk. Independent risk factors included female gender, age <45 years, HCC, and gastric cancer. Younger patients were more likely to develop severe IMH (p = 0.003). Hepatocellular injury was the most common type of IMH across all grades, with similar risks of severe progression among different IMH types. Glucocorticoid therapy improved outcomes (p = 0.011), particularly in grade 3–4 IMH, although no significant difference in outcomes was observed between groups receiving sufficient versus insufficient recommended doses.

Conclusion:

Female gender, younger age (<45 years), HCC, and gastric cancer are independent risk factors for IMH. Younger patients are more likely to develop severe IMH. Glucocorticoid therapy is beneficial, particularly in grade 3–4 IMH. Early detection, prompt intervention, and tailored management strategies may help mitigate progression and improve outcomes.

Introduction

Malignant tumors persist as a predominant cause of human mortality, and combating cancer continues to be a major global health priority. ¹ Within the tumor microenvironment, neoplastic cells exploit various immune escape mechanisms to evade immune surveillance. These include reduced immunogenicity, suppression of immune cell function, and circumvention of immune detection. ²

Immune checkpoint inhibitors (ICIs) represent an important advancement in cancer treatment. These therapeutic monoclonal antibodies target key regulatory molecules involved in T-cell inactivation, thus counteracting immune evasion by tumor cells. ³ They significantly enhance the reactivity of T cells specific to tumors, augment the activity of the immune system, and consequently exert anti-tumor effects. ICIs have demonstrated substantial efficacy across multiple malignancies, significantly improving survival outcomes. ⁴

With the widespread clinical application of ICIs, immune-related adverse events (irAEs) have become a common clinical concern. These adverse events can affect multiple organ systems, including the gastrointestinal tract, lungs, skin, endocrine glands, and liver. ⁵ Elevated liver enzymes occur in approximately 10% of patients receiving ICI monotherapy, with CTLA-4 inhibitors associated with a higher incidence than PD-1 inhibitors. Immune-mediated hepatitis (IMH), a liver-related adverse reaction to ICIs, accounts for 22% of deaths among fatalities related to anti-PD-1/PD-L1 therapy. ⁶ The incidence of ICI-induced liver injury, within the spectrum of drug-induced liver injury (DILI), is increasing. Reported incidence rates for all grades of IMH range from approximately 1.72% to 37%, varying with factors such as ICI type and patient population. ⁷

Current guidelines generally recommend discontinuing ICIs and initiating glucocorticoids (GS) or immunosuppressants for patients with grade ⩾2 IMH.^8–11 Although proactive immune modulation combined with hepatoprotective agents achieves clinical resolution in most cases, progression to fulminant acute liver failure can still occur in approximately 1.2%–3.7% of severe IMH cases. ¹² Despite the increasing recognition of IMH, relevant clinical research is limited, and the underlying mechanisms remain unclear. Our study provides a comprehensive clinical description of IMH in cancer patients treated with PD-1 or PD-L1 inhibitors at a tertiary hospital in southern China. We aim to better understand the clinical manifestations of IMH and the prognosis with or without GS.

Materials and methods

Study design and patient selection

From January 2020 to December 2022, cancer patients who received at least one dose of PD-1 or PD-L1 inhibitor therapy at a hospital in southern China were included in this retrospective study. Notably, patients receiving CTLA-4 inhibitors were not included in this study due to the limited number of such cases during the study period. The causality of hepatotoxicity was determined by evaluating the patient’s overall condition, excluding other possible causes of liver damage, such as viral hepatitis (new infections or reactivations), active alcoholic hepatitis, autoimmune liver disease, medications other than ICIs, and tumor progression. The ethics committee approved this retrospective study. The reporting of this study conforms to the STROBE statement (Supplemental Material 1). ¹³

The diagnosis and grading of IMH followed the National Cancer Institute Common Terminology Criteria for Adverse Events (CTCAE; version 5.0), ¹⁴ the Chinese guideline for diagnosis and management of DILI (2023 version), ⁸ and the EASL Clinical Practice Guidelines: Drug-induced liver injury. ⁹ IMH was defined as liver injury that was absent before ICI treatment but appeared following ICI treatment, or worsened after ICI treatment if it existed prior to treatment. It is primarily characterized by increases in serum levels of alanine aminotransferase (ALT), aspartate aminotransferase (AST), total bilirubin (TBIL), or alkaline phosphatase (ALP). The severity of IMH is categorized into four grades based on elevated levels of hepatic enzymes or bilirubin^8,9,14: grade 1, ALT and/or AST < 3 × upper limit of normal (ULN), TBIL < 1.5 × ULN or ALP < 2.5 × ULN; grade 2, ALT and/or AST 3–5 × ULN, TBIL 1.5–3 × ULN or ALP 2.5–5 × ULN; grade 3, ALT and/or AST 5–20 × ULN, TBIL 3–10 × ULN or ALP 5–20 × ULN; grade 4, ALT and/or AST > 20 × ULN, TBIL > 10 × ULN or ALP > 20 × ULN.

Treatment protocols

The treatment plan for each patient was developed based on their height, weight, and the recommended dose outlined in the corresponding medication guidelines. All HBsAg-positive patients received nucleotide analogs antiviral therapy during immunotherapy. The dosage of prednisone equivalents was recorded for all patients who received corticosteroid treatment. Patients who received the recommended dose (prednisone at 0.5–1 mg/kg/day for grade 2 IMH and methylprednisolone at 1–2 mg/kg/day for more severe IMH) were considered to have received sufficient treatment.

Data collection

Clinical and laboratory data were collected during hospitalization and included clinical characteristics as well as liver function tests (including ALT, AST, TBIL, and ALP). IMH was classified into three categories based on the R-ratio: hepatocellular liver injury (R ⩾ 5), cholestatic liver injury (R ⩽ 2), and mixed hepatocellular and cholestatic liver injury (2 < R < 5).^8,9 The R-ratio was calculated as follows: (R-ratio = (ALT measured value/ALT ULN)/(ALP measured value/ALP ULN), with AST serving as an alternative to ALT in case ALT data were unavailable). ⁸

The prognosis was defined as either improvement (downgrading of liver function impairment) or no improvement (no change in liver function grade, deterioration, or death).

Outcomes

The primary outcome was the incidence and risk factors of IMH. The secondary outcome was the improvement of hepatitis with or without steroid therapy, and across different doses of corticosteroids.

Statistical analyses

All analyses were performed using SPSS version 26.0 (IBM Corporation, Armonk, New York, USA). Continuous variables were reported as medians (interquartile range), while categorical variables were reported as percentages. The rank-sum test was used for comparisons of continuous variables, and the chi-squared test, along with multiple comparisons, was used for categorical variables. Variables with statistical significance in univariate analysis (p < 0.10) and clinical relevance were considered as candidate predictors. Before performing multivariate analysis, multicollinearity was assessed using variance inflation factors (VIF), with all variables showing VIF <5 (indicating no severe collinearity). Multivariate logistic regression was then performed using forward stepwise selection (entry p < 0.05, removal p > 0.10) based on the likelihood ratio test. p < 0.05 was considered statistically significant.

Results

Clinical characteristics of the IMH group and the non-IMH group

From January 2020 to December 2022, 933 patients diagnosed with a tumor and treated with PD-1 or PD-L1 inhibitors at a hospital in southern China were enrolled in our study. Excluding 189 patients with missing data and those caused by other diseases, the final study cohort included 744 patients (Figure 1). The distribution of tumor types among the cohort is summarized in Figure 2, with lung cancer being the most prevalent (33.6%), followed by gastric cancer (12.5%) and colorectal cancer (9.1%). Specific subtypes, such as hepatocellular carcinoma (HCC; 5.8%) and melanoma (4.0%), were also represented. Other rare tumors or those with fewer cases were collectively categorized as “other types of tumors,” primarily including head and neck tumors, lymphomas, urinary system tumors, and reproductive system tumors, among others.

Figure 1.

Flow chart of the patient inclusion process.

ICIs, immune checkpoint inhibitors; IMH, immune-mediated hepatitis.

Figure 2.

Schematic diagram of tumor type distribution (N = 744).

The median age of the patients was 57.0 (50.0, 64.0) years, with 499 (67.1%) males and 245 (32.9%) females. In all, 79 (10.6%) patients were receiving PD-1/PD-L1 inhibitors monotherapy, 367 (49.3%) combined chemotherapy, and 128 (17.2%) combined targeted therapy (Table 1). IMH of any grade occurred in 341 patients (45.8%), but only 50 cases (6.7%) developed grade ⩾3 IMH. Respectively, 245 (71.8%), 46 (13.5%), 38 (11.2%), and 12 (3.5%) patients were classified as grade 1, 2, 3, and 4 IMH (Figure 1).

Table 1.

Clinical characteristics of patients treated with immune checkpoint inhibitors.

Characteristics	Total (N = 744)	Univariate analysis			Multivariate analysis
		Non-IMH group (N = 403), n (%)	IMH group (N = 341), n (%)	p Value	OR	95% CI	p Value
Sex
Male	499	292 (58.5)	207 (41.5)	0.001	0.591	0.429–0.815	0.001
Female	245	111 (45.3)	134 (54.7)
Age
<45	106	40 (37.7)	66 (62.3)	<0.001	1.757	1.126–2.743	0.013
45–59	357	195 (54.6)	162 (45.4)	0.811
⩾60	281	168 (59.8)	113 (40.2)	0.017
Type of ICI
Anti-PD-1	710	383 (53.9)	327 (46.1)	0.577
Anti-PD-L1	25	15 (60.0)	10 (40.0)	0.552
Anti-PD-1/PD-L1	9	5 (55.6)	4 (44.4)	1.000
Type of tumor
Lung cancer	250	161 (64.4)	89 (35.6)	<0.001
Gastric cancer	93	36 (38.7)	57 (61.3)	<0.001	1.895	1.193–3.011	0.007
Colorectal cancer	68	32 (47.1)	36 (52.9)	0.217
Hepatocellular carcinoma	43	7 (16.3)	36 (83.7)	<0.001	7.866	3.417–18.108	<0.001
Other types of tumors	260	150 (57.7)	110 (42.3)	0.157
Melanoma	30	17 (56.7)	13 (43.3)	0.779
Type of immunotherapy
ICIs monotherapy	79	50 (63.3)	29 (36.7)
Combined chemotherapy	367	209 (56.9)	158 (43.1)	0.300 a
Combined targeted therapy	128	57 (44.5)	71 (55.5)	0.009 a
Combined both	170	87 (51.2)	83 (48.8)	0.074 a
Underlying liver disease
Hepatitis B (non-reactivation)	98	38 (38.8)	60 (61.2)	<0.001
Liver cirrhosis	26	7 (26.9)	19 (73.1)	0.005
Fatty liver disease	28	10 (35.7)	18 (64.3)	0.028
Liver metastases	134	69 (51.5)	65 (48.5)	0.346

Anti-PD-1/PD-L1, involves the sequential use of both anti-PD-1 and anti-PD-L1. Other types of tumors mainly include head and neck tumors, lymphomas, urinary system tumors, etc.

^aCompared with ICIs monotherapy.

Anti-PD-1, antibodies targeting the programmed death-1 receptor; anti-PD-L1, antibodies targeting the ligand of the programmed death-1 receptor; ICI, immune checkpoint inhibitor; IMH, immune-mediated hepatitis.

Our results indicated that IMH was more common in females (54.7% vs 41.5%, p = 0.001) and younger patients (p < 0.001). The type of PD-1 or PD-L1 inhibitors did not influence the incidence of IMH. Compared to other types of malignant tumors, patients with HCC or gastric cancer had an increased incidence of IMH (p < 0.001). In our study, 79 patients received PD-1/PD-L1 inhibitors monotherapy and only 29 patients (36.7%) experienced IMH, which had a lower risk of developing hepatitis compared to those receiving combined targeted therapy (p = 0.009). However, patients receiving combined chemotherapy did not increase in the incidence of liver injury (p = 0.300; Table 1).

In addition, univariate analysis also showed that underlying liver diseases, including hepatitis B (non-reactivation), liver cirrhosis, and fatty liver disease, were all associated with IMH. However, the prevalence of liver metastasis was similar in patients with and without hepatitis. Furthermore, multivariate analysis identified female sex, age <45 years, HCC, and gastric cancer as independent risk factors for the occurrence of IMH (Table 1).

Clinical characteristics of IMH patients with different grades

Patients with IMH were grouped based on the severity of liver injury, and their clinical characteristics were analyzed. No significant gender difference was observed in the severity of IMH (p = 0.860). In addition, the study demonstrated that the severity of IMH was higher in younger patients (p = 0.003). Across all groups, hepatocellular liver injury was the most commonly observed type, with no significant association between IMH grades and liver injury types. Similarly, there was no difference in ICIs using frequency and times (Table 2).

Table 2.

Clinical characteristics of patients with different severities of immune-mediated hepatitis.

Characteristics	Grade 1 (n = 245)	Grade 2 (n = 46)	Grade 3 (n = 38)	Grade 4 (n = 12)	p Value
Sex, n (%)
Male	151 (61.6)	28 (60.9)	22 (57.9)	6 (50.0)	0.860
Female	94 (38.4)	18 (39.1)	16 (42.1)	6 (50.0)
Age (years)	56.0 (48.5, 64.5)	53.5 (47.0, 63.0)	49.0 (41.0, 56.3)	46.5 (33.8, 61.8)	0.003
Type of liver injury, n (%)
Hepatocellular liver injury	112 (45.7)	28 (60.9)	18 (47.4)	10 (83.3)	0.077
Cholestatic liver injury	107 (43.7)	15 (32.6)	13 (34.2)	2 (16.7)
Mixed liver injury	26 (10.6)	3 (6.5)	7 (18.4)	0 (0.0)
ICI frequency (times)	2.0 (1.0, 5.0)	2.0 (1.0, 5.3)	3.0 (2.0,6.0)	3.0 (1.0,5.0)	0.602
Onset time of IMH (weeks)	9.0 (4.0, 24.0)	7.0 (3.8, 26.3)	10.0 (7.0, 25.5)	11.0 (3.5, 20.8)	0.772
Type of tumor, n (%)
Lung cancer	78 (31.8)	6 (13.0)	4 (10.5)	1 (8.3)	0.001
Gastric cancer	40 (16.3)	6 (13.0)	9 (23.7)	2 (16.7)	0.615
Colorectal cancer	22 (9.0)	9 (19.6)	5 (13.2)	0 (0.0)	0.121
Hepatocellular carcinoma	12 (4.9)	5 (10.9)	13 (34.2)	6 (50.0)	<0.001
Melanoma	11 (4.5)	0 (0.0)	1 (2.6)	1 (8.3)	0.309
Other types of tumors	82 (33.5)	20 (43.5)	6 (15.8)	2 (16.7)	0.031
Type of immunotherapy, n (%)
ICIs monotherapy	23 (9.4)	2 (4.3)	4 (10.5)	0 (0.0)
Combined chemotherapy	132 (53.9)	15 (32.6)	7(18.4)	4 (33.3)	0.202 a
Combined targeted therapy	33 (13.5)	18 (39.1)	15 (39.5)	5 (41.7)	0.017 a
Combined both	57 (23.3)	11 (23.9)	12 (31.6)	3 (25.0)	0.694 a
Underlying liver disease, n (%)
Hepatitis B (non-reactivation)	31 (12.7)	8 (17.4)	15 (39.5)	6 (50.0)	<0.001
Cirrhosis	7 (2.9)	2 (4.3)	9 (23.7)	1 (8.3)	<0.001
Fatty liver disease	14 (5.7)	1 (2.2)	2 (5.3)	1 (8.3)	0.664
Liver metastases	43 (17.6)	14 (30.4)	7 (18.4)	1 (8.3)	0.163
Improvement rate	167 (68.2)	32 (69.6)	26 (68.4)	5 (41.7)	0.284

^aCompared with ICIs monotherapy.

ICI, immune checkpoint inhibitor; IMH, immune-mediated hepatitis.

Patients with grade 3–4 IMH had a higher proportion of HCC (p < 0.001). In this study, only 5.1% of patients (4/79) who received PD-1/PD-L1 inhibitors monotherapy developed grade 3–4 IMH. Combined targeted therapy was associated with increased severity of IMH (p = 0.017), with 28.1% (20/71) of patients experiencing grade 3–4 IMH. Concomitant underlying hepatitis B and liver cirrhosis were associated with the severity of IMH (p < 0.001) (Table 2).

Corticosteroid dosage and hepatitis improvement

Although guidelines recommend prednisone doses ranging from 0.5 to 1 mg/kg/day for grade 2 IMH and methylprednisolone 1–2 mg/kg/day for more severe IMH, the optimal timing and dosage for corticosteroid administration remain controversial. A total of 96 IMH cases with grade ⩾2 were evaluated, of which only 52 patients received corticosteroids. Although nearly half of the patients (52.3%, 23/44) who discontinued ICIs can improve spontaneously without corticosteroid treatment. Our results showed corticosteroid administration had increased improvement rate (76.9% vs 52.3%, p = 0.011), especially in those with grade ⩾3 IMH (75.0% vs 45.5%, p = 0.033). At the same time, our result showed that the corticosteroid dosage did not affect the improvement rate, as there was no significant difference between patients who received sufficient or insufficient recommended doses of corticosteroids (Table 3 and Figure 3).

Table 3.

Improvement rates with or without corticosteroid administration.

Grade	Corticosteroid administration			Recommended dosage
	With	Without	p Value	Sufficient	Insufficient	p-Value
Grade 2	19/24 (79.2%)	13/22 (59.1%)	0.139	6/8 (75.0%)	13/16 (81.3%)	1.000
Grade 3–4	21/28 (75.0%)	10/22 (45.5%)	0.033	7/10 (70.0%)	14/18 (77.8%)	0.674
Total	40/52 (76.9%)	23/44 (52.3%)	0.011	13/18 (72.2%)	27/34 (79.4%)	0.731

Figure 3.

(a) Improvement rates with or without corticosteroids administration in G2–G4 patients. (b) Improvement rates with sufficient or insufficient dosage of corticosteroids in G2–G4 patients.

Discussion

The expanding clinical application of ICIs in various malignancies has been accompanied by a rising incidence of IMH. Published data demonstrate that the incidence of all-grade IMH associated with ICI treatment ranges from 1.72% to 37%. ⁷ Notably, a few retrospective studies have indicated higher rates of liver injury, with reported incidences for all-grade IMH reaching up to 51%–73% in certain cohorts.^15,16 Consistent with these findings, our study reported a high incidence of IMH across all grades of 45.8% (341/744) in all patients, and 35.9% (29/79) in patients receiving PD-1/PD-L1 inhibitor monotherapy. Previous systematic reviews and meta-analyses have reported a pooled incidence of grade 3–4 hepatotoxicity following ICI treatment ranging from 0.6% to 33.3%.^17–19 In our study, grade 3–4 IMH occurred in 6.7% (50/744) of patients, similar to other results. The study primarily included cases classified as grade 1–2 IMH. After excluding alternative etiologies, the high IMH incidence underscores the critical importance of close monitoring of liver function throughout the course of immunotherapy.

In this study, the enrollment of male patients exceeded that of females, potentially reflecting the disease spectrum of the underlying malignancies. Specifically, common malignancies in males include lung cancer, while females exhibit a higher propensity for breast cancer. ¹ Instead of ICIs, breast cancer patients usually undergo endocrine therapy. Despite the predominance of male patients in this study, interestingly, multivariate analysis identified that female sex was an independent risk factor for IMH development, which is consistent with previous research. ²⁰ Furthermore, other studies have reported that the risk of IMH in females is almost three times higher than that in males. ²¹ Potential contributors to this gender disparity include differences in gastric acid secretion, gastrointestinal blood flow, drug-binding protein levels, fat content, cytochrome P450 (CYP450) isoenzymes expression, and hormonal influences, impacting the pharmacokinetics or pharmacodynamics of certain drugs. ²² Moreover, females generally exhibit a higher incidence of various autoimmune diseases, encompassing autoimmune hepatitis and primary biliary cholangitis.^23,24 Some researchers have suggested that sex hormones might interact with immune modulators or signaling molecules to significantly impact the immune system.^22,25 Notably, our study did not find a significant association between sex and the severity grade of IMH. Nevertheless, a prior study has found that among patients receiving immunotherapy, females have an increased risk of severe symptomatic irAEs. ²⁶ Further investigation is needed to elucidate the complex relationship between gender and the severity of IMH.

Cancer stands as the primary cause of mortality in individuals aged 60–79 years, with a median age of diagnosis of 65 and a median age of death of 74, and patients in this age group are more than twice as likely to develop invasive cancer as younger patients. ²⁷ In our study, the median age of patients receiving ICI treatment was 57.0 (50.0, 64.0) years, indicating an overall younger age compared to the general cancer population. This discrepancy may be attributed to the poorer status of elderly patients, who may have lower tolerability for immunotherapy. In addition, economic constraints in China may lead to a reduced demand for anticancer treatment among older patients, resulting in lower utilization of ICIs. Furthermore, in our study, age under 45 years was identified as an independent risk factor for IMH, and these patients tended to exhibit higher severity grades. This aligns with previous reports that younger patients are more likely to develop IMH than older patients, and that lower age was associated with all grades and grade 3–4 liver injury.^21,28 Another study also noted a decline in the prevalence of hepatotoxicity with age, reporting that 80% of patients aged 30–50 experienced liver injury, compared to 72% of those aged 50–70, and 50% of those over 70 years. ¹⁶ This phenomenon may be attributed to the stronger immune functionality in younger patients, resulting in more intense immune responses. Indeed, advancing age is often accompanied by a decline in immune function, a concept known as immunosenescence. ²⁹ This age-related decline is thought to primarily affect T-cell-mediated immune function, ³⁰ which corresponds with the mechanism of IMH development. As such, special attention should be paid to hepatic adverse events, and other potential systemic immune adverse reactions when prescribing ICIs to younger individuals.

A previous study identified the combination of ICIs with targeted therapy as an independent risk factor for IMH. ²⁰ Consistent with this finding, our study revealed a significantly higher incidence of IMH in patients receiving PD-1/PD-L1 inhibitors combined with targeted therapy compared to those on monotherapy. Nevertheless, there is no significant difference in the occurrence of liver damage between ICIs combined with chemotherapy and ICIs alone. The mechanism of action of PD-1/PD-L1 inhibitors involves blocking immunosuppressive signals by obstructing the interaction between PD-L1 ligands on tumor cells and PD-1 receptors on T cells, thereby enhancing T-cell activity. ³¹ As the pathway leading to IMH is consistent for PD-1 and PD-L1 inhibitors, the incidence of liver injury is generally similar. In our investigation, the choice between PD-1 or PD-L1 inhibitors did not affect the incidence of IMH. However, a few earlier studies have suggested a heightened risk of any-grade and grade 3–4 ALT/AST elevation in patients treated with PD-1 inhibitors compared to PD-L1 inhibitors. ³² Hence, deeper exploration is warranted to comprehend the distinctions in IMH incidence associated with PD-1 and PD-L1 inhibitors and underlying mechanisms.

After excluding tumor progression-related liver injury, HCC was identified as one of the independent risk factors for IMH in our cohort. The incidence of IMH in patients with HCC was significantly higher than those with other cancers, such as colorectal or lung cancer, consistent with multiple previous studies.^33–36 A recent large retrospective study also reported a notably higher rate of hepatic events in ICI-treated patients with HCC compared to those with other tumors. ³⁷ This heightened risk may be attributed to the presence of overlapping antigens between normal liver tissue and HCC, increasing the probability of immune attacks on the liver tissue, leading to liver injury. Moreover, our investigation showed no significant difference in the incidence of liver injury between patients with and without liver metastases. This also indirectly suggests that primary HCC plays a more significant role in the development of IMH than tumors from other sources. Therefore, patients with HCC should be vigilant about the factors causing alterations in liver function during ICI treatment. In addition to monitoring tumor progression, it is essential to consider drug usage. This study also highlighted gastric cancer as a significant risk factor for IMH. A recent meta-analysis found considerable differences in the incidence of ICI-related liver injury across various cancer types, which may be attributed to differences in interactions between tumor biology and the immune system, influencing susceptibility to IMH. ²⁰

Of note, this study indicated that patients with liver conditions, including hepatitis B, cirrhosis, and fatty liver, exhibit a higher incidence of IMH, which is consistent with previous reports. ³⁸ Compared with other tumors, a higher proportion of patients with HCC have underlying liver diseases, which may contribute significantly to the increased incidence of IMH in these patients. Recent studies have also shown that patients with cirrhosis experience a higher rate of ICI-related liver injury, but it does not appear to affect their clinical outcome. ⁷ Hence, patients with HCC or other tumors who have underlying liver disease should be evaluated by hepatologists before and during ICI treatment, and vigilance in monitoring liver conditions is strongly recommended during treatment.

Previous studies have shown that hepatocellular injury is more common than cholestatic injury in cases of IMH.^39,40 However, findings from a French clinical trial demonstrated that 60.3% of patients with grade ⩾3 liver injury displayed a cholestatic pattern. ⁴¹ In this study, hepatocellular injury accounted for 49.3% (168/341) of cases, cholestatic injury accounted for 40.2% (137/341), and mixed liver injury accounted for 10.5% (36/341). Notably, no statistically significant difference was observed in the occurrence of severe liver injury among the three different types of IMH. These results indicate that hepatocellular liver injury may be slightly more common in clinical practice, though the difference in their relative frequencies is not pronounced. Importantly, all types of IMH have the potential to progress to severe liver injury. Consequently, in clinical practice, it is essential not to disregard any specific type of IMH.

IMH is predominantly reversible, and even severe liver injury can exhibit a substantial portion exhibiting improvement following GS administration. ³ Currently, several guidelines recommend GS for the treatment of severe IMH. However, there were also conservative views on the use of GS in certain studies, primarily concerning the potential side effects. For instance, a systematic review highlighted that about one-third of patients treated with immunosuppressants (including systemic steroids) for irAEs developed adverse events, and suggested that these complications could impact cancer treatment and patient outcomes. ⁴² The debate among global liver disease experts continues regarding whether GS therapy is essential for grade 2 or higher IMH. Although nearly half of the patients (52.3%) who discontinue ICIs can improve spontaneously without corticosteroid treatment; nevertheless, our results suggest that steroids are beneficial for IMH, especially in grade ⩾3. Patients with grade 3 or above who received GS treatment exhibited a higher liver injury improvement rate compared to those not receiving GS (p = 0.033), indicating the efficacy of GS therapy. In addition, the timing and dosage of corticosteroids need to be addressed. A recent study demonstrated similar outcomes and a lower risk of steroid-induced complications in grade ⩾3 IMH patients who received 1 mg/kg/day methylprednisolone compared to those who received high dose. ⁴³ In another recent study, patients with grade 3 also received lower doses than recommended, and the incidence of infection was low. However, this study suggested that patients with grade 4 should be treated with higher doses of GS than recommended. ⁴⁴ By contrast, our study showed no significant difference in outcomes between patients who received recommended and insufficient doses of GS. Based on these findings, low-dose GS therapy may be a viable therapeutic strategy for IMH patients, as it may improve liver function while potentially minimizing the risk of steroid-related complications.

In summary, this study contributes objective data to the understanding of the clinical safety of ICIs and the management of IMH. However, several limitations exist. As a single-center retrospective study with a limited sample size, potential biases related to different tumor types, treatment regimens, and the presence of liver diseases in patients with HCC cannot be fully excluded. A key limitation is the absence of data from patients treated with CTLA-4 inhibitors. Given the potentially distinct hepatotoxicity profiles of CTLA-4 versus PD-1/PD-L1 inhibitors, our findings primarily apply to PD-1/PD-L1 inhibitor recipients. The generalizability to CTLA-4 inhibitor therapy (monotherapy or combination) is limited. In clinical practice, cautious interpretation of our findings in conjunction with other studies is warranted. Future research should include CTLA-4 inhibitors to address this gap. Furthermore, while the results of this study provide some insights into the overall outcomes of steroid therapy, the limited dataset did not systematically capture the precise timing of liver function improvement or the adverse effects of corticosteroid treatment. This restricted our comprehensive evaluation of its risk–benefit profile. Lastly, the CTCAE grading criteria predate the ICI era and may not accurately reflect IMH severity. It is imperative to explore additional assessment criteria to enhance future diagnostic and therapeutic guidance. Future research should focus on well-designed, prospective, multicenter studies and the establishment of larger, more diverse cohorts, along with the promotion of more detailed and standardized data collection, to enhance the generalizability of findings and provide further evidence regarding the clinical safety of ICI treatment.

Conclusion

This study establishes that during PD-1/PD-L1 inhibitor therapy, certain independent risk factors significantly predispose patients to IMH: female gender, age <45 years, HCC, and gastric cancer. Notably, younger patients are more susceptible to developing severe IMH. While most cases present as mild (grade 1) and reversible, corticosteroid therapy emerges as a critical intervention for IMH, particularly in grade ⩾3 IMH, indicating that these cases require more intensive treatment. Overall, proactive monitoring in high-risk populations, early GS initiation for severe cases, and personalized treatment strategies are essential for improving patient outcomes and minimizing the long-term impact of IMH. These findings offer valuable insights for the clinical management of IMH.