Incidence and severity of immune-related hepatitis after dual checkpoint therapy is linked to younger age independent of herpes virus immunity

Melanoma patients treated with anti-PD-1 & anti-CTLA4 combinational therapy from 01/2016 to 09/2021 at the University Medical Center Freiburg, Dpt. of Dermatology were prospectively included in the discovery cohort (n = 40). A total of 111 patients were identified in clinical records. The remaining (n = 71) patients were retrospectively evaluated in the validation cohort (see Additional file 1: Fig. S2). All included patients had baseline ALT and AST levels below 2xULN and underwent screening for Hepatitis B and Hepatitis C Virus infection. Evaluation of hepatitis was based on ALT, AST and bilirubin evaluations according to CTCAE 5.0. Other adverse events were identified by retrospective evaluation of clinical records. Patients with hepatitis of other etiology were subsequently excluded from the analysis, this affected 1 patient in the discovery cohort was excluded from analysis due to alternative cause of hepatitis (acute HEV infection). 4 patients in the validation cohort were excluded from analysis due to untraceable clinical data and lost to follow-up after therapy initiation. Tumor response was evaluated by radiographic evaluation as per clinical pathways 9–12 weeks from commencement of treatment. Progression (PD) was defined by radiographic disease progression or clinically unequivocal rapid disease progression necessitating cessation of ICB treatment. Tumor regression was determined by radiographic total (CR) or partial (PR) regression of tumor sites. Stable disease (SD) was defined by unchanged radiographic diagnosis. Patients without radiographic evaluation were noted not evaluable (NE). Objective response rate (ORR) was calculated as CR + PR/(total patients-NE); Disease control rate (DCR) was calculated as CR + PR + SD/(total patients–NE). Tumor progression-free survival (PFS) was determined from therapy initiation until the date of tumor progression. Patients that switched therapy before tumor progression were censored at time of therapy switch.

Cells were thawed and counted. 1–2*10E6 cells were used for flow cytometry. Surface staining was performed in a total volume of 50 μl antibody master mix at RT for 15 min and washed twice before acquiring on BD LSR Fortessa. For intracellular staining, cells were permeabilized with FoxP3/Transcription Factor Staining Buffer Set (eBioscience) on ice for 30 min and washed twice with FoxP3 permeabilization buffer (eBioscience), followed by intracellular staining in a total volume of 50 μl antibody master mix on ice for 30 min. Cells were fixed with 2% PFA until measurement. Samples were then acquired and recorded on BD LSRFortessa™. For gating strategies see Additional file 1: Fig. S5.

Statistical analysis was performed with Graphpad version 9.0. As indicated in figure legends, data were analyzed using two-tailed Mann–Whitney test, two-tailed chi-square test, Fisher’s exact test, Kruskal–Wallis test, log-rank survival analysis, receiver-operator characteristic (ROC) analysis or pairwise Pearson correlation.

We first evaluated the incidence of hepatitis and other immune-related adverse events (irAE) in a prospectively recruited discovery cohort of n = 39 patients after initiation of anti-PD-1 & anti-CTLA4 therapy due to advanced melanoma over a period of 100 days. 87.2% of patients (34/39) developed one or more irAE (Fig. 1A). While 28.2% of patients developed a single adverse event, 15.4% developed 2 types of adverse events and 17.9% and 25.6% of patients developed 3 or more types of adverse events. Specifically, we observed a high incidence of hepatitis irAE (59%) in our cohort (Fig. 1B). While 2.6% of patients developed mild hepatitis (grade 1), 20.5% developed moderate hepatitis (grade 2) and 25.6% and 10.3% developed severe grade 3 and 4 hepatitis, respectively, requiring immunosuppressive therapy and treatment pause or discontinuation (Fig. 1C). There was no grade 5 toxicity. In sum, we observed a relatively high rate of hepatitis incidence in the first 100 days after anti-PD-1 & CTLA-4 treatment initiation for advanced melanoma.

We next aimed to understand if development of hepatitis was associated with the development of other irAE, response to treatment, gender or age. Our discovery cohort consisted of patients with an age distribution between 19–73 years, a male dominance (71.8%) and ORR of 61.5% after 3 months, reflecting the real-life setting in our tertiary clinical centre (Additional file 1: Fig. S1). As shown in Fig. 2A, there was no association of hepatitis incidence with the oncologic response after three months (p = 0.7397) nor gender (p > 0.99) (Additional file 1: Table S1). Patients with hepatitis had significantly higher co-incidence of skin, thyroid, gastrointestinal or hypophysal irAE (p = 0.018, Fig. 2B). Development of irAE was significantly associated with age, with patients over 55 years old exhibiting higher incidence for top five frequent irAE compared to younger patients under 55 (Fig. 2C). These data indicate that age might be a determining correlate of the onset of hepatitis and other irAE in our cohort.

We next analyzed the connection between age and hepatitis incidence in more detail. Of note, 85.7% of patients under age 55 had hepatitis during the first 100 days of therapy, while only 27.8% of patients older than 55 years had hepatitis during our observation period (Fig. 3A). The significantly higher incidence of hepatitis in younger patients was also connected to hepatitis-free survival, an earlier onset and higher severity of hepatitis in younger patients (Fig. 3B, C) in the discovery cohort.

To exclude the possibility that patient characteristics other than age might have dictated the association with irAE in our discovery cohort, we sought to validate these findings in a retrospective analysis of all patients treated at our tertiary center during the recruitment period to address recruitment bias. The validation cohort consisted of n = 67 patients (see methods section for inclusion/exclusion criteria and Additional file 1: Fig. S2). We noted several differences in the composition of the validation cohort, namely reduced incidence of irAE, including hepatitis (Additional file 1: Table S2), potentially connected to a shift towards older age (median age 54 years vs. 60 years in discovery and validation cohort, respectively), reduced ORR (61.5% vs. 28.6%, respectively) while gender distribution was similar (Additional file 1: Table S2). Importantly, however, we also observed a significant association of younger age with hepatitis-free survival, severity and earlier onset in the validation cohort (Fig. 4). Together, these data support age as a validated risk factor for the development of hepatitis in the first 100 days after anti-PD-1 & anti-CTLA4 therapy.

Interestingly, patients who developed hepatitis within 100 days from therapy initiation had small however significantly higher liver transaminase levels at baseline (Additional file 1: Fig. S3A). This association was observed despite the majority of patients (n = 92, 86.8%) having transaminase levels within the normal range. ROC analysis showed a slight predictive role of both baseline AST and ALT values for ICB-induced hepatitis (Additional file 1: Fig. S3B). Our study also included few patients (n = 14, 13.2%) that had elevated liver transaminases already at baseline (up to 2xULN). However, these few patients with elevated liver transaminases (above the ULN) at baseline did not show a significantly higher incidence of developing further hepatitis during therapy in comparison to those with normal liver transaminase levels (Additional file 1: Fig. S3C). In sum, these data suggest that preexisting mild liver inflammation can be associated with the onset of ICB hepatitis.

The immune system is exposed to multiple antigens over time and immunological changes associated with herpes virus infections are connected to age [16]. A previous report described baseline antiviral T cell immunity to herpes virus infections as a potential driver of hepatic irAE [17]. We wondered if this association would explain the age-associated differences in the incidence of hepatitis in our cohort, since higher hepatitis incidence was observed in younger patients who would be predicted to have lower immune memory to herpes virus infections. Herpes virus serology for CMV, EBV and HSV was determined and analyzed with respect to hepatitis incidence. However, we did not observe an association of herpes virus serology with hepatitis-free survival in our observation and validation cohorts (Additional file 1: Fig. S4). Interestingly, in pooled analysis of both cohorts, positive serology for CMV at baseline showed a non-significant trend (p = 0.0767) towards lower hepatitis-free survival (Fig. 5A). CMV status was not connected to baseline liver transaminases (Additional file 1: Fig. S3D). Since the effect of CMV positivity on hepatitis incidence was not significant, we wondered if it might be masked by different age groups. Subgroup analysis according to CMV serostatus positivity and age (cutoff 55 years) showed however again only non-significant trends towards lower hepatitis free incidence (Fig. 5B). In sum, we did not observe a significant difference in hepatitis-free survival connected to herpes-virus immunity in our cohorts.

Preexisting T cell memory is discussed to underlie immune-mediated toxicity after checkpoint therapy and would fit to a model where the immune checkpoints targeted contribute to attenuation of autoimmunity in a physiological setting. We therefore performed a detailed analysis of CD8 and CD4 T cell subsets in the discovery cohort and analyzed potential associations of these immune populations (Fig. 6, Additional file 1: Figs. 5 and 6) with age and the incidence of hepatitis. As expected [16], we observed a reduction of naïve CD8+ T cells in older patients while memory populations expanded (Additional file 1: Figs. S6A and SB). However, we observed no difference between these age-associated T cell populations and viral hepatitis (Additional file 1: Fig. S6C). An association of an expanded CD4+ effector memory T cell (TEM) population with ICB-hepatitis was reported earlier [17]. However, we did not observe different frequencies of CD4 + TEM or CD8+ TEMRA cells in patients who subsequently developed hepatitis after anti-PD-1 & anti-CTLA-4 therapy (Fig. 6A), and there was no difference in the T cell subset distribution between patients which did or did not develop hepatitis according to CMV serology status (Fig. 6B). These data indicate that differences in baseline T cell differentiation subsets are not directly linked to the onset of hepatitis after anti-PD-1 & anti-CTLA-4 checkpoint therapy.

We next tested other possible clinical, oncological (BRAF/NRAS status, stage, LDH levels, presence of metastasis), virological or hepatological characteristics of our patient cohorts that might contribute to ICB-hepatitis during dural checkpoint therapy in the pooled cohort (Additional file 1: Table S3). In addition to age, AST and ALT levels, interestingly, this analysis also indicated a significant association of anti-HBs positivity with ICB-hepatitis, while anti-HBc or HBsAg were not significantly associated, in line with a status post HBV vaccination. We interpreted this finding as an age-dependent cohort effect due to wider introduction of HBV vaccination in younger patients. Since age was the most significant variable associated with ICB-hepatitis, we next tested if age could be used as a predictor of the development of ICB-Hepatitis using receiver operating characteristic (ROC) curve analysis. Indeed, age was a relatively reliable discriminator of patients who developed or escaped therapy-induced hepatitis (n = 106, AUROC = 0.7455, p < 0.0001) (Fig. 7A). Specifically, for an age cut-off of 55 years, ROC analysis indicated specificity of 66.67% and sensitivity of 75%respectively in the pooled analysis). We did not observe a similar predictive role for herpes virus serostatus (data not shown). Interestingly, this predictive function of age was not observed with respect to tumor response in the comparable time frame (Fig. 7B). Moreover, presence of any IRAE (including non-hepatitis irAE) was also not associated with improved survival in our cohort (Additional file 1: Fig. S7). This data suggests that the immunological mechanisms behind successful anti-tumor responses and hepatic adverse events are not necessarily connected. In sum, our data highlights age as a predictor of ICB-hepatitis.