Checkpoint inhibitor induced hepatitis and the relation with liver metastasis and outcome in advanced melanoma patients

The introduction of immune checkpoint inhibitors has significantly improved the 5-year survival of patients with advanced melanoma [1]. The checkpoint inhibitors registered for the treatment of advanced melanoma are the programmed cell death protein 1 (PD-1) inhibitors nivolumab and pembrolizumab, the cytotoxic T-lymphocyte associated 4 (CTLA-4) inhibitor ipilimumab and the combination of ipilimumab and nivolumab. The use of immune checkpoint inhibitors can lead to an array of immune-related adverse events (IRAE), including checkpoint inhibitor-induced hepatitis.

Based on data from clinical trials, the incidence of grade 3–4 checkpoint inhibitor-induced hepatitis is 1–17%, depending on the checkpoint inhibitor used [2‐9]. For ipilimumab the reported incidences are between 2–9% [3, 4, 8], for PD-1 inhibitors incidences between 1 and 4% have been reported [7, 8] and for combination therapy the incidence varies between 8–17% [6, 8, 9]. Only small studies have described characteristics of checkpoint inhibitor-induced hepatitis and no risk factors have been reported [2, 10].

Severity of hepatitis can be graded based on the alanine aminotransferase (ALT) and aspartate aminotransferase (AST) level. Treatment can be continued with grade 1 (ALT or AST 1–3 times elevated). With grade 2 (ALT or AST 3–5 times elevated) the next cycle should be delayed and prednisolone started if transaminases continue to rise. Grade 3 (ALT or AST 5–20 times elevated) and grade 4 (ALT or AST > 20 times elevated) should be treated with 1–2 mg/kg prednisolone per day according to guidelines [11, 12]. If the response is insufficient within 2–3 days, mycophenolate mofetil and/or tacrolimus can be added. Typically, checkpoint inhibitor-induced hepatitis responds well to treatment with corticosteroids although tapering of prednisolone can take 6–8 weeks [13].

Incidence of IRAEs differs between different tumors types and can be correlated to treatment response. Development of vitiligo is much more frequent in melanoma patients than in patients with other solid tumors [14‐16]. Melanoma patients with vitiligo have a better overall survival compared to patients without vitiligo [17]. Lung cancer patients have a higher risk of developing pneumonitis as IRAE [18]. However, this did not seem to influence the progression-free or overall survival [19]. Based on these findings, we hypothesized that the presence of liver metastasis could be a possible risk factor for checkpoint inhibitor-induced hepatitis.

A correlation between the occurrence of hepatitis and outcome is currently unknown. Stopping of immune checkpoint inhibitor therapy and the start of immunosuppressive treatment might decrease survival. Patients with liver metastasis have a lower rate of response compared to patients with visceral disease not involving the liver [20]. All these factors could influence survival in patients with hepatitis.

The aim of this study was to assess if checkpoint inhibitor-induced hepatitis is related to liver metastasis and outcome. The secondary aim was to assess the incidence and current treatment of patients with checkpoint inhibitor-induced grade 3–4 hepatitis in a unique real-world, nationwide prospective registry.

All advanced melanoma patients in the Netherlands are registered in the Dutch Melanoma Treatment Registry (DMTR). To ensure safety and quality of care, all data regarding the type of melanoma, given treatment, incidence of grade 3–4 adverse events and treatment outcome are registered prospectively [21]. Treatment with immune checkpoint inhibition including PD-1 inhibitors (nivolumab or pembrolizumab), CTLA-4 inhibitor ipilimumab and combination therapy with nivolumab and ipilimumab or targeted therapy with serine/threonine-protein kinase B-Raf (BRAF) inhibitors (vemurafenib, dabrafenib and encorafenib) and mitogen-activated protein kinase enzymes (MEK) inhibitors (trametinib, cobimetinib and binimetinib) are registered in treatment episodes. A treatment episode starts when a treatment is started and ends when a patient dies or a different treatment is initiated. If patients are switched to a different therapy, multiple treatment episodes are available for these patients. This registry has a nationwide coverage because registration is mandatory for reimbursement and all systemic therapy for melanoma is given in the 14 designated melanoma treatment centers. Data from July 2012 until July 2013 were retrospectively entered. From July 2013 until January 2019 data is prospectively entered by trained data managers and checked by treating physicians. In compliance with Dutch regulations, the medical ethical committee of the Leiden University Medical Centre judged that the DMTR was not subject to the Medical Research Involving Human Subjects Act.

All cutaneous melanoma patients registered in the DMTR database who received at least one cycle with PD-1 inhibitor, ipilimumab or combination therapy of nivolumab and ipilimumab between May 2012 and January 2019 were eligible for inclusion. Patients with missing liver toxicity data, a follow-up of less than 4 weeks and patients with uveal melanoma were excluded.

The different treatment regimens (PD-1 inhibitor, ipilimumab and combination therapy) were analysed separately. Treatment was administered in 2-week cycles for nivolumab and 3-week cycles for pembrolizumab, ipilimumab and combination therapy. Ipilimumab treatment was stopped after four cycles. Patients with combination therapy continued with nivolumab after four cycles of combination therapy. If patients were treated with different checkpoint inhibitors in different treatment episodes, all treatment episodes were included for analysis. If patients were treated in multiple episodes with the same checkpoint inhibitor, only the first treatment episode was included. Previous treatment episodes with targeted or immunotherapy were analysed as a risk factor for hepatitis.

Checkpoint inhibitor-induced hepatitis was defined as grade 3 (ALT or AST 5–20 times the upper limit of normal) or grade 4 (ALT or AST > 20 times the upper limit of normal) according to the CTCAE, version 4.0. Lactate dehydrogenase (LDH) was measured before the start of treatment with an immune checkpoint inhibitor. Elevated LDH was defined as LDH > 250 U/L. Presence of liver metastasis was based on a CT scan or a fluorodeoxyglucose (FDG) PET CT scan performed within 3 months of the start of treatment.

2749 patients were treated with checkpoint inhibitors for advanced melanoma in the Netherlands between May 2012 and January 2019. In total 2561 patients with 3111 treatment episodes met the inclusion criteria. Treatment with PD-1 inhibitors was given in 1620 patients, ipilimumab in 1105 patients and combination therapy in 386 patients (Fig. 1). Multiple treatment episodes were registered in 550 patients.

In the largest cohort reported thus far, we observed no relation between checkpoint inhibitor-induced hepatitis and the presence of liver metastasis. The incidences of PD-1 inhibitor and ipilimumab induced grade 3–4 hepatitis were 1.8% and 2.6%, respectively, which is comparable to previously published incidences [3, 4, 7, 8]. The incidence of combination therapy-induced hepatitis was 20.7%, which is higher than previously published frequencies of 9–17% [6, 8, 9].

The use of immune checkpoint inhibitor therapy is rapidly increasing as these drugs will be approved for more indications. Immune checkpoint inhibition has currently been approved in the Netherlands for advanced stages of melanoma, lung cancer, renal cell carcinoma, bladder carcinoma, squamous cell carcinoma of head and neck and Hodgkin lymphoma [1, 22‐26]. Furthermore, immunotherapy with checkpoint inhibitors has recently been approved as adjuvant treatment in stage III melanoma and is under investigation in many tumor types as adjuvant and neoadjuvant treatment [27, 28].

As an increase in checkpoint inhibitor-induced hepatitis can be expected, it is important to identify risk factors. Presence of liver metastasis was not a risk factor for hepatitis in patients treated with checkpoint inhibitors. Similarly, in patients with hepatocellular carcinoma the incidence of grade ≥ 3 hepatitis was found to be 2% for PD-1 inhibitors and 20% for combination therapy [29]. These frequencies are comparable to our findings and also argue against an association between an affected liver and immune checkpoint inhibitor-induced hepatitis.

For PD-1 inhibitor-induced hepatitis, elevated LDH before treatment was related to an increased risk. A clear explanation of this finding is difficult. Elevated LDH is related to total tumor load and is highly prognostic for a poor progression-free and overall survival. In addition, LDH is an independent negative predictor for therapy response in PD-1 inhibitor and ipilimumab monotherapy [30, 31]. More research is needed to clarify the relation between LDH and PD-1 inhibitor-induced hepatitis.

For ipilimumab-induced hepatitis, previous treatment with immunotherapy, mainly PD-1 inhibitor monotherapy, increased the risk of hepatitis. The effect of PD-1 inhibition might partially continue after stopping the treatment. Treatment with ipilimumab after PD-1 inhibition could then result in combined inhibition of PD-1 and CTLA-4 and increase the risk of hepatitis.

For combination therapy-induced hepatitis, younger age was a risk factor in the current study. Younger patients may have a more active immune system which can lead to hepatitis or other IRAEs [32]. A recent study showed no increased rate of grade 3–4 side effects in general in younger patients compared to older adults [33]. However, for combination therapy-induced hepatitis it does seem a relevant risk factor.

Checkpoint inhibitor-induced hepatitis occurred after median 6 weeks in ipilimumab and combination therapy and after median 12 weeks for PD-1 inhibitors. This is comparable to the previously reported 6–7 weeks in ipilimumab and combination therapy and 14 weeks in PD-1 inhibitors [11, 34]. As the ranges are wide, physicians should be aware that hepatitis can still occur in the later stages of treatment and even after treatment has stopped. Checkpoint inhibitor-induced hepatitis was treated with corticosteroids in all patients except one. Second-line immunosuppression was necessary in 20% of the patients. The type of second-line therapy was not registered, but we can assume that this was mostly mycophenolate mofetil and/or tacrolimus since these drugs are added to corticosteroid treatment in case of insufficient response according to the ESMO guidelines [11]. The rate of 20% need for second-line immunosuppressive treatment is lower than the 45% (9 of 20 patients) previously reported [10].

If a new targeted or immunotherapy was started after hepatitis was resolved, recurrence of hepatitis was rare. The decision to start a new treatment is dependent on many factors including the antitumor response.

IRAEs are a sign of activated immunity which may promote an antitumor response. A recent meta-analysis showed a better survival for patients with IRAEs compared to patients without IRAEs [35]. Immunosuppressive therapy, especially anti-TNF therapy, may be associated with decreased overall survival [36]. Although treatment with immune checkpoint inhibitors was stopped earlier due to hepatitis and immunosuppressive treatment was given, checkpoint inhibitor-induced hepatitis had no negative or positive effect on progression-free survival and overall survival.

A strength of this study is the prospective nationwide coverage of all patients with advanced melanoma who were treated with immune checkpoint inhibitors. This resulted in the largest cohort of checkpoint inhibitor-induced hepatitis to our knowledge. Due to nationwide registration a reliable estimate of incidences, current treatment of checkpoint inhibitor-induced hepatitis and outcomes can be given in a real-world cohort. In addition, the size of the cohort allowed us to analyze the different types of checkpoint inhibitor-induced hepatitis separately, which has not been possible in other smaller studies up till now. A weakness of the study was that some liver-specific variables such as AST, ALT, alkaline phosphatase (AP) and gamma-glutamyltransferase (GGT), liver biopsies results and type of second-line immunosuppression were not registered.

Our study shows that the incidence of grade 3–4 hepatitis in a real-world cohort is 1.7% for PD-1 inhibitor treatment, 2.6% for ipilimumab treatment and 20.7% for combination therapy. Hepatitis was not related to liver metastasis and had no negative effect on survival. As a rise in the number of patients with checkpoint inhibitor-induced hepatitis can be expected, it is important that oncologists and hepatologists have knowledge of this IRAE. Further prospective research and evidence-based treatment guidelines are important to optimize treatment strategies for checkpoint inhibitor-induced hepatitis.