Discussion

Viral hepatitis was excluded serologically, but hepatic metastases were present from diagnosis and acetaminophen exposure was documented during an early episode. Such competing causes are common and, in some series, explain abnormal liver tests more often than true pembrolizumab hepatotoxicity [8]. Even so, the overall pattern favored pembrolizumab-associated liver injury: temporal association with exposure, improvement after interruption and corticosteroids, reproducible recurrence on rechallenge, and prolonged absence of recurrence after switching to nivolumab. Dechallenge and rechallenge patterns remain among the most practical clinical clues when liver biopsy during toxicity is unavailable [4]. Biopsy was further limited by age 82, multiple hepatic metastases, potential bleeding or sampling error, and patient preference against an invasive procedure.

The biochemical pattern was mixed and cholestatic-predominant rather than purely hepatocellular, because alkaline phosphatase rose disproportionately during recurrent episodes. That is compatible with the broader spectrum reported for anti-PD-1 liver injury [2–5]. ASCO and ESMO guidance supports withholding immune checkpoint therapy for grade 2 hepatitis, using corticosteroids when clinically indicated, and permanently discontinuing immune checkpoint therapy for recurrent or severe grade 3–4 toxicity [6, 7]. Nivolumab was therefore an individualized exception after biochemical recovery, informed consent, limited alternatives, and planned close monitoring. Retreatment after checkpoint inhibitor hepatitis can be feasible in selected patients, but published data largely pool tumour types and retreatment strategies, and direct evidence for pembrolizumab-to-nivolumab substitution after repeated pembrolizumab rechallenge failure is sparse [9, 10].

This report does not establish class-wide safety or contradict guideline recommendations for permanent discontinuation after recurrent or grade 3–4 hepatitis. Instead, it supports a narrower conclusion: after careful assessment of competing causes, biochemical recovery, and informed risk–benefit discussion, within-class PD-1 substitution may remain reasonable only as an individualized exception when therapeutic need is high and close laboratory monitoring is feasible [4, 6, 7, 9, 10]. Principal limitations are lack of liver histology during toxicity, which was not pursued because clinical improvement made biopsy unlikely to alter acute management, unresolved classification of recurrence versus second primary, and unclear documented cause of death.