The efficacy of pembrolizumab for MSI-H solid tumors was reported in KEYNOTE158 [4]. The response rate in all carcinomas was 34.4%, including 9.9% complete responses; in 22 cases of biliary tract cancer, the response rate reached 40.9%, demonstrating efficacy. In our case of intrahepatic cholangiocarcinoma, MSI-H was also detected, and pembrolizumab was administered, showing a marked long-term effect.

The KEYNOTE-158 study demonstrated that TMB-H could be a useful predictive biomarker for response to pembrolizumab in solid tumors [16]. Among the 790 evaluable patients, those with TMB-H status (≥ 10 mutations per megabase) showed a significantly higher objective response rate of 29% (95% CI: 21–39) compared to 6% (95% CI: 5–8) in non-TMB-H patients. The durability of response was also notable, with the median duration of response not reached in the TMB-H group versus 33.1 months in the non-TMB-H group. Importantly, this predictive value of TMB-H was observed independent of PD-L1 expression and was not driven by outcomes from any particular tumor type. These results suggest that TMB-H status could be a useful pan-tumor biomarker to identify patients who may benefit from pembrolizumab monotherapy across various solid tumor types.

In this case, although there was no past or family history of colorectal cancer, the MSI was high, and the variant allele frequency of the MLH1 mutation was 0.49; therefore, considering the possibility of Lynch syndrome was necessary. Hence, both MLH1 and PMS2 were negative when immunostaining for mismatch repair-related gene proteins was performed. A germline test was performed to make a definitive diagnosis of Lynch syndrome based on immunostaining results. Germline examination did not reveal any abnormalities in any of the mismatch repair-related genes, suggesting sporadic MSI-H intrahepatic cholangiocarcinoma. TMB-H was also detected simultaneously.

MSI-H is rare in biliary tract cancer, reported in approximately 1.7–3.1% of cases [5,6,7]. The clinical features of MSI-H biliary tract cancer have not been reported in large numbers, and the clinicopathological or radiological features remain unclear [17]. To date, case reports of intrahepatic bile ducts diagnosed as MSI-H and treated with PD-1 antibody (Table 2) showed no consistent trends with respect to age, sex, or disease stage [8,9,10,11,12,13]. As for tumor markers, CA19-9 was substantially high in Toshida’s report; however, in Kai’s and Ikeda’s reports, there were cases of mild elevation, which was unclear in some cases, making detailed examination difficult (Table 2). The therapeutic effect was successful in all cases and the progression-free survival was prolonged.

To date, MSI-H, TMB-H, and PD-L1 have been studied as predictors of the therapeutic efficacy of immune checkpoint inhibitors. Based on the KEYNOTE158 trial results, MSI-H has been demonstrated to be a marker of reasonable clinical promise [4,16]; however, PD-L1 expression has not necessarily been reported to be associated with efficacy. PD-L1 is not a predictor of the therapeutic efficacy of durvalumab, which was recently approved for the treatment of biliary tract cancer [3]. In our and Nakamura’s cases, a partial response was obtained with pembrolizumab treatment, despite low PD-L1. The importance of PD-L1 immunostaining may differ depending on the organs involved.

The KEYNOTE177 trial reported that pembrolizumab was less effective in patients with MSI-H colorectal cancer who had KRAS mutations, a strong driver gene, than in those without mutations [18]. Among the case reports, this and Toshida’s cases are the only ones in which comprehensive cancer genomic profiling was performed along with MSI testing [12]. Both patients showed partial responses to pembrolizumab, despite being KRAS mutation-positive. Although the number of cases was small, it was assumed that MSI-H intrahepatic cholangiocarcinoma may have a good therapeutic effect, even with KRAS mutations.

In this case, MLH1 protein deficiency was identified as the cause of MSI-H, which was consistent with the CGP genomic data showing MLH1 pathogenic mutations. Germline testing is recommended for MLH1 variants as presumed germline pathogenic variants19,20,21,22,23,24. Germline testing revealed no aberrant variants, indicating the absence of Lynch syndrome. If Lynch syndrome is suspected due to family or medical history, or if a gene mutation for which germline testing is recommended is detected in CGP testing, germline testing should be actively considered.