(P 103) LASTING COMPLETE RESPONSE TO PEMBROLIZUMAB IN MISMATCH REPAIR DEFICIENT CARDIAC SARCOMA: A GENOMIC CHARACTERIZATION

Objective: Tumor-infiltrating lymphocytes (TILs), tumor mutation burden (TMB), and PD-L1 expression are the most important predictive markers of response to immune check-point inhibitors (ICIs) in epithelial tumors. Sarcomas are traditionally considered “cold” tumors with poor response to immunotherapy. However, the genetic landscape of sarcoma is markedly heterogeneous, and rare sarcoma cases harbor a high TMB and/or high numbers of TILs. We present the case of a 73-year-old woman with a mismatch repair deficient (MMRd) undifferentiated cardiac sarcoma, who achieved a complete and durable response with the anti-PD1 inhibitor pembrolizumab after the failure of standard treatment consisting of doxorubicin together with olaratumab.

Methods: The tumor resection specimen was routinely processed to establish the histopathologic diagnosis. In addition, MMR proteins (MLH1, PMS2, MSH2, MSH6), PD-L1 (clone SP263) and CD8+ T cells were assessed by immunohistochemistry (IHC). MLH1 gene promoter methylation was evaluated by pyrosequencing. High-throughput sequencing (HTS) was performed on tumor DNA and matched constitutional DNA, using a customized 400-gene panel (full coding sequences, covering a total of 1.24 Mb). TMB and mutational signature were determined based on somatic variants.

Results: A 73-year-old woman presented in September 2016 progressive dyspnea. Magnetic resonance imaging (MRI) revealed a left atrium tumor, which was diagnosed as primary cardiac undifferentiated sarcoma (UDS), in the absence of MDM2 gene amplification. After a first line therapy with doxorubicine and olaratumab, progressive disease developed in May 2018, prompting us to analyze predictive markers in the initial resection specimen. Tumor cells showed a loss of MLH1 and PMS2 expression (with preserved MSH2 and MSH6). They were negative for PD-L1 (tumor proportion score (TPS) < 1%), while a subset of tumor-associated immune cells was positive (combined positive score (CPS) >20). TILs were abundant, with a mean of >20 intratumor CD8+ T cells per high power field (400x magnification). HTS revealed numerous somatic variants (Table 1), accounting for a TMB of 12.9 non-synonymous somatic mutations per Mb. Variants were mostly transitions and small indels in homopolymeric sequences, consistent with an MSI-associated mutational signature. Interestingly, a somatic splice variant was detected at the intron 1 / exon 2 boundary of MLH1 (c.117-3C>G), and was associated with a copy gain of the mutated allele and a loss of heterozygosity (loss of the wild-type allele). In the absence of MLH1 promoter hypermethylation (as determined by pyrosequencing), we attributed the loss of expression of MLH1 by IHC to this splice variant. Other relevant findings of HTS included a missense mutation in TP53, and frameshift deletions in FAT1 and NOTCH2 (Table 1).
Based on these results, the patient started treatment with the anti-PD-L1 antibody pembrolizumab, in May 2018 for 14 cycles in total, and was then discontinued due to side effects. The disease's follow-up by MRI showed that partial response started 6 months after the beginning of pembrolizumab, and complete response was achieved several months after discontinuation and continues to persist at the last MRI in March 2022. (Figure 1)

Conclusion: The genomic analysis of the cardiac UDS displayed several markers known to predict response to ICIs: the tumor harbored a high TMB (> 10 mutations/Mb) and a MMRd/MSI signature. Of note, inactivating mutations of both FAT1 and NOTCH2 have also been associated with ICI therapy response in epithelial tumors. Moreover, the tumor displayed an extensive infiltrate by CD8+ T cells, and PD-L1 was expressed in the tumor immune microenvironment. Based on these biomarkers and given the paucity of other treatment options acceptable to the patient, she was treated with pembrolizumab monotherapy and obtained a durable response.
Although immunotherapy may have a role in some sarcoma patients, the identification of suitable candidates who can obtain a meaningful benefit from treatment requires improvement. Considering the genomic variability between sarcomas and the heterogeneous response to ICIs, validation of new predictive markers is essential to improve the efficacy of immunotherapy in sarcomas. The use of a combination of existing biomarkers may improve their predictive power. This might allow for a more personalized approach and an increase response rate to immunotherapy in diseases that are still considered resistant.