Hyperprogressive Disease After Combined Anti-PD-L1 and Anti-CTLA-4 Immunotherapy for MSI-H/dMMR Gastric Cancer: A Case Report

Immune checkpoint inhibitors (ICI) have been developed in gastric adenocarcinomas and approved in first-line metastatic setting (in combination with chemotherapy) as well as in pretreated patients. Microsatellite instability-high (MSI-H) tumors are predicted to derive high benefit from ICI but data in gastric locations are limited. Here, we describe the case of a 68-year old patient with stage IV MSI-H gastric adenocarcinoma, referred to our center to receive immunotherapy after failure of standard of care (surgery with perioperative platin-based chemotherapy and paclitaxel plus ramucirumab at disease progression). The patient received one injection of durvalumab and tremelimumab and was hospitalized eighteen days after because of occlusive syndrome. The CT scan showed hyperprogression of the lymph nodes and hepatic lesions, compressing the gastric stump. He died few days later. Molecular analyses did not explain this outcome. To our knowledge, this is one of the first reported cases of hyperprogressive disease after combined ICI for a patient with MSI-H tumor. We review the potential causes and discuss the emerging literature regarding predictive factors of hyperprogression in the particular subset of MSI-H patients. If some data were available in retrospective studies, validation of strong predictive factors is needed to avoid such dramatic evolutions.


INTRODUCTION
Despite progresses in prevention and screening, gastric adenocarcinoma (GA) remains the third cause of cancerrelated mortality worldwide (1). Immune checkpoint inhibitors (ICI) have been developed and evaluated in several settings in GA. The benefit of nivolumab and pembrolizumab, two PD-1 inhibitors, was first shown over placebo in pretreated patients in the ATTRACTION-2 and KEYNOTE-061 trials (2,3). In the first-line metastatic setting, addition of nivolumab to chemotherapy recently improved overall survival (14.4 versus 11.1 months) for PD-L1-positive [combined positive score (CPS) ≥ 5] advanced gastric and esophageal adenocarcinomas (4).
Somatic genomic analysis of large series of gastric cancers identified distinct molecular subtypes with their own prognosis and therapeutic targets. The Cancer Genome Atlas classified 22% of gastric cancers as microsatellite instability-high (MSI-H) tumors, with deficient mismatch repair (dMMR) and high mutational burden (TMB) (5). As for colorectal adenocarcinoma, MSI-H tumors seem to have a good prognosis, and are more frequent in localized gastric cancers (6 to 20%) (6-10) than in advanced disease (2.5-3%) (10,11).
The benefit of ICI for MSI-H tumors was first provided by a small phase 2 trial evaluating pembrolizumab in refractory tumors with or without MMR deficiency (12). Efficacy was shown in both colorectal and non-colorectal MSI-H/dMMR tumors including one MSI-H gastric cancer. The KEYNOTE-158 study further confirmed the benefit of pembrolizumab for various non-colorectal MSI-H/dMMR tumors, with a 46% response rate and a 11 months median progression-free survival for the gastric adenocarcinoma subgroup (13). Later, an exploratory analysis of the KEYNOTE-062 study showed that pembrolizumab improved overall survival compared to first-line chemotherapy in the small MSI-H subgroup (50/763 patients) whereas it was only non-inferior for the overall population of PD-L1-positive (CPS ≥ 1) gastric adenocarcinomas (14,15). A recently published meta-analysis of randomized clinical trials confirmed the predictive role of microsatellite instability for PD-1 blockade efficacy (16). However, the number of MSI-H GA treated with ICI is currently low (all available data in MSI-H GA are summarized in Table 1).
Patterns of response and progression during immunotherapy may differ from what is observed with chemotherapy (17). A dramatic worsening of disease progression, known as hyperprogressive disease (HPD) has been described in a subset of patients treated with immunotherapy, especially for head and neck squamous cell (18) and non-small cell lung cancers (19), but data are lacking about HPD in MSI-H tumors for which only a few cases have been reported (20,21).
We described the case of a MSI-H/dMMR gastric adenocarcinoma patient with HPD while receiving ICI, and reviewed the potential causes and predictive factors of hyperprogression.

CASE PRESENTATION
We report the case of a 68-year old Caucasian patient with a metastatic MSI-H/dMMR gastric adenocarcinoma. This patient with no significant past medical history was treated four years ago for localized antral gastric cancer. He benefited from a subtotal gastrectomy with perioperative FOLFOX chemotherapy (5 cycles received). Pathological examination of surgical specimen revealed poorly differentiated GA, classified ypT3N1 (2N+/23), HER2-negative [immunohistochemistry (IHC) 2+, fluorescence in-situ hybridization (FISH) negative], Epstein-Barr Virus ambiguous status (positive on the first sample but not confirmed on the second sample), Helicobacter pylorinegative, with MMR-deficiency and microsatellite-instability (loss of expression of MLH1 and PMS2, caused by MLH1 promoter hypermethylation).
The patient experienced early recurrence in retroperitoneal and intraperitoneal lymph nodes after ending the perioperative chemotherapy. He received paclitaxel and ramucirumab as second-line chemotherapy, with a near-complete tumor response, followed by radiochemotherapy on residual disease.
After two years of follow-up, he presented another locoregional lymph node recurrence and was treated again with paclitaxel and ramucirumab until progression (apparition of hepatic lesions and increase of lymph nodes) occurring 7 months later.
Due to the dMMR/MSI-H status and lack of immunotherapy approval in France for non-colorectal MSI-H cancers, he was referred to our comprehensive cancer center. His general condition was good (ECOG-PS 1) but he presented a 11% weight loss during the previous 6 months. Blood analysis showed mild perturbations of hepatic function, twofold elevation of LDH and elevated neutrophil-to-lymphocyte ratio (5.7) ( Table 2). After discussion in multidisciplinary tumor board and written consent, the patient was enrolled in the immunotherapy cohort of the "MOST plus" phase II trial The patient was hospitalized in emergency eighteen days after the first injection because of intestinal obstruction and impaired general condition. The CT-scan showed significant disease progression of lymph nodes and hepatic lesions, compressing the gastric stump ( Figure 1). Laboratories exams showed grade 3 cytolysis, anicteric cholestasis, rising of neutrophil-tolymphocyte ratio (to 10.1) and LDH and elevation of tumor marker ACE to 20.9 ng/mL (normal before treatment) ( Table 2). The situation worsened rapidly with liver impairment evolving to hepatic encephalopathy and death twelve days later.
Hyperprogression was confirmed by an external review of an expert radiologist calculating the Tumor Growth Kinetics ratio (TGKR) to 7 ( Figure 1).
After this unexpected outcome, we decided to repeat molecular analyses on a liver metastasis biopsy performed a few days before starting immunotherapy.

Hyperprogression Definition
Evaluation of therapeutic response to immunotherapy can be more challenging than with conventional cytotoxic therapy, given the different patterns of tumor response. Pseudoprogressions (PSPD)with initial flare-up followed by prolonged responsesmust be differentiated from hyperprogressions in which the tumor increase continues (22). Several radiologic criteria were developed to define HPD (19). Le

Relevant Past Medical History and Interventions
68 years old Caucasian patient without significant comorbidity. Treated four years ago by subtotal gastrectomy with perioperative FOLFOX chemotherapy for a localized gastric cancer (ypT3N1). Histology: poorly differentiated adenocarcinoma, HER2-negative, EBV-ambiguous, dMMR/MSI-H, PD-L1 < 1%). RNAseq analysis: deleterious TP53 mutation. Retroperitoneal lymph node recurrence treated with paclitaxel ramucirumab followed by radiochemotherapy. Second lymph node recurrence treated again with paclitaxel ramucirumab until progression (apparition of hepatic lesions). Enrollment in the "MOST plus" trial to receive durvalumab tremelimumab combination.

Hyperprogression in Gastric Cancers: Epidemiology and Predictive Factors
Several clinical factors were found to be associated with the likelihood of HPD among various tumors. These include advanced age, cancer recurrence, prior radiation therapy, increased metastatic burden and liver metastasis (25,26).
Hyperprogressive disease for MSI-H/dMMR gastric cancer has never been reported. A few cases have been reported for gastric cancers (27-31) but not specifically for MSI-H tumors. Retrospective cohort studies reported hyperprogressive diseases rates of 10% to 29% in common gastric cancers treated with nivolumab (32)(33)(34)(35). Impaired general condition and liver metastasis were associated with HPD in two of these studies (33,35). A large sum of target lesion diameters at baseline (35) and a PD-L1 CPS score < 10 (33) were reported as potential predictive factors of hyperprogression. These studies also confirmed the better prognosis and higher response rate to immunotherapy for MSI-H gastric cancers. HPD rates were about 12-14% in MSI-H tumors and 23% in proficient-MMR tumors, but there was no statistical association in these small samples (only four MSI-H hyperprogressors) (33,35).
While having a confirmed MSI-H GA, the tumor in our case was PD-L1-negative, maybe explaining the lack of ICI efficacy. However, MSI-H tumors are associated with high numbers of infiltrating lymphocytes which can mediate antitumor response, even in PD-L1-negative tumors (36,37). Another predictive biomarker for ICI efficacy is tumor mutational burden (TMB): pembrolizumab recently obtained an FDA approval for TMB-H cancer treatment following the results of the KEYNOTE-158 study (38). However, no gastric cancer was included in this study and the TMB evaluation is not yet standardized. An interesting report showed substantial overlap between MSI and TMB-high tumors in 63 gastric cancer patients treated with ICI with only one non-responder MSI-H GA patient, bearing a PD-L1-positive but low-TMB tumor (39).
Several scores were developed to predict ICI efficacy. The Lung Immune Prognostic Index (LIPI) is a simple tool which stratifies patients in "poor", "intermediate" and "good" prognostic groups according to pre-treatment LDH (one point if greater than the upper limit of normal, defined according the limit of local laboratory) and derived neutrophil-to-lymphocyte ratio (dNLR = absolute neutrophil count/[white blood cell concentration − absolute neutrophil count]; one point if ≥ 3) (40). This score was initially developed for advanced non-small cell lung cancers and further validated in renal cell carcinoma, melanoma and gastric cancer (41,42). A multicenter retrospective analysis of patients with metastatic MSI-H/ dMMR tumors treated with ICI showed that "poor" LIPI score was significantly associated with shorter survival and higher rate FIGURE 1 | CT scan evaluations before and after treatment. CT scans performed 18 weeks before ICI treatment (first column), at baseline (second column) and for early evaluation eighteen days after the beginning of ICI treatment (third column) show the changes in lymph nodes and hepatic lesions: the left para-aortic lymph node (red arrow, line A) increased from 30mm at baseline to 40mm on day 18, known hepatic lesion increased (red arrow, line B), multiple new hepatic lesions appeared and locoregional invasion led to gastric stump compression. of fast-progression (defined by ≤ 3 months overall survival) (43). In our case, the pre-treatment LIPI score was "intermediate".
The LIPI score could be useful to identify patients at high-risk of fastprogression but prospective validation with recognized hyperprogression criteria is needed.

Causes of Hyperprogression
Pathological mechanisms for hyperprogressive disease are being actively investigated. Analysis of hyperprogressive gastric cancers showed that ICI could increase Treg cell infiltration and therefore enhance their immunosuppressive abilities (34). Infiltration and activation of M2 macrophages has also been associated with hyperprogressive disease (44). However, these results were only partially reproducible in Yamaguchi et al. case report of a patient with PD-L1 negative metastatic gastric cancer who presented hyperprogressive disease after nivolumab third line therapy: tissue sample analysis of the hyperprogressive lymph node showed PD-L1-positive macrophage increase, but also Treg decrease which was unexpected (45). In our case, we were not able to perform a liver biopsy to evaluate immune infiltration after failure of ICI because the patient's general condition worsened too quickly. Genomic analysis of hyperprogressive tumors identified an overrepresentation of EGFR alterations and MDM2/MDM4 amplifications among them (46). In our case, none of these molecular alterations was identified.

CONCLUSION
Microsatellite instability is a strong predictive factor for tumor response, regardless of tumor site of origin. However, this case reminds us that MSI-H status does not guarantee a response to ICI: even in this highly-selected population, disease progression and even hyperprogression can be observed. Despite recent efforts, the prevalence, mechanisms, and predictive factors for HPD remain unclear.
We report here the first case of HPD in a PD-L1-negative MSI-H gastric cancer patient treated with an ICI combination and describe in detail its clinical, biological and molecular characteristics. Combining MMR status with TMB and PD-L1 analyses may be of great interest to identify ICI-responders.
Validation of a combined score or other predictive factors is needed to avoid such dramatic evolutions with ICI.

DATA AVAILABILITY STATEMENT
The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found below: https://www.ncbi. nlm.nih.gov/bioproject/PRJNA755252.

ETHICS STATEMENT
Written informed consent was obtained from the individual(s) for the publication of any potentially identifiable images or data included in this article.

AUTHOR CONTRIBUTIONS
RV and AF collected data. TG and CM reviewed CT scans and calculated the TGK ratio. AB performed the molecular analyzes. RV and CF wrote the initial manuscript. All authors contributed to the article and approved the submitted version.