Prevalence of mismatch repair genes mutations and clinical activity of PD-1 therapy in Chinese prostate cancer patients

Prostate cancer (PCa) patients with mismatch repair (MMR) genes mutations are potentially responsive to immune checkpoint blockade (ICB). However, aberrations in MMR genes were rare in PCa and there is evidence that MMR genes mutations are highly ethnic specific. Thus, the prevalence and clinical characteristics of this subgroup in Chinese PCa patients are largely unknown. Furthermore, why some of these patients do not respond to ICB also remains unclear. Here, we analyzed the sequencing data from 3338 Chinese PCa patients to profile the mutation spectrum of the MMR genes. We found that in metastatic disease, the pathogenic mutation frequency of MMR genes in Chinese PCa patients was higher than that in the Caucasus population (4.8 vs 2.2%, P = 0.006) and the mutation carriers responded poorer to androgen deprive therapy (ADT) and abiraterone than non-carriers. Besides, we reported a multi-institutional cases series of 11 PCa patients with mismatch repair deficiency (dMMR) or microsatellite instability high (MSI-H) who received programmed cell death receptor-1 (PD-1) inhibitors, and performed multiplex immunohistochemistry (mIF) to explore the relationship between tumor immune microenvironment (TIME) and response to ICB. The results showed that the responders had higher density of intratumoral CD8+ T cells than non-responders. Our data suggested MMR genes mutations may be more common in Chinese PCa patients, and it is associated with poorer response to hormonal therapies. We propose that the density of intratumoral CD8+ T cells could be a promising predictor to help further subdivide the population of PCa patients who can benefit from immunotherapy.


Introduction
Immunotherapy has become an important cancer treatment modality. Cancer immunotherapy includes immune checkpoint blockade (ICB), cancer vaccines, and adoptive cell therapy. Among these, ICB, which enhances the function of anti-tumor T lymphocytes, has achieved unprecedented Bangwei Fang, Yu Wei, Hao Zeng, Yonghong Li, Shouzhen Chen are these authors contributed equally to this work. 1 3 clinical benefits in many solid tumors [1]. However, several studies showed that the efficacy of ICB is limited for advanced PCa patients [2][3][4][5]. The mechanisms underlying the poor response to ICB in PCa are not fully understood. It has been speculated that this may be related to the low expression of programmed cell death ligand-1 (PD-L1), low tumor mutation burden (TMB) and immunosuppressive microenvironment in most PCa patients [6][7][8].
Although prior evidence indicated that ICB is generally ineffective in unselected PCa patients, other studies have found that ICB shows durable response in PCa patients with mismatch repair deficiency (dMMR) [9][10][11]. DNA mismatch repair is one of an important mechanisms of DNA damage repair, which can repair incorrect base matches and insertion/deletion errors that occur during replication. dMMR status is mainly caused by pathogenic mutations in MLH1, MSH2, MSH6, and PMS2, which often leads to the occurrence of microsatellite instability high (MSI-H) phenotypes in tumors and the generation of high tumor mutation burden (TMB-H) [12][13][14]. Immunotherapy typically elicits durable anti-tumor effects in dMMR/MSI-H cancers due to the elevation of tumor neoantigens [15,16].
According to the previous sequencing studies of PCa, deleterious mutations in MMR genes are uncommon, estimated at 2-5% of cases [11,17,18]. However, most of the genetic data come from Caucasus population, and some studies have shown that MMR genes variations are highly ethnic specific [19,20]. The tumor immune microenvironment (TIME) of solid tumors mainly includes the density, localization and composition of tumor-infiltrating immune cells, and it plays an important role in the selection of patient populations that may benefit from immunotherapy [21]. Since MMR genes mutations are relatively rare in the PCa, the prevalence of MMR genes mutations, the clinical features, the characterization of TIME and the efficacy of ICB among this group of Chinese PCa patients remain poorly understood. Herein, we use the genetic sequencing data from 3338 Chinese PCa patients to depict the mutation spectrum of 4 MMR core genes and the clinical characteristics of PCa patients carrying deleterious MMR mutations. Besides, we report a multiinstitutional case series of metastatic castration resistant prostate cancer (mCRPC) with dMMR/MSI-H treated with anti-PD-1 therapy and explored the association of the TIME with the treatment outcomes through multiplex immunofluorescence (mIF).

Patients and materials
We obtained the genomic profiling data of 1502 patients treated in Fudan University Shanghai Cancer Center (FUSCC) between May 2018 and March 2022. All patients were 18 years of age or older and had histologically confirmed prostate adenocarcinoma. The disease stage at the time of genetic test and the frequencies of deleterious MMR genes carriers in disease subsets are summarized in Supplementary Table 1. In addition, we also included 1836 Chinese PCa patients with germline sequencing data which were previously reported in our study [22]. A total of 11 mCRPC patients with dMMR/MSI-H from 4 different Medical Institutions (5 from FUSCC, 2 from West China Hospital, 2 from Qilu Hospital of Shandong University, 2 from Sun Yat-sen University Cancer Center) in China were included for analyzing the response of anti-PD-1 inhibitors. Seven samples from primary tumor (6 samples pre-immunotherapy, 1 sample post-immunotherapy) were collected for mIF. Medical record review and test requisition forms (when available) were conducted for patient clinical characteristics and outcomes with a final follow-up date of June 28, 2022. The study was approved by the Committee for Ethics of Fudan University Shanghai Cancer Center, and informed consent was obtained from each patient.

Sequencing and bioinformatics
The details of sequencing experiment process and sequence data analysis were performed as described in the Supplementary Method. The clinical significance of MMR genes germline variants was assigned using the guidelines of American College of Medical Genetics and Genomics guidelines (ACMG) [23], and the flow chart of variants annotation was described in our previous studies [24,25]. Pathogenic somatic mutations of MMR genes were identified by the following process: (1) Variants reported as oncogenic or likely oncogenic in OncoKB (http:// oncokb. org) [26] were retained and neutral variants were removed; (2) remaining variants with uncertain significance were further annotated through InterVar and ClinVar [27,28]. When one of these two databases shows a variant of unknown meaning (VUS) and the other provides a non-VUS annotation, the non-VUS status is used. Only pathogenic/likely pathogenic (P/LP) variants were considered to be deleterious and further analyzed in this study.

Multiplex immunofluorescence
Multiplex immunofluorescence staining was performed by using the Akoya OPAL Polaris 7-Color Automation IHC kit (NEL871001KT) to quantify the density of specific immune cell. Tumor parenchyma and stroma were differentiated by Pan-CK staining. Details of the protocol are shown in the Supplementary Method.

Statistical analysis
The clinical characteristics of the 1502 PCa patients in our cohort and mCRPC patients treated with PD-1 inhibitors were presented using descriptive statistics. To investigate the association of deleterious alterations in MMR genes with clinical characteristics, we used the two independent sample t tests for age of onset, the Mann-Whitney U test for prostate-specific antigen (PSA) at diagnosis and the Chisquare test for Gleason score group and family history of first degree. The Kaplan-Meier method was used to estimate the time to CRPC and the time on first-line abiraterone. Differences between groups were identified using the log-rank test in the survival package (v.3.2-11). CRPC was defined as 3 consecutive rises in PSA above the nadir with a castration level of serum testosterone (< 50 ng/dL). PSA progression was defined as a 25% rise from the nadir PSA (confirmed by a second PSA rise after at least 3 weeks). All statistical analyses were performed using R (v4.0.3) programming language with two-sided p values < 0.05 considered statistically significant.

Germline and somatic alterations in MMR genes
The patients included in our study for MMR genes mutation analysis are shown in Fig. 1. A total of 3338 patients had germline sequencing data, of which 1502 patients had both somatic sequencing data. The mutation distribution of the 4 MMR core genes (MLH1, MSH2, MSH6 and PMS2) are illustrated in Fig. 2. Among the 3338 PCa patients in our cohort, a total of 77 P/LP variants were found in 66 patients in the 4 genes and 28 (36.4%) of them were not previously reported in Clinvar and dbSNP databases ( Fig. 2A, the genomic alterations details are shown in Supplementary Table 2 and Fig. 1). Twenty-two (0.7%) patients were detected to carry pathogenic germline MMR genes mutations. For the 1502 PCa patients with germline and somatic genetic data, 54 (3.6%) patients carry deleterious mutations, including 10 patients (0.7%) with germline mutations (Fig. 2B). Most of the germline mutations occurred in MSH2 (13, 59.1%), while most of the somatic mutations occurred in MSH6 (24, 37.5%). Notably, 9 of 54 patients (16.7%) had 2 or more deleterious mutation loci in the 4 genes, 7 (13.0%) of them had mutations in two different MMR genes, and 1 patient (1.9%) had both germline mutation in MSH6 and somatic mutation in PMS2. To determine the similarities and differences in the mutation profiles of the MMR genes in the Chinese and American populations, we compared the P/LP mutation frequencies in our cohort (patients with germline and somatic data) with Caucasians of the cohort by Mahal et al. [29] (Supplementary Fig. 2). (The pathogenicity of MMR gene mutations was annotated using the same approach as for our data.) The results showed that in our cohort, the incidence of P/LP MMR gene mutations in metastatic cases was higher than in the American cohort (Metastatic: 4.8% vs 2.2%, P = 0.006. Non-metastatic: 2.3% vs 1.4%, P = 0.301).

Fig. 1
Cohort summary. A total of 3338 PCa patients were included in our study for genetic data analysis. One patient had both germline mutation and somatic mutation in different MMR genes. Among these patients, 1836 patients with germline sequencing data have been reported in our previous study (22) 1 3

Clinical and pathological characteristics
The 1502 PCa patients with somatic and germline genetic data in our cohort were divided into two groups based on MMR mutation status. Clinical and pathological characteristics of these patients are summarized in Table 1  and Supplementary Table 3. Patients with P/LP MMR genes mutations had lower PSA at diagnosis (P = 0.001) than non-carriers and often had a history of cancer in a first-degree relative (38.1 vs 20.4%, P = 0.011). Besides, it is worth noting that in our cohort mutation carriers responded relatively poorly to androgen deprivation therapy and abiraterone compared to non-carriers (Fig. 3B, C) (median time from ADT start to CRPC: 13.2 vs 21.3 months, P = 0.003; median duration of first-line abiraterone therapy: 8 vs 12 months, P = 0.016), which was contrary to previously reported result [30]. In terms of age at diagnosis and Gleason grade group, no statistically significant differences were observed between the 2 groups.

Responses to anti-PD-1 therapy in mCRPC with dMMR/MSI-H
We retrospectively collected clinical treatment information of 11 mCRPC patients with dMMR/MSI-H from 4 different centers who had received anti-PD-1 therapy (pembrolizumab, tislelizumab or sintilimab) during May 2019 to June 2022. All patients were confirmed to have dMMR/MSI-H by immunohistochemistry (IHC) for loss of MMR protein expression or PCR test [31] for MSI-H before receiving immunotherapy. The clinical characteristics of these patients are presented in Table 2 and Supplementary Table 4.

TIME in dMMR/MSI-H PCa patients treated with PD-1
Among the 11 mCRPC patients who were treated with PD-1 inhibitors, 6 patients (3 responders and 3 non-responders) were available for pre-treatment tumor samples. We performed mIF to evaluate the intratumoral and stromal immune infiltrates in these patients, and the results showed that although immune cell infiltration was remarkably heterogeneous, responders had higher density of intratumoral CD8 + T cells than non-responders (Fig. 5A, B, Supplementary Fig. 3).
Patient 6 had a significant decrease in PSA after 3 doses of tislelizumab, but the MRI scan showed that the tumor size decreased moderately (Fig. 6A, B). We obtained tumor biopsy tissues from this patient after receiving immunotherapy and performed mIF, the results of which showed the infiltration of intratumoral CD4 + T cell, CD8 + T cell (especially PD1 + CD8 + T cell), CD68 + macrophages, NK cell, Treg cell and stromal CD68 + macrophage increased significantly after treatment (Fig. 6C, D, Supplementary Fig. 4).

Discussion
Although the use of new-generation antiandrogens and chemotherapy has prolonged the overall survival of mCRPC patients, the disease remains incurable and lethal. The advent of immunotherapy has dramatically changed the outlook for cancer treatment, and anti-PD-1 inhibitor was   [32]. Compared to other solid tumors, immunotherapy has not achieved a therapeutic breakthrough for PCa. The clinical benefits of ICB in unselected advanced PCa patients were modest. According to the phase II KEYNOTE-199 study, the objective response rate of unselected mCRPC patients to pembrolizumab was only 5% [5]. However, Abida et al. reported that about half of mCRPC patients (6/11,54.5%) with dMMR/MSI-H in their series responded to anti-PD-1 or anti-PD-L1 inhibitors, demonstrating the promising efficacy of immunotherapy in this particular subset of PCa patients [11].
To the best of our knowledge, our cohort is the largest Chinese PCa cohort reported to date aiming to delineate the incidence of deleterious mutations in the MMR genes, and this is the first study to assess the efficacy of anti-PD-1 inhibitors in mCRPC with dMMR/MSI-H in Asian populations. By analyzing the genetic test data of 3338 PCa patients in our cohort, we found that a total of 3.6% patients carried deleterious mutations in MMR genes, and 36.4% P/ LP variants were newly identified in our study. Compared to European ancestry in the cohort by Mahal et al., the P/ LP mutation frequencies of MMR genes in patients with metastatic PCa in our cohort were higher (4.8 vs 2.2%, P = 0.006). These results validate that MMR gene mutations are highly ethnic-specific [19,20], and suggest that it may be more common in Asian PCa patients.
Since MMR mutations are rare in PCa patients, their clinical features are largely unknown. Previous studies have shown that deleterious MMR gene mutations in PCa were related to aggressive clinical and pathological features [30,33]. Notably, Antonarakis et al. reported that dMMR PCa patients show high sensitivity to ADT and next-generation antiandrogens (median PFS: 66 months for ADT, 24 months for abiraterone and 26 months for enzalutamide), but the opposite results were reported by Abida et al. in their cohort of dMMR/MSI-H PCa patients (median PFS: 8.6 months for ADT, 9.9 months for abiraterone or enzalutamide) [11,30]. In our study, we did not find significant differences in pathological characteristics between the deleterious MMR genes mutations carriers and non-carriers, which may be explained by the relatively high proportion of patients with metastatic disease in our cohort, but found that ADT and abiraterone seemed to be of worse efficacy in patients with the MMR genes mutations. However, the number of mutations carriers included in the analysis was relatively small, leading to unstable estimation. Therefore, the association of MMR gene mutations with the clinical features and treatment outcome in PCa patients remains controversial, and larger cohort studies will be needed in the future to clarify the relationship.
Within our multi-institutional case series report, the overall response rate of dMMR/MSI-H PCa patients to anti-PD-1 inhibitors (6/11, 54.5%) was similar to prior reports [11, 34, and the result validates the relatively promising effect of ICB in this subgroup of PCa patients. To further explore the association between the TIME and treatment response, we performed mIF on the pre-treatment biopsy tissues of 6 patients. The results showed that compared to non-responders, patients who responded to PD-1 therapy had a higher density of intratumoral CD8 + T cells. According to previous studies, a high density of intratumoral CD8 + T cells was often associated with better response to immunotherapy and improved overall survival for cancer patients [35][36][37]. Although the number of patients and specimens tested in our cohort was too small to draw a definitive conclusion, these results indicate that combining the patient's TIME and MMR/MSI status may be more helpful to accurately predict the efficacy of immunotherapy. Besides, we found that one of the patients had a significant decrease in PSA after immunotherapy, while the MRI scan showed moderate reduction in tumor size. Using mIF to examine the TIME before and after immunotherapy in this patient, we found a significantly increased infiltration of various immune cells in the post-treated biopsy specimen, especially PD-1 + CD8 + T cells, which were usually considered as exhausted T cells characterized by progressive loss of effector functions [38]. This result was consistent with Alice et al. who found that most lung cancer patients responded to immunotherapy had increased PD-1 + CD8 + T cells in peripheral blood after PD-1 treatment [39]. Besides, it suggests that there may be some patients who could benefit from immunotherapy without a reduction in tumor volume at the beginning of the treatment, and this phenomenon occurs due to the infiltration of immune cells rather than the growth of the tumor [40,41].
Although previous studies have shown that cancer patients with dMMR/MSI-H phenotype generally have high immune infiltration in tumor [42,43], the TIME characteristics of this subgroup in PCa are largely unknown due to the relatively rare occurrence of dMMR/MSI-H in PCa patients. The study by Daniel et al. showed that T cell infiltration in PCa was strikingly heterogeneous, and approximately half (5/9, 55.6%) of dMMR/MSI-H PCa patients showed high T cell infiltration [44], potentially accounting for the similar response rates to PD-1 inhibitor Fig. 6 Case description of patient 6 who had significant PSA decline while without initial radiographic tumor shrinkage. A Genetic characteristics and PSA change during the treatment of patient 6. Pre-treatment biopsy specimens was used for NGS test. The second biopsy was performed after the patient had received 3 doses of PD-1 inhibitor. B Magnetic resonance image (MRI) of patient 6 before the administration of PD-1 inhibitor (top row), after 3 doses of PD-1 inhibitor (middle row) and after 5 doses of PD-1 inhibitor (bottom row). C Representative sections of tumor specimens obtained from patient 6 before the administration of PD-1 inhibitor (upper row) and after administration (lower row, the white arrows indicated PD-1 + CD8 + cells) (hematoxylin & eosin and mIF staining). D Comparison of immune cell infiltration in patient 6 before and after PD-1 inhibitor administration. Abbreviations: PD-1 = programmed cell death protein 1, PSA = prostate-specific antigen, NGS = next-generation sequencing, mIF = multiplexed immunofluorescence 1 3 in these patients. However, the majority of PCa patients present with "cold" tumors which have sparse immune cell infiltration [8]. Therefore, promoting the transformation of immune "cold" tumors into immune "hot" tumors is of great significance for the application of immunotherapy in PCa [45]. Several recent studies have shown that some cancer treatment modalities, such as radiotherapy and ADT therapy, show promising prospects in converting the TIME, through increasing local lymphocyte infiltration or enhancing the function of T cells [46][47][48].
Our study has several limitations. First, although MMR gene mutations are rare in PCa, we chose the 4 key genes (MLH1, MSH2, MSH6, and PMS2) instead of all MMR genes and we did not investigate copy number variants, which may lead to overlook some important findings. Second, it is a retrospective study and there is a selection bias associated with limited access to current new therapies. Third, not all MMR genes P/LP mutations lead to MMR protein loss or microsatellite instability, while a small proportion of dMMR patients do not harbor MMR gene mutations [18,44,49]. Therefore, describing the incidence of MMR mutations alone does not give a complete picture of the occurrence of dMMR/MSI-H in PCa. In addition, the number of mCRPC with dMMR/MSI-H who were treated with anti-PD-1 therapy is limited and due to health insurance limitations, these patients are treated with different PD-1 inhibitors (pembrolizumab, tislelizumab or sintilimab).