Homogeneous MMR Deficiency Throughout the Entire Tumor Mass Occurs in a Subset of Colorectal Neuroendocrine Carcinomas

Neuroendocrine neoplasms occur at various organ sites and comprise a heterogeneous group of epithelial tumors ranging from well-differentiated neuroendocrine tumors (NETs) to poorly differentiated carcinomas, called neuroendocrine carcinomas (NECs). NETs clinically often present as low-grade malignancies, while NECs behave more aggressively with rapid disease progression and poor long-term survival, demanding for novel treatment options.

Microsatellite instability (MSI) refers to a distinctive pattern of increased mutational load in certain tumors, typically caused by a defective DNA mismatch repair (MMR) apparatus, and is characterized by accumulated length variations of repetitive DNA sequences (microsatellites) throughout the genome. Protein loss of at least one of the major components of the MMR system—MLH1, PMS2, MSH2, or MSH6, which can easily be analyzed by immunohistochemistry (IHC)—provides strong indirect evidence for MSI. MSI can also be detected via direct analysis of a predefined panel of DNA microsatellite loci with classic polymerase chain reaction (PCR)–based methods or novel high-throughput molecular techniques [1]. Favorable response rates for immune checkpoint inhibitors in cancers with MSI have dramatically increased the clinical request for MSI testing, even in tumor types with low expected rates of MSI. This was paralleled by the 2017 site-agnostic FDA approval of the PD-1 antibody pembrolizumab for advanced cancers with MMR deficiency/MSI-high.

Data on MSI in neuroendocrine neoplasms are limited. Recent studies have found tumors with MSI among neuroendocrine carcinomas (some combined with a conventional adenocarcinoma component, MINEN) of the stomach, small intestine, and colorectum [2‐6]. The few available studies describing MSI in cohorts of NETs are basically limited to pancreatic primaries with reported frequencies between 10% (5/48) and 33% (18/55) by PCR-based methods [7, 8]. Intratumoral heterogeneity of MSI—a potential strong confounder for molecular based treatments—has been observed in several cancer types, including colorectal cancer [9‐13]. However, intratumoral heterogeneity of MSI in neuroendocrine neoplasms has not been evaluated so far.

To systematically assess potential intratumoral heterogeneity of MMR protein expression in neuroendocrine neoplasms with MMR deficiency, a cohort of 239 NETs and NECs was screened on a tissue microarray (TMA) format by IHC, followed by a large section evaluation of cancers suspected for MSI by repeated IHC and PCR analysis.

A TMA containing 199 NETs and 40 NECs of various primary sites was manufactured for the purpose of this study as a screening tool to identify tumors with MSI among the heterogeneous group of neuroendocrine neoplasms. Detection of rare events is an ideal application for TMAs. For example, earlier studies identified 42 tumors harboring IDH1 mutations by screening 15,531 prostate cancers (0.3%) or 43 tumors exhibiting CD117 overexpression in a cohort of 1654 breast carcinomas (2.6%) [17, 18]. We recently also identified MMR deficiency/MSI in 5 tumors in a series of 448 bladder cancers (1.1%), in 7 tumors in a series of 200 advanced and/or hormone refractory prostate cancers (3.5%), and in 9 tumors in a series of 479 ovarian cancers (1.8%) (unpublished data). In the present study, only one of the 4 tumors with suspected MMR deficiency/MSI based on TMA screening turned out to be MMR intact on subsequent large section analysis. The discrepant results between TMA and large section IHC were observed in a pancreatic NET G3 and attributable to suboptimal immunoreactivity of major tumor areas, which—for PMS2—resulted in still faintly detectable staining of stromal cells but negativity of tumor cells on the TMA spot in contrast to unequivocally retained MMR protein expression in cancer cells on large section. This observation may reflect somewhat higher PMS2 expression levels in a subset of stromal cells as compared with cancer cells.

Among the 239 analyzed neuroendocrine neoplasms, MSI was confirmed in 3 cases (1.3%), all representing NECs of the cecum, suggesting a considerable prevalence of MSI (30%; 3/10) in colorectal NECs, although this may not reflect the true MSI rate in colorectal NECs due to the small tumor cohort. None of the other primary NECs from different sites of the gastrointestinal tract (n = 15) and the lung (n = 7) exhibited MSI. These data fit with previous studies reporting particularly high rates of MSI in NECs of the colorectum. Sahnane et al. found MSI in 6 of 37 (16%), Olevian et al. in 2 of 29 (7%), Furlan et al. in 2 of 21 (10%), and La Rosa et al. in 3 of 35 (14%) colorectal NECs [2, 4‐6]. In the study by Sahnane et al. also 4 tumors with MSI were identified among 36 gastric NECs (11%) [2]. A high frequency of MSI (44%) based on IHC detection of MMR protein loss was further observed by Pocrnich et al. in a cohort of 18 large cell NECs of the endometrium [19]. In contrast, NECs from other organs have rarely been found to exhibit MSI. Sahnane et al. found only one other tumor with MSI among 16 NECs of the duodenum, esophagus, pancreas, and gall bladder [2].

Taken together, the available data suggest that the prevalence of MSI in NECs is site-dependent and closely related to those organ sites where the exocrine neoplastic counterparts of NECs—adenocarcinomas—are also frequently affected by MSI, such as endometrial, colorectal, and gastric adenocarcinomas. All 3 colorectal NECs with MSI in the present study were located in the cecum, corresponding well to the strong association of MSI in sporadic colorectal adenocarcinomas with tumor location in the right colon [20]. A link between MSI in conventional and neuroendocrine carcinomas of distinct organ sites is further supported by several reports of MSI in NECs with concomitant adenocarcinoma components in the affected tumors (MINENs) [2, 4, 6]. The concept of site dependency of MSI in NECs is also supported by the reported absence of MSI-high tumors among 107 small cell lung cancers by Chung et al. [21], paralleling the low frequency of MMR deficiency in non-small cell lung cancer [22‐26]. Based on the available data, MSI should be regularly tested in colorectal, gastric, and endometrial NECs. As NECs confer a very poor prognosis and no second line treatment after failure of Cis/Carboplatin and Etoposide has been established [27], MSI testing should be performed during initial diagnosis to avoid unnecessary treatment delay upon progression after first-line treatment. As colorectal NECs have less than 50% response rate upon Platin-based chemotherapy and an overall survival of less than 9 months, this patient group might in particular benefit from upfront MSI testing and initiation of immune checkpoint inhibitor therapy [27].

In contrast to NECs, none of the 199 NETs exhibited MSI in the present study. That MSI is highly infrequent in NETs fits well with data from earlier studies. MSI was not found in previous studies analyzing 56 rectal [28], 14 small bowel [29], 16 pancreatic [30], 29 gastroenteropancreatic [31], and 38 foregut/midgut [32] as well as 35 pancreatic and 34 small intestinal NETs [33] by PCR. In contrast, two earlier studies described relevant MSI rates in pancreatic NETs [7, 8]. However, as those studies were performed 10 and more years ago, the applied approach to identify or define MSI differs from current standards. House et al. reported MSI in 5 of 48 (10%) pancreatic NETs using the “Bethesda Panel,” but status of MSI was already given when merely the mononucleotide repeat locus BAT-25 was instable, contrasting the current cut-off requiring instability in 40% of loci to justify MSI [7]. Mei et al. reported MSI-high, defined as instability of at least 4 loci of an extended panel of 12 microsatellites, in 18 of 55 (33%) insulinomas [8]. In contrast, all 8 insulinomas among the cohort of 56 pancreatic NETs analyzed in the present study were MMR intact. There are, however, reports describing MMR deficiency in few individual cases of pancreatic NETs in patients with hereditary MMR deficiency (Lynch Syndrome) [34, 35].

Evaluation of the MMR status throughout all available cancer-containing tumor blocks revealed homogeneous MMR protein loss in all three NECs with confirmed MSI. This is in agreement with our previous observations of high homogeneity of MSI in prostate [36], ovarian [37], and bladder cancer (unpublished data). Overall, these data may suggest that MMR inactivation generally occurs early in tumorigenesis. Regarding treatment purposes, homogeneity of MSI reduces the risk that molecular parameters obtained from small biopsies may not be representative for the entire cancer mass, a potential strong confounder for individualized therapies. It is a limitation of our study that results for all tumors with intact MMR were derived from TMA cores and it cannot be completely ruled out that a tumor with heterogeneous MMR/MSI status has been missed among this group because a MMR-deficient tumor area has not been represented on the respective TMA spot.

PCR analysis revealed MSI-high in all 3 cases of MMR deficiency. Although the number of cases is too small to draw major conclusions, the high concordance between IHC and PCR results is not surprising. The “Bethesda Panel,” a selection of 5 specific mono- and dinucleotide repeats, has been developed based on data derived from colorectal cancers [16]. As different microsatellite loci are not equally affected by frameshifts in the state of MSI due to an association between transcriptional activity and the occurrence of instability in individual repeat loci [38, 39], a tissue-dependent pattern of MSI is conceivable. Loci typically demonstrating instability in colorectal adenocarcinomas with MSI, i.e., the loci covered by the “Bethesda Panel,” may be affected with comparable frequency in colorectal NECs exhibiting MSI, just as both tumor types certainly show genetic overlap due to their shared origin from large bowel mucosa.

In summary, the detection of MSI in 3 of 10 colorectal NECs but not in other neuroendocrine neoplasm suggests that MSI affects NECs of the colorectum in a relevant manner, similar to colorectal adenocarcinomas. In contrast, MSI is highly infrequent in NETs irrespective of tumor site. The observed complete homogeneity of MMR deficiency indicates that the use of minor biopsy samples should result in highly representative data and hints towards MSI as an early event in tumorigenesis. Testing for MSI early in the course of disease might open novel treatment options for colorectal NECs.