DNA MMR machinery is essential for the maintenance of genomic stability. The MMR machinery is composed of MSH2/MSH6 and MSH2/MSH3 that recognize single-nucleotide mismatches and small insertion/deletions that occurs during DNA replication. Subsequently hMLH1/hPMS1 Homolog 2, (hPMS2), hMLH1/hPMS1 Homolog 1(hPMS1) and hMLH1/hMLH3 are recruited to catalyze the excision and resynthesize the mismatch [1]. The dysfunction of this system, namely dMMR, results in an errors in microsatellites which consist of repeated DNA sequences of 1–6 nucleotides [2]. Thus, the alteration of the number of microsatellites sensitizes a dMMR state. This is referred to as microsatellite instability (MSI). MSI is thought to be involved in tumorigenesis and tumor proliferation due to the accumulation of repair-associated mutations in genes for tumor suppression, cell proliferation, DNA repair, apoptosis. Clinically, it can be categorized as MSI-H and MSI-low or stable (MSS) according to the frequency of MSI [3]. In sporadic dMMR tumor, it is mainly caused by MMR gene mutation due to acquired hypermethylation of the promoter region of MLH1 gene, leading to decreased expression of MMR protein [4]. On the other hand, the Lynch syndrome, which takes the form of autosomal dominant inheritance, is caused by germline mutations of the MMR-regulated genes (MLH1, MSH2, MSH6, and PMS2) or a deletion of the EPCAM gene adjacent to the upstream of the MSH2 in one allele [5,6,7].

MSI-H/dMMR solid tumors are found in various organs [8,9]. The frequency of MSI-H/dMMR in colorectal cancer (CRC) is reported to be approximately 15% [10] with Lynch syndrome-associated CRC accounting for ~20–30% of cases and sporadic MSI-H/dMMR CRC being ~70–80% of cases [11]. The frequency of MSI-H/dMMR CRC varies according to stage (~20% stage I/II, 12% stage III, and 5% in stage IV) [12]. MSI-H/dMMR CRC is more common in the right colon and the proportion of poorly differentiated adenocarcinoma is high [13]. Moreover, BRAF V600E gene mutation is found in 35–43% of MSI-H/dMMR CRC [14,15]. Since BRAF V600E gene mutations are rarely found in the Lynch syndrome-related CRC, BRAF screening in MSI-H/dMMR CRC helps to distinguish sporadic MSI-H/dMMR tumor or Lynch syndrome [16].

In gastric cancer (GAC), MSI-H/dMMR has a frequency of ~20% [17,18]. As well as MSI-H/dMMR CRC, the prevalence depends on tumor stage; the highest in node-negative stage (20%) and the lowest in metastatic disease (<5%). In esophageal adenocarcinoma (EAC), MSI-H/dMMR can be observed 3–5% due to somatic mutation since Lynch syndrome associated esophageal adenocarcinoma (EAC) is rare [19]. Regarding gastroesophageal junction (GEJ) cancer, those in Siewert type II and III are related to MSI-H/dMMR [20]. In small bowel cancer, the frequency of MSI-H/dMMR is reported to be 5–45%, which is a relatively big range and frequency [21]. MSI-H/dMMR is also associated with other GI tumors at a lower incidence (i.e., ~2–2.5% pancreatic; ~2% biliary; ~2% gallbladder, etc.) and even more rarely seen in some (Hepatocellular carcinoma (HCC) and anal cancer given different cancer etiologies) [22,23,24].

MSI-H/dMMR tumors are generally associated with a high neoantigen burden, highly immunogenic, and thus thought to respond to ICI therapy. A current exciting pathway for these tumors is initial investigation with localized MSI-H/dMMR solid tumors and the potential for organ-sparing (non-operative) approach. Cercek et al. recently published results of locally advanced MSI-H/dMMR rectal cancer patients (n = 12) who received anti-programmed death-1 (anti-PD-1) agent ICI, dostarlimab [25]. All patients had a clinical complete response (cCR). Currently, no patients had received chemoradiation (CRT), undergone surgery, progressed, or had recurrence. Additionally, Ludford et al. reported initial results giving pembrolizumab to MSI-H/dMMR tissue agnostic localized primary tumors (n = 32) [26]. Tumor types included 24 CRC and 8 non-CRC (1 endometrial, 1 gastric, 1 meningeal, 2 duodenal, 1 ampullary, 2 pancreatic). Among 30 evaluable pts, overall response rate (ORR) was 77% with 30% CR, 47% PR, 20% stable disease, 3% progression. Pathological CR (pCR) was noted in 50% of the six patients that underwent surgery. An organ-sparing approach was chosen in 15 patients and two patients had reached one year of avoiding surgery. Additionally, an ICI shift in upfront treatment for advanced MSI-H/dMMR CRC has been a recent development [27]. These patients are now recommended pembrolizumab monotherapy given superiority over chemotherapy seen in KEYNOTE-177. In the KEYNOTE-158, a phase 2 pembrolizumab study, an ORR of 40.9%, a median progression-free survival (PFS) of 4.2 months, and median OS of 24.3 months was seen in advanced, pre-treated MSI-H/dMMR biliary tract cancers (n = 22) [28]. Although dMMR/MSI-H status for GI cancer has become a biomarker that determines an indication for ICIs, factors associated with resistance are still being investigated. Further research would suggest more implications for the role of dMMR/MSI-H status as a predictive marker for immunotherapy but determine why certain MSI-H/dMMR patients do not respond will be the key to moving forward. We believe tissue-agnostic trials of MSI-H/dMMR tumors will provide answers in a quicker fashion. Exciting trials are underway in both the localized MSI-H/dMMR rectal, colon, and gastric setting and metastatic solid tumor setting as these patients will need differing treatment strategies than those proficient in MMR/MSS. Phase 3 trials in this space are described in Table 1 [29,30,31,32,33,34,35].

Given the role that PD-L1 plays in tumor immune escape, its expression has emerged as a potential biomarker to test the effectiveness of ICI. PD-L1 expression is a current exploration amongst GI tumors to determine if this holds an ICI predictive role.

For upper GI patients (gastric and esophageal patients), PD-L1 expression and ICI response is of much debate given conflicting results seen in CHECKMATE 649, KEYNOTE-811, ATTRACTION-4, JAVELIN, and ORIENT-16 [36,37,38,39,40]. These trials are described in detail in Table 2. Currently, for upper GI tumors, we feel PD-L1 expression (method and degree of positivity) needs more standardization across trial designs to determine the predictive value. It is clear with the current data that additional new biomarkers and correlating with other clinicopathological features are needed to determine those likely to benefit in the high PD-L1 combined positive score (CPS) patients. As this appears at present time not to be the ideal biomarker alone to determine ICI response.

PD-L1 expression does not appear currently to be key marker for biliary tract cancers as seen in the TOPAZ-1 trial in which the current standard practice is for upfront advanced tumors to receive durvalumab with gemcitabine plus cisplatin [41]. PD-L1 expression also did not appear to hold predictive value in the KEYNOTE-158 study [42]. ICI biomarker relevance will be needed soon; however, in advanced biliary tract tumors as results of the phase 3 gemcitabine plus cisplatin +/− nab-paclitaxel, results are expected soon [43]. Determining who would benefit most from ICIs might help determine the best upfront therapy if this trial’s results are favorable. Controversial results are noted for HCC. CHECKMATE-459, nivolumab compared to sorafenib, those with PD-L1 positive reported higher ORR [44]. While CHECKMATE-040 showed no statistical difference [45]. Of significance are updated results of the IMbrave150. IMbrave150 established atezolizumab and bevacizumab are standard front-line treatment for advanced HCC. An updated retrospective look at the tissues in this study showed that PD-L1 expression is likely of limited predictive value to determine benefit with atezolizumab and bevacizumab (median overall survival (OS) 12.6 months PD-L1 ≥1%; median OS 15.4 months for PD-L1 <1%) [46]. Additionally, results of the HIMALAYA trial of tremelimumab and durvalumab are expected to be added to the treatment choices for upfront HCC [47]. PD-L1 expression in relation to outcomes in HIMALAYA lacked reporting thus limit determination of an ICI predictive link.

For squamous cell carcinoma of the anal canal (SCCA), single agent ICIs (nivolumab or pembrolizumab) are options for refractory metastatic anal cancer [48,49]. PD-L1 expression was not required in the studies evaluating the use of these agents in this refractory patient population; however, exploratory analysis suggests higher response in those with PD-L1 expression [48]. These data, however, remain too immature for any value.

There is much work needed at understanding the predictive role of PD-L1 expression regarding ICI GI therapy as currently it has not been as precise as hoped. Questions remaining include (1) determining the standard definition for PD-L1 expression, (2) what tumor should be tested (fresh; archived) (3) why does expression not correlate to response (4) why do some non-PD-L1 expressing tumors shrink (5) is expression altered by prior therapy (6) does PD-L1 expression drive immunogenicity in the same fashion across tumors? For now, we believe PD-L1 expression correlation remains too vague for most GI tumors and continued exploration is needed to determine the role in each tumor type.