Lights and shadows of microsatellite status characterization in gastrointestinal cancers in the era of cancer precision therapy

The introduction of immunotherapy has dramatically changed the paradigm of solid tumor treatment with the creation of novel therapeutic opportunities even for tumors that currently lack valid therapeutic options in the advanced or metastatic setting. Initially, the role of deficient mismatch repair status (dMMR)/microsatellite instability (MSI) as a predictive biomarker was confined to colorectal cancer. In 2017, MSI/dMMR became the first true agnostic biomarker to stratify patient response to immune checkpoint inhibitors. MSI/ dMMR evaluation is a crucial point in diagnostic-therapeutic decision-making for most gastrointestinal cancer patients and the pathologist must be responsible for the delivery of reliable reporting in this setting. The aim of this review is to summarize the current methods available in routine diagnostics for the evaluation of MSI/dMMR status, their limitations, and potential pitfalls that can be encountered. The authors also give an overview of the role of MSI/dMMR as a prognostic and predictive biomarker in gastrointestinal cancers, with a focus on non-colorectal malignancies.

Introduction

Microsatellites are tandem repeats of short DNA motifs, composed of up to 9 base pairs (bp) and repeated many times throughout the genome¹. DNA repeated motifs include mononucleotide repeats, dinucleotides, and tetranucleotides, generally rich in adenine (A) or thymine (T)². Microsatellites are widely distributed throughout the entire genome but are predominantly found in noncoding regions including introns, 5’ and 3’ untranslated regions (UTRs), and intergenic regions^3,4. Microsatellites spontaneously mutate 10 to 100-fold faster than non-repetitive sequences⁵. The accumulation of mutations at the microsatellite loci level is defined as microsatellite instability (MSI) and is predominantly characterized by the change in the number (gain or loss) of tandem repeats due to DNA slippage during DNA replication⁶. Based on the number of mutated microsatellite loci, microsatellite status (MS) can be classified in high (MSI-H) or low (MSI-L) levels of MSI. Factors affecting microsatellite mutation rate include motif size, gender, guanine (G) and cytosine (C) content, genomic context, and most importantly the effectiveness of mismatch repair (MMR) enzymes⁷.

Role of mismatch repair system in microsatellite stability

The MMR system corrects errors occurring during DNA replication, thus preventing the establishment of mutations during cell division, through the activation of cell cycle arrest and/or mechanisms of programmed cell death (e.g. apoptosis)⁸. Alterations in the MMR system, defined as MMR deficient (dMMR) phenotype, lead to the accumulation of genetic errors, particularly within microsatellites, which are more prone to errors during DNA replication. MMR proteins are encoded by 8 genes (among them: MSH2, MSH3, MSH6, MLH1, MLH3, PMS2) heterogeneously located throughout the human genome^9,10. MMR proteins are involved in the recognition of mismatches and can be classified into two groups: MutS Homolog (MSH) including MSH2, MSH3, and MSH6, and MutL Homolog (MLH) enclosing MLH1, MLH3, and PMS2. MMR acts through an excision and re-synthesis process and can recognize and correct mismatches in both 3′ and 5′ strands. While the heterodimer MSH2-MSH6 detects single base mismatches, MSH2-MSH3 complex identifies larger (2-10bp) mutations¹¹. The heterodimer MLH1-PMS2 is the most frequent in the MutL complex¹². MLH1 and MSH2 are essential for MMR activity, whereas alterations in MSH6 or PMS2 may be, even if partially, replaced by MSH3 and MLH3, respectively (Fig. 1). Deficient MMR phenotype is associated with genetic or epigenetic alterations of genes involved in the MMR pathway, which may be either hereditary or acquired/sporadic. Acquired dMMR phenotype is usually caused by the absence of MLH1 protein expression, due to the hyper-methylation of the MLH1 gene promoter and hence silencing of the corresponding gene. Hereditary dMMR phenotype is caused by germline mutations, generally affecting only one allele. Individuals with this mutation are carriers of Lynch syndrome (LS) and present a higher predisposition to certain malignancies, which can develop when a second hit occurs¹³. Among patients diagnosed with LS, those who have germline mutations in MHL1 and MSH2, have a much higher risk of developing cancer and/or develop cancer earlier compared to subjects with MSH6 or PMS2 mutations, despite the latter being more frequent.

The role of MSI in current oncology practice

Microsatellite status has been long considered a molecular feature of endometrial (EC) and colorectal (CRC) cancer^14-16. In the CRC setting, MS is a prognostic and predictive marker. MSI-CRCs have a better prognosis and lower recurrence rate compared to MSS tumors and MSI CRC do not respond well to 5-FU-based monotherapy¹⁷. The introduction of immune checkpoint inhibitors (ICIs) has dramatically changed the role of MS as a marker. Deficient MMR phenotype results in hypermutated tumors, which produce a high quantity of neoantigens that are recognized by the immune system¹⁸. Thus, detection of MS (or MMR protein expression) has become paramount for the selection of patients eligible for ICI treatment. Regarding immunohistochemistry, the VENTANA MMR RxDx Panel was approved by FDA as a companion diagnostic to verify the expression of MMR proteins and, to date, represents the only IHC-based assay to identify patients eligible for anti-PD1 immunotherapy¹⁹. Conversely, a vast array of assays for MSI detection is available, each one with pros and cons (Fig. 2).

Methods for MSI detection

The current practice for MSI detection is based on two major categories of tissue-based testing, an indirect testing strategy based on the evaluation of the MMR complex proteins by immunohistochemistry (IHC) and a direct method based on the detection of alterations in specific microsatellite loci, generally by polymerase chain reaction (PCR) or more recently by next generation sequencing (NGS)²⁷.

MMR COMPLEX ASSESSMENT BY IMMUNOHISTOCHEMISTRY

IHC represents the first-line diagnostic test for the detection of dMMR, due to its availability in almost all pathology laboratories. The test is based on the use of specific antibodies against the four MMR proteins and should be evaluated based on the conservation or loss of their nuclear staining²⁸. Nuclear expression of all four proteins is indicative of a proficient MMR (pMMR) system. On the contrary, loss of staining of at least one of the MMR proteins is informative of a dMMR phenotype. Typically, due to the proteolytic degradation of the whole heterodimer, loss of MLH1 is associated with loss of PMS2, while MSH2 loss is accompanied by MSH6 loss. Nevertheless, isolated loss of MSH6 or PMS2 can happen. IHC presents several advantages including short testing time, analysis of samples with tumor content lower than 20%, and a direct visual correlation between stained cells and their morphology²⁹. Despite being a widespread technique, IHC can be subject to misinterpretation, for example in cases with misrecognized intratumoral heterogeneity for MMR protein expression, or due to staining artifacts³⁰. In some cases, missense or in-frame mutations, especially in the context of LS, may produce an inactive protein that is antigenically identical to the wild-type form causing false positive staining³¹. Taking into account the technical variables and the malignancy heterogeneity, IHC detection of MMR presents a sensitivity between 85% and 100% and specificity between 85% and 92%³². Based on all these considerations, loss of expression of MMR proteins at IHC indicating dMMR should be confirmed with a direct molecular test (MSI-based PCR or NGS) for eligibility to immunotherapy²⁷.

MSI DETECTION BY POLYMERASE CHAIN REACTION (PCR)

MSI testing by PCR-based assays represents the gold standard for molecular detection of MSI. The analysis requires one to a few nanograms of DNA extracted from formalin-fixed paraffin-embedded (FFPE) samples, as template for a PCR reaction, in which fluorescent primers spanning specific marker regions (loci) in microsatellites are employed. The obtained amplicons are separated based on their size by capillary electrophoresis (CE)³³. In the analysis, MSI is reported as extra DNA fragment peaks compared with normal tissue. Thus, for punctual MSI analysis, the presence of normal tissue from the same patient is required (Fig. 3).

MARKER PANELS

The first marker panel proposed for MSI detection (Bethesda Panel, 1997) consisted of two mononucleotide (BAT25, BAT26) and three dinucleotide repeats (D5S346, D2S123, D17S250) and presented sensitivity between 67% and 100% and specificity between 61% and 92%. The accuracy of PCR-based MSI analysis was improved after the introduction of panels that included exclusively mononucleotide repeat markers (BAT-25, BAT-26, NR-21, NR-24, MONO-27), with a specificity of 98.7% and sensitivity of 95.8% (revised Bethesda guidelines, 2002)^27,34. The rationale of this choice is based on the evidence that mononucleotide repeat markers are more specific and sensitive in MSI determination, and, as these markers are quasi-monomorphic in normal DNA, may eliminate the need to employ normal tissue as control^35,36. Of note, the employment of mononucleotide markers partially overcomes the issue of MSH6-deficient cancers. Due to the partial redundancy of the function of MSH6 and MSH3 proteins, when MSH6 is mutated the MSH2/MSH3 dimer is still functioning, limiting MSI to mononucleotide tandem repeats³⁷. Indeed, a study on ECs has demonstrated that the exclusive employment of mononucleotide marker panels correctly identifies 100% of MSH6-defective cases³⁷. Recently, the use of other markers has been suggested. Because sequence instability in microsatellites increases exponentially with the increase of the repeat length, the employment of 52-60 repeated adenine bases defined as long mononucleotide repeats (LMR) has been proposed³⁸. At present, a ‘mixed’ panel consisting of a combination of short mononucleotide repeats (NR-21, BAT-25, BAT-26, MONO-27) and LMRs (BAT-52, BAT-56, BAT-59, BAT-60) is available and has already been tested with good results in pan-cancer cohorts but has yet to be approved as an IVD medical device³⁹. Despite no significative improvements related to LMR employment in MSI detection were found in CRC specimens, major advantages were reported in non-CRC specimens. In fact, LMR markers can detect larger shifts in base size, thus improving sensitivity⁴⁰. Considering the wide availability of MSI markers and panels, clear, universally accepted, cut-off values are required. An empirical cut-off of 30% unstable microsatellite loci has been adopted. Accordingly, three microsatellite statuses have been defined: MSI-H, indicating instability in two or more of the 5 analyzed loci (or > 30% of analyzed loci, when more than 5 are analyzed); MSI-L, when instability at one locus of the 5 investigated loci is detected (or in 10%-30% of analyzed loci); and MSS, indicating no loci with instability (or < 10% of analyzed loci)⁴¹. From previous studies, a large amount of data has been obtained by the analysis of CRC-derived specimens and, consequently, all validated assays are generally calibrated according to CRC samples. This may be an important drawback considering that MSI profiles may differ depending on tumor types. For example, ECs, present smaller mean indels (3 bp vs 6 bp) compared to CRC⁴². Conversely, within the subgroup of gastrointestinal (GI) cancers, nucleotide shifts were the highest in the stomach (mean shift 6.4 nucleotides), followed by colorectum (5.7), small intestine (5.0) and gastroesophageal junction cancers (4.0)⁴³. It is also, important to underline that, although more than 20 microsatellite markers have been proposed in the years, PCR-based panels currently in use are generally based on 5 to 8 markers limiting the opportunity to detect MSI, especially in tumors with low rates of MSI.

NOVEL PCR-BASED MSI DETECTION TECHNIQUES

At present, PCR amplification followed by CE represents the gold standard, although novel PCR-based MSI testing methods have been developed. For example, the recently introduced droplet digital PCR (ddPCR) can be used for MSI detection. Five common mononucleotide repeat markers can be tested by ddPCR, using competitive probes based on drop-off assay that can distinguish differences as little as 2 bp. Moreover, this type of assay does not require normal tissue as control. Using this technology MSI can be detected in diluted specimens such as circulating free tumor DNA. The use of this technique in the liquid biopsy setting has shown high concordance with FFPE samples in both CRC and non-CRC ^44,45. The major pitfall that can be encountered using a CE approach for PCR amplicon detection is the presence of multiple ‘stutter peaks’ surrounding the peak of interest that may cause misinterpretation of the electropherogram. For this reason, several other post-amplification analyses were tested in the years. A highly sensitive and specific method for post-amplification microsatellite amplicon analysis is based on high-resolution melting (HRM). The technique analyses the melting (e.g., the transition from double to single-stranded form) proprieties of the obtained amplicons. Changes in amplicon length, due to microsatellite indels, will produce different melting curves⁴⁶. The Idylla MSI test is based on this principle and is a fully automated real-time PCR assay, in which seven microsatellite markers (ACVR2A, BTBD7, DIDO1, MRE11, RYR3, SEC31A, SULF2) are amplified using a fluorescent-labeled molecular beacon. Obtained PCR amplicons are then analyzed by HRM. The main advantages of the assay are that (i) normal tissue is not required and (ii) the analysis workflow is completely automated, with no additional molecular facilities required. A 5 μm section of FFPE tissue is directly loaded (no dissection is required for cellularity > 20%) on Idylla Cartridge and DNA extraction and PCR amplification are automatically performed in less than 2 hours. Considering that a low DNA input is required and no normal tissue is needed, the assay may be particularly suitable for biopsies. Agreement with IHC is higher than 96% and reaches 100% in samples with optimal tumor cell content (20% or higher)⁴⁷. An alternative testing method is represented by denaturing high-performance liquid chromatography (DHPLC). DHPLC allows amplicon separation based on their size through a solid-phase column. The major limitations of the technology are represented by the need to optimize the column temperature based on the amplicons’ melting temperature to achieve optimal amplicon denaturation⁴⁸. Other techniques, even if still at the research use-only stage, have been investigated and include, for example, the possibility to employ a real-time PCR technique based on peptide nucleic acid (PNA) probes. The high sensitivity of PNA-probes allows to detect alterations in specimens with > 5% cellularity (partially solving the cellularity issue) and, at the same time, the combination with melting curve analysis, for amplicon discrimination, avoids artifacts such as ‘stutter peaks’ typical of CE⁴⁹.

MSI DETECTION BY NEXT-GENERATION SEQUENCING

NGS has recently emerged as an alternative to PCR for MSI detection, partially overcoming the major limit of PCR-based approaches: the limited number (5 to 10) of the evaluated loci. NGS-based MSI can simultaneously detect dozens to hundreds of microsatellite loci, limiting some of the pitfalls typical of MSI detection, including ethnicity-associated variations or variations related to tumor type. NGS can simultaneously detect multiple genetic alterations; thus, MSI testing can be integrated with the evaluation of tumor mutation burden (TMB), and/or clinically relevant gene alterations^27,50. ColonCore (MS plus 36 CRC-related genes)⁵¹, MSIPlus (16 microsatellite loci plus KRAS, NRAS, and BRAF genes mutations)⁵², ColoSeq (MLH1, MSH2, MSH6, PMS2, EPCAM, APC, MUTYH)⁵³ are some examples of integrated panels, available for NGS-based MSI profiling in CRC-specimens. The use of NGS does not require healthy tissue as a control, thanks to the use of appropriate bioinformatic tools. NGS platforms present several advantages, although n unmet issue is still represented by computational analysis. These kinds of platforms require dedicated computational tools and the creation of specific algorithms to determine the percentage of unstable loci. Computational algorithms for MSI score definition can work in three different ways: comparing indels distribution of microsatellites in paired tumor vs healthy specimens (MSIsensor, MANTIS), comparing indels distribution in tumor tissue with a defined reference set on targeted gene panels (mSINGS, MSIFinder), or analyzing the number of single nucleotide variants and indels throughout the genome to detect the hypermutated status (MSIseq)⁵⁴. NGS-based MSI detection is burdened by the drawbacks of NGS-based analysis, including the possibility of introducing errors during the sequencing-by-synthesis. The error rate typically ranges from 0.002% to 2% depending on the repeat region length⁵¹. A number of repeats lower than seven is unlikely to be related to MSI status, but longer mononucleotide repeats (> 15 bp) may be more prone to artifacts due to the sequencing procedure. NGS-based protocols are more expensive in comparison with IHC or PCR and must be carried out in hub centers. The 2022 guidelines for MSI detection from the College of American Pathologists strongly recommended the use of MMR-IHC or MSI by PCR in non-CRC neoplasms, while NGS-MSI assay is allowed only in the CRC setting⁵⁶ (Tab. I).

Agreement between different MMR/MS status detection methods

Several studies have evaluated the concordance rate between IHC and molecular assays, but most have been performed in CRC or EC specimens. The majority of studies reported a discordance rate between PCR-based assays and IHC ranging from 1% to 10%, in the CRC setting⁵⁷. For example, Chen and colleagues analyzed a series of 855 CRC patients, comparing IHC with MSI molecular definition using PCR followed by CE (a panel of 5 mononucleotide loci was considered). The analysis reported 45 discordant cases and a level of concordance equal to 88.8%⁵⁸. Similarly, in a smaller cohort of endometrial carcinomas (333 patients) a concordance between IHC and PCR of 98.8% was reported (Cohen’s K: 0.98), however, in this case, a panel of both mono and di-nucleotide loci was employed⁵⁹. Regarding the concordance between different molecular assays, Dedeurwaerdere and colleagues compared the three main molecular techniques: PCR followed by CE (five mononucleotide loci plus three dinucleotide loci), Idylla MSI assay and NGS (custom panel including 15 loci), in a mixed CRC/EC (57%/43%) cohort⁶⁰. The three molecular techniques were compared with IHC used as a gold standard. In the CRC cohort, all three molecular assays achieved 100% sensitivity and specificity in comparison with IHC in samples with cellularity of 30% or higher, but PCR and NGS failed in a sample with cellularity of 20%, which was sufficient for Idylla MSI assay. In the EC cohort, all three molecular assays achieved a specificity of 100% but showed lower sensitivity. In fact, in this setting, the most sensitive method was NGS (75%) followed by PCR (67%) and Idylla MSI (58%). Limited information is available regarding the concordance among different techniques in gastric cancer. A recent work by Marino et al. reported very low discordance (0.57%; K = 0.957, 95% CI, p = 0.000) between IHC and PCR in gastric cancer specimens⁶¹.

Preanalytical factors affecting MMR/MS evaluation

Regardless of the assay employed, sample adequacy must be always taken into account. Several pre-analytical factors may affect MSI evaluation. An important factor influencing both MMR-IHC and PCR- or NGS-based MSI detection is sample fixation. In IHC, hypo-fixation or hyper-fixation may result in a reduction of immunostaining intensity⁶². On the other hand, fixation issues may result in a reduction of DNA quality in MSI testing due to high DNA fragmentation⁶³. As demonstrated by Malapelle and colleagues, DNA integrity number (DIN)≥ 4 is required for a good concordance between IHC and molecular detection of MSI⁶⁴. In this context, biopsy specimens do not generally have fixation issues compared with surgical specimens⁶⁵. However, surgically obtained tissue presents higher neoplastic cellularity, which is another key point to evaluate in sample adequacy determination. Tumor cellularity equal or higher than 20% is generally required in CRC, while non-CRC specimens may need even higher parameters^42,66. Therefore, small biopsies, especially from metastases, may result in insufficient tumor-DNA content. In the same way, an abundant mucinous component in certain neoplasms may affect DNA yield and consequently MSI determination^67,68. Several clinical and technical aspects must be taken into consideration to avoid erroneous interpretations. Hypoxic conditions may decrease MLH1 expression while neoadjuvant chemo- or radio-therapy may induce the artifactual loss of MSH6 expression in IHC^69,70. In metastatic CRC, Crisafulli and colleagues demonstrated that resistance to alkylating agents (e.g. temozolomide) is accompanied by the onset of MSH6 mutations decreasing the efficiency of MMR machinery⁷¹. If available, pre-operative biopsies should be used for MMR evaluation, otherwise caution must be used when interpreting IHC⁷².

Molecular and biological factors underlying MMR/MS discordance

As previously said, discordance between MMR and MS status can be encountered in clinical practice. The first step is the exclusion of misinterpretation of the IHC and/or molecular assay and intratumor heterogeneity as underlying causes of the discordance. However, other causes have to be taken into account. MSI phenotype is not only associated with dMMR, but also with the ultramutated phenotype due to POLE or POLD1 pathogenetic gene mutations, in the absence of MMR protein alteration⁷³. Furthermore, tumors with MSH6 isolated deficiency are often MSS at molecular analysis by PCR, due to the partial redundancy of the function of MSH6 and MSH3 proteins, as previously described³⁷. These cases can be correctly labeled as MSI only by using a panel with a sufficient number of mononucleotide markers, instead of the classic Bethesda Panel. As reported by Chen et al, pathogenic mutations of MMR genes may lead to the expression of a full-length but nonfunctional protein with retained antigenicity, and thus to a strongly positive staining or in some cases to a positive staining, but weaker than internal control⁷⁴. This phenomenon has been reported in LS cases with missense mutations and in somatic MMR genes double mutation cases.

MSI detection by liquid biopsy

Tissue biopsies present several limitations including the inability to capture tumor heterogeneity and the impossibility to perform longitudinal disease monitoring^75,76. Current European guidelines recommend the use of liquid biopsy for MSI and other predictive biomarkers in the metastatic setting only if an adequate tissue sample is not available and re-biopsy is not possible⁷⁷. Future scenarios include the use of liquid biopsy to cherry-pick patients who benefit from adjuvant therapy and to monitor response and resistance onset to immunotherapy⁷⁸. Circulating tumor DNA (ctDNA) is considered a good analyte to assess MS. Currently, the most robust assays to detect MS in liquid biopsy specimens are droplet digital PCR (ddPCR) and NGS⁷⁸. While ddPCR-based assays may be used both for tissue and liquid biopsies but analyze a limited number of loci (3 to 5 loci), NGS-based panels allow the simultaneous analysis of hundreds of loci but require a specific set-up for plasma-derived specimens. In fact, due to the fragmentation of ctDNA, assays traditionally used for tissue specimens are not suitable for the analysis of ctDNA⁷⁹. Nakamura and colleagues analyzed a cohort of advanced GI cancers using an NGS-based assay (Guardant360), obtaining an overall percent agreement of 98.2% between tissue specimens and ctDNA⁸⁰. It is important to underline that the aforementioned study, as most of the currently available studies aimed at detecting MSI in ctDNA, was focused only on advanced malignancies, that are expected to have high ctDNA shedding. In fact, the fraction of ctDNA among total circulating free DNA (cfDNA) represents a key factor in obtaining a robust concordance between liquid biopsy and tissue specimens.

MSI as a predictive biomarker in GI cancers

MSI was considered for many years as a prerogative of CRC and EC, although about 4% of solid tumors can present MSI-H phenotype and LS has been demonstrated to increase the risk of developing a wide variety of cancers. The approval by the FDA of the anti-PD-1 antibody pembrolizumab for the treatment of any solid tumor with MSI-H/dMMR phenotype has dramatically increased the efforts to identify MS status in those tumors that previously were not routinely screened for MS^81,82. Overall, MSI is very common in GI cancers with a frequency of about 15%⁸³.

GASTROESOPHAGEAL CANCER (GC)

The percentage of GCs presenting MSI ranges from 8 (Asiatic population) to 25% (Western population) depending on ethnicity⁸⁴. In general, MSI phenotype is associated with older age at diagnosis, female gender, intestinal histotype, and earlier stage at diagnosis (stages I-II). MSI GC is considered a distinct GC subtype with specific histopathologic features (mucinous and medullary histotypes and high tumor-infiltrating lymphocytes) and genetic profile (high tumor mutational burden and frequent mutations in PIK3CA, ERBB3, ERBB2, and EGFR)⁸⁵. Most MSI GC are sporadic forms associated with MLH1 promoter hypermethylation, while only 1.6% are LS-related^86,87. MSI GCs are generally characterized by better prognosis with significantly longer overall survival if compared with MSS tumors. The seminal KEYNOTE-059 trial evaluated the efficacy of pembrolizumab in previously heavily pretreated advanced gastric and gastroesophageal junction (GEJ) cancer patients. An objective response rate (ORR) based on radiologic assessment was obtained in 11.6% of the entire pembrolizumab-treated cohort while a notable ORR of 57% was observed in the MSI-H subgroup⁸⁸. Similarly, the trial CheckMate 032, conducted in metastatic chemotherapy-refractory gastric/GEJ cancer patients, evaluated the efficacy of the anti-PD-1 nivolumab and reported an ORR of 29% in the MSI-H group against 11% in the MSI-L/MSS group⁸⁹. Subsequently, the association of ICIs to chemotherapy was tested in the first-line setting. The phase III CheckMate-649 study, which investigated the efficacy of adding nivolumab to chemotherapy in previously untreated advanced or metastatic HER2-negative gastric/GEJ adenocarcinoma, showed the benefit of adding nivolumab to chemotherapy for patients with a combined positive score (CPS) > 5 and it is at present, the study with the longest follow-up survival data. Three-year follow-up data demonstrated a greater benefit of adding immunotherapy to chemotherapy in the MSI-H subgroup compared to the MSS group in both the overall population (38.7 vs 12.3 months, [HR 0.34, 0.16 to 0.74] and 13.8 vs 11.5 months [HR 0.79, 0.70 to 0.89]) and even greater in the patients with PD-L1 CPS≥5⁹⁰.

Moving to the earlier setting, some studies explored the role of immunotherapy in the neoadjuvant/adjuvant setting with no convincing results in the all comers. The KEYNOTE-585 was the first phase III study that demonstrated a significant benefit of association of the anti-PD1 pembrolizumab to standard chemotherapy in locally advanced gastric/GEJ cancer. The benefit was impressive in the MSI-H group with a reduction in the risk of death of 40% compared to 10% in the MSS group⁹¹. Based on the observation by Pietrantonio and colleagues that locally advanced MSI-H GEJ/GC could benefit less from standard perioperative chemotherapy, dedicated trials to this immunosensitive GC/GEJ population were conducted⁹². In the phase II NEONIPIGA trial, patients with locally advanced MSI-H gastric/GEJ adenocarcinomas receiving the combination of nivolumab and the anti-CTLA4-ipilimumab showed an impressive high rate of pathological complete response (pCR: 59%)⁹³. Similar results were achieved in the proof-of-concept INFINITY phase II trial where the activity and safety of neoadjuvant short-term tremelimumab combined with durvalumab was tested in MSI-H/dMMR, EBV-negative resectable gastric/GEJ cancer. A notable pCR rate of 60% and major-complete pathological response (< 10% viable cells) of 80% were reported⁹⁴. In the same setting, the PANDA study tested one cycle of atezolizumab monotherapy followed by four cycles of atezolizumab plus docetaxel, oxaliplatin and capecitabine, showing a pCR in both the two patients presenting dMMR phenotype who underwent to surgery, compared to 39% in the pMMR group⁹⁵. Globally these encouraging results support the use of immunotherapy in the neoadjuvant/perioperative setting in locally advanced resectable MSI-H/dMMR gastric/GEJ cancers and suggest the possibility to omit surgery in highly selected patients. However, further exploration and validation in larger cohorts of patients are necessary.

COLORECTAL CANCER

The frequency of MSI in sporadic CRC is estimated to be about 10-15% and MSI is generally associated with right colon tumor localization, mucinous histotype, lymphocytic infiltration, and advanced stage (IV)⁹⁶. MSI phenotype is a good prognostic factor in early-stage (II-III) CRC, while in stage IV it is associated with worse prognosis. MS in CRC is also considered an important predictive factor of response to chemotherapy and ICI treatment. MSI patients do not benefit from 5-FU adjuvant monochemotherapy but are responsive to combination treatment. As for immunotherapy, the first studies conducted in non-selected CRC demonstrated very limited effects, but the selection of patients based on MSI dramatically changed the response to treatment, obtaining unprecedented high and durable response rates, and impressive survival improvements in the MSI-H mCRC^97,98. The first study that highlighted the predictive role of MSI-H/dMMR was the study by Le DT and colleagues in which no response was observed in pMMR compared to 40% in MSI mCRC²⁶.

These promising results were then confirmed by more extensive studies, including the phase II KEYNOTE-164 in pretreated mCRC and KEYNOTE-177 in the first line which defined a new standard of treatment in this setting^99,100. The multi-cohort study CheckMate142 explored the use of monotherapy with nivolumab and the combination of ipilimumab plus nivolumab, highlighting the benefit of this combination in both previously treated and untreated mCRC. Recently, the results of the phase III trial CheckMate 8HW confirmed the impressive benefit of ipilimumab plus nivolumab compared to investigators’ choice standard treatment with a median progression-free survival that was not reached in the ICI arm and a 12-month PFS rate of 79% compared to 21% in the chemotherapy arm.

The remarkable results obtained in MSI-H mCRC patients have led to a growing interest in the possible applications of immunotherapy in earlier settings^101-103. In the NICHE study, a combination of nivolumab and ipilimumab was employed in stage I-III resectable CRCs. The study enrolled 35 patients, including 20 who presented a dMMR phenotype and 15 pMMR CRCs. A pathological response after surgery was observed in all dMMR patients, compared to 27% among the pMMR CRC patients. On this basis, the NICHE-2 study enrolled 115 high-risk stage II or stage III MSI-H/dMMR CRCs to receive one cycle of nivolumab plus ipilimumab and a second cycle of nivolumab followed by surgery. The primary endpoint was met with an amazing 98% of pathological response, 95% major (< 10% residual viable tumor), and 68% of pCR respectively, with a very favorable safety profile¹⁰⁴. At the ESMO Congress, the primary endpoint of 3-year disease-free survival was presented and none of the patients relapsed¹⁰⁵. Comparable results in terms of pCR were observed in smaller studies. The first employed the anti-PD1 toripalimab alone or in combination with celecoxib and a pCR rate of 65%, 59%, and 52.8%, respectively^106-109. The UNICORN study (NTC05845450), an academic window-of-opportunity umbrella platform trial of short-course pre-operative targeted treatments in patients with highly molecularly selected and resectable CRC, is ongoing. In this study the two cohorts of dMMR/MSI-H cancers, treated with the anti-CTLA4 botensilimab and botensilimab plus the anti-PD1 balstilimab respectively, have recently completed the planned accrual.

Locally advanced rectal cancer (LARC) requires a multimodal treatment with a combination of chemotherapy, radiotherapy, and surgery in different sequences^110,111. However, dMMR/MSI-H LARC may not share the same benefits from neoadjuvant induction chemotherapy as their MSS counterpart¹¹². The phase II study conducted by Cercek and colleagues defined a new standard. Six months of neoadjuvant therapy with the anti-PD1 dostarlimab reached 100% of clinical complete in MSI-H/dMMR stage II-III rectal adenocarcinomas¹¹³. No patients underwent radio-chemotherapy or surgery and no relapse was observed at the data cut-off. Finally, promising data derived from the combination of botensilimab with balstilimab in the NEST-1 trial¹¹⁴. MMR/MSI assessment is mandatory and crucial on all CRC specimens regardless of stage and should be primarily conducted on biopsy specimens at first diagnosis.

PANCREATIC DUCTAL ADENOCARCINOMA (PDAC)

Approximately 1-2% of PDAC are MSI-H/dMMR^115-117. MSI status is more often associated with medullary and mucinous/colloid histology and the absence of KRAS and TP53 mutations^115,118. Recently, also signet-ring variant of PDAC has been described as an additional histology potentially enriched in MSI¹¹⁹. According to preliminary data, the percentage of MSI-H/dMMR rises to 6.9%, when intraductal papillary mucinous neoplasm (IPMN)-associated PDACs are considered¹²⁰. PDACs are part of LS-associated spectrum cancers. In fact, LS patients present a 9-fold higher risk of developing pancreatic cancer compared with the general population¹²¹. Unfortunately, the role of MSI as a predictive marker in this cancer is still highly debated: on the one hand, Riazy and colleagues demonstrated that no survival advantage is appreciable in MSI-H/dMMR patients treated with 5-FU or gemcitabine-based therapies; on the other hand, Cloyd and colleagues demonstrated that in the metastatic setting, MSI-H/dMMR patients had a better outcome compared with their MSS counterpart when treated with FOLFIRINOX (16.5 months vs 11.1 months median OS)^122,123. NCCN guidelines (2020), recommend molecular MSI definition or MMR testing for ICI treatment in advanced PDAC¹²⁴. Although limited, preliminary data on ICI-based treatment in MSI-H/dMMR PDAC are promising. In the expanded proof-of-concept study by Le and colleagues 12 different MSI-H/dMMR cancer patients, including some PDACs, received pembrolizumab with a notable benefit in term of ORR and survival, supporting the hypothesis that MSI-H/dMMR cancers could be sensitive to ICIs, regardless the origin of tumor’s tissue. The KEYNOTE-158, phase II study investigated the efficacy of pembrolizumab in patients with previously treated, advanced non-colorectal MSI-H/dMMR cancers. The study enrolled 233 patients with 27 tumor types including 22 patients (9.7%) with pancreatic cancer. In the entire population, the ORR was 34.3% (95% CI, 28.3% to 40.8%), the median PFS was 4.1 months (95% CI, 2.4 to 4.9 months) and the median overall survival was 23.5 months (95% CI, 13.5 months to not reached). These results supported the tumor-agnostic approval of pembrolizumab for the treatment of advanced MSI-H/dMMR cancers. A clinical benefit was observed even in PDAC but was less evident (ORR 18.2%, 5.2 to 40.3; mPFS 2.1 month, 1.9 to 3.4; mOS 4.0 months, 2.1 to 9.8; and median duration of response 13.4 months, 8.1 to 16.0+) probably due to the intrinsic scarce immunogenicity in the pancreatic microenvironment and taking into account the later lines setting¹²⁵. Moreover, Taieb and colleagues investigated a cohort of 31 advanced MSI-H/dMMR PDACs and achieved an ORR of 48.4% and stable disease in 19.4% of patients¹²⁶. In another retrospective cohort, 55 MSI-H/dMMR PDACs cases were described with a high rate of LS (58%) and a durable control disease in 17 patients who received ICIs (median ICI duration 27.7 months (95% CI, 11.5 to not reached); disease control rate 80%)¹²⁷. Some studies focused in the identification of potential issues that may cause resistance to ICI in MSI-H/dMMR PDAC, among them the enrichment in B2M and JAK gene mutations in tumors and the presence of intratumor heterogeneity were identified¹²⁸.

BILIARY TRACT CANCERS

Biliary tract cancers (BTCs) are a heterogeneous class of malignancies with different histopathological and genetic features, resulting in different clinical behaviors. The class encompasses gallbladder cancers (GBCs) and cholangiocarcinomas (CCAs), which are classified as intrahepatic (iCCA) and extrahepatic cholangiocarcinoma (eCCA), depending on the origin site of the malignancy. According to revised Bethesda Guidelines for LS, BCTs are included within LS-associated tumors. Data on MS status in BTCs are limited by their heterogeneity and rarity¹²⁹. When considering the whole BTC group, MSI/dMMR prevalence in the Caucasian population is less than 3%, although dramatically higher rates (up to 30%) are reported in certain regions, such as Thailand. Indeed, MSI status in CCA appears to be associated with liver fluke infection¹³⁰. A study by Eluri and colleagues analyzed the largest cohort of iCCA (7565 cases) available in the literature and reported a prevalence of MSI status of 1.8%, while another study conducted exclusively on resection specimens and considering perihilar CCA (pCCA) reported a prevalence of 2.4%^131,132. A higher MSI rate was found in GBCs in which MSI is seen in 7.8% of cases, generally associated with MSH2 loss¹³³. Considering the low rate of resectable BTCs in daily practice, the work of Kai and colleagues is of great interest. Surgical resection, biopsy, and fine needle aspiration (FNA) specimens from BTCs were compared demonstrating that MSI determination can be performed successfully in 98.3% of samples, even in cases in which tissue availability is limited¹³⁴. In the abovementioned KEYNOTE-158 study, a notable ORR of 40.9% and a mOS of 24.3 months were reached in MSI-H/dMMR previously treated advanced BTC patients in which the survival from diagnosis of advanced disease is about 12 months, pointing out the need for upfront MMR/MSI status evaluation at diagnosis¹²⁵. Moreover, in another multicenter observational study focused patients with hepatobiliary (HBP) and pancreatic cancer, pembrolizumab led to an ORR of 73% for BTCs compared to 30% for PDACs¹³⁵.

Other gastrointestinal malignancies

Unfortunately, in other rare gastrointestinal malignancies data are still lacking regarding the predictive role of MSI status.

Tumors arising in the ampulla of Vater (AVC) are rare but occur relatively frequently in patients with LS and are generally associated with MSH6 germline mutations¹³⁶. Across different studies, the prevalence of MSI in AVC is approximately 18% and a strong association with the intestinal AVC subtype has been reported^116,137,138. MSI/dMMR is associated with better prognosis and MSI patients who undergo resection present a lower rate of recurrent disease¹³⁹. Interestingly, in AVC, MSI is not restricted to the intestinal phenotype but can be found also in pancreatobiliary and mixed histologies^116,140.

Characterization of MS in hepatocellular carcinoma (HCC) is still an open issue and available data are heterogeneous and highly inconclusive. MSI-H/dMMR cases in HCC seem to be rare, with a prevalence of less than 3%¹⁴¹. However, different prevalence values are often reported in the Asiatic and European populations, probably due to the different etiologies of HCC¹⁴². Inflammatory events (e.g. cirrhosis and viral infections such as HCV, HBV) may induce dysfunctions of the MMR complex¹⁴³. Moreover, as reported in a study comparing mononucleotide repeat alterations in HCC and in CRC, variations observed in HCC are less extensive than in CRC ¹⁴⁴.

Finally, data on MSI-H/dMMR in neuroendocrine neoplasms (NENs) are limited. Studies performed on gastric and colorectal neuroendocrine carcinomas (NECs) and mixed neuroendocrine-non neuroendocrine neoplasms (MiNeNs) reported a prevalence of up to 15%, mainly due to MHL1 promoter methylation. In these tumors, MSI was associated with a more favorable prognosis^145,146. Conversely, MSI-H/dMMR is rare in neuroendocrine tumors (NETs) of the small intestine and pancreas and cases of NETs in patients with LS are mostly anecdotal^147,148.

Conclusions

Since the FDA approval of pembrolizumab for the treatment of patients with unresectable or metastatic MSI-H/dMMR solid tumors, who progressed following previous treatment, MSI has become the first true agnostic biomarker, and its evaluation has become a milestone of tumor molecular characterization. IHC represents the first-line diagnostic test for dMMR detection, due to its availability in almost all pathology laboratories and its low costs. If dMMR status is unclear, MSI evaluation by molecular analysis must be performed. In this setting, the pathologist plays a central role in therapeutic decision-making. The pathologist must perform an accurate assessment of MMR protein expression and should be aware of potential pitfalls and misinterpretations to guarantee the patient’s access to the best treatment. The pathologist in charge of molecular diagnostics should also be responsible for the evaluation of the adequacy of the specimen for molecular analysis, for choosing the best assay and panel, and for providing the clinicians with a clear and concise report and additional comments to facilitate its interpretation. Among GI cancers, MSI-H/dMMR status and its detection methods have been widely studied and developed in the CRC setting. However, there are still open issues and questions to be addressed in non-CRC GI malignancies. Future perspectives include: i) the optimization of MMR/MSI detection in biopsy specimens for the use of ICIs in the neoadjuvant setting; ii) the use of liquid biopsy as a non-invasive tool for MSI evaluation; and iii) the development of NGS panels for a reliable definition of MSI status in non-CRC malignancies.

FUNDING

MF and CL are supported by European Union - Next Generation EU, Mission 4 Component 1 CUP: C53D23002200006. MF is also supported by European Union - Next-Generation EU, PNRR M4C2-Investimento 1.4- CN00000041, and by the Department of Medicine (BIRD_2024).

CONFLICTS OF INTEREST STATEMENT

MF reports research funding (to institution) from Roche, Astellas, and Diaceutics; personal honoraria as an invited speaker from Roche, Astellas, AstraZeneca, Lilly, Incyte, Bristol Myers Squibb, Agilent, Merck Serono, Pierre Fabre, GlaxoSmithKline, Novartis, and Amgen; and for participation in advisory boards of Amgen, Astellas, Roche, Merck Serono, GlaxoSmithKline, Novartis, and Janssen.

AUTHORS CONTRIBUTIONS

All authors equally contributed in the writing and conceptualization of the manuscript. All authors have read and agreed to the published version of the manuscript.

History

Received: October 2, 2024

Accepted: October 3, 2024

Figures and tables

Figure 1. Mechanism of mismatch repair. MSI (microsatellite instability) is characterized by a change in the tandem repeats caused by DNA slippage during DNA replication. DNA slippage is due to temporary dissociations between the DNA polymerase and the DNA-template strand, and results in loss or gain of repeats, when the reassociation occurs ahead or behind where it left off, respectively. The basic process of MMR (mismatch repair) can be divided into four steps: (1) MSHs (MutS homologs) proteins recognize the mismatch and stably bind the double-strand DNA (dsDNA), where the error occurs; (2) MLHs are recruited to bind the MSHs-mismatch complex (MutL homologs: MLHs, act as mediators between the MSHs and the MMR effector enzyme endonuclease MutH); (3) endonuclease MutH cleaves at GATC sites generating a ‘nick’ in the DNA which allows the excision of the mismatch; (4) the removed DNA sequence is resynthesized by DNA polymerase using the remaining strand as a template.

Figure 2. The history of microsatellite status in cancer clinical practice. Since 1993, when three distinct groups reported the presence of repeated sequences in certain tumors, microsatellite instability (MSI) acquired a key role in cancer therapeutics^20-22. In 1995, Drummond et al. demonstrated that MSH6 is a key protein of the MMR complex and is involved in the repair of single-nucleotide mismatches, while two years later (1997) Kane et al., reported that MLH1 promoter methylation is one of the major causes of dMMR in sporadic CRCs^23,24. In the same year, the Bethesda Guidelines reported criteria for the identification of colorectal cancers that needed to undergo testing for MSI. Guidelines also included a list of recommended markers (two mononucleotide and three dinucleotide sequences) defined as Bethesda Panel. Guidelines were revised in 2004. In 2002 Suraweera et al. proposed five quasi-monomorphic mononucleotide markers for MSI detection and two years later the first commercial kit was available²⁵. In the following years, several technical improvements were proposed, including high-resolution melting (HRM) to analyze microsatellite PCR products (2010), long mononucleotide repeats (LMR) as possible MSI markers (2015), and finally, the possibility of testing MSI in combination with mutations in genes (KRAS, NRAS, and BRAF) relevant for CRC patient therapeutic management by NGS. In 2015, Le and colleagues demonstrated that MSI is associated with a higher objective response to PD-1 blockade therapy in colorectal cancers²⁶. In 2017 FDA approved pembrolizumab for the treatment of patients with unresectable or metastatic MSI solid tumors that progressed after prior treatment. Three years later (2020), satisfactory results of the KEYNOTE-177 study led the FDA to approve pembrolizumab as first-line treatment for unresectable/metastatic MSI-H colorectal cancers.

Figure 3. Most common techniques used in clinical practice for mismatch repair deficiency (dMMR)/ microsatellite instability (MSI) identification. Immunohistochemistry for MMR proteins is a fast and widespread method to identify dMMR and is the first-line test to be performed. The longest-standing technique for MSI detection is based on PCR amplification of specific DNA-repeat sequences (loci) using fluorescently labeled primers. Panels of possible loci to be amplified include mononucleotide and dinucleotide short sequences and, more recently, also long mononucleotide repeats (LMR). PCR products can be analyzed by capillary electrophoresis (CE) or high-resolution melting (HRM). The introduction of next-generation sequencing (NGS) techniques has allowed the analysis of several loci simultaneously.

Table I. Advantages and disadvantages of MS/MMR detection techniques.

Technique	Mechanism	Advantages	Disadvantages
IHC	Analysis of the four proteins of the MMR complex using specific antibodies.	WidespreadShort testing timeLower costsCorrelation with morphologic features	More prone to subjective interpretationPossible artifacts due to pre-analytical issues (e.g. sample fixation)Possible false positive staining in case of missense or in-frame mutations
PCR amplification + CE	Amplification of specific loci by PCR followed by capillary electrophoresis of the amplified sequences	Gold standard for molecular MSI detectionLow DNA inputDirect detection of alterations in DNA caused by dMMR	Requires specific instrumentationLonger turnaround timesResults can be influenced by the loci analyzed in the panelLow number of analyzed loci (5 to 10)Healthy tissue is required for each tumor as controls
PCR amplification + HRM	Amplification of specific loci by PCR followed by high-resolution melting analysis of the amplified sequences	Possibility of a fully automated protocolOnly one FFPE section is requiredLow turnaround timesHealthy tissue is not required as control samples.	Higher costsCellularity of the specimen may affect final resultsOnly 7 loci are analyzed
NGS	Sequencing of dozens of loci using NGS-based techniques	Possibility to analyze dozens of lociPossibility to analyze simultaneously MS and other tumor-specific genetic alterationsPossibility to identify novel loci in MSI non well characterized tumors	Higher costSpecific bioinformatic tools for computational analysis are requiredPossibility to introduce errors during the sequencing-by-synthesisHigher DNA input is required
IHC: immunohistochemistry; MMR: mismatch repair; PCR: polymerase chain reaction; CE: capillary electrophoresis; MSI: microsatellite instability; HRM: high resolution melting; MS: microsatellite status; NGS: next generation sequencing.