Background
Microsatellite instability (MSI) is characterized by the accumulation of insertion-deletion mutations at microsatellite-repeat sequences and represents a hallmark feature of cancer cells with DNA mismatch-repair deficiency (dMMR) [
1,
2]. Inactivation of any one or a combination of MMR genes, including
MutL homolog (MLH)1,
MutS protein homolog (MSH)2,
MSH6, and
PMS2, can result in MSI. Originally, MSI was discovered to correlate with germline defects in MMR genes in patients with Lynch syndrome, where > 90% of colorectal cancer (CRC) patients exhibit this phenotype [
3,
4]. It was later recognized that MSI also occurs in ~ 12 to ~ 15% of sporadic CRCs that lack germline MMR mutations; however, in these patients, MSI manifests due to methylation-induced silencing of the
MLH1 promoter [
5,
6]. Determination of MMR deficiency by MSI status or immunohistochemical staining for MMR proteins in CRC patients has clinical significance due to its prognostic and therapeutic implications [
7]. Patients with MSI CRCs typically have better prognosis, although these cancers are less responsive to 5FU-based adjuvant chemotherapy [
8]. Recently, clinical trials demonstrated the utility of MSI status in predicting response to PD-1 blockade in advanced unresectable solid tumor patients [
9‐
11]. Additionally, MSI status was a significant predictor of the immune-related objective response rate [40% in dMMR CRC, 71% in dMMR non-CRC, 0% in MMR-proficient (pMMR) CRC] and immune-related progression-free survival rates (78, 67, and 11%, respectively) [
9].
The procedures and criteria used to determine MSI in tumors are constantly evolving; however, there remains a lack of consensus regarding the most practical and robust MSI assay allowing for inexpensive clinical use and capable of providing consistent and reproducible results in laboratories worldwide [
12]. In an effort to unify MSI analysis in CRC patients, in 1997, a National Cancer Institute (NCI) workshop recommended the use of a reference panel of five markers: two mononucleotide-repeat markers (BAT26 and BAT25) and three dinucleotide-repeat markers (D2S123, D5S346, and D17S250) [
13]. In a follow-up NCI workshop, the panel recognized that the original markers had limitations, primarily due to the inclusion of the three dinucleotide-repeat markers [
14]. First, it was noted that the dinucleotide-repeat markers were more suitable for identifying MSI-low tumors, whereas mononucleotide-repeat markers were more specific and sensitive for the determination of MSI-positive CRCs [
15]. Second, due to the polymorphic nature of dinucleotide markers, these required PCR amplification of both the tumor DNA and matching normal specimens from each individual in order to interpret the results. Third, the conventional NCI-panel markers inadequately identified MSH6-deficient CRCs. Employing a panel of five quasi-monomorphic mononucleotide-repeat markers in a pentaplex PCR obviated the need for obtaining normal DNA from each CRC patient and offered better specificity and sensitivity relative to the NCI-panel markers [
16]. Unfortunately, despite its obvious strengths, the pentaplex MSI approach gained limited acceptance for MSI-based screening of CRC patients, possibly due to a lack of clear understanding of the technical aspects of the assay and a paucity of data enabling its validation in independent laboratories. To address this concern, we previously performed a comprehensive determination of the accuracy of the pentaplex-panel markers in a large series of dMMR and pMMR CRCs by analyzing the PCR-amplified profiles of each marker in both tumor and matching normal DNA [
17]. Based on the results of that study, we found that a smaller panel of only three markers, BAT26, NR21, and NR27, were adequate or better in identifying dMMR CRCs as compared with the original panel of five mononucleotide markers [
17]. However, despite various practical and technical strengths of this panel of MSI markers for identifying MSI-positive CRCs, one of the limitations of these markers was their lack of robustness in identifying CRCs exhibiting MSH6-deficiency [
17]. These data highlighted the need for developing a more robust assay capable of addressing this important issue and successfully identifying CRCs lacking MSH6.
Therefore, in the present study, we examined a panel of four quasi-monomorphic mononucleotide-repeat markers (BAT26, NR21, NR27, and CAT25) amplified in a single multiplex PCR reaction (Tetraplex) to determine its performance in detecting dMMR CRCs. This assay was first performed in a cohort of 318 CRC specimens, comprising 105 dMMR and 213 pMMR cases. Because the frequency of MSI tumor is the highest in the endometrium [
2], we also analyzed the performance of our MSI assay in another cohort of 138 specimens with endometrial cancer (EC) exhibiting known MMR status.
Methods
CRC specimens in the test cohort
Matched germline and tumor DNA from 212 CRC patients diagnosed as pMMR by immunohistochemical (IHC) staining were collected from patients at the Okayama University Hospital (Okayama, Japan). Specimens from CRC patient with tumors diagnosed as dMMR (105) were also collected at three different institutions, including: (1) Baylor University Medical Center (Dallas, TX, USA); (2) University of Heidelberg (Heidelberg, Germany); and (3) Okayama University Hospital (Okayama, Japan). This cohort of 105 dMMR CRC tumors included 45 MLH1-deficient (dMLH1), 45 MSH2-deficient (dMSH2), and 15 MSH6-deficient (dMSH6) tumors. Tumor DNA was extracted from serial sections (5 μm) from the 105 formalin-fixed paraffin-embedded (FFPE) tumor tissues. FFPE samples were routinely stained, and representative tumor regions were identified for DNA extraction by microscopic examination. Genomic DNA was isolated from paraffin-embedded tissues using the QIAamp DNA mini kit (Qiagen, Valencia, CA). The Institutional Review Board at each of the three institutions granted approval for this study.
EC specimens for the validation cohort
A total of 138 tumor samples were collected from EC patients at Okayama University Hospital (Okayama, Japan). Among these, 23 were dMLH1, eight were dMSH2, eight were dMSH6, and two were PMS2-deficient. Tumor DNA was collected and extracted as described for the CRC tumors. The Institutional Review Board at Okayama University Hospital granted approval for this study.
MMR protein IHC
We examined MMR-protein expression for the MLH1, MSH2, MSH6, and PMS2 proteins in primary tumors from 317 CRC and 138 EC patients by IHC. Thin (5 µm) sections of representative blocks were deparaffinized and dehydrated using an ethanol gradient. Following antigen retrieval in citrate buffer (pH 6.0), endogenous peroxidase was blocked with 3% H2O2. Thereafter, slides were incubated overnight in the presence of purified mouse monoclonal antibodies against MLH1 (clone G168-15; 1:50; BD Pharmingen, San Diego, CA, USA), MSH2 (clone G219-1129; 1:200; BD Pharmingen), MSH6 (clone 44/MSH6; 1:100; BD Pharmingen), and PMS2 (clone A16-4; 1:200; BD Pharmingen). Additional incubations were performed with a biotin-conjugated secondary antibody (Vector Laboratories, Burlingame, CA, USA), the avidin–biotin–peroxidase complex (Vector Laboratories, Burlingame, CA, USA), and with biotinyl tyramide, followed by streptavidin peroxidase. Diaminobenzidine was used as a chromogen, and hematoxylin was used as a nuclear counterstain. Tumor cells were scored as negative for MMR-protein expression only if the epithelial cells within the tumor tissue lacked nuclear staining while the surrounding stromal cells were positive for MMR staining. Tumor tissue with all MMR proteins present were defined as pMMR, and those showing deficiency in at least one of the four MMR proteins were defined as dMMR.
Tetraplex system and quasi-monomorphic variation range (QMVR) definition
MSI analysis was performed using four mononucleotide-repeat microsatellite targets (CAT25, NR21, NR27, and BAT26) in a single multiplex PCR reaction (Tetraplex). Primer sequences are shown in Additional file
1: Table S1, and each sense primer was end-labeled with one of the following fluorescent markers: PET, NED, VIC, or 6-FAM. PCR conditions for the Tetraplex assay consisted of an initial 15-min denaturation step at 95 °C, followed by 35 cycles at 95 °C for 30 s, 55 °C for 30 s, and 72 °C for 30 s, with a final extension at 72 °C for 10 min. Amplified PCR products were diluted with formamide and subjected to electrophoresis using an Applied Biosystems 310 Avant automated capillary electrophoresis DNA sequencer (Applied Biosystems, Foster City, CA, USA). Allelic sizes for each of the markers were determined using GeneMapper 3.1 software (Applied Biosystems).
Determination and validation of the QMVR for each of the four MSI markers was performed by individually scoring PCR-amplification profiles, and the size of both alleles was determined for each marker and for each tumor individually as described previously [
17,
18].
Statistical analyses
We used logistic regression analysis to examine the diagnostic performance of measuring MMR status in CRCs by utilizing different strategies to define MSI. Analyses were performed using JMP (v10.0.2; SAS Institute, Cary, NC, USA).
Discussion
Colorectal cancer (CRC) is believed to initiate as a benign adenomatous polyp, which subsequently develops into an advanced adenoma with high-grade dysplasia, and finally progresses to an invasive cancer. The clinical challenge remains a better understanding of the molecular basis of an individual’s susceptibility for developing CRC, and to determine factors that initiate development of tumor, drive its progression, and determine its responsiveness to antitumor agents. By Through the efforts of The Cancer Genome Atlas (TCGA), CRCs can now at least be classified into the two clusters; a hypermutator and a non-hypermutator phenotype [
19,
20]. The hypermutated CRCs are also categorized into the following two subsets; tumors lacking DNA repair due to the mutations in the exonuclease domain of
DNA polymerase E (
POLE) characterized by an ultramutator phenotype, and a larger proportion of tumors with DNA mismatch repair deficiency (dMMR) and ensuing microsatellite instability (MSI) phenotype. This alterations in the
POLE gene are distinct from the better-known dMMR which results in the classic MSI phenotype [
20].
Recently, several studies have demonstrated that MSI is a positive predictor for immune-checkpoint blockade [
9‐
11]. Hence, MSI analysis is now becoming important not only for the screening of Lynch Syndrome patients, but has a much larger role in identifying tumors exhibiting such a hypermutator phenotype that might respond to immune-checkpoint drugs. This includes analysis of sporadic as well as hereditary cases, of cancers with defects in the MMR system. The availability of a robust MSI assay that is fast, cost-efficient, and highly accurate for identifying dMMR in solid tumors is critical for its successful application in the clinic and research. In this study, we describe the development and application of a rapid and accurate MSI assay, which uses a single PCR reaction for the amplification of four mononucleotide microsatellite markers. This assay was subsequently validated for its usefulness in identifying MSI status in a series of pMMR and dMMR CRCs, as well as in ECs.
Standard MSI analysis using an NCI panel consisting of five microsatellite markers (two mononucleotide and three dinucleotide repeats) remains the preferred method in most clinical and research laboratories [
14]. This is unfortunate, given that multiple studies have repeatedly shown that dinucleotide repeats are better suited to detecting MSI-low tumors, of which most are pMMR [
14,
15], whereas mononucleotide MSI markers offer higher accuracy for detecting dMMR tumors [
16].
We previously illustrated the usefulness of a pentaplex PCR system consisting of five mononucleotide markers, BAT25, BAT26, NR21, NR24, and NR27 [
17]. A pairwise correlation and hierarchical-clustering analysis in this study clearly showed the weakest predictive values associated with NR24 and BAT25 as compared to the remaining three markers (BAT26, NR21, and NR27). Our observation of high sensitivity and positive predictive value with the reduced panel of three markers (BAT26, NR21, and NR27) versus the use of a panel consisting of all five markers has economic implications for MSI-based assays, as the use of this smaller marker panel might result in lower CRC-screening costs in the future [
17]. One of the limitations of the NCI-panel markers is their inability to identify dMSH6 CRCs. Unfortunately, our previous study indicated that, even with the use of mono-markers in the pentaplex panel, the sensitivity of the assay to detect dMSH6 CRCs remained relatively low relative to dMMR due to the loss of other MMR proteins in CRCs [
17]. This is significant, because the MutSα, a heterodimeric complex consisting of MSH2 and MSH6, preferentially recognizes base/base mismatches, as well as small insertion/deletion loops, containing one or two unpaired nucleotides in the DNA sequence, a subsequently participates in the repair of these lesions [
21]. Therefore, one would expect that the functional loss of MutSα due to dMSH6 would lead to preferential instability in loci containing mononucleotide repeats [
22]. Therefore, in this study, we included another mononucleotide marker, CAT25, to improve the sensitivity of our assay for identifying dMSH6 tumors [
23]. In the CRC-validation set, our new four-marker panel detected dMSH6 tumors with a sensitivity of 86.7% (13 of 15 dMSH6 CRCs). This level of accuracy was also observed in the EC test set, where the system detected dMSH6 tumors with a sensitivity of 75.0% (6 of 8 dMSH6 ECs).
During the development of an assay exhibiting higher sensitivity for dMMR tumors, we considered that the most important factor for developing a more sensitive MSH6 marker would be the QMVR range. Indeed, the normal QMVR for the four markers used in this study was three to four bp. This robust range of QMVRs might improve the sensitivity for dMMR tumors, especially dMSH6 tumors. With regards to potential limitations, in our study the test set did not include any PMS2-deficient tumors. In addition, the sensitivity of this assay for detecting MSH6-deficiency is still under 90% in the both test and validation set.
Authors’ contributions
YT performed MSI and MMR IHC staining of CRC materials and drafted the manuscript. TN and AG assisted with data interpretation, designed the project, secured the funding, and drafted the manuscript. AN and YM performed MSI and MMR IHC staining of CRC materials and extracted DNA from all materials. TH and JH performed MSI and MMR IHC staining of EC materials. KN and TF obtained patient samples and clinicopathological data. CRB assisted with data interpretation and revised the manuscript. All authors read and approved the final manuscript.