Improving the performance of polymerase chain reaction for microsatellite instability testing in endometrial cancer Authors: Marta Mendiola, Victoria Heredia-Soto, Amparo Baillo, Ignacio Ruz-Caracuel, Rocio Mena, Alberto Berjón, Francisco Javier Escudero, Jorge Pedregosa, Alicia Hernandez, Jaime Feliu, David Hardisson, Andres Redondo Published: 2025 ABSTRACT Between 20% and 30% of endometrial cancer (EC) cases show mismatch repair deficiency (dMMR), and its characterisation is recommended in these tumours for molecular classification, screening of Lynch syndrome, and as a predictive biomarker for immunotherapy. The aim of this study was to explore two tests developed by Promega (OncoMate MSI Dx Analysis System and long mononucleotide repeat (LMR) microsatellite instability (MSI) analysis system). DNA from 126 EC tumours had been screened for MMR status by immunohistochemistry (IHC). Overall, 67 (53.2%) dMMRs and 59 (46.8%) proficient (pMMR) were included. The same cases were additionally explored for MMR genomic status, with 55 (43.7%) cases presenting an altered genomic pattern, and 69 (54.8%) with no genomic alterations. There were 71 (56.3%) microsatellite stability (MSS) cases for OncoMate and 69 (54.8%) for LMR, and 37 (29.4%) microsatellite instability (MSI) cases for OncoMate and 44 (34.9%) for LMR. Differences between the test assignments were significant (p < 0.001), with an increased proportion of correctly classified cases for the LMR assay, taking the IHC result as the reference. The respective sensitivity and specificity of the LMR assay was 95.5% and 68.1%, versus 86.5% and 53.5% for the OncoMate assay. In conclusion, the new LMR MSI Analysis System had a higher correlation with IHC, including cases that could be misdiagnosed due to minimal shifts, as well as higher sensitivity and specificity than the OncoMate panel. The best method regarding the use of dMMR/MSI as a response biomarker for immune checkpoint inhibitors requires further investigation.Supplementary InformationThe online version contains supplementary material available at 10.1038/s41598-025-27895-w. Introduction Mismatch repair deficiency (dMMR) appears to be a core feature in endometrial cancers (ECs), present in approximately 20%−30% of cases1,2. Its determination is now recommended in all EC cases for several reasons, including molecular classification, hereditary predisposition screening, risk stratification and treatment selection, especially because it represents a predictive biomarker for immunotherapy in advanced or recurrent tumours3–6. Immunohistochemistry (IHC) is the current recommended method for dMMR identification, based on MLH1, PMS2, MSH2 or MSH6 expression loss3. Microsatellite instability (MSI), first recognised in colorectal carcinoma (CRC), can be alternatively explored as a surrogate marker of dMMR7,8. Microsatellites are short tandem repeats distributed in the genome, prone to errors during the replication process, which are not repaired in dMMR cells. Theoretically, dMMR and MSI conditions point to the same molecular characteristics and should be highly concordant, although they do not explore the same issue: one is focused on the deficiency’s cause, and the other on its functional consequence. Their concordance has been previously explored in CRC and EC, and it has been suggested that variations could exist depending on the cancer type. Additionally, most diagnostic tests have been developed and optimized for CRC, and sensitivity could be affected in other tumor settings. In fact, some studies comparing different testing methods indicate discrepancies9. New dMMR/MSI testing methods have recently been reported, and it might be worthwhile to test them to validate their clinical utility. Polymerase chain reaction (PCR)-based assays detect changes in the length of a variety of microsatellite panels, and have established threshold to designate tumours with microsatellite stability (MSS) and instability (MSI). Alternatively, promoter methylation analysis or gene NGS sequencing potentially identify loss-of-function alterations related to dMMR or MSI phenotypes9.The aim of this study was to compare the assignments of microsatellite status categories (MSS and MSI) by the 2 PCR assays developed by Promega (OncoMate and LMR MSI) in a previously characterised EC series by IHC and mismatch repair (MMR) genomic alterations. Materials and methods Patients and samples This study was performed in an early-stage EC series from La Paz University Hospital previously explored. The experimental protocol was approved by La Paz Hospital Ethics Committee (HULP#PI3778)10. This research project was performed in accordance with relevant guidelines and regulations and the informed consent was obtained from all participants and/or their legal guardians. Immunohistochemistry MMR status was set in tissue microarrays by the expression of MLH1, PMS2, MSH2 and MSH6, as previously described11,12. MMR gene alterations DNA extracted from the selected material was employed for MMR gene status determination as previously described13. Briefly, a Qiamp DNA FFPE Tissue Kit (Qiagen GmbH, Hilden, Germany) was employed for extraction, an MS-MLPA ME011 kit (MRC Holland, Amsterdam, Netherlands) to identify MLH1 promoter methylation and a custom panel (Paragon Genomics, Hayward, CA, USA) for identifying genomic alterations in MLH1, PMS2, MSH2 and MSH6. When promoter methylation and/or coding sequence alterations were found, the case is considered “altered”, and “non-altered” otherwise. Microsatellite instability DNA was extracted from the corresponding tissue blocks by punches in areas previously selected for non-tumoral and tumoral content (higher than 40%), and was considered an optimal yield when more than 2 ng was obtained13. MSI was determined by two Promega commercial kits. OncoMate includes 5 short mono-nucleotide (nt) repeat (SMR) markers (BAT-25, BAT-26, NR-21, NR-24 and MONO-27) and 2 penta-nt markers (Penta C and D for sample recognition). OncoMate is the new CE-IVD-certified version of Promega v1.2, including the same SMR and all PCR consumables. The long mononucleotide repeat (LMR) MSI panel is composed of the 4 SMR markers included in v1.2 and OncoMate (BAT-25, BAT-26, NR-21 and MONO-27), and 4 additional LMR markers (BAT-52, BAT-56, BAT-59 and BAT-60)14. SMR typically comprises 21–27 repeats, whereas LMR presents 52–60 repeats, which is even more sensitive to defects in the MMR pathway15,16. Both kits were used according to the manufacturer’s recommendations, and amplification was performed in a Veriti thermal cycler (Applied Biosystems) with appropriate dilution recommendations. Denatured PCR products in deionised formamide were analysed on a 3130xl Genetic Analyzer using GeneMapper 4.0 software (Applied Biosystems). Allele sizing was matched between normal and tumour samples with the Internal Lane Standard 600 (Promega). Each biomarker was analysed individually and called unstable when shifts between the tallest peak in non-tumor and tumor samples are observed, or the shoulder pattern extends the range of the smallest stutter peak in the tumor sample by at least 2 bp. For both OncoMate and LMR panels, an MSI status was reported when at least 2 markers were reported as mutant, and MSS designated when only one marker was altered or no instability was observed; cases with non-tumoral material PCR to compare, or cases with one or more individual markers not informative, were set as “non-evaluable”17. Statistical analysis Associations between factors were examined for all pairs of qualitative variables of interest (i.e., the MSI and MMR by IHC and genomic status) via computation of the contingency tables and performing Fisher’s exact independence test. For all pairs of binary variables, correlation coefficients were computed as descriptors of pairwise dependence. Also, for each possible binary response (e.g., altered/non-altered MMR gene status) we evaluated the classification capability of the other reference variables (e.g., MSS/MSI Promega status) by fitting logistic regressions. The prediction performance of these generalised linear procedures was quantified by the sensitivity, the specificity and the total proportion of correct classifications from absolute frequencies, as described before18. Briefly, the sensitivity for detecting MMR status by an MSI test using the IHC dMMR status as predictor, designates the proportion of MSI patients that are detected as MSI using the IHC dMMR condition as information. Similarly, the specificity of a test is its ability to detect an individual as MSS, computed as pMMR by IHC. In the hypothesis tests, a p-value lower than 0.05 is interpreted as evidence against the null hypothesis (of independence in the case of Fisher’s test). If any of the variables involved in a statistical procedure were not available (missing), the corresponding patients were removed (only for that specific analysis). Results Descriptive results Of the 126 included cases previously classified by IHC, 67 (53.2%) were dMMR and 59 (46.8%) proficient mismatch repair (pMMR). The same cases were additionally explored for MMR genomic status, with 2 (1.6%) cases not evaluable; 55 (43.7%) presented an altered MMR genomic pattern, whereas 69 (54.8%) did not13. The same samples were screened for microsatellite status by OncoMate and LMR MSI PCR tests. The number of invalid cases for each PCR test was as follows: 18 (14.3%) for OncoMate and 13 (10.3%) for LMR MSI, which were not included for further comparisons. There were 71 (56.3%) MSS cases for OncoMate and 69 (54.8%) for LMR, and 37 (29.4%) MSI cases for OncoMate and 44 (34.9%) for LMR. Distribution of categories by technique, as well as the instability proportion of individual markers, are shown in Fig. 1a. Compared with IHC, a decrease of dMMR assigned cases was identified by the other approaches (genomic assessment and both PCR-based methods). The proportion of MSI detected by OncoMate was lower than that detected by LMR, with a significant difference (p-value 6.5e-04). Regarding the proportion of unstable markers between kits, BAT-26 and MONO-27 were the most frequently altered by OncoMate, whereas BAT-52 and BAT-59 were most frequently altered by LMR analysis (Fig. 1b). OncoMate and long mononucleotide repeat test scoring The number of unstable markers according to the sample split by the IHC and MMR gene status categories is shown in Fig. 2. The histograms show that 1 altered microsatellite is a rational cut-off to consider a sample MSS or MSI. Fig. 1Descriptive results. (a) Category distribution by immunohistochemistry (IHC), mismatch repair (MMR) gene status and the two MSI PCR tests (OncoMate and long mononucleotide repeat [LMR] MSI Analysis sytem); (b) Proportion of individual microsatellites altered by OncoMate and LMR tests. pMMR: proficient MMR; MSS: microsatellite stability; dMMR: deficient MMR, MSI: microsatellite instability, NE: non- evaluable. Descriptive results. (a) Category distribution by immunohistochemistry (IHC), mismatch repair (MMR) gene status and the two MSI PCR tests (OncoMate and long mononucleotide repeat [LMR] MSI Analysis sytem); (b) Proportion of individual microsatellites altered by OncoMate and LMR tests. pMMR: proficient MMR; MSS: microsatellite stability; dMMR: deficient MMR, MSI: microsatellite instability, NE: non- evaluable. Fig. 2Histograms representing the number of unstable markers per sample in mismatch repair (MMR) proficient (pMMR) and deficient (dMMR) populations established by immunohistochemistry (IHC, top) or MMR gene status (below) using OncoMate (left) and long mononucleotide repeat (LMR) MSI Analysis sytem (right). Colour bars indicate dMMR (red) and pMMR (blue). Y axis includes the total number of cases evaluated, consisting in the informative cases for each comparison, and x-axis the number of unstable microsatellites (MS). Histograms representing the number of unstable markers per sample in mismatch repair (MMR) proficient (pMMR) and deficient (dMMR) populations established by immunohistochemistry (IHC, top) or MMR gene status (below) using OncoMate (left) and long mononucleotide repeat (LMR) MSI Analysis sytem (right). Colour bars indicate dMMR (red) and pMMR (blue). Y axis includes the total number of cases evaluated, consisting in the informative cases for each comparison, and x-axis the number of unstable microsatellites (MS). Comparison between oncomate and long mononucleotide repeat tests MMR categories were assigned by IHC, and we explored the distribution of MSS/MSI groups by both PCR-based techniques. A detailed comparison is illustrated in Fig. 3. Of 59 pMMR cases, 58 were assigned as MSS by both PCR tests. Among the 43 dMMR, 34 were considered MSI by both PCR approaches; however, 9 cases were determined as MSS by OncoMate. The correlation between OncoMate and LMR was 82.3%, and overall coincidences were 91.2%. Fisher’s exact test gave a p-value of 2.2e-16, demonstrating dependency between the results of both techniques. Fig. 3Heat map results overview. Cases are organized by columns and techniques by lines. Top bar shows mismatch repair (MMR) status by immunohistochemistry (IHC) and bottom bar shows MMR gene sequence alterations. In the middle, results for OncoMate and Long mononucleotide repeat (LMR) MSI Analysis system, are represented by each analyzed locus and by the integrative overall result. pMMR: proficient mismatch repair; dMMR: deficient mismatch repair; MSS: microsatellite stability, MSI: microsatellite instability; NE: not evaluable. Heat map results overview. Cases are organized by columns and techniques by lines. Top bar shows mismatch repair (MMR) status by immunohistochemistry (IHC) and bottom bar shows MMR gene sequence alterations. In the middle, results for OncoMate and Long mononucleotide repeat (LMR) MSI Analysis system, are represented by each analyzed locus and by the integrative overall result. pMMR: proficient mismatch repair; dMMR: deficient mismatch repair; MSS: microsatellite stability, MSI: microsatellite instability; NE: not evaluable. Comparison between polymerase chain reaction-based tests and immunohistochemistry A crosstab showing case distribution is included in Table 1 and a results heat map is displayed in Fig. 3. Regarding categories for each technique, overall, the concordance was quite high for the pMMR (by IHC) group and those assigned as MSS for OncoMate (38 of 43) and LMR (47 of 49). Major discrepancies were observed in the dMMR population by IHC, considered as MSI by OncoMate (32 of 65) and LMR (42 of 64). Correlations between the IHC and Promega tests were 38.8% for OncoMate and 62.6% for LMR, and overall coincidences were 64.8% and 78.8%, respectively. Fisher’s exact test exploring the PCR categories versus IHC resulted in p-values of 2.96e-07 (OncoMate) and 4.89e-12 (LMR assay), demonstrating dependency between results of both techniques. Comparisons with the individual IHC markers were also performed for both PCR tests. A detailed description is included in Supplementary Tables 1 and 2. Higher correlations were observed for MLH1, PMS2 and the doublet, with statistically significant p-values for both the OncoMate and LMR independence tests (with slightly higher correlation values for the latter). MSH2, MSH6 and their doublet did not show significant correlations for either of the 2 tests. Table 1Crosstab between immunohistochemistry and promega molecular test results, with case numbers.IHCMMR genomic statusPCR testspMMRdMMRTotalalterednoTotalOncoMateMSS383371215071MSI5323732537Total43651085355108LMR MSIMSS472269155267MSI2424437744Total49641135259111MSS, Microsatellite stable MSI, microsatellite instability LMR long mononucleotide repeat, IHC immunohistochemistry, MMR mismatch repair, pMMR proficient mismatch repair, dMMR deficient mismatch repair Table 1Crosstab between immunohistochemistry and promega molecular test results, with case numbers.IHCMMR genomic statusPCR testspMMRdMMRTotalalterednoTotalOncoMateMSS383371215071MSI5323732537Total43651085355108LMR MSIMSS472269155267MSI2424437744Total49641135259111MSS, Microsatellite stable MSI, microsatellite instability LMR long mononucleotide repeat, IHC immunohistochemistry, MMR mismatch repair, pMMR proficient mismatch repair, dMMR deficient mismatch repair Crosstab between immunohistochemistry and promega molecular test results, with case numbers. MSS, Microsatellite stable MSI, microsatellite instability LMR long mononucleotide repeat, IHC immunohistochemistry, MMR mismatch repair, pMMR proficient mismatch repair, dMMR deficient mismatch repair Comparison between polymerase chain reaction tests and mismatch repair genomic status A crosstab containing distribution of cases by both categories is included in Table 1 and a results heat map is displayed in Fig. 3. As with IHC comparisons, overall, the concordance was quite high for the no genomic alterations group and MSS categories. Of 55 cases without MMR genomic alterations, 50 and 52 cases were assigned to the MSS category by OncoMate and LMR, respectively. As occurred with IHC, major discrepancies were observed in the altered MMR genomic population, with 32 of 53 and 37 of 52 assigned to the MSI category by OncoMate and LMR, respectively. Correlations between genomic status and Promega tests were 54.0% for OncoMate and 60.5% for LMR. Overall coincidences were 75.9% and 80.2%, respectively. Fisher’s exact test exploring MSI categories versus MMR genomic status resulted in p-values of 1.82e-09 (OncoMate) and 2.27e-12 (LMR assay), demonstrating dependency between the results of both techniques. Overall concordance comparisons A complete summary of sensitivity, specificity and proportion of correctly classified cases is reported in Table 2. The results indicate that, taking IHC as a reference, LMR showed higher values of these parameters than OncoMate. Moreover, LMR showed a higher correctly classified percentage than MMR genomic status. When MMR genomic status was taking as a reference, a similar values of sensitivity, but higher specificity and correctly classified cases were reached with LMR. In addition, LMR showed a higher percentage of correctly classified cases, over IHC and OncoMate. Table 2Global comparisons between tests, considering immunohistochemistry or mismatch repair genomic status as reference variables, reported as percentages.Reference(predicted) variableCompared(predictor)variableSensitivitySpecificityTotal proportion of correctly classified casesIHCOncoMate86.553.564.8LMR MSI95.568.178.8MMR Genomic status83.669.675.8MMR genomic statusOncoMate86.570.475.9LMR MSI84.177.680.2IHC68.784.275.8IHC Immunohistochemistry, LMR Long mononucleotide repeat, MMR Mismatch repair Table 2Global comparisons between tests, considering immunohistochemistry or mismatch repair genomic status as reference variables, reported as percentages.Reference(predicted) variableCompared(predictor)variableSensitivitySpecificityTotal proportion of correctly classified casesIHCOncoMate86.553.564.8LMR MSI95.568.178.8MMR Genomic status83.669.675.8MMR genomic statusOncoMate86.570.475.9LMR MSI84.177.680.2IHC68.784.275.8IHC Immunohistochemistry, LMR Long mononucleotide repeat, MMR Mismatch repair Global comparisons between tests, considering immunohistochemistry or mismatch repair genomic status as reference variables, reported as percentages. IHC Immunohistochemistry, LMR Long mononucleotide repeat, MMR Mismatch repair Summary of discrepant cases A summary of discordant case results can be observed in Table 3; Fig. 3. IHC was set as the reference technique to compare the results of the PCR approaches. Additionally, the MMR genomic status data were explored as an orthogonal technique to clarify discrepancies. Table 3 Detailed summary of discordant cases. Discordances are shown according to OncoMate and long mononucleotide repeat technique, followed by the results of immunohistochemistry (p/d MMR), OncoMate, and LMR (MSS/MSI), as well as Genomic status (altered or not). Number of cases per category are included in brackets. Discordances IHC OncoMate LMR MSI MMR genomic status IHC/OncoMate (34)pMMR (3)MSI (3)MSI (1)altered (2)MSS (2)unaltered (1)dMMR (31)MSS (31)MSI (9)Altered(16)MSS(22)unaltered(15)IHC/LMR MSI(24)pMMR (2)MSI (2)MSI (2)altered (1)unaltered (1)dMMR (22)MSS (22)MSS (22)altered (9)unaltered (13)IHC Immunohistochemistry, LMR Iong mononucleotide repeat, MSS Microsatellite stable, MSI Microsatellite instability, MMR Mismatch repair, dMMR deficient mismatch repair, pMMR Proficient mismatch repair Table 3 Detailed summary of discordant cases. Discordances are shown according to OncoMate and long mononucleotide repeat technique, followed by the results of immunohistochemistry (p/d MMR), OncoMate, and LMR (MSS/MSI), as well as Genomic status (altered or not). Number of cases per category are included in brackets. Discordances IHC OncoMate LMR MSI MMR genomic status IHC/OncoMate (34)pMMR (3)MSI (3)MSI (1)altered (2)MSS (2)unaltered (1)dMMR (31)MSS (31)MSI (9)Altered(16)MSS(22)unaltered(15)IHC/LMR MSI(24)pMMR (2)MSI (2)MSI (2)altered (1)unaltered (1)dMMR (22)MSS (22)MSS (22)altered (9)unaltered (13)IHC Immunohistochemistry, LMR Iong mononucleotide repeat, MSS Microsatellite stable, MSI Microsatellite instability, MMR Mismatch repair, dMMR deficient mismatch repair, pMMR Proficient mismatch repair Detailed summary of discordant cases. Discordances are shown according to OncoMate and long mononucleotide repeat technique, followed by the results of immunohistochemistry (p/d MMR), OncoMate, and LMR (MSS/MSI), as well as Genomic status (altered or not). Number of cases per category are included in brackets. IHC/LMR MSI (24) IHC Immunohistochemistry, LMR Iong mononucleotide repeat, MSS Microsatellite stable, MSI Microsatellite instability, MMR Mismatch repair, dMMR deficient mismatch repair, pMMR Proficient mismatch repair Regarding comparisons of each individual PCR approach and IHC, cases with genomic alterations and dMMR/MSS and without genomic alteration and pMMR/MSI could favour to think in a PCR method pitfall; meanwhile, cases with genomic alteration and pMMR/MSI and without genomic alteration and dMMR/MSS could be assumed to be a false negative IHC result. Data recorded in Table 3 show that there are no clear differences in favour of one technique, although in the case of LMR a major agreement with genomic status was observed in dMMR discrepant cases, that can be observed in Fig. 3. Discussion In addition to dMMR/MSI, tumour mutational burden and the expression of immune checkpoint programmed cell death protein 1 and its ligand (PD-1 and PD-L1) have been extensively studied as predictive biomarkers for immunotherapy in many tumour types. Regarding EC, dMMR/MSI is the predictive biomarker recommended in guidelines to select the best candidates for immunotherapy3,19. The main approaches for its determination are IHC and PCR-based assays. In this study, we focused on 2 molecular tests developed by Promega in an early-stage EC sample series selected by their MMR IHC profile, and additionally explored for alterations in MMR genes. Elevated concordance between IHC and MSI molecular assays has been reported to date in CRC series, but was lower in EC samples. The majority of comparison studies have been performed between IHC and tests based on fragment length by capillary electrophoresis (mainly Promega v1.2). Other tests based on microcapillary electrophoresis (Tape station 4200, Agilent) and high-resolution melting (such as Idylla, Biocartis or Easy PGX ready MSI, Diatech) have also been evaluated. The technical variability, as well as the heterogeneity in the sample series analysed, make the results difficult to compare. However, the same concordance tendency has been observed among studies, reaching almost 100% in CRC, but lower and more variable in EC, ranging from 67% to 95%20–24. Our study in EC is along these lines, showing a non-negligible number of discrepancies in EC samples, mainly in the dMMR group compared with the IHC and MMR genomic statuses assigned. Those are probably highlighted due to the balanced dMMR and pMMR populations of our study, compared with the 20%−30% of dMMR samples in other studies. In our series, we observed that dMMR assignment appears to be higher by IHC than by the alterations found in the MMR genes. This difference could be explained by the fact that the protein loss could be the consequence of pathway defects beyond the alterations in the corresponding 4 main genes. Another aspect to consider is that MSI is a consequence of the improper function of the MMR pathway, and it could be hypothesised that different defects can have different impacts in the microsatellite repair. Marker by marker, previous studies of MSH6-deficient cancers have observed a higher false negative rate by MSI assays25. Another study suggests that minor components of the MMR, such as PMS2 and MSH6, could manifest as an attenuated dMMR phenotype26. More recently, discrepancies have also been described in methylated MLH1 cases20,27–30. In our series, poorer correlations were observed for MSH2, MSH6 and their doublet losses. More detailed studies need to be performed to clarify this issue in EC. Additionally, MSI can vary depending on the tumour background and can be modified by the percentage of unstable markers and the number of repeated units by markers31. It has been observed that different cancer types show different MSI levels17. In this sense, the election of the panel is also important and might explain the discrepancies between CRC and non-CRC, including EC. Some authors have suggested that microsatellite PCR and electrophoresis size discrimination could be more reliable techniques in EC to assess MMR deficiency20. The OncoMate kit includes SMR, and it has been observed that instability increases with microsatellite length; thus, longer homopolymer regions are more susceptible to replication errors16,17,32. This is particularly interesting in the case of EC, in which a mild instability pattern has been described, with subtle microsatellite shifts not commonly observed in CRC and that are more difficult to identify with only SMR markers27–30,33. The LMR MSI Analysis system maintains 4 of the OncoMate markers and includes 4 additional LMRs. In a previous study, individual LMR markers showed higher sensitivity and larger shifts than the SMR from Promega v1.2 in CRC and EC samples designated by IHC. The same specificity as well as higher sensitivity were observed16. Our results are in line with this study, demonstrating a better proportion of correctly classified cases, a higher sensitivity, and also a higher specificity compared with OncoMate. This result suggests that LMR could be a feasible option to consider specifically in EC for the characterisation of MMR status by PCR. We also identified some technical issues regarding the PCR methods. The volume and concentration of tumour cells are essential for accurate MSI analysis. Previous studies have indicated the use of 5 ng of DNA with > 10% tumour cellularity for the LMR prototype kit16. We used DNA extracted from a punched tissue block with more than 40% tumoral cells. In order to maximise the dMMR/MSI EC detection rate, the re-testing or the combination of various approaches has been suggested19,28,34. Regarding our data, until more definite recommendations exist in EC, perhaps these suggestions need to be reconsidered, as discussed later in this section. We additionally studied whether the criteria for MSI status classification could impact our results. Standard criteria are based on the number of MSIs and have been established in previous studies, although some modifications have been suggested13,16,35. Regarding the distribution in our series, we support the classical MSS classification (1 MSI or none) and MSI (two or more MSIs), which are also suggested in the Promega technical sheet for both kits17. In our study, IHC was chosen as the reference technique, because it is recommended by clinical guidelines; however, we also used the MMR genomic status information to clarify these differences. Regarding comparisons of individual PCR tests and IHC results, most discrepancies were found in dMMR cases assigned by IHC, which were considered as MSS by both PCR tests. Including the MMR genomic status in the comparisons, there was no clear trend to suggest one technique in favour of the other, although major agreement was observed in the case of LMR. This variability could be due in part to intra-tumor heterogeneity, which has been described in EC36. Along these lines, although a mutations analysis in 21 hysterectomies showed differences in the profiles of various tumour regions in 71.4% of cases, this heterogeneity was captured in the profile of the corresponding aspiration biopsies37. This issue could have an important clinical consequence, given that we do not know the impact of heterogeneous tumour cell populations on diagnoses, prognoses, and therapeutic decisions. In addition, comparison between different dMMR/MSI status determination tests have shown discrepancies, pointing that dMMR phenotype and genetic MSI status could not be completely equal. Moreover, some differences in the immune checkpoint inhibitors efficacy have been reported depending on the tests. Regarding our data and that published by other authors, we consider that the election of the best molecular technique for MMR/MSI testing as biomarker to predict the response to immunotherapy needs further investigation9,38. Among the strengths of the study, it was based on the evaluation of MSI status in a large number of EC samples using and comparing 2 PCR tests: OncoMate and LMR MSI Analysis system. Although samples were selected by the current standard technique (IHC), data on MMR genomic status were also obtained, providing orthogonal information for a more accurate comparison. Some points could be considered as limitations, but also as strengths: first, this was a single-centre retrospective study; samples were obtained from the same centre with a homogeneous selection and marker interpretation. Second, the study was based on a selected population of balanced d- and pMMR groups, which although it is not representative of real-world distribution, the groups were of equal proportion, which highlights their differences. Lastly, the inclusion of only early stages could be considered in the same manner, providing homogeneous data about this population but lacking the results of advanced cases. Conclusions In our study, the LMR test achieved higher concordance with IHC and MMR genomic status results in a series of early-stage EC cases. Therefore, a more extensive PCR test, such as LMR MSI Analysis system, could be a more suitable technique for MMR testing in EC, in which shifts are relatively small. However, some discrepancies were observed, and further studies are needed to evaluate the optimal test for MMR status as a predictive biomarker. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1 Supplementary Material 1 Supplementary Material 2 Supplementary Material 2