Complementary NGS Improves dMMR Detection With IHC in CRC and Endometrial Cancer

Amin Nassar, MD

Next-generation sequencing (NGS) is a highly sensitive assay that can be utilized alongside standard immunohistochemistry (IHC) to detect mismatch repair (MMR) deficiency (dMMR) that might otherwise be missed in patients with colorectal cancer (CRC) or endometrial cancer, potentially improving the identification of those who would benefit from immunotherapy, according to Amin Nassar, MD, and Elias Bou Farhat, MD.

A retrospective study was designed to evaluate whether tumor-only NGS for MMR mutation signature could more accurately assess and identify dMMR tumors vs MMR IHC testing. The study discovered an additional 1.0% and 5.9% of NGS dMMR cases in CRC and endometrial cancer, respectively, that had been deemed intact by IHC. These patients with dMMR tumors that were missed by IHC but detected by NGS-based mutation signature (NGS/IHC discordant) experienced similar clinical outcomes with immunotherapy compared with patients who were identified as dMMR by both IHC and mutation signature (NGS/IHC concordant), across both cancer types.

Additionally, patients with discordant results had improved outcomes vs those with concordant MMR proficiency (pMMR) results. The median overall survival (OS) was not reached (NR; 95% CI, 10-NR) in patients with dMMR on NGS but intact IHC status, and was 14.9 months (95% CI, 7.3-25; P = .003) in patients with pMMR on NGS and intact IHC. The median time to treatment failure (TTF) was 44.4 months (95% CI, 10-NR) in the discordant group and 5.0 months (95% CI, 3.2-7.3; P = .001) in the pMMR concordant group.

Moreover, patients with CRC or endometrial cancer who had discordant results and received immune checkpoint inhibitors had longer OS vs checkpoint inhibitor–naive patients treated with other systemic therapies.

“Our findings suggest that revisiting some of the MMR diagnostic guidelines and emphasizing the benefits of NGS MMR signature analysis is important,” Bou Farhat stated in an interview with OncLive.^® “These patients could be candidates for immunotherapy who otherwise would not really receive it.”

“NGS is [not likely] to take the place of IHC, at least in the near future,” Nassar added during the interview. “However, it’s an important complementary test that should be done whenever available, because we don’t want to miss those [thousands of] patients [who] could derive benefit [from immunotherapy],” he emphasized.

In the interview, Bou Farhat and Nassar explained how ​​using a tumor-only NGS approach could address limitations, such as ambiguous staining patterns and sensitivity issues associated with IHC in determining MMR status; highlighted the concordance rates observed in this patient population and their association with survival outcomes; and outlined the potential clinical impact of these findings on treatment decision-making in CRC and endometrial cancer, especially concerning immunotherapy use.

OncLive: What was the rationale for assessing the concordance in MMR determination by IHC and NGS? What gaps in knowledge or unmet needs are you seeking to address?

Bou Farhat: Mismatch repair proteins are present in the cells and usually repair DNA errors during replication. Sometimes there are some genetic conditions, especially in cancers, [that render] those proteins nonfunctional, so they won’t be able to correctly repair the DNA. This leads to a higher rate of DNA accumulation. This process usually happens in colorectal and endometrial cancers. Whenever patients [have] dMMR [tumors] they respond well to immunotherapy. The gold standard right now to detect dMMR [tumors] is performing IHC staining. However, in this project we are trying to show that there is another method using NGS to detect this deficiency with a higher sensitivity.

Nassar: The standard of care right now in the United States and in Europe is IHC, just because it has a quick turnaround time and it’s accessible. One issue with this methodology is that sometimes a mutation happens in the epitope, which does not affect the expression of the protein, but affects the function. Because IHC relies on antigen-antibody binding, a mutation in the epitope that preserves the antigen but disrupts the functionality [can cause] IHC to miss the dMMR and consider the tumor to be MMR proficient. [Since] there’s a subset of patients that are being missed, [we thought to] quantify the subset and see if [we could] detect them by a different methodology. [In doing so, we could see whether] there is benefit that they gained from immunotherapy even though they were missed by IHC.

At the end of the day, NGS involves the broad scale genotyping of multiple regions in the genome. Instead of looking only at one region, which is the antigen-epitope binding site, you’re looking at multiple features. [Accordingly], we hypothesized that NGS is going to be more sensitive and more accurate than IHC. This has been a long discussion by different [organizations, including] ASCO ESMO, and the College of American Pathologists. [Up to this point], there weren’t any data showing that NGS can show that those patients derive benefits from immunotherapy. For that reason, [this approach] hasn’t made it into the guidelines yet.

What are the consequences of misdiagnosis that were elucidated in a prior study on microsatellite instability and MMR IHC status with immune checkpoint inhibitors?

Nassar: Immune checkpoint therapy is very effective in patients with dMMR. However, we also know that there’s a subset of patients that do not derive benefit. The question has always been why. In that study, [investigators] looked at patients who did not derive benefit and noticed that [many of these patients] were misdiagnosed and were not truly dMMR. This begs the question: is IHC also [producing] some false positives at times when it shouldn’t be? For that reason, it was interesting for us to investigate NGS as a potential methodology that could limit false positives and improve the detection of false negatives.

Please provide an overview of the types of cancer where FGFR alterations are most prevalent, and how these alterations contribute to oncogenesis.

The more we test, the more [FGFR alterations] we find. FGFR gene rearrangements, fusions, amplifications, and mutations are seen in diverse cancers. Interestingly, an analysis of more than 350,000 patients with solid tumors using NGS found that FGFR1-4 variants were present in nearly 3% of patient samples and copy number aberrations were present in over 4% of tumor samples. Regarding the gene, FGFR1 was the most frequently mutated gene, followed by FGFR2, at 3.6% and 1.7%, respectively. [This was] followed by FGFR3 and FGFR4, respectively. Interestingly, the cancer with the highest frequency of FGFR alterations is cholangiocarcinoma. Beyond cholangiocarcinoma, we see FGFR alterations in bladder cancer, urinary tract cancers. We see FGFR gene rearrangements in salivary gland cancer, lung cancer, prostate cancer, and many other cancers, although they [are typically present in] less than 1% [of these patients]. One molecular profiling study found FGFR2 alterations in 20 of 158 intrahepatic cholangiocarcinoma samples. Depending on the study, we see that most of these alterations are enriched in bladder and colon carcinoma.

Please elaborate on the selection criteria for inclusion in the review. What should be known about the baseline criteria of this patient population?

Bou Farhat: [The study included] patients primarily with advanced colorectal cancer, or endometrial cancer who were diagnosed by BWH pathologists. Patients had to have [undergone] IHC and an Oncopanel/Profile NGS analysis at Dana-Farber, from that same tumor. After doing that, we checked a subset of patients that received at least 1 dose of checkpoint inhibitor therapy and checked how these patients were doing in terms of survival. IHC staining was done for 4 genes: MLH1, MSH2, PMS2, and MSH6. The result was categorized as either intact or absent. If one of those proteins did not positively stain, that was considered absent.

In terms of NGS, we defined an MMR signature as the presence of more than 1 single nucleotide insertion or deletion in a homopolymer region or 2 or more mononucleotide repeats. With IHC, sometimes you might also have subclonal results [in which] some cells are stained, and some cells are not stained. This could also lead to a [misdiagnosis].

Please detail the design and methodology of this study.

Nassar: We had 2 different methodologies. For IHC, [results were either] absent or intact. For NGS, [patients were] either deficient or proficient. Patients were then divided into 3 groups. Either they had concordant results between NGS and IHC [or discordant results]. Within the concordance [category], there were 2 different groups. They could be concordantly proficient, [which means] they would have an intact IHC protein expression of all 4 proteins and a pMMR signature by NGS, or they could be concordant deficient, whereby both NGS and IHC are saying the same thing. IHC is saying it’s absent, and NGS is saying it’s dMMR. The [final discordant] group [included] patients that had dMMR tumors by NGS but had an intact IHC protein stain for all 4 proteins. That was the cohort of interest in this study.

Bou Farhat: In summary, we have 2 tests, and 2 outcomes for each test. We compared the concordance and discordance between those tests. We looked at 3 [of the resulting 4] categories, because we did not identify any patients [proficient by NGS but absent by IHC.]

Nassar: The only 3 [patients with these results] in that study were not treated with immunotherapy. That’s why we couldn’t look at their outcomes.

What were the primary findings regarding the concordance between MMR determination using IHC and the NGS-based mutation signature assessment? What were the rates of concordance?

Nassar: The first question was: what is the rate of discordance that we’re seeing, and how many [patients] are being missed by IHC but detected by NGS as [having] dMMR [tumors]? We focused on the CRC cohort because it was our largest cohort, as expected. We had analyzed White, Asian, and Black [patients]. Across the board, we found that [approximately] 1% of the patients diagnosed within each category were missed by IHC [and] detected by NGS for dMMR. That is a big number, if you consider the [number of] patients that are diagnosed in the US. As for endometrial cancer, we didn’t divide [patients] by race because we didn’t have the numbers to do so. [However], we had almost 200 patients with an endometrial cancer diagnosis that had both NGS and IHC. Close to 6% were missed by IHC but detected by NGS. Here, because the prevalence of missing by IHC is almost six-fold higher, it is still [considered to be a high] number in the endometrial cancer cohort as well.

What was the observed association between accuracy and concordance of MMR status determination using NGS and IHC and patient outcomes?

Nassar: We have the 3 groups that we discussed, 2 of which were concordant and 1 that was discordant. Our hypothesis was that patients that were discordant, i.e. detected by NGS but missed by IHC, were going to do as well as those that were detected by both as having dMMR [tumors]. That is what we saw. The OS [rates] and time to next line of therapy were similar between both the concordant [dMMR group] and those that were discordant. That was the key point that we found. We also compared [these cohorts] to our controls, which are the patients that were found to be pMMR by both NGS and intact by IHC. Patients that were discordant did much better than those that were pMMR and intact by IHC.

Not only did [patients who have discordance in NGS and IHC results] perform as well as the group that was [concordantly] detected as [being] dMMR by both NGS and IHC; they did better than those considered [pMMR] by NGS and intact by IHC. [This] tells you that these patients are deriving benefits even though they were missed by IHC.

In the subset of checkpoint inhibitor–exposed patients, how did the accuracy of MMR status determination impact treatment responses?

We then zoomed in on [the cohort of patients that were discordant]. We looked at the patients [in this category] that were treated with immunotherapy or those that were never treated with immunotherapy but received something else like chemotherapy. Our hypothesis was that [patients with] discordant data who received immunotherapy were more likely to derive benefit compared with a patient that had discordant data but did not receive any immunotherapy. Immunotherapy is approved for tumors that are dMMR. We know that those patients do better on immunotherapy than other treatments that we typically give like chemotherapy. Patients that are truly dMMR should derive a lot of benefit from immunotherapy. This is exactly what we saw. If a physician had treated a patient with discordant data with immunotherapy, those patients did better than when the patient was treated with chemotherapy. This tells you that those patients are behaving as if they’re dMMR from the get-go, but they were just missed.

Are any limitations important to note regarding this research?

Nassar: We didn’t [perform] paired-normal testing; we instead had tumor-only NGS testing [for MMR signature]. Many of these patients could [therefore] have germline mutations that could lead to lynch syndrome. [However], I would still expect NGS to perform better in this [cohort]. We just didn’t have the sequencing data of the normal cohort. Another limitation is that [our study] did have a small dataset of immunotherapy-treated patients that had discordant data. A larger dataset is key.

Considering the projected number of additional dMMR cases identified annually through NGS, how might this impact population health? What recommendations are proposed for revisiting guideline recommendations on diagnostic testing?

Bou Farhat: Analysis of MMR signature using NGS is somewhat more sensitive than IHC analysis. If this method were applied across the whole United States, I think that [many] more [patients with] CRC and endometrial cancer who could benefit from immunotherapy would be identified. We can use it as a test to [identify] those patients who are missed by IHC.

Nassar: IHC is here to stay because it’s quick, it’s cheap, and it’s [widely] available. At the end of the day, a lot of our patients are treated at community hospitals, where they don’t always have the luxury of having sequencing done, and sometimes the turnaround time for sequencing takes a couple of weeks. [However], [we should] move [NGS] earlier in the diagnostic algorithm for patients whenever possible.

If a patient has IHC showing that their tumor is intact, I would go ahead and perform NGS if possible, because catching those patients is very important. There’s a high chance that we could cure some of these patients, even though they could be advanced in their disease. Our data provide the first evidence [for this method], and I think it is worth revisiting the MMR guidelines.

What future directions and prospective studies are recommended to confirm these findings, particularly in patients with metastatic CRC and endometrial cancer?

Nassar: Using these data, [we also want to] expand [this research] to further cohorts and see whether our methodology for identifying dMMR [tumors] can apply to different panels across the country. It is worth investigating whether this exact methodology is applicable across the board and not only in our center, because we need a generalizable signature.

Bou Farhat: We must keep in mind that this analysis [was performed in] patients with advanced metastasis. There are currently some trials in late stages showing that neoadjuvant immunotherapy is helpful for these patients who are dMMR, so the earlier we detect those patients, the more likely we are to help them and prolong their survival.

Reference

Bou Farhat E, Adib E, Daou M, et al. Benchmarking mismatch repair testing for patients with cancer receiving immunotherapy. Cancer Cell. 2024;42(1):6-7. doi:10.1016/j.ccell.2023.12.001