BRAF V600E/RAS Mutations and Lynch Syndrome in Patients With MSI-H/dMMR Metastatic Colorectal Cancer Treated With Immune Checkpoint Inhibitors

Abstract Background We pooled data from 2 cohorts of immune checkpoint inhibitors-treated microsatellite instability-high/mismatch repair-deficient (MSI/dMMR) metastatic colorectal cancer patients to evaluate the prognostic value of RAS/BRAFV600E mutations and Lynch syndrome (LS). Patients and Methods Patients were defined as LS-linked if germline mutation was detected and as sporadic if loss of MLH1/PMS2 expression with BRAFV600E mutation and/or MLH1 promoter hypermethylation, or biallelic somatic MMR genes mutations were found. Progression-free survival (PFS) and overall survival (OS) were adjusted on prognostic modifiers selected on unadjusted analysis (P < .2) if limited number of events. Results Of 466 included patients, 305 (65.4%) and 161 (34.5%) received, respectively, anti-PD1 alone and anti-PD1+anti-CTLA4 in the total population, 111 (24.0%) were treated in first-line; 129 (28.8%) were BRAFV600E-mutated and 153 (32.8%) RAS-mutated. Median follow-up was 20.9 months. In adjusted analysis of the whole population (PFS/OS events = 186/133), no associations with PFS and OS were observed for BRAFV600E-mutated (PFS HR= 1.20, P = .372; OS HR = 1.06, P = .811) and RAS-mutated patients (PFS HR = 0.93, P = .712, OS HR = 0.75, P = .202). In adjusted analysis in the Lynch/sporadic status-assigned population (n = 242; PFS/OS events = 80/54), LS-liked patients had an improved PFS compared to sporadic cases (HR = 0.49, P = .036). The adjusted HR for OS was 0.56 with no significance (P = .143). No adjustment on BRAFV600E mutation was done due to collinearity. Conclusion In this cohort, RAS/BRAFV600E mutations were not associated with survival while LS conferred an improved PFS.


Implications for Practice
Patients with MSI-H/dMMR metastatic colorectal cancer (mCRC) treated with immune checkpoint inhibitors had impressive survival results.In this population, RAS and BRAF V600E mutations in tumor are not prognostic factors for progression-free survival (PFS) and overall survival.In the absence of a standardized definition, an algorithm presented in this study based on immunochemistry and molecular data, define patients with MSI-H/dMMR mCRC, Lynch syndrome, and sporadic cases.In this population, patients with Lynch syndrome had better PFS compared with those with sporadic cases.

Introduction
Immune checkpoint inhibitors (ICIs) have revolutionized the treatment and prognosis of microsatellite instability-high/mismatch repair-deficient (MSI-H/dMMR) metastatic colorectal cancer (mCRC).2][3] The phase III KEYNOTE-177 study demonstrated superiority for first-line pembrolizumab over chemotherapy in terms of median progression-free survival (PFS) of 16.5 vs. 8.2 months; hazard ratio (HR) = 0.60, 95% CI, 0.45-0.80,P = .0002). 4 The KEYNOTE-177 study did not report a statistically significant overall survival (OS) benefit with pembrolizumab vs. chemotherapy per the published statistical plan; this was likely due to high rate (60%) of crossover to ICIs in the chemotherapy arm after progression. 5Based on these data, pembrolizumab was approved by the Food and Drug Administration and European Medicines Agency.Despite high rates of response and a clinical benefit with ICIs, 20%-31% of patients with MSI-H/dMMR mCRC experience primary resistance, frequently resulting in delaying other effective therapies, autoimmune toxicities, and significant collective cost. 1,3,4Thus, it is crucial to identify a subpopulation of MSI-H/dMMR mCRC patients with primary resistance to ICIs.
Molecular heterogeneity of MSI-H/dMMR mCRC probably impairs prognosis and the efficacy of ICIs.3][4][5] For stage III colon patients with cancer receiving adjuvant FOLFOX, BRAF, or KRAS mutations are independently associated with shorter survival in those with microsatellite-stable colon cancer, but not MSI tumors. 6,7In mCRC, RAS/BRAF V600E mutations are well-known molecular modifiers of prognosis with an impact on anti-cancer therapies such as anti-EGFR targeted strategies.The results of an analysis of PFS in pre-specified subgroup of RAS mutated MSI-H/dMMR mCRC in the KEYNOTE 177 study call into question the superiority of pembrolizumab in this subpopulation (HR = 1.14, 95% CI, 0.68-2.07).Yet, RAS mutational status data were lacking in 30% of patients and this lack of effect was less apparent in the OS subgroup analysis (HR = 0.92, 95% CI, 0.48-1.75). 4,5[10] Several studies have shown that Lynch-associated CRC or endometrial cancer generally presents with more pronounced local T-cell infiltration and even a higher mutational burden compared with sporadic MSI-H CRC, which can support different responses to ICIs. 11,122][3] However, these trials lacked rigorous criteria for distinguishing Lynch-related tumors from sporadic.In fact, only proven germline MMR gene mutation should confirm Lynch-associated CRC.Whereas those with loss of MLH1/PMS2 expression associated with MLH1 promoter hypermethylation or BRAF V600E mutation and with biallelic somatic mutations of MMR genes should be classified as sporadic (Fig. 1).
Here we evaluate the impact of RAS/BRAF V600E mutational status and Lynch syndrome on prognosis of patients with MSI-H/dMMR mCRC treated with ICIs.

Patients
In this international multicenter study, we analyzed data from 2 pre-existing prospective cohorts of MSI-H/dMMR mCRC patients who received anti-PD1 monotherapy or the anti-PD1 plus anti-CTLA4 combination.The first immuno-MSI French cohort included all consecutive MSI-H/dMMR mCRC patients treated at Saint-Antoine Hospital (Paris, France) from February 2015 to December 2021.This cohort was approved by the ethics committee (N°2020-CER 2020-6).The second multicentric international cohort included MSI-H/dMMR patients with mCRC treated at centers in Italy, Spain, and the United States between November 2014 and November 2021.Ethical approval for the second cohort was provided by the Institutional Review Board of Fondazione IRCCS Instituto Nazionale dei Tumori of Milan (INT 117/15).

Molecular Data
MSI-H/dMMR status was determined by immunohistochemistry and/or multiplex polymerase chain reaction.RAS (KRAS and NRAS)/BRAF V600E mutational status, MLH1 promoter hypermethylation status, and MMR germline mutations testing were done using local practice according to international guidelines.
We developed Lynch/sporadic classification algorithm by interrogating available immunochemistry and molecular data.Patients were considered to have Lynch syndrome-associated CRC only in case of determined germline mutation and were deemed to have sporadic CRC only if loss of MLH1/PMS2 protein expression associated with BRAF V600E mutation and/ or hypermethylation of MLH1 promoter or biallelic somatic mutations of MMR genes (Fig. 1).

Radiological Analyses
Tumor radiological assessment was done ≤28 days before the first dose (baseline) of ICI and every 6-10 weeks, thereafter, according to different protocols.The radiological response was evaluated by the Response Evaluation Criteria in Solid Tumors (RECIST) 1.1 criteria with the possibility to pursue treatment beyond initial RECIST 1.1-defined progression at the treating physician's discretion in case of clinical benefit.In this multicenter cohort, scans were not reviewed centrally for the purposes of this study.When a pseudoprogression was suspected, treatment beyond RECIST 1.1 progressive disease was conditional to a locally confirmatory imaging done at 4-8 weeks after the first evidence of progression.In this case, confirmed primary progression was defined according to immune RECIST (iRECIST) criteria and imaging was retrospectively and locally reviewed by an experienced radiologist according to RECIST 1.1 and iRECIST (confirmed progressive disease). 13

Statistical Analysis
Categorical variables were described by numbers and percentages and continuous variables by means, SDs, and minimum and maximum values.
The primary endpoint was PFS defined as time from the first injection of ICIs to the first disease progression per iRE-CIST or death from any cause.Secondary endpoints were OS, defined as time from the first injection of ICIs to death, whatever the cause and overall response rate, defined as the proportion of patients achieving partial or complete response according to iRECIST criteria.Survival curves were generated using the Kaplan-Meier method.Cox regression models were used to compare OS and PFS between groups.The results were expressed as hazard ratios with 95% CIs.The following risk factors were studied: BRAF V600E , RAS mutational status Lynch syndrome, age at start of ICI therapy, sex, sidedness (left vs. right), treatment type (anti-PD1 vs. anti-PD1 plus anti-CTL4), Eastern Oncology Cooperative Group performance score (ECOG PS; 0 vs. 1 vs. 2), primary tumor surgery (yes vs. no), number of metastatic sites (≥ 2 vs. 1), ICIs used in first-line (yes vs. no).The number of variables selected for adjusted analysis was limited to 5 events per variable and selected with a P-value of < .20 in unadjusted analysis if necessary.Center was considered as stratification variable.Risk proportionality hypothesis was checked for all variables and for continuous variable, log linearity hypothesis was checked.BRAF V600E and Lynch syndrome were analyzed separately due to collinearity.Models were performed in patients with available data for all studied variables.
Given the molecular precision required for classification of patient according to our algorithm, a large number of patients with an indeterminate Lynch syndrome or sporadic was expected, therefore we planned to perform 2 analyses including (1) patients with known RAS/BRAF V600E status and (2) patients with known Lynch syndrome/sporadic status.A logrank test was used to compare PFS between groups according to germline mutation in Lynch patients.All superiority tests were 2-sided and P-values of <.05 were considered significant.Statistical analyses were performed using SAS software (version 9.4; SAS Institute Inc., Cary, NC).

Lynch vs. Sporadic
In the population with determined Lynch and sporadic status (n = 242), 84 PFS events were observed.In unadjusted and adjusted analysis of 231 patients with data available for all selected variables (80 events observed), Lynch syndrome was associated with fewer PFS events compared with sporadic type (HR = 0.40, 95% CI, 0.25-0.64,P < .001and HR = 0.49, 95% CI, 0.25-0.96,P = .036,respectively; Table 4 and Fig. 3).The type of germinal mutation in case of Lynch syndrome did not appear to impact the ICIs effect on PFS in the analysis with a limited number of patients (n = 104, Supplementary Fig. S1).There were 58 OS events were observed.Adjusted HR for analysis of OS in patients with known Lynch syndrome compared with those with sporadic CRC (54 events observed) was 0.56 (95% CI, HR = 0.25-1.22,P = .143;Supplementary Table S2).

Discussion
Our study was undertaken to answer the clinical questions in a cohort of MSI-H/dMMR mCRC patients about the impact of RAS/BRAF V600E mutational status and Lynch syndrome/ sporadic CRC on the efficacy of ICIs.We show that RAS and BRAF V600E mutations do not seem to be molecular modifiers of prognosis in patients who were treated with ICIs.This finding is not in line with PFS data on RAS mutation reported from the post-hoc subgroup analysis of the phase III Keynote 177 trial.However, the trial did not determine RAS status for all patients (30% of patients had no mutational status data) so this could have led to selection bias. 4][3][4] This work presented a large international cohort study of patients treated for mCRC by ICIs with a strict definition of Lynch syndrome.Our data suggest that Lynch syndrome is protective against PFS events.][10] Data from the subgroup analyses of previous prospective phase II trials did not support any differences in survival between these 2 groups of patients, but characterization of Lynch syndrome in these studies was done by investigators based on past medical history and other available factors collected from clinical records only without a defined algorithm, which could have led to misclassifications.2][3] The method used in this international study was based on rigorous classification with the concurrent use of immunochemistry and our designed molecular-based laboratory practice algorithm.Indeed, assigning correctly to Lynch or sporadic MSI-H/dMMR mCRC subgroups demands data, which are not always available or asked in routine practice such as MLH1 promoter methylation status testing or biallelic somatic mutational status analysis in absence of germline mutation (Fig 2).In our study, we did not find significant differences in OS between Lynch and sporadic groups.A possible reason for this lack of benefit in OS may be the low power to detect significance due to not sufficient valid sample size.However, it should be pointed out that our data are consistent with the literature since the NICHE-2 trial recently showed that preoperative 1-month therapy with ipilimumab and nivolumab achieved an increased pathological complete response rate in patients with Lynch syndrome-associated versus sporadic MSI-H primary colon cancer. 14Also, the promising results of organ preservation strategies with 100% clinical complete response in patients with MSI-H rectal cancer may be partially related to the over-representation of Lynch syndrome in patients developing MSI-H cancers in the rectum. 15Finally, our data are biologically sound since a previous study demonstrated significantly superior tumor mutational burden in patients with MSH2/MSH6 deficiency and this may lead to increase the immunogenicity of the tumor and, potentially, improved outcomes on immunotherapy, as we showed here. 16n spite of the strengths, this study has some limitations.Regarding the analysis of the role of Lynch vs. sporadic cases, a large number of cases were excluded (n = 224) due to the absence of molecular data and because they may have potentially biased the selection.The comparison of the population with determined Lynch syndrome/sporadic and the population with excluded cases indicated that there were more BRAF V600E -mutated mCRC in the analyzed population while RAS mutation was more frequent in the population with undetermined Lynch syndrome and sporadic CRC (Table 2).This observation is consistent with the fact The Oncologist, 2023, Vol. 28, No. 9 that BRAF V600E mutation is a major factor in our algorithm to clearly distinguish Lynch syndrome and sporadic CRC.
Although BRAF V600E mutation was not a prognostic factor in this ICIs-treated MSI-H/dMMR population, it was a selective marker for sporadic cases.We did find that the highest proportion of BRAF V600E mutation was seen in the sporadic group (75.8%).Moreover, on the molecular level, some misclassifications could persist in our study and in clinical practice due to the marginal phenomenon of germline MLH1 promoter hypermethylation.Still, the analysis of determined Lynch syndrome/sporadic and excluded cases populations also found that patients matched for clinical prognostic factors, namely age, performance status, number of previous lines for mCRC received, and number of metastatic sites.These results need to be prospectively validated in subgroup analysis with the addition of Lynch/sporadic cases strictly defined by the algorithm.If doing so, this classification could be used for stratification of MSI-H/dMMR mCRC patients in ICIs-based trials given the absence of a uniform standard.The continuous age variable did not meet the proportionality of risk assumption, so it was analyzed in classes defined from quartiles.Abbreviations: ECOG, Eastern Oncology Cooperative Group; HR, hazard ratio; ICI, immune checkpoint inhibitor.

Conclusion
This study demonstrated that RAS BRAF V600E mutations do not impact prognosis of MSI-H/dMMR patients treated with ICIs.Lynch syndrome-associated CRC might have a better survival compared with sporadic CRC, but this result will require further confirmation studies.

Figure 1 .
Figure 1.The algorithm for the Lynch syndrome classification according to the order of execution of MSI PCR test and MMR IHC test.IHC, immunohistochemistry; MMR, mismatch repair gene; MSI-H, microsatellite instability-high; PCR, polymerase chain reaction.

Table 1 .
Baseline characteristics of the combined cohort of 466 included patients with MSI-H/dMMR mCRC treated with ICIs.Flow chart of study reasons for Lynch and sporadic indeterminations in excluded patients.

Table 2 .
Baseline characteristics of patients with determined/undermined Lynch syndrome and sporadic CRC.

Table 2 .
Data not available for germinal mutation and MMR immuno-histochemistry (not possible to classify Lynch or sporadic).Abbreviations: ECOG, Eastern Oncology Cooperative Group; ICI, immune checkpoint inhibitor; MMR, mismatch repair.Continued *Germline mutation identified with missing data for the mutated gene.**

Table 3 .
Unadjusted and adjusted hazard ratios for progression-free survival (PFS) in 443 patients with known RAS and BRAF V600E mutational status.
* * Only patients with all the data variables available.**

Table 4 .
Unadjusted and adjusted hazard ratios for progression-free survival in 231 patients with determined Lynch syndrome and sporadic status.Only patients with all the data variables available.Abbreviations: ECOG, Eastern Oncology Cooperative Group; HR, hazard ratio; ICI, immune checkpoint inhibitor.