Sustained Disease Control in Immune Checkpoint Blockade Responders with Microsatellite Instability-high Colorectal Cancer after Treatment Termination

A mismatch repair deficiency (dMMR) or microsatellite instability-high (MSI-H) status occurs in fewer than 5% of all patients with metastatic colorectal cancer (1). MSI-H/dMMR tumor are attributed as hereditary—arising from germline mutations in one of the four mismatch repair (MMR) genes (MLH1, MSH2, MSH6, PMS2) as part of the hereditary non-polyposis colorectal cancer (HNPCC) syndrome (2, 3)—or as sporadic, arising from globally hypermethylated colorectal cancer tumors featuring loss of gene expression of MLH1 due to promoter hypermethylation (4). Via either mechanism, loss of expression and thereby homeostatic function of subsequent MMR proteins in dMMR/MSI-H colorectal cancer tumors impairs DNA repair and results in tumors characterized by high tumor mutation burden, increased tumor neoantigens, and greater cytotoxic immune cell infiltrates within the tumor microenvironment (5–9), relative to proficient mismatch repair/microsatellite stable (MSS) colorectal cancer.

Because of increased immune recognition, an MSI-H/dMMR status is a predictive biomarker for benefit with immune checkpoint blockade with anti-programmed death-(ligand) 1 [anti-PD-(L)1] treatments not only for colorectal cancer (10) but also for other solid tumors (11–13). For patients with dMMR/MSI-H advanced, unresectable colorectal cancer, the phase III KEYNOTE-177 trial (10) demonstrated improved survival with the anti-PD-1 antibody pembrolizumab relative to standard cytotoxic chemotherapy as frontline treatment and led to an FDA approval in this setting. Approximately 50% of study participants in this trial remained without disease recurrence after 2 years. Other single-arm trials evaluating anti-PD-(L)1 antibodies as monotherapy or in combination with anti-CTLA4 antibodies have reported similar long-term disease control for patients with advanced dMMR/MSI-H colorectal cancer (14–18).

Most patients with durable benefit stop therapy after 2 years of treatment for dMMR/MSI-H colorectal cancer, although the optimal length of treatment necessary for a potentially curative outcome remains undefined. With an otherwise very favorable safety profile for immunotherapy as an oncology treatment, patients in our experience often express concern about the possibility of recurrence following cessation of these agents. To provide quantifiable data useful for clinicians for guiding such conversations, we retrospectively evaluated the University of Texas – MD Anderson Cancer Center databases to describe disease recurrence and evaluated clinical and pathologic factors associated with recurrence for patients with dMMR/MSI-H advanced colorectal cancer who had stopped immunotherapy for reasons other than disease progression.

Patients assessed were included in this retrospective review approved by the Institutional Review Board at our institution if they had received therapy between 2014 and 2022 containing at least an anti-PD-(L)1 antibody as an immunotherapy treatment. All patients had a diagnosis of advanced or metastatic adenocarcinoma of the colon or rectum that was considered to be unresectable at the time of treatment start. Documentation of a dMMR or MSI-H status was confirmed in a Clinical Laboratory Improvement Amendments (CLIA)-certified laboratory by either loss of expression of at least one MMR protein on IHC, identification of microsatellite repeats using a DNA ISH assay, or commercially available next-generation sequencing testing. Sequencing for mutations in KRAS, NRAS, and BRAF^V600E were performed with a next-generation sequencing assay in a CLIA-certified laboratory. Included patients stopped immunotherapy for reasons other than disease progression (i.e., treatment benefit or unacceptable treatment-related toxicity), based upon the clinical judgment of the evaluating oncologist. Pattern of recurrence was defined as either oligometastatic or disseminated metastatic, as defined previously (19), with oligometastatic recurrence consisting of up to five metastases, three or fewer involved organs, and lack of central nervous system, peritoneal, and bone metastases. Medical records were reviewed for collection of patients’ demographic, clinical, survival, pathology, and treatment information. Written informed consent was collected from each study participant for research conduct, and all research was performed in accordance with the Declaration of Helsinki.

Characterization of patient and tumor information was reported using descriptive statistics. Progression-free survival (PFS) was defined as the time between date of final immunotherapy (i.e., start of surveillance) and either date of disease recurrence/progression or the date of death, whichever occurred first. Overall survival (OS) was calculated as the time between stop of immunotherapy and the date of last-follow-up or date of death, whichever occurred last. Median PFS and OS were estimated according to the Kaplan–Meier method using SPSS. A log-rank test was utilized to compare median survival between groups of interest. Associations between the event of recurrence and coexisting mutations (KRAS/NRAS, BRAF^V600E), metastatic organ involvement (lung, liver, lymph nodes, or peritoneum), metastatic timing (synchronous vs. metachronous), prior immunotherapy [anti-PD-(L)1 alone or in combination with anti-CTLA-4 antibodies], etiology of MSI-H status (sporadic vs. HNPCC), and duration of immunotherapy (above vs. below the median IO duration) were assessed by Fisher exact tests, with a two-sided P value less than 0.05 considered significant.

Data were generated by the authors but are not publicly available due to patient confidentiality and protection of private health information. Deidentified data may be provided upon reasonable request from the corresponding author.

Among 121 patients with MSI-H/dMMR unresectable, metastatic colorectal cancer who were identified as having been treated with immunotherapy, 64 (53%) did not experience progression on immunotherapy and were included in this retrospective study (Table 1). The median age at the start of immunotherapy was 64 years (range, 27–87). Of these 64 patients, 48 (75%) received anti-PD-(L)1 therapy alone and 16 received anti-PD-(L)1 therapy in combination with anti-CTLA-4 therapy. There were 48 patients (75%) who stopped because of prolonged treatment benefit, and the remaining 16 patients (25%) had immunotherapy stopped because of toxicity. Median exposure to immunotherapy was 17.6 months (range, 1.3–51.9). Median exposure to immunotherapy for those patients who stopped because of toxicity was 7.2 months (range, 1.3–26.0).

After a median follow-up of 22.6 months (range, 0.3–71.7) after stopping immunotherapy, 56/64 patients (88%) remained without evidence of disease recurrence (Fig. 1A). For the 8 patients with disease recurrence, 1 had immunotherapy stopped because of prior toxicity, while 7 had no prior immunotherapy toxicity but rather stopped immunotherapy due to prolonged treatment benefit according to the discretion of the evaluating clinician. There was no statistically significant association between reason for immunotherapy cessation (disease control vs. toxicity) and subsequent disease recurrence (16% vs. 6%, respectively; P = 0.40; Fig. 1B). The patient who discontinued immunotherapy after only 1.3 months due to toxicity demonstrated a reduction (though not disappearance) of disease and has remained without recurrence or progression off treatment 14 months later.

There were 8 patients who did not experience sustained disease control—5 with recurrence at the same sites as previous disease and 3 with “progression” (appearance of new lesions not previously present; lung, distant lymph nodes, peritoneum; N = 1 each). The median time to recurrence/progression was 10.4 months (range, 0.7–19.1).

At the time of immunotherapy discontinuation, there were 28 patients (44%) who had experienced a complete response (CR) radiographically, and 36 (56%) who had stable disease or some shrinkage relative to baseline. Only 1 patient in this CR subgroup experienced recurrence, whereas 7 in the stable disease/partially responding subgroup recurred or progressed. A trend toward disease recurrence or progressing was observed for those without a CR when stopping immunotherapy [OR, 6.5; 95% confidence interval (CI): 0.75–57; P = 0.09].

Seven of these 8 patients were retreated with immunotherapy after recurrence or progression—4 with an anti-PD-1 antibody alone and 3 in combination with an anti-CTLA-4 antibody (Table 2). The remaining patient was not offered anti-PD-1 therapy again due to safety concerns per the discretion of the evaluating physician. None of the patients experienced worsening of disease upon restarting their immunotherapy, and have continued to tolerate treatment well without clinically significant toxicity that would warrant drug cessation.

The median PFS for the entire cohort was 53.9 months (95% CI: 46.3–64.7), with estimated 1- year, 2-year, and 3-year PFS probabilities of 98%, 91%, and 84%, respectively (Fig. 2A). Median OS from the time that immunotherapy was stopped was not reached (95% CI: 63.4–NR; Fig. 2B). Only 2 patients died, of which one death was due to non–cancer-related medical condition.

There was no observed association between recurrence after immunotherapy cessation and metastatic organ involvement (Fig. 3A) for the liver [OR, 1.5; 95% CI: 0.3–8.7; P = 0.63], peritoneum (OR, 1.9; 95% CI: 0.4–8.7; P = 0.38), and distant lymph nodes (OR, 5.3; 95% CI: 0.6–45.6; P = 0.13). Increased recurrence following discontinuation of immunotherapy was observed for patients with lung metastases relative to those without lung metastases (38% vs. 9%, respectively; OR, 6.1; 95% CI: 1.1–33.5; P = 0.04). As seen in Fig. 3B, recurrence after immunotherapy was not associated with a KRAS/NRAS mutation (OR, 0.6; 95% CI: 0.1–3.8; P = 0.61) or BRAF^V600E mutation (OR, 1.9; 95% CI: 0.4–9.5; P = 0.45). After recurrence/progression, molecular profiling was not performed in the majority (6/8) of cases because they were predominantly started back on immunotherapy according to the MSI-H/dMMR biomarker status. For the 2 patients who did undergo repeat testing of the circulating tumor DNA, no new acquired mutations were demonstrated at the time of recurrence compared with their baseline mutation profiling results.

Immunotherapy as a single agent or as a combination treatment did not affect the likelihood for recurrence after treatment discontinuation in our study. For example, 15% of patients treated with anti-PD(L)-1 therapy alone had recurrence, compared with 0% of patients treated with combination anti-PD(L)-1/anti-CTLA4 therapy (OR, 5.2; 95% CI: 0.3–97.6; P = 0.27; Fig. 4A). Six percent of patients with HNPCC had recurrence, compared with 13% of patients with sporadic etiology of MSI-H status (P = 0. 35; Fig. 4B). Timing of metastatic onset relative to the initial presentation for colorectal cancer (Fig. 4C) did not appear relevant in forecasting recurrence (8% for synchronous vs. 20% for metachronous, OR, 0.4; 95% CI: 0.1–1.7; P = 0.20). Sixteen percent of male patients had recurrence, compared with 9% of female patients (OR 1.9; 95% CI: 0.4–8.8; P = 0.40; Fig. 4D).

We compared outcomes among the patients who stopped immunotherapy due to disease control (N = 48) and toxicity (N = 16). No difference in median PFS was observed for the group who stopped because of disease control relative to those who stopped because of immunotherapy-related toxicity [53.9 months vs. NR; HR 3.8 (95% CI: 0.88–16), Supplementary Figure].

Our data provide a retrospective, single-institution analysis that showed that nearly 90% patients with metastatic or unresectable MSI-H/dMMR colorectal cancer who did not experience initial progression on immunotherapy do not recur after treatment cessation. Reassuringly, we found that most patients with advanced MSI-H colorectal cancer who experienced initial clinical benefit on immunotherapy overall did well after cessation regardless of a variety of disease and patient factors, including clinical features that are generally considered prognostically unfavorable, such as mutations in KRAS/NRAS (20) and BRAF^V600E (21–23) and metastases to the peritoneum (24, 25). This favorable effect was likewise observed regardless of the reason for stopping treatment (i.e., disease control or unacceptable toxicity from immunotherapy). Notably, for the 8 patients who did recur or progress after initially discontinuing immunotherapy, all 7 patients who were retreated with immune checkpoint blockade have continued to experience sustained disease control. While clinical trials have described the durable benefit of immunotherapy for MSI-H colorectal cancer from the start of treatment and who continue to remain on treatment, we provide here needed information regarding patient outcomes after stopping treatment.

Prior prospective clinical trials for treatment of advanced or unresectable dMMR/MSI-H colorectal cancer, such as KEYNOTE-177, have investigated outcomes in MSI-H colorectal cancer regardless of demonstrated response to immunotherapy. In such studies, those who ultimately did not clinically benefit were also included in their analysis. We sought to focus exclusively on patients with metastatic and/or unresectable MSI-H colorectal cancer that had immunotherapy stopped for reasons other than progression of disease, such as toxicity and prolonged disease control. By including only those that satisfied these criteria yet later experienced disease recurrence, we sought to identify features that correlated with disease recurrence (or alternatively, with durable/sustained benefit while off therapy). Our goal was to provide information useful to clinicians and patients with MSI-H advanced colorectal cancer for stopping immunotherapy electively.

A post hoc analysis of the KEYNOTE-177 trial showed that a KRAS or NRAS mutation was not predictive to benefit with pembrolizumab (10). By extension, it is reasonable to surmise that patients with dMMR/MSI-H metastatic colorectal cancer harboring a KRAS or NRAS mutation may be more likely to recur if immunotherapy is stopped. In addition, a BRAF^V600E mutation is a poor prognostic biomarker for patients with colorectal cancer, regardless of the MSS status. In our series, we found that mutations in either KRAS/NRAS or BRAF^V600E were not associated with disease recurrence in patients stopping immunotherapy for reasons other than disease progression. On the basis of these findings, it does not appear that these mutations may impact favorable outcomes in patients for whom immunotherapy is stopped electively, and these findings warrant confirmation in larger studies.

Of note, the majority of patients did not achieve a complete radiographic response at the time of their initial immunotherapy discontinuation, yet most patients have continued to remain without evidence of disease recurrence or progression. There is precedent that immune checkpoint blockade may sterilize MSI-H/dMMR tumors and eradicate all tumor despite the presence of persistent lesions noted radiographically. In support of this, a recent study showed that 12 out of 13 patients with dMMR colorectal cancer who had stable radiographic disease after immunotherapy prior to surgical resection had pathologic CR (26). Therefore, we believe that patients may stop immunotherapy after an upfront period of prolonged disease control, even if a complete radiographic response has not been achieved, as the majority of these patients will continue to experience clinically favorable outcomes without evidence of disease recurrence or progression.

Recent data suggest that site of distant metastasis may affect the response to immunotherapy for patients with colorectal cancer, likely due to differences in the tumor microenvironment. For example, a trial of anti-PD-1 and anti-CTLA4 therapies for patients with MSS, metastatic colorectal cancer reported improved response rates for patients without liver metastases than for patients with liver metastases (57% vs. 0%, respectively; ref. 27). In a single-institution retrospective analysis of 41 patients with MSI-H/dMMR metastatic colorectal cancer, treatment with pembrolizumab fared worse if patients had liver metastases than if they had non-liver metastatic colorectal cancer (28). Our retrospective cohort here was different than these analyses in excluding patients who had progressed and therefore did not benefit from immunotherapy. Indeed, only 19% of patients analyzed had liver metastases, far lower than the general population of patients with metastatic colorectal cancer, for which the prevalence exceeds 50% (29, 30). In our analysis, when we stratified by site of disease metastasis, we found no association of disease recurrence with metastases to the liver. Interestingly, a possible association between lung metastases and disease recurrence/progression was observed and merits further investigation in future studies.

We acknowledge the limitation that our data arise from retrospective study performed at only a single, large-volume, academic referral center. Validation of these findings across a population-based dataset could further generalize our findings. While a dMMR/MSI-H status is a favorable biomarker that predicts benefit to immunotherapy for patients with advanced cancers, as many as 40% of patients with colorectal cancer in this setting do not experience disease control and develop disease progression through such treatments. Recurrence is always a concern for patients and oncologists following treatment cessation.

Our data reinforce the favorable outcomes starting at the time of immunotherapy discontinuation, with nearly 90% of patients achieving sustained disease control off treatment. Favorable outcomes were observed regardless for the clinical factor (durable benefit or immunotherapy-related toxicity). The optimistic prognosis for these majority of patients should reassure both patients and providers in making the decision to stop immunotherapy.

J.A. Willis reports personal fees from Bayer HealthCare Pharmaceuticals outside the submitted work. K.P.S. Raghav reports other from Merck outside the submitted work. B. Johnson reports personal fees from Iota Bioscience; grants from Gateway for Cancer Research; other from BMS, Syntrix, and Gritstone bio outside the submitted work. A. Dasari reports grants from Xencor, Eisai, Guardant Health, Enterome; grants and other from HutchMed; other from Personalis and Illumina during the conduct of the study. B.K. Kee reports other from Medtronic outside the submitted work. M.S. Lee reports personal fees from Pfizer, Delcath, Janssen, BioNTech, G1 Therapeutics, Imvax, Bayer; grants and non-financial support from EpimAb BioTherapeutics, Merck, Erasca, Arcus Biosciences, Repare Therapeutics, Trisalus Life Sciences; grants from Boehringer Ingelheim and Xilis outside the submitted work. P.H. Le reports personal fees from Bayer and Curio Scitence outside the submitted work. J.P. Shen reports grants from Celsius Therapeutics, personal fees from Nadeno Nanoscience and Engine Biosciences outside the submitted work. E. Vilar reports grants and personal fees from Janssen Research and Development; personal fees from Recursion Pharma, Guardant Health, and Rising Tide Foundation outside the submitted work. S. Kopetz reports other from Genentech, EMD Serono, Merck, Holy Stone Healthcare, Novartis, Lilly, Boehringer Ingelheim, AstraZeneca/MedImmune, Bayer Health, Redx Pharma, Ipsen, HalioDx, Lutris, Jacobio, Pfizer, Repare Therapeutics, Inivata, GlaxoSmithKline, Jazz Pharmaceuticals, Iylon, Xilis, Abbvie, Amal Therapeutics, Gilead Sciences, Mirati Therapeutics, Flame Biosciences, Servier, Carina Biotech, Bicara Therapeutics, Endeavor BioMedicines, Numab, Johnson & Johnson/Janssen, Genomic Health, Frontier Medicines, Replimune, Taiho Pharmaceutical, Cardiff Oncology, Ono Pharmaceutical, Bristol Myers Squibb-Medarex, Amgen, Tempus, Foundation Medicine, Harbinger Oncology, Takeda, CureTeq, Zentalis, Black Stone Therapeutics, NeoGenomics Laboratories, Accademia Nazionale Di Medicina, Tachyon Therapeutics, Sanofi, Biocartis, Guardant Health, Array BioPharma, Genentech/Roche, EMD Serono, MedImmune, Novartis, Amgen, Lilly, Daiichi Sankyo, Lutris, Iylon, Frontier Medicines, Xilis, and Navire outside the submitted work. M.J. Overman reports personal fees from Nouscom, Bayer, Merck, Gritstone, Pfizer, Atreca, and Janssen outside the submitted work. V.K. Morris reports other from Bristol Myers Squibb, Pfizer, BioNTech, EMD Serono, RedX Pharma; personal fees from Regeneron and Novartis outside the submitted work. No disclosures were reported by the other authors.

K. Simmons: Conceptualization, data curation, formal analysis, validation, investigation, visualization, methodology, writing-original draft, writing-review and editing. J.V. Thomas: Resources, data curation, investigation, methodology, project administration, writing-review and editing. K. Ludford: Resources, data curation, writing-review and editing. J.A. Willis: Resources, data curation, writing-review and editing. V.S. Higbie: Resources, data curation, methodology, writing-review and editing. K.P.S. Raghav: Resources, data curation, writing-review and editing. B. Johnson: Resources, data curation, writing-review and editing. A. Dasari: Resources, data curation, writing-review and editing. B.K. Kee: Resources, data curation, writing-review and editing. C.M. Parseghian: Resources, data curation, writing-review and editing. M.S. Lee: Resources, data curation, writing-review and editing. P.H. Le: Resources, data curation, writing-review and editing. M.P. Morelli: Resources, data curation, writing-review and editing. J.P. Shen: Resources, data curation, writing-review and editing. A. Bent: Resources, data curation, writing-review and editing. E. Vilar: Resources, writing-original draft, writing-review and editing. R.A. Wolff: Resources, data curation, writing-review and editing. S. Kopetz: Resources, data curation, formal analysis, methodology, writing-original draft, project administration, writing-review and editing. M.J. Overman: Conceptualization, resources, data curation, formal analysis, investigation, writing-original draft, project administration, writing-review and editing. V.K. Morris: Conceptualization, resources, data curation, software, formal analysis, supervision, funding acquisition, investigation, visualization, methodology, writing-original draft, project administration, writing-review and editing.

V.K. Morris was supported by the NIH/NCI under award number K12 CA088084, the Cancer Prevention & Research Institute of Texas (CPRIT) under award number RP220416, and the Andrew Sabin Fellow Family Foundation. V.K. Morris, S. Kopetz, and A. Dasari were supported by CPRIT under award number RP200356. J.P. Shen is a CPIRIT Scholar in Cancer Research and was supported by the NCI (L30 CA171000 and K22 CA234406), CPRIT (RR180035), the Col. Daniel Connelly Memorial Fund, and Cancer Center Support Grant (P30 CA016672).