What Does dMMR Mean on a Cancer Report?

Every time a cell copies its DNA it makes small mistakes. A dedicated set of proteins finds those mistakes and fixes them. That system is called mismatch repair.

dMMR means deficient mismatch repair. One or more of those proteins is missing. So the errors stay. Over time they pile up. The tumor ends up with far more mutations than a typical cancer of the same type.

Two methods measure the same failure from different angles.

Immunohistochemistry stains the tumor tissue for the repair proteins. It shows which ones are absent. NCI describes it as testing for the loss of mismatch repair proteins.

Polymerase chain reaction takes the other route. It looks for microsatellite instability. Microsatellites are short repeated stretches of DNA, and they change when repair fails.

That is why reports carry both labels. FDA indications name MSI-H and dMMR together. The drugs were approved on evidence that used both tests. In the pembrolizumab data, 135 of the 149 patients were tested up front, by local PCR or by staining. The other 14 were identified later, from stored samples.

A tumor carrying an unusual number of mutations produces unusual proteins. Those give the immune system something to recognize.

Two checkpoint inhibitors are approved on this marker regardless of where the cancer started.

Pembrolizumab covers unresectable or metastatic MSI-H or dMMR solid tumors, in adults and children. The disease must have progressed after prior treatment. And there must be no satisfactory alternative option.

Dostarlimab covers dMMR recurrent or advanced solid tumors in adults. An FDA-approved test must establish the dMMR status. The disease must have progressed during or after prior treatment, with no satisfactory alternative. This indication is not approved in children.

Approvals like these come from single-arm trials, which means response rates rather than survival comparisons.

Pembrolizumab. Five uncontrolled single-arm trials contributed 149 patients. Ninety had colorectal cancer. Fifty-nine had one of 14 other cancer types. The drug went in by drip either every 3 weeks at a fixed amount or every 2 weeks at an amount scaled to body weight. The pooled objective response rate was 39.6%. Eleven patients, or 7.4%, had a complete response. Forty-eight, or 32.2%, had a partial one. Response was 36% in colorectal cancer and 46% in the other types. Among responders, 78% held that response for 6 months or longer.

Dostarlimab. The GARNET trial's efficacy group was 209 patients. All had dMMR recurrent or advanced solid tumors that had progressed after systemic therapy. The objective response rate was 41.6%. Complete responses made up 9.1%, partial responses 32.5%. Median follow-up was 27.7 months. The companion diagnostic is the VENTANA MMR RxDx Panel.

For metastatic colorectal cancer the evidence goes further, because there is a randomized comparison.

KEYNOTE-177 enrolled 307 people with untreated MSI-H or dMMR metastatic colorectal cancer. Half received pembrolizumab as a drip every 3 weeks. Half received chemotherapy. That meant FOLFIRI or modified FOLFOX-6, with or without bevacizumab or cetuximab.

Median progression-free survival was 16.5 months with pembrolizumab. It was 8.2 months with chemotherapy. Objective response was 43.8% against 33.3%. Grade 3 or higher adverse events hit 56% of the pembrolizumab group, and 78% of the chemotherapy group. In the final survival analysis, median survival was not reached with pembrolizumab. With chemotherapy it was 36.7 months.

The FDA approved pembrolizumab for untreated metastatic dMMR colorectal cancer in 2020. One caveat sits inside the trial. Progression-free survival favored pembrolizumab in every prespecified subgroup but one. The exception was patients with KRAS or NRAS variants.

The marker is not tied to one organ, which is the whole point of a tumor-agnostic approval. But it is not evenly spread either.

The clearest signal comes from how the trials were built. In the pooled pembrolizumab population, 90 of 149 patients had colorectal cancer, and the remaining 59 were spread across 14 different cancer types. GARNET, the dostarlimab trial, ran a dedicated expansion cohort for dMMR and MSI-H endometrial cancers, plus a cohort for non-endometrial dMMR and MSI-H tumors.

So colorectal and endometrial cancer dominate the evidence. A dMMR result elsewhere is real, and rarer, and the data behind it is thinner.

• Which test produced this result, staining or PCR?
• Which protein was lost, and does that point anywhere?
• Is this cancer type included in the tumor-agnostic indication?
• Has prior treatment already been given, since both approvals require it?
• Is there a randomized comparison for my cancer, or only single-arm data?
• Has germline testing been arranged, and who requests it?

A dMMR result does not guarantee that immunotherapy will work. The response figures above sit around 40%. Read plainly, that means a large minority of people see their tumor shrink. It also means most do not, at least by the measure those trials used. Some who do respond will progress later.

It also does not confirm Lynch syndrome. NCI links the MSI-H phenotype to germline defects in four genes. MLH1. MSH2. MSH6. PMS2. It is the main phenotype seen in Lynch syndrome tumors. But NCI also says a tumor can show it because one of those genes was switched off by DNA methylation. That is not inherited.

Only separate germline testing tells those apart. The answer matters to relatives as much as to the person tested, because an inherited cause changes what siblings, parents and children are offered. For how biomarker results feed into treatment generally, see biomarker testing and immunotherapy.

• NCI — Agnostic Cancer Therapies (PDQ®), Health Professional Version
• NCI — Colon Cancer Treatment (PDQ®), Health Professional Version