What Does MSI-H Mean on a Report?

Scattered through human DNA are short stretches where the same one or two letters repeat over and over. These are microsatellites. They are not genes. They are more like stutters in the text.

Every time a cell divides, it copies all of that. Repeated stretches are the easiest place to slip. A copying enzyme can add a repeat or drop one. So the cell keeps a proofreading crew on hand. That crew is the mismatch repair system, and it fixes the slip before it becomes permanent.

MSI-H means the proofreading crew has stopped working. The repeat lengths have drifted. A lab can measure that drift, and when it is large the report reads MSI-high.

A report may say MSI-H, or it may say dMMR. They are two ways of looking at the same underlying failure, from different angles: one measures the instability the failure leaves behind, the other looks for the missing repair protein. Most of the time the two agree. Not always. A tumour can be dMMR on staining and come back stable on PCR, or the reverse, and how well each test performs varies by cancer type. Where the result does not fit the clinical picture, labs may repeat it or run the other method.

NCI describes both routes. Molecular genetic tests look for the instability itself in tumor tissue, usually by PCR. Immunohistochemistry takes the other approach and stains the tissue for the repair proteins. Four proteins are checked: MLH1, MSH2, MSH6 and PMS2. If one of them is missing from the tumor cells, the tumor is called mismatch repair deficient.

Which protein is missing sends the workup in different directions. Loss of MLH1 in particular is usually followed by a check for MLH1 promoter methylation and, in bowel cancer, for a BRAF V600E change, because those point to a change that arose in the tumour rather than one carried in the family. Where those checks come back negative, germline testing and genetic counselling follow. Ask which protein was lost and what the follow-on tests showed.

Both answers point at the same problem. It is worth knowing which test produced the result on a given report, because they can disagree in unusual cases.

A tumor that cannot proofread accumulates mutations at a high rate. Those mutations produce abnormal proteins. Fragments of those proteins appear on the surface of the cancer cell, and the immune system can read them as foreign.

That is the whole logic. The tumor has made itself visible. Drugs that release the brakes on T cells then have something to act on. This is why PD-1 blockade works in this setting when it fails in many others.

The clearest evidence comes from colorectal cancer. FDA's approval summary describes KEYNOTE-177, an open-label randomized trial of 307 patients with previously untreated, unresectable or metastatic MSI-H or dMMR colorectal cancer.

Patients were assigned to pembrolizumab intravenously every 3 weeks, or to the investigator's choice of mFOLFOX6 or FOLFIRI, with or without bevacizumab or cetuximab, every 2 weeks. Anyone assigned to chemotherapy was offered pembrolizumab when the disease progressed.

Median progression-free survival was 16.5 months with pembrolizumab and 8.2 months with chemotherapy, with a hazard ratio of 0.60. FDA notes that the overall survival data were not mature at the time of the progression-free survival analysis.

Tumor status in that trial was determined locally, by PCR for MSI or by immunohistochemistry for mismatch repair. That detail matters, because it is the same pair of tests likely to appear on a report.

Pembrolizumab also carries an approval that ignores where the cancer started. NCI states the indication: adult and pediatric patients with unresectable or metastatic MSI-H or dMMR solid tumors whose disease progressed after prior treatment, and who have no satisfactory alternative.

That approval rested on five uncontrolled single-arm trials covering 149 patients. The pooled objective response rate was 39.6%. Eleven patients (7.4%) had a complete response and 48 (32.2%) had a partial response. Response was 36% in colorectal cancer and 46% in the other cancers.

Dostarlimab holds a related indication for adults with dMMR recurrent or advanced solid tumors after prior treatment. In the GARNET trial, the efficacy population was 209 patients, and the objective response rate was 41.6%, with 9.1% complete responses.

Both approvals depend on an approved test rather than on a doctor's impression. NCI names the VENTANA MMR RxDx Panel as the companion diagnostic used to select patients with dMMR solid tumors for dostarlimab. In KEYNOTE-177 the testing was done in local labs instead.

Read those numbers honestly. A response rate near 40% means most people in these trials did not have a response. It is a real option, not a certainty. It is also worth asking whether the approval being discussed is the tissue-agnostic one, which requires prior treatment and no good alternative, or a cancer-specific one such as the colorectal first-line indication.

MSI-H is the main tumor pattern seen in Lynch syndrome, also called hereditary nonpolyposis colorectal cancer. The germline defects behind it sit in the same four genes: MLH1, MSH2, MSH6 and PMS2.

But NCI is careful here. A tumor can also become MSI-H because one of those genes was silenced by DNA methylation in the tumor alone. Nothing is inherited, and nothing passes to children. This is more common in older adults.

So the tumor result raises the question. It does not answer it. Germline testing, on blood or saliva rather than tumor, is what settles whether the family needs to know. Where that referral has not been offered, it is reasonable to ask about it directly.

MSI-H says nothing about stage. It does not describe how far the cancer has spread, and it is not a substitute for imaging or a staging report.

It is also not a guarantee. Some MSI-H tumors do not respond to checkpoint blockade at all. Others respond and later grow again.

And an MSI-H result is not an emergency. It is usually treated as a priority, because it can open a route that is not chemotherapy, but it is a planning finding rather than a crisis. Whether that route is right for you is a separate question: it turns on the cancer type, where you are in treatment, the exact current approval wording for the drug proposed, and your own health. Response rates describe groups, and some MSI-H tumours do not respond.

For how this fits with other molecular results, see biomarker testing and precision medicine. For how checkpoint drugs work more generally, see immunotherapy. For what a tumor test can and cannot say about inheritance, see tumor testing vs inherited genetic testing.

• FDA — Pembrolizumab for first-line treatment of MSI-H/dMMR colorectal cancer (KEYNOTE-177)
• NCI PDQ — Agnostic Cancer Therapies (Health Professional Version)