How MSI-High and Mismatch Repair Testing Guides Immunotherapy

Ask most people how cancer drugs are chosen and they will describe the organ: a breast drug, a lung drug, a colon drug. Mismatch repair deficiency broke that habit. It is a single molecular flaw that can appear in a tumor almost anywhere in the body, and where it appears, the tumor tends to respond to immunotherapy. That is why, in 2017, the U.S. Food and Drug Administration approved a cancer drug for the first time on the basis of a shared molecular feature rather than the organ where the disease started.

Key points#

• Mismatch repair (MMR) is the cell's system for fixing small DNA copying errors; when it fails, mutations accumulate rapidly.
• A tumor with many such errors is called MSI-High; the same defect seen from the protein side is called mismatch repair deficient (dMMR). They usually mean the same thing.
• A heavy mutation load makes the tumor look foreign, which is why these tumors often respond to drugs that release the brakes on the immune system.
• In 2017 this marker earned the first tissue- and site-agnostic FDA approval; the indication became a full approval in 2023.
• The marker improves the odds of response, it does not guarantee one, and it applies only to specific advanced-disease settings confirmed by an approved test.

A proofreader inside every dividing cell#

Each time a cell divides it copies roughly three billion base pairs of DNA, and copying that much text at speed produces typos. A dedicated proofreading system, DNA mismatch repair, scans the fresh strand and corrects the small insertions, deletions, and mismatched bases that the main copying enzymes miss. Four proteins carry most of the load, working as two pairs: MLH1 with PMS2, and MSH2 with MSH6. The ESMO consensus recommendations on MSI testing single out these four as the ones a laboratory examines first.

Take one of those proteins away and the proofreading fails. The typos survive, and they collect fastest in short repeated sequences called microsatellites, the stretches where the copying enzyme is most likely to stutter and slip. A tumor riddled with these uncorrected errors is labeled microsatellite instability-high, or MSI-High.

The defect can be inherited or acquired. In Lynch syndrome a person is born with one faulty copy of a repair gene, and a cancer can start once the remaining good copy is lost in some tissue. More commonly the deficiency is sporadic, often because the tumor switches off the MLH1 gene on its own. The route differs; the destination is the same, a genome that can no longer keep itself in order.

Why a broken proofreader helps the immune system#

Here is the link that turns a lab finding into a treatment strategy. A tumor that cannot fix its own errors churns out an unusually large number of mutations, and many of those mutations code for abnormal proteins the body has never encountered. Fragments of these proteins, called neoantigens, are displayed on the tumor surface like flags the immune system does not recognize as its own.

Left alone, many tumors escape by pulling a natural brake. The PD-1 pathway tells patrolling T cells to stand down, and tumors that trigger it can hide in plain sight. PD-1 inhibitors cut that brake line. In a tumor already covered in neoantigen flags, releasing the brake can set off a forceful and lasting immune attack. In a tumor with few mutations and almost nothing for T cells to notice, the same drug has little to act on. That gap in mutational load is the mechanistic reason MSI-High and dMMR predict who benefits, and it is why the marker is called predictive rather than merely prognostic.

MSI-High and dMMR: one flaw, two names#

The two labels describe the same broken pathway from opposite ends. Test the proteins and find one missing, and you have a mismatch repair deficient (dMMR) tumor. Measure the microsatellites and find them scrambled, and you have an MSI-High tumor. In the large majority of cases a tumor that is dMMR is also MSI-High, so oncologists treat the terms as near-synonyms and use whichever test the sample and laboratory support.

Two lab tests, two angles#

Immunohistochemistry (IHC) stains a tissue sample for the four repair proteins. A normal result shows all four present. Loss of MLH1 and PMS2 together, or of MSH2 and MSH6 together, points to deficiency, because these proteins prop each other up and tend to disappear in pairs. ESMO describes IHC for the four proteins as the usual first action, since it is widely available and inexpensive.

MSI testing by PCR or next-generation sequencing goes at the question the other way. Instead of asking whether the repair proteins are present, it compares the microsatellite regions in the tumor against the patient's normal tissue and measures how far they have drifted. It looks for the damage rather than the missing repairman.

The two methods agree in most cases, which is why either can support the same conclusion; when they disagree or a result is unclear, they are run together. An abnormal result also carries a second message. Certain patterns of protein loss can flag Lynch syndrome, prompting germline testing and genetic counseling that matters not only for the patient but for their relatives.

The approval that ignored the organ#

The regulatory landmark rests on this chain of biology. In May 2017 the FDA granted accelerated approval to the PD-1 inhibitor pembrolizumab for unresectable or metastatic MSI-High or dMMR solid tumors that had progressed after prior treatment and had no satisfactory alternative. As reported in The ASCO Post, the FDA called it the first tissue- and site-agnostic approval. The supporting patients had MSI-High or dMMR cancers of many different organs, united by the marker rather than by anatomy.

Accelerated approval is a conditional bargain, granted on strong early data and owed a confirmation later. The confirmation arrived. On March 29, 2023, the FDA converted the indication to full approval. By Merck's account, the decision rested on a pooled analysis of 504 patients across the KEYNOTE-158, KEYNOTE-164, and KEYNOTE-051 trials, spanning more than 30 cancer types. The overall response rate was 33.3 percent, and among those who did respond the effect held: 77 percent of responses lasted at least a year and 39 percent lasted three years or longer. The FDA approval summary in the peer-reviewed literature set out the same evidentiary logic behind the earlier accelerated approval.

Two limits keep the picture accurate. The indication is narrow: advanced disease that has progressed after prior treatment with no satisfactory alternative, confirmed by an FDA-approved test. And a one-third response rate, historic as it was, means most patients carrying the marker did not have a measured response. The biomarker shifts the odds toward benefit; it does not promise you an outcome.

What this teaches about reading evidence#

The durable takeaway is about appraisal, not about any single drug. A marker earns a tissue-agnostic indication only when three things line up across many tumor types: a plausible mechanism, a test that reproduces, and clinical benefit that lasts. MSI-High and dMMR cleared that bar because the biology, the diagnostics, and the trial results told one coherent story. Any future claim of a universal cancer biomarker deserves to be measured against the same three-part standard.