KEYNOTE-158: What the Cancer Trial Found

An idea worth explaining first

Cancer treatment has usually been organized by where a cancer started. Lung drugs for lung cancer, breast drugs for breast cancer.

KEYNOTE-158 belongs to a different idea. It asked whether one measurable feature of a tumor's DNA could predict who responds to an immunotherapy drug, no matter which organ the cancer came from.

The feature is tumor mutational burden, usually shortened to TMB. It counts how many mutations a tumor carries per megabase of DNA. A megabase is a million letters of genetic code. The theory is simple. More mutations means more abnormal proteins on the cell surface, and more for an immune system to spot.

Pembrolizumab is a checkpoint inhibitor. It blocks PD-1, releasing a brake on T cells. Our overview of immunotherapy covers how that class works.

A basket, not a comparison

KEYNOTE-158 was a phase 2 basket trial. A basket trial gathers people with different cancers who share one feature and treats them all the same way.

It ran at 81 academic and community sites across 21 countries. Enrollment ran from January 2016 to June 2019, and 1,073 people joined.

Everyone had an advanced solid tumor that could not be cured. Ten types were eligible: anal, biliary, cervical, endometrial, mesothelioma, neuroendocrine, salivary, small-cell lung, thyroid, and vulvar. Everyone had already progressed on at least one standard treatment, or could not tolerate it. And everyone was well enough to be graded 0 or 1 on the ECOG performance scale.

Participants received pembrolizumab into a vein every three weeks, for up to 35 cycles. The amount was set by the protocol and given by the treating team.

There was no comparison group. Everyone got the drug.

The result

Tumor DNA was tested with one commercial assay, the FoundationOne CDx. The cutoff for "TMB-high" was set in advance at 10 or more mutations per megabase.

Of the 790 people with usable TMB data, 102 were TMB-high. That is 13%. The other 688, or 87%, were not. Median follow-up was 37.1 months.

Tumors shrank meaningfully in 30 of the 102 TMB-high patients. That is a response rate of 29%, with a 95% confidence interval of 21% to 39%. In the group that was not TMB-high, it was 43 of 688, or 6%, with an interval of 5% to 8%.

The FDA's own summary adds detail. Within that 29% there was a 4% complete response rate and a 25% partial response rate. The median duration of response was not reached. Fifty-seven percent of responses lasted at least 12 months, and 50% lasted at least 24.

On safety, 105 TMB-high patients were assessed. Eleven had a serious treatment-related event, and 16 had one graded 3 to 5. Colitis, meaning inflammation of the colon, was the only such event that happened to more than one person. One person died of treatment-related pneumonia.

What the FDA did with it

On June 16, 2020, the FDA granted accelerated approval to pembrolizumab for adults and children with TMB-high solid tumors. The tumors had to be unresectable or metastatic. TMB-high was defined as 10 or more mutations per megabase, measured by an FDA-approved test.

Two conditions were written into that indication. The cancer must have progressed after prior treatment, and there must be no satisfactory alternative option.

The FDA approved the FoundationOneCDx assay as the companion diagnostic on the same day. The label also carries a limitation of use. Safety and effectiveness were not established in children with TMB-high cancers of the central nervous system.

This is one of a small number of tumor-agnostic approvals. Eligibility rests on a molecular feature rather than on an organ. Our page on biomarker testing explains what those tests measure.

When to get checked

This trial is about people already diagnosed with advanced cancer, so the practical warning signs concern treatment rather than detection.

Checkpoint inhibitors can turn the immune system against healthy organs. The FDA names the pattern. It includes pneumonitis in the lungs, colitis in the bowel, hepatitis in the liver, endocrine gland problems, nephritis in the kidneys, and skin reactions.

Anyone on pembrolizumab should call the oncology team promptly, rather than waiting, about:

• Diarrhea that is new, frequent, or contains blood.
• New shortness of breath or a cough that will not settle.
• Yellowing of the skin or eyes, or dark urine.
• A rash that spreads or blisters.
• Unusual fatigue, thirst, or feeling cold all the time, which can signal a thyroid or adrenal problem.

These reactions are treatable. How well that goes depends heavily on how early they are caught.

What this trial cannot tell you

There was no control group. So there is no way to know how these same people would have fared on a different treatment. Tumor shrinkage was the measure, not survival.

Even inside the group the test picked out, 71% of TMB-high patients did not respond. A 29% response rate marks a subgroup worth trying the drug in. It is not a reliable predictor for one person.

The result also rests on one assay and one cutoff. Other tests and other thresholds may not behave the same way. The 10 mutations per megabase line has been argued over since.

Finally, the ten tumor types studied left out the most common cancers. Breast, prostate, colorectal, and non-small-cell lung were not among them. Reading this as "works in any cancer" goes past what the data support, and the FDA's own indication is narrower than that.

Sources

• Marabelle A et al., Association of tumour mutational burden with outcomes in advanced solid tumours treated with pembrolizumab (KEYNOTE-158), Lancet Oncol 2020 (NCBI E-utilities record)
• FDA approves pembrolizumab for adults and children with TMB-H solid tumors
• ClinicalTrials.gov record for NCT02628067 (KEYNOTE-158)

How this article was prepared

An AI-assisted editorial system helped prepare this page. No named medical reviewer has reviewed it unless one is listed.

See an error, old source, or unclear wording? Tell us.