FDA panel doesn't buy Merck's biomarker for fast Keytruda nod in TNBC, spelling trouble for other I-O meds

An independent FDA expert panel has voted 10-0 to recommend deferring a decision on Merck's application for Keytruda for the neoadjuvant and adjuvant treatment of patients with high-risk, early-stage triple-negative breast cancer. (FDA)

Merck & Co. isn’t likely to snag an FDA go-ahead for Keytruda’s use in pre- and post-surgery triple-negative breast cancer anytime soon—and that fact should be a warning to other drugmakers eyeing approvals in early-stage cancer, one analyst said.

The immuno-oncology giant went to the agency too early with an application for approval in early breast cancer, an FDA advisory committee determined Tuesday, echoing agency reviewers' comments days before.

The panel of 10 oncology experts voted unanimously that Merck needs to wait for longer-term data from the Keytruda-522 trial before reasonably asking the agency to approve Keytruda in the so-called neoadjuvant and adjuvant settings of high-risk, early-stage TNBC.

The setback flags tough regulatory scrutiny ahead for all immuno-oncology agents across the large early-stage solid tumors market, SVB Leerink analyst Daina Graybosch wrote in a Wednesday note—a market that's key for PD-1/L1 drugs in general, and Keytruda in particular.

The expert panelists raised some of the same concerns FDA staffers highlighted in their review (PDF) ahead of Tuesday's meeting. The message from both sources was clear—and it may not just dash Keytruda's hopes for an accelerated approval by March, but point up the bar for future applications in similar indications, Graybosch said.

Merck based its application on Keynote-522, which tests adding Keytruda to chemotherapy before surgery followed by Keytruda as monotherapy after surgery. The company applied for accelerated approval using a surrogate marker—pathological complete response (pCR)—defined as the absence of residual cancer cells from tissue samples after treatment.

Thing is, though widely used in early-stage breast cancer, pCR remains a controversial marker with limited evidence that it can predict patient survival. So the FDA wanted to see a large pCR improvement.

RELATED: ESMO: Merck's Keytruda posts first-in-class victory in pre-surgery breast cancer

But at an interim analysis of Keynote-522, the absolute difference between pCR in the experimental and control arms was just 13.6 percentage points—64.8% versus 51.2%.

The FDA said it had told Merck ahead of time that even a 15% absolute improvement in pCR rate may not be enough to predict long-term outcomes in early-stage TNBC.

The story could be different for immuno-oncology drugs like Keytruda; pCR might be more closely correlated, justifying more aggressive early treatment, Graybosch said. But shifting that mindset—and changing practice—“will require strong supportive data,” she said.

And that's something Merck doesn't have, she said—and isn't likely to get from ongoing trials. 

“Neither of Merck’s ongoing early-stage TNBC trials will provide the necessary evidence, in our view,” Graybosch wrote.  The other study refers to Keynote-242, which tests adjuvant Keytruda after surgery in patients who don't achieve pCR with neoadjuvant chemo.

Keynote-522 does also measure Keytruda's ability to cut the risk of disease return or progression after surgery, as well as to extend patients' lives. Those numbers were trending in Keytruda’s favor in a third interim analysis, but they weren't yet statistically significant.

RELATED: SABCS: Merck's Keytruda shows up Roche's Tecentriq with chemo combo win in triple-negative breast cancer

Merck's expected to have its fourth interim readout from Keynote-522 in the third quarter, and Graybosch said she remains positive the study will eventually be successful. 

But survival data could take a long time to come. “In this disease, particularly for patients that have had a pathologic complete response, it takes many years of follow-up to determine the outcome,” Roy Baynes, Merck’s head of global clinical development, said in a recent interview. He acknowledged that event-free survival over time is the “hard endpoint” but insisted that the pCR data at this stage are already important for patients.

Even if Keytruda's pCR results had been strong, that marker only measures the pre-surgery treatment effect. The FDA took issue with Merck pooling data on that stage of the trial with post-surgery results in analyzing patient survival.

“There is uncertainty with whether neoadjuvant and adjuvant treatments are both needed because we cannot evaluate the contribution of each treatment phase in this trial design,” FDA staffers wrote in their review.

RELATED: How can the $23B market for PD-1 blockers grow? 'Pan-adjuvant' use is key: report

This could spell regulatory hurdles ahead for Keynote-522 and more broadly could raise “regulatory and commercial skepticism” for other neoadjuvant/adjuvant trials that don't separately analyze the two phases of treatment, Graybosch said.

A long list of PD-1/L1 trials for early-stage cancers could be at risk, she figured. These include Keytruda’s Keynote-585 trial in gastric and gastroesophageal junction (GEJ) cancer; AstraZeneca’s Niagara trial for Imfinzi in muscle-invasive bladder cancer; and Bristol Myers Squibb’s CheckMate-7FL study for Opdivo in ER-positive, HER2-negative breast cancer, among others.

Expansion into early disease in neoadjuvant and adjuvant treatment settings will be important for all PD-1/L1s. Graybosch currently expects nearly half of Keytruda’s growth through 2028 will come from these uses.

Editor's note: Carly Helfand also contributed reporting.