February 9, 2021
The National Center for Health Research is a nonprofit think tank that scrutinizes the safety and effectiveness of medical products. We don’t accept funding from companies that make those products, so we have no conflicts of interest. We welcome the opportunity to provide our views on Merck’s application for approval of Keytruda for the indication of high-risk early stage triple negative breast cancer.
Triple negative breast cancer has a lower survival rate than other breast cancers. However, chemotherapy clearly improves 5-year survival. Patients need additional treatment options but the bottom line for patients is that FDA should not approve an indication that is not proven to have clinically meaningful benefits, especially when the treatment has clear risks.
The first issue to address is whether there is evidence that immune checkpoint inhibitors (ICIs) are effective for TNBC. We agree with FDA scientists that “there is still uncertainty regarding ICIs for TNBC” based on the results from several clinical trials.
- KEYNOTE-119 failed to meet its primary OS endpoint.
- KEYNOTE-355 has not met its OS endpoint.
- IMpassion130: clinical benefits need to be confirmed
- IMpassion131 interim OS results favored control group
The second major issue pertains to pCR data in the study. The results indicate only 7.5% improvement in pCRs at IA3 (the most recent interim analysis), which the FDA scientists point out may not be clinically meaningful even if statistically significant. We agree. The problem is that it is impossible to determine how this slight improvement would affect overall survival, and even if it does, how much neoadjuvant and adjuvant use each contribute to any benefit.
FDA scientists were clear to the sponsor that there were concerns with their study design and that the application for approval was premature since the study was not yet completed. The agency made it clear that that the event free survival (EFS) study results were not statistically significant, not clinically meaningful, and did not show a “stable trend.”
FDA reviewers are clear that data on overall survival “are too immature to provide a conclusive interpretation regarding the difference in OS between treatment arms.”
What about safety?
At IA3 (the most recent interim analysis), there were 96 deaths, which FDA points out “accounts for only 32% of the events needed for the final analysis. Therefore, the OS estimate may be unreliable, and the treatment effect size reported is subject to uncertainty.”
It is notable that the study included patient-reported outcomes (PROs), but unfortunately, KEYNOTE-522 was not designed to compare differences in PROs (symptoms, side effects, health-related quality of life), nor were these patient-reported endpoints prospectively identified and statistically tested.
PRO assessments should have been more frequent, both for neoadjuvant and adjuvant treatments
Since many high-risk, early-stage TNBC patients will be cured with standard therapy, the key issue is whether this drug has benefits that outweigh the risks. The benefits are unclear. Therefore, the evidence of the drug’s “added toxicity” is worrisome. FDA scientists concluded that “Some of these toxicities may be irreversible or require lifelong medication in patients cured of their breast cancer.
Although the sponsor counted 2 deaths due to immune-mediated adverse events, the FDA counted 4. Either way, these deaths must be considered worrisome given the lack of clear evidence of a meaningful benefit. And, there are many other serious adverse events in addition to the small number of deaths. All-grade and grade ≥3 immune-mediated AEs and infusion reactions occurred more frequently in Keytruda patients compared to placebo: 43% vs. 22% for all grade AEs, and 15% vs. 2% for high grade AEs. In fact, 10% of Keytruda patients had immune-mediated AEs and infusion reactions leading to hospitalization compared to 1% of placebo. These included the following relatively high number of adverse events:
- Infusion reactions (18%),
- Hypothyroidism (15%),
- Severe skin reactions (6%)
- Hyperthyroidism (5%), adrenal insufficiency (3%), pneumonitis (2%), and thyroiditis (2%).
It is important to note that these adverse events were not resolved at the last assessment in the study for 19% of Keytruda patients. It is also important to note that 16% of the Keytruda patients initiated thyroid hormone replacement during the study.
In summary, we agree with FDA scientists that the deaths are “particularly concerning in this curative disease setting.”
- “All grade and grade ≥3 immune-mediated AEs were increased in [Keytruda] patients.’
- Some “may be severe or lifelong.”
- The adjuvant treatment has fewer adverse events but “has not demonstrated a significant effect on any efficacy endpoint, and may be adding risk without benefit.”
Based on our analysis, we agree with the overall conclusions made by FDA scientists:
- Neoadjuvant Keytruda “confers only a small absolute improvement in pCR rate of questionable clinical meaningfulness.”
- Event-free survival and overall survival are “immature and unreliable.”
- “The design and results of KEYNOTE-522 do not currently support a role for adjuvant [Keytruda].”
- “Supportive data of clinical benefit … are lacking.”
- Adding Keytruda “is associated with increased toxicity … which may be severe, irreversible, and/or require life-long medication in potentially curable and otherwise healthy patients.”
In conclusion, the FDA and the medical community do patients no favors to approve a treatment that is not proven to benefit them and at the same time is proven to cause harm for a substantial percentage of patients. The studies should be continued to determine whether the benefit of adding Keytruda to other treatments outweigh the risks.
This written statement was submitted to the FDA on February 8, 2021 and an oral version with PowerPoint slides was presented at the FDA Advisory Committee meeting on February 9, 2021.
We are pleased that the FDA Advisory Committee agreed with our views and voted 10-0 on February 9 in favor of deferring an FDA regulatory decision until the study is completed.