AMAG’s Explanation For Makena's Failed Confirmatory Trial Faces Skeptical US FDA
Executive Summary
Post hoc analyses show no convincing evidence that PROLONG trial's failure to meet its coprimary efficacy endpoints stemmed from baseline risk differences in subjects compared to an earlier trial, agency says; advisory committee is being asked whether to withdraw accelerated approval or keep the drug on the market, with or without a new confirmatory study.
AMAG Pharmaceuticals Inc. may be hard-pressed to persuade a US Food and Drug Administration advisory committee that Makena, an accelerated approval drug for preventing preterm birth, should remain on the market given agency skepticism about the company’s view of why the confirmatory efficacy trial failed.
At a 29 October meeting of the Bone, Reproductive and Urologic Drugs Advisory Committee, the company will assert results from the unsuccessful PROLONG trial were influenced by baseline differences in the enrolled population relative to an earlier trial that led to accelerated approval, and that post hoc analyses from the failed study suggest a treatment benefit for higher-risk patients enrolled in the US.
Advisory Committee Voting Questions
-
Do the findings from Trial 003 verify the clinical benefit of Makena on neonatal outcomes?
-
Based on the findings from Trial 002 and Trial 003, is there substantial evidence of effectiveness of Makena in reducing the risk of recurrent preterm birth?
-
Should the FDA: (A) Pursue withdrawal of approval for Makena; (B) leave Makena on the market under accelerated approval and require a new confirmatory trial; or (C) leave Makena on the market without requiring a new confirmatory trial?
In a briefing document released 25 October, the agency takes a dim view of this argument.
Post hoc, exploratory analyses suggest no evidence of statistically significant interactions by US/non-US region or patient risk factors for preterm birth and a treatment effect with Makena (hydroxyprogesterone caproate injection, also known as HCP or 17P), the agency said.
There also was not “consistent, convincing evidence of treatment benefit within a specific subpopulation across the two trials,” the FDA’s briefing document states.
Three Options
For the first time in eight years, the FDA is asking its external experts whether it should withdraw a drug or indication under accelerated approval for failure to verify clinical benefit. (Also see "Makena’s Accelerated Approval Will Hang In Balance At US FDA Panel Meeting In October" - Pink Sheet, 23 Sep, 2019.)
The agency poses three voting questions to the panel, the last of which asks whether or not Makena should remain on the market. (See box.)
The agency describes the considerations that should go into panelists’ choice on the third voting question.
“Vote (A) (withdraw approval) may be appropriate if you believe the totality of evidence does not support Makena’s effectiveness for its intended use,” the draft questions state. “Discuss the consequences of Makena removal.”
A vote to require a new confirmatory trial while the drug stays on the market (option B), may be appropriate if panelists believe the totality of the evidence supports Makena’s effectiveness in reducing the risk of recurrent preterm birth, but there is no substantial evidence of effectiveness on neonatal outcomes. “Vote (B) would also reflect a belief that a new confirmatory trial is necessary and feasible,” the questions state.
Panelists voting for B are asked to discuss how the existing data provide substantial evidence of effectiveness based on the surrogate endpoint of gestational age at delivery. The FDA also seeks discussion on key study elements – including study population, control, dose and efficacy endpoints – for a new confirmatory trial.
“Vote C (leave Makena on the market without a new confirmatory trial) may be appropriate if you believe Makena is effective for reducing the risk of recurrent preterm birth and that it is not necessary to verify Makena’s clinical benefit in neonates,” the questions state. The agency requests panelists voting for this option discuss how the existing data provide substantial evidence of effectiveness and “why it is not necessary to verify Makena’s clinical benefits in neonates.”
Re-examining Accelerated Approval
Makena is the only pharmacotherapy approved to reduce the risk of recurrent preterm birth. It received accelerated approval in 2011 based on results from a National Institutes of Health study (Trial 002, also known as the Meis trial), in which Makena (HPC 250mg) intramuscular injection reduced the proportion of women who delivered <37 weeks gestation, from 55% (placebo) to 37% (Makena).
The confirmatory trial (PROLONG, also known as Trial 003), had two coprimary efficacy endpoints: delivery prior to 35 weeks gestation; and a neonatal morbidity/mortality composite index, which was considered reflective of a direct clinical benefit.
Trial 003 randomized a total of 1,708 women from nine countries, with the US accounting for 23% of the enrolled population. Data were available for 1,651 live neonates.
Trial 003’s failure to confirm an effect on gestational age of delivery, the surrogate endpoint used in the original trial, “raises the question as to whether Makena’s accelerated approval is still supported by substantial evidence of effectiveness for the reduction in recurrent preterm birth.” – FDA
The trial failed to demonstrate a statistically significant treatment effect for the proportion of women delivering prior to 35 weeks (11% Makena compared to 12% placebo, p=0.72) or neonatal composite index (5.4% Makena compared to 5.2% placebo, p=0.84).
“Also, no differences between Makena and placebo were seen in the secondary outcomes related to other gestational ages at delivery (<37 weeks [23% Makena vs. 22% placebo, p=0.57]), <32 weeks gestation [4.8% Makena vs. 5.2% placebo, p=0.70]) or for the individual components of the neonatal index,” the FDA’s briefing document states.
Since Trial 003 failed to demonstrate a reduction in adverse neonatal outcomes from preterm birth, as measured by the neonatal morbidity/morality composite index, “the clinical benefit of Makena remains unverified,” the agency said.
In addition, Trial 003 failed to confirm an effect of Makena on gestational age of delivery, the surrogate endpoint used in Trial 002 to support accelerated approval. “This raises the question as to whether Makena’s accelerated approval is still supported by substantial evidence of effectiveness for the reduction in recurrent preterm birth.”
Differences In Risk Factors
AMAG asserts the PROLONG results were adversely influenced by differences in the study population relative to patients enrolled in the Meis trial. (Also see "Makena Might Stay On Market Despite Failed Confirmatory Trial In Pre-Term Birth " - Pink Sheet, 8 Mar, 2019.)
“While the entry criteria of Meis and PROLONG were similar, the study population in PROLONG was different than that of Meis, with the latter comprised of a higher risk population,” AMAG said in its briefing document.
“When comparing demographics and baseline characteristics from PROLONG and Meis, the differences across race and other potential surrogates of socioeconomic status that have been linked to higher rates of PTB were noteworthy, with most of those differences … driven by the ex-US PROLONG subset population,” the company said. “As a result, key differences in baseline risk associated with PTB even within the PROLONG study population, notably US vs. ex-US subset populations, make the applicability of the efficacy data particularly challenging in the US.”
“Exploratory analyses of PTB rates by baseline risk suggest an increasing treatment benefit associated with 17P with increasing levels of baseline risk for recurrent PTB.” – AMAG
AMAG said that compared to the US PROLONG subset and patients enrolled in the Meis trial, the ex-US PROLONG population represented a cohort with a lower baseline risk for preterm birth based on the following factors:
-
Prior spontaneous PTB: In ex-US PROLONG, 11% had more than one prior spontaneous PTB, compared to 27% in US PROLONG and 32% in Meis.
-
Race/ethnicity: In ex-US PROLONG, only one patient was Black or African American, compared to 29% in US PROLONG and nearly 60% in Meis. Hispanics or Latinos accounted for approximately 8% of patients in ex-US PROLONG, 14% in US PROLONG and 15% in Meis.
-
Marital status: In ex-US PROLONG, 4% of patients were unmarried with no partner, compared to 31% in US PROLONG and 50% in Meis.
-
Substance use: In ex-US PROLONG, approximately 4% of patients reported any substance use during pregnancy (smoking, alcohol or illicit drugs), compared to 28% in US PROLONG and 26% in Meis.
“Exploratory analyses of PTB rates by baseline risk suggest an increasing treatment benefit associated with 17P with increasing levels of baseline risk for recurrent PTB,” the company said. “Treatment effect was observed at <37, <35, and <32 weeks gestation for the highest risk group (Meis), while the lowest risk group (ex-US PROLONG) showed no effect. Trends favoring 17P emerge in the US PROLONG subset as the population becomes more similar to that of Meis, with increased effect at <35 and <32 weeks, but not at <37 weeks gestation.”
Despite the failure of PROLONG to meet its primary efficacy outcome, “the strong efficacy data from the Meis study, previous supporting clinical trial data in the US, and trends favoring treatment benefit for 17P in post hoc analyses focused on patients enrolled in the US, coupled with a favorable safety profile, support the continued use of 17P,” the company said.
‘Convincing Evidence Of Efficacy’ Lacking
However, the agency appears unconvinced by AMAG’s argument premised upon baseline differences in the study populations.
“Generally, FDA does not support unplanned exploratory subgroups analyses, especially when the overall result does not demonstrate efficacy,” the agency's briefing document states. “When such post hoc subgroup analyses are used to search for evidence of benefit, there is a high probability that any observed favorable subgroup results are due to chance alone. Therefore, FDA considers exploratory analyses hypothesis-generating.”
“When such post hoc subgroup analyses are used to search for evidence of benefit, there is a high probability that any observed favorable subgroup results are due to chance alone.” – FDA
Nevertheless, the agency conducted its own exploratory analyses and found no treatment difference by US/non-US region or by race, and no evidence of a treatment effect within the US subgroup in Trial 003.
Exploratory analyses “do not provide convincing evidence of efficacy over placebo in any subpopulation and there is no statistically significant interaction between Makena” and any of the risk factors for preterm birth that were described in AMAG’s briefing document, the agency said.
“Although these risk factors may have an impact on the overall PTB or neonatal composite index rate, there was no evidence in Trial 003 that they impact the treatment effect nor was there consistent convincing evidence of an effect within a specific subpopulation across the two trials,” the FDA said.
Safety Profile Unchanged
The safety profile of Makena appears unchanged based on the results of Trial 003, the agency said.
The confirmatory trial was powered to exclude a doubling of the risk of fetal/early infant deaths with Makena, and this level of increased risk was successfully ruled out.
“When the data from Trial 002 and 003 were pooled, there was no difference in the overall incidence of fetal/early infant deaths with Makena compared to placebo in either trial," the FDA said. "There appeared to be a trend toward an increase in stillbirths in both trials; however, the numbers are small, precluding reliable determination of risk.”
Utilization Data
The agency’s briefing document includes data from a Sentinel query, which was conducted to try to characterize the context of real-world use of HPC, the drug substance in Makena but which also is compounded. The query found that the use of HPC or progesterone remains low among pregnancies having a related medical conduction, including history of preterm delivery (15%).
The agency also analyzed use patterns for injectable HPC and oral, vaginal and injectable dosage forms of progesterone based on outpatient prescription dispensing data.
Dispensed prescriptions for injectable HPC (Makena and generics) increased from 16,600 in 2014 to 106,000 in 2018. However, this represents a small proportion of the total use of progesterone products, for which there were 4.6m prescriptions dispensed in 2018, the agency said.