Press Releases Are ‘Incomplete Substitutes’ For FDA Letters

FDA analysis finds few matches between ‘complete response’ letters, press releases. But few specific factors about sponsors and NDAs could clearly be shown to be predictive of disclosure.

FDA looked at many different attributes of sponsors and applications to try to determine if any characteristics were predictive of communication of “complete response” letter contents. The small dataset did not turn up many statistically significant associations, but the findings are interesting in the portrait they paint of the applications that received CRLs.

FDA counted 687 "statements" describing specific application deficiencies in the 61 CRLs identified for new molecular and biologic entities from 2008-2013, FDA’s Peter Lurie and colleagues explained. Press releases "matched" CRL statements when the company covered the same issue; press releases are much shorter than CRLs and were not required to provide the same level of detail. FDA found a median of eight statements per CRL.

While there were press releases with at least one matching statement for 61% of the CRLs, only a small fraction of the total number of CRL statements were matched by a statement in a press release. Of the 687 FDA statements, only 93 were matched in press releases: 14%.

The study did not find any strongly predictive variables in the statement dataset. No characteristic of statement matches "was associated with a matching rate exceeding 26% at the statement level," Lurie, et al., noted.

Chemistry, manufacturing and controls information was most likely to be included in press releases with a 25% matching rate (16 of 63 CRL statements were reflected in press releases). Clinical pharmacology information was most likely to be omitted (76 of the 82 CRL statements were omitted from press releases, for a 7% matching rate).

A need for a new clinical trial was one of the variables more likely to be communicated in press releases. The study found a 49% matching rate when "FDA recommended a new trial for safety or efficacy reasons," a recommendation appearing in 45 statements in 32 CRLs. Nonetheless, the study also found that sponsors are not likely to communicate the substance of some serious safety concerns. "Most findings associating the drug with a higher mortality rate went unmentioned in press releases," it states.

"Insufficient evidence of efficacy" was the most common efficacy concern, with 53 statements in 29 letters, the study reported. "Requires new clinical trial for efficacy" was the most commonly matched efficacy subdomain – of 28 CRL statements, 16 (57%) were matched in press releases.

Matching rates were "higher among public companies and lower among large companies," the authors observed (see chart). The findings are not surprising, as private companies are not subject to the same SEC regulations that require publicly traded firms to disclose information of material impact, and a non-approval action is more likely to make a material impact on a small firm with few products than on a big pharma behemoth.

Who’s More Likely To Get A CRL?
Confidence intervals in the study were large and the authors acknowledged that the task of quantifying and classifying FDA and corporate statements is "inherently subjective," but they concluded that the analysis shows clearly that "press releases are generally an incomplete source of the reasons for FDA non-approval of applications."
	Number of CRLs (n=61)	CRLs with at least one matched statement (n=37)	Relative risk of matched statement (95% Confidence Interval)
Sponsor Characteristics
Publicly traded	52 (85%)	36 (69%)	6.23 (0.97 to 39.90)
Private	9 (15%)	1 (11%)
Large (>750 employees)	46 (75%)	25 (54%)	0.68 (0.47 to 0.98)
Small (<750 employees)	15 (25%)	12 (80%)
Application Designations
Orphan	17 (28%)	13 (76%)	1.40 (0.96 to 2.04)
Not orphan drug	44 (72%)	24 (55%)
Priority Review	12 (20%)	8 (67%)	1.13 (0.71 to 1.79)
Standard Review	49 (80%)	29 (59%)
First-In-Class	20 (33%)	14 (70%)	1.25 (0.84 to 1.85)
Not first-in-class	41 (67%)	23 (56%)
Advisory Committee Status
Referred to AdCmte	35 (57%)	22 (63%)	1.08 (0.72 to 1.65)
Not Referred to AdCmte	26 (43%)	15 (58%)
AdCmte Favored Approval	18 (53%)	11 (61%)	0.98 (0.58 to 1.66)
AdCmte Did Not Favor or Vote On Approval	16 (47%)	10 (63%)
FDA Approval
Subsequently Approved	25 (41%)	17 (68%)	1.22 (0.82 to 1.82)
Withdrawn or Pending	36 (59%)	20 (56%)
Source: Selected data from "Comparison of content of FDA letters not approving applications for new drugs and associated public announcements from sponsors: cross sectional study" by Peter Lurie, Harinder Chahal, Daniel Sigelman, Sylvie Stacy, Joshua Sclar and Barbara Ddamulira, published in BMJ June 10, 2015.