The change in the proportion of voting ODAC members with FCOIs over time. FCOIs indicates financial conflicts of interest; ODAC, Oncologic Drugs Advisory Committee.
eTable 1. Characteristics of included meetings.
eTable 2. List of meetings, drugs discussed and citations to published trials.
eTable 3. Breakdown of conflicts of interest with designated monetary values for advisory committee members and voting consultants.
Customize your JAMA Network experience by selecting one or more topics from the list below.
Tibau A, Ocana A, Anguera G, et al. Oncologic Drugs Advisory Committee Recommendations and Approval of Cancer Drugs by the US Food and Drug Administration. JAMA Oncol. 2016;2(6):744–750. doi:10.1001/jamaoncol.2015.6479
The US Food and Drug Administration (FDA) advisory committees influence decisions relating to the regulatory approval of drugs in the United States. Outside of the field of oncology, prosponsor voting bias has been observed among members with financial conflicts of interest (FCOIs).
To explore factors associated with Oncologic Drugs Advisory Committee (ODAC) recommendations and the influence ODAC members’ FCOIs on the drug approval process.
Design, Setting, and Participants
Retrospective analysis of 82 ODAC meeting transcripts between January 2000 and December 2014. Analysis was restricted to meetings at which votes were cast relating to oncologic drugs. The influence of methodology of trials supporting approval and frequency and type of self-reported FCOIs of voting members was explored using logistic regression.
Main Outcomes and Measures
ODAC recommendation for drug approval and subsequent FDA approval.
Eighty-two transcripts of ODAC meetings between January 2000 and December 2014 were available for analysis. During the time period analyzed, ODAC members voted on 68 applications in 79 meetings (the remaining 3 meetings included voting questions regarding postmarketing safety or trial design). There was agreement between ODAC recommendations and final FDA approval; FDA approval was received for all 41 drugs that ODAC recommended approval. Additionally, the FDA approved 7 out of 41 agents that were not recommended for approval by ODAC (κ = 0.83). In 51 of 79 meetings, more than 1 trial was available to support the indication of a particular drug, and favorable ODAC recommendations were more likely when this was the case (odds ratio [OR], 1.82; 95% CI, 1.19-2.78; P = .01). Availability of randomized data did not appear to be important with selected single-arm phase 2 trials leading to recommendations for approval, especially in rare diseases. There has been a significant reduction in FCOIs over time (31 of 77 voting members [40%] in 2000 vs 0 of 20 voting members in 2014 [0%]; P < .001). Recommendations for approval were made in 28 of 47 meetings with members reporting FCOIs while among meetings with no reported FCOIs, recommendations for approval were made in 13 of 35 meetings (OR, 1.19; 95% CI, 0.97-1.46; P = .10). No significant association between ODAC recommendations and FDA approval was observed for members with FCOIs with the sponsor (OR, 1.79; 95% CI, 0.97-1.46; P = .19 and OR, 3.48; 95% CI, 0.84-14.35; P = .09, respectively) compared with members with FCOIs with competitors (OR, 1.06; 95% CI, 0.78-1.44; P = .72 and OR, 0.94; 95% CI, 0.69-1.28; P = .69, respectively).
Conclusions and Relevance
Availability of multiple trials is associated with higher odds of ODAC recommendation and drug approval. Availability of randomized data appears less important. Declaration of FCOIs among ODAC members was frequent during the time period of interest but has decreased significantly over time. There is no apparent association between FCOIs and ODAC recommendations and subsequent FDA approval.
The US Food and Drug Administration (FDA) advisory committees influence decisions relating to the regulatory approval of drugs in the United States, thus having a substantial influence on patterns of clinical practice. These committees, which are asked typically to advise on complex cases, meet publicly to discuss the premarket approval process and issues highlighted by the FDA. The committees then vote on 1 or several questions posed in advance by the FDA. Although the FDA is not bound by the committees’ recommendations (which are strictly advisory), there is a high rate of concordance between advisory panel votes and the eventual FDA approval decision,1 suggesting that an advisory panel’s recommendation plays a critical role on decisions regarding approval. Despite their critical role in regulatory approval, factors influencing the recommendations of FDA advisory committees are not well described.2
The role of financial conflicts of interest (FCOIs) in the conduct and reporting of clinical trials3 and recommendation of specific therapies4 has received widespread attention. While the influence of FCOIs of unselected FDA advisory committees has been studied previously,5 less is known about the effect of such FCOIs in the approval process of cancer drugs.
Here, we analyze the influence of the Oncologic Drugs Advisory Committee (ODAC) on the FDA’s oncologic drug approval process and factors associated with both ODAC recommendations and final FDA approval.
Question: What factors influence cancer drug approval, and is there evidence for voting bias among US Food and Drug Administration (FDA) Oncologic Drugs Advisory Committee (ODAC) members with financial conflicts of interest?
Findings: Votes in favor of drug approval were more common when multiple trials supported the application, but randomization appeared less important with approvals occurring for drugs studied in single-arm trials, especially in rare diseases. No significant associations between financial conflicts of interest and ODAC recommendation or subsequent FDA approval were observed.
Meaning: The availability of multiple trials improve the odds of ODAC recommendation and drug approval, which appear to be free of financial conflict of interest–associated bias.
We searched the FDA website6,7 to identify all ODAC meetings between January 2000 and December 2014. We focused on drugs used in the treatment of adult or pediatric malignancy, including both antineoplastic and supportive care agents.
The ODAC meeting transcripts were the primary data source.6,7 If transcripts were not available, other sources were consulted, including webcasts, agendas, summaries of minutes, waivers, slides of presentations,6,7 labels, and medical officer review documents.8 If available, individual conflict of interest documents were also consulted for all committee members and voting consultants.
Meetings that covered a single topic over more than 1 day were considered a single meeting, while a meeting that discussed multiple products sequentially over 1 day was considered multiple meetings (1 meeting for each product), as were meetings discussing multiple indications for a single product (1 for each indication).
According to section 3.4 from the Health Research Ethics Authority of Newfoundland and Labrador, Canada, research based on the review of published and/or publicly reported literature does not require ethics review.9
All data were extracted by one author (A.T.) using predesigned electronic forms. Data were then verified by a second author (E.A.). Data extraction was conducted between September 2014 and April 2015. The body of the transcript was used to extract the meeting date, drug name, sponsor name, origin of sponsor (US vs non-US) and topic (proposed indication vs postsafety discussion), as well as data on advisory committee members (ODAC permanent members, including those voting and those recused or with waivers that allowed discussion but no voting) and voting consultants (temporary members).
For applications submitted to the ODAC meeting we collected data regarding submission type (initial indications vs supplemental indications), type of application (New Drug Application or Biologic Licensing Application) and proposed approval type (regular approval vs accelerated approval). Transcripts were also reviewed to identify clinical trial information, including number of trials (both pivotal trials and other supportive trials of any phase) supporting the proposed indication. For pivotal trials, we collected data on sample size (largest trial when more than 1 trial was included), trial design (randomized vs single-arm), phase of clinical trial (phase 2 vs phase 3), US accrual vs low and/or no US accrual (<10% of patients from the United States), treatment intent (prevention, adjuvant and/or curative, palliative, supportive), primary end points supporting the application (overall survival vs surrogate and/or intermediate end points), the result of primary analysis and whether or not approval was based on a subgroup analysis of the pivotal trial. We used the end point definitions that were used in the applications submitted to the FDA.6,7 For studies with coprimary end points, we identified the primary end point chosen by the FDA to support approval. The application number was also collected to identify whether or not the drug was approved. Drug labels were used to identify the date of approval and indication for products approved, as well as to confirm information collected on transcripts.8
Data on conflicts of interest for advisory committee members and voting consultants were collected as specified in form FDA 341010 and extracted from the executive secretary’s oral statement at the beginning of each meeting. We also extracted data on restrictions or recusals. For each voting member with a declared conflict, we entered each conflict as a separate entry. Consequently, each member could have multiple conflicts of various types at each meeting, including investments, employment, consultancy and/or advisory capacity, research funding, and speakers’ bureau activities or lectures. Specific details of the conflict and whether it involved the sponsor or a competing company were also collected if available.
The objectives of meetings varied with many meetings not holding formal votes. As such, we restricted our analysis to meetings at which dichotomous votes were cast for at least 1 question relating to a specific oncologic product. We selected a single question for each meeting using hierarchal categories described previously.5 Specifically, the following order of priority from highest to lowest was used: (1) questions considering whether to recommend approval of a drug, approval of an indication or withdrawal of a drug, including questions on whether both safety and efficacy had been established or if the drug had a favorable risk-benefit profile; (2) questions considering whether to recommend accelerated approval of a drug; (3) questions considering whether either safety or efficacy had been established for a drug; and (4) questions considering if more trials should be performed before drug approval. For each included question, we first used the meeting transcript to determine how the FDA advisory committee as a whole voted, with responses coded as either favoring or not favoring the application.
To identify reasons why ODAC members voted in favor or against drug approval, we identified those transcripts where members who voted stated their vote and the reasons why they voted as they did. We grouped these reasons in 3 major categories: efficacy, safety, and other (ie, trial design and statistical analysis, unmet medical need, confirmatory trials in progress, or issues relating to accrual in the United States). We reported the proportion of members explaining their vote for every category.
Finally, FCOIs were categorized as major or minor based on their effect on the potential for bias11 as we previously reported.4 Major FCOIs included employment, a significant investment interest (>$50 000 in value), or participation in speakers’ bureau. Other FCOIs were described as minor. If an author declared several stocks under $50 000, the FCOI was classified as minor.
Data were reported descriptively as proportions, medians, and ranges where appropriate. Association between predictors of approval and both ODAC recommendation and subsequent FDA approval were assessed using univariable logistic regression analysis and reported as odds ratios (OR) and their respective 95% CIs. To avoid overfitting in this small data set, multivariable analysis was not conducted. Data analyses were conducted using SPSS version 21 (IBM Corp). All statistical tests were 2-sided, and statistical significance was defined as P < .05. The Benjamini-Hochberg method was used to adjust for multiple hypothesis testing, ensuring the false discovery rate was below 0.05.
A total of 165 meetings held by ODAC advisory committee were analyzed. Of these, 139 (84%) were product-specific meetings, and 93 meetings (67%) included voting questions. After excluding meetings discussing devices or diagnostic tests (n = 4), noncancer drugs (n = 5), and meetings discussing general questions (n = 2), transcripts from 82 meetings were available for analysis. In these meetings, ODAC discussed and voted on 60 oncology products. Details of products are provided in Table 1. Products were sponsored by 43 pharmaceutical companies, and in 34 cases (79%) the sponsor was a US-based pharmaceutical company.
During the time period analyzed, ODAC members voted on 68 applications in 79 meetings (the remaining 3 meetings included voting questions regarding postmarketing safety or trial design). Characteristics of included meetings are shown in eTable 1 in the Supplement and are summarized in Table 2. Citations for trials supporting application are shown in eTable 2 in the Supplement.
The committee voted favoring the drug in 41 cases (50%) and against the drug in 41 cases (50%). There was a very high level of agreement between ODAC committee recommendations and subsequent FDA approval; FDA approval was received for all 41 drugs for which ODAC recommended approval. Additionally, the FDA approved 7 out of 41 agents which were not recommended for approval by ODAC (κ = 0.83) (eTable 1 in the Supplement).
Of interest, a total of 28 meetings (35%) discussed applications that were not supported by phase 3 data. Among these, ODAC recommended approval in 20 cases (71%) that were all subsequently approved by the FDA. Generally, these approvals were in hematological malignancies (n = 17 [85%]), especially acute leukemia (n = 8 [40%]). Overall, 18 applications (90%) were given orphan designation,12 41 applications (60%) were submitted for regular approval, and 27 applications (40%) for accelerated approval. Applications for accelerated approval were also in hematological malignancies predominantly (n = 20 [74%]), especially acute leukemia (n = 8 [30%]). Twenty-six applications (96%) were for later lines of therapy in advanced disease, while 1 was for neoadjuvant therapy.
Predictors of ODAC recommendation and subsequent FDA approval are shown in Table 3. Use of nonrandomized phase 2 studies to support registration was associated with higher odds of ODAC recommendation and FDA approval. There was also a significant positive association between the number of trials discussed at ODAC meetings and both ODAC recommendations and FDA approval.
Reasons of individual ODAC members for voting in favor of an application were documented in 45 meetings (55%). Among these, no members voted in favor of recommendation in 10 applications (22%), and no member voted against recommendation in 9 applications (20%). When ODAC committee members voted in favor of an application (n = 35), main reasons supporting voting included efficacy alone (n = 20 [57%]), safety alone (n = 2 [6%]), both efficacy and safety (n = 3 [9%]), efficacy and other elements (n = 6 [17%]), and other elements only (n = 4 [9%]). When ODAC committee members voted against recommendation (n = 36), reasons supporting voting included efficacy alone (n = 13 [36%]), safety alone (n = 7 [19%]), efficacy and safety (n = 5 [14%]), efficacy and other elements (n = 4 [11%]), and other elements alone (n = 7 [19%]).
A conflict of interest statement was read prior to 79 of the 82 meetings analyzed, and no details about conflicts of interest were available for 3 meetings. At 47 meetings (59%), at least 1 conflict was declared for at least 1 committee member voting. In 32 meetings (41%), no conflicts were reported for all committee members.
Of the 1017 voting members, 167 (16%) disclosed one or more FCOIs, and 1 member declared a nonfinancial conflict of interest (not remunerated). Among the 167 voting members with FCOIs, transcripts contained details regarding FCOIs for 109 members (65%) who voted in 36 meetings (77%). Presence of conflicts without specific details in the transcript was only observed in meetings prior to 2002. Including members with more than one FCOI, a total of 194 FCOIs were declared. The most common conflict type for both advisory committee members and voting consultants was stock ownership and other investments (n = 91 [47%]) followed by consultancy (n = 71 [36%]) (eTable 3 in the Supplement). Nineteen of 109 voting members with disclosure details (17%) reported major FCOIs, while 99 of 109 (91%) disclosed at least 1 minor FCOI. Eighteen members (17%) had FCOIs with the pharmaceutical company sponsoring the discussed drug, 78 (71%) had FCOIs with 1 or more competitor companies, and 13 members (12%) disclosed FCOIs with both the sponsor and 1 or more competing firms. There has been a significant reduction in the proportion of voting members with FCOIs over time (31 of 77 [40%] in 2000 vs 0 of 20 [0%] in 2014, trend P < .001) (Figure). There was an apparent association between declaration of FCOIs for at least 1 voting member and ODAC recommendation for approval (P = .04). However, this was determined to be a false discovery. There was no association between FCOIs for at least 1 voting member and final FDA approval (P = .11).
At least 1 member was recused in 11 of 79 meetings (14%). Transcripts did not report the reason for the recusal in any case. Restrictions on the activities of advisory committee members or voting consultants (allowed to participate in discussion but not to vote) were observed in 6 meetings (8%). The transcript explained the reason for restriction in 2 cases (stock ownership of competitor >$100 000 and an employer's contract with a competitor <$100 000 a year).
No significant association between ODAC recommendations and FDA approval was observed for members with FCOIs with the sponsor (OR, 1.79; 95% CI, 0.97-1.46; P = .19 and OR, 3.48; 95% CI, 0.84-14.35; P = .09, respectively) compared with members with FCOIs with competitors (OR, 1.06; 95% CI, 0.78-1.44; P = .72 and OR, 0.94; 95% CI, 0.69-1.28; P = .69, respectively) (Table 3).
The approval of new cancer drugs by the FDA is based, usually, on the demonstration of efficacy with acceptable safety in adequate and well-controlled studies.13,14 However, predictors of recommendations for approval by advisory committees and of subsequent approval are less clear.
In the current analysis, we have observed that the main predictor of a favorable vote from ODAC members and of final FDA approval was the number of trials supporting the application. This result is consistent with the FDA’s published recommendation for at least 2 adequate and well-controlled studies, each convincing on its own, being required to establish efficacy.15,16 Our data show that a large proportion of applications supported by nonrandomized trials were approved, many under the FDA Accelerated Approval Program. This likely reflects flexibility in the requirement for high methodologic quality relating to orphan disease sites. The FDA Accelerated Approval Program17,18 allows approval of drugs for patients who have serious or life-threatening diseases based on surrogate end points that are reasonably likely to predict clinical benefit such as survival or symptom benefit. The applicant is then required to conduct postapproval trials to confirm clinical benefit, referred to as phase 4 commitments.
Many of the drugs approved based on single-arm trials were studied in less common hematologic malignancies. ODAC members and the FDA have recognized that single-arm trials may be appropriate support for approval for rare cancers and diseases where there is no available therapy.19-21 However, single-arm studies have limitations as they only evaluate monotherapy and cannot identify the true effect of the new drug because there is no adequately controlled comparative arm.
The FDA has been criticized for not exerting sufficient oversight on researchers’ financial ties.22-25 The FDA has expressed difficulties in finding expert advisory committee members with no FCOIs26 and did promote stringent guidelines on the disclosure of conflicts of interest26-28 in an attempt to adequately protect the drug approval process from bias. Our study also analyzed the effect of FCOIs in FDA oncology meetings. Despite studies in oncologic trials and guideline development reporting an increasing proportion of authors with FCOIs4,29 we observed that over time the frequency of FCOIs among ODAC members decreased significantly. There were no significant associations between the number of committee members with FCOIs and both ODAC recommendation and subsequent FDA approval.
In 2002, the FDA published a new guideline on the disclosure of FCOIs,27 and subsequent to this, the number of declared FCOIs decreased. Since 2005, the number of FCOIs increased again, and in 2008, with the issue of the FDA Amendments Act,30 declarations of FCOIs substantially decreased. In our review, conflict rates were lower than those reported previously for the time period between 2001 and 2004.5,31 It is not clear why our results are discordant with prior data, but this observation may be explained by our assessment being more contemporaneous than prior data and being focused on oncologic products alone.
Strengths of our study are the inclusion of drugs that were both approved and rejected. To our knowledge, information about failed submissions is only publicly available by the FDA in those cases discussed by the ODAC members. Therefore, greater transparency in the approval process is warranted. Potential limitations of our study include the potential for poor external validity because the ODAC is asked for their opinion particularly on complex cases (eg, controversial design, suboptimal study end points, a concerning safety profile, or limited efficacy), so our data may not be generalizable to the majority of regular submissions. Other limitations include the variable source documents used to extract data, as well as the small sample size of included meetings. As such, results should be interpreted with caution. Additionally, data on individual voting decisions and FCOI declarations for members were available for only a proportion of meetings and led to some uncertainty in our estimates. Finally, our FCOIs analysis was based on self-reporting as captured by the conflict of interest statement read at the beginning of every meeting. A better method to evaluate FCOIs would be extracting data from the FDA 3410 form10 itself, but this information is not available publically in the majority of cases. Furthermore, self-reporting of FCOIs is subject to unconscious bias and may underestimate the presence of financial relationships.32
ODAC members and the FDA were more likely to approve drugs when more than 1 trial supporting the application is available. Methodological quality such as the use of randomization appears less important with approvals occurring for drugs studied in single-arm trials in diseases or those for which there is no available therapy. Declarations of FCOIs among voting members of ODAC were frequent but have decreased significantly over time. There was no apparent influence of FCOIs on ODAC recommendation and subsequent regulatory approval.
Corresponding Author: Eitan Amir, MD, PhD, Division of Medical Oncology, Princess Margaret Cancer Centre, 610 University Ave 5-124, Toronto ON M5G 2M9, Canada (email@example.com).
Accepted for Publication: December 21, 2015.
Published Online: March 3, 2016. doi:10.1001/jamaoncol.2015.6479.
Author Contributions: Drs Amir and Tibau had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Tibau, Ocana, Amir.
Acquisition, analysis, or interpretation of data: Tibau, Anguera, Seruga, Templeton, Barnadas, Amir.
Drafting of the manuscript: Tibau, Amir.
Critical revision of the manuscript for important intellectual content: Tibau, Ocana, Anguera, Seruga, Templeton, Barnadas, Amir.
Statistical analysis: Amir.
Administrative, technical, or material support: Tibau, Anguera, Templeton, Amir.
Study supervision: Tibau, Ocana, Anguera, Amir.
Conflict of Interest Disclosures: None reported.
Disclaimer: All authors confirm that all parts of this manuscript are original and no element has been reproduced from other sources.