National Comprehensive Cancer Network Guideline Recommendations of Cancer Drugs With Accelerated Approval

This cross-sectional study evaluates the basis for the supporting evidence and treatment preference ratings for cancer drugs and indications recommended by the National Comprehensive Cancer Network (NCCN) guidelines committees.


Introduction
In the US, many new cancer drugs are approved via the accelerated approval pathway of the Food and Drug Administration (FDA). 1,2Under this pathway, certain drugs for serious conditions can be approved based on clinical trials that show improvements in surrogate measures, which are only reasonably likely to predict a clinical benefit. 2,3After the drugs are approved, the manufacturers are required to complete confirmatory trials to verify clinical benefit. 4However, drugs in the accelerated approval program have experienced substantial delays in drug manufacturers' completion of confirmatory studies, 2,5 have been priced at high levels despite uncertain clinical benefit, 6 and have been slow to be withdrawn when confirmatory trials fail to demonstrate a clinical benefit. 4,7,8ese regulatory challenges have been particularly pronounced among cancer drugs.For example, checkpoint inhibitors are now FDA-approved to treat more than 85 oncological indications, 9 including dozens of accelerated approvals based on surrogate measures, such as response rate and progression-free survival. 10However, their efficacy and the strength of the correlation between surrogate measures and clinical end points 11,12 vary substantially depending on cancer histological characteristics and subtypes, 9,10,13 as acknowledged by the FDA 10 and its Oncologic Drugs Advisory Committee. 14though the accelerated approval status is included on drug labels, it is not clear whether this information factors in the treatment decisions of patients and oncologists.6][17] The NCCN guidelines combine evidence and expert opinion and are used by both public and private insurers to determine cancer drug coverage. 18[22] The NCCN guidelines frequently make recommendations on medications that extend beyond FDA-labeled indications; more than one-third of guideline recommendations relate to non-FDAapproved (off-label) uses. 18For some cancer drugs with accelerated approval, the NCCN guidelines continue to recommend their use even after confirmatory studies fail to demonstrate a clinical benefit or their indications are withdrawn by manufacturers. 7,23 is important to understand whether FDA regulatory status plays a role in the decisions and ratings by professional society guidelines, which in turn inform patients, oncologists, and payers.
Given the centrality of the NCCN guidelines in cancer treatment and coverage determination, we undertook a cross-sectional study to analyze the NCCN guidelines' assessments for cancer drug indications that received FDA accelerated approval compared with cancer drug indications that received FDA regular approval.

Methods
In accordance with the Common Rule, this cross-sectional study was exempt from ethics review and informed consent because it was not considered human participant research.We followed the

Cohort
We identified cancer drugs that were granted accelerated approval from the program's inception in 1992 through June 30, 2022, using a public FDA list. 24For each drug, we reviewed the FDA approved labeling as of October 19, 2022, to identify all indications, including indications that were granted regular approval and accelerated approval.Indications with no NCCN guidelines (eg, naxitamab for neuroblastoma) and those that were tumor-type agnostic (eg, dabrafenib plus trametinib for solid tumors with BRAF V600E alteration) were excluded.We also excluded indications for gemtuzumab ozogamicin because the drug was withdrawn then reapproved with a different dosing strategy and a different indication.All analyses were performed at the drug-indication level.

Exposure and Outcome
The exposure variable for this study was FDA regulatory status as of October 19, 2022.Exposure included indications that were initially granted regular approval, accelerated approval indications that were converted to regular approval after confirmatory studies, accelerated approval indications that were withdrawn by the manufacturer (typically after failed confirmatory studies), and indications that remained under accelerated approval. 7e relevant NCCN Clinical Practice Guidelines in Oncology 25 for each indication were reviewed as of February 10, 2023.These guidelines are updated frequently, and all guidelines reviewed in this study had been updated as of April 4, 2022.
Each drug-indication combination was assessed for its level of evidence rating and level of treatment preference rating given by the relevant NCCN committee.These committees, typically comprising 20 to 30 subspecialty experts, meet multiple times per year to appraise and discuss the most recent evidence to reach a consensus on practical recommendations for clinicians.The committees assign ratings to treatment approaches that reflect the level of evidence supporting the treatment's use, as well as preference ratings that compare the treatment's efficacy with that of other available treatment options (Table 1). 26idence ratings were category 1 (high level of evidence with uniform consensus), category 2A (lower level of evidence with uniform consensus), category 2B (lower level of evidence without uniform consensus but with no major disagreement), and category 3 (any level of evidence but with major disagreement).We separately defined when drug indications were removed or excluded from the guidelines.Categories of treatment preference ratings included preferred, alternative preferred, other recommended, or useful in certain circumstances.
Compared with indications with current accelerated approval, those converted to regular approval had higher evidence ratings (3% vs 38% category 1; P < .001)and were similarly likely to be preferred (40% vs 47%; P = .61).Converted indications were also similarly likely to have category 1 evidence vs those with regular approval (38% vs 47%; P = .26)and similarly likely to be preferred (47% vs 58%; P = .16) Among the 21 withdrawn accelerated approval indications, 8 (38%) remained in the NCCN guidelines, 6 of which had category 2A evidence ratings (Table 3).One of these indications was listed as preferred (romidepsin for T-cell lymphoma).In all 8 cases, the guidelines were updated after the withdrawal, and these recommendations remained.

Discussion
In this cross-sectional study of 315 cancer drug indications, those with accelerated approval were less likely to receive the highest evidence rating and preferred status in the NCCN guidelines compared with those that were granted regular approval.However, most indications, including those with both accelerated and regular approvals, had uniform expert endorsement despite low-quality evidence   The NCCN guidelines often identified indications with accelerated approval as having lowerlevel evidence than indications with regular approval, a finding consistent with that of prior studies. 2,3,27Indications that were granted accelerated approval were also less likely to be rated as preferred treatments than indications with regular approval.This finding is problematic because the FDA's accelerated approval program is specifically designed to expedite approval of drugs that fulfil unmet medical needs; the guideline ratings suggest that, often, other regimens are preferred over those with accelerated approval, which raises questions about whether the basic regulatory criteria for accelerated approval were met.
Although the lower evidence standard for drugs with accelerated approval is well recognized, we also found that regular approvals were frequently supported by lower-level evidence: only 47% of indications that were initially granted regular approval and 38% of accelerated approvals that were converted to regular approval were rated as having category 1 evidence.Previous studies have demonstrated that the FDA has increasingly relied on nonrandomized trials and studies using surrogate measures of efficacy as trial end points, rather than clinical outcomes, for confirmatory studies of drugs with accelerated approval and for drugs with regular approvals. 7,10This finding is problematic due to the frequency with which drugs with promising efficacy in phase 2 studies fail to show benefit in phase 3 randomized clinical trials, 28 especially in oncology. 29tter clinical evidence is needed to guide the optimal use of many cancer drugs and ensure they offer clinically meaningful benefits for patients.The large number of drugs with category 2A evidence raises concerns about the clarity and usefulness of this rating to practicing clinicians.For example, in B-cell lymphoma, selinexor, a novel drug tested in 1 single-arm phase 2 trial of 134 patients, is labeled category 2A evidence, and ibrutinib, which has been tested in thousands of patients across multiple randomized trials, is also labeled category 2A.Among the NCCN's definitions for its thresholds for consensus, 26 only 85% of votes are required to be classified as uniform consensus, and the NCCN does not specify thresholds for its delineation between high-level and lower-level evidence.Clearer definitions or thresholds for high-level and lower-level evidence, potentially with the addition of an intermediate category of evidence, may also make the recommendations more informative.Although the NCCN committees include academic oncologists from a broad range of institutions, these levels of consensus may be undermined by the widespread financial conflicts of interest among committee members 30,31 ; thus, it is possible that, in some cases, the NCCN guidelines represent optimistic interpretations of evidence supporting novel drugs.

Figure 1 .
Figure 1.National Comprehensive Cancer Network (NCCN) Category of Evidence Ratings Stratified by US Food and Drug Administration (FDA) Approval Status

Figure 2 .
Figure 2. National Comprehensive Cancer Network (NCCN) Category of Preference Ratings Stratified by US Food and Drug Administration (FDA) Approval Status

Table 1 .
National Comprehensive Cancer Network (NCCN) Categories of Level of Evidence and Consensus and Treatment Preference RatingsWe descriptively analyzed level of evidence and treatment preference ratings for cancer indications, stratified by approval status.We used Fisher exact test, with 2-sided hypothesis tests and a significance level of P < .05established a priori, to compare evidence and preference ratings of indications with accelerated approval against those with converted accelerated and regular approvals.Fisher exact test was chosen to analyze these nominal variables due to a small sample size.
Other recommendedMay be somewhat less efficacious, more toxic, or based on less mature data; or significantly less affordable for similar outcomes Useful in certain circumstances May be used for select patient populations defined with recommendationStatistical Analysis

Table 2 .
Characteristics of Cancer Indications by Cancer Subtype and FDA Approval Status Abbreviations: AA, accelerated approval; FDA, US Food and Drug Administration; NSCLC, non-small cell lung cancer.a Percentage represents the column percentage.JAMA Network Open | Health Policy NCCN Recommendations of Cancer Drugs JAMA Network Open.2023;6(11):e2343285. doi:10.1001/jamanetworkopen.2023.43285(Reprinted) November 14, 2023 4/10 Downloaded from jamanetwork.comby guest on 11/18/2023 Downloaded from jamanetwork.comby guest on 11/18/2023 (ie, category 2A).Additionally, over one-third of accelerated approval indications that had been withdrawn by the FDA continued to be recommended in the guidelines at a level representing uniform consensus.

Table 3 .
Withdrawn Accelerated Approval Indications Remaining in the NCCN Guidelines a US Food and Drug Administration definition: PD-L1stained tumor-infiltrating immune cells of any intensity covering 1% or greater of the tumor area.NCCN definition: PD-L1-stained tumor-infiltrating immune cells covering 5% or greater of the tumor area.