PRO label claims in the US and the EU for mCRPC products
This review provides a critical evaluation and comparison of PRO claims approved by the FDA and the EMA for five recently approved products for mCRPC. Some concordance was seen between the FDA and the EMA for pain claims granted to enzalutamide and abiraterone based on the BPI-SF worst pain item. Baseline BPI-SF worst pain item data were included in both the US labels and EU SmPCs for enzalutamide and abiraterone. This concordance in the acceptance of pain claims may be explained in part by the fact that patient-assessed core symptoms of a disease such as pain are well-accepted primary and secondary efficacy endpoints in registration trials, according to the EMA reflection paper on HRQOL measures [[
14]]. Similarly, the FDA PRO guidance highlights the assessment of pain intensity using a single-item measure as an obvious way to measure the impact of treatment on pain [[
10]]. Furthermore, the FDA guidance for industry on cancer clinical trial design cites symptoms as a direct efficacy endpoint that can be used to support product approval [[
33]].
However, even for the generally accepted concept of pain, there was still some discordance between the FDA and the EMA in the claim language allowed for mCRPC products reviewed. The primary difference in claim language between the US label and the EU SmPC for abiraterone was that the US label focused on the opiate use change, which was supported by the delay in pain progression, whereas the EU SmPC mentioned the time to pain progression more directly. The EMA also granted additional claim language in the SmPC for abiraterone, including “time to pain degradation,” “proportion of patients with pain palliation,” and “proportion of patients with pain progression” based on the BPI-SF worst pain item. No further claims were granted in the US label for this product. This discordance may be explained in part by the FDA’s comments within the abiraterone DAP: “Other secondary endpoints measured but not listed and not evaluated in the review included proportion of patients experiencing pain palliation using BPI-SF and analgesic score and time to pain progression…Reasons for not including them in the review are as follows: only a portion (<50%) of patients had data for the endpoints not listed as key secondary endpoints; measuring the endpoints was less objective, and their regulatory acceptability had not been evaluated by the Agency in terms of reliability, validity, ability to detect change, and interpretability in the study patient population; no pre-specified plan for multiple comparisons adjustment; changes in these endpoints do not constitute a basis for marketing approval or disapproval of abiraterone acetate for the proposed indication” [[
11]].
Notably, a recent communication prepared by the FDA highlighted the following challenges in pain palliation measurement in cancer clinical trials [[
34]]:
• Pain intensity and analgesic use assessment tools must be demonstrated to be reliable, valid, and sensitive to changes over time, consistent with FDA PRO guidance criteria.
• Enrollment eligibility criteria should ensure that patients are experiencing pain that is attributable to cancer at baseline.
• The mode of data collection (e.g., paper, electronic, internet, interactive voice response system, or interviewer-administered) must demonstrate measurement properties in keeping with the PRO guidance principles.
• Optimal timing for pain and analgesic assessment is over several consecutive days (e.g., daily over the 7-day period prior to a scheduled study visit).
• A pain palliation responder should be defined using both pain and analgesic use criteria, incorporating an analysis of tumor response that will support evidence of pain palliation response.
• There is a risk of missing data and inadvertent unblinding
Basch and colleagues stress the critical importance of tracking analgesic use with a content-valid analgesic log to ensure that pain palliation observed is truly the result of the treatment being studied rather than the result of an increase in analgesic use [[
34]]. Furthermore, the analgesic log should be administered using the same schedule and recall period (e.g., past 24 hours) as the pain intensity assessment during clinical studies evaluating treatment. FDA feedback on the abiraterone DAP and Basch and colleagues’ [[
34]] publication regarding assessment of pain palliation in cancer clinical trials highlight the many factors considered by the FDA when considering pain claims for a new cancer product.
Although the EMA granted additional pain claims to abiraterone based on the BPI-SF average pain item and to cabazitaxel based on the PPI, similar claims for these products were not granted by the FDA for these PRO measures. Based on publicly available information provided in the DAP for cabazitaxel, the following comments were made by FDA reviewers during the end-of-phase-2 meeting (EOP2) in response to questions submitted by the sponsor regarding the PPI and pain assessment: “We recommend that you submit the final version of the PPI and the AS [analgesic score] in the exact format it is administered in your protocol with instructions on how the instrument will be administered, directions explaining how scores will be derived, and how the statistical analyses will be applied; open-label data are only appropriate for labeling if results are convincing and conclusive; pain intensity should be assessed at screening, and then continued eligibility by pain score should be verified at baseline (i.e. before randomization/dosing); pain intensity should then be recorded daily, over the duration of the trial. There should also be evidence of efficacy over the entire duration of treatment; assessment of the ‘worst pain’ will provide more reliable results than ‘average pain’ over 24 hours.” However, it should be noted that the EOP2 meeting for cabazitaxel took place in 2006 and does not appear to represent the most current thinking by the FDA regarding pain assessment. As discussed earlier by Basch and colleagues, daily collection of pain and analgesic assessment for 7 consecutive days prior to an office visit is recommended over continuous daily collection of pain [[
34]]. This recommendation is likely due to higher patient burden and higher noncompliance with continuous daily collection over the length of the study period. Furthermore, the DAP for cabazitaxel indicated that the most frequent and most severe protocol violations for the phase 3 registration study were for missing pain assessments or analgesic scores, with less than 50% of each treatment group reporting analyzable data. Although similar explanations from FDA reviewers were not available in the abiraterone DAP, the FDA’s stated preference for the evaluation of “worst pain” over the past 24 hours within the cabazitaxel DAP may also partly explain why the claim was not granted for abiraterone based on the BPI-SF “average pain” item.
The largest area of discord identified between the EMA and the FDA was for HRQOL claims. HRQOL claims were granted by the EMA for both abiraterone (“time to degradation in FACT-P total score”) and radium Ra 223 dichloride (“Relative to placebo, the decline in quality of life was slower for Ra-223 during the on-treatment period as measured by EQ-5D utility index score [−0.040 versus −0.109; p = 0.001], EQ-5D self-reported VAS [−2.661 versus −5.860; p = 0.018], and the FACT-P total score [−3.880 versus −7.651, p = 0.006] but did not reach published minimally important differences. There is limited evidence that the delay in loss of HRQOL extends beyond the treatment period”). Similar HRQOL claims for these products were not granted by the FDA.
Comments by FDA reviewers found within the radium Ra 223 dichloride DAP may provide some insight into why a HRQOL claim was not granted by the FDA for this product. Specifically, the reviewer comments on the PRO results from the key pivotal trial for radium Ra 223 dichloride stated that “FACT-P and EQ-5D total scores showed a slight improvement for patients receiving Ra-223 when compared to placebo. When evaluated by visit, the statistical significance in the difference between groups decreased over time. Whether this was due to the loss of anti-cancer effect of Ra-223 with time or was due to decreased data completion rates is not known, although both are likely contributors. The FDA SEALD cites limitations specific to the current Ra-223 application including small number of assessments (maximum of 4 for FACT-P), low rate of completion at week 24+ as well as a small observed magnitude of treatment effect. Despite the limitations in choice of instrument, frequency of assessments and completeness of data, the quality-of-life results are supportive of the overall application in that, on average, the available data appear to trend toward an improvement in the Ra-223 arm and do not show a detriment to quality of life measures or pain in patients treated with Ra-223 when compared to placebo.” Similar comments from FDA reviewers were not available within the abiraterone DAP to further explain why HRQOL claims were not granted for this product.
As demonstrated by the SmPC language for both cabazitaxel and radium Ra 223 dichloride, the EMA provided more comprehensive coverage of both the positive and negative PRO outcomes from the registration studies for these products, whereas negative PRO outcomes were discussed within the DAP but did not translate into label language in the US.
The results of our research on PRO label claims for mCRPC products recently approved in the US and the EU further support the conclusions from several earlier studies with similar objectives [[
10],[
35]-[
37]]. Demuro and colleagues’ comparison of PRO label claims granted from 2006 to 2010 for new drug entities or biologic licensed agents by the FDA and the EMA indicated that the EMA is more likely than the FDA to grant PRO claims and typically does so for higher-order constructs (e.g., HRQOL), whereas PRO claims in the US are most often limited to those based on symptom improvement [[
19]]. Similar findings also resulted from earlier research by Coombs and colleagues when PRO claims in the US and the EU were compared for oncology products [[
37]].
Evaluation of instrument properties for selected PRO measures against FDA guidance criteria
Of the measures evaluated, the BPI-SF worst pain item has the strongest measurement properties in patients with mCRPC. Most importantly, content validity has been established in this population, a key criterion necessary for FDA acceptance in supporting labeling claims. Furthermore, although evidence of known-groups validity was not identified for this measure, test-retest reliability, construct validity, and responsiveness have all been established in patients with mCRPC [[
27],[
34]]. Other key strengths of this measure include the brief (24-hour) recall period; the single-item assessment (low burden on patients to complete); the 11-point numeric rating scale, which is well-accepted by patients, clinicians, and the FDA; and information available to assist interpretation of clinical trial results in patients with mCRPC with bone metastases [[
31]]. Based on this evidence and the ability to achieve successful labeling claims of “delay in pain progression” for enzalutamide and abiraterone in both the US and the EU, as well as additional claims in the EU for “time to pain degradation,” “proportion of patients with pain palliation,” and “proportion of patients with pain progression,” the BPI-SF worst pain item is recommended for future evaluation of pain in trials evaluating treatment for patients with mCRPC and for pursuit of similar labeling claims in both the US and the EU.
The identified evidence in support of the BPI-SF average pain item is weaker when compared with FDA PRO guidance criteria. Without documented evidence of content validity in patients with mCRPC, the FDA is unlikely to grant PRO claims based on this single item. However, the BPI-SF average pain item is recommended for publications supporting future FDA-approved products for mCRPC and may achieve EMA-approved SmPC claims if the data are convincing.
Of the three pain measures evaluated, the PPI component of the MPQ has the least evidence to support its future use in clinical trials evaluating mCRPC treatments. When considering evaluation of pain in patients with mCRPC and if the desire is to achieve a product label claim in the US or the EU, the PPI is not recommended.
The FACT-P was the only HRQOL measure evaluated against FDA PRO guidance criteria. Key strengths for the FACT-P are established reliability (test-retest and internal consistency), construct validity, known-groups validity, responsiveness, and established definition for clinically meaningful changes to assist with interpretation in clinical trials evaluating patients with mCRPC [[
32]]. In particular, the FACT-P was responsive in both phase 3 registration studies for abiraterone as well as phase 3 studies for enzalutamide and docetaxel for patients with mCRPC but was less responsive in the registration study for radium Ra 223 dichloride described in the SmPC [[
7],[
28],[
29],[
34]]. Based on further examination of the two EMA-approved SmPCs that included FACT-P claims, there was clear positive claim language favoring abiraterone in the SmPC for “risk of FACT-P total score degradation” and “time to degradation in FACT-P total score,” whereas the analysis of change from baseline in registration studies for radium Ra 223 dichloride resulted in an SmPC claim that scores on this measure did not reach the published minimum important difference. Given the limitation on content validity for the FACT-P in patients with mCRPC, this measure is not recommended to support US labeling claims for future products. Because the FACT-P is validated and clearly very responsive in studies evaluating treatment for mCRPC, the FACT-P should still be considered for the evaluation of PC-specific HRQOL to provide data for US publication and support of SmPC claims in the EU.
More recently, the National Comprehensive Cancer Network FACT Prostate Symptom Index-17 (NCCN-FACT FPSI-17), a new prostate symptom index, was developed based on qualitative input from patients with advanced (stage 3 and 4) castration-resistant PC, which can be used to examine the effectiveness of noncurative treatments in advanced PC [[
38]]. Although the NCCN-FACT FPSI-17 appears to have initial evidence of content validity, reliability, and construct validity in patients with mCRPC, publications documenting its responsiveness in studies evaluating treatment for patients with mCRPC are currently not available [[
38]]. Once additional studies are completed to further validate this measure and establish responsiveness, it may potentially be considered for inclusion in future studies for treatment of mCRPC designed to obtain US labeling claims focused on PC symptom improvement, provided all FDA PRO guidance criteria beyond content validity are also sufficiently met in this population.
Notably, the concept of fatigue, a prevalent disease- and treatment-related symptom of patients with mCRPC, was measured using a fatigue-specific PRO measure (Brief Fatigue Inventory) in registration trials for both abiraterone and enzalutamide based on information provided in each product’s DAP and EPAR [[
4]]. However, neither the FDA nor the EMA granted a fatigue claim. This finding is not surprising given the recent views on fatigue assessment expressed by an FDA representative that fatigue is a multidomain concept not measurable with a single item; it is believed that patients do not use the term “fatigue”; problems with instrument content validity have not allowed conclusion of benefit; and, finally, a clear link between fatigue and the disease or treatment has not been found [[
39]]. To address these concerns, a consortium research project entitled Patient-Reported Outcomes of Fatigue–Cancer has been established “to define cancer related fatigue and determine how it should be measured from a patient perspective” [[
40]]. The Cancer Fatigue–Symptom Severity Assessment, a new multidimensional fatigue measure, has been developed out of this consortium, is currently undergoing psychometric validation, and is planned for drug development tool qualification by the FDA [[
40]]. This measure may eventually be accepted by the FDA and the EMA to support fatigue-specific PRO labeling claims for new cancer products.
There are some notable limitations to our research. First, although this study included the majority of recently approved products for mCRPC, not all mCRPC products were included [[
10]]. Second, any PRO measures included in clinical trials for mCRPC products reviewed and mentioned in the DAP or EPAR for a product but not resulting in US label or EU SmPC claims were excluded from this review. Third, our review of regulatory feedback during the approval process for each of the products was limited to the information that was made publicly available on the FDA and EMA websites. The drug manufacturers may possess additional proprietary information that was not included. Finally, the nature of our literature searches to identify the desired information supporting each PRO measure of interest was targeted rather than systematic. Thus, it is possible that other information exists either to support or refute the information presented in this paper.