Many patients enter cancer clinical trials with baseline symptoms.1 Notably, the current clinician reporting mechanism for symptomatic adverse events (AEs) via the Common Terminology Criteria for Adverse Events (CTCAE)2 does not formally distinguish between symptoms present at baseline vs those that develop during a trial. Therefore, AE estimation in clinical trials may include symptoms that predate trial entry. This raises concern that the cumulative incidence of patient-reported AEs may be high, particularly if preexisting symptoms related to other causes (eg, comorbidities, prior treatment) are attributed to study drugs.
Just as patients are better positioned to detect symptomatic AEs during a trial,3-5 we hypothesized that they are also better at reporting baseline symptoms. As such, we anticipated that patient reporting would facilitate better understanding of pretrial symptoms. Moreover, if baseline symptom information were available, we might be able to adjust AE analyses to remove preexisting symptoms from tabulations, thereby enabling a focus on symptomatic AEs that are incident during the clinical trial.
To test these suppositions, we completed a retrospective analysis (NCCTG N0591/Alliance) of legacy clinical trials from the National Cancer Institute (NCI)-supported National Clinical Trials Network group the Alliance for Clinical Trials in Oncology that included clinician reporting of AEs (via CTCAE) and patient reporting of analogous symptoms (via patient-reported outcome [PRO] questionnaires) at baseline and throughout the clinical trial. Because this was a retrospective analysis, institutional review board approval was not obtained for this specific study, and all data collection for the included trials was approved by their respective institutional review boards, with informed consent provided by enrolled patients.
For clinician-reported CTCAE symptoms, the incidence of baseline grade of 1 or greater was compared using McNemar’s test6 with the maximum CTCAE grade postbaseline (ie, none vs grade ≥1), as well as with worsened CTCAE grade postbaseline (ie, no worsening vs worsening by ≥1 grade). For patient-reported measures, the incidence of baseline score of greater than 1 was compared with maximum score postbaseline (ie, score of 0 vs score ≥1), as well as with worsened score postbaseline (ie, no worsening vs worsening by ≥1 point for 0-4/0-5 scales or ≥2 points for 0-10 scales). All P values were 2-sided and considered statistically significant if P < .05, with no adjustments made for multiple comparisons.
Twenty-six clinical trials (1996-2015) were identified (n = 2608; mean [SD] age, 60.1 [12.2] years; 1611 [61.8%] women; 2492 [93.1%] white), with 24 distinct AEs captured via 23 PRO questionnaires, including the Brief Fatigue Inventory, Functional Assessment of Anorexia/Cachexia Therapy, Functional Assessment of Cancer Therapy-General, Linear Analogue Self Assessment, Lung Cancer Symptom Scale, Pittsburgh Sleep Quality Index, Profile of Mood States, SF-36, Symptom Distress Scale, Symptom Experience Diary, and UNISCALE. At baseline, symptom reporting was significantly more prevalent for patients vs clinicians (20/24, 83%) (Table).
When examining worst postbaseline symptom rates without using baseline adjustment, 21 of 24 (88%) were significantly different between patients and clinicians, with 16 of 21 (76%) having a higher rate by patient reporting. In contrast, the Figure displays differences between patient and clinician symptom reporting rates after adjusting for baseline symptoms. Across clinical trials, 16 of 24 (67%) symptoms were significantly different between patients and clinicians; but only 9 of 16 (56%) had a significantly higher rate via patient report.
At time of clinical trial enrollment, patients may have elevated symptoms that are unrelated and potentially misattributed to investigational treatments if not detected and adjusted for in analyses. This study found that clinical investigators detect fewer baseline symptoms compared with patients. Moreover, we tested a method to adjust for baseline symptoms when tabulating incident symptoms during clinical trials, to avoid misattribution. We found that without using this method, patients report more symptoms than clinicians about three-quarters of the time, whereas with this method this is reduced to about half. The lack of sample ethnic/racial diversity may limit result generalizability. Nevertheless, use of the baseline adjustment method can provide clinicians, patients, pharmaceutical sponsors, and drug trialists with greater confidence about the meaningfulness of symptom and tolerability data in clinical trials, particularly for patient-reported data.
Corresponding Author: Thomas M. Atkinson, PhD, Department of Psychiatry & Behavioral Sciences, Memorial Sloan Kettering Cancer Center, 641 Lexington Ave, 7th Flr, New York, NY 10022 (atkinsot@mskcc.org).
Accepted for Publication: October 15, 2019.
Published Online: December 26, 2019. doi:10.1001/jamaoncol.2019.5566
Author Contributions: Dr Dueck had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Atkinson, Dueck, Lafky, Sloan, Basch.
Acquisition, analysis, or interpretation of data: Atkinson, Dueck, Satele, Thanarajasingam, Sloan, Basch.
Drafting of the manuscript: Atkinson, Dueck.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Atkinson, Dueck, Satele, Sloan.
Administrative, technical, or material support: Lafky.
Conflict of Interest Disclosures: Dr Dueck reported grants from National Cancer Institute during the conduct of the study. Dr Thanarajasingam reported grants from National Institutes of Health during the conduct of the study. Dr Lafky reported grants from National Institutes of Health during the conduct of the study. Dr Basch reported grants from National Cancer Institute, grants from Patient-Reported Outcomes Research Institute, personal fees from Memorial Sloan Kettering Cancer Center, personal fees from Dana-Farber Cancer Institute, personal fees from Research Triangle Institute/CMS, personal fees from CareVive Systems, personal fees from Sivan Healthcare, and personal fees from Self Care Catalysts outside the submitted work. No other disclosures were reported.
Funding/Support: Research reported in this publication was supported by the Cancer Moonshot (US National Cancer Institute U01 CA233046), as well as the National Cancer Institute of the National Institutes of Health under award numbers U10CA180821, U10CA180882, and UG1CA189823 (to the Alliance for Clinical Trials in Oncology NCORP Research Base; Jan C. Buckner, MD, contact PI), U10CA180790, U10CA180838, and P30CA008748, which partially supports the Patient-Reported Outcomes, Community-Engagement, and Language Core Facility used to complete this investigation.
Role of the Funder/Sponsor: The funding agencies had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Dislaimer: The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health (https://acknowledgments.alliancefound.org).