How Vital Are Patient-Reported Outcomes?

In this issue of the Journal, Efficace et al. (1) thoughtfully explore whether the blinding of patients in randomized-controlled clinical oncology trials influenced patient-reported outcome (PRO) results between treatment arms. They report that the experimental treatment was not more frequently favored over the standard treatment in open-label trials than in blinded trials. Although a noninferiority analysis would have been preferable, this important study strengthens the validity of PRO results even from open-label cancer randomized clinical trials. The implications of this study are noteworthy because of 2 key trends in clinical oncology trial design. First, PROs are becoming more common in cancer clinical trial design as primary or secondary endpoints. Second, as this study reports, nearly 70% of the trials reviewed were categorized as open label. This analysis is important as it will help guide the interpretation of future trials incorporating PROs, as well as support the design of novel trials. It further supports the finding that PROs serve as objective and vital measures in clinical oncology trials.

The clinical relevance of this issue, however, is even greater as PROs become more integrated into standard oncology practice. We’ve known for some time now that PROs serve as independent predictors for cancer outcomes, often superseding classic prognosticators for long-term survival (2). Other studies have demonstrated that validated PROs can be more sensitive than provider-based toxicity scores and can help illuminate the subtle benefits of advanced technologies (3). A recent landmark study has taken these findings even further and randomly compared the integration of real-time PROs vs routine care of patients with metastatic cancer (4). This pivotal study, confirmed in yet another randomized trial (5), found that the integration of real-time PROs was associated not only with improved quality of life (and fewer emergency department visits) but also with increased survival. PROs have come full circle from initially predicting survival to now actually enhancing survival itself. Imagine how expensive it would be to clinically utilize a personalized molecular agent that not only serves as a prognostic marker but also clinically significantly improves survival? Now we have such a vital and cost-effective strategy with PROs, which represent the whole “person” in this new era of personalized medicine (6).

The logical next step would then be to fully integrate PROs into the real-world setting. Recently, at the Henry Ford Cancer Institute (HFCI), we have integrated real-time PRO monitoring throughout our entire clinical cancer enterprise, which spans 5 hospitals and multiple outpatient centers. This process begins at cancer diagnosis and is measured throughout each patient’s care journey during treatment to survivorship. Our real-time monitoring uses the validated Patient-Reported Outcomes Measurement Information System developed by the National Institutes of Health (7), including the option of the streamlined computer-assisted technology. The PRO results are immediately available in the electronic medical record for provider review.

How vital are PROs? We believe that PROs are so essential that they should be viewed over time as a novel type of vital sign. In this regard, at HFCI, severe PRO scores are transmitted in real-time to our Oncostat clinic, an urgent cancer care clinic present at various HFCI locations. Just like they react to high or low blood pressure findings, the Oncostat team now responds in real-time to high or low PRO results to contact the patient and make appropriate referrals to the cancer pain team, psycho-oncology, palliative care, or other supportive oncology services as needed.

The implementation of real-time PROs into our standard of care (and our Oncostat model) has raised many important questions. At a fundamental level, what are the clinical thresholds or combinations of PRO score levels (or changes in scores) that should trigger the need for clinical intervention? As we have embedded PROs into the surgical, medical, and radiation oncology clinics, we also recognize the importance of customizing the frequency of PROs assigned to patients based on their scores and other clinical factors to help reduce survey burden.

There is also a strong need to better understand the interplay of cancer stage, treatment, PRO scores, overall health, and functional status. Importantly, we need to carefully study how socioeconomic status, race, ethnicity, computer literacy, and/or social determinants of health may influence PRO scores. Indeed, we believe that PROs can be leveraged to help promote health equity among underserved populations by providing an unfiltered voice for the patients and facilitating improved communication that may help address unconscious biases. Studies are currently exploring intricate relationships between PROs, molecular markers, and/or radiomic findings for patients (8). In a recent publication, Morin et al. (9) deployed artificial intelligence to continually update prognostic estimates as the patient’s cancer care journey unfolded. We recommend the integration of PROs into such artificial intelligence models to further refine the precision medicine treatment strategy for cancer patients. As we further develop these opportunities, how do we optimally educate and partner with our primary care, palliative care, pain team, nursing colleagues, and beyond to continue to improve patient care, quality of care, and the overall patient experience?

Clearly, more studies and real-world data are needed. As we embark together on this exciting PRO journey into the cancer clinic, one thing is vitally clear—our patients want us to PROceed (10)!

Funding

None.

Notes

Role of the funder: Not applicable.

Disclosures: BM discloses having received research funding to his department from Varian, ViewRay, and Philips. SC has no conflicts of interest to report.

Author contributions: Writing- original draft: SSC, BM. Writing- reviewing and editing: SSC, BM.

Data Availability

Not applicable.