Studying a Rare Disease Using Multi-Institutional Research Collaborations vs Big Data: Where Lies the Truth?

Multi-institutional collaborations provide granularity lacking in epidemiologic data sets to enable in-depth study of rare diseases. For patients with superficial, high-grade soft tissue sarcomas of the trunk and extremity, the value of radiation therapy (RT) is not clear. We aimed to use the 7-institution US Sarcoma Collaborative (USSC) and the National Cancer Database (NCDB) to investigate this issue.

All adult patients with superficial truncal and extremity high-grade soft tissue sarcomas who underwent primary curative-intent resection from 2000 to 2016 at USSC institutions or were included in the NCDB from 2004 to 2013 were analyzed. Propensity score matching was performed. End points were locoregional recurrence-free survival (LRFS), overall survival (OS), and disease-specific survival (DSS).

Of 4,153 patients in the USSC, 169 patients with superficial high-grade tumors underwent primary curative-intent resection, 38% of which received RT. On multivariable Cox-regression analysis, RT was not associated with improved LRFS (p = 0.56), OS (p = 0.31), or DSS (p = 0.20). On analysis of 51 propensity score-matched pairs, RT was still not associated with increased LRFS, OS, or DSS. Analysis of 631 propensity score-matched pairs in the NCDB demonstrated improved 5-year OS rate associated with RT (80% vs 70%; p = 0.02). The LRFS and DSS rates were not evaluable.

Granular data afforded by collaborative research enables in-depth analysis of patient outcomes. The NCDB, although powered with large numbers, cannot assess many relevant outcomes (eg recurrence, DSS, or complications). In this study, the approaches yielded conflicting results. The USSC data suggested no value of radiation and the NCDB demonstrated improved OS, contradicting all randomized controlled trials in sarcoma. The pros and cons of either approach must be considered when applying results to clinical practice, and underscore the importance of randomized controlled trials.