The center of the maze.
The Study: This pair of NCDB analyses (here and here) look at survival in patients with locally advanced (T3-4 or N+) rectal cancer who achieve a pCR to neoadjuvant chemoradiation (nCRT) and go on to either observation or adjuvant chemo (aCT). Overall, close to 70% of eligible patients received nCRT. The pCR rate was 9%, and less than 30% of those went on to get adjuvant chemotherapy. Rates of overall survival at 5 years was higher for patients who received aCT (88→ 95%), primarily driven by the subset with clinically positive nodes. Here’s where George Chang’s editorial takes issue. How should we interpret results that show a pretty significant benefit when a large body of randomized data doesn’t? Veeeeery carefully, that’s how. Unless we find out that oncologists are actually West World hosts programed to assign therapies at random (we're not saying it's impossible), then every treatment choice in the NCDB is inherently biased. Even with fancy strategies like propensity score matching. As Dr. Chang puts it, "it begs the question of why the [two performance status]-matched patients were allocated different treatments in the first place.” It sure does. What this study certainly shows is, despite its endorsement in national guidelines, aCT is infrequently used in the US for patients with a pCR to nCRT. TNT anyone?
Bottom Line: Less than one third of patients who achieve a pCR to neoadjuvant therapy go on to get adjuvant chemo, and these patients have a 7% absolute reduction in the risk of death. Which is causing which (risk of death ↔ receipt of adjuvant chemo) is nothing short of a maze, and the NCDB probably won't get us to the center. | Polanco & Dossa, JAMA Oncol 2018