Association Between Signet Ring Cells and Three-Year Conditional Survival following Trimodality Therapy for Esophageal Adenocarcinoma: A Bayesian Analysis
K. G. Mitchell1, H. Feldman2, D. Milton1, M. B.. Antonoff2, W. Hofstetter3, R. Mehran4, D. Rice5, S. Swisher1, A. Vaporciyan2, G. Walsh6, R. Lin1, P. Thall1, R. Rajaram7 1University of Texas MD Anderson Cancer Center, Houston, Texas 2The University of Texas MD Anderson Cancer Center, Houston, Texas 3MD Anderson Cancer Center - University of Texas, Houston, Texas 4UT MD Anderson Cancer Center, Houston, Texas 5The University of Texas MD Anderson Cancer Center, Dept. of Thoracic, Houston, Texas 6The University of Texas, M.D. Anderson Cancer Center, Houston, Texas 7University of Texas-MD Anderson Cancer Center, Houston, Texas
Fellow, Cardiothoracic Surgery (4+3) University of Texas MD Anderson Cancer Center Houston, Texas, United States
Disclosure(s):
Kyle G. Mitchell, MD, MSc: No financial relationships to disclose
Purpose: Although signet ring cell (SRC) histology is associated with resistance to neoadjuvant chemoradiotherapy and worse overall survival (OS) in esophageal adenocarcinoma (EAC), its prognostic relationship among patients who survive the early postoperative period is unknown. We therefore examined the impact of SRC status on conditional survival in EAC. Methods: All patients with EAC who underwent trimodality therapy at a single institution (2006-2018) were identified. Patients who underwent salvage resection and those who died within 90 days postoperatively were excluded. Multivariable Bayesian survival regression was performed using a gamma model with non-informative prior distributions for all covariate effects on survival time. Conditional survival analyses were repeated using a three-year landmark with a log-normal Bayesian survival model, using the same noninformative prior distributions. For the coefficient of each covariate in each model, the posterior probability of a beneficial effect (PBE; Pr β < 0|Data), which is the probability that the death rate was lower for larger values of the covariate, was calculated. A PBE near 1 corresponds to a large beneficial effect, near 0 corresponds to a small beneficial effect, and 0.50 to no effect. Results: Of 631 patients who met inclusion criteria, most were men (556, 88%), smokers (405, 64%), and underwent R0 resection (590, 94%). SRCs were present in 16.0% (101) of patients. Those with SRC tumors were more likely to experience an incomplete resection (R1/R2: 11% [11/101] vs 6% [30/530]) and had higher pathologic stage (ypStage III: 44% [44/101] vs 28% [149/530]) compared to patients without SRCs. After a median follow-up time of 44.4 months, SRC was associated with shorter postoperative OS (median survival time 45.8 months [95% CI 31.0-96.7] vs 79.8 [63.0-107.2]; Figure 1). Upon multivariable analysis (MVA), the absence of a SRC component was moderately associated with improved OS (PBE = 0.88). However, upon MVA of 3-year conditional OS (N = 357), SRC status no longer retained a prognostic effect (PBE = 0.53 for absence of SRC), whereas higher pathologic stage was strongly associated with worse additional OS (Table 1). Conclusion: The presence of SRC portends worse OS following trimodality therapy for EAC. However, this prognostic impact is dynamic and abates by three years postoperatively. In contrast, higher pathologic stage remains associated with both poor overall and conditional survival. These findings may inform postoperative patient counselling and surveillance protocols.
Identify the source of the funding for this research project: None
