br When considered in the context
When considered in the context of PORTEC-3 and GOG-258, our re-sults suggest that CT should be used as the first post-operative strategy for patients likely to both tolerate and benefit from RT. Such patients should likely be selected based on their expected ability to tolerate RT, a sufficiently high risk of LRR such that the absolute Clozapine-N-oxide in LRR is large, and a relatively low competing risk of distant failure, since a concurrent or subsequent distant failure would render pelvic control unimportant. To date, no randomized prospective trial has included a treatment arm in which CT is given before RT or sequenced in a “sand-wich” fashion as a comparator to multi-agent CT or RT alone. To our knowledge, no other retrospective analysis has estimated the associa-tion of sequence of adjuvant multi-agent CT and RT with clinical out-comes, nor are we aware of a prospective or retrospective database with sufficient sample size with which these findings could be vali-dated. The NCDB is therefore a valuable and appropriate resource for these analyses.
As the NCDB does not provide central pathologic or dosimetric re-view, we utilized strict exclusion criteria for the selection of our patient cohorts in order to achieve as homogeneous of a patient sample as pos-sible. The GOG-258 and PORTEC-3 were obliged for the purposes of ac-crual to enroll not only patients with Stage IIIC-IVA endometrioid uterine cancer but also patients with earlier stage disease with high-risk histology [1,2]. For these analyses, however, we included only pa-tients with “Type 1” uterine cancer, classically defined as Grade 1–2 endometrioid adenocarcinoma, to potentially improve patient homoge-neity given the benefit of central pathologic review was not available. Sensitivity analyses using an expanded cohort of women that included those with Grades 1, 2, and 3 endometrioid histology yielded results that were similar to those seen when only patients with “Type 1” dis-ease were included. Separate analyses should be considered for patients with “Type 2” disease in future analyses. An additional limitation of these analyses is the inability to consider clinical outcomes outside of OS, as the NCDB does not report locoregional or distant failures. Also unavailable to us for the purpose of cohort creation and inclu-sion in multivariable models includes detailed information regarding the number of CT cycles planned and received, type of CT regimen prescribed and delivered, therapy tolerability, and more granular data regarding RT technique and dosimetry. Lastly, compared with prospective trials, observational cohort analyses are limited due to potential selection bias and the inability to control for an unmea-sured confounder. To help mitigate this, multiple bias-reducing sta-tistical techniques were utilized to provide reliable and robust estimates, such as performing each multivariable analysis using matched cohorts of patients that were well-balanced along all clini-copathologic and relevant treatment criteria. In addition, a sensitiv-ity analysis demonstrated the TR estimates were robust to the introduction of a possible unmeasured confounder.
Kaplan-Meier estimates of overall survival and multivariable parametric accelerated failure time models of propensity score-matched pairs of women who received adjuvant radiotherapy and/or chemotherapy.
Variable Univariate analysis of propensity score-matched cohorts
Multivariable AFT model
RT after CT vs CT only
RT before CT vs CT only
RT after CT vs RT only
RT before CT vs RT only
CT only vs RT only
Accelerated failure time (AFT) models for the matched cohorts were covariate-adjusted and inverse probability-weighted using propensity-weighted matched cohorts. Significance de-termined by Log-Rank test (univariate analyses) or Wald test (multivariable analyses). OS: overall survival; mo: months; y: years; N: number; RMST: restricted mean survival time; TR (95% CI): time ratio (95% confidence interval); P: P-value; RT: radiotherapy; CT: chemotherapy; ref.: reference level.
In conclusion, these hypothesis-generating findings support further consideration of adjuvant regimens that include the delivery of CT be-fore RT. For women with locally advanced Type 1 EC, treatment with multi-agent CT before RT (or potentially employing a sandwich ap-proach) may optimize control of both locoregional and distant disease, ultimately leading to longer PFS and OS. The findings presented here support the consideration of intracellular digestion regimen in the design of future pro-spective trials.
Supplementary data to this article can be found online at https://doi.
We acknowledge the NCDB for collecting the data used in these anal-yses. The NCDB and the participating hospitals are the source of the data. The NCDB has not verified and are not responsible for the statisti-cal validity of the data analysis or conclusions.