LAUSR

With the advent of large-scale multicenter data repositories, prognostic data has evolved, causing paradigm shifts in therapeutic approaches and impacting informed discussions with patients and their families. This is especially true with the implementation of neoadjuvant treatment protocols in oncology. For patients with rectal cancer, neoadjuvant chemotherapy and radiotherapy are the standard of care for patients with stage II-III disease, often downstaging the disease. Historically, prognosis in this patient population has been controversial, in part due to variation in responses to neoadjuvant therapy. While it is relatively well accepted that a pathologic response to neoadjuvant therapy results in a favorable prognosis, data regarding the magnitude of this effect are conflicting [13]. Specifically, data regarding the relative importance of a patient’s presenting clinical stage versus postoperative pathologic stage have been lacking. To address this, we queried the National Cancer Database (NCDB), a dataset large enough to provide insight into a straightforward prognostic question [4]. Methods and major findings are summarized in Figure 1. Although the NCDB is subject to variability in local practices for staging and neoadjuvant therapy protocols, this heterogeneity also improves generalizability of our findings. We tried to overcome this by strict inclusion/ exclusion criteria intended to focus the study on pertinent patient populations, which may have introduced a selection bias. We excluded patients with local excision and those with positive surgical margins, as these patients are at increased risk for local recurrence. Provision of adjuvant radiotherapy was rare. As the indications for adjuvant radiotherapy as well as its impact on outcomes are unclear, these patients were also excluded. Although these exclusions seem appropriate for a study assessing the prognostic value of clinical versus pathologic stage for patients with rectal cancer receiving neoadjuvant therapy, careful statistical analysis remains necessary to ensure the reliability of the results. We employed inverse probability weighting by generating predicted probabilities of inclusion for each subject, including subjects excluded from the main analysis, and applying heavier statistical weight to subjects that were under-represented in the main analysis. When applying the inverse probability weights, pathologic staging yielded highly accurate prognostic information compared to clinical staging, similar to the main analysis. Despite these limitations, the resultant data were definitive. For a patient undergoing neoadjuvant radiation Editorial