Disqualification of Neoadjuvant Rectal Score Based on Data of 6596 Patients From the Netherlands Cancer Registry

Abstract Background The neoadjuvant rectal score (NAR) was developed as a surrogate endpoint for overall survival in patients with rectal cancer after neoadjuvant treatment. We aimed to validate the NAR score in patients from the Netherlands Cancer Registry database. Patients and Methods We studied patients with rectal cancer treated with long-course neoadjuvant therapy followed by surgery in the Netherlands between 2007 and 2014. The probability of concordance with overall survival and the goodness of fit of several models were evaluated using Harrell's concordance index ( c index) and the Akaike information criterion (AIC), which is used to compare the quality of statistical models. Results The NAR score resulted in a c index of 0.665. We found that single pathologic parameters (pT or pN) have similar concordance as the NAR formula ( c index of 0.663 and 0.655, respectively). A combination of pT and pN resulted in better concordance with the true endpoint, overall survival ( c index 0.684), and a simple Cox regression model with the 3 parameters included in the NAR formula (cT, pT, and pN) improved the concordance even more ( c index 0.689). When the AIC index was compared for all models, the NAR score model showed the worst fit to the true endpoint. Conclusion We found no additional value for using the NAR formula as a surrogate endpoint for overall survival in rectal cancer patients treated with neoadjuvant therapy.