In their letter, Dolan and McMillan raise several important points. First, many studies have concluded that the neutrophil-to-lymphocyte ratio (NLR) is a cancer biomarker. Second, the majority of studies analyzed… Click to show full abstract
In their letter, Dolan and McMillan raise several important points. First, many studies have concluded that the neutrophil-to-lymphocyte ratio (NLR) is a cancer biomarker. Second, the majority of studies analyzed NLR as a dichotomized variable. In contrast, we kept NLR in its continuous form; our study found no evidence that NLR is prognostic or predictive of survival in patients with muscle-invasive bladder cancer. Guidelines recommend analyzing variables in their continuous form. The statistical shortcomings of dichotomization, including the risk of spurious findings, have been shown elsewhere in detail. Furthermore, a continuous variable can be split at many different cutpoints; this inflates researcher degrees of freedom and may raise issues of multiple comparisons. We appreciate the opportunity to highlight these concerns regarding the existing NLR literature. Therefore, our study kept NLR as a continuous variable. Analytic checks supported the modeling assumptions of this approach (eg, a linear functional relationship and proportional hazards). However, for the sake of transparency, we present here the results using dichotomized NLR at the thresholds of 3 and 5, as per the request of Dolan and McMillan. In multivariable models adjusted for age, sex, tumor T category, and treatment arm, NLR was neither prognostic for overall survival (hazard ratio [HR], 1.32 [95% confidence interval (95% CI), 0.97-1.79; P 5.07] for a cutpoint 3 and HR, 1.17 [95% CI, 0.76-1.81; P 5.46] for a cutpoint 5) nor predictive of the benefit from neoadjuvant chemotherapy (HR, 0.97 [95% CI, 0.56-1.70; P 5.92] for a cutpoint 3 and HR, 0.63 [95% CI, 0.291.38; P 5.25] for a cutpoint 5). We note that, in this case, the threshold for statistical significance would be below the traditional level of .05 due to multiple comparisons. In short, the dichotomized results do not change our study’s conclusions. FUNDING SUPPORT No specific funding was disclosed.
               
Click one of the above tabs to view related content.