Magnetic Resonance with Diffusion and Dynamic Perfusion-Weighted Imaging in the Assessment of Early Chemoradiotherapy Response of Naso-Oropharyngeal Carcinoma
, Eleonora Bicci 1, Linda Calistri 1, Chiara Lorini 2, Pierluigi Bonomo 3, Andrea Borghesi 4, Antonio Lo Casto 5, Francesco Mungai 6, Luigi Bonasera 6, Giandomenico Maggiore 7 and Cosimo Nardi 1,*
Round 1
Reviewer 1 Report
I only have a few suggestions to improve the manuscript.
AUC ignores incidence of the outcome, which can lead to misleading conclusions. The authors should complement AUC values with R^2 or scaled Brier values, which are measures of predictive accuracy that account for outcome incidence.
Figure 2 should also include spaghetti plots of pre-post values over time with the two groups of recurrence and post-CRT changes represented with different colors.
The choice of multiplying sensitivity and specificity to choose an optimal cutpoint is suboptimal due to ignoring incidence of the outcome. The same combination of sensitivity and specificity at different incidence will increase/decrease the rate of false positives, which will have an impact on patients (care and quality of life) and staff (workload). This sample is too small to be able to choose optimal cutpoints, but the Discussion section should also include a statement about how appropriate cut-offs should balance the benefits to the true positives (e.g., increased survival) versus the costs to false positives (e.g., unnecessary procedures).
Author Response
Point 1: “AUC ignores incidence of the outcome, which can lead to misleading conclusions. The authors should complement AUC values with R^2 or scaled Brier values, which are measures of predictive accuracy that account for outcome incidence".
Point 2: “Figure 2 should also include spaghetti plots of pre-post values over time with the two groups of recurrence and post-CRT changes represented with different colors”.
Response 1 and 2: The program (SPSS®) we used does not allow to obtain those parameters or to create that suggested graph, therefore if the referee deems it essential to add this, we need more time for revision.
Point 3: “The choice of multiplying sensitivity and specificity to choose an optimal cut point is suboptimal due to ignoring incidence of the outcome. The same combination of sensitivity and specificity at different incidence will increase/decrease the rate of false positives, which will have an impact on patients (care and quality of life) and staff (workload). This sample is too small to be able to choose optimal cut points, but the Discussion section should also include a statement about how appropriate cut-offs should balance the benefits to the true positives (e.g., increased survival) versus the costs to false positives (e.g., unnecessary procedures)”.
Response 3: Thank you for your suggestions. The statement “Finally, the study design did not allow the calculation of the outcome incidence. For this reason, a discussion of the appropriateness of the cut-off values with respect to the rate of false positives was not possible. Future studies with a different design should help in choosing appropriate cut-off values that balance the benefits to true positives (e.g., increased survival) versus the costs to false positives (e.g., unnecessary procedures)” has been added in the Discussion section (lines 337-341).
Reviewer 2 Report
The authors present a very challenging, yet interesting topic. Naso-oropharyngal carcinoma is a non-trivial task to detect and diagnose. The methods and scientific merit of this work are clear. The presentation is excellent.
I only have one suggestion to improve the paper: I feel like it lacks a clinical implication which should be stated and the processing seems complicated. Please add to the discussion whether your results can easily be used in clinical routine or if this is still a research field.
Author Response
Point 1: "I only have one suggestion to improve the paper: I feel like it lacks a clinical implication which should be stated, and the processing seems complicated. Please add to the discussion whether your results can easily be used in clinical routine or if this is still a research field".
Response 1: Thank you for your suggestions. The statement "Considering the relative complexity of DWI and DCE-PWI parameters that have been used and the low number of retrieved patients, the results obtained in our study are currently available for research purposes only." has been added in the Discussion section (lines 343-345).
