Reply to “Survival analysis in a prediction model for early systemic recurrence in breast cancer”

We thank Qiu et al. for their interest in our study. Referring to the Transparent Reporting of a Multivariable Prediction Model for Individual Prognosis or Diagnosis statement, they remarked on four potentially insufficient descriptions in our article in terms of a prediction model study. The four limitations were as follows: patient followup information, missing data handling methods, 95% confidence intervals (CIs) for the model performance, and a patient selection diagram. Two of these items were not described in the article; however, the remaining two items already were stated in the article or seemed unnecessary to include in the article. Qiu et al. also mentioned a potential flaw in the multivariate analysis technique used in our study. However, the Cox model recommended by them could not be used to predict the probability of recurrence and validate the models in our study. In the following sections, we respond to each of the five items individually.

Cancer October 15, 2022 Correspondence Reply to "Survival analysis in a prediction model for early systemic recurrence in breast cancer" We thank Qiu et al. for their interest in our study. 1 Referring to the Transparent Reporting of a Multivariable Prediction Model for Individual Prognosis or Diagnosis statement, 2 they remarked on four potentially insufficient descriptions in our article in terms of a prediction model study. The four limitations were as follows: patient follow-up information, missing data handling methods, 95% confidence intervals (CIs) for the model performance, and a patient selection diagram. Two of these items were not described in the article; however, the remaining two items already were stated in the article or seemed unnecessary to include in the article.
Qiu et al. also mentioned a potential flaw in the multivariate analysis technique used in our study. However, the Cox model recommended by them could not be used to predict the probability of recurrence and validate the models in our study. In the following sections, we respond to each of the five items individually.

MISSING DATA HANDLING METHODS
The authors stated that the methods for handling missing data should have been included in the article. Because of the strict word limit of the original article, we were unable to include the missing data handling methods. In our study, we used the pairwise deletion (available case analysis) technique for univariate analyses and the listwise deletion (complete case analysis) technique for multivariate analyses and internal validation.

PATIENT SELECTION DIAGRAM
Qiu et al. recommended that we present a patient selection diagram in our article. However, the patient selection procedure, including the inclusion and exclusion criteria, was included in the Patient Selection and Patient Characteristics sections of our article. Therefore, we did not draw a diagram because there was no additional information.

MULTIVARIATE ANALYSIS TECHNIQUE
Qiu et al. recommended using the Cox regression model instead of the logistic regression model in our study. However, the logistic model was an appropriate statistical method for our study, whereas the Cox model could not accomplish the purposes of our study.
The Cox model can analyze censored data; however, it cannot calculate individuals' estimated probabilities, which are essential for model validation. In contrast, the Cancer October 15, 2022 logistic model can calculate the estimated probabilities of individuals. The main objectives of this study were to establish the predictive models with the training and validation cohorts and to publish the online tool for estimating the probability of 5-year distant recurrence-free survival rates (https://www.thera nosti cs.jp/sln/show). Therefore, we constructed the predictive models with logistic regression analysis.

ACKNOWLEDGMENT
This study was supported by a research grant from Sysmex.

CONFLICTS OF INTEREST
Tomo Osako has received honoraria from Diaceutics and consulting fees from Chiba Cytopathology Laboratory outside the submitted work. Hitoshi Tsuda has received research funding from Taiho, Goryo Chemical, and Roche Diagnostics and scholarship donations from Chugai, Takeda, and Eli Lilly outside the submitted work. Shinzaburo Noguchi is an advisor to Sysmex and has received consulting fees and research funding from Sysmex; he is an advisor to AstraZeneca and Nittobo and has received honoraria and/or consulting fees from AstraZeneca, Eli Lilly, Chugai, Nittobo, and Sysmex outside the submitted work; and he holds joint patents not related to this study with Sysmex. The other author made no disclosures.

FUNDING INFORMATION
Sysmex.