Fig. 2

Nomogram established based on CARMN’s relationship with better cisplatin-based neoadjuvant chemotherapy response in breast cancer. A, B Texture feature selection for predicting response of neoadjuvant chemotherapy using lasso logistics regression. A Using minimum criteria of 10-fold cross-validation to select parameter lambda(λ) of LASSO model, and selected minimum λ turns out to be 0.0361. The upper x axis shows number of variants response to different λ values. The left dot line shows the minimum of λ and the right one shows λmin -SD. B LASSO coefficient profile plot of 12 texture features according to λ value selected in (A) which is also drawn as vertical line. Six optimal features are finally selected by lasso coefficient. C Forest plot illustrating factors selected by LASSO regression in predicting neoadjuvant chemotherapy response using multivariate logistics regression. D A nomogram is built for evaluating cisplatin-based neoadjuvant chemotherapy pCR rate based on multivariate logistics regression. E ROC curve of nomogram and other models for predicting cisplatin-based NAC response. F Calibration curve of the prediction model. X axis means the predicted probability of pCR. Y axis means the actual probability of pCR. The line between opposite angles means the ideal line. The other two lines means the actual prediction performance of the nomogram. G Decision curve of multivariate logistics model. The red line represents the multivariate logistics model including CARMN and clinicopathological factors, the blue one represents model with only clinicopathological factors. H ROC curve of nomogram in a validation set. I CARMN expression in ER- breast cancer with different neoadjuvant chemotherapy response. H Distribution of CARMN expression level in ER- breast cancer with different neoadjuvant chemotherapy response. ER: estrogen receptor; HER2: human epidermal growth factor receptor 2; TNBC: triple negative breast cancer; NAC: neoadjuvant chemotherapy; pCR: pathological complete response. Error bars represent means ± SD, *P < 0.05