Figure 6.

A pretreatment serum protein panel to predict the response to preoperative chemotherapy

(A) Unsupervised consensus clustering of pretreatment serum protein levels identified four clusters of patients.

(B) Pretreatment serum protein levels of four clusters.

(C) Treatment response in four clusters of patients.

(D) Lauren classification of tumors in four clusters of patients. p value was calculated by chi-squared test.

(E) ROC curve demonstrates treatment response predictive accuracy of pretreatment serum response predictive score (PSRscore).

(F) Formula of PSRscore.

(G) Treatment response in patients with high/low PSRscores.

(H) Post-treatment stromal PD-L1 level in patients with different PSRscores (high [H]: N = 18, low [L]: N = 14). p value was calculated by chi-squared test.

(I) The pretreatment patient stratification strategy based on PSRscore.

See also Figure S6.