TABLE 2.

Univariate and multivariate analyses of various predictive factors for pCR in different cohorts

		Univariable analysis	Multivariable analysis
		P value a	P value b	OR	95% CI
Beijing training cohort
Age	≥60/ < 60	.099
Gender	Male/female	.206
Tumor location	Upper, middle/lower	.194
Tumor differentiation	moderately, poorly/well differentiated	.675
Clinical TNM stage	II/III	1.000
Chemotherapy regimen c	1/2, 3	.593
Discriminant score	high/low	<.001	NA
Beijing validation cohort
Age	≥60/ < 60	.398
Gender	Male/female	.557
Tumor location	Upper, middle/lower	.258
Tumor differentiation	moderately, poorly/well differentiated	.503
Clinical TNM stage	II/III	.171
Chemotherapy regimen c	1/2, 3	.261
Discriminant score	high/low	<.001	<.001	98.633	9.335‐1042.113
Entire Beijing cohort
Age	≥60/ < 60	.229
Gender	Male/female	.241
Tumor location	Upper, middle/lower	.702
Tumor differentiation	moderately, poorly/well differentiated	.198
Clinical TNM stage	II/III	.312
Chemotherapy regimen c	1/2, 3	.353
Discriminant score	high/low	<.001	<.001	15.345	6.209‐37.925
External validation cohort
Age	≥60/ < 60	.957
Gender	Male/female	.222
Tumor location	Upper, middle/lower	.153
Tumor differentiation	moderately, poorly/well differentiated	.010	.173	0.274	0.042‐1.768
Clinical TNM stage	II/III	.043	.083	0.167	0.022‐1.266
Chemotherapy regimen c	1/2, 3	.629
Discriminant score	high/low	<.001	.002	55.384	4.544‐675.056

^aχ² or Fisher exact tests.

^bLogistic regression analysis with a forward stepwise procedure and likelihood ratio test.

^c1, platinum/paclitaxel; 2, platinum/fluorouracil; 3, platinum/others.

Abbreviations: pCR, pathological complete response; OR, odds ratio; CI, confidence interval; NA, not applicable for logistic regression analysis when covariance matrix could not be determined under the condition.