Table 2. Multivariate analysis for OS and EFS and bootstrap resampling for variable in the multivariate model.

Patients	Variable a	Cox regression		Bootstrap resampling
		P value	Hazard Ratio (95.0% CI)	Regression Coefficient	Std. Error	Bootstrap b
						P value	95.0% CI
overall cohort
	OS
	I/HD vs. SD cytarabine-based chemotherapy	0.002	0.261 (0.110–0.618)	-1.311	0.439	0.001	-2.294 – -0.562
	EFS
	I/HD vs. SD cytarabine-based chemotherapy	0.002	0.321 (0.155–0.666)	-1.068	0.361	0.002	-1.857 – -0.467
AML1/ETO-positive patients
	OS
	c-kit high vs. c-kit low	0.049	2.810 (1.003–7.872)	1.033	0.995	0.048	-0.039–2.483
	I/HD vs. SD cytarabine-based chemotherapy	0.012	0.233 (0.075–0.721)	-1.458	1.214	0.008	-3.157 – -0.411
	EFS
	c-kit high vs. c-kit low	0.033	2.739 (1.086–6.910)	1.008	0.809	0.030	0.054–2.346
	I/HD vs. SD cytarabine-based chemotherapy	0.013	0.298 (0.115–0.771)	-1.211	0.609	0.005	-2.483 – -0.400
AML1/ETO-positive and wtc-kit patients
	OS
	c-kit high vs. c-kit low	0.003	5.086 (1.732–14.933)	1.626	0.651	0.005	0.637–3.222
	EFS
	c-kit high vs. c-kit low	0.002	4.093 (1.695–9.888)	1.409	0.490	0.004	0.599–2.527

^a Variables considered for model inclusion were: c-kit expression (high vs. low), c-kit mutation status (mutation vs. wild-type), WBC count (10×10⁹/L increase), bone marrow blasts (10% increase), age (10-year increase), cytarabine-based chemotherapy (intermediate/high dose- vs. standard-dose), HSCT (allo- vs. no, auto- vs. no) and CR achievement (1 vs. ≥ 2 courses). Only variables significantly associated with outcomes in univariate analysis were included in the multivariate model.

^b bootstrap results are based on 1000 bootstrap samples. SD, standard dose; I/HD, intermediate/high dose.