Table 3.

Associations between biomarkers and ovarian cancer risk^*

	All ovarian cancer (n = 34)		Serous ovarian cancers (n = 16)
	HR (95% CI)†	P	HR (95% CI)‡	P
Individual biomarker analyses
CA125	1.42 (1.18 to 1.70)	<.001	1.62 (1.24 to 2.12)	<.001
HE4	1.45 (1.08 to 1.96)	.015	1.59 (1.09 to 2.34)	.017
Mesothelin	1.35 (1.01 to 1.80)	.045	1.46 (0.97 to 2.21)	.072
B7-H4	1.20 (0.86 to 1.67)	.284	1.02 (0.61 to 1.69)	.947
DcR3	1.09 (0.71 to 1.68)	.696	1.22 (0.69 to 2.15)	.487
Spondin-2	1.58 (1.11 to 2.23)	.010	1.74 (1.08 to 2.80)	.022
CM1	1.34 (1.18 to 1.53)	<.001	1.42 (1.19 to 1.70)	<.001
CM2	1.18 (1.09 to 1.29)	<.001	1.23 (1.09 to 1.39)	<.001
CM3	1.44 (1.19 to 1.74)	<.001	1.69 (1.29 to 2.21)	<.001
CM4	1.44 (1.16 to 1.79)	.001	1.62 (1.19 to 2.21)	.002
Stepwise multivariable analyses§
CA125	1.48 (1.23 to 1.78)	<.001	1.60 (1.22 to 2.09)	<.001
HE4			1.52 (1.05 to 2.19)	.026
Mesothelin	1.46 (1.07 to 1.98)	.016

^*Associations were obtained by use of standardized time-dependent biomarker levels that were modeled individually and in stepwise multivariable Cox models. CI = confidence interval; CM1 = composite marker 1, defined as the sum of CA125, HE4, and mesothelin; CM2 = sum of CA125, HE4, mesothelin, B7-H4, DcR3, and spondin-2; CM3 = maximum of CA125, HE4, and mesothelin; CM4 = maximum of CA125, HE4, mesothelin, B7-H4, DcR3, and spondin-2; DcR3 = decoy receptor 3; HE4 = human epididymis protein 4; HR = hazard ratio.

^†Cox regression model adjusted for age at Carotene and Retinol Efficacy Trial (CARET) enrollment, race (white vs African American), and history of female relatives with breast or ovarian cancer (yes/no). P values based on Wald statistics. All statistical tests were two-sided.

^‡Cox regression model adjusted for age at CARET enrollment and female relatives with breast or ovarian cancer.

^§Cox regression models used a forward selection procedure at 5% level with adjustments as above.