Novel 18-gene signature for predicting relapse in ER-positive, HER2-negative breast cancer

Breast Cancer Research

Table 3 Statistical analysis of TransATAC validation cohort

Score	No. of patients (relapses)	Univariate comparisons					Multivariable comparisons
		Univariate comparisons					CTS + signature vs CTS				CTS + signature
		LRχ²	p Value	HR (95% CI)	P diff	C-index (SE)	ΔLRχ²	p Value	HR (95% CI)	P diff	C-index (SE)
0–10 years
CTS	314 (59)	48.7	< 0.001	2.16 (1.79–2.62)	–	0.674 (0.018)	–	–	–	–	–
10-Year signature		28	< 0.001	1.98 (1.54–2.55)	Reference	0.671 (0.026)	7.9	0.005	1.49 (1.13–1.96)	Reference	0.709 (0.021)
Early signature		33.4	< 0.001	2.06 (1.62–2.61)	0.334	0.678 (0.024)	10.3	0.001	1.55 (1.19–2.02)	0.48	0.711 (0.020)
Late signature		18.1	< 0.001	1.72 (1.34–2.20)	0.000	0.642 0(.029)	4.3	0.037	1.33 (1.02–1.74)	0.004	0.700 (0.022)
0–5 years
CTS	314 (26)	29	< 0.001	2.04 (1.53–2.74)	–	0.679 (0.023)	–	–	–	–	–
10-Year signature		14.1	< 0.001	2.05 (1.41–2.98)	0.833	0.678 (0.037)	3.2	0.073	1.46 (0.97–2.19)	0.77	0.712 (0.029)
Early signature		14.9	< 0.001	2.00 (1.42–2.81)	Reference	0.672 (0.035)	2.8	0.096	1.40 (0.95–2.06)	Reference	0.705 (0.028)
Late signature		8.9	0.003	1.77 (1.22–2.57)	0.138	0.648 (0.042)	1.7	0.19	1.31 (0.87–1.97)	0.65	0.705 (0.031)
5–10 years
CTS	270 (33)	19.8	< 0.001	1.84 (1.33–2.54)	–	0.657 (0.026)	–	–	–	–	–
10-Year signature		13.9	< 0.001	1.93 (1.37–2.72)	0.027	0.663 (0.036)	4.8	0.028	1.53 (1.05–2.22)	0.14	0.696 (0.030)
Early signature		18.6	< 0.001	2.11 (1.52–2.94)	0.091	0.681 (0.032)	8	0.005	1.70 (1.19–2.43)	0.14	0.708 (0.028)
Late signature		9.3	0.002	1.68 (1.21–2.34)	Reference	0.636 (0.038)	2.7	0.099	1.36 (0.95–1.94)	Reference	0.686 (0.031)

CTS Clinical Treatment Score, LR Likelihood ratio
Both univariate and multivariable analyses are presented for years 0–10, years 0–5, and years 5–10 separately. Likelihood ratio test based on Cox proportional hazards models for univariate and multivariable analyses. Differences in likelihood ratio values (ΔLRχ²) were used. CTS was used as a covariate in the multivariable regressions. For each score, HRs per SD change are presented

ISSN: 1465-542X