Table 3.

Logistic regression models for pathological complete response in the training and testing datasets.

	Univariate model			Multivariate model
	OR	95% CI	P-value	OR	95% CI	P-value
Training set
lncRNA predictive score	2.72	2.06–3.64	4.49 × 10-12	1.60	1.06–2.44	2.65 × 10-2
Grade (ref = 1)
2	2.10	0.39–39.00	4.84 × 10-1	0.50	0.07–10.30	5.54 × 10-1
3	12.32	2.56–221.60	1.44 × 10-2	1.51	0.20–31.29	7.25 × 10-1
ER (ref = negative)
Positive	0.22	0.14–0.36	8.38 × 10-10	0.82	0.35–1.87	6.41 × 10-1
PAM 50 subtypes (ref = Basal-like)
Her2+	0.43	0.16–0.98	5.83 × 10-2	0.50	0.13–1.53	2.57 × 10-1
Luminal A	0.06	0.02–0.14	4.51 × 10-9	0.26	0.06–0.95	5.12 × 10-2
Luminal B	0.34	0.16–0.65	1.96 × 10-3	0.92	0.32–2.69	8.85 × 10-1
Testing set
lncRNA predictive score	2.69	2.09–3.51	6.44 × 10-14	1.83	1.27–2.67	1.33 × 10-3
Grade (ref = 1)
2	3.21	0.63–58.67	2.63 × 10-1	1.48	0.27–27.67	7.13 × 10-1
3	13.95	2.93–250.25	1.00 × 10-2	2.75	0.51–51.32	3.42 × 10-1
ER (ref = negative)
positive	0.19	0.12–0.29	1.11 × 10-13	0.41	0.20–0.85	1.52 × 10-2
PAM 50 subtypes (ref = Basal-like)
Her2+	0.97	0.49–1.87	9.18 × 10-1	1.38	0.63–2.99	4.15 × 10-1
Luminal A	0.05	0.01–0.13	4.60 × 10-7	0.30	0.06–1.05	8.27 × 10-2
Luminal B	0.28	0.17–0.44	3.84 × 10-8	1.02	0.48–2.16	9.56 × 10-1