Table 1. Patients information.

Patient	Gender	Age	Tumor type	Clinical stage	Tumor size	IHC
1	Female	58	Invasive breast carcinoma	pT2N0	3.0×2.8×2.0cm	CK5&6(−), E-cadherin(1+), EGFR(−), ER(+,95%), HER2(2+), Ki-67(+,30%), P53(−), PR(+, 60%), TOP2A(2+)
2	Female	53	Invasive breast carcinoma	pT1cN0	1.2×1.2×1.0cm	CK5&6(−), E-cadherin(2+), EGFR(−), ER(+, >90%), HER2(1+), Ki-67(+, 25%), PR(+, 10%)
3	Female	54	Invasive breast carcinoma	pT1cN0	1.5×1.2×1.2cm	CK5&6(−), E-cadherin(2+), EGFR(−), ER(+80%), HER2(2+), Ki-67(+20-30%), P53(−), PR(+10%), TOP2A(1+)
4	Female	47	Invasive lobular carcinoma	pT1cN0	1.4×1.2×1.2cm	CK5&6(−), E-cadherin(−), EGFR(−), ER(+, >95%), HER2(1+), Ki-67(+, 10%), P53(−), PR(+, >95%), TOP2A(1+)
5	Female	64	Invasive breast carcinoma	pT1N0	1.8×1.5×1.3cm	CK5&6(−), E-cadherin(1+), EGFR(−), ER(+80%), HER2(1+), Ki-67(+40%), P53(−), PR(+, <10%), TOP2A(2+)
6	Female	38	Invasive breast carcinoma	pT1cN0	1.8×1.5×1.3cm	CK5&6(−), E-cadherin(3+), EGFR(−), ER(+, 80%), HER2(1+), Ki-67(20%+), P53(−), PR(+, 80%), TOP2A(10%+)
7	Female	38	Invasive breast cancer with focal DCIS	pT1N0	2.2×1.8×1.3cm	CK5&6(−), E-cadherin(3+), EGFR(−), ER(+, >90%), HER2(1+), Ki-67(+5%), P53(−), PR(+, 25%)
8	Female	43	Invasive breast cancer with focal DCIS	pT1bN0	1.0×0.5×0.5cm	CK5&6(−), E-cadherin(2+), EGFR(−), ER(+, >90%), HER2(1+), Ki-67(+, 30%), P53(2+), PR(+, 70%), TOP2A(1+)
9	Female	63	Invasive ductal carcinoma	pT1N0	1.8×1.3×1.1cm	CK5/6(−), E-cad(2+), EGFR(1+), ER(+90%), HER2(2+), Ki-67(+, 30%), P53(−), PR(+1-10%), Top2A(1+)
10	Female	57	Invasive breast carcinoma	pT1N0	1.3×1.0×1.0cm	CK5&6(−), E-cadherin(2+), EGFR(−), ER(80%), HER2(1+), Ki-67(30%), P53(+), PR(80%), TOP2A(1+)
11	Female	53	Invasive breast carcinoma	pT1cN0	1.4×1.4×1.2cm	CK5&6(−), E-cadherin(3+), EGFR(−), ER(+, >95%), HER2(−), Ki-67(+, 30%), P53(−), PR(+, 80%), TOP2A(2+)

To calculate the recurrence score, we converted the relative transcripts expression fold from both RT-MLPSeq method and RT-qPCR method to ΔCt value (−log₂) and finally obtained the RS value based on the algorithm presented by Paik, Shak, Tang, et al. in which ΔCt values from real time PCR system were used as input data [3]. As shown in Figure 6, recurrence scores from RT-MLPSeq was highly consistent with RT-qPCR results (R²=0.9309).