Background The presence of high levels of stromal tumor infiltrating lymphocytes (TILs) has been associated with better prognosis in early triple-negative breast cancer (TNBC). The Immunoscore® (IS) is a prognostic tool, which categorizes the densities of spatially positioned CD3 and CD8 cells in both invasive margins (IM) and the center of the tumor (CT), yielding a five-tiered classification (0–4). High IS values have been reported to predict improved outcomes in colorectal cancer.

Methods The cohort consisted of 52 TNBC patients (pts) who previously received neo-adjuvant anthracycline and taxane based chemotherapy. Quantitative analysis of the immune cells was carried out using a computer-assisted image analysis in different tumor locations for CD3 and CD8 T-cell markers. Additionally, we measured stromal TILs according to the internationals TILs working group. Pre-treatment tumor samples were immune-stained for CD3 and CD8 T-cell markers and stromal TILs. The relationship between various clinical pathological factors including tumor size, glands, stage and immune factors were analyzed by Chi2 and Fischer exact test. The logrank test and the Kaplan Meyer methods were used to estimate relapse free survival.

Results The median age of the patients was 50 years (27–84 years). Tumor sizes were categorised as T1 = 9 patients (17%), T2 = 41 patients (77%) and T3 = 3 patients (6%). Patients with positive glands = 19 (36%) patients and patients without gland involvement = 34 (64%). Stage grouping included stage 1 = 5 (9%), stage IIA = 33 (63%) patients, stage IIB = 9 (17%) patients, stage III = 6 (11%) patients. The median Ki-67 was 45% (5 – 90%). The median density of CD3 CT cells = 1190 mm² (range 34 – 4614), CD3 IM = 1855 mm² (range 57 – 6190), CD8 CT 508 mm² (range 17 – 2486) and CD8 IM 805 mm² (range 90 – 3156). The median percentage of stromal TIL’s was 5% (0 – 60%). Patient with an IS of 0 = 4 patients (8%), IS 1 = 3 (5%), IS 2 = 20 patients (38%), IS 3 = 24 patients (45%) and IS 4 = 2 patients (4%). The pathological complete response (pCR) rate of the entire cohort was 62%. A positive correlation was found between TILs and CD3 CT (R = 0.641, p < 0.0000), CD8 CT (R = 0.5623, p < 0.0000), CD3 IM (R = 0.6099, p < 0.0000), and CD8 IM. (R = 0.5010, p < 0.0010). TILs correlated with immunoscore (R = 0,3603, p < 0.0087). There was no correlation between TILs and Ki-67 (R = 0.1497, p < 0.2943). On univariate analysis, factors associated with higher pCR included nodal status (positive = 42,11% vs negative = 73,53% (p<0,02362) and Ki67 <40%= 33,33% vs =40% = 76,47% (p<0,00235). A high density of CD3 (> than 1100 mm²) and CD8 (> than 400 mm²) positive T-cells in the CT was associated with higher pCR (CD3 CT: 30% vs 70%, p=0.00489 and CD8 CT: 30% vs 70%, p=0.03344). Analysis of CD3 (> than 1200 mm²) (CD3 IM: 12% vs 88%, p=0.0.02367) and CD8 in the IM (> than 550 mm²) was also significant for an association with pCR (CD8 IM:23% vs 77%, p=0.03). High IS (3+4= 73%) vs intermediate (2=55%) vs low (0+1=43%) showed a numerical difference, however, did not reach a statistical significance with pCR (p=0.111). Analysis of TILs = 20% showed a pCR of 76% compared to patients with TILs < 20% with a pCR of 54% (p < 0.12295). A Ki67 =40% was associated with pCR of 76% compared to patients with Ki67 < 40% with a pCR 33% (p < 0.00235).The median time to progression of the patients not attaining a pCR was 1600 days, compared to patients who did attain a pCR with a median PFS not reached yet, but exceeds 1800 days. The median time to progression of patients with Ki67 < 40% was 1700 days while the patients with Ki67 =40% has not been reached yet (p < 0.03). At 1800 days, 95% of patients with CD3 > 1100 mm² did not relapse, compared to 75% patients with CD3 = 1100 (p < 0.03).

Conclusions This exploratory study shows that analysing CD3 and CD8 in the centre of the tumor and invasive margin might be more sensitive than examination of TILs in TNBC patients.

Ethics Approval The study was approved by Pharma-Ethics - (Institution’s Ethics Board), approval number 170516563.