A total of 96 tumor samples from the Department of Pathology of Hospital San José TecSalud (Monterrey, Mexico) were collected for the present study. IRB approval was obtained from the Ethics Committee for Research at Tecnológico de Monterrey, and the National Bioethics Commission (code id: CONBIOETICA19CE100820130520), and also was granted in accordance with the Declaration of Helsinki. Informed consent was obtained from each patient prior to tumor sample collection.

Tumor tissue from patients with breast cancer was prospectively collected at the Breast Cancer Unit between August 2011 and August 2015. Demographic, familiar, clinical, tumor grade, lymph node status, ER, PR, and HER-2/neu data were recorded.

Tumor samples were collected using ultrasound-guided core needle biopsies. Tumor cylinders (4–6) were obtained from each patient for hematoxylin-eosin (H&E) staining and immunohistochemistry prior to NAC. Tumor sampling for microarray studies was performed as previously described (18).

Prior to standard NAC, the center of the breast tumor was marked by a charcoal suspension injection. If the tumor was palpable, the injection was performed by a breast surgeon. The injection in non-palpable tumors was performed by a radiologist under ultrasound guidance. The charcoal suspension was used to localize the tumor following NAC for pathological analysis.

The NAC regimen consisted of doxorubicin i.v. (≥60 mg/m²) + cyclophosphamide IV (600 mg/m²) every 3 weeks for 4 doses, followed by paclitaxel 80 mg/m² by i.v. infusion for 1 h weekly for 12 doses.

The histological grade of the core needle biopsies was obtained prior to neoadjuvant therapy, using the Bloom-Richardson score (19). The stage of breast cancer was determined according to the American Joint Committee on Cancer (AJCC) (20). The assessment of the percentage of TILs was performed as per the recommendations of the International TIL Working Group 2014 in breast cancer (21). The patients were divided into three categories according to TIL percentage: 1–19%, low (20); <49%, intermediate; and ≥50%, high (21).

All biopsy specimens were fixed in 10% neutral buffered formalin between 6–24 h and embedded in paraffin at room temperature. Tissue sections (4-µm thick) were obtained on coated glass slides prior to NAC treatment and stained with H&E. The H&E stain was performed by deparaffinizing sections, washing twice with xylene for 10 min each. The sections were subsequently rehydrated twice using a descending alcohol gradient (100, 95 and 70%). Next, the sections were briefly washed in distilled water, stained in Harris hematoxylin solution for 8 min, washed in running tap water for 5 min, treated with 1% acid alcohol for 30 sec and washed with running tap water for 1 min at room temperature. Lastly, the sections were treated with 0.2% ammonia water for 30 sec, rinsed in 95% alcohol (10 dips) and counterstained in eosin-phloxine solution for 30–60 sec at room temperature. The stromal component of TILs was evaluated within the borders of the invasive carcinoma using a magnification of ×200 (ocular, ×10; objective, X20). The following areas were excluded from the study: Tumoral borders; areas around the in situ carcinoma; normal breast lobules; and tumor areas exhibiting artifacts; necrosis and hyalinization. The cancer immunophenotype experiments were performed on the core needle biopsies, prior to neoadjuvant therapy. The evaluation of the pathological response following NAC was determined using the de Miller-Payne and MD Anderson Cancer Center (Residual Cancer Burden) system (22).

ER, PR, HER-2, Ki67, CD3⁺, CD4⁺ and CD8⁺ IHC analysis were performed with a Ventana BenchMarck GX^® autostainer (Roche Diagnostics). Sections of 5-mm were obtained, paraffin slides were deparaffinized using 2 changes of xylene for 10 min each and hydrated using graded alcohol and distilled water (2 changes of 100% ethanol, 2 changes of 95% ethanol and 2 changes of distilled water) for 10 min each at room temperature (15–25°C). Heat-induced epitope retrieval with citrate buffer was performed. Slides were then cooled and rinsed with distilled water, rinsed in tris buffered saline with 20 ml of Tween-20 for 5 min. Slides were then rinsed with 3% hydrogen peroxide, followed by a rinse with a wash buffer and covered with 300 µl of protein block for 5 min at room temperature. Slides were treated with Dako; Agilent Technologies Inc. monoclonal primary antibodies to HER2/Neu (1:500; Clone 4B5 for 32 min at 37°C), ER (1:500; Clone SP1), PR (1:500; Clone 1E2), Ki67 (1:100; Clone 30–9), CD3⁺ (1:500; clone 2GV6), CD4⁺ (1:500; clone SP35) and CD8⁺ (1:500; clone SP57). Except for HER2/Neu all the other primary antibodies were incubated for 16 min at 37°C. Slides were then rinsed with wash buffer and secondary antibody Dako Agilent Technologies Inc. Envision labeled polymer HRP anti-rabbit (1:100; cat. no. K4002) was applied for 5 min at 15–25°C. After the secondary reagent, DAB was applied for 10 min and the slides were rinsed with distilled water. Counterstaining was done with hematoxylin for 3 min at room temperature and slides were washed in tap water at room temperature. Slides were then blued in ammonia water, rinsed in tap water, dehydrated in graded alcohol (95 and 100% ethanol), cleared in xylene (2 changes) for 10 min each at room temperature and coverslipped for microscopic examination. The immunohistochemistry were performed following the American Society of Clinical Oncology/College of American Pathologists guidelines (Tables I and II) (23).

TIL-positive immunohistochemical scoring was performed by 2 breast pathologists from Hospital San José TecSalud, Tecnológico de Monterrey (Monterrey, Mexico). CD3⁺, CD4⁺ and CD8⁺ TILs were counted in 5 randomly selected high-power fields using a Carl Zeiss^® Axio Lab A1 light microscope (magnification, ×40), from which the average was calculated. TIL count in the stroma was rated as follows: + (1–25 cells); ++ (26–50 cells); and +++ (≥51 cells); as previously described (24). TIL cases with <25 cells were considered as ‘low TIL count’, and those with >25 cells (++, +++) as ‘high TIL count’ (Fig. 1) (24).

RNA extraction of tissue biopsies was performed according to the manufacturer's instructions, using RNeasy Fibrous Tissue Mini kit (Qiagen, Inc.), which is used to obtain high-quality RNA in small tissue biopsies. The RNA quality was assessed by capillary electrophoresis using the Experion™ Automated Electrophoresis Station (Bio-Rad Laboratories, Inc.). The mean RNA Quality index was 7.84 (range, 6.1–9.8). The RNA concentration was determined using a NanoDrop 8000 spectrophotometer (Thermo Fisher Scientific, Inc.). The mean RNA concentration of the tested samples was 269.02 ng/µl, (range, 38.30–999.13 ng/µl). If several core tumor samples were collected during the procedure, 2–3 core samples were used to isolated total RNA and another 2–3 for the immunohistochemistry (IHC) assay.

Gene expression analysis was performed using frozen fresh tissue tumor samples. The tumor tissue samples from the biopsies were immediately treated with an RNase inhibitor preserver solution (RNAlater; Thermo Fisher Scientific, Inc.) (25). Selected RNAs were used for microarray hybridization and gene expression analysis, which were conducted using the GeneChip 3′IVT Express kit (Thermo Fisher Scientific, Inc.), following the manufacturer's protocol. A recommended quantity of 100 ng total RNA input was used for each sample. Poly-A controls (Thermo Fisher Scientific, Inc.), which are exogenous positive controls that monitor the entire target preparation, were used for all samples. The hybridization mixture was prepared and applied to the GeneChip Human Genome U133 Plus 2.0 array (Thermo Fisher Scientific, Inc.), measuring >43,000 transcripts that represented >20,000 human genes. Washing and scanning processes were performed in the Fluidics Station 400 and GeneChip Scanner 3000 7G, respectively. The preliminary data analysis was completed using the Affymetrix Micorarray Suite software version 5.0.0.032 (Thermo Fisher Scientific, Inc.).

Normalization was performed using Robust Microarray Analysis (RMA) and quantile normalization. One sample exhibited poor quality and was removed from the analysis. The probes whose mean expression (log scale) was <4 (in a logarithmic scale resulting from RMA) were removed. A gene expression signature was performed by t- and Kolmogorov tests with a multiple comparison correction using the false discovery rate (FDR method (26). Probes with significant P-values in both tests (FDRt and Kolmogorov test <0.05%) were considered positive. Analyses were performed in R software [R Core Team. (2014)] (27).

Interaction analysis of the selected genes was performed using GO (28,29) and KEGG (30–32) databases through the online tool STRING: Functional protein association networks version 11.0 (33). STRING analysis demonstrated functional enrichments of selected genes: GO analysis of Biological Process, Molecular Function, SCellular Component and KEEG Pathways. The combined score was computed by combining the probabilities from the different evidence channels and corrected for the probability of randomly observing an interaction (34).

Categorical variables are presented as whole numbers and percentages, while continuous variables are described using the median and interquartile range. The χ² test was used to evaluate the association between pathological response and TILs. Descriptive and association analysis was performed using GraphPad Prism 6 for Windows version 6.01 software (GraphPad Software Inc.). Due to the small sample size and lack of power, specific association methods (exlogistic and firthlogit) were used.

A total of 96 tumor samples were prospectively collected at the Breast Cancer Unit of Hospital San José TecSalud between August 2011 and August 2015. All patients were female. A total of 26/96 samples were diagnosed as TNBC. These 26 patients with TNBC comprised the study population.

The median age was 49-years old (range, 43–56 years); 80% of the tumors were poorly differentiated (grade 3) and 20% moderately differentiated (grade 2). Lymphovascular invasion was present in 18 (73%) patients and ipsilateral lymph node metastasis in 8 (30.7%) patients. Only 1 (3.4%) patient had metastatic disease upon presentation.

Tumor staging according to the AJCC revealed that 5 patients had (19.2%) stage I, 15 (57.6%) stage II, 4 (15.3%) stage III and 2 (7.6%) stage IV disease. All patients received standard NAC. The clinicopathological characteristics of the cohort are summarized in Table III.

The pathological response to NAC was assessed, and it was identified that 4 (15.3%) patients had pNR, 16 (61.5%) pPR and 6 (23%) pCR. TIL assessment demonstrated low, intermediate, and high TIL counts in 8 (30.7%), 10 (38.4%), and 8 (30.7%) patients, respectively. The CD3 count was high in 22 (85.6%) and low in 4 (15.3%) of the specimens. Notably, the CD4 count was low in 22 (85.6%) and high in 4 (15.3%) specimens. The CD8 count was low in 21 (80.7%) and high in 5 (19.2%) samples. The CD counts determined by microscopy were associated with the number of lymphoid cells; CD3 showed a higher TIL count of lymphoid cells (>25 lymphoid cells), meanwhile CD4 and CD8 were associated with a low TIL count (<25 lymphoid cells). With regard to pCR cases, 50% had a high number of TILs, 16% an intermediate number, and 33% a low number. With regard to the patients with pPR, 25% had a high number of TILs, 50% an intermediate number and 25% a low number. Finally, none of the patients with pNR exhibited a high number of TILs, while 25% of patients with pNR had an intermediate number and 75% pNR a low number (Table IV).

Using immunohistochemistry, CD3 and CD4 counts were significantly associated with pPR and pCR (P=0.04), but CD8 was not associated with pathological response (P=0.75) (Table V). On the contrary, microarray analysis presented in a heat map demonstrated that CD3, CD4 and CD8 were significantly associated to pathological response (P<0.05; Fig. 2). This difference may be due to the heat map being an average of a global analysis that includes all pathological responses (pCR, pPR, and pNR) (35). In addition, microarray analysis determines the RNA messenger protein expression whereas IHC only analyzes the protein expression (36). In addition, most pCR cases had a high CD3 count (83.3%). No additional associations between TILs and other clinicopathological parameters were identified in the study cohort.

Gene expression analysis of selected genes [CD3, CD4. CD8, CD20, CD45, forkhead box P3 (FOXP3), interleukin 6, programmed cell death 1 and CD274 molecule] showed that CD3, CD4, CD8, CD45 and CD20 exhibited a high expression in the patients with pPR and pCR, whereas a low expression of the aforementioned genes was observed in patients with pNR (Fig. 2).

GO analysis showed that the predicted target genes exhibited a stronger interaction among CD68, IL6, FOXP3, PTPRC, CD274, CD4, CD8A, PDCD1, CD3G, CD3D, CD247 and CD3E, a weaker interaction with KRT20, MS4A1 and CD3EAP, and no interaction with SPATA2 and SNCA. A total of 14 genes were involved in immune response, 7 in T-cell receptor signaling pathway, 9 in lymphocyte activation, 8 in T-cell activation, and 7 in T-cell differentiation (background gene, 131–1560; FDR, 2.90E-9-3.07E-10; Fig. 3A). Simultaneously, the KEGG database demonstrated 8 genes that were involved in the T-cell receptor signaling pathway, 7 genes involved in hematopoietic cell lineage and TH17 cell differentiation, and 5 genes involved in primary immunodeficiency and Th1/Th2 cell differentiation (background gene 88–99; FDR, 1.33 E-7-7.39E-13; Fig. 3B).