Prognostic Value of “Basal-like” Morphology, Tumor-Infiltrating Lymphocytes and Multi-MAGE-A Expression in Triple-Negative Breast Cancer

“Basal-like” (BL) morphology and the expression of cancer testis antigens (CTA) in breast cancer still have unclear prognostic significance. The aim of our research was to explore correlations of the morphological characteristics and tumor microenvironment in triple-negative breast carcinomas (TNBCs) with multi-MAGE-A CTA expression and to determine their prognostic significance. Clinical records of breast cancer patients who underwent surgery between January 2017 and December 2018 in four major Croatian clinical centers were analyzed. A total of 97 non-metastatic TNBCs with available tissue samples and treatment information were identified. Cancer tissue sections were additionally stained with programmed death-ligand 1 (PD-L1) Ventana (SP142) and multi-MAGE-A (mAb 57B). BL morphology was detected in 47 (49%) TNBCs and was associated with a higher Ki-67 proliferation index and histologic grade. Expression of multi-MAGE-A was observed in 77 (79%) TNBCs and was significantly associated with BL morphology. Lymphocyte-predominant breast cancer (LPBC) status was detected in 11 cases (11.3%) and significantly correlated with the Ki-67 proliferation index, increased number of intratumoral lymphocytes (itTIL), and PD-L1 expression. No impact of BL morphology, multi-MAGE-A expression, histologic type, or LPBC status on disease-free survival was observed. Our data suggest that tumor morphology could help identify patients with potential benefits from CTA-targeting immunotherapy.


Introduction
Breast cancers that have no estrogen, progesterone, or human epidermal growth factor receptor 2 (HER2) receptor expression are called triple-negative breast carcinomas (TNBCs) [1,2].Compared to other subtypes, this group of tumors has almost no therapeutic options besides chemotherapy and consequently is characterized by poor prognosis [1][2][3][4][5][6].Due to the lack of expression of estrogen, progestrone, and HER2 receptors, specific targeted therapy is not effective and chemotherapy was, until recently, the only available systemic form of therapy [1,2,4,5,7].Previous studies have associated TNBC with a more aggressive

Discussion
Based on genetic profiling, most TNBCs belong to the "basal-like" subgroup [11].However, this concept's immunohistochemical, morphological, and clinical implications are still debated.While genetic profiling represents the golden standard in identifying "basal-like" intrinsic subgroups, the role of other, more conventional methods is unclear."Basal-like immunophenotype" can be determined by the expression of one or more basal cell markers, CK5/6, CK14, or EGFR, with a sensitivity of 78-86% and specificity of up to 100% [12,13].Morphological identification proved less reliable because of the large variety of histologic features associated with "basal-like" morphology, including syncytial growth pattern, high mitotic index, large central acellular/necrotic zone, pushing borders, dense lymphocytic infiltrate at the invasive front, and the presence of metaplastic and medullary elements.
Most TNBCs in our study (81.4%) were classified as NOS, while 18 (18.6%)tumors could be assigned to specific histologic types, which was consistent with the results of previous studies [41].We compared their clinicopathologic features and found that invasive NOS carcinomas had a significantly higher Ki-67 proliferation index, higher histologic grade, and higher accumulation of itTIL compared with special-type carcinomas.The prognosis of the special TNBC subtypes remains controversial [21,42,43], but we found no difference in DFS between them and the NOS carcinomas.
In agreement with previous reports [44], in our study, "basal-like" morphology was observed in 49% of TNBCs, and specific morphological characteristics were more often detectable in TNBCs of NOS histologic subtype with higher Ki-67 proliferation rate and high histologic grade [3,45].In contrast, we found no correlation between age, LPBC status, and tumor size with BL morphological characteristics [12,[14][15][16][17][18][19].Most importantly, although BL breast cancers were reported to display aggressive clinical behaviour and were associated with worse prognosis [46], a statistically significant impact of "basal-like" morphology on DFS was not observed.
TNBC is characterized by higher levels of TIL compared to other breast cancer subtypes [47][48][49], in particular, by stromal lymphocyte infiltration that is easily reproducible [26,50,51] and is commonly used in everyday practice.In addition to stromal lymphocytes, intratumoral lymphocytes (it-TIL), which are in direct contact with tumor cells, and the formation of lymphoid aggregates (LA) at the tumor margin or inside the tumor were also investigated, which were divided into two categories as primary and secondary LA, depending on the presence of germinal centers.Although a correlation of itTIL and LA with high tumor grade has been reported in previous studies, we found no correlation with the evaluated clinicopathological parameters [52,53].
In agreement with the previous studies, tumors in which ≥50% of the stromal compartment was infiltrated by lymphocytes were classified as LPBC [26,27,29].In our study, the proportion of LPBCs to TNBCs was 11%, which is within the currently reported ranges [54].We found that LPBC tumors had a higher itTIL accumulation [26], but also a higher expression of PD-L1.
Cancer cells use various strategies to evade the immune response to the tumor.The PD-L1 protein overexpressed on tumor cells binds to the programmed cell death protein 1 (PD-1) receptors on T lymphocytes [55].Patients with PD-L1-positive TNBCs are therefore excellent candidates for therapy with immune "checkpoint" inhibitors, which restore an adequate antitumor response by inhibiting PD-L1-to-PD-1 binding [56,57].In addition, the LPBC tumors in our series exhibited a higher Ki-67 proliferation index, as previously reported [56].However, in contrast to previous studies [58][59][60][61], we did not observe a significant impact of LPBC status on DFS, which is likely due to the relatively small size of our study cohort and the correspondingly low number of LPBCs.
It has been proposed that CTA favours tumorigenesis by regulating cancer cell proliferation, apoptosis, invasiveness, and metastatic properties, possibly by promoting epithelialmesenchymal transition (EMT) [62].EMT is widely accepted as an important milestone in cancer progression, but morphological changes that occur in cancer cells, and especially tumor microenvironment are still poorly understood because genetic and phenotypical changes do not always correlate as expected [63].In contrast with previous reports [32,34,64], in our study, a significant correlation between multi-MAGE-A protein expression and BL morphology was found, with 89% of TNBC with BL morphology being multi-MAGE-A positive.To the best of our knowledge, this is the first study that confirmed this correlation.The association of BL morphology with characteristic tumor microenvironment and CTA expression supports the growing evidence for the immunogenic properties of CTA in TNBCs and may offer new insights to cancer research if given the time and attention it deserves.
MAGE-A genes are frequently expressed in BCs and are considered essential future immunotherapy targets [65].However, we did not find any significant correlation between multi-MAGE-A expression and other clinicopathological parameters, including larger tumor size, recurrence, and poor overall survival [62].
The retrospective design of our study and the relatively small size of our study cohort are the main limitations of our study.The correlation between "basal-like" morphology and multi-MAGE-A expression may provide an easily reproducible option for selecting patients suitable for immunotherapy, but these results need to be confirmed in future larger, prospective studies.

Patients
The medical data of primary surgically treated breast cancer patients in four large Croatian clinical centers from January 2017 to December 2018 were evaluated.A total of 124 non-metastatic TNBCs were identified.In addition to the non-metastatic stage, additional exclusion criteria were non-neoadjuvant treatment, availability of tissue samples, and oncologic follow-up of less than 24 months at one of the institutions involved in the study.Of the patients with available tissue samples, 10 chose to be treated at another institution and 17 did not meet the required criteria of 24 months of regular follow-up as they either missed regular appointments or died of unspecified diseases.Complete follow-up data up to 1 January 2021 were obtained for 81 patients, with a mean duration of 43.3 months.
Disease-free survival (DFS) was calculated as the time between the first surgical procedure and until diagnosis of recurrence or distant metastases.Patients underwent a mastectomy, quadrantectomy, or tumorectomy (47.4%, 49.5%, and 3.1%, respectively), and 96 of them underwent a dissection of the axillary lymph nodes or sentinel lymph node biopsy.All patients who had undergone conservative surgery were treated with postoperative radiotherapy.Adjuvant chemotherapy was administered to 90 patients and omitted in 7 patients due to comorbidities or patients' choice not to be treated with chemotherapy.
All histologic evaluations were performed independently by two researchers (I.M., T. Č.).Grading of tumors was conducted according to Elston and Ellis [66], histologic tumor type was determined concordantly with the World Health Organization (WHO) tumor classification, and disease staging with the TNM Classification of Malignant Tumors [20,67].

Institutional Review Board Statement
The study was designed and written to the ethical standards of the institutional and national research committee and the 1964 Helsinki Declaration, as well as its later amendments or comparable ethical standards.The Institutional Review Board of the University Hospital of Split approved the study under the approval number: 2181-147/01/06/M.S-22-03; date of approval: 9 May 2022.

Outcomes of the Study
The primary outcome of the study was a correlation of "basal-like" morphology and multi-MAGE-A expression and their prognostic significance in TNBCs.The secondary aims were the correlation of high TIL infiltration and histologic subtype of TNBC with other clinicopathological characteristics.
Because of tumor heterogeneity, the accumulation of TIL and expression of PD-L1 and multi-MAGE-A were assessed on whole tissue sections.Multi-MAGE-A was considered positive if a cytoplasmic and/or nuclear reaction was observed in ≥10% of tumor cells [34] (Figure 1).PD-L1 expression was evaluated independently by two pathologists (S.T. and I.M.) and considered positive if discernible PD-L1 staining of any intensity was observed in the tumor-infiltrating immune cells covering ≥1% of the tumor area and contiguous peritumoral stroma.TILs were assessed by two independent pathologists (I.M. and T.Č.) in concordance with the International TILs Working Group's [26] recommendations.Areas with necrosis, fibrosis, polymorphonuclear leukocytes, and technical artifacts were removed from the analysis.Stromal TIL accumulation (sTIL) was evaluated by determining the percentage (%) of stromal infiltration at the tumor margin and within the tumor, delineated as sTIL peripheral and sTIL central, respectively, and combined as sTIL total.Tumor-infiltrating lymphocytes in contact with tumor cells were classified as itTIL.In accordance with previous studies, tumors with ≥50% of the stromal compartment (sTIL total) infiltrated by lymphocytes were classified as LPBC [26,27,29].The existence of lymphoid aggregates (LA) at the tumor margin or inside of the tumor was noted, and LAs were divided into two categories, based on the existence of germinal centers.LAs without germinal centers were noted as primary LA, and those with germinal centers as secondary LAs.
Ki-67 proliferating index was evaluated by counting 1000 tumor cells at the hot spots and the periphery of the invasive component [71].
BL morphology was considered positive if characteristic features such as syncytial growth pattern, high mitotic index, large central acellular/necrotic zone, pushing borders, dense lymphocytic infiltrate at the invasive front, and the presence of metaplastic elements were present [12,[14][15][16][17][18][19] (Figure 2).TILs were assessed by two independent pathologists (I.M. and T. Č.) in concordance with the International TILs Working Group's [26] recommendations.Areas with necrosis, fibrosis, polymorphonuclear leukocytes, and technical artifacts were removed from the analysis.Stromal TIL accumulation (sTIL) was evaluated by determining the percentage (%) of stromal infiltration at the tumor margin and within the tumor, delineated as sTIL peripheral and sTIL central, respectively, and combined as sTIL total.Tumor-infiltrating lymphocytes in contact with tumor cells were classified as itTIL.In accordance with previous studies, tumors with ≥50% of the stromal compartment (sTIL total) infiltrated by lymphocytes were classified as LPBC [26,27,29].The existence of lymphoid aggregates (LA) at the tumor margin or inside of the tumor was noted, and LAs were divided into two categories, based on the existence of germinal centers.LAs without germinal centers were noted as primary LA, and those with germinal centers as secondary LAs.
Ki-67 proliferating index was evaluated by counting 1000 tumor cells at the hot spots and the periphery of the invasive component [71].

Statistical Analysis
Acquired data were analyzed using SPSS Statistics 20 (IBM, Armonk, New York, NY, USA).Median and interquartile range (IQR) were used to describe the distribution of quantitative data, whereas categorical data were described with absolute numbers and percentages.Statistical significance was set to p < 0.05, and all confidence intervals were given at 95% level.The statistical significance of differences in categorical, demographic, and clinicopathological characteristics was calculated using the chi-squared and log-rank tests.As the Shapiro-Wilk test indicated a statistically significant deviation from a normal distribution of all numeric variables, the median and interquartile range were also used.The Mann-Whitney U test analyzed the statistical significance of differences found in quantitative variables between two groups.lymphocytes in contact with tumor cells were classified as itTIL.In accordance with previous studies, tumors with ≥50% of the stromal compartment (sTIL total) infiltrated by lymphocytes were classified as LPBC [26,27,29].The existence of lymphoid aggregates (LA) at the tumor margin or inside of the tumor was noted, and LAs were divided into two categories, based on the existence of germinal centers.LAs without germinal centers were noted as primary LA, and those with germinal centers as secondary LAs.
Ki-67 proliferating index was evaluated by counting 1000 tumor cells at the hot spots and the periphery of the invasive component [71].

Conclusions
Expression of multi-MAGE-A CTA was observed in 77 (79%) TNBCs and was significantly associated with BL morphology.Our data suggest that tumor morphology could help identify patients with potential benefits from CTA-targeting immunotherapy.

13 Figure 1 .
Figure 1.Strong cytoplasmic immunohistochemical staining of multi-MAGE-A in tumor cells.Multi-MAGE-A expression was considered positive if ≥10% of the tumor cells showed cytoplasmic and/or nuclear positivity.

Figure 1 .
Figure 1.Strong cytoplasmic immunohistochemical staining of multi-MAGE-A in tumor cells.Multi-MAGE-A expression was considered positive if ≥10% of the tumor cells showed cytoplasmic and/or nuclear positivity.

Figure 2 .
Figure 2. TNBC with "basal-like" morphology.At the periphery, the viable tumor tissue with pushing borders and peritumoral lymphocitic infiltrate can be seen, while the central part of the tumor is replaced by extensive fibrosis and necrosis.

Table 1 .
Correlation of clinicopathologic characteristics with LPBC status.
* Only one patient had a tumor with histological grade 1 and was excluded from the analysis; ** One paraffin block was worn out before the multi-MAGE-A IHC slides were prepared.Abbreviations: LPBC-lymphocytepredominant breast cancer; IQR-interquartile range; NOS-not otherwise specified; TIL-tumor-infiltrating lymphocytes; BL-"basal-like"; MAGE-A-melanoma antigen gene A; PD-L1-programmed death-ligand 1.

Table 4 .
Correlation of multi-MAGE-A expression with clinicopathological characteristics.
* Only one patient had a tumor with histological grade 1 and was excluded from the analysis; one paraffin block was worn out before the multi-MAGE-A IHC slides were prepared.MAGE-A-melanoma antigen gene A; IQR-interquartile range; NOS-not otherwise specified; TIL-tumor-infiltrating lymphocytes; LPBC-lymphocytepredominant breast cancer.

Table 5 .
Log-rank test and Cox regression univariate analyses for DFS in correlation with main variables studied in 97 TNBCs.