The tumor microenvironment (TME) consists of dynamic interactions between extra cellular matrix (ECM), cancer-associated fibroblast (CAFs) and tumor-associated immune cells (TAICs). These components influence and condition tumor metabolism (1), tumor progression (2), and treatment response (3). Recent advances in immunotherapy have led to a growing interest in studying and targeting the tumor microenvironment in order to achieve effective clinical treatments in cancer patients. Notable mentions include examples of immune checkpoint blockade therapies targeting PD-1, PDL-1 interaction (pembrolizumab and nivolumab), CTLA4 blockade (ipilimumab), etc., which aid in tumor evasion from the immune response. In addition, recent studies showcasing immune cell densities, spatial interactions, single-cell profiling results in different tumors, added to the importance of evidence-based translational research. Novel technologies like multiplex immunofluorescence and other high plex tissue profiling assays identify the composition of tumor microenvironment, calculate densities of the different cells present in the tumor microenvironment, identify their topographical distribution and correlates the TME and tumor cells, enabling a comprehensive profiling of the tumor immune landscape. Cornerstone advances, revealing distinct immunologic phenotypes led to the identification of mechanisms of resistance and potential targets to aid in cancer therapy discoveries.

Breast cancer is a leading cause of cancer incidence and deaths worldwide (4). Despite Invasive Lobular Carcinoma (ILC) being the second most common subtype of breast cancer with a distinct biology, most of the studies researching the tumor microenvironment have been done in invasive ductal carcinomas(IDCs) (5–7). ILCs have distinct histological subtypes with the classical subtype being the most common and having a quiescent nature whereas the pleomorphic variant -constituting approximately five percent of total ILC cases-is known to be an aggressive variant (8).

Desmoplasia in tumor tissue is a characteristic feature and stromal fibroblasts and cancer associated fibroblasts (CAFs) are a part of this reaction that form due to the interplay between cancer cells and the surrounding milieu (9–11). While the CAFs resemble normal residing fibroblasts or myofibroblasts, they have a diverse role characterized by increased collagen formation and a greater ECM protein production along with pro-tumor factors that support tumor growth and progression (12–15). Desmoplasia in tumors is caused by different mechanisms like cross-linking of collagens, fiber elongation and realignment (16). CAFs secrete metalloproteinases which are instrumental in desmoplasia and collagen deposition. The interplay of collagen and CAFs might play an intricate role in tumorigenesis especially collagen I, III,V,VI (17). However, it is believed that the turnover or balance of these different collagen types drives the anti-tumorigenicity or pro-tumorigenicity and much remains to be explored regarding the intricacies of interaction of collagen subtypes and CAFs (17).

There are many probable hypothesis about origins of CAF’s, but it is still unclear how they originate (15, 18). Some of the plausible theories include (i) recruitment and activation of resident fibroblasts (19); (ii) epithelial-mesenchymal transition (EMT) of resident epithelial cells (20); (iii) endothelial to mesenchymal transition (EMT) of resident endothelial cells (21) and (iv) differentiation of bone marrow mesenchymal cells (22). A minority of studies have also shown an antitumorigenic capacity of CAFs. For e.g. PD-L1 expression in CAFs of non-small cell lung carcinoma tissues has been demonstrated to correlate with good prognosis, and depletion of CAFs in pancreatic cancer was associated with poor prognosis (23–26).

CAFs heterogeneity is based on phenotypic markers, gene expression profiling, and functionality (11, 27–29). Until now, no single pan-specific CAF marker has been identified and neither a unifying approach to defining, nor a standardized nomenclature for CAF subpopulations has yet emerged. CAF have been shown to express high levels of alpha-smooth muscle actin (α-SMA/Acta2) (30), PDGFR α or β (31), CD90 (Thy1) (32), fibroblast activation protein (FAP) (33), fibroblast-specific protein 1 (FSP/S100A4) (34), integrin β1/CD29 (28, 35), podoplanin (Pdpn) (36), osteonectin (Sparc) (37), caveolin 1 (Cav1) (35) and vimentin (32). Since CAFs are an essential component of the tumor environment, much interest has arisen to explore their utility as actionable targets in cancer treatment (38).

In the current study, we aimed to profile the CAFs in the tumor microenvironment of breast ILC by comparing two different subtypes (Classic ILC vs Pleomorphic ILC), and explore their spatial distribution concerning the malignant cells using an automated mIF panel image analysis approach.

Two different histological variants of invasive lobular carcinoma: Classic ILC (n=6) ( Figure 1A ) and Pleomorphic ILC (n=6) ( Figure 1B ) were compared using paraffin embedded (FFPE) samples from individuals diagnosed with non-metastasized classic ILC and pleomorphic ILC we analyzed the tumor microenvironment using multiplex immunofluorescence. Six of these cases were of classic ILC histology and remaining six were of pleomorphic ILC histology. All cases with classical features were nuclear grade 2 and all pleomorphic cases (n=6) were nuclear grade 3. Six were Luminal B, five were Luminal A and one case was triple negative invasive lobular carcinoma with pleomorphic histology.

To detect CAF subpopulations, we performed multiplex immunofluorescence to detect markers of hematopoietic cells (CD45), epithelial cells (pan-cytokeratin [CK]), and fibroblasts (α-smooth muscle actin [α-SMA], fibroblast activation protein (FAP), fibroblast specific protein 1 [FSP-1], and Thy-1) on surgically resected invasive lobular carcinoma of the breast ( Figure 2 ). We evaluated fibroblasts as cells that are negative for CD45 and CK and positive for at least 1 fibroblast marker (blood vessels were excluded from the region of interest to avoid discrepancies with pericytes positivity for alpha-SMA). Fibroblasts were quantified in epithelial (CK⁺) and stromal (CK^-) compartments of both diseases. As a proxy for heterogeneity, fibroblasts were classified based on marker expression patterns (single-, double-, triple-, or quadruple-marker⁺) and assigned to 1 of 19 subpopulations based on possible expression patterns of the 4 markers. We observed substantial variation in the subsets of CAFs types and magnitude of infiltration in our breast ILC cohorts, demonstrating the heterogeneity of potential immune responses to the two different tumor subtypes.

We found and evaluated nineteen different colocalized phenotypes of CAFs in the tumor, stroma and overall tissue compartments comparing the TME of classic ILC and Pleomorphic ILC. Out of nineteen phenotypes, we found statistically significant differences(p-value<0.05) between classic and pleomorphic ILC in five of the phenotypes, considering overall tissue compartments. The density (cells/mm²) of Total A-SMA+, A-SMA+FAP+S100+ and A-SMA+S100 CAFs were higher in classic ILC while densities of FAP+S100+ and S-100+ CAFs were higher in the pleomorphic subtype. In the tumoral compartment, although a statistically significant difference in the density of CAFs was found only in A-SMA+S100+ CAFs (being higher in classic ILC), an “approaching significance” or a trend was seen with Thy-1+, Total FAP+, FAP+ only and FAP+S100+ CAFs being higher in pleomorphic subtype. The results are summarized in Supplementary Table 3 and Figures 3A, B , 4A–D and Supplementary Figures 1–4 ).

Next, our study aimed to find if there was a difference in the proximity of CAF subtypes to the tumor between the two histologies. Nearest neighbor distances were calculated, and statistically significant differences (p-value<0.05) were observed between the distances of CAFs from the tumor cells in classic and pleomorphic ILC. The A-SMA+, A-SMA+/S100+ phenotypes were closer to the tumor cells in the classic subtype and the FSP1+ only phenotype was closer to the tumor cells in pleomorphic carcinomas. In addition, there were two other phenotypes namely A-SMA+/Thy-1+(closer to tumor cells in classic ILC) and FAP+/S-100+ (closer to tumor cells in pleomorphic ILC), which showed a trend of significance ( Supplementary Table 4 and Figures 5A, B , 6A, B ).

We also correlated CAFs differential expression to clinic-pathological variables in each of the tumor subtypes. We did not find any significant correlation of any of the CAF subtype to tumor size, nodal status or Ki67 index.

In this study, we investigated the cancer associated fibroblasts composition in two different subtypes of invasive lobular carcinoma using specially designed multiplex immunofluorescence panels. In this cohort, we had six cases each of classic ILC and Pleomorphic ILC, and we investigated the tumor neighborhood for CAF subtypes and found significant differences in the TME of these two subtypes.

The role of CAF in metastasis is well established, when analyzed as a global population (40) and they serve different functions in tumor biology such as proliferation, metastasis and drug resistance (41–43). CAFs are heterogenous and express different markers either individually or in co-expression and the composition also varies across different tumor types (44, 45)]. CAFs originate from various types of cells, including resident fibroblasts, bone marrow-derived ancestors, or epithelial carcinoma cells that undergo epithelial-mesenchymal transition (17, 46).

CAFs are heterogenous and so care has to be taken in selecting the markers to be used for particular tumor type. For instance, in a study by Ries et al (45) on Pleural mesotheliomas, FAP which is a CAF marker was also shown to be expressed from mesothelioma cells in addition to the fibroblasts. Hence, they propose that FAP is not a good marker to characterize CAFs in Pleural mesotheliomas. Previous studies by Costa et al. and Pelon et al. utilizing flow cytometry have used different CAF markers (SMA, CD29, FAP, FSP1, PDGFRβ, CAV1) profiling them into four different CAF subtypes (CAF-S1 to CAF-S4) across breast and ovarian cancers (CAF-S1: CD29^Med FAP^Hi FSP1^Low-Hi αSMA^Hi PDGFRβ^Med-Hi CAV1^Low; CAF-S2: CD29^Low FAP^Neg FSP1^Neg-Low αSMA^Neg PDGFRβ^Neg CAV1^Neg; CAF-S3: CD29^Med FAP^Neg FSP1^Med-Hi αSMA^Neg-Low PDGFRβ^Med CAV1^Neg-Low; CAF-S4: CD29^Hi FAP^Neg FSP1^Low-Med αSMA^Hi PDGFRβ^Low-Med CAV1^Neg-Low) (28, 47). According to their data, CAF-S1 and CAF-S4 are detected solely in tumors, the CAF-S2 and CAF-S3 subsets are expressed both in tumors and normal tissues. In addition, the studies show that CAF-S1 subtype, which have high Alpha SMA and FAP positivity have clearly demonstrated an immunosuppressive function in breast cancer and attract FOXP3 regulatory T-cells.

Our study which uses multiplex panels also used markers such as A-SMA and FAP used in the studies by Costa et al. and Pelon et al (28, 47), with the difference in two of the markers for CAF identification. Our comprehensive specialized CAF multiplex panel which used protein-based detection consisted of most commonly used markers for CAF: Alpha-SMA, FAP, S100 and Thy-1 with CD45 and Pan CK to identify hematopoietic cells and tumor cells respectively. Out of these Alpha-SMA, FAP, FSP1, Thy-1 have been established as CAF markers in other studies (48, 49). A similar multiplex panel to characterize CAFs have been used by Rimm et al (50) in metastatic melanomas with the absence of S100 since they used a 5-plex based panel. We carefully selected these particular markers due to the methodology in use, as the combination of antibodies in multiplex immunofluorescence should be selected with astute care and be best representative of the research question in case. Since our study allowed us for a 7 plex technique we inculcated S100 in our panel in addition to the panel used by Rimm et al (50) to make our panel more comprehensive and be best representative for breast carcinoma Cohort.

Similar to the categorizing technique used by Costa et al. (28), we identified nineteen different phenotypic combinations of CAFs based on colocalization of above markers. This approach helped us to extensively subtype CAFs which can lead to more comprehensive studies at the molecular level in the future. Whereas single cell techniques like cytometry do help in identifying the tumor microenvironment, the power of our study lies in profiling the tumor microenvironment in a better spatial context.

The raw data is confidential and can be obtained after appropriate request to the corresponding author and following appropriate institutional guidelines.

The studies involving humans were approved by MDACC Institutional Review Board. The studies were conducted in accordance with the local legislation and institutional requirements. The human samples used in this study were acquired from a by- product of routine care or industry. Written informed consent for participation was not required from the participants or the participants’ legal guardians/next of kin in accordance with the national legislation and institutional requirements. Written informed consent was obtained from the individual(s) for the publication of any potentially identifiable images or data included in this article.

HB: Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Software, Validation, Visualization, Writing – original draft, Writing – review & editing. QD: Resources, Supervision, Validation, Writing – review & editing. RP: Investigation, Methodology, Software, Writing – review & editing. HI: Methodology, Writing – review & editing. NR: Writing – review & editing. AS: Supervision, Visualization, Writing – review & editing. IW: Funding acquisition, Project administration, Resources, Supervision, Writing – review & editing. EP: Methodology, Supervision, Writing – review & editing. MR: Conceptualization, Funding acquisition, Methodology, Project administration, Resources, Supervision, Validation, Visualization, Writing – original draft, Writing – review & editing.