Evaluation of invasive breast cancer samples using a 12-chemokine gene expression score: correlation with clinical outcomes

A unique 12-chemokine gene expression score (CS) accurately predicted the presence of tumor-localized, ectopic lymph node-like structures (TL-ELNs) and improved overall survival (OS) in primary colorectal cancer and metastatic melanoma. We analyzed the correlation between CS, clinicopathological variables, molecular data, and 366 survival in Moffitt Cancer Center’s Total Cancer Care (TCC) patients with non-metastatic breast cancer. Affymetrix gene expression profiles were used to interrogate the CS by the principal component method. Breast tumors were classified as high or low score based on median split, and correlations between clinicopathologic variables, PAM50 molecular subtype, and ELN formation were analyzed using the TCC dataset. Differences in overall survival (OS) and recurrence-free survival (RFS) in the larger KM Plot breast cancer public datasets were compared using Kaplan-Meier curves. We divided the Total Cancer Care (TCC) breast cancer patients into two groups of high or low CS. Mean CS was 0.24 (range, 2.2–2.1). Patients with higher CS were more likely to be white (172 vs. 159; p = 0.03), had poorly differentiated tumors (112 vs. 59; p <0.0001), ER/PR negative (41 vs. 26) and HER2 positive (36 vs. 19; p = 0.001), and contain TL-ELNs. Higher CS scores were also seen in the basal and HER2+ molecular subtypes. In the KM Plot breast cancer datasets higher CS patients demonstrated superior OS (HR = 0.73, p = 0.008) and RFS (HR 0.76, p = <0.0001), especially in basal and HER2+ patients. High CS breast tumors tend to be higher grade, basal or HER2+, and present more frequently in Caucasians. However, this group of patients also shows the presence of TL-ELNs within the tumor microenvironment and has better survival outcomes. The CS is a novel tool that can identify breast cancer patients with tumors of a unique intratumoral immune composition and better prognosis. Whether or not the CS is a predictive response marker in breast cancer patients undergoing immunotherapy remains to be determined.


Background
Breast cancer represents 14.0% of all new cancer cases in the United States, with 231,840 new cases and an estimated 40,290 deaths in 2015, comprising 6.8% of all cancer deaths [1]. Local and systemic treatments including surgery, chemotherapy, radiation, and endocrine therapy have all improved outcomes significantly for breast cancer patients [2]. However, the number of patients relapsing despite these treatments requires development of novel treatment modalities. One such modality that is garnering more attention recently is the use of immunotherapies [3]. Understanding how the various types of breast cancer interact with the immune system is important in informing us how to effectively utilize promising immune-oncology agents.
Although breast cancer is not perceived as a particularly immunogenic tumor when compared with melanoma and renal cell carcinoma as examples, molecular profiling of breast tumors has revealed that a subset demonstrate a high level of immunoregulatory gene activation [4]. Multiple investigators have reported that tumor-infiltrating lymphocytes and certain gene expression profiles related to immune signaling appear to have prognostic and/or predictive implications for breast cancer [especially the human epithelial growth factor receptor 2-positive (HER2+) type] [5][6][7][8][9]. These studies highlight the potential importance of the immune response in breast cancer patient outcomes. However, there are distinct types of immune cell infiltrates that can have different effects on tumor behavior. Characterization of the underlying mechanisms regulating immune infiltration in breast tumors can elucidate the key determinants for a successful host anti-tumor immune response. Secretion of chemokines within the tumor microenvironment and how certain co-morbidities like diabetes can affect the tumor chemokine milieu have gained attention as important factors that shape tumor lymphocyte infiltration [10,11].
Chemokines act as trafficking signals for various immune cells and are important in orchestrating the spatial distribution of the immune response in a host. They also can directly affect the growth and progression of cancer cells [12]. Utilizing gene expression profiles can provide a more global assessment of immune signaling and cell populations using in silico methods such as CIBERSORT [13]. Certain chemokines have been associated with formation of a specific type of well-organized immune infiltrate known as tumor-localized, ectopic lymph nodelike structures (TL-ELNs) [14]. It is hypothesized that these ELNs represent potent chemokine signaling gradients in the tumor microenvironment that attracts not only T cells but also activated B cells responding to specific tumor-associated antigens presented by co-localized dendritic cells [14]. Coppola and associates identified a unique 12-chemokine  (CCL2, CCL3, CCL4, CCL5, CCL8, CCL18, CCL19,  CCL21, CXCL9, CXCL10, CXCL11, and CXCL13) gene expression signature (GES) from a metagene grouping with overwhelming enrichment for immune-related and inflammation-related genes in primary colorectal cancer [15]. Messina and associates subsequently interrogated the 12-chemokine GES score (CS) across genomic arrays of 14,492 distinct solid tumors (primary and metastatic) of different histologies using the Total Cancer Care (TCC) database [16]. They found that this CS accurately predicted the presence of TL-ELNs and showed an association with improved overall survival in stage IV melanoma.
Since little was known about the effect of these aforementioned 12 chemokines on the breast tumor microenvironment, we sought to explore the relationship between the CS, presence of TL-ELNs, molecular subtype, and patient outcome in annotated breast cancer samples.

Patient inclusion criteria
A complete description of the TCC biobanking program has been previously published [17]. A retrospective review was performed on selected female patients with stage I to III breast cancer who were diagnosed between 1988 and 2012 and received primary surgery at the Moffitt Cancer Center. Patients may have received adjuvant therapies at Moffitt or at other locations. Snapfrozen tumor specimens from initially resected primary breast tumors were used for the gene expression profiles. Of the 813 unique gene expression files available, we included those from breast primary tumors only for which full clinical information was available. Patients chosen for study had available clinical and follow-up data within Moffitt Cancer Center's electronic medical record system along with a genomic expression profile of their primary tumor. We excluded patients who had received any form of neoadjuvant therapy and with de novo metastatic disease, resulting in 366 patients in total.

Pathologic analysis of tissue sections
Histological sections corresponding to 28 cases (prepared from the mirror image of the portion of tumor submitted for the mRNA microarray analysis) were retrieved from the Moffitt Cancer Center Anatomic Pathology Division's repository as a pilot analysis to study correlation between the CS and tumor-localized, ectopic lymphoid node-like structures (TL-ELNs). Half of the specimens were from the top 10 th percentile 12-chemokine gene expression scores and half were from the bottom 10 th percentile. All of the specimens were 10% formalin fixed and paraffin embedded. Random representative hematoxylin and eosin (H&E)-stained sections through all selected tissue blocks were evaluated for the presence or absence of TL-ELNs. To further characterize the TL-ELNs, tissue sections were stained using the avidin-biotin complex method with retrieval under high pH. Pre-diluted monoclonal antibodies to CD3, CD4, CD8, and CD20 (Ventana Medical Systems, Tucson, AZ, USA) were used for the manual morphometric analysis of TL-ELNs by brightfield microscopy. To ensure pathologic concordance, two pathologists at our institution reviewed the tissue sections. Scores of 0 to 3 were assigned based on the following features: 0 = no lymphoid infiltrate noted in slide, 1 = 1 group lymphoid infiltrate, 2 = 2 groups of lymphoid infiltrate, and 3 = 3 or more groups of lymphoid infiltrate. Both pathologists were blinded as to the 12-chemokine gene expression scores of the individual samples. Clinical information was accessible only to the principal investigator and authorized collaborators, and all samples were anonymously coded before analysis. The Fisher's exact test was used to test the association of 12-chemokine expression scores and H&E staining scores. The McNemar test was used to analyze the strength of agreement between the scoring methods of the two pathologists.

Patient variables and outcomes analyses
We compared clinical and pathological factors of patients with low versus high 12-chemokine gene expression scores calculated by principal component analysis (determined by median split). Correlation between clinicopathologic factors and the 12-chemokine gene expression score was tested using the chi-square test with the exact method using Monte Carlo estimation. Kaplan-Meier curves were created for both overall survival (OS) and recurrence-free survival (RFS), and logrank tests were used to compare 12-chemokine gene expression scores and 12-chemokine gene expression scores stratified by antibody status [estrogen receptor (ER) or progesterone receptor (PR) positive], ER and PR negative (with HER2 negative or missing), and HER2 positive. Multivariable survival models were fit using Cox proportional hazards model. Final models were chosen using backward selection, with a removal alpha of 0.05. All p values were two-sided unless otherwise stated and considered statistically significant at the 0.05 level. The final multivariate survival model incorporated age, pathologic stage, and ER status based on this criteria. All statistical analyses were performed using SAS (version 9.4; SAS Institute; Cary, NC, USA). A second log rank Kaplan Meier OS and RFS analysis was done on the larger KM Plot breast cancer dataset [20] due to the small number of events within the TCC cohort using the NED no evidence of disease, ER estrogen receptor, PR progesterone receptor, HER2 human epidermal growth factor receptor 2 A higher chemokine score was associated with Caucasian race, higher grade, ER-status, and HER2+ status

Results
Our study included 366 TCC patients who met our inclusion criteria, with 183 patients in the low and 183 patients in the high 12-chemokine gene expression score groups. Low and high 12-chemokine gene expression score groups were compared regarding patient demographics, tumor characteristics, treatment variables, and survival status.  Table 1). When we compared 12-chemokine gene expression score with PAM50 molecular subtype, higher score correlated more with basal and HER2-positive subtypes (Fig. 1a). Based on receptor status by immunohistochemistry, higher 12-chemokine gene expression scores were associated with triple-negative breast cancer (p = 0.0007) and HER2-positive tumors (p = 0.0002) (Fig. 1b).
The analysis of overall and recurrence-free survival in the KM Plot dataset demonstrated that patients with high 12-chemokine gene expression tumors had superior RFS (HR = 0.85, p = 0.018) and OS (HR = 0.63, p = <0.0012). The RFS was superior in patients with high 12-chemokine gene expression in the basal subtype (HR = 0.51, p = <0.0001), HER2 subtype (HR = 0.57, p = 0.0085), and luminal B subtype (HR 0.73, p = 0.0054). There was no RFS difference noted in the luminal A subtype (Fig. 2). A similar survival analysis was performed on the smaller TCC dataset, which showed a trend toward improved RFS in the HER2 subtype (Additional file 1). The number of OS and RFS events in the TCC dataset limited the power to fully evaluate the 12-chemokine gene expression score in relation to outcomes and was primarily used to evaluate associations between the 12-chemokine gene expression scores and clinicopathologic variables.
Of the 67 H&E tumor tissue slides analyzed histologically for immune cell infiltration, 28 were scored as 0 with absence of TL-ELNs and 39 were positive for TL-ELNs. Both the 12-chemokine gene expression score and immune cell staining score were associated with each other for both of the pathologists' scores (p < 0.001). There were no significant differences between these two scoring methods (p = 0.052), and kappa strength of agreement of 0.6148 indicated substantial strength of    (Tables 2  and 3). Immune cell infiltrate scores of 0 were noted in 31/34 (88.6%) and 28/28 (100%) of slides corresponding to low 12-chemokine gene expression score by each individual pathologist, respectively. Between 71 and 75% of tumors with high 12-chemokine gene expression scores were scored 1-3 for TL-ELNs on random sections evaluated for each tumor (Fig. 3). The immunohistochemistry stains of the TL-ELNs demonstrated perifollicular presence of CD3+ CD4+ and CD3+ CD8+ T cells with strong staining for CD20 centrally showing clustering of mature B cells (Fig. 4).  Gene expression levels of BTLA, CD274, CD69, CTLA-4, granzyme B, IDO, interferon gamma, IL10, IL2, IL6, LAG3, PD-1, PRF1, STAT1, LIGHT were all significantly higher in the 12-chemokine gene expression high group ( Table 4). The enrichment of these genes indicates that the 12-chemokine gene expression score also identifies tumors with higher levels of an activated Th1-skewed cytotoxic T cell infiltrate.

Discussion
The increasing awareness surrounding the importance of the host immune response in determining breast cancer outcomes provides new opportunities to integrate this information into treatment algorithms. Efforts to systematically describe the immune response in breast cancer by entities such as the TIL working group are critical to implementing this new system in the clinic [21]. However, given the complexity of the immune response and the need to personalize immunotherapy, it is becoming prudent to use molecular markers to dissect out what immune regulatory pathways are active in a given patient's tumor [22]. The data presented herein indicate that certain chemokine genes can identify breast tumors enriched for tumor-localized, ectopic lymph node-like structures, and potentially provide a causal mechanism for why the tumor is inflamed in this manner.
Our study demonstrates that a 12-chemokine gene expression signature can identify a group of breast cancers with more favorable long-term outcomes. This is despite the fact that this group also contains greater number of tumors with traditionally adverse pathologic factors such as higher grade, ER negativity, and HER2 overexpression. In contrast to other immune infiltrate scoring methods, the chemokine score can provide a mechanistic explanation for why a particular tumor is forming TL-ELNs and exhibiting higher levels of activated T cell infiltrates. Another advantage of this approach is that chemokine scores can be obtained from limited core biopsies (i.e., prior to neoadjuvant therapy) while whole tissue sections would be required to histologically evaluate for the presence/absence of TL-ELNs in a tumor. An important question is what tumorspecific molecular features are conducive to the emergence of the high chemokine score phenotype. Future analyses should focus on analyzing other datasets combining RNA sequencing data that can provide information on mutational load and specific mutations or epitopes associated with a high chemokine gene expression score.
In our study, the TL-ELNs have the appearance of typical peripheral lymph nodes and are constructed of the necessary immune components, with CD3+, CD4+, and CD8+ T cells appearing in the parafollicular cortex or marginal zones and with some dispersion into the follicle and CD20+ B cells concentrated in the center of the follicle. The formation of these TL-ELNs is likely a different process compared to the lesser organized, dispersed infiltration of stromal tumor-infiltrating lymphocytes. Approximately 20% of invasive breast cancers Bolded gene probes are those with false discovery rate of <1% across all representative probes for a particular gene contain perivascular TL-ELNs [20]. In particular, these infiltrates were associated with medullary breast cancers in one analysis, possibly accounting, in part, for its favorable prognosis [22]. Our study sheds light on the role of chemokine gene signaling in the tumor microenvironment and the formation of TL-ELNs, which potentially provides novel therapeutic opportunities. These may include manipulating TL-ELN-negative tumors to become TL-ELNpositive ones or isolating antibodies from the reactive B cell clones resident within TL-ELNs that potentially target tumor-associated antigens. Investigation of tumor chemokine gene expression scores in groups of breast cancer patients treated with checkpoint inhibitors and comparing its association with programmed death ligand 1 staining and clinical response is another possibility. In this respect, the chemokine score may prove useful to select patients for checkpoint blockade therapy. For chemokine-scorelow breast tumors, increasing levels of key chemokines may ultimately prime those patients to respond more effectively to subsequent immunotherapies. Using immune gene expression signatures to personalize immunotherapy approaches could be critical in the future to maximizing clinical benefit in breast cancer patients.

Conclusions
The 12-gene chemokine score evaluated in our study was associated with ectopic lymph node formation in breast tumors, increased gene expression of immune signaling pathways, and improved outcomes. The chemokine score should be further explored as a prognostic factor and predictive marker for emerging immunotherapy approaches in breast cancer patients.