Stromal ColXα1 expression correlates with tumor-infiltrating lymphocytes and predicts adjuvant therapy outcome in ER-positive/HER2-positive breast cancer

Background The breast cancer microenvironment contributes to tumor progression and response to chemotherapy. Previously, we reported that increased stromal Type X collagen α1 (ColXα1) and low TILs correlated with poor pathologic response to neoadjuvant therapy in estrogen receptor and HER2-positive (ER+/HER2+) breast cancer. Here, we investigate the relationship of ColXα1 and long-term outcome of ER+/HER2+ breast cancer patients in an adjuvant setting. Methods A total of 164 cases with at least 5-year follow-up were included. Immunohistochemistry for ColXα1 was performed on whole tumor sections. Associations between ColXα1expression, clinical pathological features, and outcomes were analyzed. Results ColXα1 expression was directly proportional to the amount of tumor associated stroma (p = 0.024) and inversely proportional to TILs. Increased ColXα1 was significantly associated with shorter disease free survival and overall survival by univariate analysis. In multivariate analysis, OS was lower in ColXα1 expressing (HR = 2.1; 95% CI = 1.2–3.9) tumors of older patients (> = 58 years) (HR = 5.3; 95% CI = 1.7–17) with higher stage (HR = 2.6; 95% CI = 1.3–5.2). Similarly, DFS was lower in ColXα1 expressing (HR = 1.8; 95% CI = 1.6–5.7) tumors of older patients (HR = 3.2; 95% CI = 1.3–7.8) with higher stage (HR = 2.7; 95% CI = 1.6–5.7) and low TILs. In low PR+ tumors, higher ColXα1 expression was associated with poorer prognosis. Conclusion ColXα1 expression is associated with poor disease free survival and overall survival in ER+/HER2+ breast cancer. This study provides further support for the prognostic utility of ColXα1 as a breast cancer associated stromal factor that predicts response to chemotherapy. Electronic supplementary material The online version of this article (10.1186/s12885-019-6134-y) contains supplementary material, which is available to authorized users.


Background
Breast cancer is the second leading cause of cancerrelated death in women. HER2 targeting therapies, such as trastuzumab and pertuzumab, prolong survival in breast cancer patients [1,2]. HER2-positive breast cancers constitute a heterogeneous disease at the molecular level. The repertoire of somatic genetic alterations in these tumors varies according to estrogen receptor (ER) status and correlates with "intrinsic" subtype [3][4][5][6]. Early studies focusing on breast cancer molecular subtypes showed that Luminal-B subtype has poorer outcome than luminal-A subtype. In fact, the overall survival in untreated luminal-B subtype is similar to basal-like and HER2-positive tumors, which are widely recognized as high risk [3]. The treatment outcome for patients with ER+/HER2+ cancer, sometimes referred to as Luminal HER2 type, is variable [7][8][9]. Therefore, additional biomarkers are needed for risk stratification and treatment response prediction. We previously reported that increased stromal Type X collagen α1 (ColXα1) expression in ER+/HER2+ tumors is associated with poor response to neoadjuvant therapy. Our findings provided initial evidence that a specific stromal collagen subtype in the breast tumor microenvironment predicts therapy response [10].
Tumor associated stroma is composed of a matrix of fibronectin, matrix metalloproteinases, collagens and other connective tissue proteins which undergo constant remodeling [11]. Tumor-stromal interactions play a key role in tumorigenesis and metastasis influencing the prognosis of several human malignancies [12][13][14]. Some studies have reported that tumor microenvironment components like tumor infiltrating lymphocytes (TILs) and tumor-associated stroma predict response to therapy and breast cancer progression [15]. In this study we aimed to evaluate the prognostic and predictive value of ColXα1 expression in the tumor stroma of ER+/HER2+ breast tumors in the adjuvant setting.

Patients and tissue samples
The study was approved by the ethics committees of Lifespan Medical Center (467617-9) and Women Infants Hospital (797108-3). The need for consent was waived by the IRB.
A retrospective search was performed in the cancer registry database for breast cancer patients who received adjuvant therapy at Lifespan Medical Center and Women and Infant Hospital in Rhode Island between 2007 and 2013. We identified 164 ER+/HER2+ cases who received adjuvant chemotherapy and HER2targeted therapy. All original tumor slides were reviewed and histological features were recorded. Immunohistochemistry for ER, PR, and HER2 expression were classified according to the CAP/ASCO guidelines [16,17]. ER and PR positive cases were further classified as lowpositive (1-10%) and positive (> 10%).

Tumor-associated stroma and TILs analysis
We morphologically evaluated the amount of intratumoral stroma and TILs on tumor samples. The stromal evaluation protocol has been previously published [10]. Briefly, the amount of tumor-associated stroma was scored as 0 to 2: 0 for absent or minimal stroma (< 10%), 1 for mild to moderate amount of stroma (10-40%) and 2 for abundant stroma (> 40%). The TILs were evaluated based on criteria published by Denkert et al. [15]. Briefly, iTILs are defined as lymphocytes in direct contact with the tumor cells, whereas sTILs are defined as lymphocytes in the surrounding stroma with the percent of the tumor or stromal volume comprised of infiltrating lymphocytes. The results were evaluated in increments of 10 (0-1% was scored as 0, with all other estimates rounded up to the next highest decile [for e.g. 11-20% was scored as 20]). sTILs and iTILs were added to calculate TILs. The trends were similar for each lymphocyte fraction (data not shown). We chose to analyze sTILs as they are considered as the most consistent metric (as recommended by the International TILs Working Group) [18]. While analyzing survival, the sTILs data was divided into three groups. Group 1: Rare TILs in the stroma (0-10%). Group 2: low density TILs (11-59%) and Group 3: high density TILs (> = 60%).

Gene expression and pathway analysis
TCGA RNA-seq data for breast invasive carcinoma was downloaded from the Firehose Broad GDAC [19]. TCGA clinical data was downloaded from the TCGA data archive in September 2015 (http://cancergenome. nih.gov/).

Statistical analysis
Fisher Exact test and Chi-square test were used when appropriate. T-test or analysis of variance (ANOVA) was used to compare continuous variables. For time-to-event measures, the Kaplan-Meier method was used to estimate the empirical survival and log-rank estimates were used. Relationships between variables were assessed using Pearson and Spearman correlation analysis as noted. Multivariate analysis was performed using the Cox proportional hazards model. Statistical analysis was performed utilizing SPSS v25 (SPSS, Chicago, IL, USA).
A P-value < 0.05 was considered statistically significant. All P-values reported are two-sided.

Correlation of ColXa1 with other factors and TILs
Overall ColXα1 positivity was present in 143/164 (87%) cases. ColXα1 expression was directly proportional to the amount of stroma in the tumor (p = 0.024) and correlated with the presence of LVI (p = 0.008). There was a significant inverse relationship between TILs and ColXα1 expression by immunohistochemistry (r = − 0.47; p < 0.001, Table 2). A significantly higher proportion of tumors with rare and low density TILs were ColXα1 positive when compared to the high density TILs group (P < 0.001, Table 1

Correlation of ColXa1 and TILs and tumor associated immune cells
In the TCGA cohort, a similar trend of inverse relationship between the tumor immune microenvironment and ColXα1 expression by RNA-seq was noted (Table 2). There was an inverse relationship between ColXα1 and lymphocytes (r = − 0.33; p = 0.001), Naïve B cells (r = −

Discussion
The HER2+ breast cancers are a heterogeneous group with variable tumor aggressiveness and response to therapy [20,21]. HER2+ tumors can be further classified according to the hormone receptor status. Luminal-B subtype is defined as ER+ tumor with increased tumor cell proliferation which is usually assessed by KI-67 [22]. The ER+/HER2+ tumors are also classified as luminal-B subtypes, and sometimes referred to as Luminal HER2 subtype [6]. In our cohort of Luminal B HER2 subtype, most tumors (95%) had high KI-67 expression (> = 20%).  Although some studies report better outcome in Luminal HER2 breast cancer when compared to ERnegative HER2-positive breast cancer [20][21][22][23], others have found no significant difference in long term survival [24][25][26]. Metastatic sites and recurrence patterns of HER2+ tumors are ER dependent. Bone is the most common site of distant metastasis for Luminal HER2 subtype, which is similar to Luminal A cancers, whereas ER-negative HER2-positive cancers have the higher rates of locoregional recurrence and these tend to initially metastasize to visceral organs such as lung [23,26,27].
In ER+/HER2+ breast cancer, HER2 overexpression predicts poor response to hormone therapy [28,29]. Cross talk between HER2 and ER results in persistent activation of the hormone receptor downstream signaling, even in the presence of hormone treatment [30]. The tumor microenvironment consists of non-cellular elements such as collagens, glycosaminoglycans, proteoglycans and hyaluronic acid. Collagens are major components of the extracellular matrix of breast tumors. Collagen expression patterns have been linked with aggressive tumor behavior and drug resistance [31]. Together with hyaluronic acid, collagen accumulation is also associated with high tumoral interstitial pressure, vascular collapse and drug resistance [32,33]. Myriad collagen genes are critical in tissue development and physiological function, as evidenced by the range of collagenopathies [34]. Nevertheless, the mechanistic role of collagen subtypes in cancer progression has been largely overlooked, although they frequently emerge as components of tumor stromal expression signatures [35,36]. A variety of collagen subtypes are highly expressed in breast tumors contributing to its dense structure. The alignment of collagen fibers has been proposed to indicate progression in breast tumors [37,38]. ColXα1and its hexagonal network that plays a role in tissue stiffness has been associated with chemoresistance and breast cancer progression [39]. Collagen type X mRNA expression is up-regulated in a variety of human malignancies when compared to normal tissue, including breast tumors [40]. In colorectal cancer, Huang and colleagues demonstrated that ColXα1 expression was significantly higher in the tumor stroma compared with normal tissues. ColXα1enhanced proliferation, migration, invasion of colon cancer cells and knockdown of ColXα1 inhibited tumorigenesis in vivo [41]. Similar to the findings in the colorectal tissue, we reported that ColXα1 was not expressed in normal breast tissue [10]. Interestingly ColXα1 is present in a periductal location surrounding ductal carcinoma in situ, and within the tumor associated stroma in select invasive breast carcinomas, particularly in ER+/HER2+ cancers [42]. We had earlier identified that presence of ColXα1, determined by gene expression analysis and immunohistochemistry, predicted neoadjuvant therapy response in ER+/HER2+ tumors. In the current study, we demonstrated that ColXα1 expression predicts long term outcome in ER+/HER2+ tumors in the adjuvant setting as well. In our cohort most of the patients were treated before the results of the Z-11 trial were published and axillary dissection was performed in a significant number of cases. Consistent with other studies [3,[7][8][9], ER+/ HER+ tumors exhibited aggressive features; three quarters of the cases had LVI, and half of the cases had lymph node metastasis.
We found an inverse relationship between TILs and ColXα1 expression, and that low TILs and ColXα1 both contribute to the poor treatment response and prognosis. We did not found differences in TIL or ColXα1 expression with age. The relationship of ColXa1 expression and imaging findings, like mammographic density, can be investigated in future studies.
In addition, ER and PR positivity is defined by > 1% of nuclear staining of the tumor cells. In practice, a large amount of information is lost when one labels a tumor as a ER or PR-positive, because a tumor in which 10% of cells exhibit weak ER or PR staining is biologically different from one that demonstrates strong intensity staining in about 90% of cells. Although the vast majority of hormone receptor positive tumors show strong immunoreactivity, approximately 20% of tumors exhibit variable ER/PR expression. Raghav et al. [43] demonstrated no significant impact on survival and benefit of endocrine therapy in low ER/PR+ cases, which was similar to triple negative tumors. In the NSABP B-14 clinical trial Baehner and coworkers reported greater benefit from tamoxifen in patients with higher ER expression [44]. We analyzed our cases based on ER/PR expression and found that in the low PR+ group of 48 (29%) tumors, OS was significantly lower in ColXα1 high tumors. ColXα1 expression was associated with poor outcome in low PR+ tumors but not in high PR tumors. Like PR, other tumor stromal factors and tumor  3 Interaction and Impact of PR status, ColXα1 expression and number of metastatic nodes on survival. ColXα1 expression levels stratify only low PR+ tumors into prognostic groups (top left). Presence of lymph node metastasis was not a risk factor for patients with low PR levels but was a risk factor for patients with high PR levels. Nodes were grouped as follows: Node score = 0, 0 nodes, Node score = 1, 1-3 nodes, Node score = 2, > 3 nodes microenvironment variables, such as ColXα1 have a potential to predict the treatment response and outcome.

Conclusion
Our findings indicate that extent of ColXα1 expression in the ER+/HER2+ breast tumor stroma is prognostic. Increased ColXα1 expression is associated with shorter OS and DFS in both neoadjuvant as well as adjuvant settings. Further molecular or proteomic studies are needed to refine the definition of ColXα1 positive/overexpressing tumors. The relationship of TILs and ColXα1is intriguing, which needs further investigation, including correlation with imaging studies.

Additional file
Additional file 1: Table S1. Raw data for all cases in this study All data generated or analyzed during this study were included and de-identified (XLSX 26 kb)