Prediction of distant metastatic recurrence by tumor-infiltrating lymphocytes in hormone receptor-positive breast cancer

Background Breast cancer subtypes are known to have different metastatic recurrence sites. Distant metastases are often observed during the post-operative course in patients with human epidermal growth factor receptor 2 (HER2)-enriched breast cancer and triple-negative breast cancer, but are relatively rare in those with hormone receptor-positive and HER2-negative (HR+/HER2−) breast cancer. Tumor-infiltrating lymphocytes (TILs) serve as an index to monitor tumor immune microenvironment and may possibly predict the prognosis and therapeutic effect in breast cancer. This study aimed to investigate the correlation between TIL density and recurrence site in HR+/HER2− breast cancer. Methods In stages I–II of HR+/HER2− breast cancer patients who underwent surgery as the first treatment and received adjuvant endocrine therapy (except adjuvant chemotherapy), forty-two patients relapsed after surgery. TILs were evaluated using needle biopsy specimens for the diagnosis of breast cancer. Morphological assessment was conducted using conventional hematoxylin and eosin staining. Results Six patients had no TILs density. In them, local recurrence was significantly less (p = 0.022), while distant metastases were significantly more (p = 0.015) compared to those in patients with TIL density. Therefore, for the prediction of distant metastases in HR+/HER2− breast cancer without chemotherapy, TILs could be used as predictors in univariate analysis (p = 0.015, odds ratio [OR] = 0.127), although not as independent factors (p = 0.285, OR = 0.144). Conclusions Our findings indicate that TILs may predict distant metastatic recurrence in stages I–II of HR+/HER2− breast cancer in patients who do not undergo chemotherapy. Supplementary Information The online version contains supplementary material available at 10.1186/s12905-021-01373-7.

breast cancer intrinsic subtypes [1,3]. Distant metastases are frequently found during the post-operative course in patients with human epidermal growth factor receptor 2 (HER2)-enriched breast cancer and triple-negative breast cancer (TNBC), while they are relatively rare in patients with hormone receptor-positive and HER2-negative (HR+/HER2−) breast cancer. Therefore, early detection of locoregional recurrence after surgery for HR+/ HER2− breast cancer has been suggested to be vital for good prognosis [7]. However, HR+/HER2− breast cancer can also be distant from the first site of recurrence.
Invasion and metastasis of cancer involve the molecular pathological features of not only the cancer but also the surrounding interstitial region, commonly called the tumor microenvironment (TME) [8,9]. TME consists of cancer-related fibroblasts, neovascular cells, and tumor-infiltrating lymphocytes (TILs). Recently, a correlation between TILs and lymph node metastasis was reported in gastric cancer and melanoma [10,11], and that between TILs and axillary lymph node metastasis was reported in TNBC [12,13]. In all these reports, lower TIL density was shown to more likely cause lymph node metastasis. However, reports on the TIL relationship with distant metastases are few compared to those on its relationship with lymph node metastases. We speculate that if TILs are involved in lymph node metastasis, which is a risk factor for distant metastasis, TILs should also be involved in distant metastasis.
We hypothesized that the site of the first recurrence may vary based on TIL density in HR+/HER2− breast cancer. Therefore, this study aimed to investigate the correlation between TIL density and recurrence site in HR+/HER2− breast cancer, specifically in patients who did not receive chemotherapy either before or after surgery.

Patient background
In patients with stages I-II of HR+/HER2− breast cancer underwent surgery as the first treatment and received adjuvant endocrine therapy, except for adjuvant chemotherapy, between 2007 and 2015 at the Osaka City University Hospital, 42 patients were found to have recurrence and were enrolled in the present study. Breast cancers and their subtypes were diagnosed by core needle biopsy or vacuum-assisted biopsy. HR+/HER2− breast cancer was defined as estrogen receptor (ER)-and/or progesterone receptor (PgR)-positive and HER2-negative by immunohistological staining of the tissue. Expression of Ki67 was also examined using immunohistological staining and the cutoff value for Ki67 was set at 20% based on previous reports [14]. CT, US, and bone scintigraphy were used to evaluate the cancer stage. Cancer progression was evaluated according to the 8th edition of the TNM classification of malignant tumors (8th edition) by the Union for International Cancer Control [15]. All patients underwent either mastectomy or breast-conserving surgery, and the latter received radiation therapy in the remaining mammary gland post-surgery. For patients diagnosed with axillary lymph node metastasis by preoperative imaging, axillary lymph node dissection was also performed. In the case of breast cancer diagnosed without axillary lymph nodes metastasis, sentinel lymph node biopsy was performed using a combination of radioisotope and dye methods, as reported previously [16,17]. Metastasis was examined pathologically by slicing sentinel lymph node at 2-mm thickness [18,19]. Axillary lymph node dissection was performed in cases where metastasis in the sentinel lymph node was greater than 2 mm, also called macro-metastasis. The biological characteristics (ER/PgR/Ki67/HER2) of the resected specimens were re-examined, and the results validated those of the biopsy tissue. Adjuvant chemotherapy administration was decided at the discretion of the attending physician, taking into consideration the patient's wishes and comorbidities. In the present study, disease-free survival (DFS) was defined as the time from surgery to recurrence or death; progression-free survival (PFS) was defined as the time from recurrence to relapse or death due to breast cancer after the next treatment; post-recurrence survival (PRS) was defined as the time from recurrence to death due to breast cancer.

Histopathological evaluation of TIL density
TIL density was evaluated pathologically using biopsy specimens. The pathological diagnosis and examination were jointly performed by two breast pathologists. The definition and evaluation method for TILs followed the International TILs Working Group 2014 [20]. Specifically, the density of infiltrating lymphocytes was averaged on full sections, at least five fields, of the tumor stroma. The results were divided into four groups according to previous reports (score 3: > 50%, score 2: > 10-50%, score 1: ≤ 10%, and score 0: absent) ( Fig. 1) [21][22][23][24].

Statistical analysis
All statistical analyses were performed using the JMP software package (SAS, Tokyo, Japan). For evaluating the correlation between two groups for each clinicopathological feature, Pearson's chi-square test was used. The odds ratio (OR) and 95% confidence interval (CI) were calculated using logistic analysis. Multivariable analysis was performed with the multivariable logistic regression model. To evaluate the hazard ratio (HR) and 95% CI related to survival outcomes, such as PFS or PRS, Cox proportional hazards models were used for univariate analysis and Cox regression models were used for multivariate analysis. Significance was defined at p value < 0.05.

Ethics statement
Written informed consent to participate in the study was obtained from each subject in accordance with the Declaration of Helsinki. Each patient or their family was fully informed of the nature of the investigation in this study prior to obtaining their written informed consent. The study protocol was approved by the Ethics Committee of Osaka City University (Approval No. # 926). Table 1 shows the clinicopathological features of fortytwo HR+/HER2− breast cancer patients with recurrence who had never received chemotherapy. The median age at the time of operation was 59 years (range 37-79 years). The median tumor size was 20.5 mm (9.5-49.3 mm), and no evidence of axillary lymph node metastasis was found by pre-operative imaging, in all cases. Therefore, sentinel lymph node biopsy was performed for all. Micro-metastasis was identified in three cases (7.1%) and macro-metastasis was found in seven cases (16.7%). The latter underwent axillary lymph node dissection, and pathological examination revealed < 3 axillary lymph nodes with metastases. The expression of ER in one patient (2.4%) was negative, while that of PgR was positive. Three patients (7.1%) were negative for PgR expression and seven (16.7%) had high Ki67. Eighteen cases (42.9%) underwent breastconserving treatment and received post-operative radiation therapy for the residual mammary gland. None of the 42 patients received post-operative radiation therapy in the axilla or supraclavicular fossa. Post-operative pathological findings showed lymphatic invasion in 21 patients (50.0%) and venous invasion in three patients (7.1%). Six patients (14.3%) were diagnosed with nuclear grade 3. All patients received adjuvant endocrine therapy; 10 patients (23.8%) were treated with tamoxifen and seven patients (16.7%) were treated with a combination of tamoxifen and a luteinizing hormonereleasing hormone agonist. The remaining 25 patients (59.5%) received anastrozole. The median DFS time was 1462 days (range 132-3300 days). The first recurrence site was local in 18 (42.9%) patients and regional lymph node in 13 (30.9%) patients, while distant metastases were observed in 11 (26.2%) patients. Lung metastasis, detected in seven (16.6%) patients, was the most common distant metastasis. None of the patients showed simultaneous locoregional recurrence and distant metastasis. Six (14.3%) patients had TIL density higher than 10%, while in six other patients (14.3%), TIL density was not detected.

Correlations between clinicopathological features and recurrence sites
The correlations between clinicopathological features and recurrence sites are listed in Table 2. According to sentinel lymph node biopsy results, two out of the seven cases with metastasis had regional lymph node recurrence and five had distant metastatic recurrence. Thus, in patients with axillary lymph node metastases, local recurrence was significantly less (p = 0.012), and distant metastasis was significantly more (p = 0.003) than that in patients without axillary lymph node metastases. In addition, in patients with longer DFS, distant metastatic recurrence was significantly higher (p = 0.014). Further, patients who underwent mastectomy had significantly more locoregional recurrence (p = 0.020). Examination of the correlations between operative procedures and clinicopathological features revealed the following significant differences in patients who underwent mastectomy: older age (p = 0.001), higher frequency of tumor with diameter more than 30 mm (p = 0.006), and lower Ki67 (p = 0.012) (Additional file 1: Table S1).
Focusing on TIL density, no significant difference was observed when the cutoff value of TIL density was set at 10%. However, of the six patients with no TIL density, two had regional lymph node recurrence and four had distant metastatic recurrence. Therefore, in patients with no TIL density, local recurrence was significantly less (p = 0.022) and distant metastasis was significantly more (p = 0.015) than that in patients with TIL density.

Correlations between clinicopathological features and TILs
We examined the correlation between clinicopathological features and TILs (Table 3); TIL density tended to be absent in patients with tumor size > 20 mm (p = 0.078) and 10% or more in patients with tumor size < 20 mm (p = 0.078). Macro-metastasis was found by sentinel lymph node biopsy in 4/6 patients with no  TIL density. Significantly more lymph node metastasis was found in patients with no TIL than that in patients with TIL (p < 0.001). Significantly higher frequency of venous invasion was found in patients with TIL ≥ 10% than that in patients with TIL < 10% (p = 0.007).

Prediction of distant metastases in patients with HR+/ HER2− breast cancer without chemotherapy
Examining distant metastasis predictors based on the results, axillary lymph node metastasis (p = 0.003, OR = 16.723) and operative method (p = 0.026,

Prognosis after recurrence
Using univariate analysis with PFS after recurrence, no clear predictive factor could be identified, since treatment after recurrence was variable (Additional file 2: Table S2).
On the other hand, as per multivariate analysis with PRS, lymph node metastasis during surgery was found to be a poor prognostic factor (p = 0.042, HR = 17.339) ( Table 5).

Discussion
There are various reports on the prediction of distant metastasis; however, the recurrence site in these reports differs depending on the subtype, and tumor size and lymph node metastasis are common risk factors [2,[25][26][27]. In addition, young age [2], histopathological grade, and lymphovascular invasion have been reported as risk factors [2,26], although there are reports suggesting otherwise, as well. In this study, a combination of "BCT and radiation therapy" as postoperative procedures was found to be predictive factor for distant metastasis. The predictive factor status may be attributed to the fact that "BCT and radiation therapy" group was significantly younger, and had larger tumor diameter and higher Ki67 compared to the "mastectomy" group. They are the risk factors listed above. Although each parameter alone was not a predictor, they were found to turn into a predictor when combined. Formation of distant metastases in cancer generally follows the concept of "seed and soil. " [28]. TME corresponds to the "soil" as per this concept [8,9]. TILs are also included in the cells that constitute the TME. In breast cancer, TILs have been reported to vary by subtype. In particular, HER2-enriched breast cancer and TNBC have been reported to show significantly higher TIL density than HR+/HER2− breast cancer, and TILs have been proven to predict the therapeutic effect of chemotherapy [13,[29][30][31]. In contrast, there are very few reports examining the correlation between TILs and clinicopathological factors or therapeutic effects in HR+/HER2− breast cancer.
In this study, we excluded patients who had undergone neoadjuvant/adjuvant chemotherapy based on three reasons. First, chemotherapy affects the immune microenvironment, including TILs, in preoperative chemotherapy study, and related changes may affect prognosis [32]. TILs are also predictors of the therapeutic effect of adjuvant chemotherapy; therefore, adjuvant chemotherapy may also affect the tumor immune microenvironment. Secondly, axillary lymph node metastasis before adjuvant chemotherapy was diagnosed based on image only, and therefore, the diagnosis was not accurate. Lastly, different chemotherapy regimens are known to have different effects, and neoadjuvant chemotherapy causes more local recurrence than adjuvant therapy [33]. There are some reports on the relationship between recurrence sites and TILs. Park et al. [34] reported patients with TILs > 10% in early-stage TNBC to show significantly more locoregional recurrence than those with lower TILs. In cervical squamous cell carcinoma, low TILs are also likely to cause distant metastasis [35]. Moreover, distant metastasis is reportedly predicted from pathological features, including lymphocytes of lymph nodes in breast cancer with lymph node metastasis, although not from the density of lymphocytes around the tumor [36]. Bidwell et al. [37] have shown innate immune escape to promote bone metastasis, based on clinical data and experiments in mice. Some studies using breast cancer cell lines in vivo also reported that immunosuppression in the tumor immune environment increased the risk of lung metastasis [38,39]. In this study, TIL density was suggested to possibly be a distant metastatic predictor, although not an independent factor. We previously reported that TILs may also be involved in lymph node metastasis in HR+/HER2− breast cancer [40]. In this study, TILs were considered to be strongly correlated with lymph node metastasis, and were involved in distant metastasis prediction as well as prognosis after recurrence.
The greatest limitation of this study was that very few cases of recurrence were examined. Another limitation was that the patients received different types of adjuvant endocrine therapy. It will be necessary to examine by accumulating future cases. Moreover, younger and lymphovascular invasion, which were previously reported as risk factors, were not found to be predictors in this study. TILs may also be involved in venous invasion. Furthermore, considering that the TIL subtypes have different functions, detailed studies are necessary to identify the organs that are prone to metastasis. Immunohistochemical staining is needed to determine TILs subtypes. Nevertheless, TILs can be easily evaluated using needle biopsy specimens for the diagnosis of breast cancer and are highly useful. Filipits et al. [41] predicted distant metastases using a combination of the above-mentioned predictors and TILs may be considered to be an additional predictor. Currently, postoperative follow-up involves patient interview, palpation, and mammography. Whole-body imaging for asymptomatic breast cancer patients post resection surgery is not recommended [4][5][6]. However, late diagnosis of distant metastasis, that is, after emergence of detectable symptoms, leads to impairment of the patient's quality of life. Moreover, the risk of reduced treatment options, owing to the deterioration in the general condition as a result of recurrence, prevails. Therefore, it is important to identify and diagnose breast cancer that is prone to distant metastasis.

Conclusions
In stages I-II of HR+/HER2− breast cancer, locoregional recurrence is common and distant metastasis is relatively rare. This study suggested TILs to possibly be one of predictors of distant metastatic recurrence in stages I-II of HR+/HER2− breast cancer in patients who do not undergo chemotherapy.