Monitoring the responsiveness of T and antigen presenting cell compartments in breast cancer patients is useful to predict clinical tumor response to neoadjuvant chemotherapy

This prospective study was approved by the ethics committee of the Instituto Nacional de Cancerología - Bogotá (reference number ACT-018 May 2012). The patients and all healthy donors had signed an informed consent form before blood samples were taken. A total of 560 patients with pathological diagnosis of breast cancer were interviewed at the Instituto Nacional de Cancerología and the Clínica del Seno (Bogotá-Colombia) between 2012 to 2015; of these patients, 36% were eligible to be treated with Doxorubicin and Cyclophosphamide (A/C) scheme as neoadjuvant chemotherapy and 22% of total patients overexpress Her2/neu; a total of 17 patients with ductal invasive carcinoma (DIC) were included in the study. After informed consent had been signed, two blood samples were taken (20 mL each) one to three days before the first dose of chemotherapy and eight to ten days after third dose of A/C chemotherapy. Healthy women (HD), were used as controls (age-matched). PBMCs were isolated by density gradient with Ficoll Hypaque (GE) and cryopreserved in liquid nitrogen in freezing media (RPMI-1640 50%, FBS 40% and 10% of DMSO) until used. Clinical data of the included patients is shown in Table 1; clinical response was evaluated by residual cancer burden (RCB) clasification [10]. RCB was calculated based on primary tumor bed area, overall cancer cellularity, percentage of cancer that is in situ disease, number of positive lymph nodes and diameter of largest metastasis.

For the analyses of different sub-populations and phenotype of T and APC we use specific staining panels. For ex vivo sub-populations in the PBMCs obtained from patients before and after treatment and HD we quantified in a single tube: (i) regulatory T cells, (ii) Myeloid-derived suppressor cells (MDSCs), and (iii) myeloid DCs and plasmacytoid DCs by the combination of the following antibodies: CD4-BV510, CD25-APC-Cy7, CD127-PECy5, FoxP3-Pacific Blue, Lin1-FITC (CD3, CD14, CD16, CD19, CD20, CD56), CD15-FITC, CD13-PE, CD33-PE, HLA-DR-PE Dazzle 594, CD11c-Alexa Fluor 700, CD123-PECy7 (all from Biolegend); and Arg1-APC (R&D Systems). The gating strategy for ex vivo subpopulations is depicted in Fig. 1b. For the phenotype of mature DCs, the following antibodies were used: Lin1-FITC, HLA-DR-PE Dazzle 594, CD11c-Alexa Fluor 700, CD123-PECy7, CD83-PECy5, CCR7-Alexa Fluor 647, and CD86-PE (all from Biolegend), the gating strategy is depicted in Fig. 2a. For the analysis for TCR internalization and T cell activation markers, the following antibodies were used: CD3-FITC, CD154-APC, CD69-PECy7, CD25-PE (all from Biolegend). Finally, for cytokine secretion, we measure in the supernatant by Cytometric Bead Array (CBA) human Th1/2 and inflammatory cytokines (BD) of DCs after maturation and T cell activation. Flow cytometry data was acquired using FACS Aria II (BD) and analyzed using FlowJo Software (Tree Star Inc.).

The phenotype and functional capacity of monocyte-derived DCs was evaluated in vitro after exposure of PBMCs to IL-4 and GM-CSF as described by Martinuzzi et al. [11], and maturated with pro-inflammatory cytokines in combination with Type I interferons as described by Mailliard et al. [12, 13]. Briefly, after induction of immature DCs (iDCs), a combination of IFN-γ (R&D systems), IFN-α (Intron-A- ROCHE), TNF-α, IL-6, IL-1β (all from Cellgenix), and Poly I:C (Sigma-Aldrich). For DCs maturation phenotype, the expression of CD11c+, HLA-DR+ and Lin1- was analyzed by flow cytometry (FC) in total PBMCs (Fig. 2a), and secretion of IL-12p70 was quantified by CBA (BD Biosciences) in the supernatant of mature DCs culture. Simultaneously, for the determination of responsiveness of T cells, 5 × 10⁶ PBMCs/mL were stimulated 24 h with a mixture of anti-CD3, CD28, and CD2 microbeads (Miltenyi Biotec) in a ratio 2:1 (PBMC:beads) cultured in AIM-V media (Thermo Fisher Scientific). After stimulation, we quantify the internalization of TCR (reduction of CD3 MFI) and expression of CD69, CD25 and CD154 (MFI and percentage) in CD3+ T cells as shown in Fig. 3a and c.

All immunological data were normalized against unstimulated controls (delta between stimulated T cells or mature DCs with unstimulated T cells or iDCs respectively). Since most of the readouts did not present a normal distribution, non-parametric tests were applied to test for statistical differences between groups with Two-way ANOVA with Turkey’s multiple comparison tests. For paired samples, we used Wilcoxon test. Receiver Operating Characteristic (ROC) curves analysis were done with Prism 5 software (GraphPad). Factorial Analysis with Principal Component Analysis (PCA) was made in SPSS (IBM – version 20), the component matrix was rotated using Varimax rotation to facilitate the interpretation of PCA, two principal components were extracted, and PC loadings of variables and PC scores of samples were a plot in a two-dimensional graph. The multifactorial categorical analysis was done using Generalized estimating equations (GEE) in STATA 13 software [14].

Neoadjuvant anti-tumor therapy with A/C is used in BC patients to induce a reduction in the tumor size before surgery. It is well known that anti-tumor therapy with A/C can produce a clinical response evidenced by the reduction in RCB in most patients. The monitoring of tumor response in 17 patients with BC during neoadjuvant chemotherapy showed that after treatment pathological complete response (pCR) observed in four of 17 patients (based on RCB index), with a tumor shrinkage in 16/17 patients who experienced a significant reduction in tumor area (Fig. 1a). Based on tumor response in this cohort, we explored the behavior of different immunological readouts before and after chemotherapy that could be associated with the clinical response. It is well known that tumor escape from immunosurveillance is favored by infiltration of the tumor by different populations of suppressor cells such as CD4+ CD25+ FoxP3+ regulatory (Tregs) [15], suppressor macrophages [16], Myeloid Derived Suppressor Cells (MDSCs) [17], and immature DCs, that inhibit tumor-specific T cells favoring escape from immune surveillance. Different reports indicate that the expansion in blood of some of these cells is a common finding in patients with BC [18‐20], and only certain reports explore how anti-tumor therapy modulates the blood levels of these cells in colorectal cancer [21]. To analyze whether the anti-tumor therapy induces changes in the levels of suppressor cells, we compared ex vivo in peripheral blood in a group of patients with BC (n = 12), the levels of Tregs, MDSC, and DCs before and after three cycles of antitumor therapy (Fig. 1b). Once baseline levels of each population in peripheral blood from healthy donors were established (HD, n = 10), these were compared with those from patients with BC before and after treatment (n = 12). As shown in Fig. 1c, the ex vivo analysis of Tregs, MDSCs and DC (both myeloid and plasmacytoid), present in PBMCs of patients before and after chemotherapy showed no significant differences with the levels detected in HD. Altogether these results suggest that in the cohort of patients analyzed, the levels of different populations of suppressor cells in peripheral blood do not undergo significant changes after chemotherapy.

Through different mechanisms, the tumor microenvironment modulates the functional capacity of T and APC compartments [8, 20]. This tumor microenvironment in BC affects the maturation capacity of DCs [20, 22] and function of T cells [23]. However, it is unknown if chemotherapy with A/C restores the responsiveness of T and APC compartments and whether this can be assessed in peripheral blood of treated patients. To evaluate the effect of anti-tumor therapy on the functional capacity of the APC compartment, we established an in vitro system in order to analyze the expression of several maturation markers on DC derived in situ from monocytes [11] and on plasmacytoid and myeloid DCs present in PBMCs (cells CD123+ or CD11c + within Lin-1-/HLA-DR+ population respectively – Fig. 2a), after stimulation with a cocktail of pro-inflammatory cytokines [12]. As expected, after stimulation of PBMCs with cytokines, monocyte-derived DCs of HD showed a positive response to the pro-inflammatory stimulus that was evidenced by increased expression of CD83, CD86, and CCR7 in comparison with immature DCs (Fig. 2b). We choose the expression of CD83 as a key marker to identify mature DCs; then we compared the delta percentage of mature DC minus the percentage of immature DCs in HD and BC patients before and after treatment (Fig. 2c). In contrast to what was observed in DC of HD, in the patient group, monocyte-derived DCs and myeloid DC before therapy exhibited a reduced expression of CD83 in response to the maturation stimuli (p < 0.001 and p < 0.01, respectively) that was restored in myeloid DC after chemotherapy (Fig. 2c). Interestingly, it was found that in response to a cytokine cocktail, the expression of CD83 in plasmacytoid DCs of patients (either before or after chemotherapy) and HD was rather similar (Fig. 2c). Besides the measurement of CD83 to evaluate the functionality of the DC we also measured the secretion of IL-12p70 (IL-12) in culture supernatants of PBMCs stimulated with the combination of pro-inflammatory cytokines described by Mailliard et al. [12]. While IL-12 was clearly detected in cells of HD in the presence of the cytokine cocktail, a significant reduction in the secretion of IL-12 in cells of patients before chemotherapy was observed, and furthermore, the production of IL-12 was significantly recovered after three doses of A/C (Fig. 2d). Altogether, these results show a remarkable defect in the APC compartment of patients before therapy that is recovered after chemotherapy.

As observed in DCs, we hypothesized that T cells could also have a functional defect in BC patients. To evaluate the capacity of T cells to respond in vitro, we stimulated patients’ and HD’s PBMCs with anti-CD3/CD28/CD2 beads for 24 h. As expected, after the in vitro stimulation, T cells from HD showed efficient TCR internalization, evidenced by the reduction of CD3 MFI (Fig. 3a). Paired analyses demonstrated that the internalization elicited by the stimulus was statistically significant in the healthy individuals examined (Fig. 3b left panel). Fig. 3b right panel, shows that TCR internalization in BC patients before therapy was compromised (p < 0.05). Following TCR internalization, several activation signals on T cells like increased expression of the alpha chain of the IL-2 receptor (CD25), CD154 (CD40L) and CD69 are associated with this phenotype [24, 25]. We evaluated the activation phenotype of CD3+ T lymphocytes in response to the in vitro stimulation in HD; we found an increased expression of CD25 (not shown), CD154, and CD69 in response to anti-CD3/CD28/CD2 beads (Fig. 3c). However, when we compared the delta of MFI (MFI of stimulated cells minus MFI of unstimulated cells) in BC patients before and after chemotherapy, we observed a small difference in the expression of CD25 (p = 0.081) between HD and BC patients before treatment (Fig. 3d, left panel). The expression of CD69 show a significant impairment in BC before and after chemotherapy compared to the expression levels of HD (Fig. 3d, central panel). The limited expression of CD154 elicited by the stimulus observed in BC patients before therapy compared to HD showed a partial recovery after chemotherapy (Fig. 3d, right panel). Together, these results suggest a dysfunctional capacity of T cells in BC patients before treatment.

The efficient activation of T cells against tumors is a multi-step process that relies not only on the capacity of APC to stimulate T cells via TCR/MHC interactions [26] but also in the ability of activated T cells to stimulate on APCs the expression of costimulatory molecules such CD83 and the production of IL-12 via the stimulation by CD154 (expressed by T cells upon activation) of CD40 on APCs [27, 28]. For these reasons, we propose that the dysfunction of T and APC compartments evidenced here in BC patients are correlated and, furthermore, that by assessing the functional performance of these two compartments is possible to discriminate between BC patients and HD status. Our results so far suggest an associated defect in the functional capacity of APCs and T cells in patients with BC before the anti-tumor therapy in agreement with a diverse array of suppressive mechanisms of tumor cells that hampers immune surveillance of tumors [29]. To clarify different possible correlations that may exist between the cells responsible for the immune response against tumors, using the proposed in vitro model, we compared several immunological readouts in HD and cancer patients. To do this, we used multivariate analysis (factor analysis with Principal Component Analysis - PCA), to simultaneously evaluate different parameters assesed in PBMCs. We selected the variables (Additional file 1: Table S1) that best describe the behavior of the samples by the matrix of components of each variable in the PCA (Additional file 2: Figure S1A). Examining the scores of the PCA in the samples, we observed a clear separation between HD and BC patients before anti-tumor treatment; samples of some patients after chemotherapy have an intermediate behavior between HD and the same patient before receiving treatment (Additional file 2: Figure S1B). This result would suggest that this in vitro model may be useful to monitor the immune and clinical responses in BC patients along adjuvant chemotherapy. Finally, we compared the sensitivity and specificity of several immunological determinants analyzed to differentiate between HD and BC patients; for this, using ROC curves we examined the area under the curve (AUC) of the internalization of the TCR (Additional file 2: Figure S1C, AUC = 0.67; p = 0.05). The multiparametric analysis done by PCA represent the complexity of the immune system involved in the response of BC patients to A/C chemotherapy. The analysis of this complexity with our model suggest that by assessing the functionality of T/APC compartments in blood it is possible to differentiate between HD and BC patients.

Predicting clinical response of tumors to neoadjuvant chemotherapy remains a formidable challenge. Despite that molecular testing of TOP2A and the in situ analysis of tumor landscape after neoadjuvant chemotherapy are promising readouts useful to predict tumor response and survival in treated BC patients [3, 30], the usefulness of functional analyses of peripheral blood leukocytes to predict tumor responsiveness to chemotherapy remains unknown. To address this possibility, the immunological readouts of all functional studies performed with peripheral blood leukocytes from patients before chemotherapy were categorized first and then analyzed by a multifactorial categorical analysis (by general estimating equations - GEE). This was done in order to calculate coefficients for each explanatory variable that best fit a model that explain the behavior of a given clinical parameter based on immune readouts (Additional file 3: Table S2). Taking into account three immunological readouts after chemotherapy: CD3 internalization and CD69 expression in T cells and the IL-12 production in DCs, became evident that the explanatory values of CD69 and IL12 (Additional file 3: Table S2) are useful for predicting tumor response to chemotherapy. All three values are associated with the expression of estrogen receptor but not to the expression of progesterone receptors, Her2/neu or KI-67 by the tumor (Additional file 3: Table S2). Altogether, these results suggest that responsiveness of the T and APC compartments and tumor clinical response are two components prompted by chemotherapy that somehow are related.

To further confirm this, we examined which immune characteristics are associated with the best fitting of tumor clinical response to chemotherapy. We found that after three doses of chemotherapy with A/C, TCR internalization is correlated with tumor response to the treatment quantified by RCB index (Fig. 4a).

Finally, we assessed the predictive value of immunological readouts (before chemotherapy) to predict beforehand the clinical outcome before treatment, to do this, the performance of readouts before treatment was cross-checked with RCB classification after chemotherapy. Higher levels of TCR internalization (p < 0.01), and matured plasmacytoid DCs (p < 0.05) were found in patients with better tumor response (pCR or RCB-I) compared with patients with higher RCB (RCB-II) (Fig. 4c). Based on these results, we established the sensibility and specificity of this immunological readout in predicting tumor response to A/C treatment by doing a ROC curve. We found that the ROC curves of CD3 internalization and DC maturation have a high AUC (0.816 and 0.825 respectively) that discriminate patients who respond from those who do not respond (Fig. 4d). These results let us argue that high levels of mature plasmacytoid DCs and the internalization of CD3 detected in peripheral blood before chemotherapy after in vitro stimuli as useful biomarkers for predicting clinical responsiveness of tumors to primary chemotherapy with A/C in BC patients.