Background
Pre-clinical experimental evidence suggests that tumor treatment with some chemo-radiotherapy regimens induce in tumor cells immunogenic cell death (ICD) that promotes the antigenicity and immunogenicity of tumors [
1]. The immunogenicity of tumor cells dying via ICD is favored by cross-presentation of antigens by DCs to anti-tumor CD8 T-cells responsible for controlling the tumor. Retrospective studies have confirmed that cancer patients treated with Doxorubicin having mutations in molecular components involved in recognition of tumor cells that die by ICD have shorter overall survival and a higher risk of metastatic disease [
2].
Clinical evidence on the immunogenicity of tumors induced by anti-tumor therapy has shown that a good clinical response to Doxorubicin is correlated with changes in immune contexture of the tumor [
3,
4]. Furthermore, the study of biomarkers in colon cancer to predict clinical response has identified immunological signatures in the tumor microenvironment with predictive and prognostic value [
4,
5]. The efforts to demonstrate a relationship between immunogenicity of tumors induced by chemotherapy and anti-tumor immune signatures in breast cancer (BC) patients with clinical response to treatment have yielded some evidence in this direction [
6,
7]. Despite, these studies for the identification of biomarkers with the potential to predict chemotherapeutic responses in BC are encouraging, blood-based monitoring systems to predict clinical response to treatment does not exist. In the case of BC patients under neoadjuvant therapy, the identification of predictive markers of clinical response using whole blood or PBMCs is desirable because this would help the adjustment of the chemotherapy regimes in trying to achieve pathological complete responses (pCR) in all patients treated.
Tumor growth is the result of tumor escape of immune surveillance due to a poor performance of T and antigen presenting cell (APC) compartments [
8]. Although experimental evidence suggests that primary chemotherapy with Doxorubicin induces ICD that favors anti-tumor responses and changes in the contexture of the tumor, the effect of Doxorubicin on T and APC compartments in patients under primary chemotherapy is yet to be demonstrated. We hypothesized that a favorable clinical response of BC tumors to neoadjuvant therapy with Doxorubicin and Cyclophosphamide (A/C) will revert suppression in these two compartments. In a recent study, we design an in vitro system to monitor the specific anti-tumor response before and after anti-tumor therapy [
9]. Our results suggest that the status of disease-free survival and a complete clinical response is supported by tumor-specific T lymphocytes induced by anti-tumor treatment. To generate clinical evidence that chemotherapeutic agents inducing ICD restores immunosurveillance of the T and APC compartments in cancer patients with clinical tumor response to Doxorubicin, in the present work we studied a group of 17 patients with BC in neoadjuvant therapy (three cycles of A/C), whose tumors experienced significant clinical response after chemotherapy. This behavior of the tumor prompted us to investigate whether a favorable clinical response to primary chemotherapy (A/C) is correlated with the better performance of T cells and APCs interaction. To do this, we compared the immunological performance of T and APC compartments in peripheral blood of these patients before and after chemotherapy. We found that the overall suppression of these two compartments perceived before treatment is reversed after chemotherapy and this recovery correlates with clinical response. Altogether our results let us argue four things: first, the unresponsiveness to stimuli of T/APC compartments observed in these BC patients before treatment starts to recover after three cycles of A/C; second, primary chemotherapy reestablished the crosstalk between T/APC compartments; third, the recovery of this crosstalk is correlated with the clinical response of the tumor and, fourth, monitoring T/APC compartments may be useful to identify predictive biomarkers of tumor responsiveness to treatment.
Discussion
Based on parameters used by us to measure tumor-specific T cells generated in response to anti-tumor therapy [
9], in the present study, we monitored a series of immunologic parameters in BC patients’ PBMCs obtained before and after chemotherapy with A/C trying to establish, first, the capacity of neoadjuvant chemotherapy to reestablish immune responsiveness and second, the usefulness of immunological readouts to predict clinical tumor response prior to treatment. We evaluated the expansion and phenotype of Tregs, MDSCs, and DCs present in patients’ PBMCs. These three populations play a critical role in tumor escape of immune surveillance [
4,
31]. Statistically significant differences between the levels of Tregs and MDSCs found in samples of patients’ PBMCs (obtained either before or after chemotherapy) with those observed in control’s PBMCs were not found. The fact that these measurements have usually been made in BC patients with advanced disease and not in patients with newly diagnosed primary tumors and before neoadjuvant chemotherapy, as in our case, may explain these results.
MDSCs are a heterogeneous population of myeloid cells that accumulate in cancer patients inhibiting T cell-mediated immune responses through the production of NO, Arginase and reactive oxygen and nitrogen species, which foster tumor infiltration by Tregs [
32,
33]. By studying the role of MDSCs in inhibiting immune surveillance of BC tumors, Verma et al., reported the increase of MDSCs from two different sources: monocyte-derived (cells CD11b + CD14+ CD124+ CD33+) and PMN-derived (cells CD11b + CD14- HLA-DR- CD66b + CD124+ CD15+) in peripheral blood of BC patients in neoadjuvant chemotherapy [
34]
. On the other hand, Yu et al. [
35], reported an increase of cells Lin- HLA-DR- CD14- CD15- CD13+ CD33+ that produce IDO in BC patients with advanced tumors. Using the same markers employed by Yu et al. (except for IDO), we did not find differences in the amounts of MDSC Arginase + in BC patients before chemotherapy in comparison with controls. The analysis of a complex population such as MDSC in different studies using different sets of markers makes difficult the comparison between studies.
On the other hand, the impact of A/C chemotherapy in the levels of MDSCs and Tregs has been a matter of debate. The use of A/C was associated with a significant increase of MDSCs in the blood of newly diagnosed BC cancer patients correlated with disease stage and metastatic tumor burden [
36]. In contrast, a more recent study shows a decrease in the levels of MDSCs and Tregs in blood attributable to the cytotoxic effect of A/C on these cells [
34]. After three cycles of A/C, we did not observe variations in levels of MDSCs or Tregs (neither CD127- nor FoxP3+). The difference between our results and those of others may be explained because the measurement of MDSCs and Tregs in blood pre- and post- chemotherapy have not been previously analyzed. In short, the contrasting results regarding the behavior of MDSCs and Tregs during anti-tumor therapy argues for the need for standardized methods for monitoring these two cell populations in patients during treatment.
IL-12 produced by DCs is a key point in cancer immunotherapy as it promotes CTLs that secrete IFN-γ a cytokine with recognized anti-tumor activity [
37]. This evidence suggests that evaluating the immune competence of DCs to produce IL-12 and to mature in response to a pro-inflammatory stimulus is useful to assess the immune surveillance of tumors. Very recently a whole-blood assay that was used for monitoring the immune competence in cohorts of healthy women and BC patients at different progression stages prior any treatment evidenced unresponsiveness of patients’ BDCA3 DCs to interferon alpha [
38]. In another study, Della Bella et al., reported a decrease in the absolute number of myeloid DCs in whole blood of BC patients’ ex vivo [
20]. This reduction that was associated with a decrease in CD119 (IFN-γR) and increased expression of CD83 without altering the expression of CD80 and CD86 in response to LPS was correlated with the severity of BC. Although we did not observe marked differences in percentages of DC populations among HD, and BC patients pre- and post-treatment, after three doses of chemotherapy we found a substantial recovery of CD83 expression and production of IL-12 in response to a cocktail of cytokines used by Mailliard et al., [
12] to derive type I alpha DCs in situ [
11]. An increased production of IL-12 was detected after tumor removal in the study by Della Bella et al. [
20], this and that the clinical tumor response to A/C correlates with the production of IL-12 and CD83 expression by DCs in the present study suggest that the responsiveness of DCs to the pro-inflammatory stimuli used here is useful for monitoring the recovery of immune surveillance by DCs during neoadjuvant treatment with A/C. In the same vein, results of preclinical studies in mice show that the A/C promotes recovery of immune surveillance associated with antigen presentation, increased expression of CD83 and IL-12 production by DCs [
39]. However, it is possible that IL-12 production by DCs has different prognostic value depending on the state of the disease, our results suggest that in early stages of treatment it promotes the recovery of the immune-surveillance and a favorable clinical response compared to its production after treatment that apparently favors tumor relapse [
38].
We observed a more efficient TCR internalization and the CD154 (CD40L) expression on T cells after chemotherapy. CD154 is expressed on both CD4 and CD8 T cells upon TCR stimulation. However, the consequence of activated CD4 Th1 cells expressing CD154 is better known [
40,
41]. In this regard, we speculate that the recovery of CD154 by Th1 cells may foster CD8 surveillance in BC patients treated with AC by promoting competent DCs after cognate CD40/CD40L interaction that probably stimulates IL-12 secretion as well as the up-regulation of adhesion and co-stimulatory molecules by DCs (e.g., CD83), all of which have been shown to occur after CD40 cross-linking on these two cell types [
40‐
48]. On this perspective, the responsiveness of the T and APC compartments after therapy observed in our patients argues in favor that neoadjuvant therapy reestablishes the cross talk between these compartments and that this is essential for immune surveillance (Additional file
4: Figure S2).
By multivariate PCA analysis, it was possible to integrate TCR internalization, CD83 expression and IL-12 production by mature DCs, with some immunological readouts (Additional file
1: Table S1). Despite neither parameter when were considered individually allows to discriminate between HD and patients, the PCA allowed us to segregate HD individuals from donors in the patient group clearly. In this regard, it is evident that after treatment, the behavior of variables in some patients becomes like those observed in the control group (HD). Finally, by using ROC curves, the TCR internalization allowed us to differentiate the immune response between HD and patients. Taken together these results lead to propose that the recovery of crosstalk between T and APC compartments induced by A/C therapy reflects the restoration of immune surveillance and is a good prognostic factor in BC patients treated with neoadjuvant A/C (Additional file
4: Figure S2).
Finally, it is of great interest to define biomarkers able to predict clinical response to chemotherapy in BC patients, in this regard candidate biomarkers are tumor infiltration by CD8+ T cells [
3] and TFH [
49] and in situ expression of markers such as HMGB-1 and autophagy [
50]. We propose that the proper TCR internalization and IL-12 production in response to treatment are potential biomarkers to predict tumor size reduction after three months of chemotherapy. The correlation between clinical response and ex vivo levels prior therapy of plasmacytoid DC CD83+ (a cell that produces type-I IFN important to activate anti-tumor responses) suggests this marker as useful for predicting clinical response to treatment. This result is consistent with the description of a type I IFN-related signature that predicts clinical responses to anthracycline-based chemotherapy in several independent cohorts of BC patients [
51].
Acknowledgements
This study was supported by funding from the Universidad Nacional de Colombia. DIB, Vicedecanatura de Investigación Universidad Nacional Medical School; funds from a joint grant between Fundación Salud de Los Andes, Universidad Nacional, and COLCIENCIAS. The authors express their gratitude to Dr. Fabio Méndez CEO at the Fundación Salud de Los Andes (FSA) and FSA for their generous support. The authors would also like to thank Dr. Bernardo Camacho and to personnel at the Hemocentro Distrital for their kind assistance in obtaining blood buffy coats from volunteers. Finally, our deepest gratitude to patients and healthy volunteers for their generous denotation of blood samples used in this study.