Quintela-Fandino et al. Breast Cancer Research (2020) 22:124 https://doi.org/10.1186/s13058-020-01362-y RESEARCH ARTICLE Open Access Immuno-priming durvalumab with bevacizumab in HER2-negative advanced breast cancer: a pilot clinical trial Miguel Quintela-Fandino1,2,3* , Esther Holgado4, Luis Manso5, Serafin Morales6, Begoña Bermejo7,8,9, Ramon Colomer1,10,11, Juan V. Apala1,2, Raquel Blanco12, Manuel Muñoz1, Eduardo Caleiras13, Vega Iranzo9,14,15, Mario Martinez16, Orlando Dominguez17, Javier Hornedo3, Lucia Gonzalez-Cortijo3, Javier Cortes18,19, Ariadna Gasol Cudos6, Diego Malon2, Antonio Lopez-Alonso1, María C. Moreno-Ortíz12, Silvana Mouron1 and Santos Mañes12* Abstract Background: Preclinical research suggests that the efficacy of immune checkpoint inhibitors in breast cancer can be enhanced by combining them with antiangiogenics, particularly in a sequential fashion. We sought to explore the efficacy and biomarkers of combining the anti-PD-L1 durvalumab plus the antiangiogenic bevacizumab after bevacizumab monotherapy for advanced HER2-negative breast cancer. Methods: Patients had advanced HER2-negative disease that progressed while receiving single-agent bevacizumab maintenance as a part of a previous chemotherapy plus bevacizumab regimen. Treatment consisted of bi-weekly durvalumab plus bevacizumab (10 mg/kg each i.v.). Peripheral-blood mononuclear cells (PBMCs) were obtained before the first durvalumab dose and every 4 weeks and immunophenotyped by flow-cytometry. A fresh pre- durvalumab tumor biopsy was obtained; gene-expression studies and immunohistochemical staining to assess vascular normalization and characterize the immune infiltrate were conducted. Patients were classified as “non- progressors” if they had clinical benefit (SD/PR/CR) at 4 months. The co-primary endpoints were the changes in the percentage T cell subpopulations in PBMCs in progressors versus non-progressors, and PFS/OS time. (Continued on next page) * Correspondence: mquintela@cnio.es; smanes@cnb.csic.es 1 Breast Cancer Clinical Research Unit – Clinical Research Program, CNIO - Spanish National Cancer Research Center, Melchor Fernandez Almagro, 3, 28029 Madrid, Spain 12 Department of Immunology and Oncology, Centro Nacional de Biotecnología/CSIC, Darwin, 3, 28049 Madrid, Spain Full list of author information is available at the end of the article © The Author(s). 2020 Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data. Quintela-Fandino et al. Breast Cancer Research (2020) 22:124 Page 2 of 14 (Continued from previous page) Results: Twenty-six patients were accrued. Median PFS and OS were 3.5 and 11 months; a trend for a longer OS was detected for the hormone-positive subset (19.8 versus 7.4 months in triple-negatives; P = 0.11). Clinical benefit rate at 2 and 4 months was 60% and 44%, respectively, without significant differences between hormone-positive and triple-negative (P = 0.73). Non-progressors’ tumors displayed vascular normalization features as a result of previous bevacizumab, compared with generally abnormal patterns observed in progressors. Non-progressors also showed increased T-effector and T-memory signatures and decreased TREG signatures in gene expression studies in baseline—post-bevacizumab—tumors compared with progressors. Notably, analysis of PBMC populations before durvalumab treatment was concordant with the findings in tumor samples and showed a decreased percentage of circulating TREGs in non-progressors. Conclusions: This study reporting on sequential bevacizumab+durvalumab in breast cancer showed encouraging activity in a heavily pre-treated cohort. The correlative studies agree with the preclinical rationale supporting an immunopriming effect exerted by antiangiogenic treatment, probably by reducing TREGs cells both systemically and in tumor tissue. The magnitude of this benefit should be addressed in a randomized setting. Trial registration: (www.clinicaltrials.gov): NCT02802098. Registered on June 16, 2020. Keywords: Durvalumab, Bevacizumab, HER2-negative breast cancer, Vascular normalization, Immuno-priming Background represents a potentially interesting chemo-free regimen. Single-agent immune checkpoint inhibitors have shown Several recent trials combining small-molecule antiangio- moderate efficacy in advanced breast cancer [1–5]. Con- genics plus immune checkpoint inhibitors have shown en- siderable efforts are directed towards enhancing their ac- hanced activity particularly in settings where high VEGF tivity, most often in combination with chemotherapy, levels are known to drive tumor growth [20]. The role of but also in chemotherapy-free regimens [6]. VEGF as a driver of breast cancer progression, however, Several factors modulate the response of cancer cells has been extensively questioned in light of the long-term to anti-PD-1/PD-L1 agents, including tumor mutational results of phase III trials with antiangiogenics [21–24]. burden (TMB), PD-L1 expression, and immune infiltra- The efforts invested in studying immune-boosting com- tion [7, 8]. PD-L1 expression rates [9, 10] are usually binations, however, have paid little attention to the timing low in breast cancer. Several strategies have been sug- of administration of the “immune enhancer”. The natural gested to enhance the efficacy of immunotherapy, in- rhythms of innate and adaptive response could be used to cluding the manipulation of tumor microenvironment to enhance the efficacy of anti-PD-1/PD-L1 agents by com- enhance T effector cell infiltration and access to the bining them with agents in the right sequence [25, 26]. In tumor [11]. Recently, it has been suggested that combin- that sense, it is possible that pre-treatment with a VEGF- ing antiangiogenic therapies and immunotherapies might A inhibitor could exert an immuno-priming effect that increase the effectiveness of immunotherapy [12]. would increase the chance of response to subsequent im- VEGF-A (Vascular Endothelial Growth Factor-A), the mune checkpoint blockade and be more effective than main soluble factor responsible for tumor angiogenesis, is concurrent therapy [26], while causing less adverse events also involved in creating an immunosuppressive tumor [12]. We sought to explore a sequential chemotherapy- microenvironment: it prevents dendritic cell (DC) precur- free regimen combining the anti-PD-L1 agent durvalumab sors from evolving into mature antigen-presenting DCs with the antiangiogenic agent bevacizumab. In order to and promotes the development of tumor-associated mac- take advantage of the potential “sequential-effects”, pa- rophages [13, 14]. It also fosters the proliferation of im- tients had to have experienced disease progression while munosuppressive cells, limits T cell recruitment into on bevacizumab monotherapy prior to entering the trial tumors, and promotes T cell exhaustion [15]. Tumor ves- (as maintenance treatment after a chemotherapy plus bev- sel and stromal abnormality correction (which can be acizumab regimen). In addition, we sought to find traits in achieved by antiangiogenics) can restore the immune peripheral-blood mononuclear cells (PBMCs) and tumor function [16]. Preclinical evidence in breast, pancreatic, biopsies that informed about the immune priming, in an and brain tumors and melanoma [17, 18] and clinical re- attempt to search for novel biomarkers and obtain a better sults in renal cell cancer [19] have shown that antiangio- understanding of the biology of this clinical scenario. genic treatment was followed by more efficient T cell extravasation and tumor infiltration, turning immuno- Methods logically ‘cold’ into ‘hot’ tumors [17, 18]. Thus, combining We performed a prospective, open-label, multicentric, antiangiogenics with immune checkpoint inhibitors single-arm, phase IB investigator-initiated study. The Quintela-Fandino et al. Breast Cancer Research (2020) 22:124 Page 3 of 14 study was conducted in accordance with the Declaration CTC AE V.4.03 (Common Terminology Criteria for Ad- of Helsinki and Good Clinical Practice standards and verse Events). Figure 1 depicts basic trial design. registered at Clinicaltrials.gov (NCT02802098). Ethics A peripheral blood sample was obtained on the day of approval was obtained from the Madrid Regional Gov- the first durvalumab dose in order to analyze peripheral- ernment Ethics Board for Drug Research and the blood mononuclear cells (PBMCs) subpopulations by Spanish Agency for Medicine and Health Products flow cytometry (FACS). Blood sampling for FACS ana- (AEMPS). lysis was repeated on weeks 4, 8, 12, 16, 20, and 24 and at the end-of-treatment visit. Patients underwent a Study population tumor biopsy between 5 and 7 days prior to the first Women ≥ 18 years old were eligible if they had a non- study treatment. curable, locally advanced or metastatic HER2-negative PBMCs immunophenotyping was performed as fol- histologically confirmed breast cancer. Given the aca- lows: blood (~ 30 ml) was collected in CPT Cell Prepar- demic nature of this study, and the limited funding as a ation Tubes (BD) and PBMC isolated by centrifugation. result of being an independent study, in the context of Immunophenotyping was performed by staining 2 × 106 current FDA status for bevacizumab, the only option to PBMCs with a combination of antibodies (Table S1) execute a pilot trial in order to detect a signal of “prim- proposed by the Human Immunology Project Consor- ing effect” was to nest a state-of-the-art set of correlative tium to characterize T cell subtypes, regulatory T cells studies in patients that were still receiving bevacizumab (TREG), dendritic cells (DC), monocytes, and natural as a part of previously FDA-approved bevacizumab+ killer (NK) cells [27]. Stained samples were acquired in a chemotherapy combination regimens followed by beva- Gallios cytometer (Beckman Coulter); the gating strategy cizumab maintenance. A clinical trial where candidates for each panel was performed with the Kaluza software were started on bevacizumab monotherapy was deemed and is shown in Supplementary Figure S1, S2, S3. The unrealistic and possibly unethical by the investigators, ratio between effector and non-effector T cell subtypes and thus, the key inclusion criterion was that it was was calculated by considering the non-effector group the mandatory to have been receiving bevacizumab mono- sum of the percentages of naïve and central memory T therapy maintenance for a minimum of 6 weeks as a cells (TCM), whereas the effector group was constituted part of any bevacizumab-containing regimen prior to by the sum of the percentages of terminally differenti- registration, but there was no limit to the number of ated effector (Teff) and effector memory T cells (TEM). previous treatment lines. The only restriction was that Regarding immunohistochemistry studies, archival and no previous bevacizumab-containing regimens were baseline tumor samples were assessed centrally for PD- allowed (i.e., it had to be the first bevacizumab- L1 expression with the Ventana SP263 assay (Ventana containing regimen that the patients were exposed to). Medical Systems, Inc., Oro Valley, Arizona). Immuno- Other inclusion criteria included negative pregnancy histochemical stainings were performed on 2.5-μm tissue test, ECOG (Eastern Cooperative Oncology Group) 0–1, sections, using an automated protocol developed for the adequate organic function (according to standard defini- Autostainer Link automated slide staining system tions), lack of uncontrolled brain metastases, LVEF (Left (DAKO, Agilent). All steps were performed on this Ventricular Ejection Fraction) > 45%, and life expectancy staining platform using validated reagents, including > 24 weeks. Patients suffering from concurrent severe deparaffinization, antigen retrieval (cell conditioning), conditions, taking anticoagulants, receiving immunosup- and antibody incubation and detection. The following pressants, or with a history of clinically significant bleed- antibodies were used for IHC: CD8 FLEX (Clone C8/ ing or thromboembolic events within 6 months or study 144B, DAKO; #IR623/IS623), CD4 FLEX (Clone 4B12, entry were excluded. Previous treatment with immuno- DAKO; #IR649), and FoxP3 (Clone 236A/E7, Abcam; therapy was not allowed. Non-measurable disease (e.g., ab20034). Corresponding tissue sections were acquired bone disease only) was allowed. and digitalized using the AxioScan.Z1 system (Zeiss). Digitalized images were automatically analyzed with the Trial design and correlative studies ZEN 2.3 lite software (Zeiss). The percentage of FoxP3 The treatment consisted of 10 mg/kg i.v. of durvalumab positivity was considered as ratio of FoxP3-positive cells plus 10 mg/kg i.v. of bevacizumab every 2 weeks. One on total number of cells. cycle was defined as 28 days. Hormone-positive patients Combined pericyte and endothelial cell staining for were not allowed to receive hormonal therapy while on assessing vessel architecture normalization was per- trial. CT (computed tomography) scans were performed formed with confocal microscopy. For immunofluores- every 8 weeks and disease burden was evaluated accord- cence staining, deparaffinization and antigen retrieval of ing to RECIST 1.1 criteria (Response Evaluation Criteria 2.5-μm tissue sections were performed. Then slides were in Solid Tumors). Toxicity was graded according to NCI washed twice with TBS-T (0.1% Tween 20), and then Quintela-Fandino et al. Breast Cancer Research (2020) 22:124 Page 4 of 14 Fig. 1 Trial design. Patients had to be receiving maintenance treatment with bevacizumab alone as a part of a previous bevacizumab-containing regimen for advanced disease, after discontinuing the companion agent because of the usual clinical practice reasons (cumulative toxicity or achievement of maximal disease response). Pre-trial bevacizumab maintenance was allowed at 5 mg/kg weekly, 10 mg/kg q2w, or 15 mg/kg q3w. When patients experienced disease progression while on bevacizumab treatment become candidates for the trial. Bevacizumab treatment was never stopped, and the first durvalumab dose was scheduled for infusion on the next planned bevacizumab dose. All patients switched to a 10 mg/kg weekly bevacizumab schedule in case they were receiving it on a different one. A fresh tumor biopsy was obtained within a time-window of 7 days prior to the first durvalumab dose. In addition, a PBMC sample was obtained on day 1 prior to the first durvalumab dose, and repeated periodically until disease progression; an additional sample was harvested at the end-of-treatment visit (28 days after coming-off trial). Treatment continued until disease progression, unacceptable toxicity or investigator decision blocked in TBS-T with 2.5% bovine serum albumin for sourceforge.net/projects/bbmap/) to remove adapter se- 30 min before the indicated primary antibody was ap- quences and polyA tails. Trimmed reads are aligned to plied. Endothelial cells were detected with a mouse the GRCh38/hg38 genome assembly with STAR v 2.5. monoclonal antibody against CD31/PECAM-1 (clone Read counting is performed with HTSeq and differential JC/70A, NovusBio; #NB600-562) and pericytes were de- gene expression analysis, between the responder and tected with a rabbit monoclonal antibody against NG2 non-responder conditions, is done with DESeq2. Gene (clone EPR9195, Abcam; #ab139406). Cells were then set enrichment analysis (GSEA) versus the Immune- washed with TBS-T and incubated with Alexa Fluor SigDB molecular signatures database [28] was performed 488- or 555-conjugated secondary antibody (Molecular on a ranked list of DESeq2 data, where log2FC of genes Probes). Cell nuclei were stained with DAPI (4’,6-diami- showing greater than 1.2 absolute fold change was di- dino-2-fenilindol) (Sigma). After washing, coverslips vided by their corresponding p value. were mounted onto glass microscope slides with Pro- Long™ Gold antifade reagent (Invitrogen). Images were Aims and statistical analysis acquired using Leica-TCS SP5 MP confocal microscope, The sample size was designed to study immunodynamics with a HCX PL APO 63× 1.4 NA oil-immersion object- in peripheral blood and in tumor, while gathering effi- ive using LAS AF version 2.6 software. cacy and toxicity data. Thus, the primary aim was to Finally, gene-expression studies in tumor samples were compare different PBMC subpopulations (at baseline or performed with RNAseq: total RNA from formalin-fixed during treatment) among patients showing benefit or paraffin-embedded tissue samples was isolated using the not from the combination. Since the patients enrolled in Maxwell® RSC RNA FFPE kit (#AS1440, Promega) with this trial were at advanced treatment lines and one of the Maxwell® RSC Instrument (Promega) according to the two study drugs was one to which they had already the manufacturer’s protocol. Sequencing libraries were experienced treatment failure (bevacizumab), we consid- prepared with the “QuantSeq 3′ mRNA-Seq Library ered that those patients that did not experience disease Prep Kit” (Lexogen, Cat.No. 015). Directional cDNA li- progression at the standard landmark for immune- braries are initiated by reverse transcription with oli- oncology drugs evaluation time (16 weeks) were experi- godT priming and eventually sequenced in single-read encing clinical benefit of the study drugs. Those patients format in a HiSeq 2500 instrument (Illumina). were termed “non-progressors,” as opposed to those Sequencing read alignment and quantification and dif- showing progressive disease at week 16 (“progressors”) ferential gene expression analysis was performed in the for biomarker analysis purposes. Sample size was deter- Bluebee Genomics Platform, a cloud-based service pro- mined according to the ability to detect at least a 10% vider (www.bluebee.com). Briefly, reads are first trimmed difference in any given PBMC subpopulation between with bbduk from BBTools (Bushnell B., BBMap, https:// progressors and non-progressors. Setting alpha and beta Quintela-Fandino et al. Breast Cancer Research (2020) 22:124 Page 5 of 14 in 5% and 80%, respectively, a minimum of 24 patients Results was deemed necessary to discriminate such effect. All Patients and treatment patients receiving at least one dose of study treatment From June 2016 to July 2018, 26 patients were ac- were included in the safety and efficacy analysis. crued at the 8 study sites. Their basic demographic The co-primary aim was to determine the overall survival and clinical characteristics are shown in Table 1. (OS) and progression-free survival (PFS). Secondary aims One patient was found to not having a documented were to study the safety and toxicity of the combination. disease progression to bevacizumab maintenance The differences between the percentages of leukocyte sub- before entering the trial and was excluded from the types between progressors and non-progressors were com- analysis. On average, patients had been exposed to pared using a two-tailed Student’s t test after Arcsin 8.4 months of continuous bevacizumab dosing be- transformation of the data. Variances were compared with the fore registration. Patients had been on up to 7 lines F test, and Welch’s correction applied when suspected to have of therapy for metastatic disease. Approximately 2/3 not-equal standard deviation. CD4, CD8, and Tregs infiltration; and 1/3, respectively, were hormone-positive and PD-L1 staining; and vascular normalization were compared triple-negative breast cancer patients (Table 1). A with a T-test. OS and PFS estimates were compared with the CONSORT diagram describing trial accrual and log-rank test and Kaplan-Meier curves. All tests were two- populations for safety, efficacy, and correlative stud- tailed and performed with SPSS V.19 software. ies is shown in Fig. 2. Table 1 Demographic and baseline clinical characteristics Characteristic Value Age (median, range) 54.1 (34.5–77.4) Tumor subtype Hormone-positive 16 (64%) TNBC 9 (36%) Histology Ductal 18 (72%) Lobular 4 (16%) Other 3 (12%) Time (months) from diagnosis of metastatic disease (median, range) 18.1 (2–116) Time (months) receiving bevacizumab before registration (median, range) 8.4 (1.6–24.7) Number of previous treatment lines for metastatic disease 1 3 (12%) 2 9 (36%) 3 3 (12%) 4 7 (28%) 5 2 (8%) 7 1 (4%) Companion drug in the pre-trial regimen: Capecitabine 13 (52%) Paclitaxel 10 (40%) Docetaxel 1 (4%) Cisplatin 1 (4%) ECOG 0 15 (60% ECOG 1 10 (40%) PD-L1 expression: Positive (> 1%) 4 (16%) Negative (0%) 16 (64%) Unknown 5 (20%) Quintela-Fandino et al. Breast Cancer Research (2020) 22:124 Page 6 of 14 Fig. 2 CONSORT diagram. Twenty six patients underwent trial screening but one patient was deemed ineligible since she was not found to have documented progressive disease to ongoing bevacizumab maintenance. The rest of the patients (N = 25) received at least 1 durvalumab dose and were included in the safety and efficacy analysis. All patients had at least one baseline PBMCs sample for the immune subpopulation analysis; of them, only 6 consented for the pre-treatment tumor biopsy that was subsequently used for immunohistochemistry and gene-expression studies Efficacy Patients experiencing SD or PR showed similar disease- Database lock was performed on February 2019, when pa- control duration (Fig. 3e). No complete responses were tients had been receiving the study regimen for a median observed. Best objective overall responses were as follows: of 13 weeks (range 4.1 to 57.3 weeks). At this timepoint, 5 PR was observed in 2 patients (8%); SD was registered in patients remained on active treatment for 24+, 28+, 28+, 10 (40%); with the exception of one patient that did not 32+, and 58+ weeks. Median progression-free survival reach the first CT scan (4%, non-evaluated; clinical PD), (PFS) and overall survival (OS) were 3.5 months and 11 the remaining 12 patients (48%) experienced PD (total months, respectively (Fig. 3a and b). PFS was not signifi- number of PD: N = 13; 52%). According to the hormonal cantly different for TNBC (triple-negative breast cancer) status, hormone-positive patients achieved SD in 6/16 and hormone-positive patients (Fig. 3a). Regarding OS, cases (37.5%), whereas the rest of the patients (10/16; there was a trend for improved outcomes for the 62.5% patients) showed PD as best response (one of them hormone-positive patients (median OS for TNBC, 7.4 was not evaluated but experienced clinical PD). The two months; median OS for hormone-positives, 19.8 months; observed responses (2/9, 22.2%) occurred in TNBC pa- P = 0.11; Fig. 3b). The best changes in disease burden and tients; 2 additional TNBC patients (22.2%) experienced disease burden changes over time among the 21 patients SD, whereas the remaining TNBC patients (N = 5; 55.6%) with measurable disease are shown in the waterfall plot experienced PD as best response. The comparison be- and spider plot (Fig. 3c and e). Overall, 28% of the patients tween TNBC and hormonal patients regarding the CBR (N = 7) experienced some degree of tumor shrinkage (al- did not show statistically significant differences (P = 0.73). though 5 of the 7 patients with tumor shrinkage did not achieve partial response criteria according to RECIST 1.1; Differences in peripheral blood mononuclear cells Fig. 3c). The changes in tumor burden were sustained for (PBMCs) immunophenotyping between progressors and at least two cycles among most of the patients achieving non-progressors SD or PR, as shown in Fig. 3d. Clinical benefit (CR+PR+ Seven patients were classified as “non-progressors” whereas SD) was observed in 15/25 patients (clinical benefit rate − eighteen were classified as “progressors” (statistics section). CBR − 60%) at 8 weeks; at 16 weeks, the CBR was 44%. The PBMCs analysis of the baseline samples (before the Quintela-Fandino et al. Breast Cancer Research (2020) 22:124 Page 7 of 14 Fig. 3 Overall efficacy data. a Kaplan-Meier curves (PFS) for the whole population (left) and split by hormonal subtype (TNBC, 2.6 months; hormone-positive, 3.3 months; log-rank P value, 0.84). b Kaplan-Meier curves (OS) for the whole population (left) and split by subtype (TNBC, 7.4 months; hormone-positive, 19.8 months; log-rank P value, 0.11). c Waterfall plot showing best percentage change from baseline in the sum of the longest diameters of target lesions. This plot depicts the changes among the N = 21 patients with measurable disease; the remaining patients (N = 4) had bone disease only (non-measurable but evaluable disease, according to RECIST 1.1). The patient that was non-evaluated due to clinical PD was one of the patients with non-measurable disease. d Swimmer plot depicting the time and durability of response of the 25 patients included in the trial. e Spider plot displaying the longitudinal change from baseline in the sum of the longest diameters of target lesions. It depicts the changes among the same N = 21 patients as in (c). Patients in (c), (d), and (e) are labeled according to their hormonal status: red dot, hormone-positive; black dot, TNBC first durvalumab dose; N = 25) yielded the following results: immune populations (Figure S4B, C). Conversely, detailed first, no differences in the major generic populations (T analysis of T cell subtypes showed an unusually high fre- cells, monocytes, DCs and NK cells) among progressors quency of CD4+ T cells in the blood of non-progressors and non-progressors (Figure S4A) were found. In addition, compared to that of progressors (Fig. 4a). Interestingly, we no baseline differences were evident either in innate observed a strong reduction in the baseline percentage of Quintela-Fandino et al. Breast Cancer Research (2020) 22:124 Page 8 of 14 Fig. 4 Immune cell sub-populations associated to clinical benefit in peripheral blood mononuclear cells. a Frequency of T-cell subtypes determined by immunophenotyping of PBMC from peripheral blood in non-progressors (red) and progressors (blue) patients in the baseline sample. b Differential Treg percentage in baseline samples among progressors and non-progressors. c Non-progressors displayed a significant increase in central memory—and a trend to increase in effector memory—CD8+ T-cells 4 weeks after the first treatment dose, compared with progressors. d These changes were not evident for CD4+ T cells circulating immunosuppressive T regulatory (Treg) cells in subsets (Fig. 4d). No other clear differences between non- non-progressors compared to progressors (Fig. 4b). No progressors and progressors in the frequency of specific T- other differences between progressors and non-progressors cell or innate subpopulations (Figure S5) were found. were evident for naïve, effector (Teff), effector memory (TEM) and central memory (TCM) CD4+ or CD8+ T cell Gene-expression patterns and vascular normalization in subtypes (Figure S4D, E). The ratio of effector (Teff + TEM) tumors from progressors and non-progressors and non-effector (Tnaïve + TCM) populations were not corre- In an attempt to understand the PBMCs findings in the lated with the clinical outcome (Figure S4F, G) either. context of the mechanism of action of the combination, Regarding the longitudinal analyses in on-treatment sam- we studied tumor biopsies. Six patients (3 non- ples, it was revealed that CD8+ TEM and CD8+ TCM were progressors and 3 progressors) had available a baseline increased after the first durvalumab dose compared to the tumor biopsy (i.e., after bevacizumab exposure and before baseline levels (Fig. 4c); this pattern was attenuated along the first durvalumab dose). Non-progressors’ tumors did subsequent cycles and not observed for CD4+ memory not show significant differences in CD8, CD4, or Treg Quintela-Fandino et al. Breast Cancer Research (2020) 22:124 Page 9 of 14 infiltration compared to the progressors (Fig. 5a–c). In HER2-negative breast cancer patients. Preclinical research addition, the six baseline biopsies were negative for PD-L1 suggests that antiangiogenic agents could exert a priming staining (Fig. 5d). Morphologically, however, the examin- of the immune microenvironment [12–15, 17–19, 30]. Par- ation of the vascular network of the baseline samples ticularly, the sequence in which these agents are adminis- displayed pericyte coverage of the endothelial cells in non- tered (sequential versus concurrently with immunotherapy) progressors as opposed to progressors (Fig. 5e), indicating might be of key importance [25, 26]. The efficacy assess- vascular normalization in response to pre-trial bevacizu- ment and the correlative studies results deserve attention. mab exposure in the former [29]. In our heavily pretreated population, we observed a Despite the lack of meaningful morphological or nu- PFS of 3.5 months and an OS of 11 months. The re- merical differences in the tumor immune infiltrate, gene sponse rate according to RECIST 1.1 was low (8%), but expression revealed profound differences in its func- the combination achieved a CBR of 60% at 8 weeks and tional status. When gene expression levels were queried 44% at 16 weeks. The CBR was not significantly different for gene set enrichment (GSEAs) against the ImmuneSig among TNBC- and hormone-positive patients. Current database [28], we found the following differences in im- evidence suggests that immune checkpoint inhibitors are munological pathways between non-progressors and more effective in TNBC [31, 32]. We only observed progressors: First, congruently with the observations in tumor responses in TNBC; conversely, OS was longer in peripheral blood, non-progressors displayed a reduction the hormonal subtypes. Our sample size precludes de- of GSEAs related with suppressive activity of FoxP3 Treg finitive conclusions about the long-term efficacy of the cells, but an enrichment in signatures related to stimu- combination, particularly regarding a selective effect of lated CD8 T cells (Fig. 5f and h), which suggest an in- the combination over one subtype or the other. Most creased effector activity in the tumor tissue. Conversely, likely, such overall survival differences may be explained enrichment in active FoxP3 Treg and inactive CD8 cells by the intrinsic biologic differences between TNBC and signatures were observed in progressors (Fig. 5g and h). the hormonal subtype, which is more benign and usually In addition, we observed an enrichment in GSEAs of un- displays a 2- to 3-fold longer overall survival in the stimulated resting DCs in progressors versus non- metastatic setting. Regardless, the CBR compares favor- progressors’ tumors (Fig. 5h). Finally, memory T cell ably in relation with the existing evidence with PD-1/L1 GSEAs were increased in non-progressors versus pro- inhibitors monotherapy in similar populations. Although gressors patients (Fig. 5f and h). trial-to-trial comparisons have to be performed with ex- treme caution, the results of PD-1/L1 inhibitors mono- Safety therapy trials are as follows: in a single-agent One hundred and six cycles have been administered. pembrolizumab study in PD-L1, hormone-positive breast Table 2 shows the incidence of grade 1–2 adverse events cancer showed a 20% 16-weeks CBR [4]. Two trials that related with the study drugs. There were two grade 3 accrued a mixed population of hormone-positive and events possibly related to the study drugs, registered in TNBC patients, testing the efficacy of durvalumab plus the same patient: a patient with lung lymphangitic carcin- tremelimumab [5] and avelumab [3] respectively, achieved omatosis and previous history of asthma and smoking a < 30% 4-month CBR. In a monotherapy trial with pem- habit was admitted with grade 3 respiratory insufficiency, brolizumab in previously untreated TNBC, the disease grade 3 bronchospasm, and grade 2 infection during cycle control rate was 23.8% [1], whereas in a previously treated 2. She received supportive measures, antibiotics, and pred- cohort this rate decreased to just 7.6% [2]. The 4-month nisone and continued on durvalumab + bevacizumab after CRB observed in our heavily pretreated cohort is encour- discharge without further side effects. The patient was di- aging and could be supporting the immuno-priming ef- agnosed as COPD exacerbation; thus, the relationship fects of bevacizumab. However, our design does not with the study drugs remains unclear. No grade 4 events distinguish between “priming” effects (versus no-priming) were registered. Four events qualified as severe adverse or even whether it is justified or not to maintain bevacizu- events (SAEs): three of them each registered and deemed mab in the combination phase. A definitive conclusion in related to disease progression in three patients (pain wors- this regard would require a 2 × 2 randomized design com- ening, pancreatitis in a patient with a pancreatic metasta- paring the inclusion or not of a bevacizumab priming sis, and general status decline); the last SAE corresponded phase followed by durvalumab monotherapy or in com- to the patient with grade 3 side effects. bination with bevacizumab, together with pre-priming, pre-durvalumab, and on-treatment PBMC serial sampling. Discussion A recent randomized trial in advanced HER2-negative dis- In order to ascertain the potential priming role of antian- ease without targetable molecular alterations in patients giogenic treatment for subsequent immunotherapy, we not showing progressive disease after 6–8 chemotherapy conducted a pilot biomarker and efficacy study in advanced cycles found a lower progression-free survival but longer Quintela-Fandino et al. Breast Cancer Research (2020) 22:124 Page 10 of 14 Fig. 5 Tumor tissue immunodynamics and gene expression studies suggest immuno-priming by bevacizumab. Representative immunohistochemistry images of a CD8+ infiltration, b CD4+ infiltration, and c Treg infiltration in tumors from non-progressor and progressors patients (upper panels). The lower panels represent the geometric mean and standard deviation of the quantitation data from all available tumor biopsies. d The six analyzed tumors were negative for PD-L1 expression (applying the 1% boundary), regardless of experiencing benefit or not from the treatment combination. e Confocal imaging showing representative fields containing normalized blood vessels from a non-progression (the whole microvessel wall—CD31-positive endothelial cells—is covered by NG2-positive pericytes) and a progressor patient (who, in turn, displays vessel abnormality—lack of pericyte coverage and tortuous architecture). The chart represents the quantitative differences between the average percentage of microvessel wall covered by pericytes in non-progressors versus progressors; **P < 0.01. f Functionally representative GSEAs of the main regulated pathways in non-progressors’ tumors; NES, normalized enrichment score (the higher NES, the higher functional enrichment); both corrected (false discovery ratio (FDR)) and uncorrected P values are shown. g Same as in (f) for non-responders. h Further enriched GSEAs in responders and non-responders ranked by their NES; all of them with FDR < 0.001. Scalebars: a–c 20 μm; d 100 μm; e 25 μm Quintela-Fandino et al. Breast Cancer Research (2020) 22:124 Page 11 of 14 Table 2 Adverse events (grades 1 and 2) these partners in for combination with immunotherapy in Event Grade 1 Grade 2 Total future trials in breast cancer exploring this concept. Asthenia 4 (16%) 1 (4%) 5 (20%) The immune-priming hypothesis to explain the clinical Headache 2 (8%) 1 (4%) 3 (12%) benefit from the combination is further supported by the immunologic biomarkers found in peripheral blood. Al- Proteinuria 2 (8%) 0 (0%) 2 (8%) though current evidence of a response-prone immune- Hypertension 1 (4%) 1 (4%) 2 (8%) environment (in PBMCs) before treatment with PD-1/L1 Diarrhea 2 (8%) 0 (0%) 2 (8%) inhibitors is lacking in breast cancer [1–5], studies in per- Pruritus 0 (0%) 1 (4%) 1 (4%) ipheral blood in other tumor types have shown bio- Hyporexia 0 (0%) 1 (4%) 1 (4%) markers of benefit from immunotherapy [37–39]. Our Nausea 0 (0%) 1 (4%) 1 (4%) patient population had been exposed to up to 25 months to bevacizumab, and at this point, patients that experi- Pneumonitis 0 (0%) 1 (4%) 1 (4%) enced benefit from subsequent durvalumab displayed an Thrombosis 0 (0%) 1 (4%) 1 (4%) obvious phenotype consisting in a decreased amount of Infection 0 (0%) 1 (4%) 1 (4%) immunosuppressive Treg cells in peripheral blood (Fig. 4). Elevated amylase 1 (4%) 0 (0%) 1 (4%) Those patients that had available tissue biopsy showed a Elevated lipase 1 (4%) 0 (0%) 1 (4%) highly congruent gene-expression program, consisting on Rash 1 (4%) 0 (0%) 1 (4%) decreased expression of gene sets related to Treg cells in the tumor microenvironment (Fig. 5f–h). Different studies Xerosis 1 (4%) 0 (0%) 1 (4%) have established a correlation between aberrant VEGF ex- Hypothyroidism 1 (4%) 0 (0%) 1 (4%) pression and the proliferation of Treg cells in the tumor Abdominal pain 1 (4%) 0 (0%) 1 (4%) microenvironment [40–42]. Our results not only support Vomiting 1 (4%) 0 (0%) 1 (4%) this local intratumor effect of VEGF blockade on Treg ex- Pyrexia 1 (4%) 0 (0%) 1 (4%) pansion, but suggest that this effect might be systemic Arthralgia 1 (4%) 0 (0%) 1 (4%) given the low frequency of Treg cells in peripheral blood of non-progressors as well. A future randomized study Xerophtalmia 1 (4%) 0 (0%) 1 (4%) should confirm whether or not the low number of Treg cells in these patients is a direct consequence of previous overall survival of durvalumab compared to maintenance bevacizumab therapy. chemotherapy apparently due to the TNBC patients that Vascular normalization is the expected positive effect were PD-L1 positive [33]. Thus, the role of durvalumab is that translates pharmacodynamic engagement by antian- still unclear in advanced HER2-negative breast cancer, giogenics [16]. The fact that the patients that experienced and a randomized study should have a sufficiently large benefit from the combination and showed the described T size in order to be able to stratify patients according to cell phenotype in tumor and peripheral blood were also their hormonal status and PD-L1 expression. Given the those that experienced a vascular normalizing response to current indications of bevacizumab in breast cancer, and the previous bevacizumab exposure (Fig. 5e) further supports fact that the present study was an investigator-initiated trial the hypothesis that antiangiogenics can exert an immuno- with limited funding, our study had to rely necessarily on pa- priming effect, at least in some cases. Obviously, not all tients currently receiving maintenance bevacizumab and seek patients experience vascular normalization in response to for a signal of biologic plausibility for the priming effects in antiangiogenics, which seems to be a factor required for the set of correlative studies. That restrictive inclusion cri- immuno-priming [17, 18, 30]. We have previously shown teria may constitute on itself a selection bias towards patients that the normalization rate is agent-dependent (ranging with relatively indolent disease, who do not show immediate from 1/3 of the patients with bevacizumab [43] to up to ¾ progression while in maintenance bevacizumab (evidenced of the patients with nintedanib [44], in HER2-negative by the average time of 8.4 months before entering the trial breast cancer). These differences should be taken into ac- under bevacizumab-based regimens; Table 1). In fact, bevaci- count for designing future trials. In any case, future zumab itself might not be the most effective VEGF-targeting biomarker-based studies in this context should be guided drug [21–24] at least in breast cancer, but the investigator- by peripheral blood sampling; access to fresh tumor biop- initiated nature of this trial made bevacizumab the only real- sies in metastatic patients is usually limited (only 25% of istic antiangiogenic option for the study. The impressive our patients consented to fresh biopsy), and thus, conclu- effects of other antiangiogenics with high affinity against sions may not be extrapolable to real-world patients. VEGFR2 such as lenvatinib (in gastric [34] and endometrial Finally, the observed GSEAs suggesting a dysfunctional cancer [35]) or axitinib (in kidney cancer [36]) in combin- DC phenotype in tumors from progressors (Fig. 5h) are ation with pembrolizumab should make at least consider congruent with the role of VEGF in DCs’ maturation Quintela-Fandino et al. Breast Cancer Research (2020) 22:124 Page 12 of 14 (progressors also did show vascular abnormality, trans- associated to activated Teff cells and reduced immunosup- lating inadequate effect in clearing VEGF by bevacizu- pressive Treg in their tumors, suggests a positive immune- mab, which alters DC maturation) and the lack of priming effect of bevacizumab. Taken together the positive benefit from durvalumab. Tolerogenic DCs are associ- efficacy signal, the low toxicity rates, the finding of candidate ated to the expansion and differentiation of Treg cells biomarkers of activity in peripheral blood, and the biological [45]. The observed decrease of Treg signatures in the rationale, our data justify a larger prospective clinical trial tumor microenvironment might thus be associated to aiming to define the magnitude of the benefit derived from improved DC maturation. antiangiogenic priming in this setting. The main weaknesses for interpretation of our data are patient heterogeneity and the limited sample size. The Conclusions heterogeneity was not only originated by mixing TNBC The combination of bevacizumab and durvalumab and hormone-positive patients: trial patients received a shows promising efficacy in a heavily pre-treated cohort variable number of treatment lines prior to maintenance of HER2-negative breast cancer patients previously ex- with bevacizumab and also were on maintenance for posed to chronic bevacizumab. Peripheral blood and highly variable times (from 1.6 to 24.7 months). Whereas tumor immunodynamics suggest that pre-treatment with patients with many previous treatment lines and long bev- bevacizumab immuno-primes at least a fraction of pa- acizumab maintenance time might be inherently more in- tients, making them more prone to benefit from durva- dolent than the remaining, disease progression also selects lumab. A prospective randomized trial should confirm for cancer features associated with refractoriness to im- the role of Tregs as potential biomarkers of activity and munotherapy [46–48]. Trials evaluating PD-1/L1 inhibi- the magnitude of benefit of this chemo-free regimen. tors in more advanced disease lines have shown limited activity compared with earlier lines [1, 2, 31]. Taken to- Supplementary information gether, these features might complicate the interpretation Supplementary information accompanies this paper at https://doi.org/10. of the overall efficacy data and the immunology correlative 1186/s13058-020-01362-y. studies, at the PBMCs and tumor infiltration levels. Importantly, no new safety concerns related to durvalu- Additional file 1 : Supplementary Figure S1: Flow cytometry gating strategy for T cells. Representative gating strategy used in T cell mab were detected (Table 2). Although the sample size is immunophenotyping based on the exclusion of dead cells, the selection relatively low, all the observed toxic events fit into those of the lymphoid cells by size and complexity, and then the surface expected either for bevacizumab or durvalumab and do expression of CD3, CD4, CD8, CCR7 and CD45RA. The combination of these markers allowed the identification of naïve cells (CCR7+CD45RA+), not suggest any negative drug-drug toxic interaction. It Teff (CCR7−,CD45RA+), TCM (CCR7+CD45RA−), and TEM (CCR7−CD45RA−). has been proposed that antiangiogenics can decrease the Staining with the CD38 marker allowed the identification of activated amount and severity of immune-related adverse events cells, which were minimally detected only in Teff and TEM subpopulations. (IRAEs) [12]. This trial was not designed for prospective Additional file 2 : Supplementary Figure S2: Flow cytometry gating strategy for Treg cells. Representative gating strategy used in Treg cell identification of IRAEs; however, the 8% incidence of diar- immunophenotyping based on the exclusion of dead cells, the selection rhea and 4% incidence of rash (plus 2 grade 3 respiratory of the lymphoid cells by size and complexity, and then the surface events in the same patient, with unclear relationship to expression of CD3, CD4, CD25, CD127, CCR4, HLA-DR and CD45RO markers. After the initial selection of CD4+ T cells (CD3+CD4+), Treg cells study drugs)—the side effects traditionally most frequently were identified as CD25+CD127low double positive cells. The combination classified as IRAEs—is not particularly high. of CCR4 and CD45RO markers allowed the identification of naïve Treg Several trials have studied the simultaneous administration (CCR4+,CD45RO−) and memory Treg (CCR4+CD45RO+), whereas HLA-DR positivity identified activated cells in these Treg subpopulations. of antiangiogenics plus immune checkpoint inhibitors in dif- Additional file 3 : Supplementary Figure S3: Flow cytometry gating ferent malignancies [36, 49, 50]. To our knowledge, we are strategy for innate cell populations. Representative example of the gating the first in reporting on the sequential combination in breast strategy used for immunophenotyping of the indicated innate cancer. Our immunophenotyping analyses highlight the pre- subpopulations. After the exclusion of dead cells, monocytes were selected by size and complexity; subtypes were identified by CD14 and dictive role of specific T cell subtypes for the clinical re- CD16 staining, allowing the identification of classical monocytes (CD14+ sponse to PD-L1 blockade, pointing towards several effector CD16−), alternative monocytes (CD14−CD16+) and intermediate T-lymphocyte subpopulations as biomarkers of activity monocytes. For DC and NK, leukocytes were selected by size and complexity in the live cells, and T- and B-lymphocytes excluded by stain- (Fig. 4). In the past, single-agent immunotherapy studies in ing with lineage-specific antibodies. Dendritic cells (DC) and NK cells breast cancer have not shown clear PBMC subpopulations were selected within the CD20− and CD14− population; NK cells were predicting benefit, suggesting that the response-prone envir- identified as CD56+ cells (both CD16+ and CD16−) and DC as HLA- DR+CD16− cells. DC subtypes were further defined by CD11c+ (myeloid onment is not obvious in peripheral blood. The fact that in DC) or CD123+ (plasmacytoid DC). our study non-progressors showed a differential PBMC pat- Additional file 4 : Supplementary Figure 4: Immunophenotyping of tern compared to the remaining patients after previous ex- leukocyte populations in the baseline sample of responders and non- posure to bevacizumab, together with the concurrence of responder patients. (A-D) Analysis of the indicated leukocyte subtypes in the baseline blood sample of the patients stratified according to their vascular normalization and gene expression programs Quintela-Fandino et al. Breast Cancer Research (2020) 22:124 Page 13 of 14 clinical response. The percentages of the main leukocyte subtypes (A), Author details CD4+ and CD8+ T cells (B), monocyte subtypes (C) and DC subtypes (D) 1 Breast Cancer Clinical Research Unit – Clinical Research Program, CNIO - are shown. Spanish National Cancer Research Center, Melchor Fernandez Almagro, 3, 28029 Madrid, Spain. 2Medical Oncology Department, Hospital Universitario Additional file 5 : Supplementary Figure 5: Longitudinal effects of de Fuenlabrada, Fuenlabrada, Spain. 3Medical Oncology Department, treatment on leukocyte populations. For each subpopulation, the time- Hospital Universitario Quiron, Pozuelo de Alarcon, Spain. 4Medical Oncology dependent increment or decrement during treatment is shown, using Department, Hospital Universitario Ramon y Cajal, Madrid, Spain. 5Medical the baseline sample as reference. Red charts: responders; blue charts: Oncology Department, Hospital Universitario 12 de Octubre, Madrid, Spain. non-responders. 6 Medical Oncology Department, Hospital Universitari Arnau Vilanova, Lleida, Additional file 6 : Supplementary Table 1: Antibodies used for Spain. 7Medical Oncology Department, Hospital Clínico Universitario, immunophenotyping in peripheral blood. Valencia, Spain. 8INCLIVA, Valencia, Spain. 9CIBERONC, Instituto Carlos III, Madrid, Spain. 10Medical Oncology Department, Hospital Universitario La Princesa, Madrid, Spain. 11Facultad de Medicina, Universidad Autónoma de Abbreviations Madrid, Madrid, Spain. 12Department of Immunology and Oncology, Centro PBMCs: Peripheral-blood mononuclear cells; OS: Overall survival; Nacional de Biotecnología/CSIC, Darwin, 3, 28049 Madrid, Spain. PFS: Progression-free survival; TMB: Tumor mutational burden; VEGF: Vascular 13 Histopathology Core Unit – Biotechnology Program, CNIO - Spanish Endothelial Growth Factor; DC: Dendritic cell; ECOG: Eastern Cooperative National Cancer Research Center, Madrid, Spain. 14Medical Oncology Oncology Group; LVEF: Left ventricular ejection fraction; CT scan: Computed Department, Hospital General Universitario de Valencia, Valencia, Spain. 15 tomography scan; RECIST 1.1: Response Evaluation Criteria in Solid Tumors; Medicine Department, Universitat de Valencia, Valencia, Spain. 16Pathology NCI CTC AE V.4.03: Common Terminology Criteria for Adverse Events; Department, Hospital Universitario 12 de Octubre, Madrid, Spain. 17Genomics FACS: Flow cytometry; CPT: Cell Preparation Tubes; TREG: Regulatory T cells; Core Unit – Biotechnology Program, CNIO - Spanish National Cancer NK: Natural killer cells; TCM: Central memory T cells; Teff: Terminally Research Center, Madrid, Spain. 18ION Institute of Oncology, Quironsalud differentiated effector T cells; TEM: Effector memory T cells; Group – Madrid & Barcelona, Barcelona, Spain. 19Vall d’Hebron Institute of IHC: Immunohistochemistry; DAPI: 4’,6-Diamidino-2-fenilindol; GSEA: Gene set Oncology, Barcelona, Spain. enrichment analysis; TNBC: Triple-negative breast cancer; CBR: Clinical benefit rate; CR: Complete response by RECIST 1.1; PR: Partial response by RECIST 1.1; Received: 15 June 2020 Accepted: 25 October 2020 SD: Stable disease by RECIST 1.1; PD: Progressive disease by RECIST 1.1; COPD exacerbation: Chronic obstructive pulmonary disease exacerbation; SAEs: Severe adverse events; IRAEs: Immune-related adverse events References Acknowledgements 1. Adams S, Loi S, Toppmeyer D, Cescon DW, De Laurentiis M, Nanda R, Winer EP, Not applicable. Mukai H, Tamura K, Armstrong A, et al. Pembrolizumab monotherapy for previously untreated, PD-L1-positive, metastatic triple-negative breast cancer: Authors’ contributions cohort B of the phase II KEYNOTE-086 study. Ann Oncol. 2019;30(3):405–11. -Conception and design: MQF, LM, RC, SMo, and SMa. -Patient accrual: MQF, 2. Adams S, Schmid P, Rugo HS, Winer EP, Loirat D, Awada A, Cescon DW, EH, LM, SM, BB, RC, JVA, VI, JH, LGC, JC, DM, and AGC -Data acquisition and Iwata H, Campone M, Nanda R, et al. Pembrolizumab monotherapy for analysis: MQF, RB, MM, EC, MM, OD, ALA, MCMO, SMo, SMa. -Manuscript previously treated metastatic triple-negative breast cancer: cohort A of the writing and approval: all authors. phase II KEYNOTE-086 study. Ann Oncol. 2019;30(3):397–404. 3. Dirix LY, Takacs I, Jerusalem G, Nikolinakos P, Arkenau HT, Forero-Torres A, Boccia Funding R, Lippman ME, Somer R, Smakal M, et al. Avelumab, an anti-PD-L1 antibody, in MQF is a recipient of the following grants: AES - PI16/00354 funded by the patients with locally advanced or metastatic breast cancer: a phase 1b JAVELIN ISCIII and co-funded by the European Regional Development Fund (ERDF) Solid Tumor study. Breast Cancer Res Treat. 2018;167(3):671–86. and B2017/BMD3733 (Immunothercan-CM) - Call for Coordinated Research 4. Rugo HS, Delord JP, Im SA, Ott PA, Piha-Paul SA, Bedard PL, Sachdev J, Groups from Madrid Region - Madrid Regional Government - ERDF funds. Tourneau CL, van Brummelen EMJ, Varga A, et al. Safety and antitumor SM is a recipient of the following grants: Spanish Ministerio de Economía y activity of pembrolizumab in patients with estrogen receptor-positive/ Competitividad (MINECO) (SAF2017-83732-R; AEI/FEDER, EU) and co-funded human epidermal growth factor receptor 2-negative advanced breast by Comunidad de Madrid (B2017/BMD3733; Immunothercan-CM). RC is a re- cancer. Clin Cancer Res. 2018;24(12):2804–11. cipient RC is a recipient of the ISCIII grants PIE15/00068 and PI17/01865. The 5. Santa-Maria CA, Kato T, Park JH, Kiyotani K, Rademaker A, Shah AN, Gross L, study was also funded by CRIS Contra el Cancer Foundation and Astra Blanco LZ, Jain S, Flaum L, et al. A pilot study of durvalumab and Zeneca Spain. Astra Zeneca Spain provided durvalumab. tremelimumab and immunogenomic dynamics in metastatic breast cancer. Oncotarget. 2018;9(27):18985–96. 6. Adams S, Gatti-Mays ME, Kalinsky K, Korde LA, Sharon E, Amiri-Kordestani L, Availability of data and materials Bear H, McArthur HL, Frank E, Perlmutter J, et al. Current landscape of The results from gene-expression studies by RNA-seq in tumor samples were immunotherapy in breast cancer: a review. JAMA Oncol. 2019;5:1205–14. deposited in Gene Expression Omnibus (GEO, NCBI) under the accession 7. Buder-Bakhaya K, Hassel JC. Biomarkers for clinical benefit of immune number GSE139050 . checkpoint inhibitor treatment-a review from the melanoma perspective and beyond. Front Immunol. 2018;9:1474. Ethics approval and consent to participate 8. Havel JJ, Chowell D, Chan TA. The evolving landscape of biomarkers for Ethics approval was obtained from the Madrid Regional Government Ethics checkpoint inhibitor immunotherapy. Nat Rev Cancer. 2019;19(3):133–50. Board for Drug Research and the Spanish Agency for Medicine and Health 9. Mittendorf EA, Philips AV, Meric-Bernstam F, Qiao N, Wu Y, Harrington S, Su Products (AEMPS). X, Wang Y, Gonzalez-Angulo AM, Akcakanat A, et al. PD-L1 expression in All participants gave informed consent to participate before taking part of triple-negative breast cancer. Cancer Immunol Res. 2014;2(4):361–70. the study. 10. Ali HR, Glont SE, Blows FM, Provenzano E, Dawson SJ, Liu B, Hiller L, Dunn J, Poole CJ, Bowden S, et al. PD-L1 protein expression in breast cancer is rare, Consent for publication enriched in basal-like tumours and associated with infiltrating lymphocytes. Not applicable. Ann Oncol. 2015;26(7):1488–93. 11. Stambrook PJ, Maher J, Farzaneh F. Cancer immunotherapy: whence and Competing interests whither. Mol Cancer Res. 2017;15(6):635–50. MQF has received research funds from Astra Zeneca. MQF has received a 12. Fukumura D, Kloepper J, Amoozgar Z, Duda DG, Jain RK. Enhancing cancer Research Award from Astra Zeneca Foundation (“Jovenes Investigadores – immunotherapy using antiangiogenics: opportunities and challenges. Nat Cancer 2016”). Rev Clin Oncol. 2018;15(5):325–40. Quintela-Fandino et al. Breast Cancer Research (2020) 22:124 Page 14 of 14 13. Ohm JE, Carbone DP. VEGF as a mediator of tumor-associated results from phase II randomized trial SAFIR02-IMMUNO. SABCS Proc 2019, immunodeficiency. Immunol Res. 2001;23(2–3):263–72. 2019:GS3–G02. 14. Johnson B, Osada T, Clay T, Lyerly H, Morse M. Physiology and therapeutics 34. Kawazoe A, Fukuoka S, Nakamura Y, Kuboki Y, Wakabayashi M, Nomura S, of vascular endothelial growth factor in tumor immunosuppression. Curr Mikamoto Y, Shima H, Fujishiro N, Higuchi T, et al. Lenvatinib plus Mol Med. 2009;9(6):702–7. pembrolizumab in patients with advanced gastric cancer in the first-line or 15. Lapeyre-Prost A, Terme M, Pernot S, Pointet AL, Voron T, Tartour E, Taieb J. second-line setting (EPOC1706): an open-label, single-arm, phase 2 trial. Immunomodulatory activity of VEGF in cancer. Int Rev Cell Mol Biol. 2017; Lancet Oncol. 2020;21(8):1057–65. 330:295–342. 35. Makker V, Rasco D, Vogelzang NJ, Brose MS, Cohn AL, Mier J, Di Simone C, 16. Jain RK. Normalizing tumor microenvironment to treat cancer: bench to Hyman DM, Stepan DE, Dutcus CE, et al. Lenvatinib plus pembrolizumab in bedside to biomarkers. J Clin Oncol. 2013;31(17):2205–18. patients with advanced endometrial cancer: an interim analysis of a 17. Schmittnaegel M, Rigamonti N, Kadioglu E, Cassara A, Wyser Rmili C, multicentre, open-label, single-arm, phase 2 trial. Lancet Oncol. 2019;20(5): Kiialainen A, Kienast Y, Mueller HJ, Ooi CH, Laoui D et al: Dual angiopoietin- 711–8. 2 and VEGFA inhibition elicits antitumor immunity that is enhanced by PD- 36. Rini BI, Plimack ER, Stus V, Gafanov R, Hawkins R, Nosov D, Pouliot F, 1 checkpoint blockade. Sci Transl Med. 2017;9(385):eaak9670. Alekseev B, Soulieres D, Melichar B, et al. Pembrolizumab plus axitinib versus 18. Allen E, Jabouille A, Rivera LB, Lodewijckx I, Missiaen R, Steri V, Feyen K, sunitinib for advanced renal-cell carcinoma. N Engl J Med. 2019;380(12): Tawney J, Hanahan D, Michael IP et al: Combined antiangiogenic and anti- 1116–27. PD-L1 therapy stimulates tumor immunity through HEV formation. Sci Transl 37. Hodi FS, Lawrence D, Lezcano C, Wu X, Zhou J, Sasada T, Zeng W, Giobbie- Med. 2017;9(385):eaak9679. Hurder A, Atkins MB, Ibrahim N, et al. Bevacizumab plus ipilimumab in 19. Wallin JJ, Bendell JC, Funke R, Sznol M, Korski K, Jones S, Hernandez G, Mier patients with metastatic melanoma. Cancer Immunol Res. 2014;2(7):632–42. J, He X, Hodi FS, et al. Atezolizumab in combination with bevacizumab 38. Kamphorst AO, Pillai RN, Yang S, Nasti TH, Akondy RS, Wieland A, Sica GL, enhances antigen-specific T-cell migration in metastatic renal cell Yu K, Koenig L, Patel NT, et al. Proliferation of PD-1+ CD8 T cells in carcinoma. Nat Commun. 2016;7:12624. peripheral blood after PD-1-targeted therapy in lung cancer patients. Proc 20. Chen DS, Hurwitz H. Combinations of bevacizumab with cancer Natl Acad Sci U S A. 2017;114(19):4993–8. immunotherapy. Cancer J. 2018;24(4):193–204. 39. Krieg C, Nowicka M, Guglietta S, Schindler S, Hartmann FJ, Weber LM, 21. Miller KD, O'Neill A, Gradishar W, Hobday TJ, Goldstein LJ, Mayer IA, Bloom Dummer R, Robinson MD, Levesque MP, Becher B. High-dimensional single- S, Brufsky AM, Tevaarwerk AJ, Sparano JA, et al. Double-blind phase III trial cell analysis predicts response to anti-PD-1 immunotherapy. Nat Med. 2018; of adjuvant chemotherapy with and without bevacizumab in patients with 24(2):144–53. lymph node-positive and high-risk lymph node-negative breast cancer 40. Tada Y, Togashi Y, Kotani D, Kuwata T, Sato E, Kawazoe A, Doi T, Wada H, (E5103). J Clin Oncol. 2018;36(25):2621–9. Nishikawa H, Shitara K. Targeting VEGFR2 with Ramucirumab strongly 22. Earl HM, Hiller L, Dunn JA, Blenkinsop C, Grybowicz L, Vallier AL, Gounaris I, impacts effector/ activated regulatory T cells and CD8(+) T cells in the Abraham JE, Hughes-Davies L, McAdam K, et al. Disease-free and overall survival tumor microenvironment. J Immunother Cancer. 2018;6(1):106. at 3.5 years for neoadjuvant bevacizumab added to docetaxel followed by 41. Yang J, Yan J, Liu B. Targeting VEGF/VEGFR to modulate antitumor fluorouracil, epirubicin and cyclophosphamide, for women with HER2 negative immunity. Front Immunol. 2018;9:978. early breast cancer: ARTemis Trial. Ann Oncol. 2017;28(8):1817–24. 42. Terme M, Pernot S, Marcheteau E, Sandoval F, Benhamouda N, Colussi O, 23. Bell R, Brown J, Parmar M, Toi M, Suter T, Steger GG, Pivot X, Mackey J, Dubreuil O, Carpentier AF, Tartour E, Taieb J. VEGFA-VEGFR pathway Jackisch C, Dent R, et al. Final efficacy and updated safety results of the blockade inhibits tumor-induced regulatory T-cell proliferation in colorectal randomized phase III BEATRICE trial evaluating adjuvant bevacizumab- cancer. Cancer Res. 2013;73(2):539–49. containing therapy in triple-negative early breast cancer. Ann Oncol. 2017; 43. Quintela-Fandino M, Morales S, Cortes-Salgado A, Manso L, Apala JV, Munoz 28(4):754–60. M, Gasol Cudos A, Salla Fortuny J, Gion M, Lopez-Alonso A, et al. 24. Vrdoljak E, Marschner N, Zielinski C, Gligorov J, Cortes J, Puglisi F, Aapro M, Randomized phase 0/I trial of the mitochondrial inhibitor ME-344 or Fallowfield L, Fontana A, Inbar M, et al. Final results of the TANIA placebo added to bevacizumab in early HER2-negative breast cancer. Clin randomised phase III trial of bevacizumab after progression on first-line Cancer Res. 2019;26:35–45. bevacizumab therapy for HER2-negative locally recurrent/metastatic breast 44. Quintela-Fandino M, Lluch A, Manso L, Calvo I, Cortes J, Garcia-Saenz JA, Gil- cancer. Ann Oncol. 2016;27(11):2046–52. Gil M, Martinez-Janez N, Gonzalez-Martin A, Adrover E, et al. 18F- 25. Rothschilds A, Tzeng A, Mehta NK, Moynihan KD, Irvine DJ, Wittrup KD. fluoromisonidazole PET and activity of neoadjuvant nintedanib in early Order of administration of combination cytokine therapies can decouple HER2-negative breast cancer: a window-of-opportunity randomized trial. toxicity from efficacy in syngeneic mouse tumor models. Oncoimmunology. Clin Cancer Res. 2017;23(6):1432–41. 2019;8(5):e1558678. 45. Phillips BE, Garciafigueroa Y, Engman C, Trucco M, Giannoukakis N. Tolerogenic dendritic cells and T-regulatory cells at the clinical trials 26. Rothschilds AM, Wittrup KD. What, why, where, and when: bringing timing crossroad for the treatment of autoimmune disease; emphasis on type 1 to immuno-oncology. Trends Immunol. 2019;40(1):12–21. diabetes therapy. Front Immunol. 2019;10:148. 27. Maecker HT, McCoy JP, Nussenblatt R. Standardizing immunophenotyping 46. Bertucci F, Ng CKY, Patsouris A, Droin N, Piscuoglio S, Carbuccia N, Soria JC, for the human immunology project. Nat Rev Immunol. 2012;12(3):191–200. Dien AT, Adnani Y, Kamal M, et al. Genomic characterization of metastatic 28. Godec J, Tan Y, Liberzon A, Tamayo P, Bhattacharya S, Butte AJ, Mesirov JP, breast cancers. Nature. 2019;569(7757):560–4. Haining WN. Compendium of immune signatures identifies conserved and 47. McGranahan N, Rosenthal R, Hiley CT, Rowan AJ, Watkins TBK, Wilson GA, species-specific biology in response to inflammation. Immunity. 2016;44(1): Birkbak NJ, Veeriah S, Van Loo P, Herrero J, et al. Allele-specific HLA loss and 194–206. immune escape in lung cancer evolution. Cell. 2017;171(6):1259–71 e1211. 29. Jain RK. Normalization of tumor vasculature: an emerging concept in 48. Billan S, Kaidar-Person O, Gil Z. Treatment after progression in the era of antiangiogenic therapy. Science. 2005;307(5706):58–62. immunotherapy. Lancet Oncol. 2020;21(10):e463–76. 30. Huang Y, Goel S, Duda DG, Fukumura D, Jain RK. Vascular normalization as 49. Motzer RJ, Penkov K, Haanen J, Rini B, Albiges L, Campbell MT, Venugopal B, an emerging strategy to enhance cancer immunotherapy. Cancer Res. 2013; Kollmannsberger C, Negrier S, Uemura M, et al. Avelumab plus axitinib 73(10):2943–8. versus sunitinib for advanced renal-cell carcinoma. N Engl J Med. 2019; 31. Schmid P, Adams S, Rugo HS, Schneeweiss A, Barrios CH, Iwata H, Dieras V, 380(12):1103–15. Hegg R, Im SA, Shaw Wright G, et al. Atezolizumab and nab-paclitaxel in 50. Socinski MA, Jotte RM, Cappuzzo F, Orlandi F, Stroyakovskiy D, Nogami N, advanced triple-negative breast cancer. N Engl J Med. 2018;379(22):2108–21. Rodriguez-Abreu D, Moro-Sibilot D, Thomas CA, Barlesi F, et al. 32. Schmid P, Cortes J, Pusztai L, McArthur H, Kummel S, Bergh J, Denkert C, Atezolizumab for first-line treatment of metastatic nonsquamous NSCLC. N Park YH, Hui R, Harbeck N, et al. Pembrolizumab for early triple-negative Engl J Med. 2018;378(24):2288–301. breast cancer. N Engl J Med. 2020;382(9):810–21. 33. Florence Dalenc IG, Thomas Filleron, Amelie Lusque, Thomas Bachelot, Monica Arnedos, Mario Campone, Marie-Paule Sablin, Herve Bonnefoi, Marta Publisher’s Note Jimenez, Jacquet Alexandra, Fabrice Andre: Durvalumab compared to Springer Nature remains neutral with regard to jurisdictional claims in maintenance chemotherapy in patients with metastatic breast cancer: published maps and institutional affiliations.