A Feasibility Study of Cyclophosphamide, Trastuzumab, and an Allogeneic GM-CSF–Secreting Breast Tumor Vaccine for HER2⁺ Metastatic Breast Cancer

The availability of multiple drugs specific for HER2 (trastuzumab, lapatinib, pertuzumab, and ado-trastuzumab emtansine) has revolutionized the treatment of HER2⁺ metastatic breast cancer (1, 2). However, not all patients benefit and survival gains could be further enhanced. Innovative therapies that complement and synergize with these drugs are urgently needed to improve disease outcomes.

Immune-based therapies, including vaccines and immune checkpoint blockade, have several unique properties (3). First, immune-based therapies recruit the host tumor-specific immune response rather than targeting the tumor directly. Second, the host immune system can recognize an unlimited number of overexpressed and/or mutated targets differentially expressed by tumor cells relative to normal tissue. Third, immune-based therapies confer a durable treatment response due to immunologic memory. Although the evaluation of immune checkpoint blockade has been limited in patients with breast cancer, multiple vaccines have been tested. Despite a good safety profile, immune responses have been modest and inconsistently associated with clinical benefit. Most of these vaccination regimens were limited by some combination of immune tolerance, established disease burdens, or the use of suboptimal therapeutic targets and/or vaccination regimens.

Cytokine-modified tumor cell vaccines that secrete granulocyte-macrophage colony-stimulating factor (GM-CSF) can induce robust T cell–dependent immunity that clears established tumors in preclinical models (4). Multiple clinical studies have demonstrated the single-agent safety and bioactivity of these agents in patients with cancer, but clinical benefit remains unproven (5, 6). This lack of success is probably because vaccination alone is not potent enough to induce immune responses sufficiently robust to overcome immune tolerance and suppression and lyse bulky disease. More recently, clinical trials combining GM-CSF–secreting vaccines with the immune checkpoint agent ipilimumab demonstrated enhanced T-cell activation associated with survival benefit in prostate and pancreas cancer (7, 8). These trials provide clinical proof of principle for combination immunotherapies.

Standard cancer drugs may also augment immune-based therapies, either at approved doses or at an immune-modulating dose and schedule (9). We showed that low-dose cyclophosphamide enhances a HER2⁺ GM-CSF–secreting vaccine in both immune tolerant neu transgenic mice (10), and in patients with metastatic breast cancer (11). Low-dose cyclophosphamide relieves the suppressive influence of CD4⁺CD25⁺ regulatory T cells (Treg) in neu mice, allowing the recruitment of potent, tumor-specific CD8⁺ T cells (12, 13). The monoclonal antibody (mAb) trastuzumab also has immunomodulatory activity (14). We showed that HER2-specific mAbs can augment HER2-specific CD8⁺ T-cell responses and tumor-free survival in vaccinated, tumor-bearing neu mice (15); furthermore, the trastuzumab-like mAb 7.16.4 enhances immune priming, augmenting primary and memory CD8⁺ T-cell responses in vaccinated neu mice (16). In patients with breast cancer, trastuzumab-based chemotherapy induces T-cell responses within both the peripheral blood and tumor microenvironment (17, 18).

We, therefore, evaluated the immunologic and antitumor activity of a GM-CSF–secreting, HER2⁺ whole-cell breast cancer vaccine given in sequence with immune-modulating doses of cyclophosphamide and weekly HER2-specific mAb in tolerant neu mice, and in patients with metastatic breast cancer. The clinical study was designed to assess the safety, clinical benefit, and immunologic activity of this combination immunotherapy regimen in patients with measurable or evaluable HER2⁺ metastatic breast cancer.

Neu-N (neu) mice were derived from the colony of Guy and colleagues (19), bred to homozygosity as verified by Southern blot analysis (20), and maintained at Johns Hopkins University. Experiments were performed with 8- to 10-week-old mice using AAALAC-compliant protocols approved by the Animal Care and Use Committee of the Johns Hopkins University School of Medicine.

The GM-CSF–secreting vaccine cell lines GM (mock) and neuGM (neu-specific), the NT2.5 neu-expressing tumor cell line, and the NT2.5B7.1 antigen-presenting cell line were derived and grown as previously described (4, 10). GM-CSF levels were verified by ELISA, and neu and B7.1 levels were verified by flow cytometry biannually. All cell lines used were regularly tested and validated to be Mycoplasma free; no other authentication assay was performed.

Hybridoma cells secreting the anti-rat neu mAb 7.16.4 (21) were grown in athymic nude mice, and ascites were collected (Harlan Bioproducts for Science). 7.16.4 was purified over a T-Gel column (Pierce Biotechnology) using the Biologic LP purification system (Bio-Rad), dialyzed into PBS, and >90% purity determined by SDS-PAGE analysis using Criterion 4%–15% Tris–HCl Pre-cast Gels and the Criterion Cell (Bio-Rad). Nonspecific isotype-matched antibody served as the control. Antibodies were confirmed endotoxin free by the Limulus amebocyte lysate test (BioWhittaker).

Neu mice were challenged with 5 × 10⁴ NT2.5 tumor cells, and vaccinated 3 days later. Vaccine cells were irradiated before s.c. injection in both hind limbs and the left forelimb. Cyclophosphamide (Bristol-Myers Squibb) at 100 mg/kg and 7.16.4 at 100 μg were injected i.p. 1 day before vaccination; 7.16.4 was then given i.p. weekly. The tumor and vaccine cell doses, and the cyclophosphamide and mAb dose and schedule, have been optimized previously (10, 20, 22). Mice were monitored for tumor onset and growth twice weekly, and tumors were measured in two perpendicular dimensions with calipers.

CD8⁺ splenic lymphocytes were purified by negative selection (Dynal Biotech, Invitrogen). CD8⁺ T cells (10⁵) were incubated in duplicate with 10⁴ target cells (NT2.5B7.1 cells stimulated with IFNγ for 2 days) at 37°C overnight (22) on precoated IFNγ-specific ELISPOT plates and developed according to the manufacturer's protocols (R&D Systems). IFNγ-secreting CD8⁺ T cells were enumerated using the Immunospot counter (Cellular Technology, LTD). The number of spots in control wells was averaged and subtracted from the number of spots in each well containing both CD8⁺ T cells and targets.

A Student t test was applied to determine the statistical significance of differences between treatment groups, with P < 0.05 being significant. Analyses were performed using GraphPad Prism, version 3.0a for Macintosh (GraphPad Software). All experiments were repeated at least twice, with 5 to 10 mice per group.

The clinical study was an open-label, single-arm feasibility study to evaluate the safety and clinical benefit associated with the administration of a fixed dose of allogeneic GM-CSF–secreting breast tumor vaccine cells given in sequence with low-dose cyclophosphamide and weekly trastuzumab. A Simon two-stage design (23) was used to evaluate 20 vaccinated patients.

The clinical study was conducted in accordance with the principles of Good Clinical Practice and the ethical principles stated in the Declaration of Helsinki. It was approved by the Johns Hopkins University School of Medicine Institutional Review Board, the NIH Recombinant DNA Advisory Committee, and the FDA Center for Biologics Evaluation and Research.

Twenty-two patients with HER2⁺ metastatic breast cancer were enrolled at the Johns Hopkins Sidney Kimmel Comprehensive Cancer Center (Baltimore, MD) between November 14, 2006, and July 13, 2009; 20 patients were treated on study. Eligible patients had an Eastern Cooperative Oncology Group (ECOG) performance status of 0 to 1, and a histologic diagnosis of HER2⁺ breast cancer by IHC 3+ staining or gene amplification ≥2.0 by FISH. Prior chemotherapy must have been completed ≥28 days before vaccination; concurrent endocrine and/or bisphosphonate therapy was allowed. Other requirements included a cardiac ejection fraction ≥45%, adequate end-organ function, and negative HIV and pregnancy tests. Stable treated central nervous system disease was allowed. Key exclusion criteria included past/current autoimmune disease, non–protocol-specific treatment or parenteral steroids within 28 days of vaccination, and past/current second malignancy (except superficial nonmelanoma skin cancer, bladder cancer, tamoxifen-related endometrial cancer cured by hysterectomy, and cervical carcinoma in situ).

Written informed consent was obtained from each research participant. Baseline studies included computed tomography of the chest, abdomen, and pelvis; bone scans; complete blood count with differential; chemistry profile; absolute eosinophil count; and echocardiogram or multiple gated acquisition scan.

Patients received three monthly vaccinations, with a boost 6 to 8 months from trial entry. During active vaccination cycles, patients received cyclophosphamide, 300 mg/m² on day −1 and 5 × 10⁸ vaccine cells on day 0, on a backbone of weekly trastuzumab given at the standard dose (2 mg/kg, with a 4 mg/kg loading dose as necessary); trastuzumab was given on day −1 the week of vaccination. During the interim period between cycles 3 and 4, trastuzumab could be given weekly (2 mg/kg) or every 3 weeks (6 mg/kg). Cyclophosphamide-modulated vaccination was given every 4 to 6 weeks for three cycles, with a fourth cycle 6 to 8 months after the first cycle. Patients with disease progression were taken off study.

Vaccine development and manufacturing have been described in another publication (24). Briefly, the parent cell lines T47D (HER2^low) and SKBR3 (HER2^high) were genetically modified by plasmid DNA transfection to secrete GM-CSF. Clinical lots were prepared from two subcloned cell lines secreting bioactive levels of GM-CSF, 2T47D-V, and 3SKBR3-7. On day 0, serum-free, cryopreserved, irradiated vaccine cells were thawed and mixed to create a HER2⁺ vaccine that secreted GM-CSF levels of about 300 ng/10⁶ cells per 24 hours. Vaccine cells were injected intradermally, evenly distributed over three lymph node areas. Anesthetic lidocaine cream was applied to the injection sites before vaccination.

Toxicities were graded using the NCI's Clinical Trials Common Terminology Criteria for Adverse Events Version 3.0 (CTCAE v.3.0). Toxicity monitoring included clinical assessment and complete blood counts on days 3 and 7; chemistry profiles were measured before and after each cycle and on day 7.

The primary endpoint of clinical benefit was defined as complete response + partial response + stable disease (CR + PR + SD) at 6 months; clinical benefit at 1 year was also determined. Exploratory endpoints of progression-free survival and overall survival were defined as time to first disease progression or death from the eligibility date.

Serum was collected at baseline and after each vaccination, and on days 0, 1, 2, 3, 4, 7, and 14 of each cycle. Serum was separated from whole blood by centrifugation, and frozen in 1-mL aliquots at −80°C. Serum GM-CSF levels were measured as described previously (11). Serum TGFβ was measured by Luminex pre- and postvaccination according to the manufacturer's protocols.

One hundred micrograms each of two MHC class II HER2 epitopes (p369 and p776; ref. 25), with mutated k-ras as a negative control, was injected intradermally on the back as previously described (11). Erythema and induration were assessed 2 to 3 days after injection.

Delayed-type hypersensitivity (DTH) was considered positive if induration increased by ≥5 mm from baseline.

Whole blood (200 mL) was collected in heparin tubes pre- and postvaccination; 20 mL whole blood was collected on days 0, 1, 2, 3, 4, 7, and 14. Peripheral blood mononuclear cells (PBMC) were isolated by Ficoll-Hypaque gradient centrifugation; interface cells were harvested and washed twice. PBMCs were resuspended in AIM-V medium (Invitrogen) with 10% dimethylsulfoxide, frozen at −80°C for 1 to 2 days, and stored in liquid N₂ until analysis (8).

Treg and myeloid-derived suppressor cell (MDSC) analyses were performed with modifications to published methods (26–28). Briefly, 10⁶ PBMCs were stained for 30 minutes at 4°C with: Pacific blue-CD4 (BD Pharmingen), FITC-CD25 (BD Pharmingen), PE-Cy7-CD45RA (eBioscience), PerCP-Cy5.5-CCR7 (BD Pharmingen), APC-CTLA-4 (BD Pharmingen), FITC-Lineage cocktail (BioLegend), APC-HLA-DR (BD Pharmingen), PE-Cy7-CD33 (eBioscience), and/or PE-CD11b (eBioscience). For Treg analysis, cells were fixed for 40 minutes at 4°C and permeabilized with Transcription Factor Buffer Set (BD Pharmingen), followed by staining with PE-FoxP3 (BD Pharmingen) for 30 minutes at 4°C. Data were acquired with a Gallios cytometer (Beckman Coulter) and analyzed using the FlowJo V7.6.5 software (TreeStar, Inc.).

T-cell analyses were conducted with modifications to published methods (29, 30). Monocytes were isolated using the MACS CD14 isolation kit (Miltenyi Biotec), and cultured in 6-well plates (2 × 10⁶ cell/mL) in RPMI-1640 medium containing 5% pooled human serum (PHS), 50 ng/mL GM-CSF (Miltenyi Biotec), and 50 ng/mL IL4 (Miltenyi Biotec). Cells were cultured for 48 hours and subsequently incubated for 18 hours with 10 ng/mL IL1β (Miltenyi Biotec), 10 ng/mL IL6 (Miltenyi Biotec), 10 ng/mL TNFα (Miltenyi Biotec), and 0.5 to 1 μg/mL PGE₂ (Sigma-Aldrich). The quality of monocytic dendritic cells (MoDC) was assessed by staining with PE-Cy7-CD80 (BioLegend), FITC-CD83 (BioLegend), Pacific blue-CD86 (BioLegend), APC-HLA-DR (BD Pharmingen), and/or APC-IL-12 (BD Pharmingen). Overlapping peptide mixtures were used to stimulate CD8⁺ T cells (JPT Peptide Technologies). PepMix HER2 (extracellular and intracellular domains) is a pool of 15-mer peptides that span the entire HER2 protein and overlap by 11 amino acids; PepMix CEF Pool and PepMix HIV served as positive and negative controls, respectively. Peptides were resuspended in DMSO, stored at −80°C, thawed the day of assay, and diluted to the required concentration.

Isolated, activated MoDCs were mixed with peptides at 1 μg/mL in RPMI-1640 medium containing 5% PHS and incubated at 37°C for 2 hours. CD8⁺ T cells were purified from PBMCs using the Dynabeads Untouched CD8 T-cell kit (Invitrogen). HER2-specific CD8⁺ T cells were indistinguishable from the background by flow cytometry directly ex vivo, and were, therefore, stimulated in vitro by coculturing peptide-loaded MoDCs with autologous CD8⁺ T cells at a ratio of 1:3 in the presence of IL7 (R&D Systems) and IL2 (R&D Systems) at concentrations of 10 ng/mL and 10 U/mL, respectively. On days 8 and 15, T cells were restimulated with MoDCs. The cells were refed with IL7 and IL2 every 3 days. The cells were harvested on day 15 after 4 hours of restimulation and assayed by intracellular cytokine staining (ICS) using FITC-CD8 (eBioscience), PE-IFNγ (BioLegend), APC-TNFα (BioLegend), and PE-Cy7-IL2 (BD Pharmingen). Sample data were acquired by a Gallios cytometer (Beckman Coulter) and analyzed using the FlowJo Version 7.6.5 software. The proportions of polyfunctional CD8⁺ effector T-cell subsets were classified by cytokine secretion.

The trial database was closed on October 16, 2013. A Simon two-stage design was used (23). Nine patients were to be enrolled and treated in the first stage. At the end of the first stage, if 2 or more patients demonstrated clinical benefit, 11 additional patients were to be enrolled and treated. This design had an α level of 0.08 with a power of 0.86 to detect a clinical benefit (CR + PR + SD) of 0.45 from a null hypothesis rate of 0.20. At the end of the second stage, if 7 or more patients demonstrated clinical benefit, the null hypothesis was rejected. On the basis of this study design, there was a 44% chance of early stopping if the true response rate was 0.20, and a 4% chance of early stopping if the true response rate was 0.45.

The study met criteria for full accrual, and data from 20 patients were used in the analysis. Tumor responses (CR + PR + SD) were determined by RECIST 1.0. Clinical benefit was estimated using the binomial distribution along with the 95% confidence interval (CI). Continuous data were summarized using mean and SD. Categorical data were represented using percentage or proportion. Survival probability was estimated using the Kaplan–Meier method (31). The CI of median survival time was constructed by the method of Brookmeyer and Crowley (32). Paired and unpaired Student t tests were used for pre- and postvaccine comparisons and between-group comparisons. All P values are reported as two-sided, and all analyses were conducted using SAS software (version 9.2 SAS Institute).

Neu mice exhibit preexisting immune tolerance specific for rat HER2 (neu), and are a stringent model for developing clinically relevant cancer immunotherapies (10, 20, 22). We first evaluated whether the trastuzumab-like mAb 7.16.4 could augment the activity of neu-specific vaccination alone or sequenced with immune-modulating doses of cyclophosphamide in these mice. Treating tumor-bearing neu mice with 7.16.4 and mock vaccination delayed tumor growth relative to neu-targeted vaccination alone, which is ineffective in this model (Fig. 1). Adding cyclophosphamide or 7.16.4 to neu-specific vaccination also delayed tumor outgrowth, increasing tumor-free survival from 0% (vaccine or 7.16.4 alone) to 10% (vaccine + cyclophosphamide) and 30% (vaccine + 7.16.4; Fig. 1A and B). Adding both cyclophosphamide and 7.16.4 to neu-targeted vaccination increased tumor-free survival to about 60% (Fig. 1C). This relative antitumor efficacy was reflected in the numbers of tumor-specific IFNγ-secreting CD8⁺ T cells by ELISPOT, where the highest levels were detected in the setting of cyclophosphamide + 7.16.4 + vaccine (Fig. 1D).

On the basis of this preclinical model, we conducted a clinical trial testing cyclophosphamide-modulated vaccination with trastuzumab in patients with HER2⁺ metastatic breast cancer. The intervention and data collection schedules are shown in Fig. 2. Twenty eligible patients were vaccinated, with an age range of 34 to 69 years (Table 1). All had HER2⁺ metastatic disease, 13 had estrogen receptor–positive and/or progesterone receptor–positive disease. The mean disease-free interval to relapse from first diagnosis was 32 months (range, 0–130 months); 7 (35%) patients presented with metastatic breast cancer at diagnosis. Eighteen patients (90%) received prior trastuzumab for metastatic breast cancer, and 3 patients received prior trastuzumab as adjuvant therapy. Eleven (55%) patients were on concurrent endocrine therapy, and the majority (70%) received concurrent bisphosphonate therapy for skeletal metastasis. The mean and median sums of the longest tumor diameter at study entry were 40.7 mm and 24.5 mm, respectively.

All patients (100%) received at least one vaccination, 15 (75%) received at least three vaccinations, and 8 (40%) received all four vaccinations. Two patients came off-study for reasons unrelated to disease progression, one after three vaccine cycles due to relocation, and another after two vaccine cycles due to study demands. A third off-study event was related to both a vaccine-related serious adverse event and simultaneous disease progression. All other off-study events before cycle 4 were due to breast cancer progression.

The most common adverse events were local vaccine site reactions, including erythema, induration, pruritus, and/or discomfort (Table 2). These self-limited local reactions occurred in all individuals, and increased in intensity but not duration with subsequent vaccination cycles. The most common systemic adverse events were fatigue, urticaria, pruritus, and fever. Seven patients (35%) developed urticaria distant from the vaccine site; of these, 3 (15%) developed grade 2–3 urticaria that required parenteral intervention with diphenhydramine and/or steroids. Of these, one was a vaccine-related serious adverse event requiring overnight hospitalization and steroid treatment. No evidence of clinically significant autoimmunity was detected. Despite the use of trastuzumab, no decrement in cardiac function was detected (data not shown).

Serum GM-CSF levels were measured as an indicator of the vaccine's life span following injection. Serum GM-CSF levels peaked at 48 hours, with peak levels persisting across all four cycles as previously observed (11). Trastuzumab did not alter the timing of the peak or the kinetics of GM-CSF decay (data not shown).

The clinical benefit rates of weekly trastuzumab with cyclophosphamide-modulated vaccination at 6 months and 1 year were 11 of 20 (55%; 95% CI, 32%–77%; P = 0.013) and 8 of 20 (40%; 95% CI, 19%–64%), respectively. One partial response and no complete responses were observed. Overall median progression-free survival and overall survival were 7 months (95% CI, 4–16) and 42 months (95% CI, 22–70), respectively (Fig. 3A). The 5-year survival rate was 6 of 20 (30%, 95% CI, 12–54).

We previously demonstrated the feasibility of using DTH responses specific for HER2-derived peptides and HER2-specific antibody responses by ELISA as immune response biomarkers for this cell-based vaccine (11). Both because our preclinical data demonstrate a correlation of HER2-specific CD8⁺ T cells with tumor-free survival in vaccinated neu mice, and because trastuzumab confounds the assessment of HER2-specific antibody responses, we assessed HER2-specific DTH and measured HER2-specific CD8⁺ T cells by ICS as biomarkers of vaccine-induced immunity in this study. HER2-specific DTH developed in 7 of 20 patients (35%; 95% CI, 15%–59%); 4 of these (57%; 95% CI, 18%–90%) showed a clinical benefit. There was a trend toward longer progression-free survival and overall survival for patients who developed HER2-specific DTH following vaccination (Fig. 3B and C), although this was not statistically significant due to the small sample size of this study. Median progression-free survival for DTH-positive versus DTH-negative patients was 9 months (95% CI, 4–46) versus 4 months (95% CI, 3–14; P = 0.58). Median overall survival durations for DTH-positive versus DTH-negative patients were 70 months (95% CI, 40–70) versus 29 months, respectively (95% CI, 13—not reached; P = 0.11). Larger randomized studies are required to rigorously evaluate the possible relationship between DTH and survival.

HER2-specific CD8⁺ T-cell function before and after vaccination was assessed in vitro by ICS using commercially available peptide mixtures containing overlapping epitopes spanning the entire HER2 protein, a representative analysis is shown in Fig. 4A. Both the average level of CD8⁺ T cells secreting IFNγ (0.501% versus 0.785%, P < 0.0001), IL2 (0.567% versus 0.683%, P = 0.0006) and TNFα (0.751% versus 1.525%, P < 0.0001), and the proportion of polyfunctional CD8⁺ T cells secreting more than one cytokine (0.023% versus 0.735%, P < 0.0001) increased across the vaccination cycles (Fig. 4B and C). In exploratory analyses, increased polyfunctional CD8⁺ T cells were possibly associated with longer progression-free survival (HR, 0.44; 95% CI, 0.2–0.98; P = 0.04) and OS (HR, 0.43; 95% CI, 0.13–1.36; P = 0.15).

We also assessed peripheral markers of immune suppression by FACS (Treg and MDSC) and Luminex (TGFβ). Although an association between decreased TGFβ levels and immunity in patients treated with a HER2 peptide vaccine and concurrent trastuzumab was reported previously (33), we found no difference in serum TGFβ levels before and after vaccination (43,125 versus 46,936 pg/mL, P = 0.21). Tregs decreased across all vaccine cycles, and both overall and CTLA-4⁺ Tregs decreased from baseline to postvaccine cycle 1; MDSCs also decreased from baseline to postvaccine cycle 1 (Fig. 5). Exploratory analyses suggested that neither baseline nor decreased levels of peripheral Tregs were associated with survival benefit. However, decreased numbers of peripheral MDSCs were possibly associated with both longer progression-free survival (HR, 0.5; 95% CI, 0.29–0.85; P = 0.010) and overall survival (HR, 0.52; 95% CI, 0.31–0.89; P = 0.016) baseline levels of MDSCs were not. To further explore possible associations between immunologic biomarkers and survival, a subgroup analysis of 12 patients was performed comparing 6 individuals with OS ≤ 24 months with 6 individuals with OS ≥ 60 months. Both decreased MDSCs and increased polyfunctional T cells following vaccination were possibly associated with OS, with P values of 0.012 and 0.028, respectively.

The results from this study of a HER2⁺ GM-CSF–secreting breast tumor vaccine with low-dose cyclophosphamide and weekly trastuzumab supports the following seven conclusions. First, in the clinically relevant neu transgenic mouse, adding both cyclophosphamide and 7.16.4 (a trastuzumab-like mAb) to vaccination increases tumor-free survival and IFNγ-secreting tumor-specific CD8⁺ T cells relative to adding either cyclophosphamide or 7.16.4 alone to vaccination. Second, up to four sequential cyclophosphamide-modulated vaccinations are safe and well tolerated in patients with HER2⁺ metastatic breast cancer receiving weekly trastuzumab. Third, this immunotherapy was associated with a 6-month clinical benefit rate of 55%, and median progression-free survival and overall survival of 7 months and 42 months, respectively. This study met its primary endpoint, as the target therapeutic outcome was a 6-month clinical benefit rate of 45%. Fourth, this combination immunotherapy induced new or increased HER2-specific T-cell immunity in patients with HER2⁺ metastatic breast cancer as measured by HER2-specific DTH and CD8⁺ T cells by ICS. There was a trend toward longer progression-free survival and overall survival in patients who developed HER2-specific DTH relative to those who did not. Fifth, polyfunctional HER2-specific CD8⁺ T cells progressively expanded across the vaccination cycles. Sixth, both Tregs and MDSCs decreased across the vaccination cycles. Finally, hypothesis-generating analyses suggest that both decreased MDSCs and a greater proportion of polyfunctional HER2-specific CD8⁺ T cells may be associated with longer progression-free survival and overall survival. A comparative trial powered for survival will be required to rigorously evaluate these associations.

Trastuzumab has revolutionized the management of HER2⁺ breast cancer, and remains the major backbone of therapy for early- and late-stage disease (1, 2, 14). However, intrinsic and acquired trastuzumab resistance remains a significant limitation to its clinical utility. In patients, trastuzumab-based chemotherapy is associated with increased levels of HER2-specific antibodies and T cells (17), and intratumoral lymphoid nodules develop in the setting of neoadjuvant trastuzumab-based chemotherapy (18). These observations suggest that leveraging the ability of trastuzumab to activate adaptive immunity could increase its efficacy. One clinical study combined a peptide vaccine with weekly trastuzumab, demonstrating increased levels of HER2-specific T cells, epitope spreading, and decreased serum levels of the immune-suppressive cytokine TGFβ (33). We showed that HER2-specific mAbs alone induce low levels of neu-specific T cells in neu transgenic mice. The addition of HER2-specific mAbs to neu-specific GM-CSF–secreting vaccination more effectively augments neu-specific CD8⁺ effector T-cell function and prolongs tumor-free survival in these mice (15). We further showed that the trastuzumab-like mAb 7.16.4 augments locoregional immune priming through binding the Fc receptor of dendritic cells in similarly vaccinated neu mice, augmenting CD8⁺ T-cell proliferation, cytokine secretion, and central memory development (16). Here, we build on these prior studies to show that the combination of low-dose cyclophosphamide, trastuzumab, and a GM-CSF–secreting HER2⁺ allogeneic cell-based vaccine is safe, bioactive, and associated with clinical benefit in patients with HER2⁺ metastatic breast cancer.

A significant challenge in optimizing combination tumor cell vaccine-based immunotherapy is the lack of clear biomarkers for assessing immune response and drug–drug interactions. We previously demonstrated the feasibility of using DTH responses specific for two MHC Class II HER2-derived peptides and HER2-specific antibody responses by ELISA before and after vaccination as immune response biomarkers for this cell-based vaccine (11). Because the use of trastuzumab here confounds the assessment of HER2-specific antibody responses, we measured HER2-specific DTH and peripheral HER2-specific CD8⁺ T cells by flow cytometry. We found that HER2-specific DTH developed in 7 of 20 patients (35%); 4 of these patients showed a clinical benefit. In addition, there was a trend toward longer progression-free survival and overall survival in patients who developed HER2-specific DTH relative to those who did not. These data are consistent with other small clinical trials that have shown an association between enhanced tumor-specific DTH and survival (34–37). Cytokine-secreting CD8⁺ T cells also increased with vaccination in patients with metastatic breast cancer. While vaccine-induced CD8⁺ T cells from neu mice could be detected by ELISPOT directly ex vivo, vaccine-induced CD8⁺ T cells from patients with metastatic breast cancer were detected by flow cytometry only after in vitro stimulation. This differential sensitivity in detecting vaccine-induced CD8⁺ T cells is likely explained by differences in the host (transgenic neu mice are genetically homogenous, whereas patients with metastatic breast cancer are genetically and biologically heterogeneous), the vaccine (the murine vaccine specifically targets HER2, whereas the human vaccine delivers multiple tumor antigens where responses to HER2 are used as a sentinel measure of vaccine activity), the T cells (assayed fresh from mice as opposed to frozen from patients with metastatic breast cancer), and the assays (ELISPOT is more sensitive than flow cytometry). We used ICS by flow cytometry to rapidly evaluate HER2-specific CD8⁺ T cells that secrete one or more cytokines in patients with metastatic breast cancer.

Polyfunctional antigen-specific T cells may be indicators of disease control and/or survival in patients with chronic infections and cancer (38–40). Polyfunctional T cells secrete multiple relevant cytokines (IFNγ, TNFα, and IL2) that reflect the quality of the immune response as a more informative correlate of sustained protective immunity than T cells that secrete only a single cytokine (IFNγ or TNFα). Durable T-cell polyfunctionality correlates with long-term disease control in HIV (41) and malaria patients (42), but has been less studied in vaccinated patients with cancer. A telomerase peptide vaccine given with temezolomide to patients with metastatic melanoma induced polyfunctional T cells (43). Here, we demonstrate the progressive expansion of polyfunctional HER2-specific T cells across vaccination cycles, where a greater proportion of polyfunctional CD8⁺ T cells may be associated with longer progression-free survival and overall survival. A comparative clinical study powered for survival is required to formally evaluate this hypothesis.

Tregs and MDSCs are significant mediators of immune tolerance and suppression in patients with cancer. We showed previously that low-dose cyclophosphamide depletes Tregs in neu mice, facilitating the vaccine-mediated recruitment and activation of latent, high-avidity tumor-specific T cells and tumor rejection (10, 12, 13). Here, we demonstrate a decline in peripheral Tregs and MDSCs in vaccinated patients with metastatic breast cancer. Several clinical trials have explored depleting Tregs in the setting of vaccination for cancer, with enhancement of antigen-specific immune responses to vaccination (11, 44–46). Notably, a randomized phase II study evaluating a multipeptide renal cell carcinoma vaccine alone or sequenced with low-dose cyclophosphamide demonstrated that cyclophosphamide-modulated vaccination reduced Tregs and induced T-cell responses specific for multiple peptide antigens, increasing the OS of vaccinated patients (46). An association between levels of circulating MDSCs and immune response or clinical outcomes has been reported in several immunotherapy trials, and these results are consistent with our findings (46–49).

This combination immunotherapy was safe. Given the known cardiac toxicity of trastuzumab (14), cardiac function was assessed every 3 months on study. No patient developed a significant drop in left ventricular ejection fraction. These data suggest that the addition of cyclophosphamide-modulated vaccination to weekly trastuzumab did not increase cardiac toxicity, although the sample size is limited. Notably, urticaria was observed in a significant number of patients in this study; 3 patients required parenteral diphenhydramine and/or steroids. It remains unclear if urticaria may be associated with higher levels of immunity or clinical benefit. We observed clinical benefit rates of 55% at 6 months and 40% at 1 year. The median progression-free survial in this study was 7 months, the median overall survival was 42 months, and the 5-year survival rate was 30%. Although the sample size is small, these clinical results compare favorably with those from other trials of HER2-directed therapy for metastatic HER2⁺ breast cancer that do not include cytotoxic chemotherapy. The seminal trials of single-agent trastuzumab for the first-line and second/third-line treatment of patients with metastatic disease showed clinical benefit rates of 48% and 56% at 6 months (50, 51); median overall survival in these seminal trials was 24 and 13 months. More recently, novel mAb-based HER2-directed therapies were approved for metastatic HER2⁺ breast cancer progressing on prior trastuzumab therapy (2). Concurrent pertuzumab and trastuzumab therapy is associated with a clinical benefit rate of 50%, and a median progression-free survival of 5.5 months (52) and ado-trastuzumab emtansine therapy is associated with a median progression-free survival and overall survival of 9.6 months and 30.9 months, respectively (53).

In conclusion, this allogeneic GM-CSF–secreting breast tumor vaccine is safe and bioactive, and confers clinical benefit when sequenced with low-dose cyclophosphamide and standard trastuzumab in patients with HER2⁺ metastatic breast cancer. It induces HER2-specific immunity as measured by DTH and polyfunctional CD8⁺ T cells, two measures of a high-quality immune response. We observed a clinical benefit rate of 55% at 6 months, with progression-free survival and overall survival of 7 months and 42 months, respectively. Since this study began, two additional mAb-based therapies with survival benefit were approved for metastatic HER2⁺ breast cancer (pertuzumab and ado-trastuzumab emtansine). The data presented here are encouraging relative to results with other HER2-directed combination therapies, including these new agents. Further investigation of GM-CSF–secreting vaccines in combination with low-dose cyclophosphamide and HER2-specific mAb therapies is warranted.

E.M. Duus is the Director of Clinical Research at Helsinn Therapeutics. V. Stearns reports receiving a commercial research grant from MedImmune. L.A. Emens receives research funding from Genentech, Inc. L.A. Emens and E.M. Jaffee receive research funding from Roche, Inc. Under a licensing agreement between Aduro, Inc. and the Johns Hopkins University, the University and E.M. Jaffee are entitled to milestone payments and royalty on sales of the GM-CSF–secreting breast cancer vaccine. The terms of these arrangements are being managed by the Johns Hopkins University in accordance with its conflict of interest policies. L.A. Emens is a member of the Cellular, Tissue and Gene Therapies Advisory Committee at the FDA. No potential conflicts of interest were disclosed by the other authors.

Conception and design: E. Garrett-Mayer, L.A. Emens

Development of methodology: G. Chen, E. Garrett-Mayer, L.A. Emens

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): G. Chen, R. Gupta, S. Petrik, J.M. Leatherman, N.E. Davidson, M. Berg, J. Fetting, E.M. Duus, S. Atay-Rosenthal, A.C. Wolff, V. Stearns, E.M. Jaffee, L.A. Emens

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): G. Chen, R. Gupta, J.M. Asquith, E.M. Duus, S. Atay-Rosenthal, X. Ye, L.A. Emens

Writing, review, and/or revision of the manuscript: R. Gupta, J.M. Leatherman, M.M. Daphtary, E. Garrett-Mayer, N.E. Davidson, J. Fetting, E.M. Duus, S. Atay-Rosenthal, X. Ye, A.C. Wolff, V. Stearns, E.M. Jaffee, L.A. Emens

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): M. Laiko, J.M. Leatherman, J.M. Asquith, M.M. Daphtary, K. Hirt, J.N. Uram, L.A. Emens

Study supervision: J.M. Leatherman, T. Dauses, L.A. Emens

Other (regulatory affairs): M.M. Daphtary

Other (review of imaging at baseline and evaluation of disease progression and response to therapy): S. Atay-Rosenthal

The authors thank Mark Greene at the University of Pennsylvania for providing the 7.16.4 hybridoma.

This work was supported by the Department of Defense Clinical Translational Research Award W81XWH-07-1-0485 (to L.A. Emens), Genentech, Incorporated (to L.A. Emens), and in part by American Cancer Society RSG CCE 112685 (to L.A. Emens), the Specialized Programs in Research Excellence (SPORE) in Breast Cancer P50CA88843 (to N.E. Davidson, E.M. Jaffee, and L.A. Emens), and the Rapid Access to Investigational Drugs (RAID) Program of the National Cancer Institute/NIH (to L.A. Emens). It was also supported by funding from the Gateway Foundation (to L.A. Emens), the Safeway Foundation (to L.A. Emens), Avon Foundation (to E.M. Jaffee and L.A. Emens), and in part by the Johns Hopkins University Institute for Clinical and Translational Research, Climb for Hope, and the Josephine A. Peiser Foundation. E.M. Jaffee is the Dana and Albert “Cubby” Professor of Oncology.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.