Rapid Generation of Sustainable HER2-specific T-cell Immunity in Patients with HER2 Breast Cancer using a Degenerate HLA Class II Epitope Vaccine

Breast cancer is a common disease with approximately 2,000,000 new cases and approximately 600,000 deaths annually, worldwide. HER2 overexpression defines a subset (15%–20%) of breast cancers (1). HER2 overexpression results in an aggressive tumor that, without treatment, is pathologic displaying rapid progression and frequent recurrence (2). The implementation of antibody-based therapeutics that target HER2, such as trastuzumab, has significantly impacted mortality in individuals with HER2⁺ breast cancer (3–5). While many mechanisms of action have been ascribed to the clinical activity of trastuzumab, recent studies have demonstrated an important role for the adaptive immune response. Foremost among immune effectors, endogenously generated HER2-specific antibodies, cytotoxic T cells, and helper CD4 T (Th) cells are important in disease protection mediated by trastuzumab-containing regimens and have been associated with greater objective tumor responses and significantly improved disease-free, progression-free, and overall survival in both humans and murine models (6–9).

Induction of an efficient coordinated adaptive immune response against infectious diseases and cancer is dependent on the generation of Th cell responses (10–13). They are critical to immune surveillance and when antigen is encountered they produce numerous cytokines that coordinate activation of various adaptive and innate immune effectors such as cytotoxic T cells, B cells, macrophages, and NK cells (10–13). On the basis of these important roles, we envision that the addition of Th cell vaccination to trastuzumab-based therapy would result in improved disease-free survival (DFS). Consistent with this rationale, we developed a vaccine that consists of four naturally presented tumor-associated epitopes derived from HER2 that bind to at least 12 of 15 common HLA-DR molecules (14). The vaccine represents a significant novel advance over prior HER2 vaccines. Notably, the vaccine was formulated on the basis of actual binding affinities of peptides to 15 different HLA-DR variants that broadly represented the North American population, including Caucasians, Asians, and African Americans (15). This is in contrast to our prior studies vaccines which were selected based solely on prediction to a single HLA-DR molecule resulting in selection of several epitopes with limited binding to most HLA-DR variants (16). Thus, the new vaccine was predicted to generate in vivo HER2-specific T-cell immunity to naturally processed and presented antigen in the vast majority of people regardless of HLA-DR background, which contrasts with other advanced HER2 peptide vaccines which are designed for use specifically in patients with select HLA backgrounds (17–19). Therefore, the primary advantage of this new vaccine is its broad applicability for individuals of all HLA-DR subtypes. In this study, this new vaccine was tested in a phase I clinical trial for safety and immunogenicity in patients with HER2⁺ breast cancer. In accordance with instructions from the FDA, the vaccine was tested following completion of trastuzumab-based regimens as a safety precaution with the ultimate goal of testing the vaccine concurrently with trastuzumab-based regimens.

This study enrolled women ≥18 years old with resected HER2⁺ stage II–III breast cancer that had completed HER2-directed systemic therapy. Eligibility criteria included an Eastern Cooperative Oncology Group performance status of 0–1, and adequate hematologic, renal, and hepatic function. Exclusion criteria included autoimmunity, systemic steroid use within 90 days of registration, baseline left ventricular ejection fraction (LVEF) < 55%, history of trastuzumab-related cardiac toxicity, pregnancy, or breast feeding. Patients received a tetanus booster prior to HER2 vaccination unless they had one within 1 year of registration. Written, informed consent was obtained from all patients. The study was conducted in accordance with recognized ethical guidelines including the U.S. Common Rule and approval by institutional review board was obtained in accordance with a filed assurance approved by the Department of Health and Human Services (Clinicaltrials.gov #: NCT01632332).

The HER2-derived peptides (Polypeptide Systems) that constituted the vaccine included NLELTYLPTNASLSF [p59, amino acid (aa) positions 59–73], HNQVRQVPLQRLRIV (p88, aa 88–102), LSVFQNLQVIRGRIL (p422, aa 422–435), and PIKWMALESILRRRF (p885, aa 885–899; ref. 14).

Patients were vaccinated monthly six times intradermally in the forearm or the outer upper leg with 500 μg/peptide and 125 μg of GM-CSF (Sargramostim, Genzyme). Patients underwent assessments at multiple time points: prior to registration, within 14 days after registration, prior to vaccination, 4 weeks after completion of vaccinations, and 3, 6, 12, 18, and 24 months after last vaccination thereafter until disease recurrence or a maximum of 30 months. Assessments included physical examination, blood work, cardiac monitoring, toxicity assessments, and research blood draws. Peripheral blood mononuclear cells (PBMC) and sera for immune monitoring were isolated as described previously (20, 21).

Vaccinations were discontinued if patients developed a grade ≥2 allergic reaction, autoimmune reaction, injection site ulceration, neurologic difficulties, or grade ≥3 adverse events. Vaccination was discontinued if there was a (i) ≥10% LVEF decrease from pretreatment to below the institutional lower limit of normal, (ii) ≥15% LVEF decrease from pretreatment, or (iii) any LVEF decrease with symptoms of congestive heart failure.

Enzyme-linked immunosorbent spot (ELISpot) was used to determine frequencies of antigen-specific T cells (21). Antigens were cyclin D1 (control) peptide (14, 22), vaccine peptides, HER2 extracellular domain protein (ECD, Sino Biological), HER2 intracellular domain protein (ICD, Cell Sciences), or tetanus toxoid (TT; List Biological Laboratories). Peptides were plated at 10 μg/mL, TT at 1 μg/mL, ECD at 25 ng/mL, and ICD at 100 ng/mL.

Antibody (IgG) responses were assessed using a typical ELISA (21). Antigens tested included 20 ng/mL of peptides, or 100 ng/mL of TT, ECD, or ICD.

The primary endpoint was the percentage of patients who developed a severe toxicity (grade 3–5 adverse events, NCI Common Terminology Criteria for Adverse Events version 4.0). DFS was defined as time from registration to disease recurrence, second primary cancer, or death without disease recurrence or second primary cancer, and estimated with the Kaplan–Meier method (23). The laboratory endpoints for the reported analyses herein were levels of antigen-specific T cells and antibodies which were compared using either the Wilcoxon matched pairs two-tailed test or the Friedman test followed by post hoc Dunn Multiple Comparison's Test (Instat, V.3. or Prism V. 8., GraphPad). Unpaired data were compared using the Mann–Whitney two-tailed test. A patient was considered to have responded if they had developed a ≥3-fold increase in HER2-specific T cells or antibodies at any point during vaccine period (21). If T-cell immunity was undetectable, a positive response was defined as ≥50 antigen-specific T cells/million PBMCs.

Twenty-two patients with HER2⁺ breast cancer enrolled in study (Table 1). Twenty of 22 patients completed all six immunizations. One patient withdrew after developing a grade 1 injection site reaction and another due to headaches. Twenty-two patients were evaluable for toxicity assessments. Overall, vaccination was well-tolerated with only grades 1 and 2 adverse reactions, mostly injection site reactions and fatigue (Table 2). One patient developed a grade 3 increased international normalized ratio (INR), deemed unlikely related to treatment. Two patients had LVEF drops of 10% or more but remained in the normal range.

The mean frequencies of T cells specific for HER2 were compared between pretreatment and posttreatment samples. PBMCs were available from 21 of 22 patients. In Fig. 1A the preimmunization and highest postimmunization T-cell frequencies are plotted as described previously (21, 24, 25). The median number of postvaccination period samples was six (range, 4–6). The preimmunization (baseline) mean T-cell frequency to p59 was 80 ± 21 (±SEM, n = 21) T cells/million PBMCs which increased to a mean of 552 ± 63 T cells/million PBMCs. T cells increased from 70 ± 20 to 443 ± 54 T cells/million PBMCs for p88, from 80 ± 22 to 380 ± 55 for p422, and from 102 ± 31 to 356 ± 50 for p885.

The mean T-cell frequency to ICD increased significantly from 72 ± 28 to 347 ± 56 T cells and increased from 98 ± 29 to 398 ± 62 T cells for the ECD, demonstrating that vaccine-elicited T cells are recognizing naturally processed antigens. Consistent with tetanus boosters given prior to HER2 vaccination, the mean TT-specific T-cell frequency increased from 695 ± 62 to 891 ± 61 TT-specific T cells/million PBMCs. The HER2 T-cell frequency, which is the paired sum of the ICD- and ECD-specific T-cell responses, is also reported for comparison with TT. As shown, the mean T-cell frequency to HER2 increased significantly from 170 ± 50 to 754 ± 120 antigen-specific T cells/million PBMCs. It is notable that this mean postvaccination HER2 T-cell frequency is approximately 85% of the magnitude of the postvaccination response to TT, which is a large 150 kDa protein with many foreign epitopes. Reactivity to control cyclin D1 peptide, in contrast, did not significantly increase during the vaccinations.

Increased mean antibody responses to the epitopes were generally small (Fig. 1B). Sera for antibody assessments were available from 21 of 22 patients. The median number of vaccination period serum samples was six (range, 5–6). The preimmunization mean antibody concentration to p59 was 1.1 ± 0.15 μg/mL (± SEM) and increased to 1.7 ± 0.21 μg/mL, postimmunization. To p88, antibodies increased from 1.2 ± 0.2 to 2.3 ± 0.6 μg/mL, to p422 from 1.0 ± 0.12 to 1.6 ± 0.22 μg/mL, and to p885 from 0.8 ± 0.13 to 1.1 ± 0.18 μg/mL. These increases led to an increase in antibodies that bound whole ICD (1.6 ± 0.33 to 2.8 ± 0.4 μg/mL) and ECD proteins (2.5 ± 0.40 to 4.2 ± 0.35 μg/mL). The mean TT antibody concentration increased from 6.6 ± 0.4 to 9.1 ± 0.8 μg/mL. Similar to the T-cell analysis, the paired sum of the ICD- and ECD-specific antibody revealed that HER2-specific antibody immunity increased from 4.1 ± 0.61 to 7.1 ± 0.62 μg/mL (∼77% of the mean high TT antibody response).

The percentage of patients that responded in the vaccine period was analyzed. Responses to all the peptides were frequent with 71%, 86%, 67%, and 76% to p59, p88, p422, and p885, respectively, in 19 of 21 evaluable patients (Fig. 2A). Although mean levels of antibodies increased significantly, antibody response rates (n = 20 evaluable) were minimal at 10%, 5%, 5%, and 0% for p59, p88, p422, and p885, respectively, (Fig. 2B), using the threefold criteria. Nineteen of 21 evaluable (90%) patients responded to vaccine with a T-cell response, all responding to multiple epitopes (Fig. 2C). The median and mean numbers of epitopes that patients responded to with T cells were 3 and 3, respectively. Four of 20 evaluable patients responded with antibody responses (Fig. 2D). Those patients who did not respond to vaccine peptides had preexistent T-cell immunity (Fig. 2E). Responders (n = 15–18; depending on the peptide) had low preexisting mean immunity ranging from 31 T cells/million PBMCs for p422 to 42 T cells/million PBMCs for p88. In contrast, nonresponders (n = 3–7) had higher preexisting immunity ranging from 180 million T cells/million PBMCs for p422 to 316 T cells/million PBMCs for p885 (Fig. 2E). Thus, at the end of vaccination, 100% of patients demonstrated immunity, the majority through vaccine augmentation. High preexisting immunity in a fraction of patients is consistent with prior findings (14). Significant correlations between peptide-specific responses and protein-specific responses demonstrate that the epitopes are naturally processed antigen (Fig. 2F–I).

The T-cell data were also examined at each individual timepoint over the course of vaccination, enabling an analysis of the time to maximal immunity (Fig. 3A–H). For this analysis, there were samples from 21, 20, 21, 20, 21, 21, and 20 patients for time points 0, 1, 2, 3, 4, 5, and 6, respectively. Time to maximal immunity was estimated to be 2, 2, 2, and 3 months for p59, p88, p422, and p885, respectively, with a median time of 2 months. This coincided with maximal immunity at 2 months to the ICD and ECD (Fig 3E–F). T-cell immunity to the cyclin D1 peptide remained low (Fig. 3G), whereas T-cell immunity to TT was high at baseline and maintained over the course of vaccinations (Fig. 3H). As a comparison, the mean levels of total HER2-specific T cells (red line, Fig 3H) were calculated by paired summing of the ICD- and ECD-specific T-cell frequencies. Levels reached maximal levels at 2 months and remained on average at 57% ± 2% of the level of T-cell immunity to TT.

In contrast, levels of HER2 peptide–specific antibodies tended to increase only slightly or remain stable over the vaccine period (Fig. 3I–P). There were samples from 20, 21, 21, 20, 21, 20, and 18 patients for time points 0, 1, 2, 3, 4, 5, and 6, respectively, for longitudinal assessment of antibody levels. Time to maximal immunity was 5, 4, 4, and 6 months for p59, p88, p422, and p885 with a median of 4.5 months. Paired analysis showed significant increases in the means for p59, p88, and p422 but not for p885 or the cyclin D1 peptide. Despite lack of antibody immunity to p885, there were significant increases in the levels of antibodies to the ICD of HER2 which increased from 1.5 ± 0.3 μg/mL (±SEM) to a maximum of 2.6 ± 0.4 μg/mL (Fig. 3M). Mean levels of antibodies to ECD protein also rose in parallel, increasing from 2.5 ± 0.4 μg/mL (±SEM) to a maximum of 3.4 ± 0.3 μg/mL (Fig. 3N). Mean levels of TT-specific antibodies increased from 6.6 ± 0.4 to 8.3 ± 0.9 μg/mL (Fig. 3P). For comparison, Fig. 3P shows the combined ICD and ECD mean antibody levels which increased throughout the course of immunization and ranged from 59% to 72% of the TT-specific antibody levels.

A critical outcome of an efficacious vaccine is persistent immunity (26). To determine whether vaccination resulted in the generation of immune memory, blood specimens were collected over 24 months after the last vaccination. Similar to Fig. 3, we determined the frequency of IFNγ-producing T cells specific for the HER2 peptides, HER2 protein, and control peptide which were compared between pretreatment samples and multiple posttreatment observation period samples (Fig. 4A). PBMCs were available from 20 of 21 patients. Specifically, there were samples from 20, 20, 19, 17, 15, and 10 patients for time points 0, 9, 12, 18, 24, and 30 months (time following start of vaccinations), respectively.

T-cell immunity was retained to all of the vaccine epitopes and HER2 protein at a consistently elevated level over a period of 2 years following the final vaccination. In contrast to the vaccination period, antibodies to the peptides appeared to robustly increase after the vaccinations were stopped, measurable at month 9 following start of vaccination (i.e., 3 months after last vaccine; Fig. 4G–L). Elevated antibody responses to both p59 and p88 did not increase after month 9 but appeared to persist over the observation period. In contrast, antibody levels to p422 and p885, despite the initial increase, returned to baseline levels at 2 years. Elevated mean levels of antibodies to HER2 protein, which were increased during the vaccine period, were also stably elevated out to 30 months without signs of diminishing.

The highest observation period T-cell frequencies and antibody concentrations are depicted in Fig 4M and N, respectively. The mean of the highest observation period T-cell frequencies for p59 was 389 ± 44 (SEM) T cells/million PBMCs, 322 ± 47 T cells/million PBMCs for p88, 197 ± 24 T cells/million PBMCs for p422, and 195 ± 49 T cells/million PBMCs for p885 (Fig. 4M). The mean of the highest observation period T-cell frequencies for HER2 protein was 379 ± 46 T cells/million PBMCs. All the means were significantly higher than the mean of the highest observation period T-cell frequencies for control cyclin D1 peptide, which was only 22 ± 5 T cells/million PBMCs. The mean of the highest observation period antibody concentration for p59 was 2.2 ± 0.3 (SEM), 2.3 ± 0.3 for p88, 1.8 ± 0.2 for p422, and 2.4 ± 0.3 μg/mL serum for p885 (Fig. 4N). The mean of the highest observation period antibody concentration for HER2 protein was 8 ± 1 μg/mL serum. All the means were significantly higher than the mean of the highest observation period antibody levels for control cyclin D1 peptide, which was only 0.57 ± 0.07 μg/mL serum.

Finally, the fraction of patients demonstrating T-cell immunity to whole HER2 protein remained high into the observation years. Although the numbers of evaluable patients in each period (year 1 observation and year 2 observation periods) varied, T-cell response rates to ICD remained above 75%, to ECD above 72%, and to either ICD or ECD above 85%. In the final period (i.e., year 2), 12 of 12 (100%) of evaluable patients demonstrated elevated immunity to HER2.

Although the trial was not designed to evaluate clinical outcomes, all patients remain alive at last vital status follow-up with a median of 2.3 years following trial enrollment (range, 0.1–3.4 years). With a median of 26.8 months of disease assessment follow-up, only two patients developed recurrence (one nodal recurrence 11.9 months and one in-breast recurrence 26.6 months, after trial enrollment). This translated into a 2-year DFS rate of 94.7% [95% confidence interval (CI), 68.1%–99.2%]. The in-breast recurrence was deemed likely a new primary rather than a local recurrence as the new lesion occurred in the opposite quadrant of the breast (9 vs. 2 o'clock position). The patient who developed nodal recurrence did not complete the vaccination treatment and only received four vaccinations. There were no distant recurrences.

The goals of this trial were to assess the safety and immunogenicity of a novel broad coverage HER2-targeted Th-cell vaccine. The vaccine was found to be safe overall. Furthermore, the vaccine was highly immunogenic and rapidly expanded T cells that recognized naturally processed HER2 antigens in 90% of patients. Patients with weaker vaccine-mediated responses were those that demonstrated elevated immunity prior to vaccine administration. The vaccine also elicited HER2-specific antibodies. Both elevated T-cell and antibody immunity were stably elevated at least 2 years after last vaccine.

One potential reason for the good safety profile is that the vaccine targets CD4 T cells whose activity is regulated by HLA class II, which is typically only expressed on immune and stromal cells (21). Because normal tissues shed very little HER2, it is likely that there is very little presentation of HER2 on HLA class II in normal tissue that would be targeted by vaccine-induced T cells (27). Historically, we have found very few HER2-specific T-cell responses to the vaccine HLA class II peptides in normal healthy individuals, a finding which is indicative of either a lack of prior exposure to the epitopes or to tolerance. A lack of prior exposure is most likely because the possibility of tolerance is not supported by data that the epitopes are highly immunogenic in the majority of patients (14).

A key finding is that immune responses were rapidly induced with a median time to maximal immunity of 2 months. This finding contrasts our previously published study using a degenerate FRα-based HLA class II vaccine in which we observed a median time to maximal immunity of 5 months in healthy disease-free patients previously diagnosed with either ovarian (n = 14) or breast cancer (n = 8). One potential reason for the differential response could be prior exposure to antigen, which could have resulted in higher levels of immune memory. Patients in this trial all demonstrated expression of HER2 in their cancers, whereas expression of FRα was not required in the FRα vaccine trial due to prior observations that FRα can be expressed on recurrent lesions even if absent on the primary lesion (21). Indeed, in that prior study, only 12 of 20 evaluable patients demonstrated FRα expression, which could have skewed the time to maximal immunity. In addition, in the FRα trial, patients were not previously exposed to immune therapy, whereas the patients in this trial received trastuzumab, which has been shown to induce endogenous immunity including activation of helper CD4 T cells as previously described above (7–9).

While the epitopes were designed primarily to stimulate T cells, antibody responses were also assessed. While increased levels of antibodies to the individual peptides were observed during the vaccine period, the responses were minor and did not meet response criteria. Considering that small increases of even 25% are associated with improved survival, perhaps the response criteria in this study were too rigid (7). Nonetheless, despite the low level response, levels of antibodies to the whole HER2 protein steadily and significantly increased during the vaccination period. Interestingly, we did observe stronger, but brief, increases in peptide-specific antibodies after vaccination ended, some of which persisted for reasons that were unclear but could have been due to the Th1-skewing nature of GM-CSF (28–30). Th1 T cells are known to favor the development of cell-mediated immunity and do not release high levels of cytokines (IL4 and IL5) that activate B cells and augment antibody responses. GM-CSF is a key regulator of Th2-mediated antibody-based immunity and is known to suppress both IgG and IgA responses in infectious disease models (31). In addition to suppressing Th2 induction, GM-CSF also induces subsets of regulatory T-cells that suppress production of autoantibodies (32). Thus, the immune system has multiple mechanisms that restrict antibody development that when modulated or removed could have resulted in increased antibody production that we observed in the trial.

An important finding from this trial is that vaccination resulted in HER2 protein–specific T-cell and antibody immunity that persisted well into the observation period. Specifically, time course analysis showed that the mean T-cell immune responses were stabilized at 15 months following the first immunization and continued unabated at significantly elevated levels at least to 2 years after the final vaccination. This is an ideal outcome for a vaccine for prevention of relapse (33). Despite the observation of persistence, we do not yet know whether the immune responses are sustainable over longer periods of time that are commensurate with the patterns of relapse that are observed in patients with HER2⁺ breast cancer, which can range from several months to several years (34). If the immune response is not sustainable, one possible remediation is the inclusion of boosters. In other previously reported HER2 vaccine trials, boosters are required to sustain HER2-specific immunity long term, and in at least one trial, there was some evidence to suggest that boosters may impact disease recurrence (19, 35).

The vaccine may synergize with trastuzumab or other HER2-specific antibody-based therapies to improve survival, as has been suggested previously (24). While designed for safety and given after trastuzumab therapy, a secondary objective was patient survival. Given the observation that only two of 22 patients recurred within the 24 months following last vaccination, 2-year DFS (94.7%, measured from after completion of adjuvant treatment and prior to vaccination initiation) compared favorably with historical data from the adjuvant trastuzumab trials. For example, in the combined analysis of trials NCCTG N9831 and NSABP B-31, 3-year recurrence-free survival (measured from just prior to start of all adjuvant therapy) of similar patients with stage II and III HER2⁺ breast cancer was 89.22% (95% CI, 87.72%–90.76%; personal communication with study team for applicable subset of patients reported by Perez and colleagues; ref. 3). Importantly, there was no distant recurrence observed in our cohort. This vaccine, given its broad patient coverage and ability to induce immune memory, should be evaluated for its ability to increase the cure rate for women with HER2 breast cancer, a disease controlled by the immune system (7–9, 25, 36). Our group has recently developed a phase II trial in which this vaccine will be coadministered with standard-of-care trastuzumab-based therapy in the adjuvant setting in patients with HER2⁺ breast cancer. The trial will ask whether the inclusion of vaccinations with trastuzumab-based regimens, along with periodic boosters, extends the invasive DFS period in patients at high risk for relapse.

K.L. Knutson reports receiving commercial research grants from Marker Therapeutics. M.S. Block reports receiving commercial research grants from Merck, Genentech, Bristol-Myers Squibb, Pharmacyclics, and Marker Therapeutics. G. Wilson is an employee/paid consultant for Marker Therapeutics. No potential conflicts of interest were disclosed by the other authors.

Conception and design: K.L. Knutson, L. Karyampudi, G. Wilson, A.C. Degnim

Development of methodology: K.L. Knutson, M.S. Block, C.L. Erskine, A.B. Dietz, A.C. Degnim

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): K.L. Knutson, M.S. Block, C.L. Erskine, T.J. Hobday, D. Padley, T.K. Mangskau, S. Chumsri

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): K.L. Knutson, M.S. Block, N. Norton, S. Chumsri, A.C. Dueck, G. Wilson, A.C. Degnim

Writing, review, and/or revision of the manuscript: K.L. Knutson, N. Norton, T.J. Hobday, A.B. Dietz, M.P. Gustafson, S. Chumsri, A.C. Dueck, L. Karyampudi, G. Wilson, A.C. Degnim

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): K.L. Knutson, M.P. Gustafson, D. Puglisi-Knutson, T.K. Mangskau, A.C. Degnim

Study supervision: K.L. Knutson

This work was supported by grants from the U.S. NIH (P50CA116201-Mayo Clinic Breast Cancer SPORE) and P30CA015083-Mayo Comprehensive Cancer Center Grant, Marker Therapeutics, Inc., and the U.S. Department of Defense (W81XWH-18-1-0563, W81XWH-18-1-0564, W81XWH-16-1-0265 and W81XWH-16-1-0266, all to K.L. Knutson and A.C. Degnim). The authors are grateful to the administrative and clinical staff in the Mayo Clinic Comprehensive Cancer Center for their assistance throughout the trial. The authors are grateful to the patients with breast cancer who participated in this study.

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.