Anti-HER2 CD4+ T-helper type 1 response is a novel immune correlate to pathologic response following neoadjuvant therapy in HER2-positive breast cancer

A progressive loss of circulating anti-human epidermal growth factor receptor-2/neu (HER2) CD4+ T-helper type 1 (Th1) immune responses is observed in HER2pos-invasive breast cancer (IBC) patients relative to healthy controls. Pathologic complete response (pCR) following neoadjuvant trastuzumab and chemotherapy (T + C) is associated with decreased recurrence and improved prognosis. We examined differences in anti-HER2 Th1 responses between pCR and non-pCR patients to identify modifiable immune correlates to pathologic response following neoadjuvant T + C. Anti-HER2 Th1 responses in 87 HER2pos-IBC patients were examined using peripheral blood mononuclear cells pulsed with 6 HER2-derived class II peptides via IFN-γ ELISPOT. Th1 response metrics were anti-HER2 responsivity, repertoire (number of reactive peptides), and cumulative response across 6 peptides (spot-forming cells [SFC]/106 cells). Anti-HER2 Th1 responses of non-pCR patients (n = 4) receiving adjuvant HER2-pulsed type 1-polarized dendritic cell (DC1) vaccination were analyzed pre- and post-immunization. Depressed anti-HER2 Th1 responses observed in treatment-naïve HER2pos-IBC patients (n = 22) did not improve globally in T + C-treated HER2pos-IBC patients (n = 65). Compared with adjuvant T + C receipt, neoadjuvant T + C — utilized in 61.5 % — was associated with higher anti-HER2 Th1 repertoire (p = 0.048). While pCR (n = 16) and non-pCR (n = 24) patients did not differ substantially in demographic/clinical characteristics, pCR patients demonstrated dramatically higher anti-HER2 Th1 responsivity (94 % vs. 33 %, p = 0.0002), repertoire (3.3 vs. 0.3 peptides, p < 0.0001), and cumulative response (148.2 vs. 22.4 SFC/106, p < 0.0001) versus non-pCR patients. After controlling for potential confounders, anti-HER2 Th1 responsivity remained independently associated with pathologic response (odds ratio 8.82, p = 0.016). This IFN-γ+ immune disparity was mediated by anti-HER2 CD4+T-bet+IFN-γ+ (i.e., Th1) — not CD4+GATA-3+IFN-γ+ (i.e., Th2) — phenotypes, and not attributable to non-pCR patients’ immune incompetence, host-level T-cell anergy, or increased immunosuppressive populations. In recruited non-pCR patients, anti-HER2 Th1 repertoire (3.7 vs. 0.5, p = 0.014) and cumulative response (192.3 vs. 33.9 SFC/106, p = 0.014) improved significantly following HER2-pulsed DC1 vaccination. Anti-HER2 CD4+ Th1 response is a novel immune correlate to pathologic response following neoadjuvant T + C. In non-pCR patients, depressed Th1 responses are not immunologically “fixed” and can be restored with HER2-directed Th1 immune interventions. In such high-risk patients, combining HER2-targeted therapies with strategies to boost anti-HER2 Th1 immunity may improve outcomes and mitigate recurrence.

Utilizing a prospective cohort, we have recently demonstrated a progressive loss in anti-HER2 CD4 + T-helper type-1 (Th1) immunity across a tumorigenesis continuum in HER2 pos BC [10]. Interestingly, HER2-specific Th1 responses are preserved in healthy volunteers and patients harboring HER2 neg (0-1+) invasive breast cancer (IBC). In patients with HER2 pos IBC, this anti-HER2 Th1 deficit is not impacted by standard therapies (i.e., surgical resection, radiation, or T + C treatment), but can be restored following HER2-pulsed type-1-polarized dendritic cell (DC1) vaccinations. Moreover, depressed anti-HER2 Th1 responses predict an increased risk of subsequent recurrence in patients treated with adjuvant T + C [10]. These observations prompted us to investigate whether similar depressed anti-HER2 Th1 responses are observed in another known harbinger of recurrence, non-pCR status following neoadjuvant T + C [8]; conversely, we hypothesized that preservation/ restoration of anti-HER2 Th1 responses may be associated with pCR.
In this study, we identified elevated anti-HER2 CD4 + Th1 response as a novel systemic immune correlate to pCR following neoadjuvant T + C in patients with HER2 pos IBC. Relatively depressed anti-HER2 Th1 responses in patients with non-pCR are not attributable to host-level T cell anergy, loss of immunocompetence, or increase in circulating immunosuppressive phenotypes. Importantly, this anti-HER2 Th1 deficit in patients with non-pCR is not fixed, and can be corrected with CD4 + Th1-directed immune manipulations via HER2-targeted DC1 vaccinations. To the best of our knowledge, these observations represent the first demonstration of a modifiable host-level oncodriver (HER2/neu)-specific immune disparity that is associated with pathologic response to neoadjuvant T + C. These findings may have important implications for immune monitoring and/or design of adjunctive immune therapies to improve outcomes in trastuzumab-treated HER2 pos BC patients.

Study design
After approval by the Institutional Review Board of the University of Pennsylvania, 87 patients with HER2 pos IBC were enrolled in a non-biased fashion (Table 1). Eligible patients had histologically confirmed IBC, HER2/ neu overexpression (i.e., immunochemistry (IHC) 3+ or 2+/fluorescence in situ hybridization (FISH)-positive) confirmed at our institution, no evidence of distant metastasis, and were not receiving immunosuppressive medications. Informed consent was obtained from all participants. Anti-HER2 CD4 + Th1 responses of recruited subjects were analyzed prospectively. Anti-HER2 Th1 responses in treatment-naïve (i.e., not receiving definitive therapy at enrollment) stage I-III HER2 pos IBC patients (n = 22) were established as an immunologic "baseline", and were compared with Th1 responses in stage I-III HER2 pos IBC patients who had completed T + C treatment (n = 65; i.e., either neoadjuvant or adjuvant T + C plus definitive surgery). In patients treated with T + C, analyses were stratified by sequence of chemotherapy (i.e., neoadjuvant versus adjuvant), and further sub-stratified by pCR and non-pCR status within the neoadjuvant cohort (Fig. 1). pCR was defined as absence of residual invasive cancer on pathologic examination of resected breast specimen(s) and sampled lymph nodes (i.e., ypT0/Tis ypN0).
Four patients with non-pCR were recruited to our adjuvant HER2-pulsed DC1 vaccination trial (NCT02061423); anti-HER2 Th1 responses in these patients were compared pre-immunization and post-immunization.
An empiric method of determining anti-HER2 Th1 response specificity was employed [10]. A positive response to an individual HER2 peptide was defined as: (a) threshold minimum of 20 SFC/2 × 10 5 cells in experimental wells after subtracting unstimulated background; and (b) ≥2-fold increase in antigen-specific SFCs over background. Three metrics of anti-HER2 Th1 response were defined for each cohort: (a) responsivity (proportion of patients responding to ≥1 peptide), (b) repertoire (mean number of reactive peptides), and (c) cumulative response across 6 peptides (SFC/10 6 cells). A sample calculation is illustrated in Additional file 1: Figure S1. Inter-assay precision of ELISPOT assays was validated as described previously [14].

Vaccination procedure and trial design
We have initiated a phase I adjuvant HER2-pulsed DC1 vaccination trial for patients with HER2 pos IBC with residual disease following neoadjuvant T + C (NCT02061423). Eligible patients are 18 years or older, have Eastern Cooperative Oncology Group (ECOG) performance status score of 0 or 1, and have biopsy-proven stage I-III HER2 pos IBC. The primary endpoint of this trial is safety/feasibility; however, we report an interim analysis of anti-HER2 immune responses following vaccination (a secondary endpoint) in recruited patients (n = 4) as proof of principle of its immunogenicity in this heavily pre-treated population. Monocytic dendritic cell precursors (CD14 pos peripheral blood monocytes) were obtained from subjects via tandem leukapheresis/countercurrent centrifugal elutriation. Dendritic cells (DCs) were cultured overnight in macrophage serum-free medium (Cellgro, Manassas, VA, USA) with granulocyte monocyte colony stimulating factor (GM-CSF, 250 IU/mL; Berlex, San Pablo, CA, USA) and IL-4 (1000 u/mL; R&D Systems, Minneapolis, MN, USA) -these are considered immature DCs (iDCs). The following day, iDCs were pulsed with the aforementioned six HER2 major histocompatability class (MHC)-II promiscuous-binding peptides (42-56, 98-114, 328-345, 776-790, 927-941, 1166-1180). After 8-12 h incubation, IFN-γ (1000 U/mL) was added; the following day, National Institutes of Health (NIH) reference standard lipopolysaccharide (LPS) was added (10 ng/mL) to achieve full DC activation to a DC1 phenotype 6 h before harvest. For HLA-A2.1 pos patients, DC1 were pulsed with two MHC class I binding peptides (369-377, 689-697). Harvested cells were washed and lot release criteria of >70 % viability, negative Gram stain, and endotoxin <5 EU/kg confirmed.
Immunizations were administered in the NIH-designated General Clinical Research Center at the Hospital of the University of Pennsylvania. Injections comprised 10-20 × 10 6 HER2-pulsed DC1s suspended in 1 mL sterile saline, and administered by ultrasound guidance into groin lymph nodes [12,15]. Immunizations were administered once weekly for 6 weeks, followed by three booster doses spaced 3 months apart.

Statistical analysis
Descriptive statistics summarized distributions of patient characteristics and immune response variables. Data transformation of the cumulative response variable (natural log or square root) was applied to meet the assumptions of parametric testing, where applicable. The unpaired or paired Student's t test (parametric continuous data), Mann-Whitney (non-parametric continuous data), and chi square (χ 2 ) tests (categorical data) were used for two-group and univariate comparisons between pCR and non-pCR cohorts. To determine independent correlates of pCR, variables with a trend toward significance on univariate testing (p <0.20) were entered into a forward, stepwise multivariable logistic regression model (p <0.05 for entry, p <0.10 for exit). A p value <0.05 was considered statistically significant. All tests were two-sided. Analyses were performed using Prism 5.0 (GraphPad Inc., La Jolla, CA, USA) and SPSS version 22 (IBM Corp, Chicago, IL, USA).

Anti-HER2 T cell immune responses correlate strongly with pCR
In the cohort receiving neoadjuvant T + C, patients achieving pCR (cohort E; Fig. 1 Fig. 3a). Of note, median duration from initiation of neoadjuvant T + C to study enrollment did not differ between pCR and non-pCR cohorts (23.5 vs 26.5 months, p = 0.44).
Anti-HER2 Th1 responsivity is independently associated with pCR following multivariable analysis The independent association between IFN-γ + anti-HER2 Th1 responses and pCR was evaluated by controlling for confounding from relevant demographic and clinicopathologic characteristics. Upon univariate testing, pCR and non-pCR cohorts did not differ significantly by age, menopausal status, race, BMI, comorbidity, presence of LVI, nuclear grade, or utilized T + C regimens. However, pCR patients were more likely to have ER/PR neg tumors compared with patients with non-pCR (68.8 % vs 29.2 %, p = 0.02). Although pCR patients demonstrated a trend toward presentation at lower (i.e., stage II) clinical stage (68.8 % vs 41.7 %, p = 0.12) and less frequent need for mastectomy (50.0 % vs 79.2 %, p = 0.09), these comparisons did not reach statistical significance ( Table 2).

Anti-HER2 Th1 deficit in patients with non-pCR can be corrected with HER2-targeted CD4 + Th1 immune interventions
We have previously demonstrated that intranodally injected HER2-pulsed DC1s elaborate abundant IL-12p70 and polarize naïve CD4 + T-cells to IFN-γ/TNF-α-producing anti-HER2 Th1 in vivo [12,18]. When employed in HER2 pos ductal carcinoma in situ (DCIS) and patients with stage I HER2 pos IBC in phase I/II trials, autologous   HER2-targeted DC1 vaccination resulted in durable anti-HER2 Th1 immunity; pCR rates approached 25 % with substantial loss of target antigen in the remainder of patients (unpublished data) [15,19]. In order to determine the impact of HER2-Th1targeted immune interventions in high-risk non-pCR patients, four non-pCR patients (cohort G; Fig. 1) were recruited to our phase I adjuvant HER2-pulsed DC1 vaccination trial (NCT02061423); demographic and clinicopathologic characteristics of enrolled patients are detailed in Table 3. Subjects received six weekly injections followed by three booster doses at three-month intervals. Vaccination-induced anti-HER2 Th1 responses were followed prospectively; Th1 reactivity in individual patients pre-vaccination and post-vaccination is illustrated in Fig. 6a. In vaccinated subjects, evaluable anti-HER2 Th1 responses measured 6 months post-vaccination (i.e., prior to the third booster) indicated significantly improved anti-HER2 Th1 repertoire (3.7 ± 0.5 post-vaccination vs 0.5 ± 0.5 prevaccination, p = 0.014) and cumulative response (192.3 ± 16.4 vs. 33.9 ± 19.4 SFC/10 6 , p = 0.014) compared with pre-vaccination levels (Fig. 6b). Vaccinations were welltolerated, with only two cases of grade-1 toxicity observed.

Discussion
In the present study, we identify a novel systemic immune correlate to pathologic response following neoadjuvant HER2-targeted therapy in patients with HER2 pos IBC. Although not globally improved in all patients treated with T + C, anti-HER2 CD4 + T-cell immunity is more robust in patients achieving pCR compared with their non-pCR counterparts despite controlling for relevant demographic and tumor-related confounders. HER2specific Th1, but not Th2, CD4 + T-cells appear to be the dominant contributor to the circulating anti-HER2 IFN-γ + immune disparity; this anti-HER2 Th1 deficit is not attributable to host-level T cell anergy, lack of immunocompetence, or preponderance of immunosuppressive phenotypes in non-pCR patients. Importantly, this anti-HER2 Th1 deficit is modifiable, and can be corrected with HER2-pulsed DC1 vaccinations. In high-risk non-pCR patients, strategies to boost anti-HER2 Th1 immunity may be of benefit.
Pathologic complete response following neoadjuvant administration of HER2-targeted therapies is a reliable surrogate for favorable long-term outcomes in HER2 pos BC [7,8]; in fact, the Food and Drug Administration (FDA) supports pCR as a trial endpoint for drug approval [20]. Conversely, non-pCR portends a worse overall prognosis. Recent investigation has elucidated tumor cell-level mechanisms that account for suboptimal responses to HER2-targeted therapies, including overexpression of EGFR, cMYC, or ERBB3, and mutational loss of PTEN or activation of PI3K [21]. Beyond these factors, and the known association between ER negativity [8] which is not readily modifiableand pCR, there is a relative void in our understanding of host-level factors that impact response to HER2-directed therapies. In the current study, heightened circulating anti-HER2 CD4 + Th1 immune responses correlate strongly with pCR; conversely, the association of an anti-HER2 Th1 immune deficit with non-pCR warranted a search for therapeutic strategies that might correct this deficit. Fortunately, even in these heavily pre-treated patients, the Th1 deficit did not appear to be immunologically fixed and could be rectified with appropriate HER2-directed Th1 interventions. Thus, while a strategy such as withholding HER2targeted therapies in patients with negatively prognostic tumor-level genetic alterations (e.g., PI3K mutations) is impractical [22], augmenting the depressed anti-HER2 Th1 immunity in non-pCR patients may be more feasible as an adjunct to existing HER2-targeted therapies to improve clinical outcomes.
CD4 + Th1 cells have emerged as critical components of antitumor immunity. Via expression of T-bet and IFN-γ, Th1 cells indirectly mediate antitumor effects by enhancing CD8 + cytotoxic T-lymphocyte and NK cell function [23]. In addition, via elaboration of IFN-γ and TNF-α, HER2-specific Th1 cellsin synergism with trastuzumab-mediated HER2 blockadedirectly promote senescence and apoptosis, as well as HER2-specific CD8 + T cell targeting of HER2overexpressing tumors in vitro [10,24]. Indeed, the association between improved HER2 369-377 -specific CD8 + T-cell  immune responses and tumor eradication in pCR patients may reflect the ready availability of CD4 + T cell help. Moreover, a recent genomic analysis from the NCCTG-N9831 trial demonstrated a strong association between increased relapse-free survival in adjuvant trastuzumab-treated patients and a signature of immune function genes, including IFN-γ and TNF-α [21]. In the present study, a relative decay in circulating anti-HER2 T-bet + IFN-γ + (i.e., Th1), but not GATA-3 + IFN-γ + (i.e., Th2), phenotypes is associated with persistence of HER2 pos tumors following neoadjuvant T + C. Taken together, these data suggest that abrogation of immunologic, particularly anti-HER2 Th1 function, may represent a HER2 pos tumor-driven mechanism to evade immune surveillance during T + C treatment. Immune interventions aimed at restoring anti-HER2 Th1 function may be valuable in improving pathologic response following neoadjuvant T + C.
In parallel with these observations, growing evidence indicates that robust cellular immune responses in the tumor microenvironment are associated with improved outcomes in BC [25], particularly in HER2 pos subtypes [26]. Furthermore, an analysis from the GeparQuattro trial suggested that tumor-infiltrating lymphocyte (TIL) density correlates with pCR following neoadjuvant T + C; for every 10 % increase in TIL levels, a 16 % increase in pCR rates was observed [27]. The sizeable increase in circulating anti-HER2 Th1 populations in pCR patients in the present study may represent a systemic corollary to such immune-related changes in the tumor microenvironment, and lend further support to evidence that intact immune functionality, in addition to HER2-signaling inhibition, is critical in mediating antitumor effects following T + C treatment [28]. What is not immediately evident from our analysis, however, is whether the heightened anti-HER2 Th1 responses in pCR patients represent preservation of erstwhile immunity, or immune restoration following T + C treatment. If the latter is true, these data may further support immune restorative neoadjuvant interventions in order to improve pathologic response.
Other limitations merit discussion. First, given the retrospective and exploratory nature of the study design, the findings herein should be interpreted as hypothesisgenerating and warrant large-scale validation. Second, despite minimal demographic/clinical variability between treatment-naïve T + C -treated HER2 pos -IBC cohorts, the global lack of improvement in anti-HER2 Th1 responses following T + C treatment must be interpreted with caution, since these data were derived from an unpaired comparison between independent patient samples. Finally, while encouraging, definitive conclusions regarding the immune restorative impact of HER2-directed DC1 vaccination in high-risk non-pCR patients, cannot be drawn until completion and final reporting of this ongoing trial.
The translational implications of these findings bear emphasis. As discussed, they may justify addition of HER2-targeted Th1 immune interventions to neoadjuvant T + C regimens and/or in the adjuvant setting for high-risk non-pCR subgroups. Moreover, in light of our recent demonstration that depressed anti-HER2 Th1 immunity correlates with subsequent recurrence in patients treated with adjuvant T + C [10], monitoring high-risk patients with non-pCR for real-time fluctuations in anti-HER2 Th1 immunity may complement existing radiographic surveillance, and identify critical opportunities for therapeutic intervention. Incorporation of anti-HER2 Th1 immune detection protocols in future clinical trial design, particularly those investigating neoadjuvant HER2-targeted therapies, appears justified.

Conclusions
In summary, this is the first description, to our knowledge, of a critical association between anti-HER2 CD4 + Th1 immunity and pCR following neoadjuvant T + C in HER2 pos IBC patients. Although our data cannot confirm causality, the dramatic IFN-γ + anti-HER2 Th1 deficit observed in non-pCR patients following neoadjuvant T + C raises the possibility that immune rescue with HER2-Th1 interventions may complement standard HER2-targeted strategies in improving outcomes in these high-risk patients. While correction of the anti-HER2 Th1 immune deficit has already been observed in non-pCR patients recruited to our HER2-DC1 vaccination trial, longitudinal follow up and larger-scale studies will establish if such immune manipulations ultimately mitigate recurrence in such patients.