Patients with MBC were enrolled in the phase II clinical trial at Barbara Ann Karmanos Cancer Institute (KCI) and Wayne State University (WSU) in Detroit, Michigan between July 2010 and August 2014. It was monitored by the KCI data safety monitoring committee. All patients signed an informed consent form prior to enrolment.

Patients with stage IV BrCa with HER2 amplification of 0–2+HER2 (0–2+) were enrolled into the phase II clinical trial (figure 1A) to evaluate if HER2 BATs infusions could extend OS and progression free survival (PFS). PBMC were collected from patients using a single apheresis, activated with anti-CD3 mAb and expanded with interleukin 2 (IL-2) to ATC. The ATC were harvested after 12–16 days of culture, armed with HER2Bi,3 and cryopreserved in aliquots for infusions.5 After apheresis, patients received four cycles or 4 months of their oncologist’s choice chemoT. The original protocol included a single dose of lymphodepleting cyclophosphamide 1.0 gm/m² (Cy) on day −7 prior to the first dose of HER2 BATs. Immune studies in the first five patients (#1–5) suggested that the addition of Cy to 4 cycles or 4 months of chemotherapy was impairing endogenous immune. Therefore, further immune depletion was eliminated from the protocol. In an effort to increase immune responses, the last five patients (#28 – #32) were given low dose IL-2 300 000 IU/m² subcutaneous injections daily and granulocyte-macrophage-colony stimulating factor (GM-CSF) 250 µg/m² two times per week. A dose of 20×10⁹ HER2 BATs per infusion (a total dose of 80×10⁹ HER2 BATs) was selected based on safety profiles, immune and clinical responses in our phase I clinical trial.5 HER2 BATs were then thawed and infused once a week for 3 weeks followed by a booster infusion of 20×10⁹ given 12 weeks after the third infusion. The protocol schema is shown in figure 1A.

Eligibility criteria: Eligible patients were ≥18 years with histologically proven HER2 negative BrCa as defined by a fluorescence in situ hybridization (FISH) ratio <2.2. Radiological/clinical evidence of metastases was required with noted progression after prior therapy. Patients were estrogen and/or progesterone receptor positive(ER/PR+) (HR+) or TNBC. Eligible patients had progressed after ≥1 line of hormonal and/or chemoT with Karnofsky ≥70% or Eastern Cooperative Oncology Group score of 0–1 and life expectancy ≥3 months.

Required initial laboratory data included: lymphocyte count is >500 mm³; granulocytes >1000/mm³; platelet count >50 x 10⁹/L; hemoglobin 80 g/L; blood urea nitrogen <1.5 times normal; serum creatinine <1.8 mg/dL; creatinine clearance ≥60 mL/min (can be calculated using the Cockcroft & Gault equation); bilirubin <1.5 times normal; alanine aminotransferase, aspartate aminotransferase, and alkaline phosphatase <5 times upper normal; negative HIV, hepatitis B or hepatitis C infection serology; left ventricular ejection fraction ≥45% at rest (multiple-gated acquisition scan or echocardiogram and pulmonary function tests of forced expiratory volume in 1 s -forced expiratory volume in 1 s, diffusing capacity, and forced vital capacity ≥50% of predicted.

Measurable or evaluable metastatic disease documented by radiograph, computerized tomography (CT) scan, positron emission tomography (PET)/CT scan, magnetic resonance imaging (MRI), bone scan, or physical examination was required. Each patient was required to have at least one bi-dimensionally measurable lesion that had not been irradiated with at least one diameter being ≥10 mm for liver, lung, skin lesions, and ≥15 mm for lymph node metastases. Biopsy of recurrent site(s) was not required. Biopsy of accessible sites before and after HER2 BATs was an option.

The heterconjugation of OKT3 (Miltenyi, Auburn, California, USA) with trastuzumab (Herceptin, Genentech) is well described.3 In brief, OKT3 was cross-linked with Traut’s reagent and trastuzumab was cross-linked with sulfosuccinimidyl-4-(N-maleimidomethyl)cyclohexane-1-carboxylate (Sulfo-SMCC). The two cross-linked mAbs were heteroconjugated overnight to produce anti-CD3 × anti-HER2 BiAb.3

T cells from the aheresis product were activated with 20 ng/mL of OKT3 and expanded in 100 IU/mL of IL-2 for 12–14 days in Roswell Park Memorial Institute Medium (RPMI-1640 medium) containing 2% human serum without antibiotics in GE Bioreactors as described.5 Harvested ATC were armed with HER2Bi at a preoptimized concentration of 50 ng/106 ATC as described.5 Armed ATC were washed twice to remove unbound HER2Bi and cryopreserved in four aliquots. The HER2 BATs were tested for specific cytotoxicity, immune subsets, pathogens, mycoplasma, and endotoxin prior to release for clinical infusion.5

The harvested ATC product and PBMC at multiple time points after BATs infusions from the patients were evaluated to assess changes in phenotype by staining for CD3⁺, CD4⁺, CD8⁺, CD25⁺, CD19⁺, CD20⁺, CD45RO⁺, CD45RA⁺, CD19+, CD20+, CD56⁺, CD127⁺, CD11b+, CD33+, and HLA-DR⁺ cells. Cryopreserved PBMC from the time of apheresis served as baseline controls.

The number of specific cytotoxic T lymphocytes (CTL) was assessed with fresh PBMC from patients using SK-BR-3 (BrCa target) and K562 cells (natural killer (NK) cell target) as targets for IFN-γ Elispots to measure CD8-mediated memory CTL activity and CD4-mediated helper responses. IFN-γ EliSpots produced by PBMC were assessed after 18 hours of stimulation with SK-BR-3 or K562 as well as spontaneous IFN-γ EliSpots produced by PBMC at an effector to target ratio of 1:1 as previously described.6

Cytokines were measured in the serum at selected time points using a 25-plex human cytokine Luminex Array (Invitrogen, Carlsbad, California, USA) using Bio-Plex system (Bio-Rad Lab., Hercules, California, USA). The multiplex panel included IL-1β, IL-1 receptor antagonist (IL-1RA), IL-2, IL-2 receptor (IL-2R), IL-4, IL-5, IL-6, IL-7, IL-8, IL-13, IL-17, tumor necrosis factor (TNF-α), IFN-alpha (IFN-α), IFN-γ, GM-CSF, macrophage inhibitory protein (MIP-1α), MIP-1β, IFN-γ-induced protein-10 (IP-10), monokine induced by IFN-γ (MIG), Eotaxin, Regulated on Activation Normal T Cell Expressed and Secreted (RANTES) and monocyte chemotactic protein (MCP)-1. The limit of detection for these assays is <10 pg/mL. The cytokine levels were calculated from a standard curve.

The primary endpoint was to determine if infusions of HER2 BATs would increase the TTP beyond the estimated 2 months expected after 3 cycles of chemoT and to confirm the safety of HER2 BATS. SD (SD) and OS were measured from the first infusion. Published trials on chemoT alone reported a median TTP of around 1.6 months when calculated from completion of 3 cycles of chemoT without additional treatment. We estimated the median TTP would be roughly 2 months under traditional treatment (estimated 4-month TTP was 25%). We hypothesized that HER2 BATs treatment would improve the median TTP by 2 months (estimated 4-month TTP of 50%) using one-stage design test for a 4-month TTP ≤25% vs a 4-month TTP ≥50%. If ≥9 out of 26 patients had not progressed by the 4-month follow-up, we could declare that the treatment was effective on TTP. The secondary endpoint was OS among all patients which will be investigated from enrolment and from the first BATs infusion. The Kaplan-Meier (K-M) method was used to estimate TTP and OS. TTP and OS was examined among the chemoT-sensitive (ChemoS) and chemoT-resistant (ChemoR) groups. The secondary endpoints were to assess response rates. TTP and OS were measured from enrolment of the date of the first infusion. Descriptive statistics were used to evaluate immune monitoring using Prism (GraphPad, V.8.0). We used a paired t-test to compare their levels at pre-, mid- and post-infusion time points separated by their chemoT response status (ChemoS or ChemoR). We also applied Cox regression models to examine if chemoT response status, each biomarker level at prefirst infusion, or the change between preinfusion and postinfusion would be associated with TTP or OS. Biomarkers in ELISpot were log10-transformed in data analysis to improve their normality of distribution.

Table 1 summarizes patient age, type of disease, status of disease after chemoT, total dose and number of HER2 BATs infusions, those with SD at 4 weeks after starting infusions, TTP, OS, tumor markers, and number of lines (L) of hormonal therapy, chemoT, radiation or antibodies. There were 32 evaluable HER2 negative patients who received therapy out 42 patients who were originally enrolled. Eight patients had TNBC and 24 patients had HER2-HR+MBC. The first five patients received a 1.0 g of Cy/m² intravenously 7 days prior to the first HER2 BATs infusion. Since immune responses did not appear to be as robust as seen in the phase I study, Cy was stopped after the first five patients (#1–5). In an effort to increase endogenous immune responses, the last five patients were given IL-2 300 000 IU/m² daily and GM-CSF 250 µg/m² two times per week (#28 – #32).

There were no severe treatment related toxicities. The HER2 BATs related grade 1–2 side effects were chills, hypotension, fever, hypertension, headaches, fatigue, nausea, and vomiting (online supplemental table S1). There were no irreversible grade 3 side effects that persist at grade 3 >72 hours.

There was one complete response (CR), 8 partial responses (PR) and 14 patients with SD after chemoT in 32 patients prior (table 1). Fifteen patients were SDor better at 4 weeks after the third BATs infusion. Nine patients who progressed during chemoT were able to receive some infusions of HER2 BATs. The lines (L) of therapy are summarized in table 1. Eight patients developed a PR response to chemoT and one patient (IT 20082) with progressive metastatic TNBC became stable after chemoT for 4.9 months with survival of 14.6 months. Interestingly, in the group of 8, 2 patients (IT 20073 and IT 20095) with TNBC who developed PRs to chemoT survived 12.4 and 12.2 months, respectively. Of note, the T cells from IT 20029 with chemosensitive disease who had received 9 lines of therapy expanded to a dose of 72.3×10⁹; IT 20029 was stable for 5.8 months and survived 17.4 months. On the other hand, T cells from IT 20085 with TNBC who had received 1 line of chemoT did not expand. Yet, the patient’s disease was stable for 4.9 months and the patient survived 42 months after receiving a dose of 10.7×10⁹ HER2 BATs.

The primary goal of this study was to attain ≥9 of 26 (34.6%) patients stable or better at 4 months after the last infusion. Although we were not successful in achieving the primary goal, 9 of 32 (28.1%) evaluable patients were stable for >4 months after their last infusion. The nine patients who were stable at 4 months included 3 TNBC and 6 HER2-HR+ patients. One of the nine patients received a total of only 17×10⁹ HER2 BATS. The mean TTP for all 32 patients was 2.7 months. The mean TTP was 1.4 months for the 8 TNBC patients and 2.5 months for the 24 HER2-HR+ patients (K-M curve not shown). The swimmer’s plot shows the TTP and OS for individual patients (figure 1B).

The median OS (with 95% CI) for all 32 patients, 8 TNBC patients, and 24 HER2-HR+patients is 13.1 (8.6–17.4), 12.3 (2.1–17.8), and 15.2 (8.6–19.8) months, respectively (figure 2A). For the chemoS group (n=25), the median OS was 14.6 (9.6–21.8) months, whereas the median OS for the chemoR (n=7) group was 8.6 (3.3–17.3) months (figure 2B and K–M, NS). In the HER2-, HR+ patients, the chemoS group had a median OS of 16.5 (9.1–24.2) months with a TTP 3.1 (1.4–4.2) months (figure 2C) whereas the patients in the chemoR group (n=8) had shorter median OS of 8.6 months with a TTP of 1.3 months.

The TNBC patients with chemoS disease (n=6) had a median OS of 12.3 months and the two TNBC patients with chemoR disease had median OS of 10.6 months (figure 2D). The TTP for the chemoS TNBC patients was 3.2 months whereas the TTP for the chemoR TBNC patients was 1.7 months.

Table 1 summarizes total lines of therapy before the chemoT given as part of this protocol for all patients. Online supplemental table S3 provides the mean total lines of therapy prior to and after chemoT and the proportion of each group of patients who received 1, 2, 3, and ≥4 L. The mean L of chemoT for HR+, TNBC, and all patients at enrollment were 3.75, 2.38, and 3.4 L. After receiving the chemoT part of the protocol, the mean L of chemoT for HR+, TNBC, and all patients becomes 4.75, 3.28, and 4.4 L, respectively. The striking finding is that 50, 25, and 43.6% of the HR+, TNBC, and all patients were heavily pretreated with ≥4 L of chemoT.

The median OS for first (68–83 × 10⁹), second (60–65 × 10⁹), third (40–57 × 10⁹), and fourth (11–38 × 10⁹) cell dose level quartiles were 12.1, 17.2, 10, and 13.2 months, respectively, and were not significantly different (figure 2E). Although the median OS of 21.9 months appeared to be higher for patients who received four infusions than the median OS of 15 months for patients who received 2–3 infusions, it was not significant (figure 2F, p=0.11). Patients IT20085 and IT 20101 highlight how cell doses and tumor markers did not correlate with TTP or OS. IT20085 with TNBC disease, who received only 1 line of chemoTand a cell dose of 10.7×10⁹ HER2 BATs, had a TTP of 6.7 months and OS of 42.0 months. IT 20101 with HER2-HR+ disease, who received three lines of chemoTand four infusions totaling 83.1×10⁹ HER2 BATs, had a TTP of 6.7 months and OS of 55.0 months. The differences may be due to heterogeneity of different individual immune systems.

In order to address the question of whether four cycles or 4 months of chemoT prior to receiving HER2 BATs infusions affected the TTP and OS, TTP and OS were analyzed from the time of enrollment prior to chemoT. The median OS from the time of initiating chemo for all 32 patients, HER2-HR+patients, and TNBC patients were 17.0 (12.0–21.8), 19.0 (12.0–24.0), and 15.1 (5.9–221.3) months, respectively (online supplemental table S2). The median OS for the chemoS and chemoR groups were 18.8 (13.7–24.0) and 11.6 (5.9–21.3) months, respectively. The key caveat is that patients who progressed during their 4 cycles of chemoT were not included in the analysis. The median TTP from enrollment for all 32 patients, HER2-HR+patients, and TNBC patients were 6.2 (5.5–7.5), 2.9 (5.6–7.7), and 5.8 (4.0–8.2) months, respectively (online supplemental table S2). For the chemoS group (n=25), the median TTP from the time of enrollment was 6.4 (5.8–8.2) months, whereas the median TTP for the chemoR (n=7) group was 5.5 (3.8–6.4) months. In the HER2-, HR+ patients, the chemoS group (n=19) from the time of enrollment had a median TTP of 6.4 (5.8–9.1) months whereas the five patients in the chemoR group had shorter median OS of 5.6 months with a TTP of 2.6 months (online supplemental table S2).

A log-rank (Mantel-Cox) test was used to ask whether those who had SD, PR, or CR survived longer than those who had PD after 4 months or 4 cycles of chemoT, The differences (figure 2G) between those who were SDor better and those who had PDwas highly significant (p=0.029). These data show that those who were clinically stable after chemoT survived longer than those who had PDjust prior to IT.

Cox regression models were used to determine if chemoT-response status was associated with TTP. The hazard rate for TTP in chemoS patients was 39% of that in chemoR patients, indicating that chemoS patients progressed significantly slower than chemoR patients (HR=0.39, p=0.042). For OS, the hazard rate in chemoS patients was 43% of that in chemoR patients, which indicated that chemoS patients had significantly better OS than those chemoR patients (HR=0.39, p=0.068).

Immune phenotyping and CD4/CD8 ratios in PBMC obtained pre-IT, mid-IT, and post-IT are summarized for patients who were stable and patients who had progressed in online supplemental table S4. Online supplemental figure S1 shows CD4/CD8 ratios in patients with SD (n=9) vs patients who had PD (n=23) at study time points. Although there were no statistically significant differences between patients who had SD compared with patients with PD at any time point, the data suggest there was an increase in CD4/CD8 ratios in patients who had SD when compared with the CD4/CD8 ratios in those with PD at the post-IT time points from the pre-IT baseline. There were no statistical correlations observed between immune phenotypes and clinical responses except for the proportion of CD19 +B cells.

The proportion of CD19 +B cells was the only immune cell populations that was significantly associated with TTP or OS. The hazard rate increased 5% with one percent increase of CD19 +B cells at pre infusion #1 indicating that patients progress significantly faster if their CD19 +B cells were higher at pre infusion #1 (HR=1.05, p=0.029) compared with patients with lower CD19 +B cells at pre infusion #1

We examined whether there were any correlations between clinical responses and the proportion of T cell subsets (CD3, CD4, CD8, CD45RA, CD45RO, CD19, CD56, Tregs, MDSC) in HER2 BATs product or PBMC phenotypes at multiple time points during and after immunotherapy (IT). There were no correlations between T cell subsets in the ATC product or PBMC after HER2 BATs infusions and OS and TTP.

Tumor markers decreased in 13 of 23 (56.5%) patients with evaluable tumor markers. In those patients whose tumor markers decreased, the 4 of 8 (50%) who were evaluable patients received four infusions each (total dose of 64.8–80 x 10⁹ HER2 BATs). There was no correlation between the number of doses or the total amount of HER2 BATs received and tumors markers after IT. Ten of 23 (43.4%) patients had increases in their tumor markers.

Fresh PBMC from 32 patients were tested for CTL activity by measuring the IFNγ ELISpots on exposure to SK-BR-3 or K562 at pretherapy (pre-IT), during therapy (mid-IT or preinfusion #3), and 1 month after treatment (post-IT). A significant increase (p<0.02) in the numbers of T cells in PBMC secreting IFNγ ELISpots in response to SK-BR-3 cells in PBMC from patients with SD at post-IT compared with pre-IT baseline (figure 3). There was also a significant increase (p<0.02) in innate immunity responses as seen in IFNγ ELISpots against NK specific K562 cells by PBMC from patients with SD at mid-IT compared with pre-IT baseline (figure 3, lower panel). These data clearly show immune responses directed at SK-BR-3 and K562 cells could be detected mid-IT or post-IT.

A panel of 25 cytokines and chemokines were tested in serum samples at pre-IT, mid-IT and post-IT from patients who received three or four HER2 BATs infusions. Online supplemental table S4 shows fold change for each cytokine or chemokine at pre-IT and post-IT, the fold change, and the p values. Th₁ and Th₂ cytokines (IL-2, IL-2r, IL-12, TNF-α, IFN-ɣ, GM-CSF, IL-6, IL-4, IL-10, IL-13), and chemokines (IP-10, MIG, MIP-1α, MIP-1β) using a multiplex array with the Luminex System (figure 4A–C). The fold changes were all highly significant for those listed in online supplemental table S5. The upper panel of figure 4A shows that Th₁ cytokine levels IL-12, IL-2, IFN-γ, TNF-α and IL-2R increased significantly (p<0.0001 p<0.0005) at mid-IT and post-IT compared with pre-IT levels. The fold increases for Th₁ cytokines IL-2, IL-2R, IL-12, TNF-α, IFN-ɣ, GM-CSF were 23.8, 1.6, 46.2, 14.5 times baseline, respectively. The fold increases for the chemokines IP-10, MIG, MIP-1α, and MIP-1β were 2.3, 5.7, 25.0, and 10.9 times baseline, respectively. Among the Th₂ cytokines (figure 4B), IL-4, IL-6, IL-10, and IL-13 showed significantly increased levels at mid-IT and post-IT (p<0.0005) compared with pre-IT levels (figure 4B). IL-6 levels were significantly elevated at mid-IT and post-IT (p<0.001) without concomitant increases in clinical toxicities. Intriguingly, IFNγ induced chemokines IP-10 and MIG levels (figure 4C) increased significantly at mid-IT and post-IT along with T cell recruiting chemokines MIP-1α and MIP-1β (p<0.0001 and p<0.0003) compared with pre-IT levels. The mean = Th₁[IL-2+IFNγ]/Th2 [IL-4+IL-10] ratio in the lower right of panel of figure 4, shows a Th₁-type response induced as a function of HER2 BAT infusions, ratio increasing from 1 at pre-IT to 3.5 at post-IT. It should be noted that fold changes of serum cytokines/chemokines were greater than 10-fold for IL-2, TNF-α, IFN-ɣ, MIP-1α, and MIP-1β. These findings were consistent with increased specific IFN-γ production observed by ELISpot analysis of patient PBMC exposed to SK-BR-3 tumor cells at the post-IT time point. Together these immune data show that there is activation of a systemic endogenous immune effect induced by infusions of HER2 BATs. Other cytokines and chemokines including IL-1β, IL-1RA, IL-5, IL-7, IL-8, IL-17, Eotaxin, RANTES and MCP-1 did not increase or remained unchanged from pre-IT (data not shown).

IL-12p70, produced mainly by activated monocytes, is the principal cytokine for polarizing T cell responses toward Th₁. Furthermore, IL-12 is known to enhance the cytotoxic functions of NK and CD8 +T cells. We observed an increase in serum levels of IL-12 at mid-IT and pos-tIT time points in patients with MBC (figure 4A). These data provide evidence that HER2 BATs infusions induce systemic Th₁-type antitumor immunity that may activate monocytes during the process of T cell engagement of the tumor to produce IL-12.