CD4⁺ T-Helper Type 1 Cytokines and Trastuzumab Facilitate CD8⁺ T-cell Targeting of HER2/neu–Expressing Cancers

HER 2/neu (HER2) is amplified in a number of solid malignancies, including breast, ovarian, gastric, and pancreatic cancers (1). This HER2 receptor tyrosine kinase (RTK) is critically involved in early uncontrolled growth, enhanced invasiveness, and metastatic spread (2, 3). Although the combination of HER2-targeted monoclonal antibodies (trastuzumab) with chemotherapy has dramatically improved outcomes in patients with HER2-overexpressing (HER2^pos) breast cancer (4, 5), significant resistance to therapy occurs, leading to recurrence (6).

There are emerging data that both CD4⁺ and CD8⁺ T-cell antitumor responses are critical in these aggressive tumors. Not surprisingly, enhanced infiltration of these immune cell subsets is associated with favorable clinical outcomes in HER2^pos tumors (7). The tumoricidal activity of antigen-specific CD8⁺ CTLs is also widely appreciated; indeed, CD8⁺ T cells recognizing the HLA-A2–restricted peptide 369–377 (HER2_369–377; KIFGSLAFL) have been identified in tumors from breast and ovarian cancer patients (8). Controversy exists, however, whether this epitope is actually processed and presented by HER2-expressing cancers. Utilizing HER2_369–377 (with adjuvant) to vaccinate patients with HER2^pos tumors generated postimmunization HER2_369–377-reactive CD8⁺ T cells that failed to recognize HLA-A2^pos tumor cells expressing HER2 (9). In addition, HER2_369–377 peptide vaccination in GM-CSF (E75) induced immune responses and improved clinical outcomes in patients with low HER2-expressing (1+)—but not in classically HER2^pos (3+ or 2+/FISH-positive)—breast cancer patients (10). The failure of HER2-specific CD8⁺ T-cell recognition may be explained by evidence that HER2 overexpression downregulates MHC class I expression by inducing defects in the antigen-processing machinery (APM; refs. 11–14), thereby mediating escape from immune surveillance.

In the current study, we attempted to reconcile this controversy by demonstrating that HER2_369–377 is endogenously presented by HER2-expressing cancer cells, and naturally recognized by HER2_369–377-specific CD8⁺ T cells in a class I–dependent manner. Furthermore, we demonstrate a critical cooperation between CD4⁺ T-helper type 1 (Th1) cytokines IFNγ/TNFα and HER2-targeted antibody trastuzumab in mediating restoration of class I expression and facilitating HER2_369–377-CD8⁺ T-cell targeting of HER2-overexpressing cancers. Concomitant induction of PD-L1 on HER2/neu-expressing cells by IFNγ/TNF and trastuzumab, however, has minimal impact on type 1–polarized dendritic cell (DC1)–sensitized HER2_369–377-CD8⁺ T-cell–mediated cytotoxicity. Although activation of EGFR and HER3 signaling significantly abrogates IFNγ/TNFα/trastuzumab-induced class I restoration, EGFR and HER3 receptor blockade rescues class I expression as well as HER2_369–377-CD8⁺ cytotoxicity of HER2/neu-expressing cells. As such, our novel findings have important implications for vaccine design and T-cell–directed therapies in patients with HER2-expressing cancers.

HER2-expressing breast cancer cell lines SK-BR-3 and BT-474 (HER2^high), MCF-7 (HER2^intermediate), MDA-MB-231 (HER2^low; American Type Culture Collection), and ovarian cancer cell line SK-OV-3 (HER2^intermediate) stably transfected with the HLA-A2 gene (SK-OV-3^A2; kind gift of Mary Disis, University of Washington) were immediately resuscitated and maintained in RPMI supplemented with 10% FCS (Cellgro). HLA-A2 status was verified (LABType SSO) by the Clinical Immunology laboratory at the Hospital of the University of Pennsylvania. HLA-A2/HER2 status of cell lines was verified by flow cytometry (Fig. 1A, data not shown).

HER2-expressing cells were treated with the following, either alone or in designated combinations: rhTNFα, rhIFNγ (BD Biosciences), trastuzumab (Genentech), lapatinib (Santa Cruz Biotechnology); rh-EGF (BD Biosciences), rh-Heregulin (Sigma-Aldrich); neutralizing anti-EGFR (LA1) and/or anti-HER3 (H3.105.5) antibodies, or IgG1 isotype control antibody (all Millipore); and neutralizing anti–PD-1 (MIH1; ref. 15) or IgG1 isotype control (eBiosciences). Specific cell treatments are detailed in Supplementary Methods.

HER2_369–377-specific CD8⁺ TCR clones were generated as previously described (16). Briefly, high-avidity HER2_369–377-reactive CD8⁺ T cells were isolated after in vitro HER2_369–377 stimulation of CD8⁺ T cells obtained from a patient with HLA-A2^posHER2^pos ductal carcinoma in situ who was previously vaccinated with HER2_369–377-pulsed autologous DC1. The vaccination protocol is summarized in Supplementary Methods. CD8⁺ T-cell stimulation with HLA-A2–matched or unmatched HER2-expressing cell lines, with subsequent sorting for tumor-activated HER2_369–377 T cells, allowed for isolation of HER2_369–377-specific CD8⁺ TCR. Primary human CD8⁺ T cells genetically modified to express this HER2_369–377 TCR specifically bound HER2_369–377-containing HLA-A2⁺ tetramers (16). GFP-transduced and nontransduced clones served as controls.

Monocyte-derived DCs from four HLA-A2^pos HER2^pos-DCIS donors, who had undergone HER2_369–377-pulsed DC1 vaccinations (Supplementary Table S1), were matured to a DC1 phenotype (high IL12-secreting) via IFNγ (1,000 U/mL) and LPS (10 ng/mL; ref. 17), and pulsed with HER2_369–377 peptide (Genscript; 50 μg/mL) 2 hours before harvest. DC1s were sensitized with postvaccination CD8⁺ T cells, isolated by immunomagnetic separation (EasySep; Stem Cell Technologies), by coculturing at a 20:1 (T cell:DC1) ratio in RPMI 1640 + 5% human serum. IL2 (30 IU/mL) was added on day 2. As previously described (17), this technique generates high functional-avidity HER_2369–377-specific CD8⁺ T cells. After 6 to 7 days, sensitized CD8⁺ T cells were harvested and tested 1:1 with HLA-A2 transporter (TAP)–deficient T2 hybridoma cells pulsed with HER2_369–377 or control class I peptides (1 μg/mL), and tested against MDA-MB-231 with or without HER2/HLA-A2 siRNA transfection, SK-OV-3^A2, MCF-7, and SK-BR-3 cell lines as indicated. HER2_369–377-TCR–transduced, GFP-transduced, and nontransduced CD8⁺ T-cell clones were also tested against T2 and HER2-expressing cells. Supernatants and tumor cells were harvested after 24 hours, and CD8⁺ T-cell recognition (by IFNγ ELISA) and cytotoxicity (by flow cytometry) were assessed.

For HER2 or HLA-A2 silencing, 3 × 10⁵ MDA-MB-231 cells were transfected with HER2, HLA-A2, or nontargeting siRNA sequences (25 nmol/L; ON-TARGETplus; Dharmacon) using RNAi Max Lipofectamine (Life Technologies) in serum-free medium. After 1 hour, medium was supplemented with 10% FBS; 20 hours later, cells were serum-starved for 48 hours, followed by coculture with HER2_369–377-sensitized or HER2_369–377-TCR–transduced CD8⁺ T cells. Before coculture, aliquots from individual treatment groups were obtained to confirm HER2/HLA-A2 silencing via flow cytometry or Western blot (Supplementary Methods), or both.

Cell suspensions were prepared in FACS buffer (PBS + 1% FCS + 0.01% azide); 7-AAD (viability stain), and FITC/APC/phycoerythrin (PE)–conjugated mouse anti-human CD11c, CD8, HLA-ABC, HLA-A2, HER2, IFNγRα/β, TNFαR1, PD-L1, PD-1, or subclass-matched controls (BD Bioscience) were utilized as indicated. HER2_369–377/HLA-A*0201 tetramers (MCL) were used to identify HER2_369–377-specific CD8⁺ T cells. Flow cytometry was performed using BD FACSCalibur; datasets were analyzed using CellQuest Pro software.

Following cell treatments as indicated, CFSE-labeled tumor cells were cocultured 1:1 with HER2_369–377-CD8⁺ T cells for 24 hours. Cells were harvested, stained with 7-AAD and FITC:anti-CD8, and subjected to flow cytometry. CFSE-positive, but not CD8-positive, cells were gated and analyzed; the percentage of apoptotic cells was calculated as 7-AAD⁺/(7-AAD⁺+7-AAD⁻) × 100%. Cytotoxicity was calculated as percentage of apoptotic tumor cells in CD8⁺ coculture minus background (i.e., apoptotic cells in tumor culture alone).

Refer to Supplementary Methods.

Capture and biotinylated detection antibodies and standards for IFNγ (BD Pharmingen) were used according to the manufacturer's protocols.

One-way ANOVA with post hoc Tukey paired testing was used for all ≥3-group comparisons. Student t test (parametric) or Mann–Whitney tests (nonparametric) were used for two-group comparisons. P ≤ 0.05 was considered statistically significant. Analysis was performed using Prism 5.0 (GraphPad Inc.).

The HER2_369–377 peptide is widely regarded as the immunodominant epitope recognized by lymphocytes from HLA-A2^pos patients with breast/ovarian cancer (8). HER2_369–377 was used to immunize mice transgenic for both HLA-A2.1 and human CD8; postimmunization splenocytes recognized HLA-A2^posHER2^pos human tumor cells (18). In spite of supportive preclinical evidence, the failure of CTLs from HER2_369–377-immunized patients to recognize HER2^pos tumor cells (9) has generated skepticism regarding the utility of this peptide for HER2-directed immunotherapy. More recently, several groups have attempted to explain this phenomenon by demonstrating that overexpression of a signal-competent HER2 RTK dramatically impairs MHC class I expression and APM components, thereby impairing CD8⁺ recognition of HER2-expressing cancers (12–14).

In the present study, although HER2^low MDA-MB-231 maintained robust surface class I expression, class I expression was severely diminished on HER2-overexpressing SK-BR-3 cells; postvaccination patient-derived CD8⁺ T cells sensitized with HER2_369–377-pulsed autologous DC1s recognized MDA-MB-231, but not SK-BR-3, cells (Fig. 1A). While refractory to HER2_369–377-sensitized CD8⁺-mediated lysis, SK-BR-3 cells were significantly vulnerable, however, to natural killer (NK)–mediated antibody-dependent cell-mediated cytotoxicity (ADCC; Supplementary Fig. S1). These data reinforce evidence that HER2 overexpression, and the associated downregulation of class I expression, reduces susceptibility of tumor cells to class I–dependent CD8⁺-mediated, but not to class I–independent NK-mediated, lysis.

Next, we evaluated the ability of postvaccination patient-derived high-avidity HER2_369–377-CD8⁺ T cells to recognize antigen-loaded target cells in IFNγ release assays. By IFNγ ELISA, DC1-sensitized HER2_369–377-CD8⁺ T cells showed highly specific recognition of HER2_369–377-loaded T2 cells, compared with control peptide–loaded T2 cells. A similarly specific recognition of HER2_369–377-loaded T2 cells, compared with control peptide–loaded T2 cells, was observed when cocultured with HER2_369–377-TCR–transduced CD8⁺ T cells, but not with nontransduced or GFP-transduced CD8⁺ T-cell controls (Fig. 1B).

Importantly, in order to determine if HER2_369–377-CD8⁺ T cells could recognize HER2-expressing cancer cells, DC1-sensitized HER2_369–377-CD8⁺ T cells were cocultured with HLA-A2^pos HER2^low-expressing MDA-MB-231 cells with or without HER2 siRNA transfection. Compared with nontargeting (NT) control siRNA, HER2 siRNA transfection resulted in depletion of HER2 protein expression by Western blot, as well as loss of HER2 surface expression by flow cytometry. Notably, HLA-A2 expression on MDA-MB-231 cells remained unaffected by HER2 interference (Fig. 1C, top). While NT siRNA-transfected MDA-MB-231 cells were specifically recognized by DC1-sensitized HER2_369–377-CD8⁺ T cells, this recognition was abrogated by silencing HER2 in HER2 siRNA-transfected MDA-MB-231 cells (Fig. 1C, bottom left). In order to corroborate these observations, a similarly specific recognition of NT siRNA-transfected MDA-MB-231 cells was demonstrated by HER2_369–377-TCR–transduced CD8⁺ T cells, compared with control nontransduced or GFP-transduced CD8⁺ T cells. This HER2_369–377-specific tumor recognition, however, was eliminated by silencing HER2 expression in MDA-MB-231 cells (Fig. 1C, bottom right).

In an effort to explore if HER2_369–377-CD8⁺ recognition of HER2-expressing cancer cells was contingent on HLA-A2, DC1-sensitized HER2_369–377-CD8⁺ T cells were cocultured with nontransfected, NT siRNA, and HLA-A2 siRNA-transfected MDA-MB-231 cells. HER2_369–377-CD8⁺ T cells only recognized nontransfected or NT siRNA-transfected, but not HLA-A2–silenced (via HLA-A2 siRNA), MDA-MB-231 cells (Fig. 1D). Together, these data indicate that HER2_369–377 is endogenously presented by HER2-expressing cancers; importantly, high-avidity HER2_369–377-specific CD8⁺ T cells can naturally recognize the HER2_369–377 epitope on HER2^low cancer cells maintaining abundant class I/HLA-A2 expression, but not on surface class I/HLA-A2–deficient HER2^high cells.

The immune escape provoked by HER2 overexpression on cancer cells warrants a search for strategies that restore surface MHC class I expression and improve sensitivity to CD8⁺-mediated recognition and lysis. Although CD4⁺ Th1 cytokine IFNγ upregulates surface class I expression in HER2-overexpressing murine models in vitro (11)—restoring CD8⁺-mediated lysis and/or tumor cell rejection in vivo (19)—it is comparatively less effective in reverting class I suppression in human HER2-driven tumors (13, 20). HER2 signaling is also increasingly recognized in activating the MAPK and PI3K/AKT signal transduction pathways (21), suggesting that targeting these pathways may influence class I expression (14, 22). In view of this evidence, we evaluated the effect of HER2-targeted tyrosine kinase inhibitors trastuzumab and lapatinib, as well as Th1 cytokines IFNγ and TNFα, on class I expression in HER2-expressing cancers.

A spectrum of HER2-expressing cell lines (MDA-MB-231, MCF-7, SK-OV-3^A2, BT-474, and SK-BR-3) was treated with IFNγ, TNFα, or trastuzumab alone, or in designated combinations. Compared with untreated tumor cells, treatment with TNFα or IFNγ alone increased class I expression in select (TNFα: BT-474; IFNγ: SK-OV-3^A2, BT-474), but not all, HER2-expressing cells. Dual IFNγ and TNFα treatment, however, significantly restored class I expression on all HER2-expressing cell lines evaluated (P < 0.05). Treatment with trastuzumab alone had little impact on class I expression compared with that in untreated cells; however, the combination of trastuzumab, IFNγ, and TNFα dramatically enhanced class I expression on all cells [MDA-MB-231 (P = 0.015), MCF-7 (0.05), SK-OV-3^A2 (P < 0.001), BT-474 (P < 0.0001), and SK-BR-3 (P < 0.001)]. Interestingly, class I expression was restored more effectively following triple therapy with trastuzumab/IFNγ/TNFα than with dual IFNγ/TNFα treatment in HER2^high [BT-474 (P = 0.006); SK-BR-3 (P = 0.03)], but not in HER2^intermediate (MCF-7 or SK-OV-3^A2) or HER2^low (MDA-MB-231), cells (P > 0.05; Fig. 2A).

Next, we determined if variability in MHC class I expression was related to IFNγ/TNFα receptor expression or treatment dose. By flow cytometry, IFNγRα/β (Supplementary Fig. S2A) and TNFαR1 (data not shown) expression was qualitatively similar across all cell lines tested. In HER2^intermediate/HER2^high cells, a dose–response relationship for class I expression was observed with increasing rhIFNγ doses (250—2,000 U/mL). A dose-saturation effect beyond the 1,000 U/mL rhIFNγ dose was observed following combination treatment with TNFα or TNFα/trastuzumab (Supplementary Fig. S2B); consequently, this standard dose was utilized for further experiments. The addition of trastuzumab to lower IFNγ concentrations (i.e., 250 or 500 U/mL) improved sensitivity of HER2^intermediate MCF-7/SK-OV-3^A2—but not HER2^high BT-474/SK-BR-3—cells to class I restoration. These data raise the intriguing possibility that trastuzumab—in concert with even moderate levels of Th1 cytokines in the tumor microenvironment—may aid CD8⁺ recognition of tumors without classical HER2 overexpression.

Consistent with findings in HER2^pos-gastric/esophageal cancer cells (22), treatment of HER2^high SK-BR-3/BT-474 cells with lapatinib alone did not restore class I expression appreciably. Moreover, compared with both IFNγ/TNFα treatment and trastuzumab/IFNγ/TNFα treatment, an attenuated upregulation in class I expression was observed when lapatinib was combined with IFNγ/TNFα (Supplementary Fig. S3). Mechanisms underlying this observation warrant investigation; importantly, these findings have clinical implications and may explain the superior head-to-head clinical efficacy of trastuzumab versus lapatinib observed in HER2^pos breast cancer patients (23).

We next sought to determine the impact of class I restoration on HER2_369–377-CD8⁺ T-cell recognition and lysis in HER2^{intermediate/high} cancers. Compared with untreated cells, treatment with TNFα, IFNγ (data not shown), or trastuzumab alone (Fig. 2B) did not augment HER2_369–377-CD8⁺ T-cell recognition or lysis of HER2^intermediate MCF-7/SK-OV-3^A2, and HER2^high SK-BR-3 cells. Dual treatment with IFNγ and TNFα, however, significantly enhanced HER2_369–377-CD8⁺-mediated recognition and lysis of HER2^intermediate MCF-7 and SK-OV-3^A2, but not HER2^high SK-BR-3, cells. The addition of trastuzumab to IFNγ/TNFα rendered HER2^high SK-BR-3 cells susceptible to both recognition and lysis by HER2_369–377-CD8⁺ T cells; the addition of trastuzumab to IFNγ/TNFα did not incrementally improve recognition or lysis of HER2^intermediate MCF-7/SK-OV-3^A2 cells (Fig. 2B and C).

Because IFNγ is known to induce immunosuppressive programmed death (PD)-ligand-1 (PD-L1) in cancer cells (24, 25), we examined the effect of Th1 cytokine/trastuzumab combinations on PD-L1 expression in HER2-expressing cell lines. Consistent with previous findings (25), constitutive PD-L1 expression was observed only in basal breast cancer subtype MDA-MB-231 cells. In all other HER2-expressing cells (SK-OV-3^A2, MCF-7, SK-BR-3), despite negligible endogenous levels, PD-L1 expression was inducible with IFNγ but not with TNFα or trastuzumab treatment alone. Dual IFNγ/TNFα treatment further enhanced PD-L1 expression in all cells; however, addition of trastuzumab to IFNγ/TNFα did not incrementally improve PD-L1 expression (Fig. 3A).

Next, we evaluated the competing effects of cytokine/trastuzumab-mediated class I and PD-L1 upregulation on HER2_369–377-CD8⁺ T-cell recognition of these tumors. Using HER2_369–377/HLA-A*0201 tetramers, scant PD-1 expression on DC1-sensitized HER2_369–377-reactive CD8⁺ T cells was observed (mean, 1.21 ± 0.2% of CD8⁺ T cells; Fig. 3B). Not surprisingly, treatment with PD-1–neutralizing, compared with isotype-control, antibody did not significantly affect HER2_369–377-CD8⁺ lysis of HER2^intermediate MCF-7/SK-OV-3^A2 or HER2^high SK-BR-3 cells (Fig. 3C). Although these data may explain the minimal impact of inhibitory PD-1/PD-L1 interactions on HER2_369–377-CD8⁺ cytotoxicity following a single in vitro DC1 sensitization, the upregulation of PD-L1 following Th1 cytokine/trastuzumab treatment justifies exploration of combination therapy with HER2-targeted antibodies, HER2-Th1 immune interventions, and PD-1/PD-L1 axis inhibition in HER2-expressing cancers.

The HER2 RTK signaling domain is activated upon heterodimerization with other HER family members (EGFR, HER3, HER4) or upon homodimerization (26). Given the inability of trastuzumab to inhibit EGFR/HER2 and HER2/HER3 heterodimers (27), escape signaling via EGF and/or HER3 receptors has been implicated as an important mechanism of resistance to trastuzumab (28). Therefore, we investigated the impact of EGFR and HER3 signaling on class I expression in vitro. HER2^high BT-474 and SK-BR-3 cells were pretreated with EGF (EGFR ligand) or Heregulin (HER3 ligand) and subjected to treatment with trastuzumab and IFNγ/TNFα. Activation of signaling via EGFR/HER3 together and HER3 alone, but not EGFR alone, rendered HER2^high BT-474 (Fig. 4A) and SK-BR-3 (data not shown) cells significantly resistant to the class I–restoring effect of trastuzumab/IFNγ/TNFα. Importantly, inhibition of EGFR and HER3-driven signaling with anti-EGFR and anti-HER3 antibodies rescued EGF and Heregulin-mediated resistance to class I restoration (Fig. 4B), and consequently HER2_369–377-CD8⁺ T-cell lysis (Fig. 4C), of HER2^high cells.

We have previously shown that HLA-A2^pos donor-derived CD8⁺ T cells, sensitized with HER2_369–377-pulsed DC1, directly recognize HER2-expressing cancer cells via an IL12-dependent mechanism (17). In the current study, we extend these observations by demonstrating that the HER2_369–377 epitope is not only endogenously presented on HER2-expressing cancers, but also naturally recognized on class I–abundant HER2^low, but not on class I–deficient HER2^high, cancer cells by HER2_369–377-specific CD8⁺ T cells. Our findings, therefore, may explain the clinical benefit paradoxically observed in patients with HER2^low (1+), but not classically HER2^high (3+), tumors following HER2_369–377 vaccinations in GM-CSF (10).

Because HER2-driven class I downregulation impedes CD8⁺ T-cell recognition of even immunodominant epitopes, such as HER2_369–377 in HER2-overexpressing cancers (13), we uncover a critical collaboration between cellular (IFNγ and TNFα) and humoral (trastuzumab) immunity in restoring class I expression in HER2-overexpressing cells, thereby rendering them susceptible to HER2_369–377-specific CD8⁺ T-cell–mediated targeting. Intriguingly, although synergism between Th1 cytokines IFNγ and TNFα appears sufficient for class I restoration and HER2_369–377-CD8⁺ T-cell targeting of HER2^{low/intermediate} cancers, class I restoration and ensuing targeting of HER2-overexpressing (HER2^high) cells is critically dependent on the cooperation between IFNγ/TNFα and trastuzumab-mediated HER2 blockade. Recent evidence suggests that MAPK signaling inhibition predominantly regulates MHC class I induction in HER2-overexpressing cancer cells, whereas preserved PI3K/AKT signaling further enhances class I expression compared with MAPK inhibition alone (13, 22). In the setting of trastuzumab-mediated inhibition of presumably both MAPK and PI3K/AKT signaling, it is possible that IFNγ/TNFα potentiate class I upregulation in HER2^high cells by (i) more complete abrogation of MAPK signaling and/or (ii) preferential rescue of PI3K/AKT signaling. Future studies should investigate signaling cascades underlying such effects.

The translational relevance of these novel observations bears emphasis. Combinations of HER2-targeted antibodies and immune interventions incorporating CD4⁺ helper epitopes may potentiate anti-HER2 CD8⁺ T-cell–directed immunotherapies and improve clinical outcomes in patients with HER2-overexpressing tumors. Indeed, we have recently initiated a phase I trial in HER2^pos breast cancer patients investigating such combinations in the neoadjuvant setting. Ultimately, given the impact of EGFR/HER3-mediated escape signaling on class I expression as well as CD8⁺ T-cell–mediated cytotoxicity in trastuzumab-treated HER2^pos cancers, a multivalent strategy targeting other HER family members—in addition to HER2—may prove most effective.

No potential conflicts of interest were disclosed.

Conception and design: J. Datta, S. Xu, B.J. Czerniecki

Development of methodology: J. Datta, S. Xu, C. Rosemblit, J.B. Smith, D.J. Powell Jr, B.J. Czerniecki

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): J. Datta, S. Xu, C. Rosemblit, J.B. Smith, J.A. Cintolo, D.J. Powell Jr, B.J. Czerniecki

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): J. Datta, S. Xu, C. Rosemblit, J.A. Cintolo, B.J. Czerniecki

Writing, review, and/or revision of the manuscript: J. Datta, S. Xu, C. Rosemblit, J.A. Cintolo, D.J. Powell Jr, B.J. Czerniecki

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): J. Datta, S. Xu, J.B. Smith, B.J. Czerniecki

Study supervision: D.J. Powell Jr, B.J. Czerniecki

This study was supported by NIH grant R01 CA096997 and Pennies in Action Foundation.