Augmented HER-2–Specific Immunity during Treatment with Trastuzumab and Chemotherapy

Purpose: Passive immunotherapy with antitumor antibodies has the potential to induce active tumor immunity via the opsonic enhancement of immunogenicity of tumor antigen. We have assessed whether immune sensitization to the HER-2/neu tumor antigen occurs during treatment with the anti-HER-2/neu monoclonal antibody trastuzumab.

Experimental Design: Twenty-seven patients treated with trastuzumab and chemotherapy were assessed for the induction of HER-2/neu–specific immunity. Sera and peripheral blood mononuclear cells obtained before and after trastuzumab therapy were compared for the presence of anti-HER-2/neu endogenous Igλ antibodies and HER-2/neu–specific CD4 responses by ELISA and enzyme-linked immunospot, respectively.

Results: Anti-HER-2/neu antibodies were detectable in 8 of 27 (29%) patients before trastuzumab treatment and in 15 of 27 (56%) patients during trastuzumab treatment. In the overall study population, anti-HER-2/neu humoral responses significantly increased during therapy (P < 0.001) and were not associated with development of an anti-idiotypic response. In 10 evaluable individuals, 6 showed augmented HER-2/neu–specific CD4 T-cell responses during therapy. Of the 22 individuals treated for metastatic disease, those patients showing objective clinical responses exhibited more frequent (P = 0.004) and larger (P = 0.006) treatment-associated anti-HER-2/neu humoral responses.

Conclusion: Humoral immune sensitization occurs during treatment with chemotherapy and trastuzumab. Further studies are warranted to investigate whether augmented anti-HER-2/neu humoral and cellular immunity contributes mechanistically to clinical outcome.

Antitumor monoclonal antibodies (mAb) represent a major advance in the therapy of cancer. In the past decade, six “naked” antitumor mAbs have been approved and many others are in preclinical and clinical development. Although the clinical experience is still immature, the potency of this class of therapeutics as single agents in front-line therapy ranges from 15% to 50% (1–5). Combination therapies of antitumor antibodies and active cytotoxic agents are more effective and have increased median survival in CD20⁺ lymphoma (6, 7). In the treatment of HER-2⁺ breast cancer, the HER-2/neu mAb trastuzumab in combination with paclitaxel increases survival in the metastatic setting (8) and its addition to chemotherapy enhances disease-free survival in the adjuvant setting (9). Although multiple underlying synergistic mechanisms may contribute, the ability of epidermal growth factor receptor family member–targeted antibodies (10–12) and small-molecule protein tyrosine kinase inhibitors (13, 14) to render tumor cells more susceptible to chemotherapy-induced cellular apoptosis has suggested that chemosensitization is central to clinical response (15). However, antitumor antibodies, as opposed to protein tyrosine kinase inhibitors, contain Fc domains that may also mechanistically contribute via the induction of an innate and/or adaptive antitumor immune response.

Recent genetic studies have provided strong evidence for the importance of the Fc domain in the efficacy of antitumor antibodies; in murine systems, Fcγ receptor (FcγR) engagement was required for efficacy of antitumor antibodies in several tumor antigen models, including HER-2 (16). Furthermore, four clinical studies have shown a positive correlation between the presence of favorable FcγR polymorphic alleles with higher affinities for IgG and improved clinical outcomes in rituximab-treated patients (17–20). These studies have established that Fc-FcγR interactions are critical to antitumor antibody efficacy in the mouse and are correlative with clinical outcome in patients. Indeed, natural killer cells are recruited to tumor sites in patients during therapy with trastuzumab and chemotherapy but not with chemotherapy alone (21), providing supportive evidence for the potential involvement of antibody-dependent tumor cell cytotoxicity by FcγR-bearing effectors in situ.

In addition in their roles as opsonins, antitumor antibodies are predicted to enhance dendritic cell internalization and antigen presentation of tumor antigen via endocytosis and phagocytosis of tumor antigen–containing immune complexes and antibody-opsonized tumor target cells, respectively (22, 23). Although murine studies are supportive of the concept of immune-complex–mediated induction of tumor immunity (22, 24), evidence for the enhancement of immunity in antitumor antibody-treated patients is lacking. We hypothesized that tumor responses in patients treated with combination chemotherapy and trastuzumab would be accompanied by alterations in antitumor immunity and therefore have investigated the HER-2/neu immunologic response in these patients.

Study design. The study was reviewed and received prior approval by the institutional review boards of Columbia University Medical Center, University of California at Los Angeles School of Medicine, and the Mayo Clinic. Informed consent was obtained from nonpregnant adults with HER-2–positive solid tumors who were to receive trastuzumab with chemotherapy as clinically indicated. Thirty-five patients were recruited, of which 34 patients had adenocarcinoma of the breast and 1 patient had metastatic bladder cancer (Table 1). Eight patients did not complete study due to either death (three patients) or self-removal (five patients). Patients receiving paclitaxel were also administered dexamethasone as a premedication. Patients with signs of significant myelosuppression (i.e., absolute neutrophil count <1,000, absolute lymphocyte count <400, hemoglobin <8.0, platelets <90,000) were excluded. HER-2/neu overexpression was documented in 19 patients who were HercepTest 3⁺ (Dako Corp.) or HercepTest 2⁺ and confirmed by fluorescence in situ hybridization gene amplification (8 patients). Blood samples were collected before treatment and after ≥8 weeks of weekly treatment with trastuzumab. Sera were prepared from clotted tubes and stored at −80°C. Peripheral blood mononuclear cells (PBMC) were obtained from EDTA blood tubes using Hypaque-Ficoll centrifugation. Buffy coat samples were washed in PBS and counted before storage in liquid nitrogen in human AB serum/10% DMSO at 1 × 10⁶/mL.

Clinical assessment. Patients with metastatic disease were restaged after 8 weeks of therapy, or sooner if clinical progression was suspected, as part of their routine clinical care. All sites of known metastatic disease were reassessed by physical exam, bone scan, computed tomography, positron emission tomography, or magnetic resonance imaging as clinically indicated. Objective responses, partial responses, mixed responses, stable disease, and progressive disease were defined according to WHO criteria (25, 26). None of the patients treated in the adjuvant setting showed evidence of clinical recurrence during the observation period.

HER-2 ELISAs. ELISA plates were coated with 5 μg/mL HER-2 extracellular domain protein in PBS. HER-2/neu protein was purified by trastuzumab affinity chromatography from culture supernatants of baby hamster kidney cell–produced extracellular domain (27). Control plates were coated with tetanus toxoid (10 plaque-forming units/mL; Aventis Pharmaceuticals) diluted 1:100 in PBS. Plates were blocked with either 0.5% pig gelatin (Sigma) or 1% bovine serum albumin (Sigma) in PBS for HER-2/neu–coated and tetanus-coated plates, respectively, before diluted serum samples were added at room temperature for 2 h. Plates were subsequently washed four times in PBS/0.05% Tween 20. Igλ antibodies were detected with 1 μg/mL biotinylated anti-human Igλ (BD PharMingen) for determination of patient-derived anti-HER-2 responses. Serum trastuzumab concentrations were determined with 1 μg/mL anti-human Igκ antibody (BD PharMingen). ELISAs were developed with streptavidin-horseradish peroxidase (diluted 1:10,000; Southern Biotech), and A₄₅₀ values were compared with a trastuzumab standard curve.

All samples were assayed in triplicate and the data are presented as the relative A₄₅₀ calculated as follows: [serum sample mean A₄₅₀ anti-HER-2/neu (sample on HER-2/neu or tetanus antigen-coated wells) − mean A₄₅₀ background (sample on uncoated/blocked wells)] / [mean A₄₅₀ (anti-HER-2/neu–positive standard sera on HER-2/neu or tetanus-coated wells) − mean A₄₅₀ (positive standard on uncoated blocked wells)]. Normalization in this manner to the internal standard reference of pooled patient sera from trastuzumab-treated patients was done on every ELISA plate and served to eliminate plate-to-plate variation. All serologic assays were repeated at least twice for each individual patient. A humoral response was considered positive by a relative A₄₅₀ index of >0.2 or a titer <1/100.

Igκ depletion. To deplete serum of Igκ antibodies, 850 μL of 1:25 diluted sera were applied in PBS to 1 mL of anti-Igκ beads [0.5 mg anti-Igκ mAb coupled to N-hydroxysuccinimide–activated agarose beads (Amersham)]. After 30 min, the unbound fraction was applied two subsequent times to eluted, regenerated anti-Igκ agarose beads. Each round of depletion resulted in >90% quantitative reduction in Igκ levels as determined by ELISA. κ-Depleted serum and nondepleted serum dilutions were equalized for total protein concentration based on results of Bradford assays. Antibody titers after κ depletion were assayed in the HER-2/neu and tetanus ELISAs as above.

Anti-idiotypic trastuzumab capture ELISAs. ELISA plates coated with 1.5 μg/mL trastuzumab in PBS were blocked with 0.5% pig gelatin in PBS. Diluted sera or goat anti-human F(ab′)₂ IgG (standard) was incubated with trastuzumab-coated plates for 2 h at room temperature followed by washing in PBS/0.05% Tween 20. Trastuzumab-bound IgG was detected with addition of 250 ng/mL of biotinylated trastuzumab, washed, and then developed with streptavidin-horseradish peroxidase.

FcγR polymorph determination. FcγRIIA and FcγRIIIA polymorphic allele status was determined by genomic PCR approaches as described by Wu et al. (28), with two minor modifications; the FcγRIIIA PCRs were done at annealing temperatures of 58°C and 60°C for the T and G reaction, respectively.

HER-2/neu–specific CD4 IFN-γ enzyme-linked immunospots. Twelve HER-2/neu peptides, each known to bind to multiple HLA-DR molecules (29), were used to detect T-cell responses by the enzyme-linked immunospot (ELIspot) method (30, 31). Four of the HER-2/neu helper peptides, p98, p369, p927, and p776, have been previously described in detail (32, 33). The remaining seven peptides (all 15-mers), designated by the position of the first amino acid, p62, p77, p83, p88, p350, p783, and p976, are recently identified epitopes that exhibit high-affinity binding to a variety of HLA class II molecules.⁷

Both phorbol 12-myristate 13-acetate/ionomycin and pooled cytomegalovirus, EBV, and Flu viral peptides (CEF) were used as positive controls. In brief, cryopreserved PBMCs were cultured at 2 × 10⁵ per well in 96-well plates for 7 days in medium containing individual HER-2/neu class II peptides (each at 10 μg/mL) or in the absence of any antigens (no-antigen control wells). In some cases where patient material was lacking, the number of peptides was reduced to nine peptides to accommodate. In these cases, all of the time points were assessed with the same panels of peptides. Interleukin-2 (10 units/mL) was added at day 5, and on day 7, peptide and 2 × 10⁵ per well irradiated autologous PBMCs were added as antigen-presenting cells. On day 8, the cells were gently transferred to the ELIspot plate for detection of spots (Mabtech AB). ELIspots were developed, dried, and read with an AID Immunospot ELIspot reader as previously described (30). Peptide-specific immune reactivity was determined by subtracting the background spots in the no-antigen–containing wells. A positive response was defined as the peptide-specific spots that were statistically higher (triplicates) than control wells using a two-tailed t test (P < 0.05). A zero response was assigned if the peptide-specific wells were not different than control (i.e., no peptide) wells. The counts for each peptide were summed and presented as the total HER-2/neu–specific T cells assessed at each time point. Note that, although the peptides are known to bind to multiple HLA-DR alleles, it is difficult to rule out that they could contain embedded peptides that could stimulate CD8 T cells; however, as previously reported, HER-2/neu–specific CD8 T-cell responses are typically lower by at least one order of magnitude even in vaccinated patients (30). Changes between preimmune and postimmune responses were considered positive if there was at least a doubling for increases and a halving for decreases.

Statistical analysis. The antibody response of a single patient (ON:33) deviated significantly from the norm and was identified, by box plot of change in λ anti-HER-2 (post-pre) response, as an outlier. This patient exhibited the highest preexisting anti-HER-2/neu binding activity in the cohort, which remained positive during therapy, but, in contrast to all other 26 individuals studied, decreased in absolute magnitude. Accordingly, P values are provided for statistical analysis that either includes these outlier data or, as indicated (§), excludes this outlier.

Anti-HER-2/neu humoral immunity is induced during trastuzumab therapy. To assess whether immunologic responses to HER-2/neu were altered during therapy with chemotherapy and trastuzumab, patients were enrolled in an observational study before initiation of therapy. Twenty-seven patients completed the study and provided pretreatment and 8-week posttreatment sera and PBMCS for collection and storage. Of these 27 subjects, 26 were breast cancer patients and 1 patient was treated for HER-2/neu–positive metastatic bladder cancer (Table 1). Of the 26 breast cancer patients, 21 (81%) were treated in the stage IV setting with weekly trastuzumab and either vinorelbine (13 patients), paclitaxel (8 patients) or no chemotherapy (1 patient) and 5 (19%) individuals were treated in the high-risk adjuvant setting on protocol NCCTG N-9831 (9) and received either paclitaxel followed by trastuzumab or concurrent paclitaxel and trastuzumab.

Sera were analyzed by ELISA for evidence of a humoral response to HER-2/neu. For these ELISA-based assays, plates were coated with tetanus toxoid as a control antigen or with human HER-2/neu extracellular domain protein. Although there was a range of anti-tetanus Ig activity between patients, activities of individual patients did not vary during therapy, indicating that overall specific Ig levels were not influenced by treatment. In contrast, anti-HER-2/neu Igλ responses were induced in several patients during therapy (Fig. 1; Table 1). The Ig response of the λ subclass was specifically addressed to prevent the spurious detection of trastuzumab, a human IgG1 κ mAb present at high concentrations in the sera of treated patients. As previously recognized, anti-HER-2/neu antibody responses were detected in 8 of 27 (29%) patients before the initiation of trastuzumab therapy, consistent with preexisting immune recognition of HER-2/neu (34). During therapy, anti-HER-2/neu Igλ responses were detectable in 15 of 27 (56%) patients. Anti-HER-2/neu responses increased across the entire population of 27 patients after 8 weeks of therapy (P = 0.002 and P < 0.001§), with increases evident in 12 of 27 (44%) treated individuals (Table 2).

Preexistent humoral immunity was more frequently observed (P < 0.05, χ² test) in HercepTest 3⁺ patients (8 of 19 patients) than in HercepTest 2⁺ patients (0 of 8 patients), but there was no evidence for an association of preexistent humoral immunity and clinical outcome (Table 2). Most patients lacked detectable endogenous HER-2/neu Igλ antibodies before initiation of trastuzumab, and in this group, 9 of 19 developed detectable endogenous anti-HER-2/neu Igλ antibodies. Of the eight patients with circulating anti-HER-2/neu antibodies before receiving trastuzumab, the binding activities increased in three and were unchanged or decreased in the other six individuals. Overall, preexistent anti-HER-2 levels were not related to posttreatment anti-HER-2 antibody levels (Pearson r = 0.44).

To examine whether humoral responses were durable, anti-HER-2 levels were examined at later time points in 10 of the patients that exhibited increases in anti-HER-2 Igλ responses during initial therapy. Mean anti-HER-2 levels were not significantly different across the population at early and late time points (Fig. 1B). Individually, eight of the 10 patients showed persistent humoral immunity after more than 20 weeks of ongoing trastuzumab therapy, indicating that humoral immunity was sustained in most patients (data not shown).

Induced humoral immunity correlates with favorable clinical response. Although limited by the statistical power of this small patient population size, we have addressed whether induction of endogenous humoral responses during treatment occurred more frequently or was of greater magnitude in patients who responded clinically (Fig. 2; Tables 2 and 3). When restaged after 2 months of treatment, 8 of the 22 (36%) patients with metastatic disease exhibited an objective clinical response, 6 (27%) had stable disease or a mixed response, and 8 (36%) exhibited progressive disease. Based on a one-way ANOVA model, there was a significantly greater increase of Igλ anti-HER-2/neu in the objective response group than in the combined mixed response, partial response, and progressive disease group (P = 0.004 and 0.006§), and conversely, there was a marginally significantly smaller increase of Igλ anti-HER-2/neu in the progressive disease group than in the combined mixed response and objective response group (P = 0.056). Similarly, the frequency of occurrence of antibody responses in individuals was related statistically to clinical outcome. Among the clinical groups, treatment-associated increases in anti-HER-2/neu Igλ immunity occurred frequently in patients exhibiting objective responses (6 of 8 objective response patients, 75%) and significantly more often than in patients that did not show clinical objective responses (progressive disease, mixed response, and partial response groups) in whom rises in anti-HER-2/neu Ig responses occurred in just 2 of 14 (14%) patients [P = 0.004, χ² test (P = 0.002§)]. Thus, induction of an endogenous anti-HER-2/neu humoral response during therapy with trastuzumab and chemotherapy is associated with clinical response. Administration of chemotherapy was not required for the induction of humoral immunity because enhanced anti-HER-2 Igλ levels were seen in two of the three patients that received trastuzumab without concomitant chemotherapy (patients 1, 22, and 34).

It has been noted that clinical responses occur more frequently in rituximab patients harboring either the FcγRIII V158 or the FcγRIIA R131 alleles (17, 18). Correlative studies in trastuzumab-treated patients have yet to be reported. FcγRIIA and FcγRIIIA polymorphic allelic status in this group of 27 trastuzumab-treated patients is reported in Table 1. Conclusions are limited, however, by the small sample size of the studied population about the association of the FcγRIIIA V158 or FcγRIIA R131 alleles with either antibody or clinical response.

Endogenous anti-HER-2/neu Igλ responses do not correlate with serum trastuzumab levels or the appearance of idiotypic antibodies. The presence of trastuzumab in sera obscured the specific detection of endogenous anti-HER-2/neu IgG/κ with available human IgG secondary reagents. Thus, to confirm the specificity of the anti-λ secondaries and to rule out the potential for this artifactual contribution of trastuzumab, serologic levels of trastuzumab were determined in all treated patient samples. Had trastuzumab contributed to the detection of Igλ antibodies, one would have expected a correlation between serum trastuzumab levels and anti-HER-2/neu Igλ levels. Importantly, serum trastuzumab levels and anti-HER-2/neu Igλ titers did not correlate significantly (Pearson r = 0.3806; Fig. 3A), discounting this potential source of artifact.

Further studies were done to eliminate the possibility that the ELISA detection of HER-2/neu–specific λ antibodies was due to indirect binding to HER-2/neu through trastuzumab, as might be expected, for instance, by the potential presence of idiotypic anti-trastuzumab antibodies. Therefore, the presence of high-titered anti-idiotypics was ruled out in all patients: no anti-idiotypic antibodies were detected in any patient using trastuzumab-coated plates in an assay whose sensitivity for detection was 20 ng/mL (Fig. 3B). In a second direct experimental approach, trastuzumab antibodies were removed from patient sera by three successive rounds of Igκ depletion, by anti-κ affinity chromatography. κ-Depleted sera were then reassayed for Igλ anti-HER-2/neu reactivity, which remained unchanged despite a >99% reduction in Igκ anti-HER-2/neu reactivity (data not shown; Fig. 3C). Thus, detection of anti-HER-2/neu antibodies is not confounded by the presence of trastuzumab antibodies in the sera.

Treatment with trastuzumab and chemotherapy augments HER-2/neu–specific CD4 T-cell immunity. PBMC samples were available for testing for pretreatment and at least one posttreatment CD4 T-cell responses from nine individuals. Eight early (i.e., <15 weeks) posttreatment samples showed that 50% of patients developed elevated HER-2–specific CD4 T cells early in the course of treatment (Fig. 4A). Nine late (i.e., ≥15 weeks) posttreatment samples showed that 78% of patients had elevated CD4 T cells (Fig. 4B). Figure 3C shows persistent immunity in four patients, all of whom showed clinical benefit. Overall, there was remarkable concordance between the enhancement of HER-2–specific antibody responses and the development of augmented CD4 T-cell immunity, consistent with the development of a T-dependent humoral response (Table 1). In the two patients evaluated in the adjuvant setting (patients 19 and 22), CD4 T-cell responses were of the greatest magnitude and were sustained for at least 4 months (Fig. 4C). HER-2–specific CD4 T-cell responses were detectable in all four evaluated patients who showed objective clinical response (patients 3, 11, 14, and 33). In contrast, of the four patients in whom objective responses were not evident (patients 4, 21, 24, and 30), only one showed a treatment-associated increase in HER-2/neu–specific CD4 frequency. Responses against the individual peptides are shown in Table 4. Three of the patients exhibited broad reactivity to several HER-2 epitopes.

Responses to phorbol 12-myristate 13-acetate/ionomycin and the CEF peptide pool were compared between pretreatment and posttreatment samples to determine if there were differences in responses to noncancer stimuli. As shown in Fig. 4D, the responses to the control stimuli were not different between the two populations. The mean number of spots per million PBMC for the pretreatment samples in the phorbol 12-myristate 13-acetate/ionomycin–containing wells was 2,177 ± 634 (mean ± SE; n = 11), which was not statistically different than the response in the posttreatment samples (3,056 ± 323; n = 21; P = 0.18). The mean number of spots per million PBMC for the pretreatment donors in the CEF peptide–containing (Fig. 4E) wells was 424 ± 223, which also was not statistically different than the response in the posttreatment samples (95 ± 73; P = 0.1).

We provide evidence for the induction of a humoral and cellular HER-2/neu–specific immune response during the treatment of HER-2/neu–overexpressing malignancies with trastuzumab and chemotherapy in the high-risk adjuvant breast cancer and stage IV setting. Enhanced endogenous humoral responses were seen in 44% of treated patients and, interestingly, were of greater magnitude and more frequently observed in clinically responding patients. This association of endogenous anti-HER-2/neu humoral immunity with favorable clinical outcomes may simply be a marker of clinical response, possibly the consequence of immunomodulatory effects of enhanced tumor cell apoptosis/necrosis occurring in clinically responding patients receiving trastuzumab and chemotherapy. Alternatively, the association could be causal and indicate that augmented tumor immunity contributes to the efficacy of trastuzumab and chemotherapy.

There is substantial preclinical experimental support for the idea that antitumor antibodies in their roles as opsonins can promote the immunogenicity of both human and murine tumor antigens. Immunization of mice with dendritic cells pulsed with antibody-opsonized tumor antigens acquired via either FcγR-mediated endocytosis (22, 24) or phagocytosis (35) induces CD4- and CD8-mediated tumor immunity. Dhodapkar et al. (23) showed that cross-presentation mediated by FcγRs on human dendritic cells can enhance the presentation of multiple myeloma antigens to patient-derived T cells, thus suggesting that uptake of antibody-opsonized tumor cells and cellular fragments by antigen-presenting cells could lead to antigenic/epitope spreading and the induction of immunity to several tumor-associated antigens. Herein, we provide data supportive of this concept in patients by showing the first direct evidence for an induced immunologic response in antibody-treated cancer patients and further studies will be important to assess whether concomitant immunity to other tumor-associated antigens is also induced in trastuzumab-treated patients as predicted by an antigenic cascade. Anti-HER-2/neu antibodies enhance the potency of HER-2/neu–expressing whole-cell vaccines in mice, suggesting that mAbs may enhance priming of effective tumor immunity (36). In trastuzumab-treated patients, the opsonic enhancement of HER-2/neu immunogenicity may be the consequence of trastuzumab bound to either shed soluble HER-2/neu extracellular domain protein or to HER-2/neu bearing necrotic/apoptotic tumor or normal cells.

We cannot formally address whether the coadministered chemotherapy contributes to the occurrence of immunologic responses. However, chemotherapy is not an absolute requirement because three of the patients showing immunologic responses (patients 1, 22, and 34) received trastuzumab alone. Patient 1 received trastuzumab in the metastatic setting, whereas patients 22 and 34 received trastuzumab alone after completing adjuvant chemotherapy. Chemotherapy and/or dexamethasone (administered in patients as an antiemetic and to prevent paclitaxel and cremophor hypersensitivity responses) have been traditionally perceived as detrimental to the induction of immunity via myelosuppression and inhibition of lymphocyte and antigen-presenting cell function. However, a recent vaccine trial found no evidence for chemotherapy-associated impairment of T-cell responses in 28 patients randomized to receive either vaccine alone or vaccine concurrently with docetaxel/dexamethasone (37). Preclinical studies have shown that administration of chemotherapy enhances the immunostimulatory capacity of coadministered vaccines (38). Indeed, there is accumulating evidence that supports the notion that chemotherapy could enhance tumor antigen priming through multiple mechanisms, including (a) promotion of tumor cellular apoptosis/necrosis, thus increasing the antigenic load available for uptake by antigen-presenting cells, and (b) inhibition of regulatory T-cell function (39).

The assessment of the anti-HER-2/neu humoral response has been limited to the anti-HER-2/neu response of the Igλ subclass, as detection of IgG responses was complicated by the high serum concentrations of trastuzumab, an IgG1 κ antibody. Most patients who exhibited Igλ anti-HER-2/neu increases also showed an increase in anti-HER-2/neu IgM levels (data not shown). Determination of levels of class-switched antibodies of the IgG subclass has not been possible to date because a screen of several secondary reagents recognizing human IgG2, IgG3, and IgG4 subclasses has failed to identify a reagent specific enough in our assays to avoid detection of trastuzumab (present in the sera at 20-600 μg/mL). The concordance in 10 individuals of the presence of augmented HER-2/neu–specific CD4 cell responses with the presence of increased humoral responses, however, makes it likely that the humoral responses observed are the product of a T-dependent response. Further analysis of additional patients will be required to determine whether augmented CD4 immunity occurs significantly more frequently in clinically responding patients. HER-2/neu–specific IFN-γ–producing CD4 cells would be expected to contribute as effectors of antitumor inflammatory responses or through the provision of T-cell help for CD8 T-cell responses. Preliminary ELIspot assays of six HLA A0201⁺ patients in this cohort have not yet revealed induction of a CD8 anti-HER-2/neu response using two previously defined HER-2/neu immunodominant epitopes (369-377:KIFGSLAFL and RLLQETELV 689-697; refs. 40, 41).

What role could endogenous HER-2/neu antibodies play in treatment responses? The high levels of trastuzumab already present in treated patients might suggest that, for both antibody-dependent tumor cell cytotoxicity and inhibition of HER-2 signaling, growth-regulatory consequences are already saturated and optimized. However, endogenous antibodies might bind epitopes distinct from the trastuzumab-binding site on the juxtamembrane region of HER-2/neu. By binding distinct epitopes, endogenous antibodies could provide additive roles promoting trastuzumab-mediated antibody-dependent tumor cell cytotoxicity or furthering HER-2/neu signaling perturbation.

Recent clinical data have suggested that favorable clinical outcomes in patients treated with the anti-CD20 mAb rituximab occur more frequently in patients harboring allotypic FcγR alleles conveying higher affinity for IgG (17–19). The data presented here of 22 patients treated in the metastatic setting lack the necessary statistical power to appropriately address this question in trastuzumab-treated patients. With regard to immune sensitization, activating FcγR haplotypes did not strongly segregate with the occurrence of endogenous humoral response, although again the strength of this conclusion is limited by sample size. Uptake of HER-2/neu immune complexes through activating FcγRs would be expected to enhance the HER-2/neu T_helper CD4 cell response (42), thus predicting that activating FcγR subtype would be contributory. However, immune complex uptake by other receptors, including the inhibitory FcγRIIB receptor and/or complement receptors, on antigen-presenting cells and B cells, respectively, could also positively regulate immune complex enhancement of activation of anti-HER-2/neu–specific B cells (43, 44).

The data provided here are the first to show immune sensitization during treatment with antitumor antibodies. The induction of CD4 and endogenous humoral immunity suggests that therapeutic antibodies not only provide passive immunotherapy through antibody-dependent tumor cell cytotoxicity but also can promote active immunity. Although these data do not prove causality, they nevertheless suggest that strategies aimed at promoting the vaccinal effect of opsonic antitumor antibodies would augment immunologic memory and thereby enhance durable clinical benefit.

We thank Chao Yu and Courtney Erskine for technical assistance.