Developing a Rational Tumor Vaccine Therapy for Renal Cell Carcinoma: Immune Yin and Yang

Until recently, the most successful therapy for metastatic renal cell carcinoma (RCC) has been that of single-agent, high-dose interleukin 2 (IL-2; ref. 1). The addition of other proinflammatory agents, such as IFN-α, or adoptive cellular therapy with ex vivo expanded autologous effector cells, such as lymphokine-activated killer cells or tumor-infiltrating lymphocytes, to IL-2 has failed to significantly improve clinical outcome compared with IL-2 alone (1–5). Vaccine strategies have also had limited benefit for patients with metastatic RCC (6). Recently introduced “directed therapies,” such as bevacizumab, sorafenib, sunitinib, and temsirolimus, have had an effect on the survival of metastatic RCC patients but rarely induce durable complete remissions (7–10).

Immunotherapy with high-dose IL-2, which mobilizes immune effector cells that recognize and destroy cancer, can induce complete durable remissions but only in few metastatic RCC patients. The reason why IL-2 therapy fails in most metastatic RCC patients may be explained, in part, by a persistent imbalance between proinflammatory/stimulatory (Yang) and regulatory (Yin) pathways (11–13).

The CD8⁺ CTLs mediate tumor destruction and represent a common pathway for many immunotherapeutic approaches. CD8⁺ tumor-specific T cells naturally exist in the lymphocyte memory compartment and can be identified in the peripheral blood of cancer patients (14, 15). Memory CD8⁺ T lymphocytes consist of a heterogeneous mixture of two lineage groups of cells that evolve in parallel: central memory CD8⁺ T cells, which recirculate through lymph nodes, and effector memory CD8⁺ T cells, which are mostly found in peripheral tissues (16). Central memory and effector memory T-cell responses can be enhanced by dendritic cell (DC)–presenting tumor antigens (17). Thus, DC vaccination approaches may help to optimize antitumor CTL generation.

Tolerogenic pathways that involve regulatory cells are a mechanism of tumor-specific anergy (18–20). CD4⁺CD25^+high regulatory T (T_reg) cells function in a non–antigen-specific manner through T cell to T cell contact or other contact-dependent mechanisms or through expression of immunosuppressive cytokines. CD8⁺CD28⁻ suppressor T (T_s) cells inhibit immune responses via cell-to-cell interactions or using antigen-presenting cells as a bridge to the CD4⁺ T helper cell in an antigen-specific, MHC-restricted manner (21–23). Under experimental conditions in murine models, tumor-DC vaccinations can overcome tolerance and enhance therapeutic outcome (24). IL-2 and IFN-α can also contribute in overcoming these regulatory pathways (25, 26). Other inhibitory loops include both natural killer T cells and a subpopulation of inhibitory DCs (27, 28). The contribution of each of these regulatory compartments to tumor tolerance in the cancer patient is currently unknown (29, 30).

The lack of ability of cancer patients' CTLs to mount a complete response against tumor (anergy) may also be due to the loss of the T-cell receptor ζ chain (TCRζ), defective signaling downstream from TCRζ, and/or activation of inhibitory pathways by regulatory cells (31, 32). Some of these defects can be corrected by subsequent cytokine signals, including IL-2 (33). Agents that exploit proinflammatory activities that both enhance CTL function and disrupt regulatory pathways may synergistically act in causing immune-mediated tumor destruction. We hypothesize that immunotherapy using mature DC vaccine administered intranodally, IL-2, and IFN-α designed to shift the balance between immune stimulatory and inhibitory (regulatory) pathways will enhance therapeutic benefit.

DCs are the most potent antigen-presenting cells and are able to activate proinflammatory tumor-specific immune pathways as well as disrupt regulatory pathways (13). Immature DCs sample and process antigens and then migrate to T cell–rich areas in lymphoid organs (13, 34). In lymphoid organs, mature DCs, which express costimulatory molecules, present antigen and stimulate antigen-specific T cells to proliferate. In the tumor-bearing host, in vivo DC maturation and function seem inhibited (35–37). Treatment of tumor-bearing mice with an ex vivo generated DC vaccine reduces T-cell tolerance and enhances antitumor immunity (24, 38). Although the optimal route of DC administration is not defined, DC cross-priming of T lymphocytes requires cell-to-cell contact, suggesting that direct administration of mature DCs to lymph nodes would be advantageous. This is supported by mouse studies that confirm enhanced protective systemic immunity by intranodally administered DC-tumor vaccines, as well as preliminary data in human studies using intranodal vaccinations, which suggest that this route is superior in eliciting CD8 T-cell responses (39, 40). The source and type of antigens that are best suited for exploiting DC biology to initiate immunity remain unclear; defined tumor peptides, autologous tumor lysate, apoptotic tumor cell bodies, DC-tumor cell fusion, and genetically engineered DCs to express antigen have all been used with some evidence of immune stimulation (41–44). The uses of therapeutic tumor vaccines for metastatic RCC in the past have shown a low level of clinical activity (6). The poor clinical performance, in humans, of therapeutic cancer vaccines is due to the lack of ability of the vaccines, when given alone, to effectively influence proinflammatory and regulatory pathways needed to augment and maintain CTL response. Alternatively, antigen-bearing DCs may not have been appropriately matured ex vivo with inflammatory signals to effectively stimulate protective antitumor immune responses. This suggests the need for postvaccination signaling as immune manipulation. Combining therapeutic vaccines with single-agent proinflammatory cytokines, such as IL-2, or with inhibiting regulatory pathways with anti–CTL antigen-4 (CTLA-4) antibody does not seem to improve clinical outcomes, further suggesting the need for multitargeted combinations (45). The approach we have used differs from those previously reported by adopting an intranodal DC vaccination, followed immediately by combining IL-2 and IFN-α cytokine therapies.

Over the past few years, disruption of the CD4⁺ T_reg function through the blocking of CTLA-4 has been a clinical focus of immunotherapy. It is not clear whether anti–CTLA-4 antibody blockade directly works on the T_reg cells or by blocking negative signals to effector T cells. CTLA-4 molecule is constitutively expressed on CD4⁺ T_reg cells and competes with CD28 (a T-cell stimulatory receptor) to bind to B7.1/B7.2 on antigen-presenting cells (45). When CTLA-4 is disabled, it is believed that CTLs are released from inhibition and the regulatory effects of T_reg cells are diminished. Disabling CTLA-4 can be achieved by a blocking antibody (anti–CTLA-4; ref. 46). Preliminary clinical trials of blocking anti–CTLA-4 antibodies in melanoma and RCC patients suggest that this approach has a small therapeutic effect (47, 48). In one study, the addition of IL-2 to anti–CTLA-4 antibody therapy in melanoma patients failed to show additional clinical activity over IL-2 or anti–CTLA-4 antibody therapy alone, suggesting an overlap in the mechanisms of action of these agents on CD4⁺ T_reg cells or the presence of other regulatory pathways (49).

To determine the presence of other regulatory cell populations in cancer patients, we studied peripheral blood mononuclear cells from healthy donors and advanced RCC patients for the presence of T_s cells. T-cell subpopulations were identified by multicolor flow cytometry using anti-CD8, anti-CD28, anti-CD4, and anti-CD25 antibodies (Coulter/Immunotech, Marseille, Cedex, France). Phenotypical differences in the proportion of T-cell subsets between healthy donors and RCC patients were analyzed using an unpaired t test. T_s cell function was measured by a [³H]thymidine CD4⁺ T-cell proliferation assay using irradiated KG-1 cells, which act as an antigen-presenting cell bridge (21).

Blood samples from 24 healthy donors and 29 RCC patients were analyzed. The percentage of CD8⁺CD28⁻ T_s cells was significantly higher in RCC patients (46.2%) than in healthy donors (39%; P = 0.044). The percentage of T_reg cells (CD4⁺CD25^+bright) was also significantly higher in RCC patients (38.3% versus 21.3%; P = 0.0006).

RCC patients' T_s cells consistently suppressed autologous CD4⁺ cell proliferation. The percentage of reduction in proliferation ranged from 31% to 47%. Irradiation of T_s cells before cell culture did not affect their ability to suppress CD4 proliferation. Low concentrations of IL-2 (50 IU/mL) were able to reverse the suppressive effects of T_s cells on CD4⁺ cell proliferation (Fig. 1).

Thus, we have shown a significantly higher percentage of functioning T_s cells in RCC patients. Their contribution to tumor tolerance is still unclear, but they may play a significant role along with the other regulatory cell populations. Recent data suggest that DCs made tolerant by CD8⁺ T_s cells have reduced expression of costimulatory molecules (21). Thus, one approach to overcome CD8⁺ T_s cell suppression would be to induce costimulatory molecules by maturing DCs ex vivo before clinical use.

By exploiting the proinflammatory properties (Yang) of the DC vaccine, IL-2, and IFN-α, and the ability of IL-2 and mature DCs to disrupt regulatory pathways (Yin), we hoped to improve clinical therapy for metastatic RCC patients. We have analyzed the first 13 metastatic RCC patients (11 men and 2 women) treated with DC vaccine, IL-2, and IFN-α. Patient demographics are summarized in Table 1.

Patients who met the eligibility criteria received two induction cycles of IL-2/IFN-α2a given on days 1 and 14 and three maintenance cycles at 28-day intervals (Fig. 2; ref. 50). An intranodal matured DC vaccine was given the day before each cycle. An 18 × 10⁶ IU/m² dose of IL-2 (Chiron, Inc., Emeryville, CA) was administered i.v. as 24-h continuous infusion for 5 days per cycle. IFN-α2a (Hoffmann-La Roche, Inc., Nutley, NJ) was given every other day, for three doses, by s.c. injection; at the start of each cycle, a dose of 6 MIU was used. Continuous infusion of IL-2 and/or IFN-α2a were interrupted for serious toxicity, and two dose reductions of IL-2 to 75% and 37.5% were allowed with subsequent cycles of therapy. Toxic effects that required further interruptions of therapy resulted in patients discontinuing IL-2 therapy and continuing IFN-α2a and vaccine therapies.

Patients were characterized for prognosis based on the criteria described by Motzer et al. (51). Clinical response, defined by using the Response Evaluation Criteria in Solid Tumors from the National Cancer Institute, was assessed by computed tomography of the chest, abdomen, and pelvis and bone scan before therapy, at the completion of the second induction cycle, between the second and third maintenance cycle, and at the completion of the study (52). Subsequent follow-ups for responding and stable patients consisted of the same imaging methods every 3 months thereafter or as clinically indicated. Survival was evaluated using the Kaplan-Meier method.

Tumor lysate–loaded, mature DCs were obtained from leukapheresis products under IND 11162. DCs were generated and matured over 9 days ex vivo in Lifecell tissue culture bags with serum-free AIM-V medium, 500 IU/mL of granulocyte-macrophage colony-stimulating factor (Berlex, Inc., Richmond, CA), 20 ng/mL of IL-4 (R&D Systems, Inc., Minneapolis, MN), autologous tumor lysate (1-3 tumor cell equivalents per DC added on day 5), and 50 ng/mL of tumor necrosis factor α (added on day 6; R&D Systems). These matured DCs typically express high levels of costimulatory molecules (CD80, CD86, MHC classes I and II) and up-regulate CD83 expression. DCs were harvested on day 9, and a total of 1 × 10⁷ DCs in 1 mL of Ringer's lactate were injected into two inguinal lymph nodes under ultrasound guidance. A portion of these DCs was frozen in 90% autologous serum and 10% DMSO and thawed for vaccine 2. Subsequent DC vaccines (vaccines 3, 4, and 5) were generated from monocyte precursors obtained from freshly prepared pheresis products. Requirements for release of the final DC preparation include >70% viability, negative sterility Gram stain, and an acceptable low endotoxin level.

To assess the effect of the therapy on immunologic function, we assessed tumor-specific CD4 and CD8 precursor levels using a dye dilution proliferation assay (46, 47). TCR function was assessed via a redirected cytotoxicity assay (53, 54). The dye dilution proliferation assay has a sensitivity of 10⁻⁵ and differentiates the renal cell–specific CD4⁺ and CD8⁺ T-cell populations.

Determination of the clinical response rate was the primary objective of the study. Six (46%) of the 19 patients achieved an objective partial or complete clinical response to treatment (Table 2). The two complete responders remain free of disease at >17 and 11 months of follow-up, respectively. Six patients had favorable Motzer criteria, and seven patients were in the intermediate risk group. Clinical responses were seen in both visceral (lung, liver, and adrenal) and lymph node–based disease sites. Responses were even observed in sites of bone involvement. Ten of thirteen patients remain alive from 19+ to 33+ months.

Table 3 lists the grade 3 or 4 adverse events observed during the treatment with IL-2, IFN-α2a, and DC vaccine, all of which were expected. We were particularly interested in the development of high-grade, autoimmune-like toxicity, which has been suggested to correlate with clinical responses. No grade 3 or 4 adverse events directly related to vaccine occurred. No treatment-related deaths occurred. Eighty-five percent of patients developed significant dermatitis, which is probably an autoimmune-like reaction. One patient who achieved a complete response developed glomerulonephritis, cardiomyopathy, and lung radiographic changes suggestive of an interstitial pneumonitis after three vaccine doses and four cycles (two induction and two maintenance) of IL-2 and IFN-α2a. An endomyocardial biopsy specimen was nondiagnostic, and renal biopsy was deferred. All of these manifestations normalized at ∼3 to 4 months after the completion of the therapy.

Tumor-specific T-cell precursor frequency (Fig. 3) includes a summary of CD4⁺ T-cell precursor frequencies as measured based on IFN-γ production and proliferation in response to autologous, tumor lysate–loaded DCs. Included are the four patients who received at least three vaccine doses and had blood samples obtained at the designated time points for analysis for which data are currently available. All four patients showed a relatively small increase in tumor-specific CD4⁺ T-cell precursor frequencies at a minimum of one time point after induction or maintenance phases of treatment compared with pretreatment baseline values. Three of these patients experienced a clinical partial response. No measurable increases in CD8⁺/IFN-γ⁺ precursors have been detected to date.

The effect of treatment on the CD8⁺ TCRζ pathway was determined via lytic activity in a redirected cytotoxicity assay, which uses an anti-CD3 antibody that engages the TCRζ (CD3) on the effector cell and the Fc receptor on the target P815 cell (Fig. 4; ref. 54). This assay allows a functional assessment of the TCRζ signal transduction pathway. A measurable increase in the CD8⁺ T-cell lytic activity after treatment was observed in three of six patients, suggesting enhanced signaling through TCRζ and reversal of signaling pathway defects. The other three patients had relatively stable posttreatment activity compared with pretreatment levels. Those patients with a detectable increase in lytic activity during or after their course of therapy included two patients with a partial response and one patient with a stable disease. The one complete responder of this group of six patients had a stable lytic activity.

Our broader understanding of the complexity of immune inflammatory (Yang) and regulatory (Yin) pathways in cancer patients now offers new paradigms that will affect immunotherapy strategies for patients. Our observation of functional CD8⁺ T_s cells in RCC patients and the limited clinical benefit of disruption of the CD4 T_reg cell–CTLA-4 pathway suggests that multiple regulatory pathways are likely to be operative in cancer patients. Induction of proinflammatory pathways or down-regulation of immune tolerance alone is frequently insufficient to produce antitumor responses in most cancer patients. We hypothesize that the generation of the appropriate balance of immunologic Yin and Yang may well be the key to enhancing therapeutic benefit. DCs, IL-2, and IFN-α play roles in both enhancing inflammatory responses and down-regulating or limiting peripheral tolerance mediated by T_reg and T_s pathways (22, 56). Preliminary data from our trial of autologous mature DC-tumor vaccination followed by IL-2 and IFN-α2a therapy provides support for combination immunotherapy, addressing both proinflammatory and regulatory pathways.

In our study, the administration of autologous, tumor lysate–loaded, mature DC vaccine, coupled with continuous infusion of IL-2 and IFN-α2a, resulted in a clinical objective response in 6 of 13 patients with metastatic RCC, twice the response rates historically seen with high-dose IL-2 or with continuous infusion of IL-2 with IFN-α (1, 2, 50) and better than the response rates reported to RCC tumor vaccines, including DC vaccines. We would attribute the improvement of the objective response rate observed in our study, if confirmed, to the addition of the DC vaccine component in our regimen. Optimization of this therapy using the Yin-Yang hypothesis would ultimately need to be tested in larger randomized studies.

The limitations of DC vaccination strategies for human cancers include DC biological complexity, the immunologic complexity of the host, and the production of homogeneous GMP grade product. DC function can be classified into initiating a T helper 1 cellular immune response, initiating a T helper 2 humoral immune response, or inhibiting immune responses via the induction of the T_reg function (57). Each of these functions has been associated with different DC phenotypes, although it seems that, in humans, these functional phenotypes are somewhat plastic. Thus, the best DC precursor to use for initiating in vitro cultures and the conditions that would provide the most effective T helper 1 skewing are unclear. The complexity of the host immune environment may differ between patients with the same cancer and between patients with different cancers, which may limit the general value of DC vaccination. Developing production methods that can provide high-quality, GMP-grade DC products which are homogeneous in their maturational state for a large number of patients is also a challenge. This may ultimately limit the application of DCs to a larger population of patients.

The IL-2 dose of the Negrier regimen is still quite substantial, and we observed toxicity with our regimen as expected of a high-dose IL-2–containing regimen (50). Serious adverse events were transient and were managed with standard supportive measures consistent with published guidelines and in a center experienced in administering IL-2 (58). Of note, we observed higher rates of grade 3 or 4 hypotension than that reported by Negrier et al. (50). Interestingly, we also observed a significantly higher rate of grade 3 or 4 skin manifestations, including skin peeling, erythema, and pruritus, which have been considered to be autoimmune-like effects and correlated to better clinical outcomes in other immunotherapy studies (47, 48). This, too, would support our hypothesis that combining DC vaccine with induction of inflammatory pathways and inhibition of regulatory pathways is crucial to improving clinical outcomes. We suspected a unifying autoimmune process in the one patient who developed a cardiomyopathy, glomerulonephritis, and interstitial infiltrates on his lung radiograph after three complete cycles of therapy. This patient went on to achieve a complete response, along with spontaneous resolution of the apparent treatment-related toxic effects.

We previously reported that CD8⁺ CTL activity in advanced RCC patients is impaired but can be restored with IL-2 in vitro (33). In our current study, we observed a posttreatment increase in CD8⁺ CTL activity in three of the six patients tested, although this did not necessarily correlate with the clinical responses. The lack of a consistent increase in CTL activity, particularly in patients with clinical responses, may reflect the individual heterogeneity in a functional TCRζ signaling apparatus in peripheral blood lymphocytes (11, 12, 21, 32). To isolate a more tumor-specific immune response, we used a dye dilution proliferation assay to measure IFN-γ–producing CD4⁺ or CD8⁺ T cells in response to the culture with tumor lysate–loaded, autologous DCs (53, 54). Thus far, four patients who were tested exhibited an increase in at least one time point in antigen-specific CD4⁺ T-cell precursor frequency in response to treatment.

Our encouraging preliminary results raise the possibility of enhancing the objective response rate and, particularly, the durable clinical responses with therapy that takes advantage of enhancing inflammatory and limiting regulatory pathways.

Dr. Atkins: What is the end point of your renal cancer vaccine trial?

Dr. Ernstoff: This is a classic two-stage, phase 2 trial design, employing response as a primary end point. We've already seen six responders in the first cohort of patients, which exceeds our criteria for the first stage, and thus anticipate accruing our planned 33 patients.

Dr. Vieweg: What kind of T-cell subsets do you induce during vaccination? Do you get T-cell memory? Do you induce CD4 or CD8 T-cell responses?

Dr. Ernstoff: We are still in the process of analyzing data for specific T-cell subsets. We are employing a panel using seven-color flow cytometry to determine the induction of different CD8 and CD4 T-memory and T-effector cells. The tumor-specific proliferation assay that we employ uses autologous DCs loaded with lysed tumor cells or unloaded DCs as stimulators and peripheral blood lymphocytes as responding cells. By use of the dye dilution method and subtraction of the fraction of T cells proliferating in response to unloaded DCs, we can determine tumor-specific T-cell responses in peripheral blood lymphocytes. We are still in the process of completing this assessment in all patients, but preliminary data suggest induction of tumor-specific CD4 response to treatment.

Dr. Vieweg: Are your DCs migratory?

Dr. Ernstoff: We haven't looked at the migratory nature of the DC population that we used to treat patients. We have looked at both DC phenotype and cross-presentation, so we know that there is up-regulation of CD83 and other costimulatory molecules and that the DCs are able to cross-present antigen in vitro.

Dr. Atkins: Do the DCs need to migrate if you are injecting them into the nodes?

Dr. Vieweg: Other studies suggest that (CD40L) activated DCs that are not migratory can cause tolerance.

Dr. Atkins: Is there potentially a selection bias in your study that comes from detecting recurrence early in these patients or is the patient population enrolled in this study mostly the typical population referred in with metastatic disease?

Dr. Ernstoff: Our referral pattern is such that these patients are representative of the everyday metastatic renal cell patient. Approximately half of the patients have metastatic disease at presentation and have a nephrectomy as standard of care. The others are frequently followed up in the community and sent to us only when their disease recurs.

Dr. Sosman: How durable is your treatment response?

Dr. Ernstoff: The two complete responders continue to be in complete response. The longest response is now about 18 or 19 months. Of the four partial response patients, two continue with partial response with no further therapy, probably about 10 or 11 months out, and two have progressed about that time and gone on to antivascular endothelial growth factor therapy.