A Pilot Study of Consolidative Immunotherapy in Patients with High-Risk Pediatric Sarcomas

Impressive advances in the last 30 years for patients with clinically localized Ewing’s sarcoma family of tumors (ESFT) and alveolar rhabdomyosarcoma (AR) have led to current survival rates of 60% to 70% (1–3). However, several clinical groups continue to fare poorly, and long-term toxicity related to therapy is substantial (4). ESFT patients who present with isolated pulmonary metastases have 5-year survival rates ∼30%, whereas <20% of ESFT patients with bone or bone marrow involvement at initial diagnosis survive (5–7), and ESFT patients who recur after frontline therapy also have dismal long-term survival rates (8, 9). Similarly, 5-year survival rates among patients with newly diagnosed metastatic AR are <25% (1, 10), and of the 30% to 40% of patients who present with localized disease but relapse after frontline therapy, most will eventually die of progressive disease (9, 11). Several chemotherapy regimens have shown activity in recurrent AR and ESFT (12–14), but none of these regimens are curative. Therefore, novel therapeutic approaches are needed to improve outcomes for patients with metastatic or recurrent ESFT and AR.

Advances in basic tumor immunology and preliminary evidence for clinical activity of immunotherapy in select settings (15–17) have sustained optimism that immunotherapy will find an established place in cancer therapy. Tumor vaccines are a central component of many cancer immunotherapies, and dendritic cell vaccines have shown preliminary activity in some studies (18, 19). However, as single agents, tumor vaccines have been largely unable to induce regression of established tumors (20), which is not surprising given that animal models show diminished efficacy of T-cell-mediated immunotherapy with increasing tumor burdens (21, 22). Moreover, effective immunotherapy occurs over several weeks to months (15, 23), a timeline impractical for patients with rapidly growing tumors such as pediatric sarcomas. Therefore, as concluded from an initial trial of tumor vaccination in patients with recurrent pediatric sarcomas (24), tumor vaccinations as single agents undertaken in the setting of recurrent ESFT or AR are unlikely to have meaningful antitumor activity. The current study attempted to circumvent these challenges by delivering immunotherapy as consolidation during the period of clinical remission often experienced by high-risk ESFT and AR patients following multimodal therapy. Vaccines comprising dendritic cells pulsed with peptides spanning the breakpoint region of the tumor-specific translocations present in AR and ESFT were administered with autologous lymphocyte infusions ± recombinant human interleukin-2 (rhIL-2). This report describes clinical outcomes for all patients apheresed for potential receipt of immunotherapy, clinical outcomes for the 30 patients who ultimately received immunotherapy on this study, and immune responses for 23 of 30 immunotherapy recipients from whom postvaccination lymphocytes were available for analysis. Detailed presentation of the immune reconstitution patterns in these patients was published previously (25).

Eligibility. This study was approved by the Institutional Review Board of the National Cancer Institute and in accordance with an assurance filed with and approved by the Department of Health and Human Services. Patients were eligible for initial enrollment on this study if they had (a) newly diagnosed metastatic or recurrent AR or ESFT and (b) tumors expressing a t(2:13) or t(11;22) type 1 or 2 translocation. In addition, patients enrolled with recurrent disease were required to have experienced a chemotherapy-free interval of >12 months for patients >5 years and >6 months for patients <5 years and to have a CD4⁺ count that was >400 cells/μL to assure that the T-cell harvest obtained via apheresis would be sufficient. Documentation of the t(11;22) and t(2:13) translocations was made by reverse transcription-PCR analysis of fresh or frozen tumor tissue at the time of initial diagnosis or tumor recurrence as described previously (26). Following apheresis for cell harvest, all patients received cytoreductive therapy comprising chemotherapy, radiation therapy, and/or surgery as determined by the patients’ referring medical teams (Table 1; Supplementary Table S1). On completion of cytoreductive therapy, the following eligibility criteria were required to initiate immunotherapy: acceptable performance status (Eastern Cooperative Oncology Group 0, 1, or 2), acceptable vital organ function, and no requirement for continued cytoreductive therapy during the period of immunotherapy. Tumor response to cytoreductive therapy did not, in and of itself, affect eligibility to receive immunotherapy, as two patients with progressive disease initiated immunotherapy (Table 1). Because a prolonged period elapsed between apheresis and immunotherapy (Fig. 1), informed consent was obtained before apheresis and then again before initiation of immunotherapy.

Cell harvest and immunotherapy. Approximately three blood volumes were processed via apheresis, then separated into lymphocyte and monocyte fractions via countercurrent centrifugal elutriation as described previously (27), and then cryopreserved. Fifty million cells from each fraction were assessed for tumor contamination by nested reverse transcription-PCR for the tumor-specific translocation as described previously (sensitivity, 1 tumor cell/10⁶ cells; ref. 28). See Supplementary Table S2 for primer sequences. Details of the immunotherapy administered to each cohort are shown in Fig. 1. Briefly, all cohorts received three influenza vaccinations and similar doses of autologous T cells. Cohorts 1 and 2 received rhIL-2 (moderate and low dose, respectively), whereas cohort 3 did not receive rhIL-2 but received oral indinavir. To generate the dendritic cell vaccines, cryopreserved monocytes were thawed and placed into culture for 5 to 7 days at 37°C using 10% autologous plasma or human AB serum, rhIL-4 (2,000 units/mL), and recombinant human granulocyte macrophage colony-stimulating factor (2,000 units/mL) as described previously (29). In cohorts 2 and 3, CD40 ligand (1 μg/mL) was added during the last 24 h. On the day of vaccine administration, 50% of the cells were cocultured with 10 μmol/L of the appropriate tumor-derived breakpoint peptide and 50% were cocultured with the control HPV16 E7 peptide (Table 2) for 2 h, then washed, pooled, and irradiated (2,500 cGy) before injection.

Immune response assessment. Immune responses (T-cell proliferation, cytokine production, and cytolysis) in response to influenza, the immunizing tumor translocation breakpoint peptide, and the E7 peptide were measured using cells obtained at the time of initial apheresis following cytoreductive therapy before initiation of immunotherapy and at several time points during and following completion of immunotherapy. Briefly, autologous peptide-pulsed dendritic cells were incubated with the immunizing peptides or whole influenza and then added to peripheral blood mononuclear cells at 1:10 to 1:10,000 stimulator/effector ratios, and proliferation was measured via ³H incorporation. Cytolytic T-cell responses were quantified using a ⁵¹Cr assay, and cytokine (IFN-γ + IL-5) levels were measured in culture supernatants. Positive response criteria, determined before initiation of the study, were as follows: (a) for cytolytic activity, a percent specific lysis of at least 10% as this value was >2 SD of the percent specific lysis measured from a pool of 6 normal donors; (b) for cytokine assays, >2-fold increase compared with no-peptide controls; and (c) for proliferation assays, a stimulation index of >3.0. Anti-influenza A antibody responses were done by Mayo Medical Laboratories.

Statistics. This study used a phase II trial design with α = 0.10 and 90% power to detect a targeted 25% immune response rate to the breakpoint peptide with a goal to enroll a total of 24 patients. Tumor histologies were combined for determination of response rates. Based on interim immune response and immune reconstitution endpoints, the protocol was sequentially revised leading to three cohorts. The immune response data for cohorts 1 to 3 were combined after the data were determined to be sufficiently similar by the Kruskal-Wallis test (results not shown). Statistical analyses included construction of Kaplan-Meier survival curves with two-tailed log-rank tests and Wilcoxon signed rank tests to test the significance of paired differences in immune response data over time. Survival data are calculated from the date of diagnosis for patients enrolled with newly diagnosed metastatic disease and from the date of the last recurrence detection before enrollment on this study for patients with recurrent disease.

Patient characteristics, feasibility, and toxicity. Between 1997 and 2003, 52 patients were enrolled and underwent apheresis for potential receipt of immunotherapy. Because accrual of this rare population required recruitment from long distances, one concern was that the travel required to perform precytoreductive therapy apheresis for cell harvest would result in undue treatment delays, especially for newly diagnosed patients. However, newly diagnosed patients with metastatic disease initiated frontline therapy with minimal delay from the time of diagnosis (median, 11 days; range, 1-24 days). The median time for patients with recurrent disease between recurrence detection and initiation of chemotherapy was 22 (range, 5-117) days (Fig. 1). Apheresis was well tolerated, and adequate harvests meeting phenotypic, viability, and sterility standards established as release criteria were obtained from all patients. Following apheresis, all patients received cytoreductive therapy as dictated by their referring medical teams. Of 52 patients initially enrolled and apheresed, 22 did not initiate immunotherapy due to progressive disease resulting in an unacceptable performance status (n = 11), death due to complications of therapy (n = 1), or patient choice (n = 10).

The immunotherapy regimen was well tolerated [grade 4 thrombocytopenia (n = 1), grade 3 neutropenia (n = 6), diarrhea (n = 2), elevated total bilirubin (n = 1), abdominal pain (n = 1), and skin rash (n = 3)]. Four patients required dose adjustment or discontinuation of rhIL-2. Of the 30 patients in whom immunotherapy was initiated, 23 (77%) completed the full course, whereas 7 patients did not due to disease progression (n = 6) or patient choice (n = 1). Thus, of patients with metastatic or recurrent ESFT and AR who do not develop progressive disease during frontline cytoreductive therapy, a majority will tolerate a complex immunotherapy regimen without undue toxicity.

Immune responses. We have shown previously that dose-intensive chemotherapy for AR and ESFT induces profound lymphopenia (30), T-cell subset alterations (31), and relative expansion of the CD4⁺CD25⁺ regulatory T-cell subset (25), which could potentially limit responsiveness to vaccination. To determine whether patients treated on this trial could respond to a vaccine known to be effective in a healthy population, immune responses were measured following three sequential influenza vaccines administered during the same period as the peptide-pulsed dendritic cell vaccines. Despite profound CD4⁺ depletion at the initiation of immunotherapy (25), all patients analyzed showed immunity to influenza within 6 months of completing cytoreductive therapy. As shown in Fig. 2, increased IFN-γ production in response to influenza occurred within 1.5 months of initiating immunotherapy (P = 0.05) followed by increases in influenza-specific T-cell proliferation (P = 0.05) and humoral immunity (P = 0.006) within 4.5 months of initiating immunotherapy. Most patients also showed cytolytic responses to influenza (data not shown). Thus, patients rendered lymphopenic by dose-intensive cytoreductive therapy for cancer who receive autologous lymphocyte infusions ± rhIL-2 retain a robust capacity to respond to vaccination as evidenced by the brisk influenza-directed immune responses.

With regard to immune responses to the tumor translocation breakpoint peptides, 5 of 23 patients (22%) had measurable immune responses before initiation of immunotherapy (Fig. 2D) as shown by IFN-γ production following exposure to the tumor-specific peptide in vitro. Following dendritic cell vaccination, 2 of those 5 patients maintained immune responses, and 7 patients generated new positive responses. Therefore, 9 of the 23 patients (39%) showed measurable immune responses to the vaccinating peptide (Fig. 2D), mostly limited to IFN-γ production. Importantly, in most patients with immune reactivity toward the immunizing peptide, the responses were transient, dropping below the threshold of positivity at the time of the next analysis (6 weeks later). We detected no difference in response rate to the three tumor-derived breakpoint peptides used for vaccination. Because it was likely that the breakpoint peptides used in this study would not bind to all HLA alleles (32) and because tumor-associated tolerance could limit immune reactivity toward tumor-associated peptides, the HLA-A2-binding HPV16-derived peptide E7 was used as a control. Only 3 of 12 (25%) HLA-A2⁺ patients enrolled on this study generated immune responses to E7. Thus, inadequate HLA binding of the tumor breakpoint peptides cannot fully explain the low immune response rate observed because immune responsiveness was also poor following E7 peptide-pulsed dendritic cell vaccination in HLA-A2⁺ patients where binding could be assured.

Clinical outcomes. The 5-year overall survival (OS) for all patients who underwent apheresis following initial enrollment on this study is 31% (Fig. 3A, median follow-up of surviving patients, 7.6 years) with improved survival in patients initiating immunotherapy in both ESFT and AR cohorts (Fig. 3B and C). Significant survival differences were not observed between histologic subtypes (Fig. 3C); however, patients with recurrent disease had improved OS compared with those with primary metastatic disease (Fig. 3D), although they had similar event-free survival (Fig. 3E). With regard to variables that correlated with survival in the cohort that initiated immunotherapy, we observed with no significant survival differences based on whether patients had a partial response or complete response to cytoreductive therapy (data not shown), evidence for tumor contamination in the lymphocyte product administered (Fig. 3F), differences in immunotherapy administered in cohorts 1 to 3 (Fig. 3G), immune responses measured to the tumor vaccine (Fig. 3H), total CD4⁺ counts at baseline, or postcytoreductive therapy or the frequency of CD4⁺CD25^hi cells at these time points (Supplementary Table S3). Further information regarding disease sites, cytoreductive therapy received, and status at last follow-up for the 30 patients who initiated immunotherapy is shown in Supplementary Table S1 and detailed information regarding the long-term survivors in immunotherapy recipients is shown in Table 3.

Lymphopenia-induced alterations in T-cell homeostasis provide a fertile milieu for inducing immune responses (33), especially when adoptively transferred T cells are incorporated into the regimen (15, 34). Using this paradigm (30, 35), we administered tumor vaccines with autologous T cells to patients with high-risk pediatric sarcomas following cytoreductive chemotherapy. Based on evidence that chromosomal translocations can provide targets for T-cell immune recognition (36–39) and preliminary results in ESFT and AR xenografts (40), the tumor vaccines comprised dendritic cells pulsed with translocation breakpoint peptides. In the first two cohorts, the immune restoration regimen included rhIL-2. However, because we observed rhIL-2-induced expansion of CD4⁺CD25⁺ regulatory T cells (25), the third cohort did not receive rhIL-2. Only the third cohort received indinavir, a HIV protease inhibitor that has shown immunomodulatory effects in vitro (41). The sequential modifications of the immunotherapy administered during the conduct of this trial resulted in each immunotherapy cohort being composed of limited numbers of patients receiving different regimens, a design that makes it difficult to parse the contribution of varying elements within the immunotherapy to the final outcome. One feature common to all patients initiating immunotherapy, however, was the receipt of a sizable dose of autologous T cells during the period of lymphopenia that followed standard cytoreductive therapy. Approximately 1 × 10⁸ T cells/kg were administered, a dose estimated to represent 2% of the total body T-cell pool (42) and which has been associated with antileukemic effects when administered as donor lymphocyte infusions following allogeneic transplantation (43).

With regard to feasibility, the results show that immune cells can be harvested before cytoreductive therapy without undue treatment delays, and monocytes collected from this patient population reliably generated dendritic cells meeting release criteria from all patients. Importantly, however, only 30 of 52 patients apheresed following initial enrollment eventually initiated immunotherapy due to progressive disease, which rendered them unable to meet eligibility criteria (n = 11) for immunotherapy, patient choice (n = 10), or death due to complications of therapy (n = 1). Although progressive disease was not itself an exclusion criteria for receipt of immunotherapy, it is important to note that only 2 patients with progressive disease during cytoreductive therapy initiated immunotherapy. Thus, new approaches for inducing an effective remission remain critical for improving outcomes for sizable numbers of these patient populations because the period of minimal residual neoplastic disease that is required to accomplish this therapy was not achievable in a sizable fraction of this patient population.

Nonetheless, an “intent-to-treat” analysis revealed overall 5-year survival rate for the 52 patients apheresed of 31%, which compares favorably with rates reported in the literature. For the 30 patients who initiated immunotherapy, 5-year OS was 43%; however, selection factors no doubt contribute to this favorable survival rate because patients with rapidly progressive disease did not initiate immunotherapy. The requirement for a prolonged chemotherapy-free interval as part of the eligibility criteria for enrollment with recurrent disease selects for patients with more indolent disease compared with other studies of recurrent disease in the same histologies. Indeed, some patients enrolled following “late” tumor recurrences experienced exceptionally indolent disease courses both before and following immunotherapy (Table 3, patients 6 and 11). This is illustrated graphically in Fig. 3D and E, which shows improved OS for recurrent patients compared with newly diagnosed metastatic patients despite the fact that event-free survivals were not significantly different. When evaluating only the newly diagnosed metastatic subset, of which 63% had ESFT and 37% had ARMS, where this selection factor did not come into play, the 5-year OS for the entire cohort was 23%. This is similar to recently reported OS rates of 25% to 32% and <20% for patients with metastatic ESFT (44–46) and nonembryonal rhabdomyosarcoma (10), respectively, treated with standard regimens.

Similar to findings in some other studies of immunotherapy (47), survival did not correlate with immune response to tumor peptide vaccination. This could indicate that the immunotherapy administered here did not alter the natural history of disease in these patients and that the favorable survival in the immunotherapy recipients relates to selection factors alone. Alternatively, the peripheral blood may not be optimal for measuring immune reactivity because activated immune cells traffic to tissues to affect antitumor immune responses. Finally, it is also possible that adoptive transfer of autologous T cells as done here contributed to control of micrometastatic disease independent of vaccine responses. We published previously that patients with ESFT have circulating T cells, which recognize and lyse autologous tumors (48), and in animal models of sarcoma, adoptive transfer of T cells during a period of minimal residual disease prevented the development of overt metastatic disease (49). Similarly, improved immune reconstitution measured by the lymphocyte count on day 15 following autologous stem cell transplantation correlates with improved survival in several cancers (50), and younger ESFT and rhabdomyosarcoma patients, who experience more rapid immune reconstitution (30), consistently have better outcomes than older individuals (7, 51). Thus, it remains possible that more effective immune reconstitution, rendered via young age or immunotherapy, diminishes the rate of metastatic recurrence.

Patients on this study experienced profound lymphopenia following cytoreductive therapy (25), but essentially all patients generated brisk immune responses to influenza during immunotherapy, whereas the dendritic cell vaccines were relatively ineffective. Several factors likely contribute to this finding. First, both the breakpoint region and the E7 peptide represented neoantigens for most patients, which provide a more stringent test of immune competence than the generation of responses to a recall antigen, which influenza represents. Second, although the breakpoint peptides used as immunogens in this trial bind to some HLA molecules (39), they appear poorly immunogenic overall (32). Third, the dendritic cells administered in this study may not be optimal for inducing primary immune responses. Immature dendritic cells, as administered to the first cohort, are now known to be suboptimal for inducing immune responses (52), and the CD40 ligand-matured dendritic cells used in cohorts 2 and 3 produced relatively low levels of IL-12 (data not shown), a critical immunostimulatory cytokine. Further, i.v. infusion of dendritic cells, as administered to cohort 1, appears less effective than other routes of administration (18). Fourth, although lymphopenia provides opportunities for generating immune responses, it also presents inherent barriers, including expansion of CD4⁺CD25⁺ regulatory cells capable of broad immune suppression (25), and which contribute to the suppression of antitumor immunity. CD4⁺CD25⁺ Tregs were preferentially expanded by rhIL-2 therapy in cohorts 1 and 2 of this study, and significant increases in the frequency of these cells also occurred in response to lymphopenia alone in cohort 3. Although indinavir was incorporated into cohort 3 based on preclinical data, suggesting that it could augment immune reconstitution in patients not infected with HIV, we did not observe any beneficial effect of indinavir in this study with regard to overall immune reconstitution and/or vaccine responsiveness.

In summary, consolidative immunotherapy, as delivered here, incorporates scientific principles to provide a framework for administering immune-based therapy in chemoresponsive cancer. It appears feasible for a sizable fraction of patients with high-risk pediatric sarcomas who experience a good clinical response to frontline cytoreductive therapy, and patients receiving an autologous lymphocyte infusion show sufficient immunocompetence to generate vaccine-induced immune responses to influenza. However, improvements in the regimen are needed to induce strong and sustained immune responses to tumor antigens in a majority of patients. Modifications under study include alternative approaches for inducing dendritic cell maturation (53) and the use of more immunogenic antigens. Although the final determination of the effectiveness of immunotherapies such as this will require analysis of survival outcomes in randomized studies using an intention-to-treat analysis, pilot studies such as this that carefully assess immune endpoints are essential for optimizing immunotherapy for cancer (53).

No potential conflicts of interest were disclosed.

This article is dedicated to the memory of Charles S. Carter whose innovation and dedication to the research and development of cellular immunotherapy played a pivotal role in the completion of this and many other cellular therapy trials conducted throughout the world.

We thank the patients and the families of patients who consented to receive investigational therapy on this study, the staff of the Pediatric Oncology Branch and the NIH Clinical Center for exemplary care of these patients, and Drs. Seth Steinberg and Terry Fry for careful review of the article.