Randomized Study of Intravenous versus Subcutaneous Interleukin-2, and IFNα in Patients with Good Prognosis Metastatic Renal Cancer

Approximately 40% of the patients diagnosed with renal cell carcinoma develop metastases (1, 2). Metastatic renal carcinoma is refractory to chemotherapy, and survival is generally around 1 year (3–5). IFN-α and interleukin 2 (IL-2) are widely used for first-line treatment in patients with metastases. Many phase II studies with these agents have shown tumor regressions in small subsets of patients, but very few have shown improved survival. A complete review and a pooled analysis of the literature were recently done by the Cochrane Collaboration (6). Analysis of the four available randomized trials showed that IFN-α is associated with a 3.8 months longer weighted average median survival than nonimmunotherapy controls (i.e., medroxyprogesterone or vinblastine; refs. 7–10). The benefit of single-agent IL-2 on survival has never been tested against control treatments, and comparisons with other drugs have failed to show any advantage of the cytokine (6).

In 1998, the Groupe Français d'Immunothérapie reported the results of a large multicenter trial that compared IL-2, IFN-α, and their combination, and investigated factors predictive of response and failure to these treatments (11). Three subgroups of patients were identified (12). Patients who combined several metastatic organ sites, liver metastases, and a time interval from primary tumor to metastases less than 1 year were shown to have >70% probability of treatment failure and very poor outcome [median survival, 6 months; 95% confidence interval (95% CI), 4-8 months]. This subgroup, which represents 25% of all renal cancer patients, was no longer considered candidate for cytokine treatment. The majority (55%) of patients had an intermediate chance of responding to or failing cytokine treatment; they were proposed to be enrolled into a randomized trial comparing the different cytokine treatments to a control group with medroxyprogesterone acetate. This trial failed to show any survival advantage of cytokine treatments and showed that cytokine administration was associated with a significant risk of toxicity (13, 14). Finally, the 20% remaining patients had only one metastatic site associated with the best prognosis: a 37% chance of response to a combination of both cytokines and a longer survival (median, 25 months; 95% CI, 16-34 months). The Groupe Français d'Immunothérapie proposed to test a combination of IFN-α and IL-2 in this subset of patients and to assess the importance of the administration route. If the s.c. administration of IL-2 is definitely more convenient and less toxic than the i.v. route, some authors think that the i.v. route induces more complete remissions, which ultimately translates into a gain in survival (15, 16). The PERCY Duo trial was designed to address this question in the predefined subgroup of patients who have the best chances of response to a combination of both cytokines.

Selection of patients. Patients older than 18 years were eligible if they had histologically confirmed, clearly progressive metastatic renal carcinoma, no more than one metastatic organ site, good performance status (Karnofsky score ≥90%), and normal blood and liver functions with creatinine level <150 μmol/L. All histologic subtypes of renal cell cancer were eligible. Patients with previous systemic treatment or radiotherapy within 6 weeks of randomization, or with evidence of active brain metastases, severe cardiac dysfunction, active infections, or current corticosteroid treatment were not eligible. Patients with a history of organ transplantation, or with other cancer or seizure, as well as pregnant or lactating women, were excluded. The baseline work-up consisted of thoracic, abdominal, and pelvis computed tomography scan, brain magnetic resonance imaging, or computed tomography scan and bone scan. The number of organs with metastases was determined in each patient before randomization according to a specific list of sites: lung, liver, bone, contralateral kidney, abdominal lymph nodes, superficial lymph nodes, ipsilateral renal fossa, adrenal glands, mediastinum, pleura, skin, central nervous system, and other.

Trial design. Patients were randomly assigned to receive a combination of IFN-α with either i.v. IL-2 (arm A) or s.c. IL-2 (arm B). The major end point was overall survival (OS). The sample size calculation assumed a 15% improvement in OS at 4 years with a one-sided α risk of 5% and a β risk of 20%. OS rates at 4 years were estimated to be 40% and 25% in i.v. IL-2 and s.c. IL-2 arms, respectively. Under these assumptions, 220 patients would be required to detect a significant difference between groups.

Randomization was stratified by the participating center using a block method. Treatment allocation was done centrally through a specific Web site. The protocol was approved by the ethics committee and local institutional review boards, in compliance with French laws. Written informed consent was obtained from all patients before randomization. An independent review committee evaluated the quality of the study, especially with respect to data monitoring and statistical analysis.

Treatment. In arm A, IL-2 (Proleukin, Chiron) was administered as a continuous i.v. infusion (18 × 10⁶ IU/m²) and IFN-α (Roche) was given s.c. (6 × 10⁶ IU thrice a week) as previously reported in the CRECY study (11). Induction treatment consisted of two 5-day courses of IL-2 separated by a 1-week break. This treatment cycle was repeated after 3 weeks of rest. IFN-α was given throughout each of the two treatment cycles. In patients who did not progress, maintenance consisted of four 5-day courses of the combination of IL-2 and IFN-α, separated by 3 weeks of rest.

In arm B, 9 × 10⁶ IU IL-2 were given s.c. on a 4-week schedule as follows: twice daily for 5 days during the 1st week, then twice daily for 2 days and once daily for 3 days during the following 3 weeks. After a week of rest, an identical 4-week cycle was administered. IFN-α was given s.c. as 6 × 10⁶ IU thrice a week throughout the two 4-week cycles, following the schedule previously described (17).

Detailed recommendations for dose modifications were given to the investigators. In summary, all toxic events judged as grade 3 for intensity and prolonged over 2 weeks, or events of grade 4 intensity, could justify permanent treatment discontinuation.

Tumor evaluation was done 12 weeks after treatment start in both arms; patients without disease progression were eligible for maintenance treatment as shown in Fig. 1. Crossover between treatment arms was not allowed.

Assessment of response, toxicity, and quality of life. Tumor response was classified in accordance with the WHO criteria (18). Objective complete and partial responses were assessed by thoracic, abdominal, and pelvic computed tomography scan. Tumor evaluation was done 12 weeks after treatment start and between weeks 24 and 26 in patients receiving maintenance. Adverse events were evaluated according to the National Cancer Institute toxicity criteria (19), and quality of life was evaluated through the European Organization for Research and Treatment of Cancer Quality of Life questionnaire (EORTC QLQ-C30; ref. 20); patients were proposed to complete these questionnaires just before treatment and several days before tumor assessment at 12 weeks.

Statistical analysis. OS was calculated from the date of randomization to the date of death from any cause or to last follow-up for living patients (censored observation). Survival estimates were calculated by using the Kaplan-Meier method (21). Differences in survival estimates between arms were assessed by the log-rank test (22). Median follow-up was calculated by using reverse Kaplan-Meier estimation.

Secondary end points were progression-free survival measured from randomization to disease progression or death, or censored at last follow-up; objective tumor response was assessed by descriptive statistical analysis; toxicity profiles were compared between arms using Fisher's exact test; and quality of life was measured by the global quality of life score (items 29 and 30) of EORTC QLQ-C30 v.3 questionnaire and compared between arms using a nonparametric Wilcoxon rank-sum test (23).

Survival was analyzed on an intent-to-treat basis, and no interim analysis was planned or done. According to the protocol, a one-sided log-rank test was used for the primary end point; all other tests were two-sided.

Patient characteristics. From January 2000 to January 2005, 80 patients were randomized to the combination with i.v. IL-2 and 75 patients to the combination with s.c. IL-2. Enrollment was stopped earlier than planned because of poor accrual, especially during the last year. There were two reasons for this. First, the number of patients with only one metastatic organ site selected after strict prospective control was lower than anticipated from previous trials with broader entry criteria. For instance, in doubtful cases, especially mediastinal lymph node involvement with size around 1 cm, the policy was to reject enrollment. Second, because new active agents became available through clinical trials and compassionate programs during the last year, patients were less likely to participate.

Analysis was done 14 months after the last enrollment, which corresponded to a median follow-up of 42.5 months (95% CI, 39.4-51.7).

At the time of analysis, 11 patients appeared to have multiple metastatic sites, although this was an exclusion criteria; 4 had been randomized to i.v. IL-2 and 7 to s.c. IL-2. The characteristics of patients according to their randomization group are detailed in Table 1; there were no significant differences between groups.

Treatment. Two patients randomized to i.v. IL-2 did not receive any treatment; 42 (27.5%) patients did not receive full doses of the planned cytokines during the first phase of treatment (before the first tumor evaluation). Premature discontinuation was more frequent in arm A (23 patients stopped, 20 because of toxicity) than in arm B (13 patients stopped, 9 because of toxicity). Treatment intensity during the induction treatment period was calculated and compared because the planned doses of cytokines were close between the two groups. The average percentage of IL-2 administered was 57.5% (SD 23.6) in arm A versus 84.7% (SD 21.4) in arm B (P < 0.0001), whereas patients received an average percentage of 77.0% (SD 27.7) of IFN-α in arm A versus 88.4% (SD 21.0) in arm B (P = 0.0044). After tumor evaluation, 72 patients (27 in arm A, 45 in arm B) received some maintenance treatment.

Subsequent to the immunotherapy, 35 patients (16 in arm A, 19 in arm B) received antiangiogenic agents through clinical trials or compassionate use programs. Details of these treatments were as follows: 17 patients (8 in arm i.v. IL-2, 9 in arm s.c. IL-2) were enrolled in a double-blind trial of sorafenib versus placebo, 9 patients (5 in arm i.v. IL-2, 4 in arm s.c. IL-2) received bevacizumab, and 9 (3 in arm i.v. IL-2, 6 in arm s.c. IL-2) received sunitinib.

Toxicity and quality of life. A total of 412 grade 3 or 4 adverse events were reported, 281 (68.2%) in arm i.v. IL-2 and 131 (31.8%) in arm s.c. IL-2, corresponding to 85.9% versus 74.7% patients (P = 0.08). Details of these toxic events are given in Table 2. No toxic death was observed. Hypotension, skin toxicities, elevation of creatinine, and vomiting were more frequently observed in patients treated with i.v. IL-2.

Quality of life was evaluated using the EORTC QLQ-C30: 127 (81.9%) patients completed the questionnaire at the time of randomization and 97 (62.6%) after 12 weeks.

The global quality of life scores obtained before treatment were very similar between the two groups (median, 83.3 for both arms; Wilcoxon nonparametric test, P = 0.53). Both scores decreased after treatment with again no significant difference between arms (58 in arm i.v. IL-2 and 50 in arm s.c. IL-2, P = 0.23).

Efficacy. At the time of analysis, 87 (56%) patients had died: 44 of 80 in arm i.v. IL-2 and 43 of 75 in arm s.c. IL-2. The median OS time was 33 months (95% CI, 27.0-40.2). As shown in Fig. 2, there was no significant difference in OS between randomization arms (one-sided log rank test, P = 0.202). The median survival time was 37.7 months (95% CI, 28.2-55.6) and 30.1 months (95% CI, 25.1-34.5) in arms A and B, respectively.

Figure 3 shows progression-free survival according to randomization arm; there was no statistically significant difference between arms and median times were close: 7.2 months (95% CI, 6.0-9.6) in arm i.v. IL-2, 6.2 months (95% CI, 5.1-8.5) in arm s.c. IL-2.

Response rates (Table 3) were not different at 3 months, that is, 17.9% in arm i.v. IL-2 versus 21.3% in arm s.c. IL-2 (P = 0.60) compared with 15.3% and 24% at 6 months (P = 0.17), respectively.

No progression occurred after 3 years in 50% and 26% of the 12 and 18 responding patients of arms A and B, respectively.

For decades, IL-2 and IFN-α have been used for the treatment of patients with metastatic renal cell carcinoma, although with limited efficacy. Attempts to identify robust, reproducible biological prognostic factors have failed, but two clinical parameters seem to be critical for tumor regression under these treatments, good performance status, and limited extent of the disease (6, 11, 24). The present study was conducted in a subgroup of patients with these criteria who were considered the best candidates for cytokine treatments. Our primary objective was to test whether the continuous i.v. administration of IL-2 is superior to the s.c. route when combined with IFN-α.

More toxicities were observed with i.v. IL-2 than with s.c. IL-2. Hypotension, increase in serum creatinine level, nausea, and neurologic toxicity were significantly more frequent and more severe with i.v. IL-2. As a consequence, treatment duration was significantly shortened in this group and treatment intensity was lower. These results confirmed previous observations by other authors in unselected cohorts of patients (15, 16).

Quality of life results seemed somewhat surprising because they did not confirm the general opinion of a worse effect of i.v. IL-2. We think that the evaluation of quality of life through general questionnaires and at limited points in time (before and after the end of the induction treatment period) is probably not adapted to detect relatively short periods of quality of life impairment.

In terms of activity, the response rates observed in both treatment groups (around 20%) were virtually identical to observations from previous unselected series, which was rather disappointing. The response rates obtained failed to match the estimates calculated retrospectively on a previous cohort of patients (i.e., 37% probability of response; ref. 12). This illustrates that retrospective estimates of probability can differ from actual rates observed prospectively on selected patients.

Progression-free survival, with median times between 6 and 7 months, seemed longer than commonly observed in unselected patients treated with cytokines (6, 15). However, these results are clearly inferior to those obtained with new antiangiogenic agents like sunitinib or bevacizumab combined with IFN-α (5, 25, 26).

Although the planned number of patients could not be reached and the dose intensity of treatment was significantly inferior in the group receiving i.v. IL-2, the OS at 4 years was slightly in favor of this treatment group. This could be due to a better quality of response to the i.v. combination because, among the 30 6-month responders, 50% of i.v. IL-2 patients had not progressed at 3 years compared with 26% of s.c. IL-2 patients. This superiority in terms of duration of response has also been reported in randomized trials using i.v. high-dose bolus IL-2 in nonselected patients (15, 16).

It is unknown whether adequate patient enrollment, as determined by the study design, would have changed the results, but the lack of difference between the two arms in terms of complete remission as well as progression-free survival does not support this hypothesis.

In conclusion, the combination of IFN-α with continuous infusion of IL-2 did not induce significantly higher response rate or progression-free survival in selected patients than the combination with s.c. low-dose IL-2, and the i.v. route was associated with greater toxicity. Even if response duration seems longer with i.v. IL-2, it is unreasonable to continue to recommend this regimen now that new therapies, which have been shown to double the length of progression-free survival, are available.

S. Negrier and B. Escudier are members of the Roche Pharmaceuticals advisory board.

Investigators: C. Chevreau, N. Tubiana, E. Legouffe, B. Audhuy, J.C. Eymard, M. Mousseau, N. Dohollou, F. Rousseau, B. Ceccaldi, L. Geoffrois, B. Leduc, E. Khenifar, V. Lucas, D. Coeffic, N. Meneveau, I. Mechin, C. Platini, B. Sautois. Coordinating center and statistics: E. Blanc, S. Tanguy, I. Calmes, F. Gomez, C. Ferlay.