Work

Re: Interdisciplinary Critique of Sipuleucel-T as Immunotherapy in Castration-Resistant Prostate Cancer

Year: 2012

Type: letter

Abstract: Huber et al. made three observations related to the phase III IMPACT trial leading them to suggest that the observed difference in overall survival (OS) could have been attributed to age-dependent immunodepletion resulting from leukapheresis of the control arm patients ( 1 ). Evidence from the sipuleucel-T trials and the literature refute this alternative explanation. Studies have shown age to be prognostic in castration-resistant prostate cancer ( 2 , 3 ). Most studies investigating prognostic factors in castration-resistant prostate cancer have focused on chemotherapy-treated patients in whom other prognostic factors might be expected to outweigh age. The sipuleucel-T studies exclusively enrolled patients with asymptomatic or minimally symptomatic disease, 85% of whom were chemotherapy naïve ( 4 , 5 ). In a pooled analysis of these studies, age (in 10-year increments) was a statistically significant predictor of OS in both sipuleucel-T (hazard ratio [HR] of death = 1.28, 95% confidence interval [CI] = 1.12 to 1.47) and control (HR of death = 1.31, 95% CI = 1.10 to 1.56) arms (Dendreon data on file), which would not be anticipated if the control group had experienced relative immunodepletion. Among 63 subgroups assessed for the IMPACT study report, OS favored sipuleucel-T (ie, a HR of death < 1.0) in all subgroups including above and below the median age of 71 ( 4 ); only the subgroup of patients aged less than 65 years did not favor sipuleucel-T. Given the high probability of a false-positive result with this number of subgroup analyses, it is important to seek additional data for consistency. In an analysis of the first two phase III sipuleucel-T trials, positive treatment effects were observed in patients less than 65 years of age (HR of death = 0.57, 95% CI = 0.27 to 1.21) and 65 years or older (HR of death = 0.69, 95% CI = 0.49 to 0.98), with the trend actually favoring a greater treatment effect in the younger subgroup (Dendreon data on file). Furthermore, when both IMPACT and the earlier phase III studies are dichotomized at the median age of 71, the treatment effect of sipuleucel-T is consistent in both the younger and older patients ( 4 , 5 ). The US Food and Drug Administration (FDA) performed an analysis of pooled data across the three phase III sipuleucel-T trials and obtained a favorable estimate for the treatment effect in patients younger than 65 years (HR of death = 0.92, 95% CI = 0.62 to 1.37). The FDA reviewers’ conclusion was that the “… analyses of the data from all three studies … support the hypothesis that the subgroup of patients who were less than 65 years of age also benefit from treatment with sipuleucel-T. The hazard ratio in the subgroup of Study D9902B [IMPACT] patients who were less than 65 years of age most likely resulted from chance, related to the multiplicity of comparisons…” ( 6 ). A review of recent trials with patient populations comparable to that in IMPACT demonstrates consistency of the median survival of the control arms, ranging from 16.6 to 21.5 months ( Table 1 ) ( 4 , 7 ). Huber et al. ( 1 ) cited the 27.1-month median OS shown for the control group in a subset analysis of 264 patients from the GVAX trial who enrolled with a Halabi-predicted survival of greater than 18 months. We calculated the median OS of IMPACT control patients with a more than 18-month Halabi-predicted survival (n = 114); at 26.7 months, it compares favorably to the median OS of control patients from the GVAX subset analysis, particularly given that the GVAX control arm involved active treatment with docetaxel. Furthermore, the definition of “minimally symptomatic” used for the subgroup of patients from TAX327 who had a 25.6-month median OS (n = 110) included a minimal score on a quality-of-life questionnaire ( 8 ), which was not included in the IMPACT definition. A larger TAX327 minimally symptomatic subgroup (n = 550) that did not include the quality-of-life criterion for defining symptomatology revealed a median survival of 21.3 months for all three treatment arms combined ( Table 1 ) and 19.8 months specifically in the control arm ( 9 ). Key prognostic factors and outcomes from clinical trials of patient populations comparable to those enrolled in IMPACT (4,7)* * Alk Phos = alkaline phosphatase; ECOG = Eastern Cooperative Oncology Group performance status scale; LDH = lactate dehydrogenase; NR = not reported; OS = overall survival; PSA = prostate-specific antigen. † In the IMPACT trial, sipuleucel-T (S) was compared with a control intervention (C). In the prostvac trial, recombinant vaccinia vector-PSA-TRICOM (2×10 8 pfu) followed by recombinant Fowlpox-PSA-TRICOM boost (1×10 9 pfu) + recombinant granulocyte-macrophage colony-stimulating factor adjuvant (PV) was administered to patients and was compared with patients who received the control intervention (C). Patients in the Tax 327 trial were administered 75mg/m 2 docetaxel every 3 weeks + 5mg prednisone twice daily (D3P) and were compared with both patients who received 30mg/m 2 docetaxel weekly + 5mg prednisone twice daily (D1P) and those who received 12mg/m 2 mitoxantrone every 3 weeks + 5mg prednisone twice daily (MP). No previous therapy with chemotoxic agents with the exeption of estramustine, which was allowed. The clinical trial of zibotentan compared patients who recieved 15mg zibotentan daily (Z15) with those who received 10mg zibotentan daily (Z10) and those who received an oral placebo daily (P). In the ASCENT-2 trial, the treatment arms were high-dose calcitriol daily + 36mg/m 2 docetaxel every 3 weeks + 8mg oral dexamethasone 12 hours, 3 hours, and 1 hour before docetaxel infusion for 3 of 4 weeks (CD1) vs prednisone twice daily + 75mg/m 2 docetaxel every 3 weeks and 8mg oral dexamethasone 12 hours, 3 hours, and 1 hour before docetaxel infusion (PD3). Prior chemotherapy (only estramustine monotherapy) was allowed. The atrasentan trial compared 10mg atrasentan daily (A) vs oral placebo daily (P). The median Gleason score in both treatment arms was 7; percentage of patients with Gleason score of 7 or lower was not reported. ECOG was not reported; the value listed for ECOG 0–1 represents the percentage of patients with Karnofsky Performance Status 100–80. Key prognostic factors and outcomes from clinical trials of patient populations comparable to those enrolled in IMPACT (4,7)* * Alk Phos = alkaline phosphatase; ECOG = Eastern Cooperative Oncology Group performance status scale; LDH = lactate dehydrogenase; NR = not reported; OS = overall survival; PSA = prostate-specific antigen. † In the IMPACT trial, sipuleucel-T (S) was compared with a control intervention (C). In the prostvac trial, recombinant vaccinia vector-PSA-TRICOM (2×10 8 pfu) followed by recombinant Fowlpox-PSA-TRICOM boost (1×10 9 pfu) + recombinant granulocyte-macrophage colony-stimulating factor adjuvant (PV) was administered to patients and was compared with patients who received the control intervention (C). Patients in the Tax 327 trial were administered 75mg/m 2 docetaxel every 3 weeks + 5mg prednisone twice daily (D3P) and were compared with both patients who received 30mg/m 2 docetaxel weekly + 5mg prednisone twice daily (D1P) and those who received 12mg/m 2 mitoxantrone every 3 weeks + 5mg prednisone twice daily (MP). No previous therapy with chemotoxic agents with the exeption of estramustine, which was allowed. The clinical trial of zibotentan compared patients who recieved 15mg zibotentan daily (Z15) with those who received 10mg zibotentan daily (Z10) and those who received an oral placebo daily (P). In the ASCENT-2 trial, the treatment arms were high-dose calcitriol daily + 36mg/m 2 docetaxel every 3 weeks + 8mg oral dexamethasone 12 hours, 3 hours, and 1 hour before docetaxel infusion for 3 of 4 weeks (CD1) vs prednisone twice daily + 75mg/m 2 docetaxel every 3 weeks and 8mg oral dexamethasone 12 hours, 3 hours, and 1 hour before docetaxel infusion (PD3). Prior chemotherapy (only estramustine monotherapy) was allowed. The atrasentan trial compared 10mg atrasentan daily (A) vs oral placebo daily (P). The median Gleason score in both treatment arms was 7; percentage of patients with Gleason score of 7 or lower was not reported. ECOG was not reported; the value listed for ECOG 0–1 represents the percentage of patients with Karnofsky Performance Status 100–80. Huber et al. ( 1 ) failed to account for the fact that the intravascular pool of mononuclear cells represents a small fraction of the total body pool and that these populations rapidly reequilibrate ( 10 ). The 1.5–2.0 volume leukapheresis procedure in IMPACT removed approximately 7×10 9 lymphocytes. Given that the total body pool of lymphocytes is estimated between 5×10 11 and 6×10 12 cells ( 11–13 ), leukapheresis removed only 0.1%–1.4% of the total number of lymphocytes. This small percentage would not be expected to have any substantive effect on the lymphocyte repertoire. An extensive body of literature has shown no detrimental effects of repeated apheresis procedures on healthy donors ( 14 ), and the median white blood cell, neutrophil, lymphocyte, and monocyte counts in the sipuleucel-T studies were within normal ranges at 2, 10, and 22 weeks following the third leukapheresis procedure in both the sipuleucel-T and the control groups ( 15 ). Furthermore, cells undergoing homeostatic proliferation actually tend to be more, rather than less, immunoreactive, as evidenced by multiple cancer immunotherapy protocols that induce lymphopenia before immunization and/or adoptive T-cell transfer to augment immune activation and clinical efficacy. Finally, infection-related adverse events were not increased among the control arm and were consistent with what would be expected in an elderly population over an extended period of follow-up ( 15 ). The storage of the cells from the control group did not lead to infusion of dead cells, because the median viabilities were greater than 95% for both sipuleucel-T and control groups (Dendreon unpublished data on file). The control arm intervention for the sipuleucel-T trials was chosen after careful consideration and discussion with the FDA. Having patients in the control arm undergo leukapheresis procedures and subsequent infusion of a cellular product that was identical in appearance to sipuleucel-T substantially strengthened the study design and the interpretability of the results. The sipuleucel-T data have undergone rigorous review, including an FDA advisory committee, FDA review, a Center for Medicare and Medicaid Services national coverage determination, a Technology Assessment, and the peer review of multiple publications. The issues raised by Huber et al. were considered during these reviews and not given credence. No author has received financial support for the submitted work. The IMPACT trial was funded by Dendreon Corp. P. W. Kantoff served as a paid consultant for Bellicum, BN-IT, Janssen, Bayer, Sanofi-Aventis, and Dendreon Corp. C. S. Higano has served as a paid consultant for Amgen, Astra Zeneca, Bayer, Bristol-Myers Squibb, Medivation, Millennium, Sanofi-Aventis, Teva, and Dendreon Corp; received research funding from Amgen, Bayer, Cell Genesys, Cougar, Imclone, Medivation, Millennium, Sanofi-Aventis, Teva, and Dendreon Corp; received honoraria from Amgen, Astra Zeneca, Bayer, Bristol-Myers Squibb, Medivation, Millennium, Sanofi-Aventis, Teva, and Dendreon Corp; her spouse is cofounder of Cell Therapeutics, Inc. E. J. Small has received honoraria from Celgene Corp, F Hoffman-La Roche Ltd, J&J Pharma Serv, LLC, Millennium, and Dendreon Corp and has received other remuneration from Dendreon Corp. J. B. Whitmore and M. W. Frohlich are employees of Dendreon Corp. P. F. Schellhammer has received honoraria from Amgen, Astra Zeneca, Bayer, and Dendreon Corp. The authors thank Johnathan C. Maher, PhD, of Dendreon Corp for contributing to the writing and editing of the submitted work. All authors contributed to the conception and design of the manuscript, data analysis and interpretation, writing and editing of the manuscript, and approved the final version for submission.

Source: JNCI Journal of the National Cancer Institute

DOI: 10.1093/jnci/djs279

Authors: P. W. Kantoff, C. S. Higano, E. J. Small, J. B. Whitmore, M. W. Frohlich, P. F. Schellhammer