Complexities in Interpretation of Osteosarcoma Clinical Trial Results
Cancer Therapy Evaluation Program, National Cancer Institute, Bethesda, MD
Results from a randomized factorial design study reported by Meyers et al in 20051 and 20082 make important contributions to the body of evidence on the use of liposomal muramyl tripeptide phosphatidyl ethanolamine (MTP), given with either three standard chemotherapy agents or with the same agents (administered on a different schedule) plus ifosfamide, for treatment of osteosarcoma. However, as discussed below, additional study is necessary to define if or how MTP should be used in the treatment of osteosarcoma.
The 2005 report identified event-free survival (EFS) as the primary outcome measure and presented EFS outcome data,1 whereas the 2008 article presented a reanalysis of EFS as well as overall survival based on additional follow-up.2 The 2005 analysis indicated significant interaction between the ifosfamide and MTP questions with respect to the primary end point of EFS (P = .049 for test of interaction). The 2008 article interprets a P value of .102 for the test of interaction as evidence of no interaction (by missing the .10 significance cutoff). In the context of a 2 × 2 factorial design, there is adequate power only for detecting very large interactions, so one cannot interpret a borderline P value as evidence of no interaction. As Altman and Bland3 discuss, absence of evidence is not evidence of absence, a point that is particularly relevant when discussing evidence of interactions in factorial designs.4 What can be said is that the evidence (or magnitude) of an interaction has been reduced in the 2008 reanalysis. However, the therapeutic implications of the interaction that was clearly present in the primary 2005 analysis and that is still apparent in the 2008 report (Myers et al, Fig 2A) cannot be ignored simply because the interaction test on the cumulative follow-up data slightly exceeds a cutoff value.
The key point to be made is that because of the interaction, the analysis of the MTP marginal effect pooled over chemotherapy arms should be interpreted with caution. This is because in the presence of an interaction among treatments, a pooled analysis will produce an estimate of the MTP treatment effect that may be either too large or too small depending on whether ifosfamide is present or not.4 The purpose of testing for interaction in this specific setting is to assess whether MTP can be recommended based on the pooled analysis for an individual patient regardless of whether the three-drug or four-drug regimen is used. The test is not intended to disprove interaction, because it is almost always present at some level, but rather to evaluate the degree to which it can be ignored. It seems that in this study, potential interaction effect (on EFS) should be considered in guiding clinical practice.
The highlight of the 2008 reanalysis is the finding of survival improvement for MTP when chemotherapy arms are pooled. However, the reanalysis likely lacks the ability to rule out a clinically meaningful interaction in survival (especially because there are fewer survival events than there are EFS events). Along these lines, the interpretation of the results could be enhanced by examining the estimates (and confidence intervals) of the interaction for EFS from both the 2005 and 2008 reports and overall survival from the 2008 report. Given that the effect of treatment on survival is expected to be mediated in large measure through its effect on EFS, the concern of interaction carries over to the analysis of survival, and caution is required in interpreting this pooled survival analysis. In this situation, a more clinically relevant assessment is provided by comparing the individual MTP-containing regimens to the three-drug chemotherapy control arm regimen. The drawback of this approach is that the study was not sized to allow such comparisons to be made with reasonable statistical power.
Randomized clinical trials are the gold standard for shaping clinical practice. For a study to promote a change in practice, it is important that the data are robust enough to provide unequivocal evidence that is generalizable to the relevant clinical setting. In this study, due to the interaction concerns outlined above, the results, in our view, do not meet generally accepted standards for practice-changing conclusions. However, the results do suggest the possibility for benefit from the use of MTP with (or possibly without) concurrent ifosfamide for osteosarcoma. Additional clinical evaluations will be required to reliably define the role of MTP in the treatment of osteosarcoma and to demonstrate whether any benefit that might exist for MTP requires concurrent use of ifosfamide.
The author(s) indicated no potential conflicts of interest.
|1.||Meyers PA, Schwartz CL, Krailo M, et al: Osteosarcoma: A randomized, prospective trial of the addition of ifosfamide and/or muramyl tripeptide to cisplatin, doxorubicin, and high-dose methotrexate. J Clin Oncol 23::2004,2005-2011, Link, Google Scholar|
|2.||Meyers PA, Schwartz CL, Krailo MD, et al: Osteosarcoma: The addition of muramyl tripeptide to chemotherapy improves overall survival—A report from the Children's Oncology Group. J Clin Oncol 26::633,2008-638, Link, Google Scholar|
|3.||Altman DG, Bland JM: Absence of evidence is not evidence of absence. BMJ 311::485,1995, Crossref, Medline, Google Scholar|
|4.||Brittain E, Wittes J: Factorial designs in clinical trials: The effects of non-compliance and subadditivity. Stat Med 8::161,1989-171, Crossref, Medline, Google Scholar|