2 Clarke Drive
Suite 100
Cranbury, NJ 08512
© 2024 MJH Life Sciences™ and OncLive - Clinical Oncology News, Cancer Expert Insights. All rights reserved.
Maurie Markman, MD, offers examples of clinical trials where the interpretation of study results is worthy of considerable additional discussion or where justification for the actual conduct of the study can be called into question.
Maurie Markman, MD
Although publication of clinical trial findings in a peer-reviewed journal is a vital step in disseminating emerging research developments, there is a critical need for all readers of oncology literature to approach these reports with a healthy degree of skepticism. Among the important questions that readers should ask are whether the conclusions are truly meaningful to patients and why the study was initiated in the first place.
This commentary offers examples of clinical trials where the interpretation of study results is worthy of considerable additional discussion or where justification for the actual conduct of the study can be called into question. In fact, it is not hard to find studies in the peer-reviewed oncology literature where an objective observer could lodge a serious challenge against the conclusions reached by the study investigators. The examples discussed here include a colorectal cancer study and, due to the research interests of this commentator, several studies focused on the gynecologic cancer area. However, it is virtually certain that similar arguments could be advanced in regard to multiple other areas within oncology.
Debatable Conclusions in Colorectal Study
FOLFOXIRI Versus FOLFIRI
The recent publication of the results of a phase III trial that compared two multiagent combination chemotherapy regimens (FOLFOXIRI vs FOLFIRI) plus bevacizumab as first-line treatment of metastatic colorectal cancer serves as the starting point for this commentary.1 The issue here is whether the statistically significant improvement in progression-free survival (PFS) of a median 2.4 months (P = .003) and a 12% increase the objective response rate (65% vs 53%), but nonstatistically significant superior overall survival (OS) of a median 5.2 month difference (P = .054) associated with the delivery of the FOLFOXIRI program justify its use when the regimen is also associated with a rather impressive increased risk of certain highly clinically relevant grade 3-4 toxicities (peripheral neuropathy [5.2% vs 0%]; diarrhea [18.8% vs 10.6%]; stomatitis [8.8% vs 4.3%]).
How should the study be interpreted? Should the statistically significant improvement in PFS and objective response rate in this difficult clinical setting trump the associated increased risk of distressing side effects and permit the study to be labeled as a “positive” trial? Or should the toxicity profile in the absence of a statistically significant impact on OS result in the conclusion that this regimen is not a favored alternative in this clinical setting? This is especially relevant when another approach might be the sequential delivery of agents rather than a single combination program, resulting in equivalent survival, less treatment-related toxicity, and an overall superior quality of life.
Unfortunately, and perhaps surprisingly considering the risk of serious side effects, a formal quality-of-life assessment was not performed as a component of this study. The authors of this trial appear to conclude the added toxicity is justified based on the previously stated clinical endpoints. But considering the modest documented improvements noted in the presence of a substantial increased risk of clinically relevant side effects, it would not be difficult to challenge the interpretation of these investigators.
Questionable Findings in Ovarian Trial
Pegylated Liposomal Doxorubicin Versus Topotecan
This study compared single-agent pegylated liposomal doxorubicin with single-agent topotecan as a treatment for platinum-resistant or recurrent potentially platinum-sensitive ovarian cancer.2 There was no observed difference in outcome (response rate, PFS, or OS) between the drugs in the platinum-resistant patient population. However, in the potentially platinum-sensitive group, the administration of pegylated liposomal doxorubicin was associated with a very modest improvement in PFS (median 28.9 weeks vs 23.3 weeks; P = .037) but a substantially greater difference in OS (median 108 weeks vs 71.1 weeks; P = .008).2 Somehow, a median difference of less than 2 months in PFS was translated into a median 9.2 month improvement in OS.
The investigators concluded that the administration of pegylated liposomal doxorubicin improved OS in recurrent platinum-sensitive ovarian cancer. But is this the only—or even the most likely—explanation for these unexpected and realistically unexplained findings?
An alternative perspective on the study results begins by noting that patients in this specific subgroup of the trial where the favorable impact on survival was observed potentially had platinum-sensitive disease (recurrent >6 months following the last dose of a platinum agent) when they entered the investigative program. As a result, at the time of progression on either study drug, there is a high probability the treating oncologist would select a platinum agent (most likely carboplatin) as the next therapeutic regimen. However, due to the recognized major hematologic toxicity of topotecan (delivered at the FDA-approved dose of 1.5 mg/m2 for 5 days and repeated on a 21-day schedule), it is reasonable to suggest that patients previously managed with this regimen on the trial were considerably less likely to be able to subsequently successfully receive carboplatin compared with individuals treated with the far less marrow-toxic pegylated liposomal doxorubicin.
Therefore, it is appropriate to hypothesize that rather than revealing the superiority of pegylated liposomal doxorubicin in improving OS in women with recurrent ovarian cancer, this study provided evidence for the negative impact of topotecan in preventing the adequate second-line administration of the single most important class of drugs (platinum) in this malignancy.
Flawed Rationale for Maintenance Study
Erlotinib Versus Placebo In Epithelial Ovarian Cancer
The results of a recently reported phase III trial explored the use of erlotinib, a small-molecule inhibitor of EGFR, as maintenance therapy for patients with epithelial ovarian cancer who had achieved a clinically defined complete response. Erlotinib failed to improve either PFS or OS com- pared with a placebo control.3
The fundamental question to be asked here is why this study was conducted when rather substantial previously reported clinical data in lung cancer had revealed overexpression of EGFR (present in a substantial percentage of epithelial ovarian cancers) was not a clinically relevant target for this class of agents? Is it acceptable that these patients were entered as “research subjects” into a trial whose biological basis was so seriously flawed?
By contrast, one could make a highly valid argument that the presence of activating mutations in EGFR, which have been shown to be a critically important target for tyrosine kinase inhibitors of EGFR in non—small cell lung cancer, might be a rational investigative strategy in ovarian cancer. In fact, the single ovarian cancer patient with such a mutation in a phase II trial of gefitinib, another small-molecular inhibitor of EGFR, in previously treated individuals with this malignancy demonstrated an objective response, while no objective responses were observed in the remainder of the patients in this study in the absence this mutation.4
Faulty Logic for Treatment Triplet
TAP Regimen in Endometrial Cancer
A phase III trial conducted by the Gynecologic Oncology Group compared the combination of cisplatin, doxorubicin, and paclitaxel (TAP) with the previous standard of care (cisplatin and doxorubicin) in recurrent or metastatic endometrial cancer.5 The study revealed an improvement in OS, but at the cost of considerable additional toxicity (similar to the observation of the FOLFOXIRI plus bevacizumab regimen in colon cancer).
However, the fundamental basis for the conduct of this trial must be challenged. First, it was known at the time of this study that single-agent paclitaxel was quite active in this malignancy.6,7 Second, in the closely related malignancy, epithelial ovarian cancer, the simpler and less toxic doublet combination of a platinum agent plus paclitaxel had improved OS, and there was no evidence that adding a third cytotoxic drug to a platinum-doublet further improved outcome in this setting.8
Therefore, while the statistically significant superior survival observed with the TAP regimen might conceivably have resulted from the more intensive and toxic three-drug program, it could simply have occurred due to the addition of pacli- taxel to the regimen.9,10
In fact, a subsequently conducted trial by the Gynecologic Oncology Group in the same clinical setting that directly compared the three-drug TAP regimen with the two-drug program of carboplatin/paclitaxel (the standard of care at this time in epithelial ovarian cancer) revealed the widely anticipated equivalent survival outcome between the two programs with considerably less toxicity associated with the administration of the plati- num doublet.11
Again, it must be asked: Was it acceptable that these patients were entered as research subjects into a trial that was conceptually so seriously flawed?
Maurie Markman, MD, editor-in-chief of OncologyLive, is president of Medicine & Science at Cancer Treatment Centers of America, and clinical professor of Medicine, Drexel University College of Medicine. maurie.markman@ctca-hope.com.