Regional Hyperthermia With Neoadjuvant Chemotherapy for Treatment of Soft Tissue Sarcoma

To the Editor The report by Issels et al on long-term survival in the randomized clinical trial of regional hyperthermia (RHT) plus neoadjuvant chemotherapy vs neoadjuvant chemotherapy alone for treatment of soft tissue sarcoma is misleading. This report is a post hoc analysis of the initial study results, which showed no benefit in overall survival (OS) associated with RHT (hazard ratio [HR], 0.88; 95% CI, 0.64-1.21; P = .43). Issels et al reported the result of a competing-risks analysis that indicated that the specific risk of death from sarcoma was significantly lower after RHT (HR, 0.73; 95% CI, 0.54-0.98; P = .04), though the reported number of non–sarcoma-related deaths suggests that this risk was significantly higher after RHT (15 vs 6 deaths; odds ratio, 2.76; 95% CI, 1.04-7.29; P = .04 by 2-tailed χ² test). In general, there is no gain in OS associated with neoadjuvant therapy with RHT. Instead of reporting the result as a secondary analysis, the authors presented it as the primary result of the study. They did not mention that there was no actual benefit in OS, presented the competing risks-based sarcoma-specific survival (SSS) as the initial study end point, and omitted the significant increase in the non–sarcoma-related mortality. Moreover, in the recent study there is a risk for selection bias because the authors excluded 12 patients apparently owing to withdrawal of consent or metastatic disease, though this exclusion was not included in the previous report after 3 years of follow-up. Interestingly, the authors stated the additional reason for the exclusion in the supplement: these patients did not start their allocated treatment after randomization. This additional reason for exclusion is confusing because it appears that 7 patients were excluded from the RHT arm in the most recent article, whereas the earlier study stated that only 4 did not start treatment. The recent study result is presented as a robust intention-to-treat analysis, though after the exclusions, this is a less reliable per-protocol analysis. Furthermore, the SSS is confounded by fewer cycles of chemotherapy in the control arm (median, 5 cycles) than in the RHT arm (median, 8 cycles), and apparently no adjustment for the confounding was applied. Other possible confounders, including the end point substitution, the competing risks-based nature of the end point, and selection bias resulting from the excluded patients, were not properly addressed. Thus, the conclusion that the advantage in SSS is owing to the longer follow-up may not be justified. In my opinion, this report does not show an advantage of RHT plus neoadjuvant chemotherapy vs neoadjuvant chemotherapy alone. I highlight these issues to prevent possible misinterpretation of this trial's results as positive and to avoid unproven treatments being used in clinical practice.

https://ift.tt/2BFQbBj