I was very interested to read the article by Wallace et al. outlining their experience on radiotherapy (RT) for pleomorphic adenoma (PA). Although this report reveals useful information, there are some areas for discussion on which I would like to expand.

First, the investigators designed and implemented a retrospective cohort study comparing RT versus RT plus surgery. It can be implied that the predictor study variable was surgical intervention (ie, yes or no). Had the authors demonstrated statistical significance between the 2 groups of the cohort, it would have been able to conclude that “surgery as an adjunct to RT improves the clinical outcomes of RT”. If the study’s purpose is to address the following question: “Among patients having PA removed, is RT effective in controlling the recurrence of PA after surgery?” (as in their conclusion), the investigators should enroll a cohort of subjects/patients who underwent surgery with versus without RT. The alternative, 1-tail hypothesis will be that recurrence is lower in the surgery plus RT group than in the surgery alone group. By this reason, bias of the general interpretation of the results will be minimized.

Second, the study’s results are indeed inconclusive. The authors covered most of the issues associated with tumor treatment outcome. However, any impact seems to be lost due to the inability to make any significant difference between the 2 groups of the cohort (RT versus RT plus surgery). The Kaplan-Meier survival curve does not represent any statistical significance of the comparison. It is known that many studies on oral and facial diseases have no statistical analyses, even though it is possible and informative to perform them. Many reports also lack basic univariate analyses, so that it is unclear to whom the results may be generalizable, or the analyses do not fully relate to the hypotheses or the study design .

Third, it is generally accepted that observational studies do not control the exposure (intervention). Causal interference made in this study type is complicated by the contamination of unknown or unmeasured confounding variables. The findings based on a small sample size should also be interpreted cautiously . Besides, small studies or subgroups are often underpowered, namely suffering from “false-negative or type II error”. To detect a 50% to 80% relative efficacy benefit of one treatment over another, sample size in controlled therapeutic trials should range from 50 to 200 .

Fourth, as the authors also reminded us, surgical excision is the treatment of choice for PA . RT alone is not recommended for PA because it does not reduce recurrence rates, a recurrence-free interval and the size of safety margins needed intraoperatively. Irradiation can result in tissue ischemia/fibrosis, oral and dental tissue changes, osteoradionecrosis, and retardation of the facial nerve recovery . Some studies have shown the radioresistance of PA .

Moreover, different surgical modalities contribute to different treatment outcome. Two meta-analyses revealed that enucleation and extracapsular dissection of PA of the parotid gland elicited 9 and 10 times higher rates of recurrence compared with superficial parotidectomy (SP), respectively . In the Wallace et al’s study, various surgical procedures were mixed together: 12 SP, 4 total parotidectomy and 1 enucleation. This may skew the analysis and study results.

Fifth, the authors concluded that “Patients with pleomorphic adenoma who present with positive margins or multinodular recurrence benefit from treatment with adjuvant RT”. However, they did not present or analyze any data on multinodular recurrence. This also requires further explanation.

Lastly, the authors cited that the study by Makeieff et al. was conducted at “Le Centre Hospitalier Universitaire (Montpellier, France)”. In fact, “Le Centre Hospitalier Universitaire” can be literally translated into “the University Hospital Center” (“Le” for masculine words in French means “the” in English). The readers may misunderstand that “Le” is the hospital’s name. The words “Le Centre Hospitalier Universitaire (Montpellier, France)” should be read as “the University Hospital Center of Montpellier, France”.

Taken together, Wallace et al. presented an excellent analysis and interesting information on RT for PA, but their study results need to be interpreted with caution. Outcome research requires more attention in this evidence-based era, as it would influence systematic reviews and meta-analysis in the future.