Commentary

Oral leukoplakia is a diagnosis given to a white patch that cannot be categorised. Once a histological diagnosis is made, it is useful to refer to it either by causal factor, for example, candidal leukoplakia, or by degree of dysplasia.1 An international meeting clarifying these definitions reported its findings and also suggested a method of staging these lesions.1 This has been further commented upon by Van der Waal and Axell2 and Schepman and van der Waal3 and it is a pity that this study did not adopt this methodology. This staging not only includes the different forms of dysplasia but also takes into account the size of the lesion.

A recent Cochrane systematic review on treatment of leukoplakia also provides some guidelines for future RCT in this field.4 In their paper, Lodi et al5 point out that no RCT conducted to date exceeds 15 months and yet there is evidence to show that malignant transformation increases with duration of follow-up.6 They also point out that many researchers use outcomes other than malignant transformation, for example, histological diagnosis or resolution of lesion. There are problems with using these outcomes because there is little evidence for their predictive value and it has been shown that outcomes such as dysplasia are subject to high observer variation.7, 8 Many journals have now adopted the CONSORT guidelines for reporting of RCT: this study would have been easier to follow if it had used this format.

There are many omissions in the methodology that affect the reader's ability to interpret the study. The following data are missing: how the trial population was selected; how many people did not consent to the trial; how well-matched the groups were in terms of dysplasia (the control group had no severe dysplastic lesions); what the tobacco use of the population was; what method of randomisation was used; how clinicians and patients were blinded; how many pathologists were involved; how many subjects had biopsies taken pre- and post-treatment; how the site of biopsy was determined, especially if the lesion had disappeared; how the power of the study was determined; how complete the follow-up was; and, finally, how many patients did not complete the full protocol.

The authors point out that many of these lesions are related to smoking or chewing tobacco and yet no details of these habits are provided, either at onset of the study or at its conclusion. Gupta et al9 showed how lesions can disappear after merely cessation of tobacco use. The groups should have been compared using this parameter and the level of dysplasia.

The results also do not specify at which timepoint they were taken, whether at the end of the active use of the drug or at the end of the follow-up period. If the latter, were there any regressions? Data about side effects and toxicity are only provided in the discussion and these are said to be non-existent.

Owing to these problems, it is impossible to draw any firm conclusions from this study in the form in which it is currently reported. It seems feasible to carry out further RCT with lycopene, however, using the criteria suggested by Lodi et al,5 the staging system and CONSORT guidelines.