Perspective on consensus studies in the absence of prospective trials

Prospective multicenter trials have become the Holy Grail of medical care. By providing natural history information as well as response to treatment we often find “What we knew” to be unture. Unfortunately true prospective randomized trials of adequate power are extremely uncommon. Nonetheless, the push to true “evidence based medicine” often points out how little we truly know about common disease processes. The lack of evidence based data, however, does not mean that we are not responsible for making appropriate clinical decisions based on analysis of risk benefit ratio with the best material we have available. That is why consensus studies such as “Evidence-based review of recommendations for visual function testing in patients treated with Vigabatrin” remain important in advising the best clinic practice in the absence of true multi-centered trials. There is of course the possibility that substituting “eminence based “for “evidence based” medicine can be seen as equivalent or at least of equal weight. That unfortunately is not true since even experts are limited by their experience.

This consensus study raises some important issues when trying to sort out the value of “eminence based” consensus reports. Since the early introduction of Vigabatrin in the early 1990s, it is clear that patients treated with this medication can develop evidence of toxicity to the retina. Although through the first decade of use there were only isolated clinical reports, subsequent studies as noted by the authors suggest this effect is far more common. This drug obviously should be limited to those patients who have no other alternatives. Vigabatrin clinically produces constricted visual fields and ERG changes. Multiple studies have suggested that the longer the duration and dose exposure, the higher the frequency of involvement. Although there has been some variability in studies, it is strongly suggested that the toxic effects on the retina do not reverse with discontinuation of the medication. Thus the afferent visual system effects (constricted visual fields and likely diffuse effect on the visual field) should not be regarded as potential toxicity but likely a consequence if patients are treated with high enough doses long enough. At this time we have no way of predicting which patients may be immune. What is still unknown is how significant these visual field defects are likely to be. The authors point out that we have little evidence that this has yet to produce “functional” impairment of the visual field. Since this has not been studied prospectively, it is not surprising that it really is impossible for us to predict how often patients treated with vigabatrin are eventually going to develop visual field defects dense enough to interfere with activities of daily living. This has been further compromised by the failure to specifically ask about visual symptoms in most studies even when documenting toxicity. Since even in common optic neuropathies where visual field defects have been well studied such as glaucoma, quantitative assessment of interference with daily function has been very difficult, and only over the last several years have attempts been made to assess the functional aspects of a diagnosis of glaucoma and its effects on the performance of tasks required in daily living and quality of life.

There are several issues brought up in this consensus report that I think do bear scrutiny. While it is hard to argue about the importance of establishing a baseline of quantitative assessment of visual function in patients who are likely to have toxic effects if treated long enough, the timing of establishing this baseline might be questioned. Since toxicity does not appear to occur in any significant numbers earlier than a year, insisting on baseline evaluation prior to any medication is probably unnecessary. The second question revolves around the most sensitive means of assessing afferent dysfunction. The authors correctly point out that there is limited quantitative data on the use of automated static perimetry for studying peripheral visual fields. Other studies not cited by the authors including work by Lars Frisen who noted that suprathreshold microdot perimetry (rarebit) may pick up earlier abnormalities in patients on vigabatrin, suggests that there may well be other means of assessing afferent visual function in these patients.¹ Other innovative techniques have been suggested to test kinetic visual fields in young children.² None of these however have yet to be vetted. One of the natural tendencies in consensus reports like this is to be overly optimistic about techniques that are yet to be proven. One particular instance in this report is the suggestion that automated kinetic perimetry, now available on the Octopus system and in the Humphries, may be effective in studying these patients. To date, none of these have been investigated in any sort of controlled fashion, and should be considered of theoretical value. Similarly the authors are far too pessimistic on the role of automated perimetry in young (less than 10 year old) patients. Our current computer literate “Nintendo” generation often performs extremely well on standard automated perimetry at a very young age.

A third aspect of this report is the suggestion that ERG should be a substitute for patients unable to do perimetry. While there is good evidence that ERG abnormalities also occur in these patients with increasing dose and correlation, the correlation between ERG abnormalities and the extent of visual field defects remains to be elucidated. As the authors do point out, ERG, particularly in children, requires sedation, is difficult to interpret, and thus carries its own set of risks. While other studies such as OCT may also be found to be useful at this time there is simply not enough data to make any recommendations.

Ultimately, the most important issue is the importance of a frank discussion of risk benefit ratio. For the patients who have no other therapy, the potential of allowing them to lead a life, either free of or with reduction in otherwise uncontrollable seizure activity, needs to be balanced against the known potential risk to the afferent system, specifically visual field dysfunction. None of these drugs should be started until this discussion is held with the patients and their families and if a beneficial effect is not realized they should be discontinued. Since toxicity is essentially unheard of during a short course a one month test should determine if the drug might play a therapeutic role. The role of screening in these patients (and thus any implications for how often this should be done) involves simply having a repeat discussion of the risk benefit ratio. At this time, with the absence of more quantitative data, and in particular without the ability to reverse the effects of these medications, it is that very discussion of risk benefit ratio that underlies the use of these medications. Thus any specific comments about the timing of screening have to do more with the interaction of the individual patient and their physician and whether knowing about early toxicity will change therapeutic plans. At this time I do not believe we have any data to suggest specific timing of repeat visual field testing in patients who are responding to the medication. Until such time as we have better medications with less of a toxic profile, the art of medicine will include the interaction with patients so that they can best appreciate the advantages and disadvantages of any therapeutic program.

Declaration of interest: The author reports no conflicts of interest. The author alone is responsible for the content and writing of the paper.