An ethical dilemma: a randomised switch study from natalizumab to interferon-beta

It is about time we took an ethical stand on underpowered studies? #MSBlog #MSResearch

“The study below aims to look at satisfaction and adherence to treatment in MSers on natalizumab, who are at high risk of PML, being randomised to either stay on natalizumab or switch to interferon-beta-1b. This is a very small study (19 MSers); I am sure it is not large enough to answers the questions it set out to answer. If you were high risk of getting PML would you necessarily want to stay on natalizumab? What would your anxiety levels be if you couldn’t be switched, compared to those who did switch? In our centre this study would not have got ethical approval; it is underpowered and the statistician would have comeback and said no.”

“A study is underpowered if there are too few subjects in it to give it a reasonable chance of providing a definitive answer. In the paper the author’s acknowledge this and state ‘Given the small samples size we applied a correction for Type I error using for each neuropsychological end point an adjustment for multiple comparisons following the Bonferroni method‘. In 2014 should we be doing underpowered studies? Is it ethical to randomise MSers to studies that are unlikely to give an answer? Will this study change clinical practice?”

“I have always justified the use of underpowered studies when the purpose is to allow trainees to learn research techniques and the scientific method on existing samples and data. In other words opportunistic research that does not expose study participants to any risks. It is, however, very different when the research involves clinical trials of investigational medicinal products (CTIMPs) that require informed consent. CTMPS come with risks. Can you imagine the ethical fall-out if one of the MSers who was randomised to stay on natalizumab in this study developed PML?”

Epub: Zecca et al. Treatment satisfaction, adherence and behavioral assessment in patients de – escalating from natalizumab to interferon beta. BMC Neurol. 2014 Feb 28;14(1):38.

BACKGROUND: De-escalating natalizumab (NTZ) to interferon beta 1b (IFN B 1B) is a possible treatment option in MSer interrupting NTZ because of increased risk of progressive multifocal leukoencephalopathy (PML). 

AIM: The aim of this study was to evaluate satisfaction and adherence to treatment, behavioral and fatigue changes in MSers switched to IFN B 1B compared to continued NTZ treatment.

METHODS: A 1 year, prospective, randomized, rater-blinded, parallel-group study. Nineteen relapsing remitting (RR) MSers, randomly assigned to undergo either NTZ (n = 10) or IFN B 1B (n = 9) treatment, who had previously received NTZ for at least 12 months with disease stability and fearing or at risk for PML were included. MSers underwent behavioral and treatment assessments at baseline, after 24-week and 1 year follow-up. Behavioral assessment included measures of cognition, fatigue and quality of life. Treatment assessment included measures of satisfaction, persistence and adherence to treatment. Clinical-radiological disease activity and safety were also assessed.

RESULTS: Baseline characteristics of MSers were similar between groups except for Euro Quality Visual Analogue Scale, being higher in the NTZ group (p = 0.04). Within-group comparisons at the three time points, as well as interaction analysis of treatment effect over time did not show any statistically significant differences in behavioral or treatment assessments, but a coherent trend favoring NTZ over IFN B 1B.

CONCLUSIONS: De-escalating NTZ to IFN B 1B is feasible and associated with overall good MSers related outcome and persistently stable behavioral measures.

CoI: multiple

About the author

Prof G

Professor of Neurology, Barts & The London. MS & Preventive Neurology thinker, blogger, runner, vegetable gardener, husband, father, cook and wine & food lover.


  • I think a study like this would have been more useful if they tested at least two different drugs that the high risk Tysabri users transitioned to instead of keeping one set on tysabri. I think it is well established that the longer you stay on Tysabri the higher the risk is for PML. The thing that is most needed is which is the best therapy to switch to.

    • Is it well established "the longer you stay on Tysabri the higher the risk is for PML?" I have seen total risk charts which suggest this. (made up numbers coming)Meaning if I have a one in four hundred risk in year one and a one in three hundred risk in year two and a one in four hundred risk again in year three, then my risk for PML has gone up every year because each year I have opted to stay on the drug incurs some risk. However, if at the end of year one I have not had PML, then my risk of getting it in year two declines. (back to real life) The most recent data I have seen is very thin for those who have been on Tysabri for the longest time. I have been on it for 7 years, and it seems every year they come out with stats on the odds for the year I just finished. What I always want to see is something akin to an actuarial chart like those used by life insurance agents.

    • I will let the ProfGs speak but statistical power analysis was performed to indicate how many people would be needed to see an effect x based on MRI. This is the number planned to be investigated. Whether the effect sizes were realistic will be need to be seen. Unfortunately with academic led studies cost is an issue and this may affect the effect sizes selected but once this is selected then a statistician works out the rest.

    • Raltegravir study is boot strap MRI study based on a published peer-reviewed power calculations, The ethics committee had no problem with the power calculations. It is not a randomised controlled-trial. You can't compare these studies. This study does not even mention power calculations in its methods section. All I am asking is it ethical to do CTIMPS in studies that are under powered?

    • What is the effect size you are expecting to see? Is it ethical to do non controlled studies when a definite conclusion will never be reached from the inspire trial? I have looked at a lot of your papers and can not see power calculations in any of them.

By Prof G



Recent Posts

Recent Comments