Saturday, 14 February 2015

ClinicSpeak & StatsSpeak: switching sideways or upwards

How actively do you want your MS to be managed? #ClinicSpeak #StatsSpeak #MSBlog #MSResearch

"Just as video killed the radio-star; orals DMTs are killing the injectable DMTs. The study below is essentially a trial of NEDA-2 (no clinically evident disease disease activity) and is being referred amongst UK neurologists as current best practice, i.e. you change therapy when someone breaks through clinically with a relapse. What this study is showing that a horizontal switch from IFN-beta to GA or GA to IFNbeta results in a worse outcome compared to escalation to fingolimod. This study uses real-life data collected as part of routine clinical practice and controls for bias by using a statistical method called propensity matching. The latter is about as good as it gets with real-life data. Ideally you would like this to be a clinical trial and randomise the subjects to two arms; i.e. anyone failing a 1st-line injectable clinically gets randomised horizontally to another class of injectable or vertically to the higher efficacy fingolimod. We already know that fingolimod is on average more effective than the injectables based on phase 3 trial data and its impact on end-organ damage (brain atrophy on MRI)."

"In the UK, the MS Society's early effective task force is currently designing a trial to test NEDA-2 vs. NEDA-3; i.e routine clinical practice using significant relapses to escalate treatment compared to vs. using active monitoring with frequent MRI to detect subclinical relapses to escalate vs. flipping the pyramid to start on a highly-effective treatment first. When I have presented the concept of our trial design to neurologists outside of the UK they have responded that the trial is unethical and would not be possible in their country because this is how they manage MS already."

"Please note that in the UK escalating on breakthrough relapses is controversial; many neurologists only offer a switch if they relapse is disabling. In addition, many relapses are not reported by MSers and therefore not taken into account when making a decision about switching therapies. This is why you have to monitor your relapses and why self-monitoring apps and tools are becoming increasingly important. I personally don't think there is any biological difference between mild, moderate and severe MS relapses; what they are all telling you is that your disease is active and that if you are on a DMT is not doing its job. What determines if a MS lesion causes a relapse is its location and size; a small lesion in an ineloquent site of the brain will be asymptomatic, however the same size lesion in the brainstem could cause a devastating relapse that is life threatening. We need to get away from the idea that any disease activity is acceptable; this is why we have a zero tolerance strategy or ZeTo."

"The following are a few slides to illustrate our shifting practice and how we have stolen ideas from our colleagues in other specialities; the example I give is how the gastroenterologists manage inflammatory bowel disease."

StatsSpeak: In the statistical analysis of observational data, propensity score matching (PSM) is a statistical matching technique that attempts to estimate the effect of a treatment, policy, or other intervention by accounting for the covariates that predict receiving the treatment. 

IMPORTANCE: After MS relapse while an MSer is receiving an injectable disease-modifying drug, many physicians advocate therapy switch, but the relative effectiveness of different switch decisions is often uncertain.

OBJECTIVE: To compare the effect of the oral immunomodulator fingolimod with that of all injectable immunomodulators (interferons or glatiramer acetate) on relapse rate, disability, and treatment persistence in MSers with active MS.

DESIGN, SETTING, AND PARTICIPANTS: Matched retrospective analysis of data collected prospectively from MSBase, an international, observational cohort study. The MSBase cohort represents a population of MSers monitored at large MS centers. The analyzed data were collected between July 1996 and April 2014. Participants included RRMSers who were switching therapy to fingolimod or injectable immunomodulators up to 12 months after on-treatment clinical disease activity (relapse or progression of disability), matched on demographic and clinical variables. Median follow-up duration was 13.1 months (range, 3-80). Indication and attrition bias were controlled with propensity score matching and pairwise censoring, respectively. Head-to-head analyses of relapse and disability outcomes used paired, weighted, negative binomial models or frailty proportional hazards models adjusted for magnetic resonance imaging variables. Sensitivity analyses were conducted.

EXPOSURES: MSers had received fingolimod, interferon beta, or glatiramer acetate for a minimum of 3 months following a switch of immunomodulatory therapy.

MAIN OUTCOMES AND MEASURES:  Annualized relapse rate and proportion of relapse-free MSers, as well as the proportion of MSers without sustained disability progression.

RESULTS: Overall, 379 MSers in the injectable group were matched to 148 MSers in the fingolimod group. The fingolimod group had a lower mean annualized relapse rate (0.31 vs 0.42; 95% CI, 0.02-0.19; P = .009), lower hazard of first on-treatment relapse (hazard ratio [HR], 0.74; 95% CI, 0.56-0.98; P = .04), lower hazard of disability progression (HR, 0.53; 95% CI, 0.31-0.91; P = .02), higher rate of disability regression (HR, 2.0; 95% CI, 1.2-3.3; P = .005), and lower hazard of treatment discontinuation (HR, 0.55; P = .04) compared with the injectable group.

CONCLUSIONS AND RELEVANCE: Switching from injectable immunomodulators to fingolimod is associated with fewer relapses, more favorable disability outcomes, and greater treatment persistence compared with switching to another injectable preparation following on-treatment activity of MS.

CoI: multiple


  1. Prof G,

    Thanks. Are we any nearer to establishing whether highly effective treatments early on will prevent the onset of progressive disease / disability later on? I, and I suspect many others, would opt for early highly effective treatments if the longer terms benefits were guaranteed. However, I don't really want to take the risks associated with BMT or other chemo agents, or with Tysabri (PML) if I'll still end up SPMS in 10-15 years time. What happens in other diseases where they treat early and with highly effective treatments e.g. what happens in RA?

    1. Re: "Are we any nearer to establishing whether highly effective treatments early on will prevent the onset of progressive disease / disability later on?"

      The data is clear in my mind that they delay the onset. Whether they prevent it is another story and will take many years to establish. We and others have been encouraging Alasdair Coles to deep phenotype some of the highly-active RRMSers who are in long-term remission (>10 years) post alemtuzumab to see if there is any evidence of ongoing activity or tissue damage. If there is no ongoing MS-related tissue damage this would go a long way to showing that early induction therapy is working. Please note I say MS-related; unfortunately, ageing is a process that we can't escape at present and MSers may age quicker than normals due to a loss of reserve capacity as a result of inflammatory damage.

    2. Re: "What happens in other diseases where they treat early and with highly effective treatments e.g. what happens in RA?"

      The need for joint replacements in RAers is plummeting. The following paper in 2011 reports a 50% decline and with newer more aggressive approaches in contemporary cohorts of RAers these figures continue to improve.

    3. Prof G, I get the impression from the above that progression is Age-related due to loss of reserve capacity due to inflammation damage as opposed to direct MS related damage.

      Is this different in PPMS when there is progression without loss of reserve capacity.?


    4. Prof G,

      What do you mean by deep phenotype? I get an annual MRI at Cambridge and results to date show no inflammation or new lesions. Is there a more sensitive method for checking whether the disease is inactive?

  2. This is weird... either we believe that MRI activity is a true measure of disease activity in MS - and then we should "treat to target" until there is no disease activity. Or we believe that MRI activity in not a true measure of disease activity in MS (= currently available treatments are "cosmetic treatments" only and just reduce the number of lesions) - and then we can forget it and be happy treating clinical relapses.
    As the ratio of MRI activity to clinical relapses is about 10 T2 MRI lesions:1 clinical relapse, there is a lot more going on on MRI than can be clinically detected. So do UK neurologists believe in MRI or don't they?

  3. I'm not in the Uk, but I am British. My MS has been disabling from day 1, almost 7 years ago. I have had Rebif (did nothing to slow relapses), Tysabri (had an allergic reaction so stopped), Gilenya and now Tecfidera. From what I read, this wouldn't have necessarily have happened in the UK.

    I'm also confused by the post above, in that I thought it was established that MRIs do not show the whole picture, in that they only show inflammatory disease activity. I have a relatively low lesion load for my level of disability. My disease activity is now at a level not measured by MRIs. I don't think it's a question of believing in MRIs, it's a question of understanding their limitations.

    1. I have high lesion load from day one and no clinical progression(but MRI progression) for many years now.
      There is the question of where the lesions are ( spinal cord or brain) and how old the MSer is. Complicated stuff.

    2. Sure is, and mine are spine and cervical spine, as well as brain. Lots of space in the brain so you can lots of lesions that don't "do" anything...but the spine...kerpow! (Technical ms tem... Feel free to use it, ProfG...)

  4. "We already know that fingolimod is on average more effective than the injectables based on phase 3 trial data and its impact on end-organ damage (brain atrophy on MRI)."

    Could you please provide a reference for this? In particular I would like to see the data for a head to head trial of fingolomod vs. Copaxone with atrophy as the endpoint being measured.

  5. Re: "Could you please provide a reference for this?"

    No reference as I am also not aware of phase 3 head-2-head of GA vs. Fingo. However, I am aware of several head-2-head studies of GA against IFNbeta the last being the CombiRx study (Avonex vs. GA: and Fingo vs. Avonex in the Transforms study ( The data that GA normalises brain atrophy loss in MSers is very poor. But you can interpret it as you wish. The field is moving on from individual agents to treatment strategies; it is all about horses-for-courses and treat-2-target. If you not responding to agent X you move to agent Y; the question is how long you decide to stay on agent X and the frequency of monitoring. In addition, some MSers prefer to load the dice in their favour by flipping the pyramid and going for the agents that are more likely to render them NEDA up front. It is all about choice.

  6. Prof. G isn't this study actually a type of head-2-head study between Fingo and GA?

    1. Re: "Prof. G isn't this study actually a type of head-2-head study between Fingo and GA?"

      Yes, this study is comparing fingo with GA. However, it is not a randomised double-blind study hence there may be biases that may affect the results. We simply use the propensity score to try and account for these biases. But when we assess to relative strengths of two drugs against each other we try and look at the bigger picture and include all the data in our assessment. Please note that the trials only give the average response and within the data sets there people who do better and worse than the average. What we really need are predictive markers that will allow us to make a call up front about your chances of responding to a drug or not. For example, I have a large number of patients doing very well long-term on GA, I simply wish we can identify them early.


Please note that all comments are moderated and any personal or marketing-related submissions will not be shown.