MRI findings in blinded trials should be available to treating physicians – No

In multiple sclerosis (MS) clinical trials, clinical and imaging efficacy outcomes generally are assessed by raters who do not have access to clinical information, so as to preserve blinding. Separate treating physicians are responsible for clinical care, including symptom management; initial assessment of suspected relapses and treatment of protocol-defined relapses; and assessment of adverse effects. The clinical trial team has the important job of maintaining the scientific validity of the trial, but not at the cost of the ethical obligation to each individual patient. The Declaration of Helsinki states that “considerations related to the well-being of the human subject should take precedence over the interests of science and society,” and physician researchers are ethically obligated to provide available effective treatments at the same standards than those not involved in research. ¹ Magnetic resonance imaging (MRI) is an indispensable tool in MS diagnosis and monitoring of treatment efficacy, both in clinical practice and clinical trials. The current debate concerns whether treating physicians should have ongoing access to MRI results during a trial to fulfill their ethical responsibility to provide appropriate clinical care. We argue that providing treating physicians access to MRI results is both unnecessary and inadvisable.

The question is whether knowledge and treatment of clinically silent MRI lesion activity is necessary, particularly if the activity is modest. Several studies have shown that targeting MRI “silence” does not improve long-term outcomes in MS. In a recent study, for patients who were inactive clinically, the development of new or enlarging T2 lesions over 2 years was not associated with subsequent clinical worsening over a 10-year period, as measured by Expanded Disability Status Scale (EDSS), Timed 25-Foot Walk, 9-Hole Peg Test, or Paced Auditory Serial Addition Test. ² In another study, patients with MRI lesion activity who changed disease-modifying therapy (DMT) fared no better clinically or radiographically at 18 months compared to those with MRI lesion activity who did not change DMT. ³ In a third study, modest MRI activity (<2 gadolinium-enhancing lesions, <3 new T2 lesions) was not associated with EDSS worsening over on average 6.7-year follow-up. ⁴ Thus, a low level of MRI lesion activity might not warrant intervention.

However, the importance of blinding and allocation concealment of both trial participants and treating physicians to the validity of the trial is irrefutable. On average, randomized trials that have not used appropriate levels of blinding are more likely to report favorable outcomes for the test drug and to show larger treatment effects compared to trials with effective blinding.^5–7 In fact, investigator blinding to treatment assignment has been shown specifically in MS clinical trials to prevent erroneous conclusions about treatment efficacy. ⁸ The importance of blinding also is recognized in the CONSORT Statement adopted by key medical journals and editorial groups, which urges clinical trial publications to report “how the success of blinding was evaluated.” ⁹ Due to this need for blinding, the roles of clinical management of trial participants by treating physicians and evaluation of outcome measures (clinical rating scales, neuroperformance testing, and quantitative assessment of MRI findings) by blinded raters are generally separated in MS clinical trials. However, treating physicians remain involved in the evaluation of certain outcomes, specifically relapses and adverse events.

Because the magnitude and time course of effects on MRI vary between DMTs, providing treating physicians access to MRI data throughout the study could inform them of participant assignment to a treatment expected to have less potent or delayed efficacy on MRI. This potential unblinding could bias efficacy outcome assessments that are affected by the treating clinician (e.g. whether a participant is brought in for assessment of symptoms of a potential relapse or whether the relapse is confirmed in trials in which the treating physician is responsible for adjudicating potential relapses). Potential unblinding of treatment assignment also might affect assessment of adverse effects, especially assignment of causality.

In addition to concerns regarding bias when treatment allocation is unblinded, providing MRI data to treating physicians throughout the trial might lead to interventions for participants with clinically insignificant lesions. For example, corticosteroid use in MS clinical trials generally should be restricted to treatment of protocol-defined confirmed relapses. The concern is that asymptomatic MRI lesion activity might trigger administration of corticosteroids, which could affect clinical, participant-reported, and MRI outcomes, as well as adverse events. Similarly, the presence of clinically silent MRI lesion activity might lead to discontinuation of randomized treatment or discontinuation from the trial. Differential corticosteroid use, treatment discontinuations, or study discontinuations could spuriously alter study results and conceivably might bias in favor of a less efficacious treatment.

We appreciate that substantial MRI lesion activity, even if asymptomatic, might warrant intervention with a short course of corticosteroids and/or change in disease therapy. Therefore, safeguards need to be in place in trials to detect substantial MRI activity and the development of pathology other than MS. Many institutions, including ours, require local review of study MRIs, even in blinded trials with a central MRI reading center. We suggest that treating physicians receive MRI reports with a standardized (protocol-specific) format that includes the presence of findings consistent with the known diagnosis of MS and the presence/absence of findings indicating conditions other than MS. The report either could be silent regarding MS-related MRI lesion activity or could comment on the presence/absence of lesion activity exceeding a pre-defined threshold. Alternatively, the study Medical Monitor or MRI Reading Center could notify the site of lesion activity exceeding a pre-defined threshold, though this approach creates some logistical difficulties. Periodic assessment by the Data Safety Monitoring Board (DSMB) of major group-level differences in MRI activity is an additional safeguard, though DSMBs generally review data intermittently.

In conclusion, treating physicians should not have access to MRI results in clinical trials, as it is not necessary for participant safety and care and could significantly bias trial outcomes. It is possible to implement procedures to notify treating physicians of clinically relevant MRI lesion activity, which provides adequate safeguards for participants while protecting the scientific validity of the trial.