STAR-MS Study: Predicting MS using a single 3T MRI scan

  • Research type

    Research Study

  • Full title

    STAR-MS Study: Single Test to ARrive at MS diagnosis. A prospective, longitudinal, investigator blinded, pilot study assessing the accuracy of a single 3 Tesla MRI brain scan in predicting multiple sclerosis in cases of diagnostic uncertainty.

  • IRAS ID

    155731

  • Contact name

    Nikos Evangelou

  • Contact email

    nikos.evangelou@nottingham.ac.uk

  • Sponsor organisation

    University of Nottingham

  • Duration of Study in the UK

    2 years, 0 months, 1 days

  • Research summary

    Summary of Research

    There is no single, simple test to differentiate Multiple Sclerosis (MS) from conditions that mimic it. Apart from the patient's symptoms, doctors use a brain scan (MRI) to see if there are abnormalities which are consistent with MS. In patients without typical symptoms or a typical MRI appearance, a firm diagnosis cannot be made as other neurological illnesses can mimic symptoms of MS and the MRI scans can look very similar. Further tests are often required such as lumbar punctures, resulting in delays, discomfort for the patient and additional costs to the NHS. With time the diagnosis can become clearer as patients may develop other symptoms of MS but this may take months or years, causing uncertainty.

    The most common errors are from misinterpretation of the brain abnormalities or 'lesions' seen on the MRI scan. Subsequently if a patient is misdiagnosed with MS they may receive treatment they do not need. Furthermore a delay in a firm diagnosis delays treatment for another condition.

    Pathologically (when looking at lesions using a microscope) MS lesions usually have a vein running through the centre, whereas in lesions arising from other conditions, no central vein is seen. We can therefore distinguish between patients with MS and patients without it.

    We aim to test the accuracy of a new, simple MRI scan that allows us to see the presence or absence of a vein within brain lesions, therefore improving the ability to predict MS in patients where there is uncertainty.

    Patients will only have one brain scan and then be followed up by their neurologist, who will confirm the final diagnosis. We shall then look back at the original scans to see if those with MS had veins within their lesions and if those without MS had lesions without veins.

    Summary of Results

    Thirty-eight patients (27 females) with a mean age of
    50 years (standard deviation (SD) = 15) were
    recruited. Data from one patient were excluded from
    the analysis due to excessive movement artefacts on
    the MR images. At the time of the analysis, 35 patients
    had been given a diagnosis by their neurologist.
    Median time between the index test (CV assessment)
    and the reference standard (clinical diagnosis) was
    equal to 6.5 months (range: 0–41 months). Twelve
    patients were subsequently clinically diagnosed with
    MS, according to the latest available McDonald criteria,
    and 23 patients were diagnosed with non-MS.
    Confirmed MS diagnoses were usually based on new
    typical relapses, radiological progression or supportive
    lumbar puncture results (see Table 1). The most
    common non-MS diagnosis was small vessel disease
    (SVD) (13 patients) (see Figure 1 for a summary of all
    the diagnoses). Median time between the index test
    and the last follow-up visit was 20 months (range:
    1–49 months). Patients experienced no adverse events
    as a result of the index test or the reference standard.
    Patients eventually diagnosed as MS had minimal to
    no disability at the time of their index test.
    Since the start of the study in May 2015, 57 patients
    have been approached by their neurologist about the
    study. Out of those, 38 patients agreed to take part and
    were subsequently scanned. Reasons for not participating
    included no contact after an initially positive
    response (six patients), unwillingness or inability to
    have another MRI scan (reasons included being overweight,
    problems with walking and so on; four
    patients), being too busy (four patients), incompatibility
    with study dates (three patients) and a change of
    mind (three patients). Rate of recruitment for the first
    year of the study (May 2015 to April 2016) was two
    patients a month with a screen failure rate of 0.25,
    based on 24 patients out of a total of 32 whom we
    approached about the study that year.
    In total, 785 WMLs were analysed. Patients subsequently
    diagnosed with MS had a median of 20
    WMLs (range: 3–71) while non-MS individuals had a
    median of 21 WMLs (range: 2–63) (see Figure 2 for
    the typical appearance of lesions in MS and non-MS
    patients). The proportion of all the lesions with CVs
    was 51% (standard error (SE) = 3%) in the MS group
    and 28% (SE = 2%) in the non-MS group. Figure 3
    demonstrates the distribution of lesions with CVs in
    the MS and non-MS groups. In the non-MS cohort,
    causative factors for the WMLs included smoking,
    migraine, hypertension and in one case post-radiotherapy
    changes.
    The clinical diagnoses were compared to the presence
    of the CV sign and the ideal threshold for a diagnosis
    using the CV sign was calculated. Youden’s J statistic
    was equal to 0.74 when associated with a threshold of
    >40.7% WMLs with CVs. This resulted in sensitivity
    of 100% (95% confidence interval (CI): 73.5%–
    100%), specificity of 73.9% (95% CI: 51.6%–89.8%)
    and positive and negative predictive values equal to
    66.7% (95% CI: 50.1%–79.9%) and 100%, respectively.
    The area under the curve (AUC) was equal to
    0.86, p < 0.001 (see Figure 4). We achieved 82.9.7%
    agreement with the clinical diagnoses and inter-rater
    agreement was 73%. The sensitivity and specificity of
    the diagnoses according to the rule of six were 91.7%
    (95% CI: 61.5%–99.8%) and 52.2% (95% CI: 30.6%–
    73.2%), respectively. A cross-tabulation of the results
    from the index and the reference tests, including
    demographic and paraclinical data, are found in Table
    1. There were no statistically significant differences
    in age (p = 0.139) and sex (p = 0.92) between patients
    with and without an MS diagnosis according to the
    CV sign.

  • REC name

    North West - Greater Manchester East Research Ethics Committee

  • REC reference

    15/NW/0286

  • Date of REC Opinion

    15 Apr 2015

  • REC opinion

    Further Information Favourable Opinion