MRI biomarker assessment of neuromuscular disease progression: a prospective observational cohort study

Summary Background A substantial impediment to progress in trials of new therapies in neuromuscular disorders is the absence of responsive outcome measures that correlate with patient functional deficits and are sensitive to early disease processes. Irrespective of the primary molecular defect, neuromuscular disorder pathological processes include disturbance of intramuscular water distribution followed by intramuscular fat accumulation, both quantifiable by MRI. In pathologically distinct neuromuscular disorders, we aimed to determine the comparative responsiveness of MRI outcome measures over 1 year, the validity of MRI outcome measures by cross-sectional correlation against functionally relevant clinical measures, and the sensitivity of specific MRI indices to early muscle water changes before intramuscular fat accumulation beyond the healthy control range. Methods We did a prospective observational cohort study of patients with either Charcot-Marie-Tooth disease 1A or inclusion body myositis who were attending the inherited neuropathy or muscle clinics at the Medical Research Council (MRC) Centre for Neuromuscular Diseases, National Hospital for Neurology and Neurosurgery, London, UK. Genetic confirmation of the chromosome 17p11·2 duplication was required for Charcot-Marie-Tooth disease 1A, and classification as pathologically or clinically definite by MRC criteria was required for inclusion body myositis. Exclusion criteria were concomitant diseases and safety-related MRI contraindications. Healthy age-matched and sex-matched controls were also recruited. Assessments were done at baseline and 1 year. The MRI outcomes—fat fraction, transverse relaxation time (T2), and magnetisation transfer ratio (MTR)—were analysed during the 12-month follow-up, by measuring correlation with functionally relevant clinical measures, and for T2 and MTR, sensitivity in muscles with fat fraction less than the 95th percentile of the control group. Findings Between Jan 19, 2010, and July 7, 2011, we recruited 20 patients with Charcot-Marie-Tooth disease 1A, 20 patients with inclusion body myositis, and 29 healthy controls (allocated to one or both of the 20-participant matched-control subgroups). Whole muscle fat fraction increased significantly during the 12-month follow-up at calf level (mean absolute change 1·2%, 95% CI 0·5–1·9, p=0·002) but not thigh level (0·2%, −0·2 to 0·6, p=0·38) in patients with Charcot-Marie-Tooth disease 1A, and at calf level (2·6%, 1·3–4·0, p=0·002) and thigh level (3·3%, 1·8–4·9, p=0·0007) in patients with inclusion body myositis. Fat fraction correlated with the lower limb components of the inclusion body myositis functional rating score (ρ=–0·64, p=0·002) and the Charcot-Marie-Tooth examination score (ρ=0·63, p=0·003). Longitudinal T2 and MTR changed consistently with fat fraction but more variably. In muscles with a fat fraction lower than the control group 95th percentile, T2 was increased in patients compared with controls (regression coefficients: inclusion body myositis thigh 4·0 ms [SE 0·5], calf 3·5 ms [0·6]; Charcot-Marie-Tooth 1A thigh 1·0 ms [0·3], calf 2·0 ms [0·3]) and MTR reduced compared with controls (inclusion body myositis thigh −1·5 percentage units [pu; 0·2], calf −1·1 pu [0·2]; Charcot-Marie-Tooth 1A thigh −0·3 pu [0·1], calf −0·7 pu [0·1]). Interpretation MRI outcome measures can monitor intramuscular fat accumulation with high responsiveness, show validity by correlation with conventional functional measures, and detect muscle water changes preceding marked intramuscular fat accumulation. Confirmation of our results in further cohorts with these and other muscle-wasting disorders would suggest that MRI biomarkers might prove valuable in experimental trials. Funding Medical Research Council UK.


MRI Protocol
Selection of musculature Lower limb muscles were chosen as the region for study as they are a key site of pathology in both CMT1A (lower calf) 2 and IBM (quadriceps) 1 and weakness is these areas is a key cause of disability in these patient groups. Furthermore lower limb imaging is practical compared with dedicated upper limb image: both limbs may be imaged simultaneously lowering scanning times and lower limb imaging is in our experience more comfortable for participants.

Block positioning
Imaging was performed using a multi-channel peripheral angiography coil (Siemens 'PA Matrix') and 'spine matrix' coil elements supplemented with a body surface coil for proximal thighs. Before scanning the distance between the anterior superior iliac spine and the superior border of the patella was measured and thigh-level imaging volumes were centred one third of this distance above the patella superior border. Similarly calf-level imaging volumes were centred below the tibial tuberosity by one quarter of the total distance from the tibial tuberosity to the lateral malleolus. The derived distances were recorded and used for follow up imaging.
The TE=3.45ms image was used for the region of interest (ROI) placement and as a reference for inter-method image interpolation and registration (calf) using FLIRT (FSL, FMRIB, Oxford), such that that the same ROIs could be applied to extract data from all maps.

T 2 -Relaxometry
Dual-contrast turbo-spin-echo (TSE) images (6500/13/52ms or 6500/16/56ms; 10x10mm slices with 10mm gap, iPat=2, BW=444Hx/pixel, refocusing flip angle 180º, NEX=1, 6/8 k-space sampling, 256x128matirx (thigh), 256x120matrix (calf)) were acquired. Pseudo-T 2 was calculated from the respective pixel intensities I TE1 and I TE2 from the TE 1 and TE 2 images as T 2 = ) / ln( . The different echo times were the result of altered constraints following a routine scanner software upgrade which occurred after 54 baseline and 6 follow-up scans had been completed. Analysing control values pre-and post-upgrade suggested a systematic bias between pre-and post-upgrade T2 values. By comparing the observed relationship between FF and T2 measurements pre-and post-software upgrade a correction equation was determined separately for thigh muscles (corrected T2 = 1.0606 x post-upgrade T2 + 1.1522) and calf muscles (corrected T2 = 1.0933 x post-upgrade T2 -0.0245). These corrections were applied to all post-upgrade T2 measurements to ensure pre-and post-upgrade T2 measurements were comparable. Parameters for the other quantitative sequences were not affected by this upgrade, and analysis of the control values pre-and post-upgrade indicated no systematic bias in these values was introduced by the software upgrade.

Analysis Slice Selection
For the baseline scan the fifth most superior slice was used in the thigh and the sixth slice in the calf unless all muscles were not visible in which case an adjacent slice was selected. The ROI were drawn on the follow up acquisition on the slice closest to that used for the first scan, determined on the basis of measured distance from bony landmarks (tibial plateau or tip of the fibular head) identified on the 3D-FLASH images. After extraction, all data were cross-checked for outliers and any errors rectified.

Data analysis
Whole muscle ROIs were defined to encompass the entire muscle cross sectional area (CSA) to the fascia whilst "small" ROIs were defined in a consistent anatomical location within each muscle to avoid contamination with fascia or vessels and to allow for minor anatomical movement between acquisitions (figure 2A). Left and right limb ROIs were defined for all muscles at these levels: rectus femoris, vastus lateralis, vastus intermedius, vastus medialis, semimembranosus, semitendinosus, biceps femoris, adductor magnus, sartorius, gracilis, tibialis anterior, peroneus longus, lateral gastrocnemius, medial gastrocnemius, soleus and tibialis posterior muscles.
Minor adjustments to small ROI were made where imperfect registration meant ROI were no longer wholly within the target muscle. Whole muscle ROIs were transferred to the inherently co-registered FF maps only. The whole muscle ROIs from the unprocessed Dixon acquisition were not used for T2 and MTR analysis to avoid ROI contamination with non-muscle tissue, a particular problem for these measures, at the region boundaries due to minor subject movement between acquisitions. ROIs including areas of gross artefact were excluded from the analysis. Of 248 total maps, 5 fat fraction maps (2%), 2 T2 maps (1%) and 14 MTR maps (5%) were excluded due to technical issues or widespread artefact. Artefact in the anterior right thigh of the MTR maps limited analysis in this region as reported in the application of this protocol to healthy volunteers. 4 This is demonstrated in the number of each ROI analysed in figure e1.

Statistical methods
If the follow-up duration differed from 12 months the magnitude of change was converted to an annualised value. Outcome measure 12 month SRMs were calculated for each measure as the mean change between baseline and follow-up divided by the standard deviation of that change, and categorised by magnitude according to Cohen's suggestion: <0·2 minimal responsiveness; 0·2-0·5 small responsiveness; 0·5-0·8 moderate responsiveness; >0·8 large responsiveness. Data are mean ± standard deviation (range). Isometric values are the peak torque at the fixed angle listed whilst isokinetic values are the peak torque at the fixed speed noted. Both CMT1A patients and IBM patients have significantly (p<0.01 for all) reduced strength than their matched control groups for all measurements.