The predictive value of creatine kinase, EMG and MRI in diagnosing muscle disease

Sir, MRI is increasingly employed in the investigation of suspected neuromuscular disease. High signal intensity on fat-suppressed T2-weighted and short T1 inversion recovery (STIR) muscle MR images is found in inflammatory myositis [1]. MRI is sensitive for localizing non-homogeneous muscle abnormalities allowing biopsy to be targeted [2]. Previous studies have shown that non-targeted single site biopsy may lead to the diagnosis being missed in approximately one-third of cases [3]. Suspected muscle disease is a common referral to the rheumatologist. Symptoms can be mild and non-specific and the referral is often prompted by a rise in creatine kinase (CK) which may be small. It is unclear how these patients are best investigated. Muscle biopsy is the gold standard for diagnosing muscle disease; however, the process is invasive and the diagnostic yield often low [4]. We performed a retrospective study to determine whether it is possible to predict those patients in whom muscle biopsy will be abnormal using the results of screening investigations. All patients referred for muscle biopsy from the Rheumatology Department at Coventry University Hospital from January 2001 to October 2006 were included in the study. Approval was gained from the local Research and Development Committee. Patient demographics, creatinine kinase (CK) level at time of referral (normal range 0–125 IU/l), any EMG and muscle MRI results, and the result of the muscle biopsy were recorded. All biopsies were from a single site using an open technique and were processed by the Cellular Pathology Department at University Hospital, Birmingham. A total of 64 patients underwent muscle biopsy during the study period [females 40; mean age (s.d.) 52.7 (15.5) yrs]. The majority of patients were White Caucasian (81%); the remainder were Asian (11%), Afro-Caribbean (5%) and unknown (3%). Twenty-nine (45%) biopsies showed specific abnormalities. The remaining 55% were classified as normal, showing non-specific changes or selective type 2 fibre atrophy only. The most common abnormal finding was of an inflammatory myopathy (19 out of 29, 66%): polymyositis (2), dermatomyositis (5), inclusion body myositis (2), non-differentiated inflammatory myositis (10). The remaining abnormal biopsies were consistent with primary neuropathic disease (5), vasculitis (1), granulomatous disease (1), a mitochondrial disorder (1), adult-onset acid maltase deficiency (1) and eosinophilic fasciitis (1). CK values ranged from 21 to 14 117 IU/l. All patients with CK values >1800 IU/l (approximately 15-fold the upper limit of normal) had muscle biopsy results consistent with an inflammatory myositis. However, 6 out of 19 (32%) patients with inflammatory changes on biopsy had CK values <1000 IU/l and two patients had CK values within normal limits. The sensitivities, specificities and likelihood ratios of CK, EMG and MRI for predicting abnormal muscle biopsy are shown in Table 1. All patients had CK measured prior to biopsy, 50 (78%) patients underwent EMG and 21 (33%) patients had MRI. Predictive values of screening investigations for abnormal muscle biopsy n, number of patients undergoing screening test; LR+, positive likelihood ratio; LR−, negative likelihood ratio. Predictive values of screening investigations for abnormal muscle biopsy n, number of patients undergoing screening test; LR+, positive likelihood ratio; LR−, negative likelihood ratio. A CK value >1000 IU/l (8-fold the upper limit of normal) was not sensitive but was specific for abnormal muscle biopsy (48 and 94%, respectively). Decreasing the CK threshold to >500 IU/l increased sensitivity but at the expense of specificity (66 and 77%, respectively). EMG showed similar results to lower threshold CK (sensitivity 74%, specificity 67%). The highest sensitivity and specificity was seen with MRI (92 and 89%, respectively). Combining screening modalities was no better than MRI alone (Table 1). The single false positive seen with MRI screening occurred in a patient in whom biopsy was taken from the contralateral limb to that showing abnormal signal. The single false negative was a patient in whom the biopsy revealed very mild inflammatory changes and who improved spontaneously without treatment. Biopsy abnormalities found in patients with abnormal MRI included inflammatory, neuropathic and vasculitic disease. CK values in these patients ranged from 21 to 14 117 IU/l. Myositis-specific antibodies other than anti-Jo-1 were not routinely tested. Three patients were anti-Jo-1 positive. Two had an elevated CK and abnormal biopsy consistent with inflammatory myositis. The third patient however, presented with mild myalgia but no weakness and had a normal CK and biopsy. This suggests that muscle biopsy is helpful even in patients with positive myositis-specific antibodies. The frequency of positive muscle biopsy result in this study was only 29 out of 64 (45%) meaning nearly one in two patients underwent a non-informative invasive test. CK level and EMG were poor predictors of muscle biopsy result. MRI, however, was sensitive and specific for predicting abnormal biopsy, although we accept the sample size was small. We propose that patients with suspected muscle disease should undergo MRI, particularly those with only modest rises in CK. This allows both the detection of structural abnormalities and the site of biopsy to be targeted. Biopsy in patients with negative scans should be performed with caution and patients counselled as to the low diagnostic yield. Further studies involving larger patient numbers are welcomed to confirm these findings. The authors have declared no conflicts of interest.