Objective
Osteoarthritis (OA) is currently diagnosed using clinical and
radiographic findings. In recent years MRI use in osteoarthritis has
increasingly been studied. This study was conducted to determine the
diagnostic utility of MRI in OA through a meta-analysis of published
studies.
Methods
A systematic literature search was undertaken to include studies that
used MRI to evaluate or detect osteoarthritis. MRI was compared to various
reference standards: histology, arthroscopy, radiography, CT, clinical
evaluation, and direct visual inspection. Sensitivity, specificity, positive
predictive value (PPV), negative predictive value (NPV), and receiver
operating characteristic (ROC) area under the curve were calculated. Random
effects models were used to pool results.
Results
Of 20 relevant studies identified from the literature, 16 reported
complete data and were included in the meta-analysis, with a total of 1220
patients (1071 with OA and 149 without). Overall sensitivity from pooling
data of all the included studies was 61% (95% confidence interval [CI] 53 to
68), specificity was 82% (95% CI 77 to 87), PPV was 85% (95% CI 80 to 88),
and NPV was 57% (95% CI 43 to 70). The ROC showed an area under the curve of
0.804. There was significant heterogeneity in the above parameters
(I2 >83%). With histology as the reference standard,
sensitivity increased to 74% and specificity decreased to 76% compared with
all reference standards combined. When arthroscopy was used as the reference
standard, sensitivity increased to 69% and specificity to 93% compared with
all reference standards combined.
Conclusion
MRI can detect OA with an overall high specificity and moderate
sensitivity when compared with various reference standards, thus lending
more utility to ruling out OA than ruling it in. The sensitivity of MRI is
below the current clinical diagnostic standards. At this time standard
clinical algorithm for OA diagnosis, aided by radiographs appears to be the
most effective method for diagnosing OA.