MRI is a key imaging modality for detection of osteomyelitis (OM), but the gold standard for OM diagnosis is histopathology with culture confirmation. There are scant data correlating MRI findings with bone pathology and culture results.
Retrospective review of pathology records for the years 2010-2014 using the search terms “bone inflammation,” “bone biopsy” and “osteomyelitis” to identify cases with bone pathology and a priori consideration of OM. Cases with an MRI in the week prior to the specimen were reviewed to identify clinical, MRI and pathologic data.
We identified 103 patients (pts) with complete records and an MRI in the week prior to bone biopsy/resection: 78 (75%) underwent bone biopsy [59 (75%) foot bones, 19 (24%) vertebrae] and 21 (20%) underwent bone resection; 4(3%) had both. 72/103 had a diagnosis of OM (24 on pathology only, 20 with positive bone cultures, 28 with both); 31/103 did not have OM on pathology and had negative bone cultures. Median age of the cohort (N=103) was 60 years; 71(68%) were male, 64% had diabetes, 22% had peripheral vascular disease. Both groups received a mean of 5 days of antibiotics prior to the specimen (p=0.6).
On MRI, 58 (56%) had marrow edema, 64 (62%) cortical erosions, 30 (29%) decreased T1 signal and 35 (33%) increased T2 signal intensity. MRIs were read as definite OM in 53/72 (73%) in those with proven OM and 22/31 (70%) without OM (p=0.8). Pts with and without OM did not differ with respect to the frequency of marrow edema [43(59%) vs. 15 (50%), p=0.3], cortical erosions [48(66%) vs. 16(53%),p=0.2], decreased T1 signal [18 (25%) vs. 12 (40%),p=0.1] or increased T2 signal [22(30%) vs. 13 (43%), p=0.2].
The sensitivity (Sns), specificity (Sp), positive (PPV) and negative predictive values (NPV) of MRI findings for the diagnosis of OM were: marrow edema (Sns=59%, Sp=50%, PPV=74%, NPV=34%), cortical erosions (Sns=66%, Sp=46%, PPV=75%, NPV=36%) or combined marrow edema and cortical erosions (Sns=45%, Sp= 73%, PPV=80%, NPV=36%).
Conclusion: No significant difference was observed in the frequency of MRI findings specific for OM between those with or without proven OM. The correlation between MRI and pathologic and culture-evident OM in routine practice may be less than suggested by the literature, and further data regarding their correlation in clinical practice are warranted.
D. Mildvan, None
D. C. Perlman, None