Predicting Osteomyelitis in Patients Whose Initial MRI Demonstrated Bone Marrow Edema Without Corresponding T1 Signal Marrow Replacement

SLR - November 2021 - Alyse Acciani

Reference: Sax, AJ, Halpern, EJ, Zoga, AC, Roedl, JB, Belair, JA, Morrison, WB. Predicting Osteomyelitis in Patients Whose Initial MRI Demonstrated Bone Marrow Edema Without Corresponding T1 Signal Marrow Replacement. Skeletal Radiology (2020) 49: 1239-1247

Level of Evidence: IV

Scientific Literature Review

Reviewed By: Alyse Acciani, DPM, MPH
Residency Program: Hennepin Healthcare – Minneapolis, MN

Podiatric Relevance: Aggressive treatment for suspected osteomyelitis can often be hindered by imaging when read as negative. Clinicians generally have a low threshold for ordering x-rays while treating diabetic ulcers, however, these lack the sensitivity and specificity required for diagnosis of acute osteomyelitis. Though costly, an MRI has a greater sensitivity and specificity with regards to the diagnosis of osteomyelitis. The development and implementation of a treatment algorithm along with repeat MRI images while treating an ulcer can be an effective, though aggressive, method to predict osteomyelitis in the diabetic foot ulcer.  

Methods: Retrospective analysis of 60 MRIs of pedal ulcers suspicious for osteomyelitis. Clinical data regarding the ulcer was collected including treatment regimen. Baseline MRIs were read as negative for osteomyelitis. Subsequent MRIs were reviewed and grouped into one of two groups: progression to osteomyelitis, or no progression to osteomyelitis. Data collected from the MRIs included size and depth of ulcer, region of interest (ROI), and bone marrow edema patterns. Statistical analysis included two-sample t-test and a Cox proportional hazard model. Note, ROI refers to a numerical analysis of a given area of a scanned image of particular importance.  

Results: Thirty-four of the 60 MR exams progressed to osteomyelitis. The group that progressed to osteomyelitis had joint fluid ratio average 65 percent, whereas the non-osteomyelitis group averaged 45 percent. Ultimately analysis showing 53 percent of joint effusion was most predictive of progressing to osteomyelitis. Ulcer proximity to bone averaged 6 millimeters versus 9 millimeters in the osteomyelitis to non-osteomyelitis group, respectively. However, further analysis indicated less than 3 millimeters from ulcer to bone is still a predictive value of increased risk of osteomyelitis. Ulcers with a larger mean area (4.1 square centimeters versus 2.4 square centimeters) were not predictive of developing osteomyelitis in a univariate analysis; however, when dichotomized with a cut off 3 square centimeters, the univariate analysis found it to be predictive. When looking at the treatments of the ulcers, there was no statistically significant difference between those that developed osteomyelitis when treated with IV antibiotics oral antibiotics, or wound debridement.

Conclusions: Increasing bone marrow ROI signal/joint fluid ratios on T2/STIR images were the strongest risk factors for developing osteomyelitis, while ulcer size and depth are weaker predictors. The authors suggest a definitive diagnosis of osteomyelitis can be achieved through MRI, though clinically, the gold standard is bone biopsy, which the authors suggest has a low positive yield. Interestingly, the ulcer distance-to-bone was a statistically significant predictor of osteomyelitis, and in the univariate analysis, ulcer diameter was not. Though these parameters require a large sample size to increase validity and power, it poses a clinically significant starting point for additional research.