Optimizing methods to quantify intramuscular fat in rotator cuff tears with normalization.

Abstract

To determine which normalization method may best account for confounding individual factors, such as age or BMI, when quantifying fat infiltration on MRI in patients with rotator cuff tears, the effects of normalization using three different muscles (teres major; triceps brachii; teres minor) were compared. Thirty-seven consecutive patients diagnosed with rotator cuff pathology were included. MRI fat-water sequences were used to quantify rotator cuff intramuscular fat (%fat). Three reference muscles (teres major, triceps, teres minor) were used to derive normalized %fat. Relationships between intramuscular %fat and tear size, age, and BMI in each rotator cuff muscle, before and after normalization, were compared with Fisher transformations (α = 0.05). Normalization with teres major ameliorated confounding relationships of age and BMI on rotator cuff %fat. In contrast, normalization with triceps maintained the confounding relationships between %fat and age in supraspinatus (p = 0.03) and infraspinatus/teres minor (p = 0.028). Normalization with teres minor maintained the confounding relationship between %fat and BMI in subscapularis (p = 0.039). Normalization with teres major best-maintained relationships between tear size and infraspinatus/teres minor %fat (p = 0.021). In contrast, normalization with triceps or teres minor eliminated all significant relationships with tear size. Results of this pilot study suggest normalization to teres major using MRI-based %fat quantification methods can effectively control for individual factors, such as BMI or age, and may have utility in evaluating and monitoring rotator cuff fat infiltration attributed specifically to a tendon tear.