Use of preoperative magnetic resonance imaging to predict rotator cuff tear pattern and method of repair.

PURPOSE

To determine the magnetic resonance imaging (MRI) criteria for predicting rotator cuff tear pattern and method of repair.

TYPE OF STUDY

Retrospective MRI/arthroscopy correlation.

METHODS

Sixty-six preoperative MRI scans were evaluated. The maximum medial to lateral length (L) of the tear was measured on T2-weighted coronal cuts. The maximum anterior to posterior width (W) was measured on T2-weighted sagittal cuts. The cases were divided into 3 groups: group 1, short-wide tears, L < or = W, L < 2 cm; group 2, long-narrow tears, L > W, W < 2 cm; and group 3, long-wide tears, L > or = 2 cm, W > or = 2 cm.

RESULTS

Of the 66 MRI scans, 55 were adequate for standardized measurement. Group 1, 16 cases: 15 were found at arthroscopy to be crescent-shaped tears repaired end-to-bone; 1 was repaired with interval slides. Group 2, 22 cases: all 22 were repaired side-to-side/margin convergence. Group 3, 17 cases: 12 required interval slides, 1 partial repair was performed, and 4 were repaired side-to-side/margin convergence.

CONCLUSIONS

Tear pattern and method of repair can be predicted on high-quality MRI scan. Group 1, L < or = W and L < 2 cm, predicts a crescent-shaped tear and end-to-bone repair (positive predictive value, 93.8%). Group 2, L > W and W < 2 cm, predicts a longitudinal tear and side-to-side/margin convergence repair (positive predictive value 100%). Group 3, L > or = 2 cm and W > or = 2 cm, predicts a massive contracted tear and that primary end-to-bone or side-to-side repairs are usually not possible and that interval slides or partial repair may be necessary (positive predictive value, 76.5%). The overall diagnostic model based on usable MRI scans significantly predicted arthroscopic findings (P < .001 for chi-square test).

LEVEL OF EVIDENCE

Level III, development of diagnostic criteria with universally applied reference (nonconsecutive patients).