Summary

MRI is widely accepted as an accurate method for the diagnosis of meniscal tears, however, the role of MRI in predicting the reparability of meniscal tears is still not established. We demonstrated that MRI is a powerful diagnostic tool for predicting meniscal reparability, if executed with optimized protocols and using the correct criteria.

Abstract

In this study we wanted to evaluate the predictability of MRI for differentiating reparable and non reparable meniscal tears. We classified the tears in groups with different odds of reparability: longitudinal vertical, the most commonly reparable; longitudinal horizontal and radial, less commonly reparable; complex tears, usually not reparable; bucket-handle tears, reparable only if the tear is located all along into the red zone; meniscal flaps, never reparable. Then we used two criteria: the maximum distance from the external meniscal wall to the tear and age.
All the consecutive patients undergoing arthroscopic surgery for meniscal tears in the period from January 1st, 2008 to April 30th, 2016 at the author’s institution were considered. Inclusion criteria were as follows: arthroscopic surgical repair of meniscal tears; the presence of an MRI of the knee performed in our Hospital within 4 months before surgery. Overall, 30 subjects fulfilled the inclusion criteria and were included in this study (20 men, 10 women, mean age 27,2 ± 9,7 range 14-44). In all patients an all inside suture was performed. Only one surgeon operated all the subjects. A second group of 49 consecutive patients were operated with meniscectomy, within the same period, and were considered as control group. The inclusion criteria of the control group were as follows: arthroscopic procedure of meniscectomy; age less than 45; availability of an MRI performed in our Hospital.
All the examinations were performed with 1,5 Tesla (T) scanner (Magnetom Symphony, Siemens AG, Erlangen, Germany), with phased array coils. All the examinations were performed with a standard protocol.
One single operator reviewed all the MRI of the patients included in the study, blind to their generalities, age and final surgical procedure.
The accuracy of MRI in identifying correctly meniscal tears overall was 96%(76 meniscal tears over 79). MRI performance in predicting the reparability among overall meniscal tears was as follows: sensibility 88 %, specificity 77 %, accuracy 84%. For vertical tears values of sensibility, specificity and accuracy were respectively 81%, 86%, 92%; for horizontal tears they were respectively 90%, 50%, 81%. K analysis to evaluate agreement between MRI and arthroscopy was good (k =0.65 standard error 0.09; C.I. 0.48 – 0.82). This value was even higher for the vertical tears (k = 0.69, standard error 0.02, C.I. 0.45 to 0.92) but was only moderate for horizontal tears (k= 0.43, standard error 0.21 C.I. 0.02 to 0.85). ROC curve for the overall meniscal tear had an area under the curve (AUC) of 0.85 (p<0.001); with a cutoff value of <3 mm values of sensitivity 60.0 % (C.I. 40.6 - 77.3) and specificity 95.8% (C.I. 85.7 - 99.5) were obtained. For the vertical tears the AUC was excellent (0.91; p<0.001); with a cutoff value <4 mm sensibility of 82.6% (C.I. 61.2 - 95.0) and specificity of 81.3 % (C.I. 54.4 - 96.0) were obtained. For the horizontal meniscal tears the results were less encouraging as the AUC: 0.76 (p = 0.025); with a cutoff < 5 mm sensibility 50.0% (C.I. 11.8 - 88.2), specificity 85.7 % (CI 63.7 - 97.0) were obtained.