Excellent Intra- and Interobserver Reliability of the D.H. Dejour Version 2 Classification for Trochlear Dysplasia Using Radiographs Combined With MRI With Evaluation of Shortcomings

TL;DR

This study shows that using radiographs and MRI for the Dejour V2 classification of trochlear dysplasia has high reliability, especially when simplified to two grades.

Contribution

The study introduces a simplified two-grade classification for trochlear dysplasia that improves diagnostic reliability and sensitivity.

Findings

Intrarater and interrater reliability improved to 0.95 and 0.93, respectively, with the simplified two-grade classification.

High-grade trochlear dysplasia had 97.8% sensitivity and 96.4% specificity when using the simplified classification.

The original four-type classification had moderate sensitivity for low-grade trochlear dysplasia.

Abstract

The D.H. Dejour classification (Version 2 [V2]) expanded upon the H. Dejour radiographic classification of trochlear dysplasia by adding computed tomography (CT) scans to the evaluation. Magnetic resonance imaging (MRI) then became the main investigation of choice.

To report the reliability of the Dejour V2 using a combination of radiographs and MRI instead of CT scan as per the original classification and to explore differences in the assessment of trochlear dysplasia between assessors to better understand limitations to the classification.

Cohort study; Level of evidence, 3.

This is a retrospective comparative study, conducted by reviewing a prospectively maintained institutional database, between 2 groups of patients: those with recurrent patellar dislocation, termed objective patellar instability (OPI), and control patients with no patellofemoral symptoms. Inclusion criteria were available preoperative imaging including both knee MRI and a true lateral view radiograph of the knee at 20° of flexion and no history of previous knee surgery. Imaging evaluation was performed independently by 2 orthopaedic surgeons, and each trochlea was classified according to the Dejour V2 classification. To classify, all reviewers initially used the lateral radiograph, then confirmed with MRI slice imaging.

A total of 200 patients were included in the statistical analysis (OPI, n = 123; control, n = 77). In the control group, 13% of patients presented with trochlear dysplasia type A, whereas 87% of patients had a normal trochlea. The kappa coefficient was 0.77 for intrarater reliability and 0.75 for interrater reliability, representing a substantial level of agreement. In the OPI group, 97% of patients presented a trochlear dysplasia. The kappa coefficient was 0.92 for intrarater reliability and 0.86 for interrater reliability, representing an excellent correlation between reviewers. When simplified from 4 types to 2 types of trochlear dysplasia, high-grade (supratrochlear spur present) versus low-grade (no supratrochlear spur present), the intrarater reliability and interrater reliability improved to 0.95 and 0.93, respectively, and there was a 97.8% sensitivity and 96.4% specificity for diagnosing high-grade trochlear dysplasia.

Utilizing radiographs and MRI for the Dejour V2 classification of trochlear dysplasia, we demonstrated only moderate sensitivity in diagnosing low-grade trochlear dysplasia utilizing the 4 types of trochlear dysplasia. The sensitivity for diagnosing low-grade trochlear dysplasia, along with the overall intra- and interrater reliability was improved by simplifying the classification from 4 types of dysplasia to 2 grades, low versus high grade, based on the presence of a supratrochlear spur.

Figures