BACKGROUND: Open reduction and internal fixation (ORIF) has been widely performed because the osteochondral component of the osteochondritis dissecans (OCD) lesion is the most suitable for reconstructing the joint structure. PURPOSE: To evaluate radiological healing in terms of reconstructed bony structure after ORIF with bone graft by computed tomography (CT), to identify preoperative prognostic factors for failure, and to determine the cutoff value of radiological healing for risk of failure. STUDY DESIGN: Case-control study; Level of evidence, 3. METHODS: A retrospective cohort study of 42 patients (44 knees) who underwent internal fixation with bone graft for OCD lesions of the knee from 2004 to 2018 was conducted. All patients were evaluated 6 months postoperatively, and if not healed 6 months after surgery, they were evaluated by CT periodically thereafter. Radiological healing was judged according to the following 3 criteria: (1) reossification of the OCD lesion, (2) bony continuity between the OCD lesion and basal floor, and (3) reconstructed bony surface of the femoral condyle reconstructed to match the normal joint. Then, the percentage of the radiological healing area was calculated as the ratio of the healing length to the total lesion length. The nonhealing area was calculated by multiplying the sum of the total nonhealing length. Clinical failure was defined as any definitive reoperation for the same OCD lesion, such as fragment excision, or a cartilage restoration procedure. After 6 months, all eligible patients underwent arthroscopy to check for protrusion of the absorbable pin into the joint; the removal of an absorbable pin protruding into the joint was not considered a failure. RESULTS: Clinical failure was recorded for 4 cases (9.1%). The mean overall percentage of the radiological healing area of OCD 6 months after ORIF with bone graft was 79.5% ± 24.4%, and the mean overall nonhealing area at 6 months was 87.8 ± 107.9 mm2. The percentages of radiological healing area of stable (International Cartilage Regeneration & Joint Preservation Society OCD II) lesions and femoral condylar (lateral femoral condyle + medial femoral condyle) lesions were significantly lower than unstable lesions and femoral groove lesions, respectively (P = .01 and P = .03, respectively). On receiver operating characteristic curve analysis, the cutoff points for predicting a significantly increased risk of failure were 33.9% (sensitivity, 100%; specificity, 100%; area under the curve, 1) for the percentage of radiological healing area and 222.9 mm2 (sensitivity, 95%; specificity, 100%; area under the curve, 0.956) for the nonhealing area 6 months postoperatively. CONCLUSION: A stable lesion and a femoral condylar lesion were the predictors of poor radiological healing on CT images 6 months after ORIF with bone graft. The risk of failure was increased significantly in cases with only approximately one-third of the lesion healed or in cases with large nonhealing areas at 6 months postoperatively.
