Prognostic model for outcome after L4–5 minimally invasive transforaminal lumbar interbody fusion based on a comprehensive clinical and radiological analysis

OBJECTIVE The objective of this study was to develop and test the ability of a novel multivariate model to predict outcomes following minimally invasive transforaminal lumbar interbody fusion (TLIF) for degenerative spondylolisthesis at the L4–5 segment, and conduct internal validation of the proposed models for preoperative patient selection and to improve postoperative outcomes. METHODS The authors conducted a retrospective analysis of a prospectively collected database. One hundred ninety-one consecutive patients undergoing TLIF on the L4–5 segment for symptomatic degenerative spondylolisthesis were prospectively enrolled and followed for 1 year. A comprehensive patient clinical and radiological assessment was performed at baseline and 12 months postoperatively. Regression mathematical modeling of preoperative variables was used to create the prognostic model of clinical outcomes (Oswestry Disability Index [ODI] and 36-Item Short-Form Health Survey [SF-36]). To predict the clinical outcome, 3 models were identified: 1) y0, based on the postoperative ODI score; 2) y1, based on the postoperative SF-36 Physical Component Summary (PCS) score; and 3) y2, based on the SF-36 Mental Component Summary (MCS) score. The following criteria were chosen as independent variables: age, BMI, duration of symptoms, presence of motor deficit, preoperative SF-36 PCS, preoperative SF-36 MCS, preoperative ODI score, preoperative back pain, preoperative leg pain, preoperative lumbar lordosis, preoperative interbody space height, preoperative sagittal angle, preoperative linear translation, intervertebral disc degeneration, facet joint degeneration, value of the apparent diffusion coefficient, and facet angle on the operative side. RESULTS All patient-reported outcomes improved postoperatively (median, baseline vs 12 months): ODI score from 72% to 20% (p = 0.01), visual analog scale (VAS) back pain score from 78 to 24 mm (p = 0.02), VAS leg pain score from 92 to 14 mm (p = 0.01), SF-36 PCS score from 26.13 to 41.24 (p = 0.03), and SF-36 MCS score from 22.46 to 46.27 (p = 0.01). Reoperations occurred in 6 patients (3.1%), 9 (4.7%) were readmitted within 30 days of surgery, 168 (88.0%) returned to work, and 24 (12.6%) experienced an unplanned outcome (back pain and/or lower extremity pain > 20 mm according to the VAS, > 20 points on the ODI, a reoperation, or a readmission). These results suggest that the independent preoperative variables determined by radiography and MRI allow the prediction of the clinical outcome, but they have differing roles and dominance depending on the developed predictive model. CONCLUSIONS The predictive regression models that were developed in this study using these data can improve preoperative risk counseling and patient selection for minimally invasive TLIF surgery at the L4–5 segment.