Early outcomes with virtual surgical planning software and patient-specific instrumentation in adult spinal deformity.

Abstract

Objective: Software engineering innovations have led to the development of virtual surgical planning software (VSPS) for deformity correction. VSPS uses calibrated radiographs and machine learning predictive models to simulate postoperative spinopelvic parameters based on corrective techniques and anticipated compensatory/reciprocal changes. The authors aimed to assess the safety and efficacy of deformity correction in adult spinal deformity using VSPS and patient-specific rods manufactured based on a simulated plan.

Methods: A retrospective analysis of a prospectively maintained database was conducted, and 146 patients who underwent long-segment thoracolumbar fusions with pelvic fixation (October 2015-May 2023) with a minimum of 1 year of follow-up for deformity correction consistent with the Scoliosis Research Society (SRS)-Schwab classification were identified. Patients were dichotomized into a VSPS group (61 patients, mean age 62.1 years) and a historical control group (85 patients, mean age 64.3 years) prior to implementing VSPS. Comparative analyses were performed to assess VSPS accuracy and outcomes. Equivalence analysis was performed via the two one-sided t-test method using Cohen's d = 0.5.

Results: In the VSPS group, the achieved spinopelvic parameters at 3 months were equivalent to the simulated plan for lumbar lordosis (LL), pelvic tilt (PT), pelvic incidence (PI), sacral slope, PI-LL mismatch, sagittal vertical axis (SVA), T1 pelvic angle (T1PA), thoracic kyphosis, and L4-S1 lordosis. Compared with controls, VSPS demonstrated an increased L4-S1 lordosis (p = 0.001) and decreased T1PA (p = 0.001); L4-S1 lordosis within 4.26° (p = 0.989) and T1PA within 3.85° (p = 0.969) were not significantly equivalent. VSPS demonstrated a significant increase in achievement of an SVA < 5 cm (p = 0.026), T1PA < 20° (p = 0.001), and age-adjusted T1PA (p < 0.001). The age-adjusted PI-LL mismatch (p = 0.018), PT (p = 0.002), and SVA (p = 0.021) were equivalent. There was no significant difference in the improvement of 1-year patient-reported outcome measures (PROMs), proximal junctional kyphosis (p = 0.270), or proximal junctional failure (p = 0.290) between the two groups. In the multivariate analysis, VSPS use independently predicted achievement of age-adjusted T1PA (OR 6.51, p = 0.001). The upper instrumented vertebra, number of rods and rod material, and VSPS were not predictors of complications or reoperation for hardware failure. The time interval from the first surgery was not a predictor of complications or age-adjusted spinopelvic parameters.

Conclusions: VSPS accurately predicted achieved spinal alignment at 3 months. VSPS has the potential to facilitate global spinopelvic parameter correction, particularly reflected by T1PA and L4-S1 lordosis. However, this study did not demonstrate meaningful improvements in PROMs compared with controls. To justify the increased cost and widespread adoption of this technology in its current state, a clear functional benefit should be demonstrated along with a cost-benefit analysis. Further research is necessary to delineate long-term durability and the potential impact of VSPS and patient-specific instrumentation, particularly with continual software and engineering innovations.