Quantitative Assessment of Postoperative Lumbar Morphology Using Geometric Morphometrics: Within-Subject Changes Following TLIF and Decompression.

Background: Conventional radiographic assessment of postoperative lumbar alignment relies predominantly on Cobb angle measurements, which incompletely characterize global spinal morphology. We evaluated whether geometric morphometrics (GM) provides complementary quantitative information for assessing postoperative lumbar morphology after TLIF and decompression. Methods: Ninety-nine patients (TLIF n = 70; decompression n = 29) underwent landmark-based GM analysis of pre- and post-surgical lateral lumbar radiographs (48 landmarks, L1-S1). Because TLIF and decompression patients were imaged at markedly different postoperative intervals as a matter of routine clinical practice, the primary analyses tested whether each technique's own pre-to-post shape change, adjusted for baseline L1-S1 angle, log centroid size, sex, age, number of operated levels, and degenerative spondylolisthesis, was significant (sign-flip permutation test), along with lordosis change (paired t-test, unadjusted). Cobb analyses assessed global lumbar lordosis (L1-S1) and operated level(s). Results: TLIF showed a significant within-group reduction in global lumbar lordosis at early postoperative imaging (Δ = -3.75°, paired t-test p < 0.001) and a significant, systematic change in overall lumbar shape (sign-flip permutation p < 0.001); decompression showed no significant within-group change in global lordosis (Δ = +0.47°, p = 0.733) or overall shape (p = 0.31), but a small increase in segmental lordosis at the operated level(s) (Δ = +1.44°, p = 0.036) that was not robust to sensitivity analysis. Conclusions: In this retrospective cohort, TLIF was associated with reduced lumbar lordosis and a systematic shape change at early postoperative imaging, while decompression showed preserved global lordosis and a small gain in segmental lordosis at long-term imaging. GM provided a consistent quantitative framework, with strong correlation and level-dependent agreement against manual Cobb measurement, for characterizing these within-technique postoperative changes.