Beyond sacral slope: a new quantitative paradigm for Roussouly classification

To propose a pelvic incidence (PI)-based asymptomatic spine classification, quantitatively define differences among Roussouly morphotypes, and develop a morphotype-oriented algorithm to guide surgical planning.

A retrospective analysis of 102 asymptomatic adults undergoing full-spine EOS radiographs was performed. Patients were classified according to the original Roussouly criteria. Spinopelvic parameters were measured using validated software. Receiver operating characteristic (ROC) curve analysis assessed the discriminative capacity of sagittal variables, with optimal cutoffs determined using the Youden index. A one-versus-rest ROC approach was applied for three-group comparisons. Linear regression analysis evaluated the relationship between lumbar lordosis (LL), PI, and Roussouly type.

Although mean PI increased across types, significant overlap was observed among Types 1-3 for PI < 50°. In the binary comparison between Types 1 and 2, L5-S1 lordosis showed the highest discriminative performance (AUC 0.92; 95% CI 0.848-0.997; p < 0.001), with a cutoff of 19° differentiating Type 2 (≤ 19°) from Type 1 (> 19°). In the three-group analysis (Types 1-3), L5-S1 remained the most discriminative parameter (Type 2 vs. rest AUC 0.85). Thoracolumbar junction morphology differed consistently: kyphotic in Type 1, neutral in Type 2, and neutral-to-lordotic in Type 3.

Sagittal alignment cannot be defined by PI alone. Integration of distal lordosis (L5-S1) and thoracolumbar morphology improves morphotype differentiation and may support more precise surgical planning.