What type of study is this?

September 5, 2025Open Access

Diagnostic accuracy and clinical utility of mTLICS versus TLICS and TL AOSIS in stratifying three-tier treatment for thoracolumbar injuries: focus on intermediate score range

QDQuang DaoChildren's Hospital at Westmead VNVan Son NguyenLe Quy Don Technical University VDVan Phu DangCan Tho University

Key Points

mTLICS demonstrated the highest diagnostic accuracy, particularly in patients with intermediate TLICS scores.
With an AUC range of 0.94-1.00, mTLICS significantly improved prediction of treatment options compared to TLICS and TL AOSIS.
Multinomial regression positioned mTLICS as the sole significant predictor for treatment allocation, emphasizing its clinical utility.
Despite promising results, prospective validation of mTLICS is necessary to ensure broader applicability and minimize bias.

Abstract

Thoracolumbar injury classification systems such as TLICS and TL AOSIS are widely implemented but offer limited guidance in intermediate score ranges (TLICS = 3-4), where treatment decisions are often uncertain. The modified TLICS (mTLICS) was developed to address this gap by integrating MRI-derived quantitative parameters. This retrospective study included 146 adults with MRI-confirmed thoracolumbar spine injuries (T1-L5) treated at Phu Tho Provincial General Hospital between April 2024 and May 2025. Inclusion required MRI within 7 days of trauma and complete clinical data, including ASIA grade, VAS score, and treatment modality. All cases were classified using TLICS, TL AOSIS, and mTLICS, and managed conservatively, minimally invasively, or surgically. Predictive performance for treatment allocation was assessed using ROC analysis, multinomial logistic regression, and decision curve analysis (DCA). mTLICS showed the highest diagnostic accuracy across all treatment comparisons (AUC = 0.94-1.00), particularly in the intermediate-score group (TLICS = 3-4), with AUCs of 0.991 (conservative vs. surgical) and 0.965 (minimally invasive vs. surgical). Multinomial regression identified mTLICS as the sole independent predictor of treatment allocation (OR = 31.2-1338.4; p < 0.01), while TLICS and TL AOSIS were not statistically significant. DCA demonstrated the highest net clinical benefit for mTLICS, especially within the 0.3-0.6 threshold range. The mTLICS demonstrated improved accuracy in stratifying thoracolumbar injuries across three treatment tiers and enhanced clarity in decision-making for intermediate cases. Its MRI-based components support personalized, image-guided management. Nonetheless, as clinicians at the study site were familiar with the mTLICS framework through prior academic exposure, potential incorporation bias cannot be entirely excluded. These findings should therefore be interpreted with caution, as mTLICS scores were calculated retrospectively after treatment completion and were not used prospectively to determine patient management, underscoring the need for prospective, multicenter validation to confirm its generalizability.

Ask AI

Helpful

Bookmark

View Full Paper