Abstract
Purpose
To compare the agreement and the differences between the modified Thoracolumbar Injury Classification and Severity Score (mTLICS) system and the AO Thoracolumbar Spine Injury Score (TL AOSIS) in guiding surgical decision-making for thoracolumbar fractures.
Methods
The clinical and imaging data of 100 patients with thoracolumbar fractures admitted to our hospital between January 2021 and December 2023 were retrospectively analyzed. Two orthopedic surgeons, blinded to the patients’ clinical outcomes, independently evaluated the cases using both scoring systems and provided treatment recommendations. Disagreements were resolved by a senior attending surgeon. Agreement between the two systems’ treatment categories was quantified by weighted Cohen’s κ with 95% confidence intervals from paired 3 × 3 cross-tabulations, and interobserver reliability was assessed before consensus adjudication.
Results
The two systems assigned the same treatment category in 86 of 100 patients (86.0%; unweighted κ = 0.773, 95% CI 0.666–0.881; linear-weighted κ = 0.820), with no significant asymmetry (McNemar–Bowker P = 0.160). Agreement was substantial in both the 57 neurologically intact patients (κ = 0.696) and the 43 patients with neurological impairment (κ = 0.619). Surgery was recommended by mTLICS versus TL AOSIS in 24.6% versus 19.2% of neurologically intact patients and in 90.7% versus 81.4% of those with neurological impairment. Interobserver reliability before consensus was substantial to high (linear-weighted κ = 0.88 for mTLICS and 0.75 for TL AOSIS). The two scoring systems showed substantial agreement in treatment recommendations. In burst fractures with intervertebral disc injury, mTLICS tended to assign cases to higher treatment categories than TL AOSIS, although these subgroup differences were not statistically significant.
Conclusion
mTLICS and TL AOSIS show substantial concordance as decision-support tools for thoracolumbar fractures. For burst fractures with intervertebral disc involvement, mTLICS tends to recommend surgery more often, reflecting a difference in classification behaviour rather than a demonstrated clinical advantage. Because no outcome data were analysed, whether this tendency improves patient outcomes remains to be determined.
Keywords: Thoracolumbar fracture, TL AOSIS, TLICS, mTLICS
Introduction
Thoracolumbar fractures are the most common type of spinal fracture, accounting for approximately 54–66% of cases. Without effective, guideline-directed management, affected patients are prone to severe sequelae such as kyphotic spinal deformity, intervertebral disc degeneration, and chronic low back pain. These complications pose long-term health risks and also impose substantial economic burdens on patients and healthcare systems [1–3]. Therefore, timely and accurate evaluation of thoracolumbar fractures is critical, and the implementation of reliable scoring systems is essential for guiding clinical decision-making. Currently, the Thoracolumbar Injury Classification System (TLICS) and the Thoracolumbar AOSpine Injury Score (TL AOSIS) are widely applied in clinical practice to determine surgical indications for thoracolumbar fractures. An et al. [4] found that TLICS and TL AOSIS had good agreement in the surgical recommendations for thoracolumbar fractures, but for patients with complete burst fractures or combined neurological impairment, TL AOSIS might be better than TLICS in the evaluation recommendations. To address limitations in TLICS, Lu et al. [5] introduced a “disc injury status” subclass and reduced the scoring weight of the posterior ligamentous complex (PLC) integrity subclass, proposing the modified thoracolumbar injury classification and severity score system (mTLICS). Despite these refinements, mTLICS has not yet gained widespread clinical adoption, and its advantages over TL AOSIS in disc assessment and PLC scoring adjustments remain unproven. As such, a direct comparison of these two scoring systems holds significant clinical importance. Clarifying their differences will not only help improve the consistency of treatment decisions across medical centers but also optimize the selection of surgical indications. This study aims to compare mTLICS and TL AOSIS in guiding surgical decision-making for thoracolumbar fractures by assessing the agreement and the sources of discordance between their treatment recommendations, with particular attention to how each system incorporates intervertebral disc injury into its score. Because no clinical reference standard or follow-up was available, the analysis is framed as a comparison of agreement rather than a test of which system is better.
Materials and methods
General information
We conducted a retrospective study on 100 patients with single-level thoracolumbar fractures who had complete preoperative imaging data and were admitted to the Department of Orthopedics at the Second Affiliated Hospital of Zhejiang Chinese Medical University between January 2021 and December 2023. The cohort included 64 males and 36 females, aged 25–55 years (mean age: 41.3 ± 6.9 years). Neurological status was classified according to the American Spinal Injury Association (ASIA) impairment scale: 57 patients had no neurological deficits (ASIA E), and 43 patients exhibited neurological impairments (ASIA A-D). All cases involved single-level vertebral fractures. Preoperative imaging data included anteroposterior and lateral radiographs of the thoracolumbar spine, computed tomography (CT) with three-dimensional vertebral reconstruction, and magnetic resonance imaging (MRI) performed on a 1.5T scanner. All imaging datasets were anonymized and devoid of any information or annotations related to fracture classification. This study was approved by the hospital’s ethics committee, and the requirement for informed consent was waived due to its retrospective design.
The thoracolumbar fracture morphology was determined through analysis of anteroposterior/lateral X-ray radiographs, CT with three-dimensional vertebral reconstruction and MRI [3]. The MRI diagnostic criteria for PLC injury [6] were defined as follows: Sagittal T2-weighted fat-suppression sequences were used to assess PLC integrity. Discontinuity of ligaments or high signal intensity on these sequences indicated PLC injury. The MRI classification system for intervertebral disc injury [7] classified the adjacent disc into four grades (0–3) on the basis of morphological and signal changes. Grade 0: normal disc, with normal morphology and signal. Grade 1: disc edema, characterized by high signal intensity on T2-weighted imaging (T2WI) with preserved disc morphology. Grade 2: disc rupture, iso- or hyperintense on T1WI, hypointense on T2WI, with surrounding hyperintensity. Grade 3: disc invasion into the vertebral body, manifesting as annular tears or herniation into the vertebral body, iso- or hyperintense on T1WI. In the mTLICS score, Grade 0 = 0 (no injury), Grade 1 = 1 (mild injury), and Grades 2–3 = 2 (moderate-to-severe injury). Neurological impairment was evaluated using the ASIA Impairment Scale [8].
Inclusion criteria
①Traumatic single-level thoracolumbar fracture patients confirmed by initial imaging assessment; ② Availability of complete clinical and imaging datasets, including: Preoperative anteroposterior/lateral radiographs of the thoracolumbar spine, CT with three-dimensional vertebral reconstruction. ③ MRI scans.
Exclusion criteria
①Patients with multi-segment or old thoracolumbar fractures (e.g., ≥ 2 vertebral levels involved); ② Non-traumatic thoracolumbar fractures, including pathological fractures (spinal tumor-associated fractures, infection-related fractures, or osteoporotic fractures); ③ Patients with pre-existing neurological impairments prior to the fracture event; ④ Patients with concomitant pelvic or lower limb fractures requiring surgical intervention; ⑤ Patients with incomplete clinical or imaging datasets (e.g., missing preoperative MRI or CT scans).
Research method
Two attending orthopedic surgeons and one senior chief surgeon were invited to study the mTLICS (Table 1) and TL AOSIS (Table 2) scoring criteria and were subsequently tested. After passing the assessment, the two attending surgeons independently scored the imaging data of the 100 patients. To minimize bias, the raters were blinded to the subsequent treatment methods and clinical outcomes of the patients. If the scores and treatment recommendations from both surgeons were consistent, the result was accepted. If there were discrepancies, the case was handed over to the chief surgeon for a final decision. This process helps to reduce the errors that younger doctors may make when assessing complex cases, thereby improving the consistency and accuracy of the scoring results. Statistical analysis of inconsistent factors between the two scoring systems was conducted to explore causes affecting classification, scoring, and surgical decision-making.
Table 1.
Modified thoracolumbar injury classification and severity score system (mTLICS)
| Injury Type | Features | Score |
|---|---|---|
| Morphological Injury | Compression | 1 |
| Burst fracture | 2 | |
| Displacement/rotation | 3 | |
| Distraction | 4 | |
| Neurological Impairment | No neurological deficit | 0 |
| Nerve root injury | 2 | |
| Complete spinal cord/conus medullaris injury | 2 | |
| Incomplete spinal cord/conus medullaris injury | 3 | |
| Cauda equina injury | 3 | |
| Posterior Ligamentous Complex (PLC) Injury | Intact | 0 |
| Suspected injury | 1 | |
| Injury | 2 | |
| Intervertebral Disc Injury Status | No injury | 0 |
| Mild injury | 1 | |
| Moderate-to-severe injury | 2 |
Table 2.
AO thoracolumbar injury classification system (TL AOSIS)
| Fracture Classification | Subclassification | Score | |
|---|---|---|---|
| A-Type | Compression fractures | ||
| A0 | Spinous/transverse process fracture | 0 | |
| A1 | Wedge fracture | 1 | |
| A2 | Split fracture | 2 | |
| A3 | Incomplete burst fracture | 3 | |
| A4 | Complete burst fracture | 5 | |
| B-Type | Tension band injury | ||
| B1 | Posterior column injury (bony dominant) | 5 | |
| B2 | Posterior column injury (ligament dominant) | 6 | |
| B3 | Disc-mediated anterior column injury | 7 | |
| C-Type | Translation/rotational displacement | 8 | |
| Neurological Status | |||
| N0 | Normal neurological function | 0 | |
| N1 | Transient neurological dysfunction | 2 | |
| N2 | Radiculopathy (symptoms/signs) | 2 | |
| N3 | Incomplete spinal cord/cauda equina injury | 3 | |
| N4 | Complete spinal cord injury | 3 | |
| NX | Unable to assess due to head trauma, sedation, or multi-system injuries | ||
| Patient-Specific Modifiers | |||
| M1 | Indeterminate posterior tension band (PLC) injury suggested by MRI/clinical exam | 1 | |
| M2 | Comorbidities influencing surgical decision-making | 0 | |
mTLICS is a modified TLICS classification system that incorporates an additional subclass, the “intervertebral disc injury status,” while reducing the scoring weight of the " PLC integrity” subclass. The total maximum score is 11. The “intervertebral disc injury status” subclass follows the Sander classification, which categorizes disc injuries into three grades: no injury (Grade 0), mild injury (Grade 1), and moderate-to-severe injury (Grades 2–3), corresponding to 0, 1, and 2 points, respectively. For the “PLC integrity” subclass, scores are assigned as follows: 0 points for intact PLC, 1 point for suspected injury, and 2 points for confirmed injury. The subclasses for fracture morphology and neurological impairment remain consistent with the TLICS system. The severity of the fracture is evaluated based on the total score (T) derived from all four subclasses (T = morphological injury + neurological impairment + PLC injury + disc injury), guiding clinical treatment decisions. T ≤ 3: Conservative management is recommended. T = 4: Either conservative or surgical treatment may be selected based on patient vital signs and quality-of-life requirements. T ≥ 5: Surgical intervention is indicated [5].
The TL AOSIS classifies thoracolumbar fracture morphology into three categories with subtypes: type A (compression injuries) — A0 (no or clinically insignificant injury, e.g., an isolated transverse- or spinous-process fracture), A1 (wedge-compression of a single endplate without posterior-wall involvement), A2 (split/pincer involving both endplates), A3 (incomplete burst, a single endplate with posterior-wall retropulsion), and A4 (complete burst, involving both endplates); type B (tension-band/distraction injuries) — B1 (transosseous posterior tension-band disruption, the “Chance” fracture), B2 (posterior tension-band disruption, bony and/or ligamentous), and B3 (anterior tension-band disruption, hyperextension); and type C (displacement/translation in any plane). The system comprises three components — fracture morphology, neurological status, and patient-specific modifiers — with a maximum total score of 13 (T = fracture morphology + neurological status + patient-specific modifiers). Treatment recommendations are: score < 4, conservative; score 4–5, either conservative or surgical depending on patient preference and surgeon judgment; score > 5, surgical [9].
Statistical methods
Continuous variables are presented as mean ± standard deviation and categorical variables as counts (percentages). Agreement between the two scoring systems was assessed from the paired 3 × 3 cross-tabulation of treatment categories using Cohen’s unweighted and linear-weighted kappa with 95% confidence intervals and the McNemar–Bowker test of symmetry; in the subgroup that reduced to a 2 × 2 table an exact McNemar test was used. Interobserver agreement before consensus adjudication was assessed with weighted kappa for the mTLICS treatment category, the TL AOSIS treatment category, the MRI disc-injury grade, and the posterior ligamentous complex (PLC) status. Kappa was interpreted according to Landis and Koch as slight (0.00–0.20), fair (0.21–0.40), moderate (0.41–0.60), substantial (0.61–0.80), and almost perfect (0.81–1.00) [10, 11]. All analyses were performed with IBM SPSS Statistics 25.0, and a two-sided P < 0.05 was considered statistically significant.
Results
The cohort comprised 100 patients (64 men, 36 women) with a mean age of 41.3 ± 6.9 years (range 25–55); injuries resulted from traffic accidents in 51 and falls in 49, most frequently at L1 (38%) and T12 (37%). Fifty-seven patients were neurologically intact (ASIA E) and 43 had neurological impairment (36 ASIA D, 7 ASIA C; no ASIA A or B). By AO morphology, 30 fractures were type A1, 59 A3, 1 A4, and 10 B2. The PLC was intact in 38 patients, suspected to be injured in 51, and injured in 11; the MRI disc-injury grade was none in 35, mild in 9, and moderate-to-severe in 56. Overall, 66 patients underwent surgery and 34 were treated conservatively. Baseline, neurological, morphological and treatment characteristics, both overall and stratified by neurological status (ASIA E vs. A–D), are summarized in Table 3. Actual treatment is reported for descriptive purposes only; because the cohort is retrospective and treatment decisions may have been informed by the scores, these data are not used to judge the diagnostic accuracy of either system.
Table 3.
Baseline characteristics of the cohort, overall and by neurological status
| Characteristic | Overall (n = 100) | ASIA E (n = 57) | ASIA A–D (n = 43) |
|---|---|---|---|
| Age, years, mean ± SD (range) | 41.3 ± 6.9 (25–55) | 42.2 ± 7.1 (25–55) | 40.1 ± 6.6 (27–53) |
| Sex | |||
| Male | 64 (64.0) | 33 (57.9) | 31 (72.1) |
| Female | 36 (36.0) | 24 (42.1) | 12 (27.9) |
| Injury mechanism | |||
| Traffic accident | 51 (51.0) | 26 (45.6) | 25 (58.1) |
| Fall | 49 (49.0) | 31 (54.4) | 18 (41.9) |
| Injured level | |||
| T11 | 7 (7.0) | 4 (7.0) | 3 (7.0) |
| T12 | 37 (37.0) | 23 (40.4) | 14 (32.6) |
| L1 | 38 (38.0) | 19 (33.3) | 19 (44.2) |
| L2 | 18 (18.0) | 11 (19.3) | 7 (16.3) |
| ASIA grade | |||
| E | 57 (57.0) | 57 (100.0) | 0 (0.0) |
| D | 36 (36.0) | 0 (0.0) | 36 (83.7) |
| C | 7 (7.0) | 0 (0.0) | 7 (16.3) |
| AO/TL morphological subtype | |||
| A1 | 30 (30.0) | 30 (52.6) | 0 (0.0) |
| A3 | 59 (59.0) | 16 (28.1) | 43 (100.0) |
| A4 | 1 (1.0) | 1 (1.8) | 0 (0.0) |
| B2 | 10 (10.0) | 10 (17.5) | 0 (0.0) |
| mTLICS morphology | |||
| Compression | 31 (31.0) | 31 (54.4) | 0 (0.0) |
| Burst | 69 (69.0) | 26 (45.6) | 43 (100.0) |
| Neurological status | |||
| No deficit | 57 (57.0) | 57 (100.0) | 0 (0.0) |
| Nerve-root injury | 4 (4.0) | 0 (0.0) | 4 (9.3) |
| Incomplete cord/conus | 39 (39.0) | 0 (0.0) | 39 (90.7) |
| Posterior ligamentous complex (PLC) | |||
| Intact | 38 (38.0) | 34 (59.6) | 4 (9.3) |
| Suspected | 51 (51.0) | 12 (21.1) | 39 (90.7) |
| Injured | 11 (11.0) | 11 (19.3) | 0 (0.0) |
| Disc degeneration grade (mTLICS 3-tier) | |||
| None | 35 (35.0) | 31 (54.4) | 4 (9.3) |
| Mild | 9 (9.0) | 9 (15.8) | 0 (0.0) |
| Moderate–severe | 56 (56.0) | 17 (29.8) | 39 (90.7) |
| mTLICS recommendation | |||
| Conservative | 32 (32.0) | 32 (56.1) | 0 (0.0) |
| Gray zone | 15 (15.0) | 11 (19.3) | 4 (9.3) |
| Surgery | 53 (53.0) | 14 (24.6) | 39 (90.7) |
| TL AOSIS recommendation | |||
| Conservative | 34 (34.0) | 34 (59.6) | 0 (0.0) |
| Gray zone | 20 (20.0) | 12 (21.1) | 8 (18.6) |
| Surgery | 46 (46.0) | 11 (19.3) | 35 (81.4) |
| Actual treatment received | |||
| Surgery | 66 (66.0) | 23 (40.4) | 43 (100.0) |
| Conservative | 34 (34.0) | 34 (59.6) | 0 (0.0) |
Across all 100 patients, mTLICS assigned 32 patients to conservative management, 15 to the gray zone, and 53 to surgery, whereas TL AOSIS assigned 34, 20, and 46, respectively. In the paired 3 × 3 cross-tabulation the two systems agreed in 86 of 100 patients (86.0%); agreement was substantial (unweighted κ = 0.773, 95% CI 0.666–0.881; linear-weighted κ = 0.820, 95% CI 0.726–0.914), and the McNemar–Bowker test showed no significant asymmetry between the two systems’ category assignments (χ² = 5.167, df = 3, P = 0.160; Table 4). Agreement remained substantial after stratification by neurological status. Among the 57 neurologically intact patients (ASIA E) the two systems agreed in 47 (82.5%; unweighted κ = 0.696, 95% CI 0.532–0.860; linear-weighted κ = 0.726; McNemar–Bowker χ² = 1.667, df = 3, P = 0.644). Among the 43 patients with neurological impairment (ASIA A–D) they agreed in 39 (90.7%; κ = 0.619, 95% CI 0.291–0.948); all of these patients fell into the gray-zone or surgery categories, leaving only four discordant pairs, so an exact binomial McNemar test (P = 0.125) was used in place of an unreliable asymptotic χ² (Table 5). Representative cases are shown in Figs. 1 and 2.
Table 4.
Paired 3 × 3 cross-tabulation of mTLICS and TL AOSIS treatment categories (n = 100)
| mTLICS \ TL AOSIS | Conservative | Gray zone | Surgery | Total |
|---|---|---|---|---|
| Conservative | 30 | 1 | 1 | 32 |
| Gray zone | 2 | 12 | 1 | 15 |
| Surgery | 2 | 7 | 44 | 53 |
| Total | 34 | 20 | 46 | 100 |
Observed agreement 86/100 (86.0%). Unweighted Cohen’s κ = 0.773 (95% CI 0.666–0.881); linear-weighted κ = 0.820 (0.726–0.914). McNemar–Bowker test of symmetry: χ² = 5.167, df = 3, P = 0.160. 95% CIs are asymptotic
Table 5.
Paired 3×3 cross-tabulation stratified by neurological status
| ASIA E (no neurological impairment, n = 57) | ||||
|---|---|---|---|---|
| mTLICS \ TL AOSIS | Conservative | Gray zone | Surgery | Total |
| Conservative | 30 | 1 | 1 | 32 |
| Gray zone | 2 | 8 | 1 | 11 |
| Surgery | 2 | 3 | 9 | 14 |
| Total | 34 | 12 | 11 | 57 |
| Observed agreement 47/57 (82.5%). Unweighted κ = 0.696 (95% CI 0.532–0.860); linear-weighted κ = 0.726 (0.564–0.887). McNemar–Bowker χ² = 1.667, df = 3, P = 0.644. | ||||
| ASIA A–D (with neurological impairment, n = 43) | ||||
| mTLICS \ TL AOSIS | Conservative | Gray zone | Surgery | Total |
| Conservative | 0 | 0 | 0 | 0 |
| Gray zone | 0 | 4 | 0 | 4 |
| Surgery | 0 | 4 | 35 | 39 |
| Total | 0 | 8 | 35 | 43 |
Observed agreement 39/43 (90.7%). κ = 0.619 (95% CI 0.291–0.948). Only 4 discordant pairs (the table reduces to 2×2); exact McNemar P = 0.125
Fig. 1.
A 48-year-old female presented with a complete burst fracture of L1 (AO Type A4) and was neurologically intact (ASIA E). MRI confirmed a moderate superior intervertebral disc injury with an intact PLC. Her scores placed her in the “gray zone” for both systems (mTLICS = 4, TL AOSIS = 5), and she was ultimately treated surgically
Fig. 2.
A 26-year-old male with an incomplete L1 burst fracture (AO Type A3) was neurologically intact. MRI confirmed a moderate disc injury. His mTLICS score of 4 placed him in the "gray zone," while the TL AOSIS score of 3 suggested conservative treatment. He ultimately underwent surgery
Before consensus adjudication, agreement between the two independent raters was high. For the treatment-category assignment, the linear-weighted κ was 0.882 (95% CI 0.810–0.954; observed agreement 90/100) for mTLICS and 0.753 (95% CI 0.646–0.861; 81/100) for TL AOSIS. The interobserver κ was 0.874 (95% CI 0.807–0.942; 88/100) for the MRI disc-injury grade and 0.683 (95% CI 0.557–0.808; 81/100) for PLC status (Table 6).
Table 6.
Interobserver reliability between the two independent raters
| Domain | n | Observed agreement | Unweighted κ (95% CI) | Linear-weighted κ (95% CI) | Strength |
|---|---|---|---|---|---|
| mTLICS treatment category | 100 | 90 (90.0) | 0.819 (0.713–0.924) | 0.882 (0.810–0.954) | Perfect |
| TL AOSIS treatment category | 100 | 81 (81.0) | 0.674 (0.546–0.802) | 0.753 (0.646–0.861) | Substantial |
| Disc-injury grade | 100 | 88 (88.0) | 0.797 (0.695–0.899) | 0.874 (0.807–0.942) | Perfect |
| PLC status | 100 | 81 (81.0) | 0.681 (0.559–0.804) | 0.683 (0.557–0.808) | Substantial |
The two systems gave a different treatment category in 14 of the 100 patients, and a further 12 patients were placed in the gray zone by both systems; all 26 cases are grouped by shared component pattern, with fracture morphology, neurological status, PLC status, disc-injury grade, the component sub-scores and totals of both systems, and the actual treatment, in Table 7. The 14 discordant cases followed two distinct patterns. In 11 of them mTLICS assigned the higher (more operative) category: every one of these 11 patients had a moderate-to-severe intervertebral disc injury (disc sub-score 2), so the disc component raised the mTLICS total while the TL AOSIS morphology score — most often A3 (incomplete burst, 3 points) — kept that system in the conservative or gray-zone range; 10 of these 11 patients were ultimately treated surgically. In the remaining 3 discordant cases TL AOSIS assigned the higher category, driven by morphological features that mTLICS scores only coarsely — a complete-burst (A4) fracture, a posterior tension-band (B2) injury, and the M1 modifier — and 2 of these 3 were treated conservatively. The 12 patients placed in the gray zone by both systems were neurologically intact burst fractures with mild disc injury and suspected PLC injury (8 patients) or burst fractures with nerve-root injury (4 patients); 11 of the 12 underwent surgery. Overall, 22 of these 26 patients were ultimately treated surgically. Consistent with the study design, these actual-treatment data are presented descriptively and are not used to determine which system is correct.
Table 7.
AO/TL morphological subtypes and A3-versus-A4 detail. (a) AO/TL subtype distribution
| AO subtype | Overall, n (%) | ASIA E | ASIA A–D |
|---|---|---|---|
| A1 | 30 (30.0) | 30 | 0 |
| A3 | 59 (59.0) | 16 | 43 |
| A4 | 1 (1.0) | 1 | 0 |
| B2 | 10 (10.0) | 10 | 0 |
By AO subtype the cohort comprised 30 A1, 59 A3, 10 B2, and a single A4 (complete-burst) fracture; the clinical fracture-morphology descriptor and the AO subtype were concordant in 95 of 100 patients (Table 8). Typical cases are Figs. 1 and 2.
Table 8.
Gray-zone and discordant cases (n = 26), grouped by component pattern, with both systems’ scores and actual treatment
| n | ASIA | Morphology | Neurology | PLC | Disc | mTLICS (total → cat) | TL AOSIS (subtype·total → cat) | Actual treatment |
|---|---|---|---|---|---|---|---|---|
| Part A — Discordant cases (systems disagree on category; n = 14, 7 component patterns) | ||||||||
| 1 | E | Compression | None | Suspected | 0 | 2 → Cons | A3 · 4 → Gray | Conservative |
| 1 | E | Burst | None | Intact | 1 | 3 → Cons | A4 · 6 → Surg | Surgery |
| 2 | E | Burst | None | Intact | 2 | 4 → Gray | A3 · 3 → Cons | 1 surgery / 1 conservative |
| 1 | E | Burst | None | Intact | 2 | 4 → Gray | B2 · 6 → Surg | Conservative |
| 2 | E | Burst | None | Injured | 2 | 6 → Surg | A3 · 3 → Cons | Surgery |
| 3 | E | Burst | None | Suspected | 2 | 5 → Surg | A3 · 4 → Gray | Surgery |
| 4 | D | Burst | Incomplete cord | Suspected | 2 | 8 → Surg | A3 · 5 → Gray | Surgery |
| Part B — Concordant gray-zone cases (both systems = gray zone; n = 12, 2 component patterns) | ||||||||
| 8 | E | Burst | None | Suspected | 1 | 4 → Gray | A3 · 4 → Gray | 7 surgery / 1 conservative |
| 4 | D | Burst | Nerve root | Intact | 0 | 4 → Gray | A3 · 5 → Gray | Surgery |
Discussion
In clinical practice, the management of thoracolumbar fractures relies on scoring systems to guide treatment decisions, with the TLICS and TL AOSIS being the most widely used. However, the clinical application of TLICS has revealed certain limitations, particularly its insufficient evaluation of burst fractures, which can lead to an underestimation of spinal instability and potentially result in delayed kyphosis in patients treated conservatively. By contrast, TL AOSIS provides a more detailed morphological classification and quantitative scoring that some studies suggest may be better than TLICS for evaluating complete burst fractures. A key factor in stability that is not explicitly scored as a primary component in these systems is the intervertebral disc; studies have shown that significant disc injury occurs in over half of traumatic thoracolumbar fractures and critically impacts stability. To address this gap, Lu et al. proposed the mTLICS, which incorporates a specific subclass for “intervertebral disc injury status”. Although this modification may, in theory, capture an additional injury feature, whether it confers any clinical advantage over the well-established TL AOSIS remains unknown. Our study addresses this question by directly comparing the treatment recommendations generated by the two systems, focusing on their agreement and the reasons for discordance rather than on which system is better.
This study showed good agreement between mTLICS and TL AOSIS for surgical guidance of thoracolumbar fractures. Among 57 patients without neurological impairment, the rates of surgical recommendation were 24.6% (14/57) and 19.2% (11/57), respectively. Conservative management was recommended for 32 patients by mTLICS and 34 patients by TL AOSIS, including 3 patients with mTLICS total score ≥ 4 versus TL AOSIS total score < 6. Among 43 patients with neurological impairment, the surgical recommendation rates for mTLICS and TL AOSIS were 90.7% (39/43) and 81.4% (35/43), respectively. Specifically, 8 patients with TL AOSIS had a total score of 4–5 and were recommended to be treated conservatively and surgically, while the remaining 35 patients were recommended to be treated surgically. mTLICS reclassified 4 patients from the “conservative or surgical” category under TL AOSIS into the definitive surgical category.
Guidance for patients with neurological impairment
Our analysis revealed no statistically significant differences between the two scoring systems in guiding treatment for patients with neurological impairments. Clinically, most physicians tend to recommend surgical intervention for patients with neurological deficits. Several scholars have emphasized that early surgical treatment is critical for restoring neurological function in such cases [12, 13]. Haldrup et al. [14] pointed out that early surgery is beneficial for patients with neurological impairment. In this study, surgical treatment was finally selected for patients with conservative or surgical treatment of neurological impairment. In clinical practice, we also found that when patients have nerve injury, doctors and patients are more willing to choose surgical treatment, so that patients can get a better prognosis. This is consistent with Muratore et al. [15] that prompt surgical treatment can stabilize the spine, restore neurological function, and eliminate the long-term risk of kyphosis.
Guidance for patients without neurological impairment
For patients with burst thoracolumbar fractures without neurological impairment, There is some controversy regarding the choice of appropriate treatment modality. This study found that although there was no statistically significant difference in the overall treatment recommendations between the two scoring systems, we observed a trend where the mTLICS system was more inclined to recommend surgical intervention for incomplete burst fractures involving intervertebral disc and suspected PLC injury. When mTLICS was used to evaluate these patients, the total score was 2 + 0+1 + 2 = 5, and surgical treatment was recommended. However, the total score using TL AOSIS assessment was 3 + 0+0 = 3 and conservative treatment was recommended. Finally, these patients were treated with surgery according to the doctor’s recommendation. Although the mTLICS only assigns a score of 2 for the evaluation of burst fractures as TLICS, the addition of the evaluation of intervertebral disc injury increases the total score and enters the surgical range. However, TL AOSIS was evaluated only by fracture morphology, and incomplete burst fractures were assigned a score of only 3 points, which was in the conservative treatment interval.
This difference stems from the different design emphases of the scoring systems, mTLICS is more detailed in the evaluation of intervertebral disc injury, which can be divided into three types: no injury, mild injury and moderate to severe injury. TL AOSIS, by contrast, has no standalone disc-injury category and captures the disc only indirectly—through the B3 morphological subtype (disc-mediated anterior-column/hyperextension injury) and, when the tension band is indeterminate, through the M1 modifier; in A-type burst fractures with a concomitant traumatic disc lesion but no hyperextension mechanism it therefore assigns no disc-specific points, which can leave some patients with disc injury in the conservative range. This difference in classification behaviour, rather than any demonstrated clinical advantage, is the principal source of the discordance observed in this subgroup. Some studies have pointed out that intervertebral disc injury has a non-negligible impact on the long-term prognosis of patients, and it is of great significance to pay attention to intervertebral disc injury when treating thoracolumbar fractures [16]. Because the intervertebral disc plays an important role in stress conduction of the thoracolumbar segment and is a key factor to maintain the height and stability of the corresponding segment, its injury will cause a decrease in intradiscal pressure, which will cause structural and metabolic changes, and eventually lead to intervertebral disc degeneration [17–19], A number of studies have found that about 60% of spinal instability cases caused by thoracolumbar fractures are caused by intervertebral disc injury. At the same time, vertebral collapse, delayed kyphosis and discogenic low back pain after conservative treatment are also closely related to intervertebral disc injury [20–24]. Consequently, for thoracolumbar fractures involving the disc, there is a potential tendency for mTLICS to favor surgical treatment. a tendency that aligns with clinical practice because long-term spinal stability is one of the factors surgeons consider when opting for surgery. Early intervention targeting destabilizing factors can reduce the risk of subsequent kyphosis and disc degeneration and promote long-term spinal stability.
We observed a potential trend of increased surgical recommendations with mTLICS, which, although not statistically significant, may raise concerns about over-treatment. However, a deeper analysis of the cases with inconsistent treatment recommendations revealed that these patients all had severe intervertebral disc injuries. From a prognostic perspective, existing evidence suggests that traumatic disc injuries significantly increase the long-term risk of progressive kyphotic deformity and adjacent segment degeneration [22]. Therefore, through effective evaluation and intervention for this high-risk subgroup, their long-term prognosis and societal benefits may potentially be better than conservative treatment. This view is supported by a recent international multicenter study, which reported that, although the long-term outcomes of both treatments were similar, surgical intervention significantly reduced pain and disability during rehabilitation for patients with neurologically intact burst fractures, providing sustained economic benefits [25]. In this regard, the higher surgical-recommendation rate with mTLICS may reflect more sensitive detection of a destabilizing feature rather than overtreatment. However, because the present study did not assess clinical outcomes, the possibility that this tendency leads to overtreatment in some patients cannot be excluded; this trade-off should be weighed explicitly and tested in outcome-based studies.
Treatment guidance for the “gray zone”
Both scoring systems have a “gray zone” where treatment recommendations are unclear, requiring clinical decisions to be made based on the specific clinical and patient conditions. In this study, 12 patients simultaneously fell into the gray zone of both mTLICS and TL AOSIS. An additional 3 patients were classified as gray zone only by mTLICS, with TL AOSIS recommending conservative treatment for 2 and surgery for 1. There were also 8 patients who were classified as gray zone only by TL AOSIS; in mTLICS assessment, 1 was recommended for conservative treatment and 7 for surgery. Stratified by neurological status, in 57 patients without neurological impairment, 8 were classified in the gray zone by both scoring systems, 3 were classified as gray zone only by mTLICS, with TL AOSIS recommending 2 for conservative treatment and 1 for surgery; 4 were classified as gray zone only by TL AOSIS, with mTLICS recommending 1 for conservative treatment and 3 for surgery. In 43 patients with neurological impairment, 4 were classified in the gray zone by both, and 4 were classified as gray zone only by TL AOSIS, with mTLICS recommending surgery.
This distribution reflects the subtle differences in how the two scoring systems assess different types of injuries. When a patient is in the gray zone of TLICS, Load-Sharing Classification Score often needs to be referenced for further evaluation [26]. However, the mTLICS system addresses this shortcoming by incorporating intervertebral disc injury into its scoring system. Therefore, when a burst fracture involves the intervertebral disc, the patient may fall into the gray zone or surgical range, leading to further treatment. Although TL AOSIS has no dedicated disc-injury axis and addresses the disc only indirectly (via the B3 subtype and the M1 modifier), it partly compensates through fracture morphology—assigning a higher score to complete (A4) than to incomplete (A3) burst fractures—and thereby captures some disc-associated instability through the morphological severity gradient.
A meta-analysis study has shown that for patients in the gray zone, surgery, compared to conservative treatment, can effectively prevent spinal kyphosis, lower VAS scores, and reduce the occurrence of complications [27]. These differences are of significant clinical importance: when mTLICS classifies a patient as gray zone while TL AOSIS recommends conservative treatment, it indicates that mTLICS is more sensitive to intervertebral disc injury; when TL AOSIS classifies a case as gray zone while mTLICS classifies it for surgery, it reflects the direct impact of morphological assessment on surgical decision-making. Overall, the two scoring systems showed substantial agreement in their treatment recommendations. The greater tendency of mTLICS to flag burst fractures involving intervertebral disc injury may be useful for refining treatment-stratification algorithms for ‘gray zone’ patients, but this remains hypothesis-generating and requires validation against clinical outcomes before any advantage can be claimed.
Conclusion
mTLICS and TL AOSIS are both useful decision-support tools and show substantial concordance in guiding the treatment of thoracolumbar fractures (κ = 0.773). For burst fractures involving the intervertebral disc, mTLICS tends to recommend surgery more frequently; this reflects a difference in classification behaviour rather than a demonstrated clinical advantage, and the subgroup differences observed were not statistically significant. This study has several limitations. First, it is a single-centre, retrospective analysis with a relatively small sample; the retrospective design and the absence of follow-up preclude any causal or prognostic inference. Second, it compared the treatment recommendations of the two systems rather than clinical outcomes—no functional, radiographic, or patient-reported outcomes (e.g., kyphotic progression, disc degeneration, ODI, or quality of life) and no systematic mid- to long-term follow-up were available. Third, we did not perform multivariable analysis to adjust for potential confounders such as age and fracture morphology. Fourth, although interobserver reliability before consensus was substantial to high for both scoring systems and for disc-injury and PLC assessment, the MRI disc-injury grading is derived from the Sander classification, which has not undergone broad multicentre validation and whose prognostic value and correlation with spinal instability remain uncertain. Finally, because osteoporotic and pathological fractures were excluded, the cohort comprised younger adults (25–55 years) with traumatic injuries, which limits generalizability to elderly or osteoporotic patients, in whom thoracolumbar fractures are common. Consequently, these data support agreement between the two systems rather than a clinical advantage of either; prospective, multicentre studies with standardized outcome measures and adequate follow-up are needed to determine whether the greater sensitivity of mTLICS to disc injury translates into better patient outcomes.
Acknowledgements
Not applicable.
Institutional review board statement
This study was approved by the Ethics Committee of the Second Affiliated Hospital of Zhejiang Chinese Medical University (ZCMU 2024 NO.106-01).
Generative AI
During the preparation of this work the author(s) used ChatGPT (OpenAI) in order to improve the language of the manuscript. After using this tool, the author(s) reviewed and edited the content as needed and take full responsibility for the content of the publication.
Authors' contributions
ZC A Conceptualization, Methodology, Supervision, Formal Analysis, Project administration, Writing - review & editing. H Z Data curation, Formal Analysis, Software, Validation, Writing - original draft. JW F Investigation, Data curation, Writing - original draft. B W Investigation, Data curation, Writing - review & editing. JY D Investigation, Writing - review & editing. HB L Data curation. C C Investigation, Writing - review & editing. KF L Investigation, Writing - review & editing. FZ S Investigation, Writing - review & editing. BB T Investigation, Writing - review & editing. TY L Supervision, Project administration, Writing - review & editing. JZ S Investigation, Resources, Writing - review & editing. LQ D Project administration, Writing - review & editing. WZ F Supervision, Validation, Writing – review & editing.
Funding
No funding was received for conducting the study.
Data availability
The original contributions presented in the study are included in the article, further inquiries can be directed to the corresponding author.
Declarations
Consent for publication
Not applicable.
Competing interests
The authors declare no competing interests.
Footnotes
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Han Zhang and Junwei Feng contributed equally to this work and share first authorship.
Contributor Information
Zhongcheng An, Email: andoctor@sina.cn.
Wangzhe Fan, Email: fanwangzhe2008@163.com.
References
- 1.Holmes JF, Miller PQ, Panacek EA, Lin S, Horne NS, Mower WR. Epidemiology of thoracolumbar spine injury in blunt trauma. Acad Emerg Med. 2001;8(9):866–72. [DOI] [PubMed] [Google Scholar]
- 2.Zileli M, Sharif S, Fornari M. Incidence and Epidemiology of Thoracolumbar Spine Fractures: WFNS Spine Committee Recommendations. Neurospine. 2021;18(4):704–12. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Bajamal AH, Permana KR, Faris M, Zileli M, Peev NA. Classification and Radiological Diagnosis of Thoracolumbar Spine Fractures: WFNS Spine Committee Recommendations. Neurospine. 2021;18(4):656–66. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Zhongcheng A, Yuchen Z, Guoqiang W, Hao W, Liqiang D. Thoracolumbar AO spine injury score versus thoracolumbar injury classification and severity score system in guiding thoracolumbar fracture surgery. In: Zhongcheng A, Yuchen Z, Guoqiang W, Hao W, Liqiang D, editors. Chinese Journal of Orthopaedic Trauma. 2020;22(7):598–603.
- 5.Lu WJ, Zhang J, Deng YG, Jiang WY. Reliability and repeatability of a modified thoracolumbar spine injury classification scoring system. Front Surg. 2022;9:1054031. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Lee HM, Kim HS, Kim DJ, Suk KS, Park JO, Kim NH. Reliability of magnetic resonance imaging in detecting posterior ligament complex injury in thoracolumbar spinal fractures. Spine (Phila Pa 1976). 2000;25(16):2079–84. [DOI] [PubMed] [Google Scholar]
- 7.Sander AL, Laurer H, Lehnert T, El Saman A, Eichler K, Vogl TJ, et al. A clinically useful classification of traumatic intervertebral disk lesions. AJR Am J Roentgenol. 2013;200(3):618–23. [DOI] [PubMed] [Google Scholar]
- 8.Kirshblum S, Snider B, Eren F, Guest J. Characterizing Natural Recovery after Traumatic Spinal Cord Injury. J Neurotrauma. 2021;38(9):1267–84. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Vaccaro AR, Oner C, Kepler CK, Dvorak M, Schnake K, Bellabarba C, et al. AOSpine thoracolumbar spine injury classification system: fracture description, neurological status, and key modifiers. Spine (Phila Pa 1976). 2013;38(23):2028–37. [DOI] [PubMed] [Google Scholar]
- 10.Durmaz A, İlter MH, Tuzlali H. Is magnetic resonance imaging needed for decision making diagnosis and treatment of thoracic and lumbar vertebral fractures? Eur J Orthop Surg traumatology: orthopedie traumatologie. 2023;33(1):107–15. [DOI] [PubMed] [Google Scholar]
- 11.Landis JR, Koch GG. The measurement of observer agreement for categorical data. Biometrics. 1977;33(1):159–74. [PubMed] [Google Scholar]
- 12.Lee DY, Park YJ, Song SY, Hwang SC, Kim KT, Kim DH. The Importance of Early Surgical Decompression for Acute Traumatic Spinal Cord Injury. Clin Orthop Surg. 2018;10(4):448–54. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Burke JF, Yue JK, Ngwenya LB, Winkler EA, Talbott JF, Pan JZ, et al. In Reply: Ultra-Early (< 12 Hours) Surgery Correlates With Higher Rate of American Spinal Injury Association Impairment Scale Conversion After Cervical Spinal Cord Injury. Neurosurgery. 2019;85(2):E401–2. [DOI] [PubMed] [Google Scholar]
- 14.Haldrup M, Schwartz OS, Kasch H, Rasmussen MM. Early decompressive surgery in patients with traumatic spinal cord injury improves neurological outcome. Acta Neurochir (Wien). 2019;161(10):2223–8. [DOI] [PubMed] [Google Scholar]
- 15.Muratore M, Allasia S, Viglierchio P, Abbate M, Aleotti S, Masse A, et al. Surgical treatment of traumatic thoracolumbar fractures: a retrospective review of 101 cases. Musculoskelet Surg. 2021;105(1):49–59. [DOI] [PubMed] [Google Scholar]
- 16.Peng Z, Cui Z, Kuang X, Yu C, Ruan Y, Li C, et al. Intervertebral disc injury is the mainspring for the postoperative increase in Cobb Angle after thoracolumbar burst fracture. J Orthop Surg (Hong Kong). 2022;30(2):10225536221088753. [DOI] [PubMed] [Google Scholar]
- 17.Howlett DC, Drinkwater KJ, Mahmood N, Illes J, Griffin J, Javaid K. Radiology reporting of osteoporotic vertebral fragility fractures on computed tomography studies: results of a UK national audit. Eur Radiol. 2020;30(9):4713–23. [DOI] [PubMed] [Google Scholar]
- 18.Kweh BTS, Tee JW, Dandurand C, Vaccaro AR, Lorin BM, Schnake K, et al. The AO Spine Thoracolumbar Injury Classification System and Treatment Algorithm in Decision Making for Thoracolumbar Burst Fractures Without Neurologic Deficit. Global spine J. 2024;14(1suppl):S32–40. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Wang W, Pei B, Pei Y, Shi Z, Kong C, Wu X, et al. Biomechanical effects of posterior pedicle fixation techniques on the adjacent segment for the treatment of thoracolumbar burst fractures: a biomechanical analysis. Comput Methods Biomech Biomed Engin. 2019;22(13):1083–92. [DOI] [PubMed] [Google Scholar]
- 20.Mohd Isa IL, Teoh SL, Mohd Nor NH, Mokhtar SA. Discogenic Low Back Pain: Anatomy, Pathophysiology and Treatments of Intervertebral Disc Degeneration. Int J Mol Sci. 2022;24(1):208. [DOI] [PMC free article] [PubMed]
- 21.Hu C, Zhong W, Chen Z, Peng J, Li J, Tang K, et al. Comparison of the Outcomes between AO Type B2 Thoracolumbar Fracture with and without Disc Injury after Posterior Surgery. Orthop Surg. 2022;14(9):2119–31. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Cui XL, Yang WM, Ding A, Zhu J, Zhang W, Wu H, et al. Characteristics and mechanical mechanisms of intervertebral disc degeneration in old thoracolumbar fractures with kyphosis: clinical observations and finite element analyses. BMC Musculoskelet Disord. 2024;25(1):1040. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Chow N, Gregory DE. The effect of intervertebral disc damage on the mechanical strength of the annulus fibrosus in the adjacent segment. spine journal: official J North Am Spine Soc. 2023;23(12):1935–40. [DOI] [PubMed] [Google Scholar]
- 24.Desmoulin GT, Pradhan V, Milner TE. Mechanical Aspects of Intervertebral Disc Injury and Implications on Biomechanics. Spine. 2020;45(8):E457–64. [DOI] [PubMed] [Google Scholar]
- 25.Dandurand C, Öner CF, Schnake KJ, Bransford RJ, Schroeder GD, Dea N, et al. Surgical versus nonsurgical treatment of thoracolumbar burst fractures in neurologically intact patients: a cost-utility analysis. spine journal: official J North Am Spine Soc. 2025;25(7):1494–507. [DOI] [PubMed] [Google Scholar]
- 26.Alan N, Donohue J, Ozpinar A, Agarwal N, Kanter AS, Okonkwo DO, et al. Load-Sharing Classification Score as Supplemental Grading System in the Decision-Making Process for Patients With Thoracolumbar Injury Classification and Severity 4. Neurosurgery. 2021;89(3):428–34. [DOI] [PubMed] [Google Scholar]
- 27.Li T, Yan J, Liu X, Hu J, Wang F. Efficacy and Safety of Conservative Treatment Compared With Surgical Treatment for Thoracolumbar Fracture With Score 4 Thoracolumbar Injury Classification and Severity (TLICS): A Systematic Review and Meta-analysis. Clin Spine Surg. 2024;37(5):230–41. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
The original contributions presented in the study are included in the article, further inquiries can be directed to the corresponding author.