Abstract
Background
Spinal curve flexibility can be assessed reliably by fulcrum bending (FB) radiograph for surgical planning. However, it remains unknown how curve correctability of rigid curves are different from flexible curves. This study aims to investigate the relationship of curve flexibility and surgical outcomes in adolescent idiopathic scoliosis, to investigate for any differences of surgical outcomes in rigid and non-rigid curves, and to explore whether postoperative changes were different in rigid curves.
Methods
This retrospective study included patients diagnosed with AIS, aged ≤18 years and underwent posterior spinal fusion with >2 years of postoperative follow-up at two affiliated hospitals between 2010 and 2022. Radiographic parameters were collected, including coronal Cobb angle of major curve, curve type, FB flexibility, fusion length, and implant density. Relationship of flexibility and correction rate was tested. Rigid curves (FB flexibility <50 %) were compared with non-rigid curves, in terms of curve correction rate, FB correction index (FBCI) and loss of correction at postoperative 2 years.
Results
A total of 190 patients (83.7 % females) were included. Preoperative major curve was 60.2° (SD 10.9°) with 64.4 % (SD 17.2 %) flexibility, 21.6 % (n = 41) of the patients had rigid curves. FB flexibility correlated with curve correction rate (immediate postoperative rs: 0.281, p < 0.001). After covariate adjustment, correction rate for rigid and non-rigid curves were 69.4 % and 75.5 %, with mean difference −6.0 % (95 %CI −10.6 % to −1.4 %, p = 0.013). Rigid curves had higher FBCI (175.5 % vs 110.4 %; mean difference: 65.1 %, 95 %CI 53.2 % to 77.0 %, p < 0.001). Fewer rigid curves demonstrated loss of correction, with comparable changes in correction rates (−2.5 % vs −4.6 %, p = 0.124).
Conclusion
Spinal curve flexibility significantly correlated with curve correction rate and FBCI. Rigid curves demonstrated ∼70 % correction rate, all with FBCI >100 %, indicating curve correction achieved more than estimated preoperatively. Despite rigid curves had lower curve correction rate, correction lost at 2-year follow-up was comparable between rigid and non-rigid curves.
Graphical abstract
1. Introduction
Surgical intervention for adolescent idiopathic scoliosis (AIS) aims to prevent curve progression and achieve curve correction which can be maintained long-term.1,2 Among the numerous factors associated with the success of correction surgery such as curve magnitude and curve type, spinal curve flexibility is a key factor under consideration in surgical planning. More flexible curve allows greater extent of derotation during correction manoeuvres to achieve the desired angles.3 Rigid curves are difficult to manipulate intraoperatively and frequently prompt osteotomy or anterior releases as additional measures for maximizing surgical correction.4 Fulcrum bending radiograph is one of the preoperative radiographic assessment techniques which provides the best assessment of spinal flexibility, in comparison to traction, active side bending or supine lateral bending approach.5, 6, 7 In addition to determining the lowest instrumented vertebra (LIV) and the fusion levels, fulcrum bending radiographs also assist surgeons to maximize the preservation of motion segments.8 Despite curve flexibility can be assessed reliably by fulcrum bending radiograph, little is known about how much curve correction can be achieved in rigid curves, whether rigid curves behave differently from flexible curves after surgical correction, and whether they have any preoperative difference held responsible for maintaining the corrected curves. Therefore, this study aims to investigate the relationships of spinal curve flexibility and surgical outcomes in rigid curves and non-rigid curves of AIS; to compare the surgical outcomes of rigid versus non-rigid curves, and to explore whether rigid curves demonstrate postoperative changes differently after posterior spinal fusion.
2. Methods
2.1. Study design
This was a retrospective study of patients who had undergone posterior spinal fusion for AIS at the spine specialist division of two affiliated hospitals from January 2010 to December 2022. Ethics approval was obtained with local ethics committee (UW 24–363). Inclusion criteria include patients diagnosed with AIS undergoing posterior spinal fusion at or before the age of 18. Patients were excluded if they had scoliosis of non-idiopathic nature, had previous spinal surgery, missing preoperative fulcrum bending spine radiographs, lacked postoperative 2-year follow-up, or with an osteotomy (Ponte or Smith-Petersen) as part of the surgery. Our specialist centre has a standardized routine protocol of fulcrum bending radiographs as part of the preoperative assessment of AIS surgical cases, specifically for assessing segmental spinal flexibility. All patients underwent posterior spinal fusion with pedicle screw constructs, and one to two hooks were used in the case where pedicle screws could not be inserted. Surgical techniques were standardized with the correction manoeuvres being mainly: rod rotation at concave side of the curve, followed by concave distraction and convex side compression (with/without translation), or cantilever correction, or rod rotation and direct vertebral rotation. Inferior articular process excision or release was routinely performed.
2.2. Data collection
Preoperative patient characteristics were collected, including chronological age, gender, skeletal maturity assessed by Risser staging9 and Sanders staging.10 Radiographic parameters were recorded, including coronal Cobb angle of major and minor curves (in degrees, °), Cobb angle on fulcrum bending radiographs, thoracic kyphosis and lumbar lordosis from lateral spine radiographs, curve type (modified Lenke curve type, with sagittal thoracic modifier according to the sagittal alignment between T5 and T12, or with lumbar curve modifier (A, B, C) based on the position of the apex of the lumbar spine to a centre sacral vertical line),11,12 vertebral level of fusion, upper and lower instrumented vertebrae (UIV, LIV), surgical manoeuvre and surgical implant materials, were collected. Coronal Cobb angle of major curve was measured preoperatively, immediate postoperative (within 1 week) and at post-operative 2 years.
2.3. Study outcome
Rigid curves were defined as having preoperative spinal curve flexibility of the major curve <50 %,13, 14, 15, 16, 17 which was assessed using fulcrum bending radiographs taken immediately before surgery (Fulcrum bending flexibility = (Preoperative Cobb angle – fulcrum bending Cobb angle)/Preoperative Cobb angle x 100 %).5,18 The 50 % curve flexibility was used as a threshold for surgical planning, as any curve flexibility below this threshold requires additional surgical approach to achieve optimal curve correction in posterior spinal fusion. Fulcrum bending correction index (FBCI) was calculated by (Correction rate/Flexibility) x 100 %15,18 (Fig. 1). Surgical outcomes referred to curve correction rate, which was calculated as (Preoperative Cobb angle – postoperative Cobb angle)/Preoperative Cobb angle x 100 %. Change of correction rate (%) was calculated by postoperative 2-year correction rate minus immediate postoperative correction rate, given the major curve Cobb angle change must be > 5° which allows us to ascertain such change of curve size was beyond measurement error.19 Loss of correction was defined as any negative change of >1 % of correction rate and >5° increase of major curve Cobb angle.
Fig. 1.
A FBCI of 205.0 % as demonstrated by a Lenke curve type 3C, major Cobb angle at T5-T11 67.0° T11-L3 57.1°, with a correction rate of 63.6 % and fulcrum bending flexibility of 31.0 %.
2.4. Statistical analysis
Data normality was tested and non-parametric measures were used for skewed data. Descriptive statistics were presented in mean and standard deviation (SD) and count and percentage (%) for continuous and categorical parameters respectively. Radiographic parameters were assessed independently by two raters. Intrarater and interrater reliabilities were assessed using intraclass correlation coefficient (ICC) and were found satisfactory (respective ICC: 0.985 (95 % CI 0.940–0.996) and 0.922 (95 % CI 0.640 to 0.981)).
Patients were first stratified into rigid and non-rigid curve groups. The relationships of spinal curve flexibility, FBCI, surgical outcomes and the difference between curve correction and flexibility rates were investigated using Spearman's correlation tests. Correlation coefficient (rs) ranges from −1 to 1, with fair, moderate and very strong relationships being indicated by 0.30 to 0.59, 0.60 to 0.79 and > 0.80 respectively.20 Intergroup comparison was performed to screen for any statistically significant differences in preoperative Cobb angle and surgical parameters, through Mann-Whitney U test for continuous parameters and Chi-square test/Fisher-Freeman-Halton Exact test with Bonferroni method for categorical parameters. Surgical outcomes in terms of correction rate (%) were compared between rigid and non-rigid curve groups by Quade one-way analysis of covariance (ANCOVA), with adjustment for major curve Cobb angle and any covariates with intergroup differences as identified in the previous step. Change in correction rate (%) at postoperative 2-year follow-up was evaluated using Quade ANCOVA. FBCI and flexibility rates were also compared between those patients with or without correction loss.
Statistical analyses were performed using SPSS v.29.0 (IBM, USA). A p-value of <0.05 was set as statistical significance.
3. Results
A total of 190 patients (83.7 % females) were recruited (Fig. 2), with an overall mean age of 14.7 years (SD 1.7) at surgery (Table 1). Preoperative major curve was 60.2° (SD 10.9°), and Lenke curve type I (49.5 %) and II (25.3 %) had a larger presentation. Immediate preoperative characteristics of the study cohort were presented in Table 1. Spinal flexibility of the whole cohort was 64.4 % (SD 17.2 %), and 21.6 % (n = 41) of the patients had rigid curves. Differences between rigid and non-rigid curves were detected, in curve types, fusion length, number of implants, and implant density (Table 1, Table 2).
Fig. 2.
Patient recruitment flowchart.
Table 1.
Study cohort preoperative characteristics.
| Parameters mean (SD), median | Whole cohort n = 190 | Rigid curves n = 41 | Non-rigid curves n = 149 | p valuea |
|---|---|---|---|---|
| Age (years) | 14.7 (1.7), 14.8 | 14.8 (1.6), 15.2 | 14.6 (1.7), 14.7 | 0.410 |
| Genderb | ||||
| Females | 159 | 33 | 126 | 0.532 |
| Males | 31 | 8 | 23 | |
| Risser stagingb | ||||
| 0- | 3 | 0 | 3 | 0.830 |
| 0+ | 19 | 3 | 16 | |
| 1 | 22 | 3 | 19 | |
| 2 | 31 | 6 | 25 | |
| 3 | 34 | 7 | 27 | |
| 4 | 50 | 13 | 37 | |
| 4+ | 13 | 3 | 10 | |
| 5 | 18 | 6 | 12 | |
| Sanders stagesb (n = 176) | ||||
| 1 | 1 | 0 | 1 | 0.418 |
| 2 | 5 | 2 | 3 | |
| 3a | 5 | 1 | 4 | |
| 3b | 16 | 2 | 14 | |
| 4 | 19 | 3 | 16 | |
| 5 | 9 | 3 | 6 | |
| 6 | 24 | 2 | 22 | |
| 7a | 31 | 6 | 25 | |
| 7b | 33 | 11 | 22 | |
| 8 | 33 | 7 | 26 | |
| Curve typesb | ||||
| 1AL | 35 | 15 | 79 | 0.011∗ |
| 1AR | 39 | |||
| 1B | 10 | |||
| 1C | 10 | |||
| 2A | 41 | 12 | 36 | |
| 2B | 3 | |||
| 2C | 4 | |||
| 3C | 20 | 10 | 10 | |
| 4C | 5 | 2 | 3 | |
| 5C | 7 | 1 | 6 | |
| 6C | 16 | 1 | 15 | |
| Curve magnitude (Cobb angle, degrees) | ||||
| Major curve | 60.2 (10.9), 58.2 | 62.4 (10.5), 59.9 | 59.6 (11.0), 57.4 | 0.108 |
| Minor curve (n = 88) | 46.3 (11.2), 45.5 | 47.3 (9.0), 46.9 | 45.8 (12.0), 45.2 | 0.240 |
SD: standard deviation, n: number.
∗ Statistical significance at p < 0.05.
Mann-Whitney U test.
Chi-squared Test/Fisher-Freeman-Halton Exact Test with Bonferroni correction and adjusted p value.
Table 2.
Surgical parameters of rigid and non-rigid curves.
| Parameters | Whole cohort n = 190 | Rigid curves n = 41 | Non-rigid curves n = 149 | p valuea |
|---|---|---|---|---|
| Preoperative | ||||
| Fulcrum bending flexibility (%) | 64.4 (17.2), 65.4 | 40.6 (7.3), 42.1 | 70.9 (12.8), 69.4 | <0.001∗ |
| Major curve (°) | 60.2 (10.9), 58.2 | 62.4 (10.5), 59.9 | 59.6 (11.0), 57.4 | 0.108 |
| Thoracic Kyphosis (°) | 16.4 (12.8), 13.6 | 19.4 (15.3), 19.1 | 15.6 (11.9), 13.0 | 0.250 |
| Lumbar lordosis (°) | 51.9 (12.1), 52.1 | 52.2 (15.2), 53.0 | 51.8 (11.1), 51.6 | 0.796 |
| Surgical parameters | ||||
| Upper instrumented vertebrae – count (n) | ||||
| T10 | 3 | 1 | 2 | N/A |
| T11 | 3 | 0 | 3 | |
| T2 | 34 | 8 | 26 | |
| T3 | 5 | 1 | 4 | |
| T4 | 27 | 12 | 15 | |
| T5 | 83 | 15 | 68 | |
| T6 | 28 | 4 | 24 | |
| T7 | 3 | 0 | 3 | |
| T8 | 2 | 0 | 2 | |
| T9 | 2 | 0 | 2 | |
| Lower instrumented vertebrae | ||||
| L1 | 54 | 11 | 43 | N/A |
| L2 | 24 | 5 | 19 | |
| L3 | 33 | 8 | 25 | |
| L4 | 19 | 5 | 14 | |
| L5 | 2 | 0 | 2 | |
| T10 | 1 | 0 | 1 | |
| T11 | 8 | 1 | 7 | |
| T12 | 49 | 11 | 38 | |
| Number of levels fused | 9.8 (2.1), 9.0 | 10.3 (1.9), 10.0 | 9.6 (2.1), 9.0 | 0.052 |
| Number of implants (screws, hooks) | 11.4 (2.6), 11.0 | 12.6 (3.0), 12.0 | 11.1 (2.4), 10.0 | 0.004∗ |
| Implant density | 0.86 (0.09), 0.88 | 0.83 (0.09), 0.83 | 0.87 (0.09), 0.89 | 0.011∗ |
| Implant materialsb - count (n) | ||||
| Both titanium | 181 | 40 | 141 | >0.99 |
| Both cobalt chromium | 2 | 0 | 2 | |
| Convex titanium, concave cobalt chromium | 7 | 1 | 6 | |
| Correction manoeuvreb - count (n) | ||||
| Concave rod rotation, followed by concave distraction and convex compression ± translation | 176 | 41 | 135 | 0.136 |
| Cantilever correction | 12 | 0 | 12 | |
| Rod rotation and direct vertebral rotation | 2 | 0 | 2 | |
SD: standard deviation, N/A: not applicable, n: number.
∗ Statistical significance at p < 0.05.
Mann-Whitney U test.
Fisher-Freeman-Halton Exact Test or Fisher's Exact Test with Bonferroni method and adjusted p value.
3.1. Relationships of spinal flexibility and surgical outcomes
Rigid curves had negative correlation with correction rate immediately postoperative (rpb: −0.194, p = 0.007) (Table 3). Spinal flexibility and curve correction rate had weak to fair correlations in the whole study cohort (immediate postoperative rs: 0.280, p < 0.001; postoperative 2 years rs: 0.166, p = 0.022). Such relationship was found in non-rigid curves but not in rigid curves when analysed for each group. Correlations were found between FBCI and curve correction rate (immediate postoperative rs: 0.288, p < 0.001; postoperative 2-year rs: 0.285, p = 0.001). These relationships by FBCI were stronger in rigid curves (immediate postoperative rs: 0.689, p < 0.001; postoperative 2-year rs: 0.531, p < 0.001) (Fig. 3). Fulcrum bending flexibility had strong negative correlation with FBCI (rs: −0.795, p < 0.001), and moderate negative correlation with the difference of actual correction rate and flexibility rate (rs: −0.741, p < 0.001).
Table 3.
Relationships of spinal curve flexibility and FBCI with surgical outcome at immediate postoperatively and at post-operative 2 years.
| Surgical outcome parameters | Rigid curve (yes vs no) |
Fulcrum bending flexibility (%) |
FBCI (%) |
|||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Whole cohort n = 190 |
Rigid curves n = 41 |
Non-rigid curves n = 149 |
Whole cohort n = 190 |
Rigid curves n = 41 |
Non-rigid curves n = 149 |
|||||||||
| rpb | p value | rs | p value | rs | p value | rs | p value | rs | p value | rs | p value | rs | p value | |
| Immediate postoperative | ||||||||||||||
| Correction rate (%) | −0.194 | 0.007∗ | 0.281 | <0.001∗ | 0.083 | 0.605 | 0.245 | 0.003∗ | 0.288 | <0.001∗ | 0.689 | <0.001∗ | 0.486 | <0.001∗ |
| FBCI (%) | 0.651 | <0.001∗ | −0.795 | <0.001∗ | −0.538 | <0.001∗ | −0.671 | <0.001∗ | – | – | – | – | – | – |
| TK(°) | 0.107 | 0.145 | −0.074 | 0.312 | 0.201 | 0.213 | −0.034 | 0.683 | 0.102 | 0.164 | −0.212 | 0.189 | 0.110 | 0.180 |
| LL (°) | 0.001 | 0.998 | −0.024 | 0.748 | −0.059 | 0.713 | −0.037 | 0.655 | 0.049 | 0.501 | 0.044 | 0.787 | 0.066 | 0.424 |
| Postoperative 2-year | ||||||||||||||
| Correction rate (%) | −0.134 | 0.066 | 0.166 | 0.022∗ | 0.148 | 0.354 | 0.104 | 0.207 | 0.285 | <0.001∗ | 0.531 | <0.001∗ | 0.449 | <0.001∗ |
| TK (°) | 0.122 | 0.097 | −0.073 | 0.324 | 0.173 | 0.287 | −0.031 | 0.706 | 0.038 | 0.607 | −0.261 | 0.104 | 0.031 | 0.714 |
| LL (°) | −0.030 | 0.679 | −0.076 | 0.299 | 0.077 | 0.636 | −0.158 | 0.056 | 0.070 | 0.339 | −0.188 | 0.245 | 0.142 | 0.087 |
| Difference [Correction rate (%)– Flexibility rate (%)] | 0.539 | <0.001∗ | −0.741 | <0.001∗ | −0.273 | 0.084 | −0.634 | <0.001∗ | N/A | N/A | N/A | N/A | N/A | N/A |
FBCI: fulcrum bending correction index, TK: thoracic kyphosis, LL: Lumbar lordosis, N/A: not applicable due to same/similar entities.
rs: Spearman correlation coefficient from Spearman correlation tests; rpb: Point-biserial correlation coefficient from point-biserial correlation tests.
∗ Statistical significance at p < 0.05.
Fig. 3.
Scatter plot of FBCI and Immediate Postoperative Correction Rate.
3.2. Surgical outcomes of rigid versus non-rigid curves
After adjusting for major curve Cobb angle, curve type and implant density, mean correction rate for rigid and non-rigid curves were 69.4 % and 75.5 % respectively, with mean difference −6.0 % (95 % CI −10.6 %, −1.4 %, p = 0.013) (Table 4). Rigid curves had higher FBCI (adjusted mean 175.5 % versus non-rigid 110.4 %; mean difference: 65.1 % (95 % CI 53.2 %, 77.0 %, p < 0.001). All rigid curves had FBCI >100 %. Correction rates based on postoperative 2-year Cobb angles had no significant difference between rigid and non-rigid curves. Correction achieved was much higher than indicated by fulcrum bending flexibility rate for rigid curves (28.4 % versus non-rigid 4.6 %; mean difference: 23.8 (95 % CI 18.2 %, 29.3 %, p < 0.001). Comparison without covariate adjustment had similar results. (Appendix I).
Table 4.
Intergroup comparison of surgical outcomes between rigid and non-rigid curves.
| Parameters | Rigid curves |
Non-rigid curves |
Mean difference (95 % CI) Rigid minus Non-rigid curves | Standard error | p valueb |
|---|---|---|---|---|---|
| Estimated Marginal Meansa (95 % CI) | |||||
| Preoperative Fulcrum bending flexibility (%) | 41.0 (37.4, 44.7) | 70.8 (68.9, 72.7) | −29.8 (−34.0, −25.6) | 2.1 | <0.001∗ |
| Surgical outcomes | |||||
| Immediate postoperative Correction rate (%) | 69.4 (65.4, 73.5) | 75.5 (73.4, 77.6) | −6.0 (−10.6, −1.4) | 2.3 | 0.013∗ |
| FBCI (%) | 175.5 (165.0, 185.9) | 110.4 (104.9, 115.8) | 65.1 (53.2, 77.0) | 6.0 | <0.001∗ |
| Immediate postoperative TK (°) | 18.9 (16.3, 21.5) | 16.7 (15.4, 18.1) | 2.2 (−0.8, 5.2) | 1.5 | 0.189 |
| Immediate postoperative LL (°) | 48.6 (45.1, 52.2) | 48.6 (46.8, 50.5) | 0.01 (−4.0, 4.0) | 2.0 | 0.799 |
| Postoperative 2-year correction rate (%) | 66.2 (61.8, 70.7) | 70.6 (68.2, 72.9) | −4.3 (−9.4, 0.7) | 2.6 | 0.089 |
| Difference [Postoperative correction rate (%) – Preoperative flexibility rate(%)] | 28.4 (23.5, 33.3) | 4.6 (2.1, 7.2) | 23.8 (18.2, 29.3) | 2.8 | <0.001∗ |
FBCI: fulcrum bending correction index, TK: thoracic kyphosis, LL: lumbar lordosis, CI: confidence interval.
∗ Statistical significance at p < 0.05.
Adjusted for preoperative major Cobb angle, curve type, implant density.
Quade ANCOVA analysis and post hoc test.
3.3. Postoperative changes of curve correction and rigid curves
There was 29.5 % (n = 56) of the patients experienced loss of curve correction at postoperative 2 years. Fewer patients with rigid curves experienced correction loss than those with non-rigid curves (22.0 % (n = 9) of rigid curves versus 31.5 % (n = 47) of non-rigid curves) but without statistical significance (Table 5). The changes of curve correction rates at post-operative 2 years between rigid and non-rigid curves were comparable (rigid −2.5 % versus non-rigid −4.6 %, mean difference 2.1 % (95 % CI −0.6 %, 4.8 %, p = 0.236). For patients with correction loss, there was a mean of 14.5 % (SD 5.8 %) reduction of correction rate without significant difference in rigid or non-rigid curves. Subgroup analysis of patients with loss of correction reveals their immediate postoperative correction rate was 79.7 % and 76.9 % for rigid and non-rigid curves (mean difference 2.9 % (95 % CI −6.0 %, 11.7 %, p = 0.417). Rigid curves had lower fulcrum bending flexibility (mean 39.5 % versus 70.8 % for non-rigid curves) yet higher FBCI (mean 220.3 % versus 110.8 %) (both at p < 0.001, Table 5). Adjusted mean fulcrum bending flexibility and FBCI were comparable between curves with or without loss of correction (Table 6).
Table 5.
Postoperative changes of curve correction rate per rigid and non-rigid groups.
| Surgical outcome | Rigid curves |
Non-rigid curves |
Mean difference (95 % CI) Rigid minus Non-rigid curves | Standard error | p valueb |
|---|---|---|---|---|---|
| Estimated Marginal Meansa (95 % CI) | |||||
| Change of correction rate at 2 years (%) | −2.5 (−4.9, −0.1) | −4.6 (−5.8, −3.3) | 2.1 (−0.6, 4.8) | 1.4 | 0.236 |
| Loss of correction (count)c | – | ꭓ2 | 0.233 | ||
| Yes (n = 56) | 9 | 47 | df (1): 1.423 | ||
| No (n = 134) | 32 | 102 | |||
| Patients with loss of correction n = 56 |
|||||
|---|---|---|---|---|---|
| Rigid curves |
Non-rigid curves |
Mean difference (95 % CI) Rigid minus Non-rigid curves | Standard error | p valueb | |
| Estimated Marginal Meansa (95 % CI) | |||||
| Correction loss (%) | −13.8 (−17.6, −10.0) | −14.6 (−16.3, −13.0) | 0.9 (−3.3, 5.0) | 2.1 | 0.846 |
| Immediate postoperative Correction rate (%) | 79.7 (71.7, 87.8) | 76.9 (73.4, 80.4) | 2.9 (−6.0, 11.7) | 4.4 | 0.417 |
| Fulcrum bending flexibility (%) | 39.5 (31.8, 47.1) | 70.8 (67.5, 74.1) | −31.3 (−39.7, −22.9) | 4.2 | <0.001∗ |
| FBCI (%) | 220.3 (193.1, 247.4) | 110.8 (99.2, 122.5) | 109.4 (79.7, 139.2) | 14.8 | <0.001∗ |
n: number, %: percentage, CI: confidence interval, df: degree of freedom.
∗ Statistical significance at p < 0.05.
Adjusted for preoperative major Cobb angle, curve type, implant density.
Quade ANCOVA analysis and post hoc test.
Chi-square test.
Table 6.
Comparison of spinal flexibility and FBCI between patients with or without loss of curve correction.
| Loss of Correction |
|||||
|---|---|---|---|---|---|
| Yes n = 56 |
No n = 134 |
||||
| Estimated Marginal Meansa (95 % CI) | Mean difference (95 % CI) Rigid minus Non-rigid curves | Standard error | p valueb | ||
| Fulcrum bending flexibility (%) | 65.7 (61.2, 70.2) | 63.9 (61.0, 66.8) | 1.8 (−3.5, 7.1) | 2.7 | 0.295 |
| FBCI (%) | 128.1 (116.9, 139.4) | 122.9 (115.6, 130.2) | 5.3 (−8.2, 18.7) | 6.8 | 0.849 |
CI: confidence interval.
Adjusted for preoperative major Cobb angle, curve type, implant density.
Quade ANCOVA analysis and post hoc test.
4. Discussion
Spinal curve flexibility is important for surgical planning in AIS. It provides crucial indication of the extent of structural curves and guide the selection of fusion levels, which can aid surgeons in adopting appropriate surgical approach to achieve the desired correction.21 In this cohort of AIS surgical patients, the extent of curve correction correlated with whether the curve was flexible or rigid. Although rigid curves had lower curve correction rate (by ∼6 %), their correction rate was significantly higher than that estimated on fulcrum bending radiograph when compared to non-rigid curves (whose FBCI was ∼65 % lower). Rigid and non-rigid curves demonstrated similar changes of curve correction rates at postoperative 2 years. Among those who experienced a loss of correction, curve correction rate and the amount of correction loss (14 %–15 %) were comparable between rigid and flexible curves.
4.1. Relationships of flexibility and curve correction by surgery
The use of fulcrum bending radiographs in our routine practice contributes to the accuracy of the estimated curve flexibility, which allows for more precise prediction of postoperative correction.5,22,23 Spinal curve flexibility was found correlating with curve correction achieved overall in the whole study cohort. Further probing reveals that fulcrum bending flexibility correlated specifically with immediate postoperative correction rate in flexible curves, while correlations by FBCI were stronger in rigid curves. Importantly, curve flexibility had strong correlations with the correction index, i.e. the amount of surgical correction achieved (fulcrum bending flexibility and FBCI, r: −0.795, p < 0.001). Also, spinal curve flexibility was found to have a negative correlation with the difference in correction rate versus that anticipated according to fulcrum bending flexibility. As comparing to flexible curves, rigid curves achieved much more correction than estimated preoperatively (Fig. 4). Such difference can be due to the rigidity of the curves and its correctability, surgeon's decision on the amount of correction, as well as the higher flexibility under anaesthesia intraoperatively.24
Fig. 4.
Scatter plot of Fulcrum Bending Flexibility Rate and Difference between Correction Rate and Flexibility.
4.2. Surgical outcomes between rigid and non-rigid curves
In this study, the curve correction rate achieved through posterior spinal fusion appeared significantly less (mean difference of 6.0 %) for rigid curves than flexible curves, taking into account preoperative major curve magnitude, curve type, and implant density. In fact, all rigid curves had FBCI over 100 %, indicating that the surgical procedures were very effective in terms of curve correction consuming all of the inherent curve flexibility.15,25 Despite similar strength of correlations between FBCI and curve correction rate were found in our rigid curves as those in previous study of Lenke 1 and 2 curves by Kwan et al.,26 the significantly less curve correction in rigid curves with high FBCI suggests two possibilities. One is that rigid curves may not require such high correction rate or there is difficulty to achieve as greater curve flexibility allows better derotation for correcting the deformity.27 Second, the desired correction rate as per surgical planning must account for the consequences of excessive correction. One may strive for maximal curve correction for improving patients with very rigid thoracic curves which can restrict pulmonary function,28,29 whereas over-correction was found related to postoperative coronal imbalance as in Lenke curve type 125 or type 5C thoracolumbar curves,30, 31, 32, 33 problem with coronal decompensation, lumbar curve progression below thoracic fused segment, or distal junction kyphosis.34,35 Despite the longer fusion length and higher number of implants, rigid curves had slightly lower implant density than flexible curves in this study, hence all intergroup comparison of correction rates were adjusted with implant density which is known for correlating to curve correction.36 This raises the question of whether rigid curves require higher implant density, particularly for thoracic curves, whereby the selection of LIV allows lumbar mobility and potential risk of distal junctional kyphosis.
Rigid curves can achieve approximately 70 % correction rate in this study, as comparing to the mean correction rate of 45 % in the study by Luk et al. more than two decades ago.15 The mean difference of 6 % correction rate between rigid and flexible curves was far less than that detected in Luk's study, despite the very similar values of fulcrum bending flexibility rates between the two studies. A much higher FBCI (rigid 176 % and flexible 110 % versus Luk's study rigid 107 % and flexible 96 %) was also observed. The ability to achieve higher correction rate must be examined further as this study included all curve types which we adjusted for as a covariate during comparative analyses of rigid and flexible curves, while previous study examined thoracic curves only. Further investigation about other associating factors including surgical approaches and techniques can possibly delineate if higher correction rates were due to the advancement of spinal fusion surgeries and surgical approach.
4.3. Loss of correction and spinal rigidity
Despite their lower correction rate achieved, rigid curves had fewer patients with correction loss but without statistical significance (rigid 22.0 % vs non-rigid 31.5 %, p = 0.233) at postoperative 2 years (Fig. 5). Among those patients who lost some curve correction, the correction rate achieved at surgery was very similar between rigid and non-rigid curves, yet the rigid curves had significantly 109 % higher FBCI (with rigid curve's FBCI doubled that of non-rigid curve groups) and their flexibility was ∼30 % less. Spinal flexibility may play a role in whether curve correction can be maintained. Several factors can be associated with the loss of correction, including correction manoeuvres, screw strategy or the extent of fusion.37,38 Future research is also warranted for determining whether correction loss is related to segmental flexibility,39 discal or vertebral wedging,40,41 which are beyond the scope of this study.
Fig. 5.
Case presentation (a) Non-rigid curve (fulcrum bending flexibility: 61.2 %) Lenke curve type 1AL, T5 – T11 major Cobb angle 66.6° had a correction rate of 57.3 %, FBCI 93.6 %. (b) Rigid curve (fulcrum bending flexibility: 40.9 %) Lenke curve type 1AL, T5 – T10 major Cobb angle 58.2° had a correction rate of 57.8 %, FBCI 141.4 %. Both curves had no loss of correction at 2-year follow-up.
4.4. Limitations
Limitations of this study include the lack of data about patient's activity level and muscle training between immediate postoperative and 2 years after. Also, the fewer number of lumbar curves in the study cohort requires further study focusing on each curve type. As there is no standardized protocol for the use of grade 2 osteotomy in our institute, we have excluded all the cases with osteotomy in order to eliminate its confounding effect on the amount of potential curve correction, However, this can lead to the exclusion of very rigid curves as well. Specific in-depth study of rigid curves with or without osteotomy should be conducted. It is desirable to have more severe curves with very severe rigidity, which may not be presented too frequently as our school screening programme for scoliosis is well-established.42 AIS diagnosis and intervention are performed early before curves deteriorate to severe curvature. Future investigations on other surgical outcomes, including truncal balance, are essential. Lastly, three-dimensional reconstruction of the spinal column is required ideally in order to comprehensively appreciate curve correction more than the coronal plane or sagittal alignment. This could only be achieved when all patients having biplanar stereoradiography.
5. Conclusion
For patients with AIS undergoing posterior spinal fusion, spinal curve flexibility correlated with curve correction rate and FBCI. All of the rigid curves (with fulcrum bending flexibility <50 %) were effectively treated as indicated by their FBCI >100 %, i.e. the amount of curve correction achieved was more than that estimated preoperatively. Rigid curves demonstrated approximately 70 % correction rate, with 6 % significantly lower curve correction than non-rigid curves. The amount of loss of correction at 2-year follow-up was comparable for rigid curves with the more flexible curves. Further investigations should focus on whether curve flexibility plays a role in curve correction loss, and if any preoperative predictors can be identified to increase the likelihood of maintaining the curve correction achieved by spinal fusion.
Guardian/patient's consent
This research did not require guardian/patient's consent as this was a retrospective study and they were waived by the local ethics committee.
Ethical review committee statement
Ethics approval was obtained from the institutional review board of the University of Hong Kong/Hospital Authority Hong Kong West Cluster (IRB reference number: UW 24–363).
Author contributions
PWHC: Study design, methodology, statistical analysis, project administration, manuscript-original draft and editing.
VYTH: Methodology, investigation, data curation, manuscript-review and editing.
STYC: Data curation, investigation, manuscript-review and editing.
JLKL: Data curation, manuscript-review and editing.
GCCC: Data curation, manuscript-review and editing.
JPYC: Conceptualization, study design, methodology, resources, data curation, supervision, manuscript-review and editing.
Ethical statement
We have obtained ethics approval for this study:
Institutional Review Board of the University of Hong Kong/Hospital Authority Hong kong West Cluster (HKU/HA HKW IRB).
UW 24-363.
Funding statement
This research was supported by the Li Shu Fan Medical Foundation Professorship in Orthopaedic Surgery.
Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Acknowledgement
The authors would like to thank Mr. Alex Suen and Mr. Arthur Shou for their help in this project.
Footnotes
This article is part of a special issue entitled: Severe Rigid scoliosis published in Journal of Clinical Orthopaedics and Trauma.
Contributor Information
Prudence Wing Hang Cheung, Email: gnuehcp6@hku.hk.
Victoria Yuk Ting Hui, Email: vichui@connect.hku.hk.
Samuel Tin Yan Cheung, Email: samuel00@hku.hk.
James Long Ki Lin, Email: jamesllk@connect.hku.hk.
Garvin Chi Chun Cheung, Email: garccc18@connect.hku.hk.
Jason Pui Yin Cheung, Email: cheungjp@hku.hk.
Appendix I.
Table 1.
Surgical outcomes of rigid and non-rigid curves without covariate adjustment
| Surgical outcome mean (SD), median | Whole cohort n = 190 | Rigid curves n = 41 | Non-rigid curves n = 149 | p valuea |
|---|---|---|---|---|
| Correction rate immediately postoperatively (%) | 74.2 (13.2), 74.5 | 69.3 (13.6), 69.2 | 75.5 (12.9), 75.6 | 0.011∗ |
| FBCI (%) | 124.4 (43.7), 114.4 | 177.1 (57.4), 174.6 | 109.9 (23.9), 105.5 | <0.001∗ |
| Immediate postoperative thoracic kyphosis (°) | 17.2 (8.3), 16.9 | 18.9 (9.2), 18.3 | 16.7 (8.0), 16.8 | 0.205 |
| Immediate postoperative lumbar lordosis (°) | 48.6 (11.2), 48.7 | 48.6 (11.2), 48.9 | 48.6 (11.3), 48.5 | 0.985 |
| Correction rate at postoperative 2 years (%) | 69.6 (14.3), 68.9 | 66.0 (13.3), 66.4 | 70.6 (14.5), 70.9 | 0.060 |
| Difference [Postoperative correction rate (%) – Preoperative flexibility rate (%)] | 9.8 (18.4), 8.7 | 28.6 (15.0), 28.9 | 4.6 (15.7) 3.7 | <0.001∗ |
FBCI: fulcrum bending correction index, SD: standard deviation.
∗ Statistical significance at p < 0.05.
†Mann-Whitney U test.
References
- 1.Wagner S.C., Lehman R.A., Lenke L.G. Surgical management of adolescent idiopathic scoliosis. Semin Spine Surg. 2015;27:33–38. doi: 10.1053/j.semss.2015.01.008. [DOI] [Google Scholar]
- 2.Baghdadi S., Baldwin K. Selection of fusion levels in adolescent idiopathic scoliosis. Curr Rev Musculoskelet Med. 2024;17:23–36. doi: 10.1007/s12178-023-09876-6. 20231214. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Senkoylu A., Cetinkaya M. Correction manoeuvres in the surgical treatment of spinal deformities. EFORT Open Rev. 2017;2:135–140. doi: 10.1302/2058-5241.2.170002. 20170511. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Saifi C., Laratta J., Petridis P., et al. Vertebral column resection for rigid spinal deformity. Glob Spine J. 2017;7 doi: 10.1177/2192568217699203. [DOI] [Google Scholar]
- 5.Cheung K.M., Luk K.D. Prediction of correction of scoliosis with use of the fulcrum bending radiograph. J Bone Joint Surg Am. 1997;79:1144–1150. doi: 10.2106/00004623-199708000-00005. [DOI] [PubMed] [Google Scholar]
- 6.Klepps S.J., Lenke L.G., Bridwell K.H., et al. Prospective comparison of flexibility radiographs in adolescent idiopathic scoliosis. Spine. 2001;26:E74–E79. doi: 10.1097/00007632-200103010-00002. [DOI] [PubMed] [Google Scholar]
- 7.Vaughan J.J., Winter R.B., Lonstein J.E. Comparison of the use of supine bending and traction radiographs in the selection of the fusion area in adolescent idiopathic scoliosis. Spine (Phila Pa 1976) 1996;21:2469–2473. doi: 10.1097/00007632-199611010-00012. [DOI] [PubMed] [Google Scholar]
- 8.Jain A., Ahuja K., Roberts S.B., et al. Techniques of deformity correction in adolescent idiopathic Scoliosis-A narrative review of the existing literature. J Clin Med. 2025;14 doi: 10.3390/jcm14072396. [DOI] [Google Scholar]
- 9.Bitan F., Veliskakis K., Campbell B. Differences in the risser grading systems in the United States and France. Clin Orthop Relat Res. 2005;436:190–195. doi: 10.1097/01.blo.0000160819.10767.88. [DOI] [Google Scholar]
- 10.Sanders J.O., Browne R.H., McConnell S.J., et al. Maturity assessment and curve progression in girls with idiopathic scoliosis. J Bone Joint Surg Am. 2007;89:64–73. doi: 10.2106/jbjs.F.00067. [DOI] [Google Scholar]
- 11.Lenke L.G., Bridwell K.H., Blanke K., et al. Radiographic results of arthrodesis with cotrel-dubousset instrumentation for the treatment of adolescent idiopathic scoliosis. A five to ten-year follow-up study. J Bone Joint Surg Am. 1998;80:807–814. doi: 10.2106/00004623-199806000-00004. [DOI] [PubMed] [Google Scholar]
- 12.Dang N.R., Moreau M.J., Hill D.L., et al. Intra-observer reproducibility and interobserver reliability of the radiographic parameters in the spinal deformity study Group's AIS radiographic measurement manual. Spine. 2005;30:1064–1069. doi: 10.1097/01.brs.0000160840.51621.6b. [DOI] [PubMed] [Google Scholar]
- 13.Ameri E., Behtash H., Mobini B., et al. Predictors of curve flexibility in adolescent idiopathic scoliosis: a retrospective study of 100 patients. Acta Med Iran. 2015;53:182–185. [PubMed] [Google Scholar]
- 14.Li M., Zhu X-d, Cheung K.M., et al. Surgical treatment for scoliosis extending to main thoracic spine by keyvertebral-screws technique (KVST) J Med Coll PLA. 2007;22:115–120. doi: 10.1016/S1000-1948(07)60025-2. [DOI] [Google Scholar]
- 15.Luk K.D., Cheung K.M., Lu D.S., et al. Assessment of scoliosis correction in relation to flexibility using the fulcrum bending correction index. Spine (Phila Pa 1976) 1998;23:2303–2307. doi: 10.1097/00007632-199811010-00011. [DOI] [PubMed] [Google Scholar]
- 16.Delman C., Cage J.M., Lausé G., et al. Anterior and posterior fusion for large, rigid idiopathic scoliosis: does implant density matter? World Neurosurg. 2020;134:e37–e45. doi: 10.1016/j.wneu.2019.08.123. 20190828. [DOI] [PubMed] [Google Scholar]
- 17.Sridhar V.S.P.R.J. Effect of flexibility and implant density on curve correction in adolescent idiopathic scoliosis. Int J Orthop Sci. 2017;3(3):746–750. doi: 10.22271/ortho. [DOI] [Google Scholar]
- 18.Cheung W.-Y., Luk K.D.K. Fulcrum bending radiograph and its applications. ArgoSpine News J. 2012;24:73–77. doi: 10.1007/s12240-012-0045-y. [DOI] [Google Scholar]
- 19.Morrissy R.T., Goldsmith G.S., Hall E.C., et al. Measurement of the cobb angle on radiographs of patients who have scoliosis. Evaluation of intrinsic error. J Bone Joint Surg Am. 1990;72:320–327. [PubMed] [Google Scholar]
- 20.Chan Y.H. Biostatistics 104: correlational analysis. Singap Med J. 2003;44:614–619. [Google Scholar]
- 21.Samartzis D., Leung Y., Shigematsu H., et al. Selection of fusion levels using the fulcrum bending radiograph for the management of adolescent idiopathic scoliosis patients with alternate level pedicle screw strategy: clinical decision-making and outcomes. PLoS One. 2015;10 doi: 10.1371/journal.pone.0120302. 20150813. [DOI] [Google Scholar]
- 22.Ting Hui VY., Cheung S.T.Y., Cheung J.P.Y., et al. How accurate are fulcrum bending radiographs in estimating postoperative outcomes in adolescent idiopathic scoliosis? A systematic review and meta-analysis. Clin Orthop Relat Res. 2025 doi: 10.1097/corr.0000000000003468. [DOI] [Google Scholar]
- 23.Khodaei M., Pachêco-Pereira C., Trac S., et al. Radiographic methods to estimate surgical outcomes based on spinal flexibility assessment in patients who have adolescent idiopathic scoliosis: a systematic review. Spine J. 2018;18:2128–2139. doi: 10.1016/j.spinee.2018.06.344. [DOI] [PubMed] [Google Scholar]
- 24.Liu R.W., Teng A.L., Armstrong D.G., et al. vol. 35. Spine; 2010. (Comparison of Supine Bending, Push-Prone, and Traction Under General Anesthesia Radiographs in Predicting Curve Flexibility and Postoperative Correction in Adolescent Idiopathic Scoliosis). [Google Scholar]
- 25.Ohrt-Nissen S., Cheung P.W.H., Kawasaki S., et al. Curve overcorrection predicts coronal imbalance in selective thoracic fusion in adolescent idiopathic scoliosis. Glob Spine J. 2024;14:856–861. doi: 10.1177/21925682221124526. 20220831. [DOI] [Google Scholar]
- 26.Kwan M.K., Zeyada H.E., Chan C.Y. Prediction of curve correction using alternate level pedicle screw placement in patients with Adolescent Idiopathic Scoliosis (AIS) lenke 1 and 2 using supine Side Bending (SB) and Fulcrum Bending (FB) radiograph. Spine (Phila Pa 1976) 2015;40:1605–1612. doi: 10.1097/brs.0000000000001087. [DOI] [PubMed] [Google Scholar]
- 27.Hamzaoglu A., Talu U., Tezer M., et al. Assessment of curve flexibility in adolescent idiopathic scoliosis. Spine (Phila Pa 1976) 2005;30:1637–1642. doi: 10.1097/01.brs.0000170580.92177.d2. [DOI] [PubMed] [Google Scholar]
- 28.Johari J., Sharifudin M.A., Ab Rahman A., et al. Relationship between pulmonary function and degree of spinal deformity, location of apical vertebrae and age among adolescent idiopathic scoliosis patients. Singap Med J. 2016;57:33–38. doi: 10.11622/smedj.2016009. [DOI] [Google Scholar]
- 29.Dobbs M.B., Lenke L.G., Kim Y.J., et al. Anterior/posterior spinal instrumentation versus posterior instrumentation alone for the treatment of adolescent idiopathic scoliotic curves more than 90 degrees. Spine (Phila Pa 1976) 2006;31:2386–2391. doi: 10.1097/01.brs.0000238965.81013.c5. [DOI] [PubMed] [Google Scholar]
- 30.Fang Y., Li J., Hu Z., et al. Postoperative coronal imbalance in lenke 5C adolescent idiopathic scoliosis: Evolution, risk factors, and clinical implications. Neurospine. 2024;21:903–912. doi: 10.14245/ns.2448544.272. 20240930. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Wang Y., Bünger C.E., Zhang Y., et al. Lowest instrumented vertebra selection for Lenke 5C scoliosis: a minimum 2-year radiographical follow-up. Spine (Phila Pa 1976) 2013;38:E894–E900. doi: 10.1097/BRS.0b013e31829537be. [DOI] [PubMed] [Google Scholar]
- 32.Banno T., Yamato Y., Oba H., et al. Risk factors and clinical impact of persistent coronal imbalance after posterior spinal fusion in thoracolumbar/lumbar idiopathic scoliosis. J Neurosurg Spine. 2022;37:883–892. doi: 10.3171/2022.5.Spine22385. 20220715. [DOI] [PubMed] [Google Scholar]
- 33.Chan C.Y., Kwan M.K., Saravanan S., et al. The assessment of immediate post-operative scoliosis correction using pedicle screw system by utilising the fulcrum bending technique. Med J Malaysia. 2007;62:33–35. [PubMed] [Google Scholar]
- 34.Lowe T.G., Lenke L., Betz R., et al. Distal junctional kyphosis of adolescent idiopathic thoracic curves following anterior or posterior instrumented fusion: incidence, risk factors, and prevention. Spine (Phila Pa 1976) 2006;31:299–302. doi: 10.1097/01.brs.0000197221.23109.fc. [DOI] [PubMed] [Google Scholar]
- 35.Fischer C.R., Kim Y. Selective fusion for adolescent idiopathic scoliosis: a review of current operative strategy. Eur Spine J. 2011;20:1048–1057. doi: 10.1007/s00586-011-1730-9. 20110309. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Chen J., Yang C., Ran B., et al. Correction of Lenke 5 adolescent idiopathic scoliosis using pedicle screw instrumentation: does implant density influence the correction? Spine (Phila Pa 1976) 2013;38:E946–E951. doi: 10.1097/BRS.0b013e318297bfd4. [DOI] [PubMed] [Google Scholar]
- 37.Galbusera F., Cina A., Panico M., et al. The importance of curve severity, type and instrumentation strategy in the surgical correction of adolescent idiopathic scoliosis: an in silico clinical trial on 64 cases. Sci Rep. 2021;11:1799. doi: 10.1038/s41598-021-81319-z. 20210119. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Hyun S.J., Lenke L.G., Kim Y., et al. The incidence of Adding-On or distal junctional kyphosis in adolescent idiopathic scoliosis treated by anterior spinal fusion to L3 was significantly higher than by posterior spinal fusion to L3. Neurospine. 2021;18:457–463. doi: 10.14245/ns.2142182.091. 20210412. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Yao G., Cheung J.P.Y., Shigematsu H., et al. Characterization and predictive value of segmental curve flexibility in adolescent idiopathic scoliosis patients. Spine (Phila Pa 1976) 2017;42:1622–1628. doi: 10.1097/brs.0000000000002046. [DOI] [PubMed] [Google Scholar]
- 40.Sun X., Xie Y., Kong Q., et al. Segmental characteristics of main thoracic curves in patients with severe adolescent idiopathic scoliosis. World Neurosurg. 2018;119:e174–e179. doi: 10.1016/j.wneu.2018.07.086. [DOI] [PubMed] [Google Scholar]
- 41.Kawasaki S., Shigematsu H., Tanaka M., et al. Segmental flexibility in adolescent idiopathic scoliosis assessed using the fulcrum-bending radiography method. Clin Spine Surg. 2020;33 [Google Scholar]
- 42.Luk K.D., Lee C.F., Cheung K.M., et al. Clinical effectiveness of school screening for adolescent idiopathic scoliosis: a large population-based retrospective cohort study. Spine (Phila Pa 1976) 2010;35:1607–1614. doi: 10.1097/BRS.0b013e3181c7cb8c. [DOI] [PubMed] [Google Scholar]