Radiographic Variability of Pelvic Incidence Following Surgical Correction of Thoracic and Thoracolumbar Scheuermann Kyphosis

Abstract

Study Design

Retrospective study.

Objectives

Pelvic incidence (PI) is traditionally considered a fixed anatomical parameter; however, postoperative variability has been reported in adult deformity cohorts. Data regarding PI behavior in adolescent Scheuermann’s kyphosis (SK), particularly according to deformity subtype and distal fusion level, remain limited. This study aimed to evaluate PI variability in thoracic (TSK) and thoracolumbar (TLSK) SK and to analyze its relationship with spinopelvic parameters.

Methods

102 consecutive SK patients who underwent single-stage posterior pedicle screw fusion(2015–2021) were retrospectively reviewed; 52 met inclusion criteria (TSK n=30,TLSK n=22). Thirty age-matched healthy controls were included. Radiographic parameters (TK,TLK,GK,LL,CL,PI,PT,SS,SVA,PI–LL mismatch) were measured on standardized standing whole-spine radiographs. Reliability was assessed using ICC and smallest detectable change(SDC95). Intergroup comparisons, correlation, and exploratory multivariate regression analyses were performed.

Results

Reliability for PI, PT, and SS was excellent(ICC 0.92–0.97). Preoperatively, TLSK demonstrated lower PI and SS than TSK(p<0.05). At final follow-up(mean 4.6±2.2 years), PI increased significantly in TLSK (ΔPI +6.18°), exceeding SDC95, whereas no significant change was observed in TSK. The intergroup ΔPI difference was 5.22°(p=0.050). Distal fusion(≥L3) was associated with greater ΔPI. ΔPI correlated positively with postoperative PT and SS and negatively with LL. In multivariate analysis, postoperative PT(positive) and LL(negative) remained independently associated with ΔPI (R²=0.345, p=0.001).

Conclusions

TLSK patients exhibit lower baseline PI and modest postoperative PI increase compared with TSK. Postoperative PI variability appears related to pelvic posture and lumbar alignment rather than definitive anatomical remodeling. These findings highlight the importance of comprehensive spinopelvic assessment during surgical planning and follow-up of Scheuermann kyphosis.

Keywords

Scheuermann’s kyphosis thoracolumbar Scheuermann kyphosis pelvic incidence spinopelvic alignment sagittal balance lumbar lordosis

Introduction

Scheuermann’s kyphosis (SK) is classically defined by the presence of ≥5° of anterior wedging in at least three consecutive vertebral bodies.¹ SK is commonly divided into typical and atypical forms.¹⁴ The typical variant usually has an apex in the mid-thoracic region (T7–T9) and meets Sørensen’s criteria, whereas the atypical variant more often has an apex at the thoracolumbar or lumbar levels, where classic radiographic findings such as disc space narrowing, endplate irregularities, and Schmorl’s nodes may be present, although Sørensen’s criteria are not always fulfilled.²

The balance between the pelvis and spine is essential in determining overall global spinal alignment.¹⁵ Although pelvic incidence (PI) has traditionally been considered a fixed anatomical parameter,⁶ recent studies have reported postoperative changes in PI.^1,3,4 However, these observations are largely derived from adult deformity cohorts,⁸ and postoperative PI behavior in adolescent SK remains unclear.^10,12,17 The literature has not clearly compared thoracic (TSK) and thoracolumbar (TLSK) variants with respect to PI, and the influence of fusion extending into the lower lumbar segments remains uncertain.^12,16,17 Therefore, this study aims to address this gap by evaluating PI variability across SK subtypes and its relationship with other spinopelvic parameters. While posterior spinal fusion remains the standard surgical treatment for progressive SK, motion-preserving growth-modulation approaches such as flexible posterior vertebral tethering have recently been reported, with longitudinal assessment of spinopelvic parameters (PI, PT, SS) during follow-up.²³

Material and Method

Patient Selection and Clinical Investigation

Following ethics committee approval, this retrospective single-center study was conducted at a tertiary referral spine center. A total of 102 patients with Scheuermann kyphosis who underwent single-stage posterior pedicle screw fusion between January 2015 and March 2021 were retrospectively reviewed. All surgical procedures were performed by the same spine surgery team according to standardized surgical and follow-up protocols. Inclusion criteria were: (1) diagnosis of Scheuermann kyphosis under the age of 20 according to Sorensen criteria, defined as anterior wedging of at least 5° in three or more consecutive vertebral bodies; (2) visibility of the pelvis, femoral heads, and C2 in all standing AP/lateral spine radiographs taken before and after surgery; and (3) follow-up period of at least 2 years. Exclusion criteria were postural kyphosis, congenital deformity, neuromuscular disease, previous spinal surgery (including anterior procedures), and revision surgery. Of the 102 initially screened patients, 50 were excluded. 27 patients were excluded because they did not meet the radiographic inclusion criteria required for reliable spinopelvic measurements, including incomplete visualization of the pelvis, femoral heads, or cervical spine. 17 patients were excluded because of insufficient follow-up or missing clinical/radiographic data. 6 patients were excluded because, although they had previously been diagnosed with Scheuermann kyphosis, they did not fulfill the classical Sorensen criteria. Consequently, 52 patients met all eligibility criteria and were included in the final analysis. To form the control group, thirty healthy participants were selected from our institution’s radiographic database. The control cohort was selected to match the age and sex distribution of the Scheuermann kyphosis cohort(p<0.05). These individuals had undergone full standing spinal radiography as part of routine health screening (e.g., pre-employment/occupational assessment) and only those without a known spinal deformity or history of spinal or pelvic surgery were included. To ensure comparability with the patient cohort, control radiographs were obtained using the same standardized standing lateral imaging protocol (neutral upright posture; images were repeated in case of inadequate positioning, e.g., insufficient overlap of the femoral heads). Individuals with structural spinal deformities (e.g., scoliosis or pathological kyphosis), spondylolisthesis, previous fracture/trauma sequelae, congenital or neuromuscular conditions affecting alignment, or inadequate image quality for reliable measurement were excluded from the control group.

Patients were divided into two groups according to the kyphotic apex level: thoracic Scheuermann kyphosis (TSK; apex T9 or higher) and thoracolumbar Scheuermann kyphosis (TLSK; apex T10 or below) (Figures 1 and 2).⁷ All operated patients had a preoperative global kyphosis angle greater than 70°. The lowest instrumented vertebra was selected as the sagittal stable vertebra (SSV) or SSV-1, while the upper instrumented vertebra (UIV) was chosen as the proximal end vertebra (PEV) of the kyphotic deformity. The lowest instrumented vertebra (LIV) was recorded for each patient. For secondary analyses, distal fusion was categorized as proximal LIV (L2 or higher) and distal LIV (L3 or lower), and compared between categories along with ΔPI, fusion length, and kyphosis correction magnitude. Accordingly, 30 patients were included in the TSK group, 22 in the TLSK group, and 30 in the control group.

Figure 1.

Preoperative (A) and postoperative third year (B) radiographs of a 19-year-old female with Thoracic Scheuermann Kyphosis with an apex at T9. Posterior spinal fusion (T3–L1) was achieved with use of pedicle screws and rods system

Figure 2.

Pre- (A) and postoperative third year (B) radiographs of a 18-year-old female with Thoracolumbar Scheuermann Kyphosis with an apex at T11-T12. Posterior spinal fusion (T4–L4) was achieved with use of pedicle screws and rods system

Radiological Evaluation

In the surgical group, preoperative and postoperative long-cassette standing whole-spine anteroposterior (AP) and lateral radiographs were obtained using a standardized acquisition protocol. In the control group, a single long-cassette standing whole-spine radiograph was obtained using the same protocol. Patients stood barefoot, looking straight ahead, with knees and hips fully extended; upper-extremity positioning was standardized to minimize shoulder overlap (fists on clavicles). Pelvic rotation was minimized by neutral pelvic positioning and confirmation of near-overlap of the femoral heads on lateral radiographs; images showing substantial femoral head displacement were repeated whenever possible. Radiological measurements were performed using Surgimap (Nemaris Inc., USA) and included: thoracic kyphosis (TK, T5–T12), thoracolumbar kyphosis (TLK, T10–L2), global kyphosis (GK; Cobb angle between the upper and lower end vertebrae of the kyphotic curve), cervical lordosis (CL, C2–C7), lumbar lordosis (LL, L1–S1), pelvic incidence (PI), sacral slope (SS), pelvic tilt (PT), C7 sagittal vertical axis (SVA), and PI–LL mismatch (Figure 3). Two independent spine surgeons measured PI, PT, and SS to assess inter-observer and intra-observer reliability. Intraclass correlation coefficients (ICCs), standard error of measurement (SEM), and smallest detectable change at the 95% confidence level (SDC95) were calculated and reported; observed longitudinal changes (e.g., ΔPI) were interpreted in relation to SDC95. Inter-observer ICC was calculated using a two-way random-effects model (absolute agreement), and intra-observer ICC was calculated using a two-way mixed-effects model (absolute agreement). SEM was calculated as SD×√(1−ICC), and SDC95 was calculated as 1.96×√2×SEM.

Statistical Analysis

Statistical analyses were performed using NCSS 2020 (NCSS LLC, Kaysville, Utah, USA). Continuous variables are presented as mean ± standard deviation (or median [min–max] where appropriate), and categorical variables are presented as numbers and percentages. Normality was assessed using the Shapiro-Wilk test and graphical methods. Intergroup comparisons were performed using Student's t-test for normally distributed variables, otherwise using the Mann-Whitney U test. Preoperative and postoperative comparisons were assessed using paired-sample t-tests (or nonparametric equivalents where appropriate). Categorical variables were compared using chi-square, Fisher’s exact test, or Freeman-Halton tests as appropriate. Relationships were assessed using Pearson correlation. Since pelvic incidence is geometrically related to pelvic parameters in standing radiographs (PI = PT + SS), regression models using PI as the dependent variable and including derived terms such as PT, SS, or PI-LL are mathematically cyclical; Therefore, multivariate regression analysis using PI as the dependent variable was not performed. The statistical significance level was set at p ≤ 0.05. Since this is a retrospective cohort study, no prior sample size calculation was performed, and all eligible cases were included during the study period. To provide a practical assessment of sample adequacy, a sensitivity analysis was performed using a two-sided α = 0.05 for primary group comparison (ΔPI difference between TLSK and TSK). With TLSK n=22 and TSK n=30, and using the pooled standard deviation (SD_pooled ≈ 9.23°) for ΔPI obtained from Table 4, the current sample size provides approximately 80% power to detect an intergroup difference of approximately 7.40° in ΔPI (standardized effect size d≈0.80); smaller differences may be detected with lower power. This post-hoc sensitivity analysis is provided for context only and should not be interpreted as evidence of prospective statistical power (Figures 2, and 3).

Figure 3.

Measurements of pelvic incidence (PI), pelvic tilt (PT), and sacral slope (SS)

Results

The mean follow-up time was 4.6 ± 2.2 years, and the mean fusion length was 11.4 ± 0.9 segments. Follow-up time (TLSK: 4.45 ± 2.30 years vs TSK: 4.70 ± 2.17 years; p = 0.638) and fusion length (TLSK: 11.50 ± 1.14 segments vs TSK: 11.40 ± 0.81 segments; p = 0.532) did not differ between the groups. Demographic and baseline characteristics are summarized in Table 1. By definition, the apex vertebra was located at T10 or below in 22 patients (TLSK) and at T9 or above in 30 patients (TSK) (Table 1).

Table 1.

Demographic and Baseline Characteristics of the Study Cohort (Operated Patients and Controls)

Characteristic	Total (n=82)	Operated patients (n=52)	Controls (n=30)
Gender
Female	43 (52,4)	25 (48,1)	18 (60,0)
Male	39 (47,6)	27 (51,9)	12 (40,0)
Age (years)
Mean ± SD	16,85 ± 1,65	16,67 ± 1,68	17,17 ± 1,58
Median (Min–Max)	17 (13–19)
Follow-up duration (years)^a
Mean ± SD		4,60 ± 2,21
Median (Min–Max)		4 (2–8)
Apex vertebra^a
T6		2 (3,8)
T7		12 (23,0)
T8		14 (26,9)
T9		2 (3,8)
T10		7 (13,4)
T11		11 (21,1)
T12		4 (7,6)
Length of fused segments^a
Mean ± SD		11,44 ± 0,96
Median (Min–Max)		12 (9–13)
Group distribution
TLSK	22 (26,8)	22 (42,3)	0
TSK	30 (36,6)	30 (57,7)	0
Control	30 (36,6)	0	30 (100,0)

Values are presented as n (%), mean ± SD, or median (min–max).

^aVariables apply to operated patients only (n=52). NA = not applicable.

Inter-observer and intra-observer reliability was excellent for PI, PT, and SS measurements (inter-observer ICC range 0.92–0.95; intra-observer ICC range 0.96–0.97; Table 2). The inter-observer SDC95 for PI was 3.10° and the intra-observer SDC95 was 2.40°. In the TLSK group, the mean postoperative change in PI (ΔPI) was +6.18°, exceeding both SDC95 thresholds, indicating that the observed change was greater than the expected measurement error (Table 2).

Table 2.

Inter- and Intra-observer Reliability of Pelvic Parameters (PI, PT, SS) Measured on Standing Lateral Radiographs

Parameter	Inter-observer ICC (95% CI)	Intra-observer ICC (95% CI)	SD (°)	SEM_inter (°)	SDC95_inter (°)	SEM_intra (°)	SDC95_intra (°)
PI	0.95 (0.89–0.98)	0.97 (0.93–0.99)	5.0	1.12	3.10	0.87	2.40
PT	0.93 (0.85–0.97)	0.96 (0.91–0.99)	4.0	1.06	2.94	0.80	2.22
SS	0.92 (0.83–0.97)	0.96 (0.90–0.99)	6.0	1.70	4.71	1.20	3.33

Inter-observer ICC was calculated using a two-way random-effects model (absolute agreement). Intra-observer ICC was calculated using a two-way mixed-effects model (absolute agreement). SEM = SD×√(1−ICC). SDC95 = 1.96×√2×SEM.

Compared to the control group (n=30), significant differences were observed in preoperative sagittal alignment parameters including TK, TLK, GK, LL, and CL in both TLSK and TSK patients (p ≤ 0.05), whereas there was no significant difference in PT and SVA (p > 0.05) (Table 3). Prior to surgery, TLSK showed lower PI and SS values compared to TSK (Table 4; PI MD = TLSK−TSK −7.09°, 95% CI −12.49 to −1.70; SS MD −8.04°, 95% CI −12.53 to −3.55), but these differences were no longer present at the last follow-up. Within TLSK, PI increased by +6.18° (95% CI 1.58 to 10.79) and SS by +4.45° (95% CI 1.16 to 7.75) (Table 4). The intergroup ΔPI difference was 5.22° (95% CI 0.01 to 10.42) (Table 4).

Table 3.

Comparison Between Groups

		Total	Group			p (1-3)	p (2-3)
		Total	¹TLSK (n=22)	²TSK (n=30)	³Control (n=30)	p (1-3)	p (2-3)
Gender	Female	43 (52.4)	11 (50.0)	14 (46.7)	18 (60.0)	0.473^a	0.301^a
Gender	Male	39 (47.6)	11 (50.0)	16 (53.3)	12 (40.0)
Age	Mean ± SD	16.85 ± 1.65	16.32 ± 1.81	16.93 ± 1.55	17.17 ± 1.58	0.078^b	0.566^b
Baseline GK	Mean ± SD	67.06 ± 23.06	81.59 ± 10.10	83.13 ± 12.23	40.33 ± 9.75	0.001**^b	0.001**^b
Baseline PI	Mean ± SD	43.17 ± 10.05	37.77 ± 10.42	44.87 ± 8.89	45.43 ± 9.70	0.009**^b	0.814^b
Baseline PT	Mean ± SD	11.49 ± 8.21	11.95 ± 10.05	11.07 ± 7.48	11.57 ± 7.68	0.875^b	0.799^b
Baseline SS	Mean ± SD	31.61 ± 8.61	25.73 ± 7.29	33.77 ± 8.42	33.77 ± 7.86	0.001**^b	0.595^b
Baseline LL	Mean ± SD	−61.66 ± 13.05	−63.09 ± 14.31	−69.17 ± 10.75	−53.10 ± 8.73	0.003**^b	0.001**^b
Baseline PI–LL	Mean ± SD	−18.49 ± 14.79	−25.32 ± 17.58	−24.30 ± 10.40	−7.67 ± 9.36	0.001**^b	0.001**^b
Baseline TLK	Mean ± SD	22.79 ± 18.50	44.77 ± 11.53	22.23 ± 12.79	7.23 ± 8.89	0.001**^b	0.001**^b
Baseline TK	Mean ± SD	48.88 ± 15.74	62.77 ± 11.52	53.57 ± 10.50	34.00 ± 9.35	0.001**^b	0.001**^b
Baseline CL	Mean ± SD	−16.79 ± 18.01	−22.18 ± 15.91	−26.20 ± 18.86	−3.43 ± 8.33	0.001**^b	0.001**^b
Baseline SVA	Mean ± SD	−19.55 ± 30.93	−5.82 ± 43.69	−28.23 ± 25.91	−20.93 ± 19.88	0.100^b	0.266^b

^aPearson chi-square test.

^bStudent’s t-test.

*p ≤ 0.05; **p < 0.01.

Table 4.

Preoperative, Final Follow-Up, and Change (Δ) in Spinopelvic Parameters in TLSK and TSK, Including Effect Sizes (Mean Differences) and 95% Confidence Intervals

Parameter	Timepoint	TLSK (n=22) mean ± SD (95% CI for Δ)	TSK (n=30) mean ± SD (95% CI for Δ)	MD TLSK-TSK (95% CI)	Between-group p	Within-group p (TLSK/TSK)
PI	Preoperative	37.77 ± 10.42	44.87 ± 8.89	−7.09 (−12.49 to −1.70)	0.011*
	Final follow-up	43.95 ± 12.46	45.83 ± 10.92	−1.88 (−8.41 to 4.66)	0.566
	Change (Δ)	6.18 ± 10.39 (1.58 to 10.79)	0.97 ± 8.29 (−2.13 to 4.06)	5.22 (0.01 to 10.42)	0.050*	0.011*/0.528
PT	Preoperative	11.95 ± 10.05	11.07 ± 7.48	0.89 (−3.99 to 5.77)	0.716
	Final follow-up	13.82 ± 7.89	12.60 ± 8.13	1.22 (−3.31 to 5.75)	0.591
	Change (Δ)	1.86 ± 10.05 (−2.59 to 6.32)	1.53 ± 5.87 (−0.66 to 3.73)	0.33 (−4.12 to 4.78)	0.882	0.394/0.163
SS	Preoperative	25.73 ± 7.29	33.77 ± 8.42	−8.04 (−12.53 to −3.55)	0.001**
	Final follow-up	30.18 ± 11.18	33.30 ± 8.00	−3.12 (−8.45 to 2.22)	0.246
	Change (Δ)	4.45 ± 7.44 (1.16 to 7.75)	−0.47 ± 6.61 (−2.94 to 2.00)	4.92 (0.99 to 8.85)	0.015*	0.011*/0.702
LL	Preoperative	−63.09 ± 14.31	−69.17 ± 10.75	6.08 (−0.90 to 13.05)	0.086
	Final follow-up	−51.77 ± 12.33	−50.80 ± 12.37	−0.97 (−7.94 to 5.99)	0.780
	Change (Δ)	11.32 ± 16.07 (4.19 to 18.44)	18.37 ± 9.60 (14.78 to 21.95)	−7.05 (−14.22 to 0.13)	0.054	0.003/<0.001
PI-LL	Preoperative	−25.32 ± 17.58	−24.30 ± 10.40	−1.02 (−8.84 to 6.80)	0.795
	Final follow-up	−7.82 ± 11.66	−4.97 ± 11.09	−2.85 (−9.24 to 3.54)	0.374
	Change (Δ)	17.50 ± 17.24 (9.86 to 25.14)	19.33 ± 10.11 (15.56 to 23.11)	−1.83 (−9.48 to 5.82)	0.632	<0.001/<0.001
TLK	Preoperative	44.77 ± 11.53	22.23 ± 12.79	22.54 (15.62 to 29.46)	<0.001**
	Final follow-up	12.36 ± 10.07	9.63 ± 9.57	2.73 (−2.78 to 8.25)	0.325
	Change (Δ)	−32.41 ± 16.37 (−39.67 to −25.15)	−12.60 ± 13.98 (−17.82 to −7.38)	−19.81 (−28.28 to −11.34)	<0.001**	<0.001/<0.001
TK	Preoperative	62.77 ± 11.52	53.57 ± 10.50	9.21 (3.04 to 15.37)	0.004**
	Final follow-up	32.73 ± 7.38	30.17 ± 9.89	2.56 (−2.47 to 7.59)	0.312
	Change (Δ)	−30.05 ± 13.12 (-35.86 to −24.23)	−23.40 ± 11.82 (−27.81 to −18.99)	-6.65 (−13.63 to 0.34)	0.062	<0.001/<0.001
CL	Preoperative	−22.18 ± 15.91	−26.20 ± 18.86	4.02 (−5.95 to 13.98)	0.422
	Final follow-up	-11.55 ± 15.81	−24.07 ± 17.82	12.52 (2.94 to 22.11)	0.012*
	Change (Δ)	10.64 ± 21.20 (1.24 to 20.04)	2.13 ± 19.30 (−5.08 to 9.34)	8.50 (−2.84 to 19.85)	0.139	0.028*/0.550
SVA	Preoperative	−5.82 ± 43.69	−28.23 ± 25.91	22.42 (2.96 to 41.87)	0.025*
	Final follow-up	−18.23 ± 16.61	−23.33 ± 41.16	5.11 (−13.58 to 23.79)	0.586
	Change (Δ)	−12.41 ± 42.31 (−31.17 to 6.35)	4.90 ± 41.72 (−10.68 to 20.48)	−17.31 (−40.97 to 6.35)	0.148	0.183/0.525

Note: Values are mean±SD. Δ = final follow-up - preoperative. MD = mean difference (TLSK - TSK) with 95% CI. For Δ rows, 95% CIs are provided for mean change within each group. *p ≤ 0.05; **p < 0.01.

Between TLSK and TSK, most sagittal parameters no longer differed at the last follow-up; However, the C2-C7 angle differed at the last follow-up (Table 4; MD 12.52°, 95% CI 2.94 to 22.11; p = 0.012), suggesting less cervical lordosis (smaller magnitude/less negative angle) in the TLSK group.

In the overall cohort, ΔPI showed a weak positive correlation with postoperative (last follow-up) PT (r = 0.329, p = 0.017) and postoperative SS (r = 0.381, p = 0.005), and a moderate negative correlation with postoperative LL (r = −0.423, p = 0.002) (Table 5). No significant association was observed between ΔPI and postoperative PI-LL mismatch, TLK, TK, CL, or SVA (Table 5).

Table 5.

Correlation Between ΔPI and Postoperative Radiographic Parameters (Overall Cohort)

Postoperative parameter	Pearson r	p-value
Postop PT	0.329	0.017
Postop SS	0.381	0.005
Postop LL	−0.423	0.002
Postop PI–LL	0.089	0.529
Postop TLK	0.062	0.661
Postop TK	0.051	0.719
Postop CL	0.180	0.203
Postop SVA	0.020	0.890

ΔPI indicates final follow-up minus preoperative (positive values = increase over time; negative values = decrease). Postoperative parameters (Postop) refer to values measured at final follow-up. Values are Pearson correlation coefficients (r) with corresponding two-sided p-values. *p ≤ 0.05, **p < 0.01.

In the exploratory multivariate linear regression model where ΔPI was the dependent variable, postoperative PT and postoperative LL remained independently associated with ΔPI (R² = 0.345; model p = 0.001). Higher postoperative PT was associated with a larger increase in ΔPI (B = 0.493; 95% CI 0.212 to 0.774; p = 0.001), whereas higher postoperative LL was associated with a smaller ΔPI (B = −0.384; 95% CI −0.567 to −0.201; p = 0.001) (Table 6).

Table 6.

Multivariable Linear Regression Model Predicting ΔPI

Model summary
R	0.587
R²	0.345
Adjusted R²	0.318
Std. Error of estimate	7.85
F	12.912
Model p-value	0.001
Coefficients
Predictor	B (unstandardized)	95% CI for B; p-value
Intercept	−22.958	—
Postop PT	0.493	0.212 to 0.774; p = 0.001
Postop LL	−0.384	−0.567 to −0.201; p = 0.001

Regression equation: ΔPI = −22.958 + 0.493(Postop PT) − 0.384(Postop LL).

While distal fusion typically up to L3–L4 (rarely L5) was more common in TLSK cases, TSK cases generally terminated more proximally (mostly around L1). When classified according to LIV (proximal vs distal), ΔPI was compared between categories and interpreted in conjunction with fusion length and the magnitude of kyphosis correction (ΔTK/ΔTLK) (Table 7).

Table 7.

Lowest Instrumented Vertebra (LIV) Distribution Across Groups

Group	LIV	n	% Within group	% (n = 52)
Thoracic (n=30)	T11	2	6.7	3.8
Thoracic (n=30)	T12	10	33.3	19.2
Thoracic (n=30)	L1	18	60.0	34.6
Thoracic subtotal		30	100.0	57.7
Thoracolumbar (n=22)	L3	7	31.8	13.5
Thoracolumbar (n=22)	L4	14	63.6	26.9
Thoracolumbar (n=22)	L5	1	4.5	1.9
Thoracolumbar subtotal		22	100.0	42.3
Overall total		52		100.0

When categorized by lowest instrumented vertebra (LIV), the proximal LIV group (≤L2) corresponded to TSK cases (T11–L1), whereas the distal LIV group (≥L3) corresponded to TLSK cases (L3–L5) (Table 8). ΔPI was greater in the distal LIV group (6.18±10.39°) than in the proximal LIV group (0.97±8.29°), with an MD of 5.22° (95% CI 0.01–10.42; p = 0.050). Follow-up duration and fusion length were comparable between LIV categories (p = 0.638 and p = 0.532, respectively), while thoracolumbar kyphosis correction (ΔTLK) was larger in the distal LIV group (Table 8).

Table 8.

Comparison by LIV Category (Proximal vs Distal)

Variable	Proximal LIV (≤L2) (n=30; TSK)	Distal LIV (≥L3) (n=22; TLSK)	MD (distal - proximal) (95% CI)	Between-group p	Within-group p (distal/Proximal)
Follow-up duration (years), mean ± SD	4.70 ± 2.17	4.45 ± 2.30	NR	0.638	NR
Fusion length (segments), mean ± SD	11.40 ± 0.81	11.50 ± 1.14	NR	0.532	NR
Baseline PI (°), mean ± SD	44.87 ± 8.89	37.77 ± 10.42	−7.09 (−12.49 to −1.70)	0.011*	NR
ΔPI (°), mean ± SD (95% CI for Δ)	0.97 ± 8.29 (−2.13 to 4.06)	6.18 ± 10.39 (1.58 to 10.79)	5.22 (0.01 to 10.42)	0.050*	0.011*/0.528
ΔTLK (°), mean ± SD (95% CI for Δ)	−12.60 ± 13.98 (−17.82 to −7.38)	−32.41 ± 16.37 (−39.67 to −25.15)	−19.81 (−28.28 to −11.34)	<0.001**	<0.001/<0.001
ΔTK (°), mean ± SD (95% CI for Δ)	−23.40 ± 11.82 (−27.81 to −18.99)	−30.05 ± 13.12 (−35.87 to −24.23)	−6.65 (−13.63 to 0.33)	0.062	<0.001/<0.001

LIV category is collinear with deformity subtype in this cohort (TSK cases ended proximally; TLSK cases ended distally); therefore, LIV-based comparisons should be interpreted descriptively and does not fully isolate the independent effect of distal fusion.

*p ≤ 0.05; **p < 0.01.

In TLSK, ΔPI showed a strong positive correlation with ΔPT (r = 0.731, p = 0.001), and ΔSS showed a strong positive correlation with ΔCL (r = 0.598, p = 0.003). A moderately negative correlation was observed between ΔLL and ΔCL (r = −0.425, p = 0.045) (Table 9).

Table 9.

The Relationship Between Preoperative to Final Follow-Up Radiological Parameters According to TLSK

TLSK		Δ PI	Δ PT	Δ SS	Δ LL	Δ PI–LL	Δ TLK	Δ TK	Δ CL	Δ SVA
Δ PI	r	1	0.731	0.417	-0.206	0.431	0.030	0.020	0.053	−0.178
Δ PI	p		0.001**	0.054	0.357	0.045*	0.896	0.928	0.816	0.428
Δ PT	r	0.731	1	−0.309	0.152	0.559	0.385	0.348	−0.381	0.143
Δ PT	p	0.001**		0.162	0.499	0.007**	0.077	0.112	0.081	0.527
Δ SS	r	0.417	−0.309	1	−0.521	−0.180	−0.414	−0.413	0.598	−0.449
Δ SS	p	0.054	0.162		0.013*	0.422	0.055	0.056	0.003**	0.036*
Δ LL	r	−0.206	0.152	−0.521	1	0.778	0.130	−0.176	-0.425	0.431
Δ LL	p	0.357	0.499	0.013*		0.001**	0.565	0.433	0.049*	0.045*
Δ PI–LL	r	0.431	0.559	−0.180	0.778	1	0.085	−0.198	−0.325	0.254
Δ PI–LL	p	0.045*	0.007**	0.422	0.001**		0.706	0.378	0.140	0.254
Δ TLK	r	0.030	0.385	−0.414	0.130	0.085	1	0.482	−0.342	0.433*
Δ TLK	p	0.896	0.077	0.055	0.565	0.706		0.023*	0.120	0.044*
Δ TK	r	0.020	0.348	−0.413	−0.176	−0.198	0.482	1	−0.241	0.372
Δ TK	p	0.928	0.112	0.056	0.433	0.378	0.023*		0.280	0.088
Δ CL	r	0.053	−0.381	0.598	−0.425	−0.325	−0.342	−0.241	1	−0.609
Δ CL	p	0.816	0.081	0.003**	0.049*	0.140	0.120	0.280		0.003**
Δ SVA	r	−0.178	0.143	−0.449	0.431	0.254	0.433	0.372	−0.609	1
Δ SVA	p	0.428	0.527	0.036*	0.045*	0.254	0.044*	0.088	0.003**

r: Pearson correlation test.

**p < 0.01.

*p ≤ 0.05.

Δ indicates final follow-up minus preoperative values. TK, TLK, LL, CL are Cobb angles; PI, PT, SS are pelvic parameters; SVA is sagittal vertical axis.

Discussion

Pelvic incidence (PI) has traditionally been regarded as a fixed parameter, yet recent studies have challenged this view.^4,5,9,11 Jiang et al.¹² emphasized that spinopelvic parameters in Scheuermann’s kyphosis vary according to deformity type. In line with their findings, our study demonstrated that TLSK patients had significantly lower PI and SS values compared with TSK patients, but significantly higher TLK. They attributed these differences to rigid kyphosis and compensatory hyperlordosis, particularly in the lumbar and cervical regions. Jansen et al.¹⁶ also reported that the rate of LL improvement was greater in TSK than TLSK, which is consistent with our observation of less lumbar hyperlordosis in TLSK cases.

Since pelvic incidence (PI) is derived from the sacral endplate and femoral head centers, small variations in pelvic rotation, beam projection, or femoral head overlap can affect the measured value and mimic true variation. In this study, radiographs were acquired using a standardized standing long-cassette protocol in a neutral upright position, and images were repeated when positioning was insufficient (including inadequate overlap of femoral heads) to reduce projection-related variability. Measurements were performed by two independent observers, and measurement reliability was quantified using intraclass correlation coefficients (ICCs); furthermore, the standard error of measurement (SEM) and the smallest detectable change at a 95% confidence level (SDC95) were calculated to support the interpretation of longitudinal changes. Using this approach, the postoperative PI increase observed in the TLSK group exceeded the SDC95, suggesting that the variation could not be explained solely by measurement error; whereas the smaller variation in the TSK group remained within the measurement error range. Notably, motion-preserving growth-modulation strategies for SK have also incorporated longitudinal spinopelvic assessment; in a recent cohort of flexible posterior vertebral tethering, global sagittal alignment improved while pelvic parameters (PI, PT, SS) remained largely stable during follow-up, suggesting that postoperative PI variability may differ by treatment strategy and/or pelvic posture during standing radiographs.²³

Pelvic incidence (PI), although traditionally considered a stable morphological parameter after growth,^26,28 may appear variable postoperatively when measured on standing radiographs.²⁴ First, spinopelvic fixation and the lumbopelvic construct may affect measured PI by altering the relationship between the sacrum and pelvis; studies have reported PI changes after long fusions to the pelvis, and S2-alar-iliac (S2AI) fixation with corrective maneuvers may contribute to such variation.²⁴ In ASD patients instrumented to the pelvis with S2AI screws, clinically meaningful PI changes (≥6°) have been observed in a substantial proportion immediately postoperatively, with proposed explanations including SIJ motion, S2AI screw placement, and sagittal cantilever correction techniques.²⁴ Second, in adolescent patients, PI may increase during growth and maturation; large pediatric cohorts show an increase in PI across growth stages before stabilization in adulthood.^25,26 Third, the postoperative definition/interpretation of pelvic parameters may change if the fusion extends to the lumbosacral junction (L5–S1), as L5 can be considered part of the pelvis (“new sacrum”) and pelvic parameters can be re-evaluated using the L5 superior endplate after a stable lumbosacral fusion.²⁷ Similarly, when the S1 endplate cannot be identified (e.g., lumbosacral transitional vertebrae), L5-based parameters (L5 incidence) using the superior L5 endplate have been proposed.²⁸

Rizkallah et al.¹⁰ demonstrated that patients with low baseline PI (<40°) exhibited greater postoperative changes, whereas those with higher PI values showed minimal variation. Consistent with these findings, our TLSK cohort—characterized by a mean preoperative PI of 37.7°—demonstrated a significant postoperative increase. Bederman et al.¹⁷ further suggested that in patients with low PI values, sagittal balance is maintained through adaptive changes in lumbar lordosis (LL) and pelvic tilt (PT). The postoperative increase in PI and sacral slope (SS) observed in our TLSK group supports this compensatory mechanism. While higher PI values are generally associated with greater compensatory capacity, excessively elevated PI has been linked to degenerative spondylolisthesis.^18,19 Similarly, Li et al.¹³ reported lower PI values (mean 35.1°) in patients with congenital or angular kyphosis secondary to tuberculosis, further corroborating our finding that TLSK patients present with inherently lower PI compared with both TSK patients and healthy controls.

We observed that both TLSK and TSK patients presented with significantly elevated preoperative TK, TLK, GK, LL, and CL values relative to controls, implying that compensatory adaptations in Scheuermann’s kyphosis primarily occur in the lumbar and cervical regions.¹² Moreover, PI and SS values in the TLSK group were significantly lower than those in both controls and the TSK group, consistent with the findings of Li et al, who reported kyphotic cases presenting with low PI values.¹³ In addition, the preoperative differences observed in SVA, TK, and TLK became comparable postoperatively, highlighting the effectiveness of surgery in restoring global sagittal balance.^16,21 At the final follow-up, only CL was significantly lower in the TLSK group, suggesting that thoracolumbar correction may limit the adaptive capacity of the cervical spine; similarly, Liu et al reported a tendency toward decreased cervical lordosis following thoracolumbar kyphosis surgery.²¹

In the TLSK group, moderate to strong associations were identified among spinopelvic parameters (ΔPI–ΔPT, r = 0.731; ΔSS–ΔCL, r = 0.598). Specifically, ΔPI correlated positively with ΔPT, ΔSS correlated positively with ΔCL, while ΔLL correlated negatively with ΔCL. These results indicate that changes in lumbar and pelvic alignment are closely linked to compensatory mechanisms at the cervical level. Our findings are consistent with those of Ye et al.,²⁰ who reported that increased lumbar lordosis may be accompanied by a compensatory reduction in cervical lordosis.

In our study group, postoperative variability in measured PI was associated with pelvic posture and lumbar alignment rather than behaving as a completely fixed value. ΔPI showed a weak positive correlation with postoperative PT (r = 0.329, p = 0.017) and postoperative SS (r = 0.381, p = 0.005), and a moderate negative correlation with postoperative LL (r = −0.423, p = 0.002) (Table 5). In an exploratory multivariate model using ΔPI as the dependent variable, higher postoperative PT remained independently associated with a larger ΔPI (B = 0.493; 95% CI 0.212 to 0.774; p = 0.001), whereas higher postoperative LL was associated with a smaller ΔPI (B = −0.384; 95% CI −0.567 to −0.201; p = 0.001) (R² = 0.345; Table 6). These findings support interpreting postoperative PI changes within the broader spinopelvic chain, rather than as an isolated “predictor” of PI, especially considering the geometric linkage between PI, PT, and SS in standing radiographs. Consistent with this concept, Bederman et al. The relationships reported between PI and both LL and PT in patients with Scheuermann’s disease¹⁷ and Nasto et al evaluated preoperative alignment parameters related to PI and LL in relation to the risk of postoperative joint kyphosis after posterior correction of Scheuermann’s kyphosis only.²²

This study has certain limitations. The retrospective design inherently carries the risk of selection bias, even though strict inclusion and exclusion criteria were applied. Additionally, the retrospective design limited the sample size to the current cohort. Furthermore, the follow-up period, while sufficient to evaluate short-to mid-term outcomes, may not fully demonstrate the long-term adaptations of pelvic incidence and sagittal balance. In addition, body mass index (BMI) data were not consistently available for the control group because the radiographs were obtained from an institutional screening database. Therefore, matching and comparison between groups with respect to BMI could not be performed and residual confounding related to body habitus cannot be excluded. These limitations should be considered when interpreting our results, and future prospective, multicenter studies with larger patient populations and longer follow-up are warranted.

Conclusion

Patients with thoracolumbar Scheuermann kyphosis demonstrated lower preoperative PI and SS values and showed a modest increase in radiographically measured PI during follow-up, whereas no significant PI change was observed in thoracic cases. Postoperative PI variability appeared to be associated with spinopelvic alignment and pelvic posture rather than definitive anatomical remodeling. These findings highlight the importance of comprehensive spinopelvic assessment during surgical planning and postoperative follow-up of Scheuermann kyphosis. Given the retrospective single-center design and relatively small sample size, the results should be interpreted as preliminary and hypothesis-generating rather than definitive evidence. Further prospective multicenter studies are required to confirm these observations.

Footnotes

ORCID iDs

Muhammed Fatih Serttas

Furkan Karakaş

Ethics Considerations

This study was approved by the Institutional Review Board of Sakarya University Faculty of Medicine (Approval No: E-71522473-050.01.04-285315). The study was conducted in accordance with the principles of the Declaration of Helsinki.

Consent to Participate

Written informed consent was obtained from all patients (or their legal guardians) prior to inclusion in the study.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

References

Lonner

Toombs

Mechlin

, et al.

MRI screening in operative Scheuermann kyphosis: is it necessary?

Spine Deform. 2017;5(2):124-133. doi:10.1016/j.jspd.2016.10.008.

Sardar

Ames

Lenke

. Scheuermann’s kyphosis: diagnosis, management, and selecting fusion levels. J Am Acad Orthop Surg. 2019;27(10):e462-e472. doi:10.5435/JAAOS-D-17-00748.

Place

Hayes

Huebner

Hayden

Israel

Brechbuhler

. Pelvic incidence: a fixed value or can you change it? Spine J. 2017;17(10):1565-1569. doi:10.1016/j.spinee.2017.06.037.

Maekawa

Endo

Suzuki

, et al. Impact of pelvic incidence on change in lumbo-pelvic sagittal alignment between sitting and standing positions. Eur Spine J. 2019;28(9):1914-1919. doi:10.1007/s00586-019-05891-9.

Ohya

Kawamura

Takasawa

, et al. Pelvic incidence change on the operating table. Eur Spine J. 2021;30(9):2473-2479. doi:10.1007/s00586-021-06753-z.

Schwab

Lafage

Patel

Farcy

. Sagittal plane considerations and the pelvis in the adult patient. Spine (Phila Pa 1976). 2009;34(17):1828-1833. doi:10.1097/BRS.0b013e3181a13c08.

Lin

Wang

Yang

, et al. The effect of pedicle subtraction osteotomy for the correction of severe Scheuermann thoracolumbar kyphosis on sagittal spinopelvic alignment. BMC Musculoskelet Disord. 2021;22(1):165. doi:10.1186/s12891-020-03942-7.

Menezes

Lacerda

Lamarca

. Sagittal alignment concepts and spinopelvic parameters. Rev Bras Ortop. 2023;58(1):1-8. doi:10.1055/s-0042-1742602.

Roussouly

Gollogly

Noseda

Berthonnaud

Dimnet

. The vertical projection of the sum of the ground reactive forces of a standing patient is not the same as the C7 plumb line: a radiographic study of the sagittal alignment of 153 asymptomatic volunteers. Spine (Phila Pa 1976). 2006;31(11):E320-E325. doi:10.1097/01.brs.0000218263.58642.

10.

Rizkallah

Shen

Phan

, et al.

Can pelvic incidence change after lumbo-pelvic fixation for adult spine deformity, and would the change be affected by the type of pelvic fixation?

Spine (Phila Pa 1976). 2024;49(1):E1-E7. doi:10.1097/BRS.0000000000004651.

11.

Schroeder

Noschenko

Burger

, et al. Pelvic incidence changes between flexion and extension. Spine Deform. 2018;6(6):753-761. doi:10.1016/j.jspd.2018.03.008.

12.

Jiang

Qiu

, et al. Sagittal spinopelvic alignment in adolescents associated with Scheuermann’s kyphosis: a comparison with normal population. Eur Spine J. 2014;23(7):1420-1426. doi:10.1007/s00586-014-3266-2.

13.

Sun

Guo

, et al. Analysis of spinopelvic sagittal alignment in patients with thoracic and thoracolumbar angular kyphosis. Spine (Phila Pa 1976). 2013;38(13):E813-E818. doi:10.1097/BRS.0b013e3182913219.

14.

Blumenthal

Roach

Herring

. Lumbar Scheuermann’s: a clinical series and classification. Spine (Phila Pa 1976). 1987;12(9):929-932. doi:10.1097/00007632-198712000-00002.

15.

Le Huec

Thompson

Mohsinaly

Barrey

Faundez

. Sagittal balance of the spine. Eur Spine J. 2019;28(9):1889-1905. doi:10.1007/s00586-019-06083-1.

16.

Jansen

van Rhijn

van Ooij

. Predictable correction of the unfused lumbar lordosis after thoracic correction and fusion in Scheuermann kyphosis. Spine (Phila Pa 1976). 2006;31(11):1227-1231. doi:10.1097/01.brs.0000217682.53629.ad.

17.

Bederman

Farhan

, et al. Sagittal spinal and pelvic parameters in patients with Scheuermann’s disease: a preliminary study. Int J Spine Surg. 2019;13(6):536-543. doi:10.14444/6073.

18.

Chung

Lee

. Correlation of pelvic parameters with isthmic spondylolisthesis. Asian Spine J. 2009;3(1):21-26. doi:10.4184/asj.2009.3.1.21.

19.

Barrey

Jund

Perrin

Roussouly

. Spinopelvic alignment of patients with degenerative spondylolisthesis. Neurosurgery. 2007;61(5):981-986. doi:10.1227/01.neu.0000303194.02921.30.

20.

Rider

Lafage

, et al. Spinopelvic sagittal compensation in adult cervical deformity. J Neurosurg Spine. 2023;39(1):1-10. doi:10.3171/2023.2.SPINE221295.

21.

Liu

Zhu

, et al. How does cervical sagittal profile change after the spontaneous compensation of global sagittal imbalance following one- or two-level lumbar fusion. BMC Musculoskelet Disord. 2024;25:387. doi:10.1186/s12891-024-07518-7.

22.

Nasto

Jain

Shalabi

, et al. Correlation between preoperative spinopelvic alignment and risk of proximal junctional kyphosis after posterior-only surgical correction of Scheuermann kyphosis. Spine J. 2016;16(10):1239-1248. doi:10.1016/j.spinee.2016.02.04.

23.

Aydogan

Pehlivanoglu

Erdag

Akturk

Akar

. Flexible posterior vertebral tethering for the management of Scheuermann’s kyphosis: correction by using growth modulation—clinical and radiographic outcomes of the first 10 patients with at least 3 years of follow-up. European Spine Journal. 2024;33:2677-2687. doi:10.1007/s00586-024-08297-4.

24.

Wei

Zuckerman

Cerpa

, et al.

Can pelvic incidence change after spinal deformity correction to the pelvis with S2-alar-iliac screws?

Eur Spine J. 2021;30:2486-2494. doi:10.1007/s00586-020-06658-3.

25.

Pesenti

Charles

Prost

Solla

Blondel

Ilharreborde

French Spine Surgery Society SFCR . Spinal sagittal alignment changes during childhood: results of a national cohort analysis of 1,059 healthy children. J Bone Joint Surg Am. 2023;105:676-686. doi:10.2106/JBJS.22.00977.

26.

Mac-Thiong

Berthonnaud

Dimar

2nd Betz

Labelle

. Sagittal alignment of the spine and pelvis during growth. Spine (Phila Pa 1976). 2004;29(15):1642-1647. doi:10.1097/01.brs.0000132312.78469.7b.

27.

Rajasekaran

Das

Aiyer

Kanna

Shetty

. Analysis of spinopelvic parameters with L5 as the new sacrum after fusion in high-grade spondylolisthesis: a possible explanation for satisfactory results with in-situ fusion. Asian Spine J. 2018;12(1):103-111. doi:10.4184/asj.2018.12.1.103.

28.

Dominguez

Faundez

Demezon

Cogniet

Le Huec

. Normative values for the L5 incidence in a subgroup of transitional anomalies extracted from 147 asymptomatic subjects. Eur Spine J. 2016;25:3602-3607. doi:10.1007/s00586-015-4371-6.