Staged lateral lumbar interbody fusion combined with mini-open posterior screw fixation with or without facetectomy for coronal and sagittal realignment in degenerative spinal deformity: A retrospective cohort study

Abstract

Introduction

Coronal imbalance in degenerative spinal deformity is difficult to correct, particularly when the fractional curve is rigid and the lumbosacral foundation is oblique.

Research question

Is staged lateral lumbar interbody fusion with targeted mini-open posterior correction associated with stepwise coronal and sagittal realignment and maintenance of coronal alignment at follow-up?

Material and methods

We retrospectively reviewed 51 consecutive patients treated with multi-level lateral lumbar interbody fusion followed by posterior percutaneous fixation with or without targeted mini-open osteotomy. Standing full-length radiographs were assessed preoperatively, after stage I, after stage II, and at latest follow-up. Coronal endpoints were coronal Cobb angle, coronal vertical axis, and fractional-curve take-off angle. Sagittal endpoints included lumbar lordosis, pelvic incidence minus lumbar lordosis mismatch, T1 pelvic angle, and sagittal vertical axis. Clinical outcomes were assessed preoperatively and at follow-up.

Results

Coronal Cobb angle improved from 35.62° preoperatively to 21.93° after stage I, 10.20° after stage II, and 10.91° at follow-up. Coronal vertical axis improved from 28.50 mm to 14.95 mm, and fractional-curve take-off angle from 17.28° to 5.88°. Lumbar lordosis increased from 22.40° to 35.66°, while pelvic incidence minus lumbar lordosis mismatch, T1 pelvic angle, and sagittal vertical axis also improved. Back-pain visual analog scale and Oswestry Disability Index improved significantly at follow-up.

Discussion and conclusion

Staged lateral lumbar interbody fusion with targeted mini-open posterior correction was associated with stepwise coronal and sagittal realignment, and the improvement in mean coronal alignment was maintained at follow-up.

Highlights

• •Staged LLIF improved coronal and sagittal alignment in rigid DSD.
• •Targeted FC correction reduced lumbosacral foundation obliquity.
• •Coronal balance was maintained across Qiu subtypes at follow-up.
• •ODI improvement reached MCID in most patients at follow-up.

1. Introduction

Degenerative spinal deformity (DSD) commonly manifests with coupled deformities in the sagittal and coronal planes and contributes substantially to pain and disability in the aging population (Ailon et al., 2015; Diebo et al., 2019). While sagittal imbalance has received extensive attention because of its strong association with health-related quality of life, coronal imbalance (CI) remains under-recognized despite its considerable prevalence pre- and postoperatively and its limited compensatory potential (Bao et al., 2016, 2019; Daubs et al., 2013; Glassman et al., 2005; Ploumis et al., 2015; Terran et al., 2013). Additionally, few compensatory mechanisms are available for CI patients, resulting in a low quality of life (Bao et al., 2016; Daubs et al., 2013; Glassman et al., 2005). Thus, CI in DSD should not be ignored, especially when it appears after surgery. Moreover, the preoperative CI is also correlated with increased implant failures (Cho et al., 2013).

Correction of biplanar imbalance in DSD patients is complex, especially in patients with rigid deformities, for whom the asymmetrical pedicle subtraction osteotomy technique and the sequential correction technique with a posterior approach have been proposed (Bao et al., 2019; Cecchinato et al., 2015; Thambiraj and Boszczyk, 2012).

With the rise of minimally invasive (MIS) lateral approaches, multi-level lateral lumbar interbody fusion (LLIF) can achieve indirect decompression and meaningful sagittal correction while preserving posterior musculature (Bae and Lee, 2018). However, when the fractional curve (FC) is undertreated, aggressive correction of the main lumbar curve on an oblique foundation can precipitate postoperative coronal malalignment—the so-called take-off phenomenon (Acosta et al., 2011; Anand et al., 2013; Deukmedjian et al., 2013; Mummaneni et al., 2019; Ransom et al., 2024). This risk is particularly relevant in rigid thoracolumbar/lumbar deformities. Failure to restore the horizontal foundation may lead to oblique takeoff and subsequent CI.

To address this gap, we evaluated a staged LLIF-based correction strategy in which multi-level lateral interbody fusion was followed by posterior percutaneous fixation with selective mini-open osteotomy directed at the fractional-curve region when residual deformity persisted. We hypothesized that this staged approach would be associated with stepwise improvement in both coronal and sagittal alignment and with maintenance of coronal balance at follow-up across Qiu coronal subtypes (Qiu et al., 2025).

2. Methods

2.1. Study cohort and design

This single-center retrospective cohort study was approved by the institutional review board of Xuanwu Hospital, Capital Medical University (Approval No. [2025]113-001; approval date: 2025-07-26). The requirement for informed consent was waived because of the retrospective design and the use of de-identified data. Consecutive patients with degenerative spinal deformity involving the thoracolumbar/lumbar region who underwent staged LLIF-based correction between January 2019 and January 2021 were included. All patients had failed at least 1 year of conservative treatment. Exclusion criteria were prior spinal surgery, spinal neoplasm, and spinal tuberculosis. The primary objective was to characterize radiographic changes across the two operative stages and at latest follow-up, with particular attention to coronal balance maintenance after correction of the fractional-curve region.

2.2. Surgical strategy and technique

Baseline CI was identified according to previous literature (Bao et al., 2016). Patients were selected for staged correction when severe coronal and/or sagittal deformity required stepwise correction and reassessment after anterior column release. Preoperative indications for the staged approach included one or more of the following features: pelvic incidence minus lumbar lordosis mismatch >30°, fractional-curve take-off angle >15°, or rigid thoracolumbar kyphosis.

• Stage I consisted of multi-level LLIF through the ante-psoas corridor from L1–2 to L4–5 (Fig. 1A-B). No formal anterior longitudinal ligament release or anterior column realignment technique was performed in this cohort. After stage I, full-length standing radiographs and computed tomography were obtained to reassess coronal type, sagittal alignment, fusion levels, and the need for further posterior correction.

• Stage II consisted of posterior percutaneous screw fixation (PSF) alone in patients whose residual deformity no longer required additional posterior release. For residual coronal deformity with fractional-curve take-off angle (FC-TO) ≥10°, mini-open Schwab grade 2 posterior column osteotomy was performed at the fractional-curve region, typically below L4, with selective convex-side compression and concave-side distraction to restore a more horizontal lumbosacral foundation (Fig. 1C-E). For residual sagittal malalignment, mini-open Schwab grade 2 osteotomy and/or transforaminal lumbar interbody fusion (TLIF) was performed at the kyphotic apex as indicated (Schwab et al., 2014). Preoperative CT was reviewed to identify rigid or spontaneously fused segments, defined by bridging osteophytes, disc-space ankylosis, facet-joint fusion, and/or severe asymmetric disc collapse.

Fig. 1.

Schematic illustration of the staged LLIF procedure combined with mini-open osteotomy. The first-stage multi-level LLIF (A–B) partially corrects deformity and releases the anterior column, leading to a temporary coronal shift toward the convex side. After reassessment, the second-stage mini-open osteotomy (C) is performed to release the posterior column. Following rod insertion (D), compression on the convex side and distraction on the concave side are applied at the FC to horizontalize the L4–L5 endplates and achieve final global realignment (E).

L5–S1 fusion was considered when symptoms originated from the L5–S1 segment, advanced degeneration was present, or lumbosacral horizontalization required inclusion of the lumbosacral junction. When L5–S1 fusion was required, oblique lumbar interbody fusion at L5–S1 (OLIF51) was performed during stage I, whereas MIS-TLIF/TLIF was used during stage II when direct posterior decompression or supplemental posterior reconstruction at L5–S1 was necessary.

2.3. Radiographic and clinical outcomes

Standing full-length coronal and sagittal radiographs were obtained preoperatively, after the first-stage LLIF, after PSF, and at the latest follow-up. Coronal measurements included coronal Cobb angle (CCA), coronal vertical axis (CVA; horizontal distance between the C7 plumb line and the central sacral vertical line), and a study-specific FC-TO, defined as the angle between the superior endplate of the proximal vertebra of the lumbosacral fractional-curve region, typically L4, and the horizontal plane. FC-TO was used as a coronal foundation index to quantify lumbosacral obliquity rather than as a conventional fractional-curve Cobb measurement. Sagittal measurements included pelvic incidence (PI), pelvic tilt (PT), lumbar lordosis (LL), pelvic incidence minus lumbar lordosis mismatch (PI-LL mismatch), T1 pelvic angle (TPA), and sagittal vertical axis (SVA). Coronal alignment was categorized into three types according to the Qiu coronal classification. Type A was defined as a coronal balance distance <30 mm. Type B was defined as a coronal balance distance ≥30 mm with the C7 plumb line shifted toward the concavity of the major curve, whereas Type C was defined as a coronal balance distance ≥30 mm with the C7 plumb line shifted toward the convexity of the major curve (Bao et al., 2016). Classification was reassessed at each radiographic time point.

All radiographic measurements were performed using Surgimap software (version 2.2.15.5, New York, USA). Measurements were independently obtained by two spine surgeons. Clinical outcomes, including the Oswestry Disability Index (ODI) and visual analog scale (VAS) scores for back and leg pain, were analyzed preoperatively and at the latest follow-up. Minimum clinically important difference (MCID) achievement at the latest follow-up was defined a priori as an improvement of at least 12.8 points for ODI and at least 2.0 points for VAS back pain and VAS leg pain (Copay et al., 2008; Farrar et al., 2001). To explore the relationship between radiographic realignment and clinical improvement, patient-level change scores were calculated from baseline to the latest follow-up. Improvement in ODI and VAS scores was defined as the preoperative value minus the follow-up value. Radiographic correction was defined as the preoperative value minus the follow-up value for CCA, CVA, FC-TO, PI–LL mismatch, TPA, and SVA, whereas LL correction was defined as the follow-up value minus the preoperative value.

2.4. Bone mineral density and mechanical complications

Preoperative dual-energy X-ray absorptiometry (DXA) examinations performed within 12 months before the first-stage procedure were retrospectively reviewed. Areal bone mineral density and T-scores at the lumbar spine, femoral neck, and total hip were extracted from the formal DXA reports. Bone-density status was classified according to the lowest valid T-score among the lumbar spine, femoral neck, and total hip as normal bone density (T-score ≥ −1.0), low bone mass (−2.5 < T-score < −1.0), or osteoporosis (T-score ≤ −2.5).

Postoperative and latest follow-up radiographs and computed tomography images, when available, were systematically reviewed for mechanical complications, including cage subsidence, screw loosening, rod breakage, proximal junctional kyphosis, pseudarthrosis, clinically relevant loss of correction, and revision surgery attributable to a mechanical complication. Clinically relevant loss of correction was determined by serial comparison of the post-PSF and latest follow-up radiographs in conjunction with the clinical record.

2.5. Statistical analysis

Data were analyzed using GraphPad Prism 10.0 (GraphPad Software, San Diego, CA, USA) and are presented as mean ± standard deviation. Repeated-measures analyses were used to assess radiographic changes across the four time points, with post hoc pairwise comparisons where appropriate. Patient-reported outcome measures (PROMs), including ODI and VAS scores, were compared between the preoperative assessment and the latest follow-up. Spearman rank correlation coefficients were calculated between changes in patient-reported outcomes and changes in radiographic parameters. A two-sided p value < 0.05 was considered statistically significant. Patients were additionally stratified according to the Qiu coronal classification to describe dynamic type conversion across stages; these analyses were exploratory and descriptive rather than intended to establish causal inference.

3. Results

A total of 51 patients were included, comprising 7 men and 44 women with a mean age of 70.6 ± 7.9 years. The demographic and radiographic parameters at each time point are summarized in Table 1, and detailed operative characteristics of the staged procedure are summarized in Table 2. Stage I involved 167 LLIF levels overall, corresponding to 3.3 ± 0.7 levels per patient; 7 patients (13.7%) underwent 2-level LLIF, 23 (45.1%) underwent 3-level LLIF, and 21 (41.2%) underwent 4-level LLIF. The treated LLIF segments most commonly included L2–3, L3–4, and L4–5, and an additional OLIF51 at L5–S1 was performed in 7 patients (13.7%). Stage II consisted of posterior percutaneous pedicle screw fixation in all patients; 21 (41.2%) underwent fixation alone, whereas 30 (58.8%) required supplemental posterior procedures. These supplemental procedures were not mutually exclusive. Among them, 19 patients (37.3%) underwent mini-open posterior Ponte osteotomy (Schwab grade 2), and 17 (33.3%) underwent additional posterior interbody fusion (MIS-TLIF/TLIF). At the segmental level, posterior fusion was most frequently performed at L5–S1 (9 segments, 17.6%). The mean follow-up duration was 26.3 ± 11.5 months.

Table 1.

Demographic and radiographic characteristics of patients undergoing staged LLIF-based deformity correction.

Number	51
Age (yrs)	70.61 ± 7.88
Gender M/F (% of F)	7/44 (86.27)
	Pre-op	After LLIF	P value (Pre-op VS After LLIF)	After PSF	P value (Pre-op VS After PSF)	Latest follow-up	P value (Pre-op VS Latest follow-up)
CCA(°)	35.62 ± 17.19	21.93 ± 11.59	<0.0001	10.20 ± 5.78	<0.0001	10.91 ± 7.01	<0.0001
CVA (mm)	28.50 ± 18.32	22.09 ± 14.68	0.0641	17.18 ± 14.82	0.0010	14.95 ± 10.28	0.0002
FC-TO	17.28 ± 8.83	12.42 ± 7.25	0.0043	5.67 ± 3.69	<0.0001	5.88 ± 3.79	<0.0001
PI(°)	44.53 ± 9.76	46.50 ± 10.41	0.3401	46.14 ± 9.38	0.4015	47.47 ± 8.75	0.1550
LL(°)	22.40 ± 16.97	31.14 ± 16.71	0.0128	34.33 ± 10.33	<0.0001	35.66 ± 11.82	0.0002
PI-LL mis-match(°)	22.70 ± 15.52	16.13 ± 13.16	0.0289	11.86 ± 10.27	<0.0001	13.47 ± 9.55	0.0024
PT(°)	24.61 ± 7.95	22.16 ± 8.64	0.1515	20.10 ± 8.66	0.0078	22.31 ± 7.90	0.1915
TPA(°)	23.22 ± 9.65	19.86 ± 8.55	0.0760	17.46 ± 7.12	0.0010	18.67 ± 7.85	0.0245
SVA (mm)	62.20 ± 49.30	35.68 ± 31.24	0.0026	31.29 ± 23.44	0.0001	29.55 ± 20.58	0.0005

M, male; F, female; Pre-op, preoperative; PSF, percutaneous screw fixation; CCA, coronal Cobb angle, CVA, coronal vertical axis, FC-TO, fractional-curve take-off angle, PI, pelvic incidence, LL, lumbar lordosis, PI-LL, pelvic incidence minus lumbar lordosis, PT, pelvic tilt, TPA, T1 pelvic angle, SVA, sagittal vertical axis.

Table 2.

Operative characteristics of the staged correction procedure.

Characteristic	Value
Stage I: lateral procedure
Total LLIF levels treated, n	167
LLIF levels per patient	3.3 ± 0.7
Number of LLIF levels, n (%)
2 levels	7 (13.7)
3 levels	23 (45.1)
4 levels	21 (41.2)
LLIF level included, n (%)
L1-2	23 (45.1)
L2-3	47 (92.2)
L3-4	50 (98.0)
L4-5	47 (92.2)
Additional OLIF51, n (%)	7 (13.7)
Stage II: posterior procedure
Posterior percutaneous pedicle screw fixation, n (%)	51 (100.0)
Posterior fixation alone, n (%)	21 (41.2)
Any supplemental posterior procedure, n (%)	30 (58.8)
Supplemental posterior procedure type, n (%)
Mini-open posterior Ponte osteotomy (Schwab grade 2)	19 (37.3)
Additional posterior interbody fusion (MIS-TLIF/TLIF)	17 (33.3)
Additional posterior interbody fusion level(s), n (%)
L1-L2	5 (9.8)
L2-L3	2 (3.9)
L3-L4	1 (2.0)
L4-L5	3 (5.9)
L5-S1	9 (17.6)
Mini-open Ponte segment(s), n (%)
T11-T12	1 (2.0)
T12-L1	4 (7.8)
L1-L2	6 (11.8)
L2-L3	3 (5.9)
L3-L4	2 (3.9)
L4-L5	11 (21.6)
L5-S1	2 (3.9)

OLIF51, oblique lumbar interbody fusion at L5–S1.

Values are presented as n (%), mean ± SD, or total number of treated levels. Subcategories under supplemental posterior procedures are not mutually exclusive. For posterior fusion and mini-open Ponte, involved motion segments are counted individually; multi-segment procedures contribute to each included segment.

3.1. Overall radiographic correction

Both coronal and sagittal alignments improved significantly after staged surgery and remained stable during follow-up. The mean CCA decreased markedly from 35.62 ± 17.19° preoperatively to 21.93 ± 11.59° after LLIF and 10.20 ± 5.78° after PSF, maintaining 10.91 ± 7.01° at the latest follow-up (all P < 0.0001 vs. preoperative). The CVA showed an overall decreasing trend from 28.50 ± 18.32 mm preoperatively to 22.09 ± 14.68 mm after LLIF, with significant improvement achieved after PSF (17.18 ± 14.82 mm) and maintained at follow-up (14.95 ± 10.28 mm). Likewise, the FC-TO, reflecting the horizontalization of the lumbosacral foundation, decreased from 17.28 ± 8.83° preoperatively to 12.42 ± 7.25° after LLIF (P = 0.0043), 5.67 ± 3.69° after PSF (P < 0.0001), and 5.88 ± 3.79° at follow-up (P < 0.0001). FC-TO showed continuous improvement across all timepoints, whereas CVA demonstrated an overall improving trend with further correction after PSF (Fig. 3).

Fig. 3.

Line plots showing progressive improvement of coronal parameters at each surgical stage. (A) CCA, (B) CVA, and (C) FC-TO were measured preoperatively, after LLIF, after PSF, and at the latest follow-up. CCA and FC-TO decreased significantly across staged treatment, whereas CVA showed an overall improving trend with significant improvement after posterior fixation and at follow-up.

3.2. Sagittal alignment improvement

In the sagittal plane, LL increased significantly from 22.40 ± 16.97° to 31.14 ± 16.71° after LLIF and 34.33 ± 10.33° after PSF, maintaining 35.66 ± 11.82° at the final follow-up (P < 0.001). The PI–LL mismatch decreased from 22.70 ± 15.52° preoperatively to 11.86 ± 10.27° after PSF (P < 0.001), indicating improved sagittal harmony. Parameters reflecting global balance—including TPA and SVA—showed consistent correction and stability: TPA decreased from 23.22 ± 9.65° to 17.46 ± 7.12°, and SVA from 62.20 ± 49.30 mm to 31.29 ± 23.44 mm (P ≤ 0.001), with both maintaining near-normal values at follow-up.

3.3. Stratified analysis by qiu coronal type

According to the Qiu coronal classification, 26 patients were Type A, 16 were Type B, and 9 were Type C before surgery (Table 3). There were no significant differences in age (P = 0.73) or gender distribution (P = 0.57) among groups.

Table 3.

Demographics and radiographic parameters of patients with DSD in Type A, Type B and Type C using Qiu coronal classification.

		A	B	C
	Age (yrs)	70.00 ± 8.26	70.56 ± 7.47	72.44 ± 8.10
	Gender M/F (% of F)	4/22 (84.62)	1/15 (93.75)	2/7 (77.78)
Pre-op	CCA(°)	31.83 ± 18.79	42.66 ± 13.36	34.07 ± 16.16
	CVA (mm)	13.85 ± 6.89	41.48 ± 8.94	47.73 ± 18.71
	FC-TO(°)	15.74 ± 8.30	16.17 ± 7.73	23.72 ± 10.14
After LLIF	CCA(°)	19.85 ± 11.82	28.33 ± 12.02	17.79 ± 5.21
	CVA (mm)	18.00 ± 14.53	24.05 ± 16.24	31.20 ± 6.98
	FC-TO(°)	10.63 ± 7.62	13.55 ± 5.94	15.98 ± 7.18
After PSF	CCA(°)	9.84 ± 6.00	11.23 ± 6.67	9.34 ± 3.17
	CVA (mm)	18.36 ± 15.25	14.41 ± 15.67	18.94 ± 12.83
	FC-TO(°)	4.78 ± 3.66	6.18 ± 3.71	7.17 ± 3.42
Latest follow-up	CCA(°)	10.64 ± 7.58	11.98 ± 8.11	9.50 ± 2.78
	CVA (mm)	17.55 ± 11.41	12.41 ± 8.80	14.10 ± 10.42
	FC-TO(°)	5.41 ± 3.89	5.69 ± 3.93	7.21 ± 3.52

M, male; F, female; Pre-op, preoperative; PSF, percutaneous screw fixation; CCA, coronal Cobb angle, CVA, coronal vertical axis, FC-TO, fractional-curve take-off angle.

After the first-stage LLIF, some patients with Type A deformity showed temporary worsening of CVA and were reclassified as Type B or C due to trunk shift toward the convex side. Following FC correction during the second stage, most of these patients regained balanced alignment and reverted to Type A, only one patient remained Type C. The Sankey diagram (Fig. 2) illustrates this dynamic transformation across surgical stages. Quantitatively, CCA and FC-TO decreased progressively through both stages in all groups, whereas CVA improvement was stage-dependent—it transiently deteriorated after LLIF in Type A patients but improved gradually after PSF in Types B and C. These stage-wise trends were descriptively consistent across the three Qiu subtypes, with progressive reductions in CCA and FC-TO and subtype-dependent changes in CVA.

Fig. 2.

Sankey diagram demonstrating dynamic transformation of Qiu coronal classifications before and after each surgical stage. Patients were classified as Type A, B, or C based on coronal balance patterns. Flows indicate transitions between stages—preoperative, post-LLIF, and post-PSF—showing that some Type A patients temporarily shifted to Type B or C after LLIF but reverted to Type A following FC correction. The Sankey diagram is used for descriptive visualization rather than statistical inference.

3.4. Clinical outcomes

At the latest follow-up, pain and disability had improved significantly. The mean VAS score for back pain decreased from 5.23 ± 1.54 preoperatively to 1.00 ± 0.82, and the ODI improved from 61.68 ± 14.47 to 16.45 ± 7.22 (both P < 0.001). MCID was achieved by 95.6% of patients for ODI and by 63.6% and 72.7% of patients for VAS back pain and VAS leg pain, respectively. No statistically significant correlations were identified between the magnitude of radiographic correction and improvement in ODI, VAS back pain, or VAS leg pain. Spearman correlation coefficients between ODI improvement and changes in CCA, CVA, FC-TO, LL, PI–LL mismatch, TPA, and SVA ranged from −0.168 to 0.189 (all P ≥ 0.400). Similarly, no significant correlations were observed between VAS back-pain improvement and the analyzed radiographic parameters (ρ range, 0.001–0.265; all P ≥ 0.246), or between VAS leg-pain improvement and radiographic correction (ρ range, −0.109 to 0.296; all P ≥ 0.180). Full correlation results are presented in Table 4.

Table 4.

Correlations between radiographic correction and improvement in patient-reported outcomes.

	ΔODI ρ	P value	ΔVAS-back ρ	P value	ΔVAS-leg ρ	P value
ΔCCA	0.134	0.553	0.194	0.387	0.079	0.728
ΔCVA	−0.129	0.569	0.051	0.822	0.267	0.230
ΔFC-TO	0.189	0.400	0.001	0.996	−0.072	0.749
ΔLL	−0.126	0.577	0.121	0.592	0.108	0.632
ΔPI–LL	−0.168	0.454	0.151	0.501	0.296	0.180
ΔTPA	−0.071	0.761	0.265	0.246	0.232	0.312
ΔSVA	0.072	0.757	0.046	0.844	−0.109	0.639

Positive change values indicate greater radiographic or clinical improvement. Correlations were assessed using Spearman rank correlation coefficients.

Preoperative DXA data were available for 40 of 51 patients (78.4%), with a mean interval of 3.4 ± 2.4 months between DXA examination and surgery. The mean lowest valid T-score was −1.3 ± 0.8. According to the lowest valid T-score, 17 of 40 patients (42.5%) had normal bone density and 23 of 40 (57.5%) had low bone mass. None of the patients with available DXA data met the densitometric criterion for osteoporosis.

During follow-up, radiographic cage subsidence was identified in 13 of 51 patients (25.5%). Screw loosening and proximal junctional kyphosis each occurred in 1 patient (2.0%). No rod breakage, pseudarthrosis, clinically relevant loss of correction, or revision surgery attributable to a mechanical complication was observed. Bone-density findings and mechanical complications are summarized in Table 5. No major intraoperative neurological, vascular, or visceral complications occurred. Transient postoperative femoral nerve-related symptoms occurred in 15 of 51 patients (29.4%), predominantly after the lateral stage, and were characterized by hip flexion weakness or anterior thigh dysesthesia. All improved with conservative management and had resolved by the latest follow-up, with no persistent motor deficit.

Table 5.

Preoperative bone-density status and postoperative mechanical complications.

A. Preoperative DXA characteristics
Characteristic	Value
Patients with available preoperative DXA, n/N (%)	40/51 (78.4)
Interval from DXA to surgery, months	3.4 ± 2.4
Lowest valid T-score	−1.3 ± 0.8
Bone-density category, n/N_DXA (%)
Normal bone density	17/40 (42.5)
Low bone mass	23/40 (57.5)
Osteoporosis	0/40 (0)

B. Mechanical complications during follow-up
Mechanical complication	n/N (%)
Cage subsidence	13/51 (25.5)
Screw loosening	1/51 (2.0)
Rod breakage	0/51 (0)
Proximal junctional kyphosis	1/51 (2.0)
Pseudarthrosis	0/51 (0)
Clinically relevant loss of correction	0/51 (0)
Revision surgery for a mechanical complication	0/51 (0)

Values are presented as mean ± standard deviation, or n/N (%), as appropriate. Percentages for DXA characteristics were calculated using the number of patients with available and valid DXA data as the denominator. Percentages for mechanical complications were calculated using the entire cohort of 51 patients as the denominator.

DXA, dual-energy X-ray absorptiometry.

4. Discussion

• 1Principal Findings

In this retrospective single-center cohort, we described a staged LLIF-based correction strategy for degenerative spinal deformity and evaluated radiographic changes across the two operative stages and at follow-up. The main finding was that coronal and sagittal alignment improved in a stepwise manner across staged treatment, while coronal correction was maintained at follow-up, although residual coronal malalignment remained in a small proportion of patients.

Importantly, the present study was not designed to determine comparative superiority over alternative strategies. Rather, it provides descriptive evidence that, in patients with residual deformity after lateral interbody fusion, selective stage II posterior escalation—including mini-open Ponte osteotomy and/or additional posterior interbody fusion at key distal segments—may help address transient postoperative coronal deviation and support maintenance of coronal alignment over time.

• 2Biomechanical and Surgical Rationale

CI is often underappreciated compared to sagittal malalignment, yet its impact on patient quality of life is substantial. In DSD, the FC serves as the key transitional segment linking the mobile lumbar spine to the fixed sacropelvic base. When the FC is rigid or oblique, excessive correction of the main thoracolumbar curve through lateral fusion alone may create a tilted foundation—known as the take-off phenomenon—and lead to convex trunk translation. Our data support that such transient coronal deviation can occur after multi-level LLIF, particularly in Type A patients, but can be reversed by horizontalizing the FC during the second stage.

Biomechanically, selective mini-open posterior column osteotomy complements the anterior column support achieved by LLIF by increasing local mobility where residual deformity persists after stage I (Schwab et al., 2014). In the present cohort, the segmental distribution of stage II procedures was not random: mini-open Ponte osteotomy was performed most frequently at L4–5, whereas additional posterior interbody fusion was concentrated at L5–S1. This pattern supports the concept that the key residual deformity after lateral correction often remains centered in the distal lumbar/fractional-curve region and at the lumbosacral junction. Targeted convex-side compression and concave-side distraction across these levels allow fine adjustment of the coronal foundation and contribute to the marked reduction in FC-TO observed in this cohort.

• 3Comparison With Previous Studies

Several strategies have been proposed to manage coronal and sagittal imbalance simultaneously (Baroncini et al., 2023; Tempel et al., 2014; Zhang et al., 2024). Cecchinato et al. and Thambiraj et al. described asymmetric or unilateral pedicle subtraction osteotomy as an effective but highly invasive approach for biplanar deformity correction (Cecchinato et al., 2015; Thambiraj and Boszczyk, 2012). Bao et al. introduced a sequential posterior correction technique to decompose complex deformity correction into manageable steps (Bao et al., 2019). However, these techniques are associated with considerable blood loss, prolonged operative time, and increased perioperative risk, especially in elderly DSD populations (Barile et al., 2023).

Minimally invasive lateral approaches such as LLIF and extreme lateral interbody fusion have gained popularity for reducing muscle trauma and blood loss while providing satisfactory sagittal alignment restoration (Hiyama et al., 2023). Nonetheless, postoperative convex coronal malalignment remains a known complication after lateral-only correction due to insufficient FC management. Our staged strategy integrates the advantages of minimally invasive lateral approaches with targeted posterior correction, while maintaining acceptable morbidity in this cohort.

In the first stage of surgery, multi-level LLIF was performed to release the anterior column and to partially correct the coronal and sagittal deformities. The deformity was then re-evaluated. Dividing the three-dimensional deformity into the sagittal and coronal planes allows the surgeon to correct the deformity and achieve balance in each respective plane, and further, to select the appropriate MIS-TLIF and mini-open osteotomy (usually Ponte osteotomy) in the second stage of surgery (Zhang et al., 2023). Fig. 4 shows the outcome of staged orthopedic correction in a patient with Qiu Type B. In the sagittal plane, multi-level LLIF was sufficient in obtaining sagittal correction. Along with techniques such as anterior column realignment, hyperlordotic and expandable cages, and mini-open osteotomy, appropriate sagittal alignment, and spinopelvic harmony can be achieved using lateral MIS approaches (Berjano et al., 2015; Turner et al., 2015). In the coronal plane, the thoracolumbar/lumbar main curve is often easier to correct than the distal fractional curve. When main-curve correction exceeds fractional-curve leveling, the trunk may shift toward the convex side after multi-level LLIF, producing a temporary take-off phenomenon on an oblique lumbosacral foundation. Our subtype analysis suggests that this mismatch may transiently convert some balanced or concave-shift patterns into more convex-shifted patterns after stage I. These observations reinforce the importance of reassessing the distal foundation after lateral correction and selectively adding stage II posterior procedures to re-level the fractional-curve region rather than uniformly escalating posterior release in all patients.

Fig. 4.

Representative case of a 70-year-old female patient with Qiu Type B deformity (primary lumbar curve with compensatory fractional curve). Standing full-length coronal radiographs (A–D) and sagittal radiographs (E–H) are shown sequentially at the preoperative stage (A, E), after stage I lateral lumbar interbody fusion (LLIF) (B, F), after stage II posterior correction and fixation (C, G), and at the latest follow-up (D, H). The images demonstrate stepwise restoration of both coronal and sagittal alignment and maintenance of coronal balance at follow-up.

Because inadequate correction of the fractional-curve region may lead to postoperative coronal imbalance, the residual deformity after stage I must be reassessed in a level-specific manner rather than addressed with routine uniform posterior release (Daniels et al., 2024). In our series, all patients underwent posterior percutaneous fixation during stage II, but only 58.8% required escalation to supplemental posterior procedures. Moreover, this escalation was anatomically focused: posterior fusion clustered at L5–S1, whereas Ponte osteotomy clustered predominantly at L4–5 and adjacent distal lumbar levels. These operative patterns reinforce the technical rationale of the staged strategy, namely, that posterior correction was selectively added to re-level the distal foundation and lumbosacral take-off rather than to perform extensive posterior osteotomy throughout the entire construct. We believe this selective, segment-targeted escalation is one reason why transient post-LLIF coronal deviation could be corrected while maintaining acceptable procedural invasiveness. LLIF combined with posterior release can achieve a horizontal foundation by allowing spinal manipulation of all three columns (Uribe et al., 2018). Fig. 5 shows the staged correction process in a representative patient with Qiu Type C deformity. In our cohort, this staged strategy was associated with a reduction in mean FC-TO from 17.28 ± 8.83° preoperatively to 5.88 ± 3.79° at the latest follow-up.

Fig. 5.

Representative case of a 72-year-old female patient with Qiu Type C deformity (dominant fractional curve with compensatory thoracolumbar/lumbar curve). Standing full-length coronal radiographs (A–D) and sagittal radiographs (E–H) are shown sequentially at the preoperative stage (A, E), after stage I lateral lumbar interbody fusion (LLIF) (B, F), after stage II posterior correction and fixation (C, G), and at the latest follow-up (D, H). The images show partial correction after stage I, followed by restoration of the lumbosacral foundation and global coronal balance after stage II, with radiographic maintenance at follow-up.

The magnitude and stage-wise pattern of correction in the present cohort are consistent with previous LLIF-based deformity series. Tempel et al. reported significant improvement after combined LLIF and posterior stabilization, with VAS back pain decreasing from 7.5 to 4.3 and ODI from 48 to 38 (Tempel et al., 2014). In our cohort, VAS back pain decreased from 5.23 to 1.00 and ODI from 61.68 to 16.45, although direct comparison should be made cautiously because of differences in baseline disability, patient selection, and follow-up. Hiyama et al. similarly demonstrated that the response of coronal malalignment to staged LLIF is influenced by coronal pattern and distal lumbar tilt (Hiyama et al., 2023), which is consistent with the subtype-dependent changes in CVA observed after stage I in our cohort. Zhang et al. reported that Ponte osteotomy combined with posterior lumbar interbody fusion reduced the coronal Cobb angle from 18.1° to 5.8°, CVA from 31 to 12 mm, and SVA from 56 to 31 mm, while increasing LL from 22.2° to 36.0° at follow-up (Zhang et al., 2023). Our corresponding final values—CCA 10.91°, CVA 14.95 mm, SVA 29.55 mm, and LL 35.66°—were broadly comparable, despite a larger preoperative coronal Cobb angle of 35.62°. Together, these comparisons suggest that selective posterior escalation after LLIF can achieve correction within the range reported for more uniform posterior osteotomy strategies, while avoiding osteotomy in all patients.

Although both radiographic alignment and patient-reported outcomes improved significantly, the magnitude of radiographic correction was not significantly correlated with the degree of ODI or VAS improvement at the individual-patient level. This finding suggests that clinical recovery after staged deformity correction may not be determined solely by the magnitude of global radiographic realignment. Neural decompression, baseline symptom burden, muscular condition, comorbidities, and other patient-specific factors may also contribute to postoperative improvement.

DSD is frequently accompanied by vertebral rotation, and derotation is an important component of three-dimensional deformity correction. In the present study, derotation could be applied during PSF instrumentation to enhance coronal correction when appropriate. However, derotation inevitably increases the magnitude of main curve correction. In patients classified as Qiu Type A or Type C, aggressive derotation of the main curve may increase the risk of postoperative CI if not properly matched with correction of the lumbosacral FC. Accordingly, derotation in our practice was applied selectively and cautiously, only when adequate coordination between main curve correction and FC leveling could be achieved.

This observation underscores the importance of balancing coronal correction across different curve components rather than pursuing maximal correction of a single segment (Park et al., 2025).

• 4Clinical Implications and Limitations

Clinically, this staged approach offers a reproducible and safe strategy for DSD patients—particularly elderly individuals with rigid deformities—where single-stage osteotomies may pose excessive risk. The separation of anterior and posterior procedures enables surgeons to reassess alignment between stages and plan tailored correction for both planes. Moreover, using percutaneous instrumentation minimizes paraspinal trauma and facilitates early mobilization.

Bone-density assessment showed that 57.5% of patients with available DXA data had low bone mass, although none met the densitometric criterion for osteoporosis. Radiographic cage subsidence occurred in 25.5% of the overall cohort and was therefore not negligible. Nevertheless, no clinically relevant loss of correction, rod breakage, pseudarthrosis, or revision surgery for a mechanical complication was observed during the available follow-up. These findings suggest that the observed cage subsidence did not progress to clinically relevant construct failure or loss of global alignment within the follow-up period.

Preoperative DXA data were unavailable in 21.6% of the cohort, and the absence of patients meeting the DXA criterion for osteoporosis prevented a formal comparison between osteoporotic and non-osteoporotic patients. In addition, the relatively small sample size and low number of major mechanical complications limited statistical assessment of the relationship between bone density and implant-related failure. Lumbar DXA measurements may also be affected by degenerative changes, vertebral rotation, osteophytes, and endplate sclerosis in patients with spinal deformity. Larger studies incorporating standardized hip DXA measurements and longer follow-up are required to clarify the effect of bone quality on correction durability.

Several limitations should be acknowledged. First, this was a retrospective single-center cohort study without a control group, which limits causal inference and external generalizability. Second, although stage-specific radiographs were available, postoperative pain, standing tolerance, and muscle guarding may still have influenced the measured perioperative alignment at different time points; accordingly, the stage-dependent radiographic findings should be interpreted as descriptive associations rather than isolated effects of a single procedural step. Third, although preoperative DXA data were available for most patients, they were unavailable in 21.6% of the cohort, and no patient met the densitometric criterion for osteoporosis. Consequently, the influence of osteoporosis on mechanical complications and correction durability could not be formally evaluated. Fourth, preoperative curve flexibility was not quantified using standardized bending radiographs. Although CT and the interval assessment after LLIF provided clinically useful information regarding residual rigidity, they cannot replace a formal preoperative flexibility assessment; this limitation may have influenced patient selection and the choice of posterior correction.

The corrective capacity of this staged strategy has important limits. It is most applicable when meaningful correction can be obtained through multilevel anterior-column release, interbody cage placement, posterior instrumentation, and selective Schwab grade 2 osteotomy. It should not be considered a substitute for a three-column osteotomy in patients with severe fixed sagittal malalignment, sharply angular kyphosis, extensive circumferential ankylosis, marked vertebral translation, or residual deformity that cannot be mobilized after the lateral stage. In such cases, Schwab grade 3–5 osteotomy or another more powerful reconstruction may still be required. In addition, the staged strategy entails two procedures and an interval reassessment, and temporary worsening of coronal alignment may occur after the lateral stage before definitive posterior correction.

5. Conclusion

In this retrospective cohort, staged LLIF combined with targeted mini-open posterior correction was associated with stepwise improvement in coronal and sagittal alignment and with maintenance of coronal balance at follow-up. These findings support the clinical relevance of addressing the lumbosacral fractional-curve region in rigid thoracolumbar/lumbar deformity, but they require confirmation in prospective or comparative studies.

Ethics approval and consent to participate

This study was approved by the Clinical Research Ethics Committee of Xuanwu Hospital, Capital Medical University (Approval No. [2025]113-001; approval date: 2025-07-26). The requirement for informed consent was waived because of the retrospective design and the use of de-identified patient data. All procedures were performed in accordance with the Declaration of Helsinki and relevant institutional guidelines.

Author contributions

Conceptualization: HW; Data curation: LZ, XYZ, YTW;

Formal Analysis: LZ, ZLL, XYZ, KW;

Funding acquisition: HW;

Methodology: LZ, KW, YHL; Project Administration: HW; Writing – Original Draft: LZ, KW; Writing – Review & Editing: HW.

Funding

The study was funded by the Beijing Natural Science Foundation Changping Innovation Joint Fund Project (L234059), National Key Research and Development Program of China (2023YFC2507703).

Competing interests

The authors declare that they have no competing interests.