Risk Factors for Immediate Postoperative Coronal Imbalance in Degenerative Lumbar Scoliosis Patients Fused to Pelvis.

STUDY
DESIGN: Retrospective case-control radiographic study.
OBJECTIVE: To identify risk factors for immediate postoperative coronal imbalance in degenerative lumbar scoliosis (DLS) patients fused to pelvis.
METHODS: A total of 71 DLS patients treated with deformity correction surgery were reviewed. Measurements included coronal parameters such as global coronal malalignment (GCM), major Cobb angle, L4/L5 coronal tilt and sagittal parameters. Based on the orientation of L4 coronal tilt relative to C7 plumb line (PL) preoperatively, coronal patterns were subdivided into (1) consistency pattern, L4 coronally tilts toward C7 PL, and (2) opposition pattern, L4 coronally tilts opposite C7 PL; the proportion of these 2 patterns was analyzed. Also, the proportion of type C and surgical factors were recorded. According to postoperative GCM, patients were divided into imbalanced group and balanced group.
RESULTS: Compared with the balanced group, the imbalanced group had a larger proportion of preoperative consistency pattern (79.17% vs 40.43%, P = .002), larger proportion of type C (29.17% vs 8.51%, P = .023), and lower amount of GCM correction (-4.92 ± 24.25 vs 14.52 ± 19.49 mm, P < .001). There were no significant intergroup differences regarding preoperative and postoperative major Cobb angle, preoperative and postoperative L4 or L5 coronal tilt, major Cobb correction, the amounts of correction of coronal tilt of L4/L5, osteotomy levels, osteotomy grades, instrumented levels, levels of interbody fusion, and distribution of upper instrumented vertebra.
CONCLUSIONS: In addition to known risk factors such as type C, preoperative coronal consistency pattern might be a new risk factor for postoperative coronal imbalance in DLS patients fused to pelvis.

Introduction

Restoration of spinal balance is one of the main purposes of corrective spinal deformity surgeries for patients with adult scoliosis.[1] Achieving postoperative sagittal balance has received more attention because sagittal imbalance has been proven to have great impact on health-related quality of life (HRQoL)[2]; however, recently, there have been many studies showing that postoperative coronal imbalance also affects HRQoL in adults.[3-5]

Nevertheless, restoration of postoperative coronal balance in long deformity corrections can be challenging. Ploumis et al[6] found that the prevalence of coronal malalignment in 54 adult spinal deformity (ASD) patients after deformity correction surgeries did not improve when compared with the preoperative state, although sagittal malalignment was significantly improved by surgery. Kurra et al[7] reported that there was no significant improvement in coronal malalignment in a total of 50 ASD patients treated with deformity correction and long instrumented fusion to the pelvis postoperatively versus preoperatively. For ASD patients extensively fused to the pelvis, the spine has little ability to compensate by itself once postoperative imbalance occurs, resulting in a permanently fixed decompensation.[8]

However, there are few reports about risk factors for failed restoration of coronal balance in patients with adult scoliosis fused to pelvis. Bao et al[9] studied 69 surgically treated degenerative lumbar scoliosis (DLS) patients and found that those with preoperative global coronal malalignment (GCM) greater than 30 mm and trunk shifting to the convex side (type C) had a higher rate of postoperative coronal imbalance than those who had C7 plumb line (PL) shifted to the concave side. Lewis et al[10] believed that coronal tilt of L4 and L5 were associated with postoperative coronal imbalance. Other mentioned risk factors included osteoporosis, revision surgery at the same site, and anterior surgery.[6] In this study, we investigated a series of DLS patients treated with deformity correction and instrumented fusion to the pelvis via posterior-only approach to identify the key risk factors for immediate coronal imbalance after surgery.

Materials and Methods

Patient Population

Ethics approval by PLA General Hospital (No. 20 190 602 113) was obtained prior to this study. We retrospectively collected data of DLS patients in 1 single institution between January 2015 and April 2019. Inclusion criteria were as follows: primary spinal deformity correction and fusion to the pelvis through a posterior-only approach. Exclusion criteria included the following: fusion levels <5, history of hip or knee arthroplasty, and absolute discrepancy of leg length >20 mm. A total of 71 DLS patients fused to pelvis were identified and enrolled in this study eventually.

Surgical Techniques

After exposure, pedicle screws were placed segmentally and bilaterally at each level in the construct. To obtain pelvic fixation, 17 patients were fitted with S1 screws only, 19 patients with S1 screws plus iliac wing screws, and 35 patients with S1 screws plus S2 alar-iliac screws. To improve deformity correction, spinal osteotomies were performed in all patients, including Schwab grade I (facetectomy) and grade II (Ponte osteotomy). Decompression and transforaminal lumbar interbody fusion were performed at the caudal levels (L2-S1) if foraminal compression existed or augmentation of fusion as an anterior column support was needed.

Radiographic Evaluation

Full-spine standing anteroposterior and lateral radiographs were analyzed preoperatively and at discharge from hospital or postoperatively at 2 weeks. The measurements were done using Surgimap (version 2.2.15; Spine Software, New York, NY) by 2 independent researchers, and the mean values were collected for analysis. Coronal radiographic measurements included the following: (1) GCM, defined as the horizontal distance between the C7 PL and central sacral vertical line; (2) major Cobb angle, defined as the angle between the superior endplate of the most tilted vertebra cranially and the inferior endplate of the most tilted vertebra caudally; (3) L4 coronal tilt, defined as the angle between the superior endplate of L4 and the horizontal line[9]; and (4) L5 coronal tilt, defined as the angle between the superior endplate of L5 and the horizontal line.[9] Sagittal parameters included the following: (1) thoracic kyphosis (T5-T12), (2) thoracolumbar kyphosis (T10-L2), (3) lumbar lordosis (L1-S1), (4) sacral slope, (5) pelvic tilt, (6) pelvic incidence, and (7) sagittal vertical axis.

Because spinal osteotomies were performed in all patients, osteotomy levels and osteotomy grades were also analyzed. Distribution of upper instrumented vertebra (UIV) and the levels of interbody fusion were recorded as well.

Based on the orientation of L4 coronal tilt relative to C7 PL preoperatively, coronal patterns were subdivided into the following: (1) consistency pattern, L4 coronally tilts toward C7 PL (Figure 1A); and (2) opposition pattern, L4 coronally tilts opposite C7 PL (Figure 1B).

Figure 1.

Coronal patterns based on the relationship between orientation of L4 coronal tilt and C7 plumb line. Consistency pattern (A) and opposition pattern (B). Abbreviation: GCM, global coronal malalignment.

Based on the absolute values of GCM preoperatively, coronal imbalance types were classified into 3 types: Type A, GCM is less than 30 mm; type B, GCM is ≥30 mm and C7 PL is on the concave side of the major curve; and type C, GCM is ≥30 mm and C7 PL is on the convex side of the major curve.[9]

Coronal imbalance postoperatively was defined as GCM ≥20 mm.[11] According to the absolute values of GCM postoperatively, the patients were divided into an imbalanced group (GCM ≥ 20 mm) and a balanced group (GCM < 20 mm).

Statistical Analysis

Continuous variables between groups were compared using the independent-samples t test and within groups using the paired-samples t test. Categorical variables were compared using χ2 analysis. Statistical analysis was performed using SPSS computer software (version 24; SPSS, Chicago, IL). P <.05 was set as statistical significance.

Results

Demographics and Operative Data

A total of 71 patients (11 male, 60 female; mean age at surgery: 63.24 years) were included in the final analysis (Table 1), of whom a total of 24 patients (33.80%) were imbalanced and 47 patients were balanced. The mean instrumented levels were 9.17, ranging from 5 to 16 segments. All patients received spinal osteotomies, including Schwab grade 1 in 11 cases and grade 2 in 60 cases. The mean osteotomy levels were 3.38, ranging from 2 to 5; The mean osteotomy grades were 1.85, ranging from Schwab grade 1 to grade 2. The mean levels of interbody fusion were 1.68, ranging from 0 to 4 levels. UIVs at T10 or above were seen in 45 cases, and UIVs below T10 were seen in 26 cases.

Table 1.

Demographics and Operative Parameters Between the 2 Groups.a

	Total	Imbalanced Group	Balanced Group	P Value
No. of patients	71	24	47
Sex (M:F)	11:60	2:22	9:38	.233
Age at surgery (years)	63.24 ± 7.04	62.71 ± 7.46	63.51 ± 6.88	.653
Instrumented levels	9.17 ± 2.51	8.88 ± 2.85	9.32 ± 2.33	.484
UIV (T10 or above:below)	45:26	16:8	29:18	.681
Interbody fusion levels	1.68 ± 1.31	1.54 ± 1.28	1.74 ± 1.32	.540
Osteotomy grades	1.85 ± 0.36	1.83 ± 0.38	1.85 ± 0.36	.848
Osteotomy levels	3.38 ± 0.85	3.45 ± 0.83	3.34 ± 0.87	.585

Abbreviations: UIV, upper instrumented vertebra.

a Results are given as the number or the mean ± SD unless otherwise stated.

General Correction Results

Both the coronal parameters and sagittal parameters (except pelvic incidence) were significantly corrected after the surgery. In terms of coronal parameters, GCM significantly decreased from 21.50 mm preoperatively to 14.19 mm postoperatively (P = .006). The major Cobb angle was markedly corrected from 26.50° preoperatively to 8.79° postoperatively (P < .001); L5 coronal tilt and L4 coronal tilt were significantly improved from 9.45° preoperatively to 4.59° postoperatively (P < .001) and from 14.18° preoperatively to 6.00° postoperatively (P < .001), respectively. The sagittal parameters measured preoperatively and postoperatively can also be seen in Table 2.

Table 2.

Coronal and Sagittal Parameters Before and After Surgery in 71 Patients.a

Coronal parameters	Preoperative	Postoperative	P Value
Major Cobb angle (degrees)	26.50 ± 15.73	8.79 ± 6.75	.000
Coronal L5 tilt (degrees)	9.45 ± 7.50	4.59 ± 4.46	.000
Coronal L4 tilt (degrees)	14.18 ± 9.35	6.00 ± 5.41	.000
GCM (mm)	21.50 ± 18.26	14.19 ± 13.25	.006
Sagittal parameters
Thoracic kyphosis (degrees)	13.00 ± 10.00	21.66 ± 8.49	.000
Thoracolumbar kyphosis (degrees)	14.37 ± 8.07	7.91 ± 5.65	.000
Lumbar lordosis (degrees)	17.32 ± 14.24	36.87 ± 5.95	.000
Sacral slope (degrees)	18.81 ± 10.03	28.55 ± 9.12	.000
Pelvic tilt (degrees)	27.75 ± 10.98	18.55 ± 8.50	.000
Pelvic incidence (degrees)	47.37 ± 11.30	47.47 ± 11.15	.261
Sagittal vertical axis (mm)	74.50 ± 49.16	30.42 ± 16.82	.000

Abbreviation: GCM, global coronal malalignment.

a Results are given as the number or the mean ± SD unless otherwise stated. Boldface indicates statistical significance.

Comparison Between Imbalanced and Balanced Groups

As shown in Tables 1, 3, and 4, the comparative analyses showed no significant intergroup differences regarding sex, age at surgery, osteotomy levels, osteotomy grades, instrumented levels, levels of interbody fusion, and distribution of UIV. Compared with the balanced group, the imbalanced group had a larger proportion of preoperative coronal consistency pattern (79.17% vs 40.43%, P = .002), larger proportion of type C (29.17% vs 8.51%, P = .023), lower proportion of coronal opposition pattern (20.83% vs 59.57%, P = .002), greater postoperative GCM (24.60 ± 17.58 mm vs 8.87 ± 5.20 mm, P < .001), and lower amount of GCM correction (−4.92 ± 24.25 mm vs 14.52 ± 19.49 mm, P < .001). There were no significant intergroup differences regarding preoperative GCM, preoperative and postoperative major Cobb angle, preoperative and postoperative L4 or L5 coronal tilt, the amount of major Cobb correction, and the amounts of correction of coronal tilt of L4 and L5. Table 5 shows that no significant differences regarding sagittal parameters were seen between imbalanced and balanced groups.

Table 3.

Comparison of Preoperative Coronal Parameters Between the 2 Groups.a

	Total	Imbalanced Group	Balanced Group	P Value
No. of patients	71	24	47
Consistency pattern (%)	53.52% (38/71)	79.17% (19/24)	40.43% (19/47)	.002
Opposition pattern (%)	46.48% (33/71)	20.83% (5/24)	59.57% (28/47)	.002
Type C (%)	15.49% (11/71)	29.17% (7/24)	8.51% (4/47)	.023
GCM (mm)	21.50 ± 18.26	23.81 ± 18.18	20.32 ± 18.38	.450
Major Cobb angle (degrees)	26.50 ± 15.73	26.77 ± 13.28	26.37 ± 16.97	.920
Coronal L5 tilt (degrees)	9.45 ± 7.50	9.90 ± 7.46	9.23 ± 7.59	.725
Coronal L4 tilt (degrees)	14.18 ± 9.35	16.72 ± 7.86	12.89 ± 10.40	.103

Abbreviation: GCM, global coronal malalignment.

a Results are given as the number or the mean ± SD unless otherwise stated. Boldface indicates statistical significance.

Table 4.

Postoperative Coronal Parameters and Their Changes Between the 2 Groups.a

	Total	Imbalanced Group	Balanced Group	P Value
No. of patients	71	24	47
Postoperative
GCM (mm)	14.19 ± 13.25	24.60 ± 17.58	8.87 ± 5.20	.000
Major Cobb angle (degrees)	8.79 ± 6.75	8.39 ± 6.68	9.00 ± 6.85	.721
Coronal L5 tilt (degrees)	4.59 ± 4.46	5.75 ± 5.98	4.00 ± 3.36	.119
Coronal L4 tilt (degrees)	6.00 ± 5.41	7.08 ± 6.09	5.44 ± 5.00	.231
Amount of correction
GCM correction (mm)	7.95 ± 22.99	−4.92 ± 24.25	14.52 ± 19.49	.000
Major Cobb correction (degrees)	17.72 ± 11.77	18.38 ± 9.48	17.38 ± 12.87	.737
Coronal L5 tilt correction (degrees)	4.90 ± 5.76	4.27 ± 4.52	5.23 ± 6.32	.511
Coronal L4 tilt correction (degrees)	8.20 ± 6.61	9.64 ± 5.64	7.47 ± 6.99	.193

Abbreviation: GCM, global coronal malalignment.

a Results are given as the number or the mean ± SD unless otherwise stated. Boldface indicates statistical significance.

Table 5.

Preoperative and Postoperative Sagittal Parameters Between the 2 Groups.a

	Total	Imbalanced Group	Balanced Group	P Value
No. of patients	71	24	47
Thoracic kyphosis (degrees)
Preoperative	13.00 ± 10.00	11.60 ± 8.52	13.69 ± 10.69	.485
Postoperative	21.66 ± 8.49	21.97 ± 8.36	21.51 ± 8.67	.858
Thoracolumbar kyphosis (degrees)
Preoperative	14.37 ± 8.07	15.08 ± 8.63	14.02 ± 7.89	.661
Postoperative	7.91 ± 5.65	7.06 ± 6.04	8.32 ± 5.49	.455
Lumbar lordosis (degrees)
Preoperative	17.32 ± 14.24	15.45 ± 16.87	18.23 ± 12.95	.515
Postoperative	36.87 ± 5.95	36.64 ± 5.98	36.98 ± 6.02	.847
Sacral slope (degrees)
Preoperative	18.81 ± 10.03	19.39 ± 11.98	18.52 ± 9.12	.772
Postoperative	28.55 ± 9.12	29.72 ± 9.23	27.98 ± 9.14	.522
Pelvic tilt (degrees)
Preoperative	27.75 ± 10.98	27.23 ± 10.51	28.01 ± 11.34	.814
Postoperative	18.55 ± 8.50	17.07 ± 8.63	19.27 ± 8.46	.386
Pelvic incidence (degrees)
Preoperative	47.37 ± 11.30	46.56 ± 10.58	47.77 ± 11.77	.723
Postoperative	47.47 ± 11.15	46.67 ± 10.27	47.86 ± 11.68	.722
Sagittal vertical axis (mm)
Preoperative	74.50 ± 49.16	68.00 ± 48.13	77.65 ± 50.03	.512
Postoperative	30.42 ± 16.82	28.12 ± 14.41	31.54 ± 17.97	.496

a Results are given as the number or the mean ± SD unless otherwise stated.

Discussion

Compared with patients with postoperative coronal balance, our results showed that patients with postoperative coronal imbalance had higher proportion of preoperative coronal consistency pattern, higher proportion of type C, lower proportion of coronal opposition pattern, and lower amount of GCM correction. Preoperative and postoperative major Cobb angle, coronal tilt of L4 or L5, and the amounts of their corrections were not associated with postoperative coronal imbalance.

Preoperative coronal types have been reported to have impacts on the incidence of postoperative coronal imbalance in DLS patients. Bao et al[9] studied 69 DLS patients who underwent posterior-only surgical correction and found that patients with type C were predisposed to having postoperative coronal imbalance when Smith-Peterson osteotomy (SPO) or pedicle subtraction osteotomy (PSO) osteotomies were performed. Consistent with their results, the current study also demonstrated that coronally imbalanced patients had a larger proportion of type C than coronally balanced patients (P = .023). In this case series, all patients received spinal osteotomy (Schwab grade 1 or grade 2). When performing an osteotomy to correct the deformity in the coronal plane in a patient with type C, routine maneuvers such as compression on the convex side and distraction on the concave side of the major curve may worsen the malalignment toward the convex side. To date, how to effectively correct type C coronal imbalance is still unclear. Recently, Bao et al[12] reported a multirod-based sequential correction technique to treat 7 patients with preoperative coronal imbalance type 2. After surgery, 6 patients achieved good coronal balance, and only 1 patient was left with coronal imbalance. However, the sample size was very small; large samples are needed to validate this technique.

The lumbosacral fractional curve consists of L4, L5, and S1 segments. Great attention has been paid to the correction of the fractional curve in spinal deformity correction surgery. It is even believed that correction of the fractional curve may play a critical role in obtaining coronal balance.[13] However, there is paucity of information about the relationship between orientation of L4 coronal tilt and C7 PL and their impact on the postoperative coronal balance. The current study first demonstrated that patients with coronal consistency pattern carried greater risk for postoperative coronal imbalance (Figure 2), whereas patients with coronal opposition pattern were more likely to achieve postoperative coronal balance (Figure 3).

Figure 2.

Radiographs showed that global coronal malalignment in a patient with coronal consistency pattern became worse after surgery.

Figure 3.

Radiographs showed that a patient with coronal opposition pattern obtained satisfactory correction of coronal imbalance after surgery.

In most DLS cases, a patient with type C can be classified into coronal consistency pattern and type B into opposition pattern. In our clinical practice, we found a big defect with the classification proposed by Bao et al.[9] According to the definition of this classification, the relationship between the major curve and C7 PL is a major foundation for the classification by Bao et al: that is, when GCM >30 mm, it is classified as type B if C7 PL is on the concave side of the major curve; if C7 PL is on the convex side of the major curve, it will be classified as type C. However, the big problem with the classification by Bao et al is how to decide which curve is the major curve because more than 1 curve (lumbosacral fractional curve, lumbar curve, or even thoracolumbar curve) exists in a DLS patient. It would become more confusing to decide on which curve is the major curve if the curves have the same Cobb angle. Setting a different curve as the major curve can lead to a different type. It is not rare for a spine surgeon to classify a DLS patient as type C one time but as type B the next time. This might cause serious consequences because different types of DLS patients may need different treatment strategies during surgical decision-making. On the other hand, the current classification we used was based on the relationship between orientation of L4 coronal tilt and C7 PL. Unlike in the Bao et al classification where the major curve might be difficult to determine, the L4 vertebra used in the current classification is clearly there, and orientation of L4 coronal tilt relative to C7 PL has only 1 direction in a DLS patient: either toward C7 PL (consistency pattern) or opposite C7 PL (opposition pattern). Therefore, classifying a DLS patient as having a different coronal pattern at different times might never happen while using the current classification.

Lewis et al[10] believed that leveling coronally L4 and L5 vertebrae is the key to achieve coronal balance in ASD patients fused to the pelvis. They thought that providing a more leveled base in the distal lumbar spine would provide a more horizontal platform from which to correct the major curve, so coronal imbalance can be avoided. In their reports, postoperative L4 and L5 tilts were significantly lower in the balanced group than that in the imbalanced group. In contrast to their reports, the current results showed no difference regarding preoperative and postoperative L4 and L5 coronal tilts and the amounts of their corrections between the 2 groups. The possible explanation for the difference between these 2 results might be as follows: first, it could be a result of the sample bias. Only when the lumbosacral fractional curve (L4-S1) is the primary driver of the coronal imbalance could leveling coronally L4 and L5 vertebrae be a critical factor for achieving coronal balance. Second, coronal L4 and L5 tilts are regional parameters; there might be other factors that contribute to the global coronal balance in addition to L4/L5 coronal tilt. This current study also demonstrated no significant difference regarding preoperative and postoperative major Cobb angle and the amount of correction of major Cobb angle between the imbalanced and balanced groups. Similar results were reported by Bao et al[9] and Lewis et al.[10] Furthermore, our results showed that there were significant differences regarding GCM correction between imbalanced and balanced groups. As a global coronal parameter, the amount of GCM correction directly influenced the surgical outcomes. Thus, it is worth paying great attention to this during operation correction of the global parameter.

There are several limitations in this study. First, this small sample of patients might have inherent selection bias. Second, we only analyzed the risk factors for immediate postoperative coronal imbalance. A long-term follow-up study is required to determine the maintenance of spinal coronal balance. Third, DLS patients often present with deformity in both the sagittal and coronal planes. Functional scores such as SRS-22 or Oswestry disability index (ODI) were not analyzed in this study because sagittal imbalance might be an important confounding factor. Despite these limitations, the current study identified important risk factors for immediate postoperative coronal imbalance in ASD patients fused to the pelvis. This was the first study that showed that patients with coronal consistency pattern carried greater risk for postoperative coronal imbalance.

Conclusions

In DLS patients fused to the pelvis, those with coronal consistency pattern, with type C, were at greater risk for immediate postoperative coronal imbalance.