Multiple Hemivertebrae: The Natural History and Treatment of 50 Patients

Abstract

Objective

Multiple hemivertebrae (MHV) is defined as three or more hemivertebrae, and is relatively uncommon among patients with congenital scoliosis. This study aimed to compare the natural history of different kinds of MHV and describe the surgical outcome of MHV.

Methods

In this retrospective cohort study, a total of 50 patients diagnosed with MHV were enrolled from June 2007 to June 2018. The medical records and radiographs of these patients were reviewed to summarize the characteristics of MHV. Patients with MHV were divided into the unbalanced (UB) group, partially unbalanced (PUB) group, and completely balanced (CB) group. Medical records and radiographs of MHV patients were reviewed to collect HV position, natural history, coronal and sagittal parameters. A Mann–Whitney U test was used to compare the radiographical data, such as the cobb angle of main curve and secondary curve, and Fisher's exact test was used to compare the patients in different kinds of MHV with surgical indication or not.

Results

The average number of hemivertebrae was 3.6 and the average main curve was 57.5°. Twenty‐four of 50 patients had associated anomalies, including four patients with sacral agenesis, one with tetralogy of Fallot, two with congenital imperforate anus, and 17 with Klippel‐Feil syndrome. In 22 patients who underwent MRI imaging, three patients had mild syringomyelia and three patients had diastematomyelia. The UB and PUB groups had a larger main curve and compensatory curve than the CB group. Of the 25 patients with follow‐up before surgery, the curve progression rate was highest in the UB group (12.1°/year) but similar in the PUB group (4.2°/year) and CB group (3.6°/year). All patients in the UB and PUB group met the criteria for surgery. In contrast, only 10 of 23 patients in the CB group had surgical indications. Eighteen of the 37 patients with surgical indications chose to undergo surgery and the correction rate of the main curve was 51.4%.

Conclusions

Early surgical intervention should be considered for most patients with UB or PUB MHV. For patients with CB MHV, surgical treatment may not be urgently needed at the first visit. Posterior hemivertebrectomy could be used for the treatment of MHV with satisfying radiographic outcome.

Multiple hemivertebrae (MHV) is defined as three or more hemivertebrae. Patients with MHV were divided into the unbalanced (UB) group (A), partially unbalanced (PUB) group (B, C), and completely balanced (CB) group (D). Patients with unbalanced and partially unbalanced MHV had a larger main curve than those with completely balanced MHV at the initial visit and final follow‐up before surgery. All patients in the UB and PUB group met the criteria for surgery. In contrast, only 10 of 23 patients in the CB group had surgical indications. Early surgical intervention should be considered for patients with UB or PUB MHV. For patients with CB MHV, surgical treatment may not be urgently needed at the first visit.

Introduction

Hemivertebrae (HV) are a common cause of congenital scoliosis that result from failure of formation. ¹ , ² , ³ , ⁴ , ⁵ , ⁶ , ⁷ The rate of deterioration and the severity of the spine deformities caused by HV mainly depend on the type, site, number, and their relationship to each other. ³ , ⁴ The natural history and treatment plan for patients with two or more hemivertebrae differ from those of patients with a single hemivertebra. For patients with double hemivertebrae, many authors report that double opposing hemivertebrae have lower overall progression and intervention rates than double unilateral HV. ³ , ⁵ , ⁸ Most patients with double unilateral hemivertebrae have severe deformity at the initial visit and require surgical treatment. ⁸ , ¹⁴ However, Shawen et al. reported that only 33% of patients with hemimetameric shift (HMMS) finally received surgical treatment, while the remaining patients received bracing or observation. ⁵ Based on the natural history and radiographic findings Lyu et al. proposed a clear operative indication for HMMS. ⁹

Multiple hemivertebrae (MHV; defined as three or more hemivertebrae) have been described in previous reports, but only as part of large series that have focused on congenital scoliosis. McMaster and David reported that only six of 104 patients (5.8%) with HV had MHV, whose resulting curves tended to balance each other and cause little deformity. ³ Another study of HMMS reported that nine of 354 patients had MHV based on three‐dimensional computed tomography. ¹⁰ However, most reported unbalanced MHV cause severe deformity and require surgical intervention. ³ , ⁸ , ¹⁴ Nasca et al. ¹¹ reported the largest series of patients with MHV and described the progression and treatment of 37 patients, many patients with MHV presented with a unilateral bar and contralateral HV, which produce severe deformity and rapid curve progression. ¹¹ Shawen et al. ⁵ found 14 patients with MHV among 186 with congenital scoliosis (7.5%), reported severe spine deformity caused by MHV with HMMS, especially in the thoracolumbar to lumbosacral junction.

These studies showed that the site of and relationship between HV play crucial roles in curve progression and some patients with multiple hemivertebrae could get severe spine deformity. ⁵ , ⁸ , ¹⁰ , ¹¹ However, limited information was provided about the natural history of MHV and all of these studies had small sample sizes of MHV. The progression rate of the scoliosis in balanced or unbalanced MHV was still unclear. Meanwhile, Shawen et al. ⁵ and Zhou et al. ⁸ reported the radiographic outcome after surgical treatment of double hemivertebrae and MHV together, with only a few cases of MHV. There were no published case series of MHV to evaluate the radiographical outcome of posterior hemivertebra resection. Therefore, the aims of our study were: (i) to describe the common presentation of multiple hemivertebrae which including age, sex, associated abnormalities and site of HV; (ii) to compare radiographic findings and natural history of unbalanced, partial unbalanced and balanced MHV; and (iii) to show surgical outcome of MHV after posterior hemivertebrectomy and fusion.

Methods

This retrospective study was approved by the Ethics Committee of the local hospital (No. 2019‐852). All patients in this case series were diagnosed with congenital scoliosis and had radiographic evidence of HV at one center between 2008 to 2018. Of 684 patients with hemivertebrae, 50 had MHV.

Inclusion and Exclusion Criteria

The inclusion criteria were: (i) patients with multiple hemivertebrae (MHV) (defined as three or more hemivertebrae) between June 2007 and June 2018; and (ii) age between 2 and 24 years. The exclusion criteria were as follows: (i) history of other spinal surgery, and (ii) incomplete medical records.

Groups

The patients were divided into the unbalanced (UB), partially balanced (PUB), and completely balanced (CB) groups. The UB group had all HV located on one side of the spinal column (Fig. 1A). The PUB group had continuous HV on the same side of the spine, with one or more HV on the contralateral side (Fig. 1B,C). The CB group had all HV located on both sides of the spinal column, with no continuous HV located on the same side of the spine (Fig. 1D).

Fig. 1.

(A) Unbalanced multiple hemivertebrae (MHV) with C5, C7 and T2 hemivertebrae all on the right side (patient No. 7). (B) Partial unbalanced MHV patient with T3, T11, L2, L3 hemivertebrae on the left side and T9, L6 hemivertebrae on the right side. L2 and L3 hemivertebrae located closely on the left side of spinal column (patient No. 9). (C) Partial unbalanced MHV with T1, T3 hemivertebrae on the right side and T8, T10 hemivertebrae on the left side (patient No. 22). (D) Completely balanced MHV with T4, T10 hemivertebrae on the right side and T6, T12 hemivertebrae on the left side. The thoracic curve showed a slight progression of 7° after 4 years follow‐up (patient No. 40).

Medical records were reviewed to collect demographic data and complications in the perioperative and follow‐up periods. All patients underwent routine examinations. Standing full‐length posteroanterior and lateral radiographs were routinely obtained before surgery, at 3 months postoperatively, and at least 2‐year‐follow‐up time.

Surgical Methods

In our department, patients with surgical indications for MHV met at least one of the following criteria: (i) for patients with segmental scoliosis of >50° or kyphosis 30° greater than the normal segmental sagittal angle of the corresponding level, the severe deformity leads to a corresponding change in appearance: and (ii) for patients with progressive deformity, MHV causes segmental scoliosis of >30° with evidence of curve progression of >5° in a 12‐month period preoperatively or kyphosis of >20° with evidence of progression of >5° in a 12‐month period preoperatively.

During surgery, a posterior standard median incision was made in the prone position. The posterior elements were explored and the pedicle screws were inserted into the upper and lower vertebrae, which were checked with fluoroscopy. For the patients with large main curve and structural compensatory curves, more pedicle screws were implanted into the vertebrae at the fusion level. A safety rod was applied on the concave side. Resection of the posterior elements of HV and the posterior part of the pedicle were performed. A wedge osteotomy was performed to completely remove the vertebral body and the upper and lower discs, which included the cartilage endplate. Subsequently, another rod was applied on the convex side. Gradual compression was applied while leaving the concave rod unlocked until the gap slowly closed. Anterior reconstruction with cage was performed if large hemivertebra obstructed the gap closure. The residual autogenous bone was used for posterolateral fusions after decortication.

Outcome Measures

The radiographs were analyzed to determine radiographic parameters in the coronal and sagittal plane. In the coronal plane, the Cobb angles of the main curve and secondary compensatory curve were measured. The distance between the C7 plumb line and center sacral vertical line was used to assess the coronal balance. The segmental lordosis, pelvic incidence (PI), sacral slope (SS), pelvic tilt (PT), thoracic kyphosis (TK,T1‐T12), thoracolumbar kyphosis (TLK), lumbar lordosis (LL), and sagittal balance were measured in the sagittal plane. The sagittal balance was determined as the sagittal vertebral axis (SVA). Three‐dimensional computed tomography was used to evaluate the anatomy of the HV. Magnetic resonance imaging of the whole spine was performed to assess intraspinal anomalies before surgery.

Statistical Analysis

Mann–Whitney U‐tests were performed to compare the coronal balance, SVA, PI, PT, SS, LL, TK, TLK, and Cobb angles of the main curve and secondary compensatory curve. Postoperative spine parameters were also analyzed in the patients who underwent surgery, and the preoperative data were compared with the final follow‐up data. We used a Fisher's exact test for patients with surgical indication or not. All data were analyzed using SPSS 21.0 statistical software (SPSS Inc., Chicago, IL, USA). Significance was defined as P < 0.05.

Results

Radiographical and Clinical Outcome

There were 20 male and 30 female patients with MHV in 684 patients with hemivertebrae (50/684, 7.3%). These 50 patients had a total of 179 HV. The mean age was 10.5 years (range 2–23 years). The average number of HV per patient was 3.6 (range 3–6), with the following distribution: C1–C7 (n = 5), T1–T4 (n = 28), T5–T11 (n = 82), T12–L1 (n = 21), L2–L4 (n = 31), and L5 to sacrum (n = 12). Forty‐eight patients had other congenital spine deformities, including butterfly vertebrae (39 cases), block vertebrae (14 cases), a unilateral or anterolateral unsegmented bar (17 cases) and rib synostosis (14 cases) (Table 1).

TABLE 1.

Medical and radiographic records of 52 patients with multiple hemivertebrae (MHV)

Patient no	Age/Sex	HV position	Require surgery	Get surgery	Associated abnormalities	Indications for surgery	Type of balance
1	9F	T5,9,11R(FS)	+	+	Left:6–7,9‐10RS. T6,10BV.T8‐12 unilateral bar(L).	1, 4	UB
2	13F	T3,7,9R(SS);L6R(FS)	+	+	T11‐12 block vertebrae	1, 5	UB
3	11F	T8,9,10R(SS)	+	+	Left: 5–6,7‐10,11–12RS. T6.T7.T13BV.T6‐11 unilateral bar(L).	1, 4	UB
4	14F	T3R(FS),T9,12R(SS)	+	+	Left: T1‐2RS.T6,T8,L2BV.	1, 6	UB
5	6M	T7,T8,T9L(FS).	+	+	—	1, 4, 6	UB
6	10F	T6,10,L1L (FS),S1L (FS)	+	−	Right:1–2RS. T3,7,9,L5BV. Sacral agenesis.	1, 2, 4, 6	UB
7	9F	C5R(FS),C7R(SS),T2R(SS)	+	−	C3‐4KF, T1,T3,T4BV	3.6	UB
8	12F	T2R(FS),T8R(SS),T10R(SS)	+	−	T3.T4.T6.T7.T9BV. L:2–5,6–8RS. T6‐9 unilateral bar(L)	1	UB
9	6M	T3L,9R,11L(FS),L2,3L(FS),L6R(FS)	+	+	T7BV. Left: 11–12RS.Sacral agenesis.	2	PUB
10	2F	T4,7L(SS),11R(SS)	+	+	T4BV	1	PUB
11	11M	L3R(SS),L5,L6L(FS)	+	+	L4‐S1 unilateral bar(R)	2, 5, 6	PUB
12	5F	T3,5,10L(SS),L1R(FS)	+	+	Right: 1–2RS	1,6	PUB
13	14M	T5L(SS),T9L(SS),T11L(SS),T14R(SS),L2L(FS)	+	+	C2‐4KF. Right:1–2,5‐6,8‐9RS.T6‐8BV. T2‐3,4‐8,12‐13block vertebrae. C5‐6 bar(L)	1	PUB
14	5F	L1R(SS),L6L(SS),S2L(FS), S4R(SS)	+	+	L4‐5 block vertebrae. Sacral agenesis	5	PUB
15	12F	T9,10,12R(SS),T14L(FS)	+	+	C3‐5KF.Left: T9‐12RS.T4‐6,T9‐11block vertebrae. T11‐13 unilateral bar(L).Congenital heart disease.	1, 2, 4	PUB
16	9F	C4L(SS),T1R(SS),T3R(SS)	+	+	C3‐4KF. Left: T1‐5RS.C3,C5‐7,T2,T4‐6BV. T1‐6 block vertebrae. C5‐6 bar(L)	3	PUB
17	11M	T6R(SS),L3R(FS),L6L(SS)	+	−	Right:7–8RS. Left:5–7RS. T7BV.T7‐8bar(L). Tetralogy of Fallot	2,5	PUB
18	19M	T1,T5L(SS),T11R(SS),L5L(SS)	+	−	C2‐3KF. T4,8,9BV	1, 5	PUB
19	22M	L2R(SS),L4L(SS),L6L(FS)	+	−	L5BV	1, 2, 4	PUB
20	21F	T9R(FS),L3L(SS),S1L(SS)	+	−	C3‐5KF. T3‐8 block vertebrae. Left:1–2,3‐10RS, Right:3–7RS.	1, 2, 5	PUB
21	23M	T10L(SS),L2,3R(FS)	+	−	T5‐6 block vertebrae. T3BV. L1‐4 unilateral bar(L).	2, 4	PUB
22	11F	T1,3L(SS),T8,10 R(FS)	+	−	T4BV	1, 3	PUB
23	16F	C6R(SS),T1R(SS),T6L(FS)	+	−	C2‐3KF.Right:1–3RS. T2‐4BV. C7‐T4 block vertebrae.	1.3	PUB
24	3F	C7L(SS),T4R(FS),T6R(SS)	+	−	T1 BV. T3‐5 unilateral bar(L). Congenital imperforate anus.	3.6	PUB
25	15F	L1,L3R(SS),L5L(SS)	+	−	T12‐L4 anterolateral bar. T11BV	2.4	PUB
26	4M	T1,T4,T6,T8L(SS),T11R(SS)	+	−	Right:1–4RS. T3,5,7,12BV.T3‐7 unilateral bar(R)	3	PUB
27	18M	T6L(SS,)T9L(FS),T13R(SS),L3L(SS)	+	−	C3‐4KF. Right:5–6RS. T5.T6BV. T8‐10 unilateral bar(R)	1, 6	PUB
28	18F	T3R(SS),T5L(SS),T8R(SS),T11L(SS),T13R(SS)	+	+	C4‐5KF. L2‐3 block vertebrae.T7BV.	4	CB
29	5F	T5R(SS),T10L(SS),L2R(FS)	+	+	C2‐3KF.Left:T1‐2RS.T2,3BV.T1‐3 bar(L).	1, 2, 6	CB
30	10F	T9R(FS),T11L(FS),L3R(SS)	+	+	T10BV	2,6	CB
31	12M	T6R(SS),T13L(SS),L2R(SS)	+	+	C2‐3KF. L1‐3 anterolateral bar. Microtia.	4	CB
32	9M	T4L(FS),T7R(FS),T9L(FS),T12R(SS),L2L(SS),S1R(SS)	+	+	C4‐5,C7‐T1KF. T2‐3block vertebrae. Congenital heart disease.	2, 5, 6	CB
33	18M	T5L(SS),T13R(SS),L2L(SS)	+	−	T6‐7 block BV	4	CB
34	8F	T3L(SS),T7R(SS),T11L(SS),L3R(SS)	+	−	L2‐4 anterolateral bar, T1.8.9.10BV	4	CB
35	12M	T4L(FS),T11R(SS),T13L(SS),T15R(SS),L2L(SS)	+	−	R:5–6RS,T14.L3BV	4	CB
36	16M	T3L(SS).T9R(SS),T11L(FS)	+	−	T1‐5block vertebrae,T1.2BV.L:1–2RS，R:1–3RS	3	CB
37	3F	T10R(SS),L1L(FS),L3R(SS)	+	−	T12.13BV. Congenital imperforate anus	2	CB
38	2F	T2L(FS),T11R(FS),L3L(SS)	−	−	T1.T3.T4BV	—	CB
39	16F	T9R(SS),T14L(SS),L4R(SS)	−	−	—	—	CB
40	11M	T4R,T6L,T10R(FS),T12L(SS)	−	−	C2‐3KF.T3,T5,T8,T9,T13BV	—	CB
41	5M	T7L(SS),T10R(FS),L4L(SS)	−	−	T3‐7 block vertebrae,T4BV	—	CB
42	15F	T10L,T12R(SS),T14L(FS),L2R(SS)	−	−	C2‐3KF,T8,9BV. T8‐11,L1‐3 block vertebrae	—	CB
43	4F	T3L,T6R(FS),T8L,T10R(SS)	−	−	C2‐3KF,T2,4,7,9BV	—	CB
44	5F	T5R(SS),T12L(SS),L2R(SS)	−	−	T6‐8,10,L4, S1BV.T4‐7 block vertebrae. Sacral agenesis	—	CB
45	3F	T5R(SS),T9L(FS),L2R(SS)	−	−	T4,5 blockBV.T8BV	—	CB
46	10F	T4L(SS),T6R(SS),T10L(SS),T13R(SS),T15L(SS)	−	−	Right: 6–8RS.Left: 8‐9RS.T7‐8 block vertebrae	—	CB
47	2M	T6L,T8R,T12L(FS),T14R(FS)	−	−	T9,10BV	—	CB
48	8F	T4R(SS),T7L(SS),T10R(FS)	−	−	C2‐3,C5‐7KF,C1‐2 rotatory dislocation. T1.2.5BV	—	CB
49	17M	T7L(SS),T9R(SS),T11L(SS)	−	−	C2‐3,C4‐5KF. T1BV	—	CB
50	9F	T1L(SS),T11R(SS),L1L(SS)	−	−	C7.T10BV. T12‐L2 anterolateral bar	—	CB

Abbreviations: BV, butterfly‐vertebra; CB, Completely balanced; F, female; FS, fully segmented; HV, hemi‐vertebra; KF, Kilppel‐Feil syndrome; L, lumbar; M, male; PUB, Partial unbalanced; RS, rib synostosis; SS, semi‐segmented; T, thoracic; UB, Unbalanced.

Note: Surgical indication: (1) Thoracic curve > 50°. (2) Lumbar curve > 50°. (3) Proximal thoracic curve or cervicothoracic curve > 50°. (4) Kyphosis > 30° than normal segmental sagittal angle of the corresponding level. (5) Lumbosacral hemivertebra. (6) Progressive scoliosis or kyphosis.

There were eight patients in the UB group, 19 patients in the PUB group and 23 patients in the CB group (Table 2). To evaluate deformity of coronal and sagittal plane, the paragraphic data at the last follow‐up time before surgery or without surgery was collected. The main curve was significantly smaller in the CB group than the UB group (41.1° ± 13.6° vs. 83.4° ± 17.1°, P < 0.01) and PUB group (41.1° ± 13.6° vs. 68.5° ± 14.0°, P < 0.01). The CB group also had a significantly smaller secondary compensatory curve than the UB group and PUB group. Also, the CB group had greater TLK than the UB group (17.0° ± 20.1° vs. −2.7° ± 15.5°, P = 0.02), and the CB group had similar TLK with the PUB group. Most patients in the UB group did not have severe TLK, but two had severe thoracic kyphoscoliosis.

TABLE 2.

Patient characteristics at the final follow‐up time before surgery or without surgery

Characteristics	Unbalance Group (n = 8)	Partial unbalanced group (n = 19)	Completely balanced group (n = 23)
Age (years)	10.5 ± 2.6	11.4 ± 7.0	10.1 ± 6.3
Male to female	1:7	9:10	9:14
Hemivertebrae number	3.3 ± 0.5	3.7 ± 0.9	3.6 ± 0.9
Main curve (°)	83.4 ± 17.1	67.7 ± 14.2 a	40.5 ± 13.7 a , b
Secondary curve (°)	54.9 ± 12.9	44.9 ± 14.3	33.0 ± 9.8 a , b
Coronal balance (mm)	13.8 ± 16.9	6.0 ± 18.4	2.0 ± 12.3
PI (°)	40.0 ± 4.4	41.0 ± 7.3	38.2 ± 6.2
PT (°)	5.1 ± 10.0	7.6 ± 8.0	4.9 ± 4.8
SS (°)	34.9 ± 6.2	33.4 ± 8.3	33.3 ± 7.2
TK (°)	45.0 ± 15.1	31.7 ± 22.2	38.9 ± 14.3
LL (°)	56.3 ± 9.8	44.8 ± 19.5	51.2 ± 15.7
TLK (°)	−2.7 ± 15.5	9.5 ± 21.7	16.6 ± 19.8 a
SVA (mm)	0.3 ± 28.8	11.3 ± 24.9	−0.6 ± 23.6
Require surgery	8/8	19/19	10/23 a , b

Abbreviations: LL, lumbar lordosis; PI, pelvic incidence; PT, pelvic tilt; SS, sacral slope; SVA, sagittal vertebral axis; TK, T1‐12 thoracic kyphosis; TLK, thoracolumbar kyphosis.

^a Significantly different from the unbalance group, P < 0.05,

^b Significantly different from the partial unbalanced group, P < 0.05.

Twenty‐five of 50 patients had associated anomalies. Two of eight patients in the UB group had associated anomalies, including one patient with Sacral agenesis and one patient with Klippel‐Feil syndrome. Meanwhile, 11 of 19 patients in the PUB group had associated anomalies, including one patient with Congenital imperforate anus, one patient with Tetralogy of Fallot, one patient with Congenital heart disease and Klippel‐Feil syndrome, two patients with Sacral agenesis, and six patients with Klippel‐Feil syndrome alone. On the other hand, 12 of 23 patients in CB group had associated anomalies, including one patient with congenital imperforate anus, seven patients with Klippel‐Feil syndrome alone, one patient with Sacral agenesis, one patient with Congenital heart disease and Klippel‐Feil syndrome, one patient with Microtia and Klippel‐Feil syndrome, and another one patient with C2–C3 Klippel‐Feil syndrome also had atlantoaxial rotatory dislocation.

Twenty‐two of 50 patients underwent magnetic resonance imaging to evaluate intraspinal abnormalities. Six patients (27.3%) had an intraspinal abnormality. Three patients had mild syringomyelia at the cervical or thoracic spine, and three patients had diastematomyelia. No patients had spinal compression or other abnormal neurologic findings.

Natural History of Radiographic Findings

After carefully reviewing the radiographs of all 50 patients, natural history data were available for only 25 patients. Although the mean follow‐up time was only 42.1 months (range 12–94 months), such data still provide some valuable information about the natural history of MHV (Fig. 2).

Fig. 2.

Curve progression of the patients with multiple hemivertebrae (MHV). Patients with unbalanced and partially unbalanced MHV had a larger main curve than those with completely balanced MHV at the initial visit and final follow‐up before surgery. The progression rate was greater in the unbalanced group (12.1°/year) than the PUB group (4.2°/year) and CB group (3.6°/year).

Natural history data were available for four patients in the UB group. The mean progression rate was 12.1°/year, which was higher than that of the PUB and CB groups. Two patients with a large C‐shaped thoracic curve had severe and rapid progression (case no. 4, 12.4°/year and case no. 5, 29.3°/year). The other two patients who had unbalanced thoracocervical semi‐segmented MHV (case no. 7, 3.0°/year) or an S‐shaped curve with lumbosacral HV (case no. 6, 3.8°/year) did not show such rapid curve progression, but had a significant change in appearance with a shoulder imbalance or coronal imbalance.

Eight PUB patients had mean progression rate of 4.2°/year. Two patients had a rapid progression rate of >5°/year. Case no. 11 was a 6‐year‐old boy with a unilateral bar with contralateral L5 and L6 fully segmented left HV, and an additional L3 right semi‐segmented HV; his lumbar scoliosis increased from 50° to 72° in 51 months (5.2°/year). Case no. 12 was a 2‐year‐old girl with a main thoracic curve of 50° resulting from T3, T5, and T10 left semi‐segmented HV and L2 right fully segmented HV; her thoracic curve increased to 76° in 39 months (8.0°/year). Another five patients in the PUB group had a medium progression rate of 3–5°/year. The remaining patient with right T1 and T3 and left C4 semi‐segmented HV had a low progression rate of 1.3°/year in 94 months.

The mean progression rate of 13 patients in the CB group was 3.6°/year. Compared with the PUB group, the CB group had a smaller main curve at the initial and final visits but a similar mean progression rate. This may be attributable to the two patients in the CB group who had a progression rate of >10°/year with only a 1‐year follow‐up duration. The first patient (case no. 30) was 9 years old at the initial visit and had a 41° lumbar curve with semi‐segmented L3 HV and fully segmented T11 HV; her scoliosis progressed to 53° at 10 years old. The second patient (case no. 29) was 4 years old at the initial visit and had a 55° thoracic scoliosis with L2 fully segmented HV and semi‐segmented T10 HV; repeat radiographs showed progression to 65° after 14 months. However, eight of 13 patients in the CB group had a progression rate of only 1–3°/year, and the remaining three patients had a progression rate of 3–3.5°/year.

The progression rate of kyphosis did not significantly differ between the CB group (mean 1.9°/year) and the PUB group (mean 1.5°/year), but was greater in the UB group (mean 7.0°/year). Most patients (21 of 25) had a progression rate of kyphosis of <3°/year, and 13 of 25 patients had a progression rate of kyphosis of <1.5°/year. However, two patients had a progression rate of kyphosis of >4°/year. Both of them had at least two fully segmented HV at the apex of kyphosis.

Surgical Treatment

Thirty‐seven patients had surgical indications. The percentage of patients with surgical indications was significantly lower in the CB group (10/23, 43.5%) than the PUB group (19/19, 100%) and the UB group (8/8, 100%). Five of the 23 patients in the CB group with surgical indications had coronal deformity, while another five had severe thoracolumbar or lumbar kyphosis. All 12 patients with MHV with fully segmented lumbar HV and all nine patients with MHV with a unilateral bar and contralateral HV had surgical indications.

Only 18 of the 37 patients with surgical indications finally chose to undergo surgery. The average age at surgery was 9.2 years, and the mean duration of follow‐up after surgery was 33.6 months. For the patients who underwent surgery, the mean Cobb angle of the segmental curve was corrected from 66.9° ± 22.2° preoperatively to 32.5° ± 14.7° at final follow‐up (P < 0.01), giving a correction rate of 51.4%. The mean Cobb angle of the secondary compensatory curve was corrected from 44.9° ± 16.9° preoperatively to 22.5° ± 14.9° at final follow‐up (P < 0.01), giving a correction rate of 50.0%. The mean TK (T1–T12) was changed from 43.4° ± 17.8° preoperatively to 34.9° ± 15.8° at final follow‐up (P = 0.03). The coronal balance, TLK, PI, PT, SS, LL, and SVA did not significantly differ from preoperatively to final follow‐up (Table 3).

TABLE 3.

Radiographic parameters of 18 patients underwent surgery with at least two‐year's follow‐up

Spinal parameter	Preoperative	Final follow‐up	Improvement	Z	P
Main curve (°)	66.9 ± 22.2	32.5 ± 14.7	34.4	−4.40	<0.01 *
secondary curve (°)	44.9 ± 16.9	22.5 ± 14.9	22.4	−3.81	<0.01 *
Coronal balance (mm)	4.8 ± 18.3	−1.4 ± 18.8	‐	−1.08	0.29
PI (°)	39.8 ± 6.9	39.7 ± 6.5	‐	−0.02	0.99
PT (°)	5.0 ± 8.5	5.7 ± 6.9	‐	−0.29	0.79
SS (°)	34.8 ± 7.3	34.0 ± 6.1	‐	−0.25	0.82
TK (°)	43.4 ± 17.8	34.9 ± 15.8	‐	−1.54	0.13
LL (°)	54.4 ± 15.3	49.8 ± 11.6	‐	−0.71	0.48
TLK (°)	7.8 ± 22.9	1.6 ± 8.9	‐	−0.59	0.56
SVA (mm)	3.7 ± 28.9	−2.3 ± 16.6	‐	−1.00	0.32

Abbreviations: LL, lumbar lordosis; PI, pelvic incidence; PT, pelvic tilt; SS, sacral slope; SVA, sagittal vertebral axis; TK, T1‐12 thoracic kyphosis; TLK, thoracolumbar kyphosis.

^* Statistical Significance between final follow‐up and preoperative time. (P < 0.05).

There were no serious or permanent neurologic complications, and no patient had a deep wound infection. Two patients underwent revision because of iliosacral screw failure and progressive proximal junctional kyphosis of upper unfused HV.

Case Examples of Patients Who Underwent Surgery

In the first case (Fig. 3), a 13‐year‐old girl presented with severe spinal deformity (patient no. 2). No muscle weakness or loss of reflexes on physical examination. Preoperative radiographs showed an UB spine with right T3, T7, and T9 semi‐segmented HV (82° thoracic curve) and a right L6 fully segmented hemivertebra (73° lumbar curve). No abnormalities were seen on magnetic resonance imaging of the entire spine. Posterior transpedicular resection of the L6 was performed, followed by pedicle screw instrumentation (intervertebral fusion with a PEEK cage). Six months after the initial surgery, the patient underwent resection of the T3 hemivertebra and T2–L2 fusion. At 2 years after the final surgery, the radiographic outcome was satisfactory.

Fig. 3.

(A–E) Preoperative radiograph and CT scan demonstrating the right T3,T7,T9 semi‐segmented hemivertebrae, right L6 fully segmented hemivertebra, and trunk shift to left 1.2 cm. Cobb measurements: T2–T12 82°(bending 75°), L1–L5 73°(bending 46°). (F–G) Postoperative view after resection of the L6 hemivertebrae. H‐I. The patient underwent resection of T3 hemivertebra 6 months later, and the patients showed satisfied radiographic outcome with correction rate of main thoracic curve to 19° at two‐year follow‐up.

The second case (Fig. 4). A 4‐year‐old male presented with progressive lumbar deformity (patient no. 32). Physical examination revealed no neurological abnormalities. Radiographic evaluation demonstrated left T4 and T9 fully segmented HV, a right T7 fully segmented hemivertebra, a left L2 semi‐segmented hemivertebra, and right T11 and S1 semi‐segmented HV (lumbar curve 55°). Five years later, the lumbar curve had increased to 70°. Posterior S1 HV resection and fusion were performed, followed by pedicle screw instrumentation. Postoperative radiographs showed that the deformities were substantially corrected (Fig. 4). The patient showed a good radiographic outcome at 2 years after surgery.

Fig. 4.

(A, B) Plain radiographs of patient no. 14, an 4‐year‐old boy with scoliosis due to hemivertebrae at T4, T9, L2 left side and T7, T12, S1 right side. (C, D) The patients final visit to the hospital was as a 9‐year‐old. The lumbar curve increased to 70°. F‐G, The S1 HV was excised and postoperative radiographs showed that the deformities were dramatically corrected to 46°. (H, I) The patient showed a satisfied radiographic outcome 2 years after the surgery.

Discussion

Common Presentation of Multiple Hemivertebrae and Associated Abnormalities

Multiple hemivertebrae were a relatively uncommon finding in congenital scoliosis with hemivertebra. The incidence of MHV was 7.3% in our study, which was consistent with McMaster and David³ and Shawen et al.'s findings. ⁵ In this retrospective review of 50 patients, the mean age was 10.5 years and the average number of HV per patient was 3.6. The patients with completely balanced MHV had lower main curve and lower risk of surgery than the patients with unbalanced and partial unbalanced MHV. In 18 patients who underwent surgery, posterior hemivertebrectomy showed satisfying radiographic outcome. Meanwhile, 25 of 50 patients had associated anomalies, including Klippel–Feil syndrome, sacral agenesis, congenital heart disease, congenital imperforate anus and Microtia. Six of 22 patients with MRI (27.3%) had an intraspinal abnormality, which is consistent with the accepted range of incidence of intraspinal abnormalities (19%–30%) in most congenital scoliosis series. ¹² , ¹³

Natural History and Radiographic Outcome of Three Kinds of MHV

In the present study, 25 of 50 patients with MHV had a mean 3.5 years of follow‐up after their initial hospital visit, which could partially showed their natural history. As expected, the UB group had the greatest mean progression rate (12.1°/year). However, the average progression rate was similar in the PUB group (4.2°/year) and the CB group (3.6°/year). Seven of eight patients in the PUB group had a progression rate of >3°/year, while eight of 13 patients in the CB group had a progression rate of only 1–3°/year. The mean progression rate in the CB group was markedly increased by two patients with a progression rate of >10°/year who both had fully segmented HV and caudal HV at the lumbar spine.

Most patients in the UB and PUB groups met the criteria for surgical treatment at their initial hospital visit. In contrast, only seven of the 24 patients in the CB group had surgical indications at their first visit and three more patients had progression and subsequently met the criteria for surgery. Apart from one patient with severe thoracocervical deformity, nine of these 10 patients in the CB group with surgical indications had their caudal HV located at the thoracolumbar to lumbosacral spine. These findings are consistent with McMaster's and Shawen's findings on patients with HMMS. ⁴ , ⁵ , ¹⁷ The significant progression of scoliosis or kyphosis in the thoracolumbar junction may result from a relatively unstable transitional zone of the vertebral column at the thoracolumbar junction. ⁵ Patients with thoracocervical HV often have a mild curve progression but have a significant change in appearance with an imbalanced shoulder. ¹⁵

The presence of two or more unilateral HV in the same region of the spine often causes severe scoliosis. ³ , ⁸ , ¹⁴ All patients with UB MHV in the present study presented with severe deformity. Most patients with UB MHV had one structural C‐shaped curve. However, some patients with all HV on the same side may also show two structural S‐shaped curves. Two patients with lumbosacral HV and thoracic HV on the same side presented with both structural lumbar and thoracic curves. In our department, we recommend that patients with UB MHV undergo HV resection as early as possible, as delayed surgical intervention may lead to curve progression that would require more extensive fusion. The fusion level is affected by the patient age, HV locations, and distance between MHV. If HV are located next to each other, short fusion may be feasible. However, as HV are usually separated by one or more normal vertebral bodies, long fusion is often necessary. In addition, as many patients also have a unilateral bar and rib synostosis, rib resection and long fusion should be considered. ¹¹ , ¹⁵

In PUB MHV, there are continuous HV on the same side of the spine, and one or more HV on the contralateral side. Patients with PUB MHV who have two adjacent unilateral HV, lumbosacral HV, fully segmented lumbar HV, or a unilateral bar with contralateral HV tend to present with severe deformity. ³ , ⁴ , ⁵ , ⁸ , ¹¹ , ¹⁶ Therefore, early surgical intervention may be the best choice for these patients. In young children with a non‐structural compensatory curve, respective hemivertebra resection and short fusion may be feasible. In patients with two or more structural curves, the principle is to fuse all structural curves, which brings concerns about growth deficits after long fusion in young children. Some children under the age of 12 years with severe spine deformity are treated with a combination surgery of growing rods and HV excision. ¹⁸ , ¹⁹

In CB MHV, there are no continuous HV located on the same side of the spine, which is similar to typical HMMS. Saito reported that nine of 354 patients with hemivertebrae (2.5%) had CB MHV, ¹⁰ which was similar to the prevalence of CB MHV in our study cohort (19 of 684, 2.8%), but slightly lower than that in the study by Shawen et al. (eight of 186, 4.3%). ⁵ In the present study, 10 of the 24 patients with CB MHV had surgical indications and most had only mild deformity at the initial visit. Thus, patients with CB MHV may not urgently need surgical treatment, but do require long‐term follow‐up. McMaster reported that the presence of two opposing HV at the same region (separated by one vertebra) may produce minimal spine deformity. ³ Adjacent HMMS result from a mismatch in the spine and some patients even have normal posterior components, which might be related to the mild deformity and progression. However, if the adjacent HMMS is located in the thoracolumbar or lumbar area, the patient can develop severe kyphosis. ⁹ In our study, only five patients in CB group had severe deformity in the coronal plane and progressive HV in the lumbar and lumbosacral regions. The other five patients had significant kyphosis in the thoracolumbar or lumbar regions. ¹⁶

Surgical Treatment of MHV

In planning surgical treatment for patients with MHV, it is important to minimize the fusion level and trauma. Posterior HV resection produces less trauma and causes fewer complications than anteroposterior approach. ⁷ , ²⁰ For most patients, it is not necessary to resect all hemivertebrae, and selective hemivertebrectomy has been used to treat MHV deformities. ²¹ , ²² In general, the hemivertebra that has an obvious influence on scoliosis, kyphosis, or progression should be excised first. For example, two or more close unilateral HV in the same region that usually cause severe scoliosis and progress rapidly should be resected. ³ , ⁸ , ¹⁴ It is not necessary to resect incarcerated or semi‐segmented HV that do not contribute to deformity. Lumbosacral HV are more likely to cause a progressive and severe deformity with coronal imbalance ³ and should be excised first. Theoretically, a contralateral lumbosacral hemivertebra with a lumbar curve will cause severe deformity and may need surgery immediately. ³ , ⁵ , ²³ , ²⁴ Although Nasca et al. found that the thoracocervical curve does not show a high rate of progressin, ¹¹ thoracocervical HV lead to shoulder imbalance, fixed torticollis, and facial asymmetry, which significantly affect appearance. ¹⁶ , ²⁵ , ²⁶ , ²⁷ Therefore, there is an urgent need to resect thoracocervical HV in patients with a significant change in appearance. ²⁸

Strengths and Limitations

The present study first reported the natural history of three different kinds of multiple hemivertebrae. Furthermore, posterior hemivertebrectomy and fusion could achieve satisfying radiographic outcome for the treatment of MHV. However, the present study had several limitations. First, it was a retrospective study with the inherent risk of data inaccuracy. Second, the sample size was small owing to the rarity of the condition. Third, the age span of the patients was quite large, which might have affected the results. Some patients had a late initial visit and some patients showed severe deformities without radiographs before the first visit, making it hard to evaluate their natural history. Therefore, the current study should be regarded as a preliminary study for further studies with a larger sample and longer follow‐up duration.

Conclusion

Most of the patients with unbalanced and partial unbalanced MHV had surgical indications, while surgical intervention may not be necessary for some patients with completely balanced MHV. The lumbosacral and thoracocervical HV could be excised first. Posterior hemivertebrectomy and fusion could be used for the treatment of MHV with satisfying radiographic outcome.

Author Contributions

Bowen Hu and Chunguang Zhou drafted the initial manuscript. Chunguang Zhou and Limin Liu conceptualized and designed the study, reviewed and revised the manuscript. Linnan Wang and Xi Yang carried out the initial analyses, reviewed and revised the manuscript. Yueming Song coordinated and supervised data collection, critically reviewed and revised the manuscript for important intellectual content. All authors approved the final manuscript as submitted and agree to be accountable for all aspects of the work.

Ethics Statement

This study was approved by the Ethics Committee of West China Hospital of Sichuan University (IRB number: 2019‐852). Informed consent for publication of their anonymized data was obtained from all subjects and/or their legal guardian(s). All methods were performed according to the appropriate guidelines and regulations.