Thoracogenic Spinal Deformity: A Rare Cause of Early Onset Scoliosis

Abstract

Object:

Surgery for severe congenital defects such as congenital diaphragmatic hernia, congenital heart defects, and tracheoesophageal disorders are life-saving treatments for many infants. However, the incidence of scoliosis following thoracoabdominal surgery ranges from 8­50%. There is little known about severe scoliosis acquired following surgery in infancy. We sought to evaluate patients who developed severe scoliosis following surgical treatment of congenital conditions.

Methods:

A multicenter database of patients with early onset scoliosis was queried to identify patients with a history of thoracogenic or acquired scoliosis. Patients with significant congenital spine deformities were excluded. 41 patients (1.6%) were noted to have thoracogenic scoliosis. Of those, 14 were observed, 10 were casted or braced, 17 underwent treatment with rib-based distraction rods, Shilla, or spine-based growing rod devices. Radiographs, complications, and patient characteristics were reviewed.

Results:

Mean age at scoliosis diagnosis for the 41 patients was 6.0 years. Mean time to follow-up was 2.9 years. Mean preoperative coronal Cobb angle in the surgical group was 65° and improved to 47° postoperatively (p=0.01). Mean Cobb angle for the non-operative group was 31° initially and 32° at follow-up (p=0.44). Among the 17 patients undergoing surgery for scoliosis, there were 13 complications in 7 patients, including a brachial plexus palsy following rib-based distraction rod placement. This resolved with revision of the rib hooks. There were no known complications in the non-operative cohort.

Conclusions:

Severe scoliosis can develop following thoracotomy and other pediatric surgical procedures. Work is needed to understand the pathogenesis of scoliosis in this population so as to implement preventative measures.

Level of Evidence:

IV, retrospective review of prospectively collected data

INTRODUCTION

Surgery for severe congenital defects such as congenital diaphragmatic hernia, congenital heart defects, and tracheoesophageal disorders are life-saving treatments for many infants. Though dwarfed by the obvious benefits and necessity of intervention, the development of severe scoliosis may be an undesirable consequence of thoracoabdominal surgery. Scoliosis affects 2% to 4% of the general pediatric population^7,17 but is much more prevalent in children following thoracoabdominal surgical intervention. Scoliosis is reported in 13% to 33% of children surviving congenital diaphragmatic hernia repair^6,13,16, 8.5% of children following cardiac surgery¹¹, and as many as 50% of children following tracheoesophageal fistula repair³. Severe scoliosis is also noted to occur following chest wall resection, and is often progressive^5,8.

Early onset scoliosis can range in severity and may result from a number of etiologies, most of which are poorly understood, though some do appear to be mechanical in nature^10,14. Patient age, underlying diagnoses, and curve characteristics all play a role in evaluating treatment strategies, which may include observation, bracing/casting, or surgical interventions for more significant curves. Vertical expandable prosthetic titanium rib (VEPTR®, DePuy Synthes, West Chester, PA, USA) and growth rods can be helpful interventions for severe curves in young children, though may worsen chest wall compliance and have high rates of complications¹. Improved understanding of the development and progression of severe scoliosis following thoracoabdominal surgery is necessary to help providers, patients, and families navigate treatment options and expectations.

The purpose of this study was to evaluate and characterize patients with severe thoracogenic scoliosis from a large multi-center database of children with early onset scoliosis, or disease onset prior to 10 years of age.

MATERIALS AND METHODS

We queried a multicenter database that prospectively enrolls and follows patients with early onset scoliosis. Twenty-five medical centers across the United States participate; and 2555 patients were prospectively enrolled in the registry from 2002 to 2014. Institutional review board approval was obtained at each site, and informed consent was obtained from all study participants/parents. We queried the database to identify patients with a history of thoracogenic or acquired scoliosis. Patients with congenital spine deformities were excluded from the current study.

We identified 43 patients (1.7%) with thoracogenic scoliosis. Two surgical patients (one VEPTR®, one indeterminate) had inadequate data available and were excluded. We reviewed radiographs, complications, and patient characteristics for 41 patients. Of those, 14 were observed, 10 were casted or braced, and 17 were treated surgically. Fifteen patients underwent treatment with VEPTR® devices, one underwent a Shilla procedure, and another had growing rods implanted. Six patients from the non-operative group and two from the operative group had no follow-up Cobb angles, but were included in the analysis.

Statistical analysis was performed using the Pearson chi-square test for nominal variables and the two-tailed student’s t-test for continuous variables. Matched pairs analysis was used to assess for change between pre- and post-operative values in the same patient. Observed differences that have a less than 5 per cent likelihood of occurring by chance were considered significant.

RESULTS

Mean age at presentation for the 41 patients was 6.0 years. Mean time to follow-up was 2.9 years (Table 1). Etiology for the scoliosis was heterogeneous. Diagnoses were similar among the operative and the nonoperative groups. Nineteen patients had had a previous thoracotomy, 4 had sternotomies, and 6 had chest wall resections for tumor. Three subjects in the operative group and four subjects in the nonoperative group had limited available data regarding the type of previous chest wall surgery. Four of these patients had congenital diaphragmatic hernia, one had congenital heart defect, one had tracheoesophageal fistula, and the final patient had leukemia treated with radiation.

Table 1.

Characteristics of surgical group. Age and follow-up given in years. Group means provided as mean (standard deviation).

	Age at treatment (Yrs.)	Sex	Diagnosis / associated interventions	Intervention	Cobb angle (°)	Curve levels	Follow-up (Yrs.)	Follow-up Cobb angle (°)
1	1.4	M	Chest wall resection	VEPTR	30	T1–T6	4.6	52
2	2.4	M	Congenital diaphragmatic hernia; thoracotomy	VEPTR	53	T7–L1	1.7	63
3	2.5	F	Spinal cord tumor; laminectomy	VEPTR	72	T5–T11	5.3	55
4	2.5	F	Congenital heart defect, pulmonary hypoplasia; thoracotomy	VEPTR	97	T4–T10	0
5	3.2	F	Congenital heart defect	VEPTR	86	T7–L2	8.5	34
6	3.3	M	Congenital diaphragmatic hernia; pulmonary hypoplasia	VEPTR	81	T6–T13	4.7	71
7	3.8	M	Chest wall resection	VEPTR	62	T4–L1	7.2	50
8	4.7	F	Congenital heart defect; thoracotomy	VEPTR	76	T6–L1	1.6	68
9	4.8	F	Spinal cord tumor; laminectomy	VEPTR	56	T9–L4	12.3	58
10	7.6	F	Congenital heart defect; thoracotomy	VEPTR	57	T3–T10	4.5	46
11	7.7	M	Chest wall resection; thoracotomy	VEPTR	29	T2–T9	1.3	13
12	9.1	F	Nemaline myopathy; pneumonectomy; thoracotomy	Growing rod	95	T12–L5
13	9.3	F	Congenital heart defect / Trisomy 21; sternotomy	Shilla	53	T7–T12	0.6	28
14	9.5	F	Congenital heart defect; thoracotomy	VEPTR	83	T6–T12	2.4	56
15	9.8	F	Esophageal atresia; thoracotomy	VEPTR	62	T4–T10	6.1	44
16	13.2	M	Tracheoesophageal fistula	VEPTR	72	T10-L3	2.6	49
17	17.9	F	Tracheoesophageal fistula; thoracotomy	VEPTR	33	T3–T10	9	22
MEAN	6.6 (4.3)				65 (21)		4.5 (3.3)	47 (16)

In the nonoperative group, mean age at presentation for the 24 patients was 5.4 years (Table 2). Mean Cobb angle for the non-operative group was 31° initially and 32° at follow-up (p = 0.44). Curves in the non-operative group progressed at least 10° in four patients, and improved at least 10° in another four patients. Of the curves that progressed, two had spinal cord tumors (laminectomies at L1–L2 and T11-L2), one had congenital diaphragmatic hernia repair involving a left thoracotomy, and one had radiation treatment for leukemia. Ten patients had initial curves greater than 30°; one of these had improved to less than 30° by follow-up, while an additional two progressed to greater than 30°.

Table 2.

Characteristics of non-operative group. Age and follow-up given in years. Group means provided as mean (standard deviation).

	Age at index (Yrs.)	Sex	Diagnosis / associated intervention	Intervention	Cobb angle (°)	Curve levels	Follow-up (Yrs.)	Follow-up Cobb Angle (°)	Follow-up curve levels
1	0.5	M	Congenital diaphragmatic hernia	Observed	23	T7–L3
2	0.9	F	Congenital diaphragmatic hernia	Cast	26	T3–T9	2.8	24	T5–T9
3	1.1	F	Congenital heart defect; sternotomy, laminectomy	Observed	29	T2–T6	0.6	36	T1–T6
4	1.4	F	Congenital diaphragmatic hernia; thoracotomy	Cast	34	T1–T10	2.1	48	T1–T10
5	2.1	F	Chest wall resection	Observed	37	T4–T10	2.3	41	T4–T10
6	2.3	M	Spinal cord tumor; laminectomy	Cast (then brace)	41	T3–T10	1.8	43	T2–T10
7	2.6	M	Tracheoesophageal fistula; thoracotomy	Cast (then brace)	56	T7–L2	1.8	34	T7–T12
8	3.1	F	Spinal cord tumor; laminectomy	Brace	16	T2–T11	9.2	8	T3–T6
9	3.3	M	Congenital heart defect / Trisomy 21; thoracotomy	Cast	40	T9–L2	0.7	30	T9–L2
10	3.6	M	Congenital heart defect; sternotomy	Observed	7	T3–L3	0.5	4	T12–L5
11	4.1	M	Chest wall resection	Observed	35	T5–T11	0.0
12	5.1	F	Congenital diaphragmatic hernia	Cast (then brace)	90	T5–L5	1.5	65	T6–L4
13	5.2	M	Congenital heart defect; thoracotomy	Observed	13	T3–T6	0.0
14	6.1	F	Congenital heart defect; thoracotomy	Brace	47	T12–L4	0.8	41	T12–L4
15	6.5	F	Congenital diaphragmatic hernia; thoracotomy	Observed	21	L1–L5	0.0
16	6.8	M	Chest wall resection	Observed	15	T3–T7	0.0
17	6.9	M	Congenital heart defect; sternotomy	Brace	26	T6–T11	1.7	27	T6–T11
18	7.8	M	Spinal cord tumor; laminectomy	Observed	18	T3–L1	2.3	10	T3–T9
19	8.0	M	Congenital diaphragmatic hernia; thoracotomy	Observed	20	T1–T9	0.0
20	8.6	M	Neuroblastoma; thoracotomy, laminectomy	Observed	32	T7–T11	2.7	26	T6–T11
21	10.0	M	Spinal cord tumor; laminectomy	Brace			8.6	20	T11–L3
22	10.2	M	Spinal cord tumor; laminectomy	Observed	21	T3–T6	3.2	32	T3–T9
23	10.8	F	Leukemia; radiation	Observed	47	T4–T12	2.0	63	T3–T12
24	11.4	M	Pulmonary hypoplasia; thoracotomy	Observed	22	T4–T7	1.8	21	T3–T7
MEAN	5.4 (3.2)				31 (17)		2.0 (2.3)	32 (16)

For the 17 patients in the operative group, mean age at scoliosis surgery was 6.6 years with a mean 4.5 years of follow-up. Mean preoperative coronal Cobb angle in the surgical group was 65° and improved to 49° postoperatively (p = 0.02, Figure 1). Fifteen of the seventeen patients in the operative group had initial curves greater than 30°. Mean Cobb angle at presentation was greater in the operative group compared to the nonoperative group (65° vs. 31°, p<0.001).

Fig 1.

A) Preoperative images, demonstrating left-sided rib resection and 62° curve (patient 7); B) Postoperative images, obtained following VEPTR® placement, demonstrating 24° curve.

Among the 17 surgical patients, there were 13 complications in 7 patients, including pneumonia (5), brachial plexus palsy (1), spine infection (3), device migration (1), implant failure (3 times in one patient). Three pneumonias required hospitalization, and two were treated on an outpatient basis. All infections and device problems required revision surgery. The most serious perioperative complication was in a patient with previous chest wall tumor resection who developed a brachial plexus palsy following VEPTR® placement. This resolved with revision of the rib hooks. Among the nonoperative patients, one patient died and another patient had cast problems requiring trimming of the cast in clinic.

DISCUSSION

The development of scoliosis following rib resection has been well-documented^8,5,10 and is often employed in experimental models^2,9,10. Severe scoliosis has also been identified following thoracotomy and other pediatric surgical procedures, but is less well understood^3,5,12,15 The present study characterizes the underlying diagnoses in a cohort of children with thoracogenic scoliosis from a large registry of patients with early onset scoliosis.

The exact timeline and pathophysiology for development of scoliosis following thoracotomy or similar procedures remains unclear. Despite the increased incidence and significant burden to patients and their families, there is little known about severe scoliosis acquired following surgery in infancy. Van Biezen and colleagues evaluated 160 patients undergoing thoracotomy for aortic coarctation and noted the development of scoliosis in 22% of patients, despite the absence of scoliosis prior to surgery¹⁵. Additional data may be helpful for surgeons as they counsel families about possible long-term sequelae of chest wall surgery performed at a young age. If children most at risk for scoliosis could be identified, screening measures could be put in place to ensure early access to nonoperative scoliosis treatment.

Further, it is uncertain whether thoracogenic scoliosis is due solely to an underlying diagnosis, such as a unifying defect in development resulting in both congenital diaphragmatic hernia and scoliosis, for example, or due to surgical interventions for the underlying diagnosis. Spinal cord tumor or chest wall tumor can result in scoliosis, and it is unclear whether the tumor or the surgical intervention causes the scoliosis. These patients were included in the study nevertheless, since they fall in the category of iatrogenic or thoracogenic scoliosis, or scoliosis which develops after a medical intervention. Russell et al. evaluated a number of patients undergoing congenital diaphragmatic hernia repair using a variety of surgical techniques and found no significant influence of repair technique on subsequent development of scoliosis¹³. Glotzbecker et al. reported severe curves (mean 26°, maximum 70°) following childhood resection of chest wall tumors, with tumor resection above the 6^th rib at increased risk for developing scoliosis⁵. This illustrates the need for further investigation and understanding of the mechanisms and pathogenesis of scoliosis development.

This study has a number of limitations. Though data was collected prospectively, limited information regarding the index chest wall surgery was available for analysis, especially for the non-operative group. Some centers only enroll surgical patients, so potentially some nonoperative patients with thoracogenic scoliosis were not entered in the database. Curve location was well characterized, but laterality of the thoracotomy or surgical intervention was not reliably reported, nor was direction of the curve. Additionally, the lack of a control group limits interpretation in the context of managing patient expectations and outcomes with regard to incidence of scoliosis as a result of the described procedures. In some instances, patients had more than one diagnosis, such as a syndrome, congenital heart disease, and a thoracotomy. Syndromic or neuromuscular diagnoses such as myopathy, Down syndrome, as well as spinal cord tumors may contribute to scoliosis progression. Treatment strategies may vary by center. For instance, some institutions may not offer casting for early onset scoliosis or prefer surgical techniques over casting⁴. Follow-up was limited to what was available within the multicenter registry.

CONCLUSIONS

Despite these limitations, this study does provide significant information regarding underlying diagnoses, as well as the development and progression of thoracogenic scoliosis. Based on the patients in our registry, thoracogenic scoliosis represents a very small proportion of patients with early onset scoliosis. Nevertheless, the resultant disease can be severe, requiring childhood surgical management. In this series, patients with severe early onset scoliosis were successfully treated with rib-based growing devices. Prospective studies may illuminate what surgical characteristics are associated with the development of scoliosis following early treatments in childhood. This project may raise awareness of the association between severe scoliosis and childhood chest wall procedures and stimulate further prospective research efforts.