Surgical complications of kyphoscoliosis secondary to neurofibromatosis type 1: a systematic review

Article information

Asian Spine J. 2026;.asj.2025.0436

Publication date (electronic) : 2026 May 21

doi : https://doi.org/10.31616/asj.2025.0436

Xinwei Wang^*, Ziquan Li^*, Xiaohan Ye, You Du, Yiwei Zhao, Haoran Zhang, Chenkai Li, Nan Wu, Jianguo Zhang, Shengru Wang

Department of Orthopaedic Surgery, Peking Union Medical College Hospital, Peking Union Medical College and Chinese Academy of Medical Sciences, Beijing, China

Corresponding author: Jianguo Zhang, Department of Orthopaedic Surgery, Peking Union Medical College Hospital, Peking Union Medical College and Chinese Academy of Medical Sciences, 1st Shuai Fu Yuan, Dongcheng District, Beijing 100730, China, Tel: +86-13901106754, Fax: +86-1069152809, E-mail: jgzhang_pumch@yahoo.com

Co-corresponding author: Shengru Wang, Department of Orthopaedic Surgery, Peking Union Medical College Hospital, Peking Union Medical College and Chinese Academy of Medical Sciences, 1st Shuai Fu Yuan, Dongcheng District, Beijing 100730, China, Tel: +86-1840071309, Fax: +86-1069152809, E-mail: wangshengru@foxmail.com

*These authors contribute equally to this work as the first author.

Received 2025 August 16; Revised 2025 December 18; Accepted 2026 January 5.

Abstract

Introduction

Neurofibromatosis (NF), a common autosomal dominant disorder, can be clinically classified into type 1 (peripheral) and type 2 (central). As the most prevalent type, NF type 1 (NF-1) can be identified based on some typical features, including milk café-au-lait spots, axillary and/or inguinal (crease) freckles, Lisch’s nodes (iris misshapen tumors), and neurofibromas.

NF-1 frequently affects the skeletal system [1], manifesting as dysplasia, pseudoarthrosis, osteoporosis, and other bone lesions. Scoliosis has also been observed in 10%–25% of patients with NF-1 [2] and frequently involves the thoracic vertebrae. Skeletal involvement in NF-1 can be categorized into dystrophic and non-dystrophic subtypes [3].

Common surgical strategies used for the treatment of kyphoscoliosis secondary to NF-1 include anterior-only, posterior-only, and combined anterior–posterior approaches (CAPA) [4]. The surgical treatment of NF-1-related scoliosis and kyphosis tends to be challenging and complex, which increases the risk of various types and forms of surgical complications, with risk factors including age of onset, severity of the deformity, and apical vertebral location. Generally, around 32.7% of patients with NF-1-related kyphoscoliosis develop complications following surgical treatment [5]. Aside from perioperative and non-neurological complications, common surgery-related complications include instrumentation-related complications (IRCs) (e.g., screw dislodgement, cap loosening, and rod breakage), alignment-related complications (ARCs) (e.g., curve progression, adding-on phenomenon, trunk shift, and junctional kyphosis), pseudoarthrosis, neurological damage and other perioperative complications.

Several studies have assessed the surgical outcomes of NF-1-related kyphoscoliosis; however, few have comprehensively analyzed surgical complications. In 2024, Wang et al. [5] performed a meta-analysis evaluating the clinical characteristics and surgical outcomes of NF-1-related kyphoscoliosis, but their study lacked a comprehensive analysis of surgical complications. The current systematic review incorporates a broader range of studies to analyze the impact of different surgical approaches and techniques on complication rates. Subgroup analyses were conducted to mitigate heterogeneity and allow for more precise interpretation.

Materials and Methods

This study is a systematic review of previously published literature and does not involve any human participants, human data, or human tissue.

Results

Study characteristics and quality assessment

Database searches identified 637 studies, among which 43 were published between 1984 and 2024 and ultimately met the inclusion criteria for qualitative synthesis as detailed in the PRISMA flowchart (Fig. 1). Quality assessment using the Revised Cochrane Risk of Bias Tool revealed methodological limitations in several studies (Fig. 2) [4,6–46]. Two studies (4.7%) were rated as high risk due to uncontrolled confounding factors, whereas four studies (9.3%) exhibited significant selection bias related to non-randomized designs. Selective outcome reporting affected five studies (11.6%), and one study (2.3%) was deemed critically biased due to excessive missing data. The majority of the included studies (72.1%) demonstrated low risk or minor concerns across bias domains, with strong inter-reviewer agreement (κ=0.82).

Fig. 1

Flow-chart showing the main steps of the review process according to the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) statement recommendations.

Fig. 2

Risk of bias summary using the Revised Cochrane Risk-of-Bias Tool.

The key characteristics of the included studies are summarized in Table 1 [4,6,8–48]. A total of 1,157 patients with kyphoscoliosis secondary to NF-1 after surgical treatment were identified, with an average age of 12.9 years and a follow-up period of 5 years. Most patients were diagnosed with scoliosis and kyphosis simultaneously, with an average coronal deformity Cobb and kyphosis Cobb angle of 74.7° and 54.7° before surgery, respectively.

Table 1

Study included and basic patient characteristics and surgery information

The surgical strategy for patients with kyphoscoliosis secondary to NF-1 varied according to the severity and position of deformity. Spinal fusion strategy can be simply classified into the anterior approach, posterior approach and a combination of the two. According to the included studies, 703 patients (60.8%) underwent posterior surgical approach, whereas only 37 (3.2%) and 230 patients (19.9%) underwent anterior approaches and CAPA, respectively. Instead of spinal fusion strategy, a growing-rod was used in 187 patients (16.2%).

Further details about the surgery were collected and summarized. According to the included literature with full information about the surgical records, the mean blood loss and mean operative time was 1,037.1 mL and 5.1 hours, respectively. Meanwhile, the general corrective rate of scoliosis and kyphosis was 54.1% and 53.6%, respectively. Data integration was based on the literature reports after weighting.

Description of surgical complications

Overall, 403 complications in 1,157 patients were described from the included studies (Table 2) [4,6,8–48]. The mean complication ratio was 34.8%, which ranged from 0.0% [6,7,47] to 171.4% [8]. Furthermore, 337 patients (83.6%) developed minor complications with a C-D grade of under II, whereas 66 patients (16.4%) developed major complications, with a C-D grade of III or higher.

Table 2

Complications assessment and classification from studies included

The most frequent form of surgical complications was ARCs, consisting of curve progression, adding-on phenomenon, trunk shift, junctional kyphosis etc. Among all reported complications, ARCs accounted for 30.0%. IRCs also accounted for 25.3% of the total complications, including screw dislodgement, cap loosening, and rod breakage. Meanwhile, 62 patients (15.4%) developed internal fixation failure.

Other complications were also reported during the perioperative period. Cai et al. [25], who compared the surgical outcomes of the spinal fusion strategy and the growing-rod system, reported a 31.3% perioperative complication rate, with two patients developing surgical site infection, two developing ileus, and one experiencing delayed wound healing. Other studies also reported high rates of infection (n=18) [9–15,47]. Chen et al. [32], who studied 16 patients with severe cervical kyphotic deformity secondary to NF, reported five typical perioperative complications, which included hoarseness (n=3; 18.8%) and dysphagia (n=2; 12.5%). Only one patient developed intermittent bleeding in the study of Yifei et al. [17], whereas two patients developed megableeding in the study by Iwai et al. [18]. After comparing the surgical outcomes of different approach of surgeries for dystrophic cervical kyphosis cases, Lin et al. [8] reported that four patients (4.9%) developed bronchopneumonia after treatment. In summary, different types of perioperative complications were reported at relatively low proportions in the entire cases of complications.

Subgroup analyses of surgical strategy, surgical approach, deformity location, and follow-up

Spinal fusion and growing-rod strategy

According to our summary of information, the surgical strategy for kyphoscoliosis secondary to NF-1 can be divided into mainly two parts: the spinal fusion strategy and the growing-rod strategy. Among all the studies collected, the growing-rod system was used in 208 patients (18.0%). Data on complications in the two groups were organized and compared (Table 3). Regarding the overall complication and IRC rate, the difference in incidence between the two groups was markedly significant (p<0.05). After assessment using the C-D grading system, significant differences in minor complication rates were also noted. No differences in neurological damage rates were observed between the two groups.

Table 3

Comparison of the complication rates between two major strategies of surgical treatment

Posterior approach and CAPA

Patients who underwent spinal fusion surgery were also categorized into those who underwent the posterior approach and CAPA and subsequently compared according to surgical complication rates under the different approaches. The included studies were also categorized, filtering out those that explicitly used only one surgical approach and considering those that used multiple surgical approaches but with detailed information making the distinction possible. As shown in Table 4, a total of 457 patients from all included studies underwent the posterior approach, with a scoliosis and kyphosis correction rate of 60.4% and 49.9%, respectively, a mean blood loss of 1,071.7 mL, and a mean operative time of 4.8 hours during the procedure. Conversely, 61 patients underwent CAPA, with a scoliosis and kyphosis correction rate of 55.3% and of 59.8%, respectively, a mean blood loss of 1,261.7 mL, and a mean operative time of 7.3 hours. Additionally, the preoperative Cobb angle differed significantly between the two groups, and the CAPA group showed a significantly larger preoperative kyphosis angle than did the posterior approach group. However, no significant differences in the scoliosis (62.3% vs. 55.3%) and kyphosis (49.9% vs. 59.8%) correction rates were noted between the two groups.

Table 4

Comparison of the complication rates between posterior approach and CAPA

After comparing the complication rates of the two surgical approaches, we found that patients who underwent CAPA showed significantly higher total and minor complication rates than did those who underwent the posterior approach. Regarding major complication rates, although rates were still higher in patients who underwent CAPA than in those who underwent the posterior approach, the difference did not reach significance (p=0.30). Furthermore, the probability of ARCs, ARCs, neurological damage, and perioperative complications were all higher in the CAPA group than in the posterior approach group, albeit not significantly.

Cervical kyphosis and thoracolumbar kyphoscoliosis

We further explored the impact of kyphosis location on the incidence of surgical complications given that targeting cervical kyphosis and thoracolumbar kyphoscoliosis can cause large differences in surgical choices and operative procedures. We performed a secondary screening of the included literature to distill and statistically analyze the surgical studies for cervical kyphosis only, as well as the surgical studies covering thoracolumbar kyphoscoliosis (Table 5). According to our results, 159 patients had cervical kyphosis, whereas 799 could be categorized as having thoracolumbar kyphoscoliosis. Overall, the surgical complication rate significantly lower for the thoracolumbar kyphoscoliosis group than for the cervical kyphosis group (p<0.01). A significant difference in the minor complication rate, especially in terms of IRC, was observed between the two groups. Patients in the cervical kyphosis group had a higher incidence of ARCs, whereas the thoracolumbar kyphoscoliosis group appeared to have higher rates of neurological damage, although the difference did not reach significance.

Table 5

Comparison of surgical complication rates between cervical kyphosis and thoracolumbar kyphoscoliosis

Follow-up duration

To further investigate the impact of follow-up duration on postoperative complications, patients were categorized into two groups: those with a mean follow-up of ≤5 years (n=470) and those with >5 years (n=267). As shown in Table 6, the overall complication rate was significantly lower in the shorter follow-up group than in the longer follow-up group (p=0.02). This difference was primarily driven by minor complication rates (27.0% and 35.6% in the ≤5 years and >5 years groups, respectively; p=0.04). The minor complications in the longer follow-up group showing increased rates predominantly consisted of low-grade ARCs, such as curve progression, adding-on phenomenon and junctional kyphosis, as well as mild IRCs not requiring surgical or interventional management. However, no significant differences in other complication categories, including major complications, ARC rate, IRC rate, and neurological damage rate, were observed between the ≤5 years and >5 years groups.

Table 6

Comparison of complication rates between average follow-up duration of ≤5 years and >5 years

Discussion

The surgical treatment of NF-1-related kyphoscoliosis tends to be challenging, complicated, and prone to various types and forms of surgical complications. The current study systematically reviewed existing literature for the purpose of summarizing, analyzing, discussing, and explaining information about the surgical complications of the NF-1-related kyphoscoliosis. We found a general complication rate of 34.8% in all 43 studies included, with minor and major complications accounting for 83.7% and 16.3%, respectively. Based on all information collected from the included studies, the most frequently occurring type of complications was ARCs.

The most common ARCs occurring following surgical treatment included curve progression, adding-on phenomenon, trunk shift, and junctional kyphosis [19], which have been closely associated with the patient’s curve, metabolic and bone morphology status. According to the findings of Funasaki et al. [49], risk factors for progression of NF-1-related kyphoscoliosis contains early age of onset, a large value of the Cobb angle upon initial diagnosis, an abnormal posterior convexity of the spine, scalloped vertebrae, the occurrence of severe rotation of the apex, and pencil-shaped ribs. Given that patients with dystrophic lateral kyphosis have lower bone density and are more likely to have lateral kyphosis with greater curvature, they have a higher incidence of associated complications. According to the study of Sirois and Drennan [50] in 23 patients with severe dystrophic spinal deformities, the incidence of pseudoarthrosis was 38%, with a mean deformity progression of 12.7°. Furthermore, comorbidities, such as dural dilatation and osteoporosis with hyperthyroidism, can together trigger deformity progression in NF-1-related lateral kyphosis of the spine [51].

The growing-rod technique is a widely used growth-friendly surgical strategy for NF-1-related kyphoscoliosis. Accordingly, Cai et al. [19] reported that treatment of NF-1-related kyphoscoliosis using the growing-rod technique promoted superior intraoperative blood loss, operative time, and maintenance of correction, prevention of postoperative implant deformation and loss of alignment, and preservation of spinal growth than did posterior spinal fusion surgery. However, considering the need for multiple surgeries and prolonged follow-up, concerns about its long-term safety and complication rates have been raised. Yao et al. [20], on the other hand, concluded that early spinal fusion induces fewer device-related complications and achieved a higher correction rate than did the growth rod system. Our review found that although the traditional growing-rod technique successfully maintains spinal growth and reduces early surgical trauma, it is associated with significantly higher complication rates than is spinal fusion. Specifically, IRCs rates were notably higher in the growing-rod group than in the spinal fusion group, whereas neurological damage rates remained similar. Despite being associated with significantly higher complication rates, the traditional growing-rod strategy cannot be overlooked as an excellent and effective spinal orthopedic strategy that benefits the patients by preserving the growth potential of the spine over a longer period of time and by meeting the multidimensional requirements of children in terms of outlook and growth needs, which cannot be achieved with the fusion strategy. Recent studies have focused on improving the corrective efficacy and success rates of the growing-rod technique while exploring strategies to reduce surgical complications. Wang et al. [52] found that a hybrid technique that combines vertebrectomy/hemivertebrectomy with short fusion and dual growing rods significantly reduced the risk of mechanical complications, although it may increase the risk of dural tears and neurological complications. According to Li et al. [53], the growing-rod technique combined with apical control techniques could benefit the patients by lowering the incidence of mechanical complications and revision surgery. Studies have proposed the use of magnetically controlled growing rods as a means to reduce the number of required surgeries, thereby lowering the incidence of infection-related complications. However, this approach does not appear to prevent common implant-related complications [54]. From a clinical perspective, more rigorous follow-up, exploration and establishment of more stable growth-friendly orthopedic systems, and more specialized advice and health promotion may potentially reduce the incidence of complications and thus improve patient outcomes.

When considering surgical approaches, our review found that although both the posterior approach and CAPA achieved comparable correction rates, the posterior approach was associated with a lower overall complication risk, particularly for minor complications. One important factor influencing these outcomes is patient selection. Our data showed that the CAPA group had significantly larger preoperative coronal (93.8° vs. 79.1°) and sagittal (55.5° vs. 49.0°) deformities, suggesting that CAPA was primarily selected for more severe cases. Thus, the higher complication rates observed in the CAPA group could be partly attributed to this selection bias rather than a disadvantage of the technique itself. Despite the more severe preoperative deformity in CAPA patients, the final correction rates between the two groups did not significantly differ (scoliosis: 55.3% vs. 62.3%; kyphosis: 59.8% vs. 49.9%). This finding raises an important clinical consideration regarding whether CAPA truly offers superior correction or whether similar outcomes could be achieved with a less invasive posterior approach in certain cases. Given that CAPA is associated with significantly longer surgical time (7.3 hours vs. 4.8 hours) and greater blood loss (1,261.7 mL vs. 1,071.7 mL), its application should be carefully considered, particularly in patients with borderline indications. The increased surgical duration and intraoperative bleeding may increase the risk of perioperative complications, including wound infections, coagulopathy, and neurological injury. Nevertheless, CAPA provides additional anterior column support and fusion, which may be beneficial in cases with severe deformities requiring greater stability. Previous studies have reported that posterior fusion alone has a high failure rate when kyphosis exceeds 50° (failure rate of 64% vs. 80% with vs. without anterior fusion, respectively) [55], reinforcing the idea that CAPA may be necessary in select cases. In moderate deformities, however, no clear evidence has shown whether posterior-only techniques can be optimized to achieve comparable stability and correction, potentially reducing the need for CAPA in less severe cases. Recent advancements in biomaterials may help bridge the gap between lower complication rates and effective spinal stabilization. The use of lower elastic modulus metals, such as Ti-6Al-4V alloy [56] and polymeric materials like polyetheretherketone [57], in posterior interbody fusion shown promising fusion rates and long-term clinical outcomes. These innovations could potentially enhance the effectiveness of posterior-only procedures, thereby providing an alternative to CAPA in select cases. Additionally, the application of novel materials in CAPA procedures may improve anterior column support and reduce implant-related complications, further refining the surgical strategy for severe spinal deformities. In summary, although CAPA remains a valuable technique for severe scoliokyphosis, its use should be carefully tailored based on individual patient characteristics. The choice of the surgical approach should not be based solely on complication rates but rather on a comprehensive evaluation of the patient’s deformity severity, surgical risks, and potential benefits. Future research should focus on optimizing posterior fusion techniques and integrating novel biomaterials to improve outcomes while minimizing surgical morbidity.

NF-1-related spinal deformities can affect different spinal regions, with varying levels of surgical complexity and risk. Among these, cervical kyphosis presents a unique challenge owing to its proximity to critical neural structures and its impact on head and neck posture. Our findings showed that surgery for cervical kyphosis was associated with significantly higher incidence of complications, with minor complications predominating. In contrast, patients in the cervical kyphosis group had a higher incidence of ARCs and IRCs, with neurological damage rates appearing to be higher in the thoracolumbar kyphoscoliosis group. Surgical correction of cervical kyphosis often requires more precise intraoperative navigation and stabilization techniques [58], which increases the overall risk of complications. According to Han et al. [59], cervical kyphotic deformities may involve facet joint abnormalities or extend into the thoracic spine, further complicating surgical correction. Additionally, cervical kyphosis surgery typically requires three-column correction, which, despite achieving more substantial deformity correction, also increases surgical risks, including neurological injury, internal fixation failure, and intraoperative blood loss [60]. Overall, the risk of complications following the surgical treatment of cervical kyphosis remains significantly elevated due to the complex anatomical structures of the cervical spine, the higher technical demands of the procedure, and the need for enhanced fixation stability.

Regarding the impact of follow-up duration on complication rates, our analysis revealed that patients with a follow-up duration exceeding 5 years experienced a significantly higher overall complication rate than did those who were followed up for 5 years or less (p=0.02). The higher complication rates may partially reflect the cumulative effect of prolonged observation rather than an intrinsic increase in surgical risk over time. However, we emphasize that long-term postoperative surveillance may uncover additional complications that were either not apparent or had not yet developed in the early postoperative period. The difference was particularly pronounced in minor complications, which were significantly more frequent in the longer follow-up group than in the shorter follow-up group (p=0.04). These minor complications mainly consisted low-grade ARCs, including mild curve progression, adding-on phenomenon, trunk shift, and junctional kyphosis, as well as non-critical IRCs that did not necessitate intervention or surgical management. From a mechanistic perspective, this trend may be attributed to the cumulative effects of long-term implant loading, continuous spinal growth and remodeling, and the intrinsic skeletal dysplasia associated with NF-1, which may predispose patients to gradual changes in spinal alignment over time. Interestingly, despite the significant difference in the overall and minor complication rates, no significant difference in major complication rates or neurological damage rates were observed. This finding suggests that although minor complications may accumulate over time, severe surgical outcomes may stabilize after the initial postoperative period. Importantly, most long-term minor complications were managed conservatively through close clinical observation, radiographic surveillance, physical therapy, or brace support, without the need for revision surgery. Some major complications requiring revision surgery may have also occurred within the first few years after surgery, leading to an underestimation of their long-term incidence. Further studies with longer follow-up durations and a more detailed breakdown of complication types could provide additional insights into the long-term safety profile of surgical treatment in this patient population [61].

Some strategies could also potentially reduce surgical complications. Indeed, previous reports have shown that preoperative traction was beneficial in improving pulmonary function in patients with severe NF-1-related kyphoscoliosis and may somewhat reduce the risk of neurological compromise [62]. Patients may present with postoperative neck pain and occasional neurological complications, such as nerve root damage or complete/incomplete spinal cord functional deficits [63]. The study by Zhang et al. [11] concluded that combining halo traction with posterior surgery was effective in correcting the aforementioned kyphosis and may even improve neurological deficits. We also attempted to compare the postoperative complication rates of patients who underwent preoperative halo-pelvic traction with those who did not. Accordingly, our results showed that overall complication rates were lower in the former than in the latter. However, given the small number of cases (n=91) and the possibility of incomplete documentation of complications, more clinical trial data are still needed to support the potential role of preoperative traction in reducing complication rates. To reduce the risk of neurological impairment, attention should also be directed to the use of routine intraoperative neurological testing, including transcranial motor evoked potential, and perioperative control of the patient’s blood pressure and hemoglobin levels to maintain spinal cord perfusion [21]. Moreover, pedicle screw placement is more challenging during NF-1-related scoliosis surgery, with mispositioned screws similarly leading to internal fixation failure. A study by Kassis et al. [64] showed that the rate of pedicle screw misplacement can reach as high as 40% in the thoracolumbar spine. The use of intraoperative O-arm scanning and triggered screw electromyography can help identify misplaced screws, thereby ensuring safety and improving the positional accuracy of pedicle screws [22].

This study has several limitations worth noting. First, given the rarity of NF-1, only 1,157 patients had been identified from the literature review, which is still insufficient for summarizing and analyzing complications. The reporting bias in some of the literature may have a greater impact on the conclusions. Second, the lack of individualized data on patients in some studies and the large differences in follow-up durations complicated classification and precluded some comparisons. Third, given the missing data on the management of complications in many of the included studies, more clinical evidence is still needed to support the treatment and management of surgical complications in NF-1-related kyphoscoliosis.

Conclusions

This systematic review highlights the significant surgical burden associated with kyphoscoliosis secondary to NF-1. Our findings indicate that surgical treatment for NF-1-related spinal deformities remains challenging, with an overall complication rate of 34.8%. Among the documented complications, ARCs and IRCs were the most common. Despite advancements in surgical strategies, our analysis suggests that CAPA, despite being preferred when encountering high-extent deformities, may carry a higher risk of complications than would posterior-only approaches. Similarly, the growing-rod technique, although it can preserve spinal growth, has been associated with a significantly increased risk of IRC. Cervical kyphosis corrections present an even greater surgical challenge, demonstrating higher complication rates than thoracolumbar procedures. Additionally, our data indicate that a longer follow-up duration (>5 years) is associated with a higher incidence of minor complications, underscoring the importance of long-term monitoring in NF-1 patients.

Article information Continued