Comparison between surgical fusion and the growing-rod technique for early-onset neurofibromatosis type-1 dystrophic scoliosis

Neurofibromatosis type-1 (NF-1), first described by von Recklinghausen in 1882, is a rare autosomal dominant neurocutaneous disorder. It is caused by a mutation in chromosome 17q11.2 of the NF-1 gene, which leads to a loss of function of the neurofibromin protein. This results in a wide variety of manifestations and clinical complications since it affects most organ systems, including deterioration of the skin, bones (neurofibroma), arteries, peripheral nerves, and the central nervous system. NF-1 has an incidence of 1 in 2500 to 3000 individuals and a prevalence of 1 in 4000 to 5000 individuals [1, 2]. It has been reported that 10–60% of NF-1 patients present with early-onset spinal deformities, leading to either dystrophic or non-dystrophic scoliosis [3]. Typical dystrophic NF-1 scoliosis has a short, sharp curve and can be recognized by three or more of the following features: rib pencilling, vertebral scalloping, wedging, rotation, and spindling of the transverse process.

Considering the nature of dystrophic NF-1 scoliosis, which has a tendency of curve progression, leading to poor pulmonary function and truncal height loss, fusion is usually recommended [4‐6]. Generally, the more severe the dystrophic changes are, the higher the likelihood of a scoliotic curvature. Dystrophic scoliosis used to be treated aggressively, and anterior-posterior fusion has been widely used [7, 8]. In recent years, to avoid encountering extensive plexiform tumours during the anterior approach, posterior-only fusion (PF) has been reportedly used in treating NF-1 dystrophic scoliosis, yielding good short-term outcomes [9, 10]. However, other than in one recent report [11], the long-term outcomes of early posterior spinal fusion in patients aged ≤10 years old have not been well reported.

Growth guidance using the growing-rod (GR) system is an alternative surgical treatment that has been demonstrated to be safe and effective in treating early-onset scoliosis (EOS). However, its use and outcomes in NF-1 patients have not been extensively reported in the literature [12, 13]. The purpose of this study was to evaluate and compare the medium-term surgical results and complications between these two posterior-based surgical techniques in early-onset NF-1 dystrophic scoliosis patients.

Twenty-six patients with NF-1 scoliosis were identified in our database. Among these, we excluded patients who underwent anterior-posterior fusion (3 cases, 11.5%), those who underwent one-stage posterior osteotomy with short segment fusion (2 cases, 7.7%), and those whose follow-up period was < 2 years (3 cases, 11.5%). Two patients in the PF group whose major curve was in the lumbar region were excluded based on the second inclusion criterion. Thus, eight cases of PF and eight cases of GR were included in our study. The average age of the patients (3 males, 5 females) in the PF group was 7.7 ± 2.3 years, and the mean follow-up was 56.9 ± 18.2 months. The average age of the patients (2 males, 6 females) in the GR group was 7.4 ± 1.4 years, with a mean follow-up of 51.0 ± 17.5 months. The mean age (p = 0.74), sex ratio (p = 0.58), and follow-up duration (p = 0.52) were similar between groups.

In this study, we reported the clinical outcomes of the posterior-only surgical treatment for dystrophic EOS in patients with NF-1. This is the first study to compare the outcomes of PF to GR surgery in dystrophic NF-1 scoliosis patients under the age of 10 years.

For cases of non-dystrophic EOS, treatment strategies similar to those for idiopathic scoliosis can be used [19]. Regarding dystrophic EOS cases, for which the deformity progresses rapidly [14], conservative treatment cannot control progression, and surgical intervention is often indicated [5, 6]. Surgical treatments for NF-1 are challenging to perform [20]. A few cohort studies [11, 12, 21, 22] have indicated that combined anterior-posterior fusion is recommended for EOS with NF-1. However, anterior-posterior fusion requires a more complicated operation and has a higher risk. Currently, pedicle screw-based instrumentation allows better control of the vertebral body than do Harrington rods or hook-rod-based instrumentation, thus making PF possible. PF also does not require access through the thoracic/abdominal cavity, which can reduce the difficulty and risk of complications associated with surgery. Recent studies have shown that PF surgery can yield good outcomes [10, 19]. Given that major curves can progress despite being fused, early fusion can cause inadequate thoracic growth in NF-1 patients. Therefore, delaying fusion has been recommended [23]. The GR system can delay fusion and reduce the operative time and blood loss. Thus, it has been performed increasingly more often since its development. Both PF and GR methods have been reported to be used in the treatment of scoliosis, but no systematic reviews comparing these two approaches have been reported. On the other hand, the law of diminishing returns with growth rods should not be ignored, and it should be considered that either revision surgery or definitive fusion may be more difficult to perform after growth-rod treatment.

In our comparative study, the initial curve correction rates for these two groups (PF, 52.7%; GR, 56.47%) were similar to those reported for Jain’s GR group (59%) but did not reach the rate for the anterior-posterior fusion group (66.15%) [11, 12]. Additionally, the number of segments instrumented was similar between our study in Jain’s study. In most cases, the progression of scoliosis is caused by pseudarthrosis. For patients with EOS, the crankshaft phenomenon is another important complication [24]. It was reported that 21% of EOS patients with NF-1 who underwent PF experienced the crankshaft phenomenon during the follow-up period in Greggi’s study [21]. GR reduces the occurrence of the crankshaft phenomenon, as does the posterior procedure, while allowing the trunk to elongate with growth. In our GR group, one in eight patients experienced the crankshaft phenomenon, and the follow-up correction rate remained at 52.27%, which is similar to that in the series reported by Jain et al. (50.8%) [12]. We should also be aware that decompensation in the fusion group can be related to the level selected for fusion.

The GR group favourably demonstrated T1-S1 growth and T1-T12 growth, whereas the PF group did not. In our study, the mean number of fused segments was 8.25 per patient in the PF group; the PF group showed significantly less T1-S1 growth than did the GR group, and the GR group demonstrated approximately 1.12 cm/year of growth in Jain’s series [12]. An increase in the height from T1-T12, the main parameter of the thoracic cavity, may be essential for lung growth and pulmonary function [25]. The T1-T12 growth/year in the GR group (0.60 ± 0.27 cm/year) was also significantly greater than that in the PF group (0.20 ± 0.15 cm/year) (p = 0.02), indicating that for EOS patients with NF-1, the GR approach is better than the PF approach in terms of thoracic cavity growth. Karol et al. mentioned in their study that patients with proximal thoracic deformity who require early fusion are at risk for restrictive pulmonary disease. The pursuit of alternative procedures to treat early spinal deformity is merited [26]., Yang et al. mentioned in their review that growth-friendly spine surgery has been shown to correct spinal deformity while allowing growth of the spine and subsequently lung growth [27]. Karol et al. also mentioned in their article that diminished thoracic spinal height correlates with decreased forced vital capacity [28]. We believe more studies with direct measurement of pulmonary function in GR group patient must be conducted to determine the effect of GR on pulmonary function development.

In the PF group, two patients had proximal screw dislodgment (one was displaced, and one was pulled out), one of whom underwent revision. In the GR group, one patient suffered loosening of the cap of the proximal screw, causing the rod to loosen and come out during the interval between lengthening surgeries. In the study by Jain et al., proximal construct failure was the most common implant-related complication (5/14) [12]. The correction force is high and the bone mineral density is low in EOS patients with NF-1, which are the main reasons for proximal anchor site failure. Early aggressive treatment, such as surgeries for dystrophic EOS in NF-1 patients, has been recommended. In Greggi’s report and our study, the clinical results of PF were unsatisfactory due to the high occurrence of deformity progression and its limitation of thoracic growth. The GR system, however, retained the possibility of growth and allowed fusion to be delayed in patients with NF-1 while leading to a degree of deformity progression similar to that of PF. In our study, the GR group had better outcomes in terms of blood loss, operative time, correction maintenance, and spinal trunk growth than did the PF group. However, it is noteworthy that the need for repeated surgeries was a disadvantage for the GR group in our study. This disadvantage may be improved by the newer version of magnetically controlled GRs. This newer version of GR with noninvasive distraction has been successfully used for EOS cases of various aetiologies [29, 30]. Though its efficacy in the context of NF-1 patients is currently unknown, it is reasonable to expect better outcomes with magnetically controlled GR with regard to the number of scheduled follow-up surgeries.

This study has some limitations. First, this is a retrospective study; thus, it is difficult to perform a strictly case-matched comparison. As it is a fusion technique, anterior-posterior fusion can lead to more reliable fusion outcomes than can PF. It would have been better to use the anterior-posterior fusion technique in a control group to determine the superiority of the GR technique in treating dystrophic EOS in patients with NF-1. Second, the follow-up duration was not long enough to observe the effects of different techniques on EOS patient outcomes and allow a comprehensive analysis of the cases in these children. Third, it can be argued that PF is a complete therapy, while GR is an incomplete therapy because of the need for repeated surgeries. For our study, we assessed the outcomes from the time of the initial surgery, as we intended to focus on the initial surgical treatment choice. If the efficacy of the magnetically controlled GR technique is established, it is worthwhile to compare it with PF. We did not have direct pulmonary assessment to measure pulmonary function data for all the patients, and the use of T1-S1 gain as a proxy for pulmonary function improvement is only shown as a reference but was not directly indicated. And the sample size in our study was small, and it was similar to that in the study by Akbarnia et al. [17] the statistical power of our study is relatively low because EOS with NF-1 is a relatively rare condition, and it is difficult to include a large number of patients. If possible, a multi-centre study with a larger sample size should be conducted to further investigate the problem and help us understand more about this condition.

Overall, after comparing the abovementioned parameters and clinical outcomes between the PF group and the GR group, we concluded that GRs and PF are both suitable options for treatment for dystrophic EOS in patients with NF-1, and the GR system, which allows more trunk growth, may be more appropriate for the initial treatment. More studies with direct measurement of pulmonary function must be conducted to determine the effect of GR on pulmonary development. More studies with larger sample sizes and longer follow-up periods are needed to fully assess the treatment strategies.