Techniques of spinal arthrodesis after deformity correction have been undergoing constant evolution over the last few decades. The fusion rates after COSS varies up to 30% [8, 11, 12]. It was proposed that posterior decortication of the lamina and harvesting of the spinous process enhances fusion rates along with segmental fixation [8]. The facet fusion technique on its own has only been studied by a few authors [17]. Moe et al. observed that radiological pseudarthrosis occurred in 23% of patients after articular facet fusion without instrumentation [18]. With newer instrumentation systems, higher rates of fusion are obtained irrespective of graft material and preservation of midline spinous processes [6, 9]. Recently, Yeh et al. observed that the rates of pseudarthrosis were similar among patients in whom spinous processes were harvested (5.4%) and those in whom the spinous processes were preserved (5.1%) at the end of 24 months [19]. This group observed that patients with preserved spinous processes (n = 61) had better pain scores compared with patients (n = 43) who had spinous processes harvested. The group concluded that using spinous processes as a source of local autologous bone graft material is not necessary. The pain scores were better when spinous processes were preserved. Twenty-six percent (16/61) of patients in the spinous process harvesting group required pain medications, while only 9% (4/43) of patients with preserved spinous processes required pain medications. In our study, even though there was no statistically significant difference between MISS and COSS groups, the fusion rate was better in the MISS group. In addition, through the SRS-22 questionnaire, we confirmed that the results of MISS surgery were higher in satisfaction than those of COSS surgery. In view of the results of our study and the surgical results of Yeh et al., facet fusion using pedicle screws in scoliosis surgery could induce a “potential” fusion grade that, is sufficient for fusion in immature AIS patients.
MIS techniques have been successfully adopted for adult deformity correction and, sporadically, for pediatric deformity correction [20]. The perioperative advantages of minimal blood loss, smaller incision, lower infection rate, and faster recovery have encouraged increasing number of surgeons to adopt MISS techniques [16, 17, 20, 21]. In our previous study, MISS techniques have a merit for hospital stays compared to COSS (12.0 days in MISS vs 16.2 days in COSS, p < 0.001) but MISS requires longer operative times (441 min in MISS vs 287 min in COSS, p < 0.001) [21]. However, the fusion rates using different graft materials in spine surgery have not been widely studied. The fate and strength of fusion are important to know as the bed for fusion is provided by the facets and not by spinous process and lamina.
In a previous study, facet fusion was studied in the context of MISS, and no case of pseudarthrosis was reported. However, the number of cases was small (n = 7) and rhBMP-2 was used for fusion [16]. We believe that rhBMP-2 has frequent side effects; moreover, there are no data supporting its long-term safety. Hence, rhBMP-2 should not be used in adolescents. The selection of appropriate graft materials for MISS remains a major concern. We wanted to avoid iliac crest grafting for several reasons. Donor site pain, bleeding, neurologic injury, hernia, fracture, and blood loss are common after autologous iliac crest harvesting, with reported rates ranging between 19 and 31% [22]. Skaggs et al. observed that posterior iliac crest bone grafting in spine surgery in 87 children was associated with significant donor site pain in 24% of patients and limitation of daily activities in 15% of patients even 4 years after surgery. Freeze-dried corticocancellous allografts were used for COSS and some cases of MISS [23]. Higher rates of fusion equal to those of autologous grafts have been reported by several authors and has shown no donor site complications [24, 25]. In our study, we observed pseudarthrosis rates of 16.3 and 3% after COSS and MISS, respectively. Despite no statistical differences of pseudarthrosis rate, COSS performs posterior fusion that requires large amount of graft volume and large area decortication compared to facet fusion in MISS. The facet fusion only needs to small amount of graft volume in the facet area, which contribute to sufficient fusion rate.
Knapp et al., in their study of 111 AIS patients using various types of instrumentation techniques along with allografts, observed a mean loss of correction of 5.9% and a pseudarthrosis rate of 2.7% [9]. Theologis et al., in their study of over 461 patients, found no cases of pseudarthrosis out of 199 patients with allograft after 2 years of follow-up [12]. Price et al. found a significantly higher failure rate of 28% with allografts compared to 13% with the use of iliac crest bone grafts and 11.1% with composite grafts in AIS surgery [11]. One of notable results by Price et al. was that pseudarthrosis was not significantly affect pain and implant-related complications. In our study, pain in clinical outcomes, correction loss, and complications was not statistical differences between facet fusion in MISS and posterior fusion in COSS. Therefore, facet fusion in MISS also provide comparable postoperative outcomes as much as posterior fusion in COSS.
Another concern is the high infection rates associated with allografts. All four infections observed in our study were associated with the use of allografts. Reported incidences of bacterial infection after all forms of allograft range from 4 to 13% [25]. However, the risk depends on the patient’s primary condition, operation time, and associated skin complications [26]. There is a potential risk of disease transmission with allografts, but this risk is exceedingly low. As reported by Asselmeier et al., no such cases of disease transmission have been observed since 1951 [27]. The risk of transmission with freeze-dried allografts is much less compared with that associated with fresh-frozen allografts [27]. Hence, we prefer to use freeze-dried corticocancellous allografts. From our study, the preliminary outcomes suggest that selection of facet fusion and of graft material produce similar results to COSS. However, the rate of infection was higher with the use of allografts.
Several graft substitutes have been studied, and many have shown encouraging results [9]. We wanted to analyze a graft material that possessed both osteoinductive and osteoconductive properties in order to enhance the rates of fusion using as little graft material as possible. Demineralized matrix has mainly osteoinductive properties, with some products having osteoconductive properties, as determined by the methods of preparation and sterilization [28]. DBM is an organic derivative of allograft used in surgery for AIS [6, 12]. DBM can be processed as granules, powder, or chips from human cortical or corticocancellous bone. Terminal sterilization with irradiation, ethylene oxide, glutaraldehyde, and formaldehyde has been shown to reduce the osteoinductive properties of DBM [6]. We used two types of demineralized bone: a matrix form and a cancellous chip form. The selection of two different forms was inspired by variation in rates of fusion observed in studies due to the manner in which DBM was prepared and the carrier with which DBM was combined [3].
All mechanical failures of COSS were associated with a failure of formation of a solid fusion mass at the distal end of the construct. Several authors have reported the distal extent of fusion to be troublesome. Knapp et al. observed three cases of pseudarthrosis in his series of 111 patients and noted that two of three patients had dislodgement of the distal two hooks at intervals of 4 months and 1 year after surgery, respectively [9]. Yeh et al. had one patient who developed bilateral L4 screw loosening at the end of 24 months and another patient with L3 screw breakage at the end of 19 months [19]. In the group with preserving spinous process, one patient had right L1 screw breakage at the end of 50 months, and another had left L4 screw cap loosening at the end of 27 months [19]. Betz et al. observed one case of pseudarthrosis, 12 months after surgery, with distal screw breakage at L4 and caudal fusion mass formation [6]. However, most rod breakage occurs in the first 2 to 3 years after implantation [9]. Investigations of the long-term outcomes of the newer techniques are needed for further comparison.
We also analyzed the amount of graft material used in different groups. Since different graft materials were used, a statistical comparison was inappropriate, but the amount of graft substance used for MISS was much less than that used for COSS.
Our study had a few limitations. Radiographs were used to evaluate fusion for most of the patients. Radiographs, including bending films, are helpful to rule out instability but have their own limitations. An interobserver agreement rate of 100% with the fusion criteria has been reported [10]. Even though the main limitation is variability between observers, fusion tends to be over-reported (false negative for non-unions) overall. CT is currently the investigation method of choice to confirm or rule out pseudarthrosis. Radiation exposure after a single thoracolumbar CT scan is 10 to 40 times greater than that of a standing whole-spine x-ray, and this increases cancer risk by 0.32 to 0.52% per CT scan. Subjecting asymptomatic patients to such amounts of radiation would be unethical. CT scan should be considered in symptomatic patients (those with axial back pain or radiological signs of pseudarthrosis), however, for confirmation and preoperative planning of revision surgeries. Furthermore, we instrumented all levels in the fusion mass. Hence, we had higher implant density indices (> 2) with better curve correction, and the loss of correction over time should have been minimal. However, since MISS is a novel technique associated with patient satisfaction, and the fusion rates with this technique have not yet been determined, we will continue to instrument all levels until long-term outcomes have been established. Another limitation was the relatively small number of patients; we did not perform a power analysis to determine the number of cases to be included in each group and subgroup before the study. Finally, our follow-up should be considered a preliminary analysis as pseudoarthrosis has been reported to develop as late as 6 years after scoliosis surgery.
