Maintaining adequate sagittal alignment and achieving solid fusion are crucial in adult spinal deformity (ASD) surgery, and robust internal fixation of the spinal column is essential to achieve these goals [1,2]. When posterior fixation extends to the sacrum, pelvic fixation coupled with anterior column support at L5–S1 is highly recommended to minimize mechanical complications and promote fusion at the lumbosacral junction [3,4]. Among various pelvic fixation methods, traditional iliac screws and S2 alar iliac (S2AI) screws are the most commonly used approaches [5]. Although some recent studies suggest that S2AI screws may be preferable to iliac screws [5–7], traditional iliac screws remain widely used due to their favorable biomechanical strength, ease of insertion and rod assembly, and preservation of the sacroiliac (SI) joint [8–10]. As the SI joint is not fused by the pelvic fixation, some motion is retained at this joint. Repetitive cyclical loading of iliac fixation can increase stresses at the SI joint, potentially leading to iliac screw failure (ISF).

Despite efforts to identify risk factors for ISF [2,8,11–14], this complication cannot be completely prevented, regardless of the fixation type, due to persistent cyclical loading imposed postoperatively at the SI joint. Therefore, ISF remains a significant concern with reported rates ranging from 12% to 57% [2,5,8,11–13,15–17]. Given that the primary goal of pelvic fixation is to enhance fusion and prevent mechanical failures at the L5–S1 segment by augmenting the weak S1 pedicle screws [18], it is reasonable to infer that ISF may lead to pseudoarthrosis and related fixation failure at the lumbosacral junction. Although a few studies have examined the relationship between ISF and lumbosacral pseudoarthrosis, their findings may be limited by several factors. These include the examination of a mixture of posterolateral and interbody fusion at L5–S1 [5,11], a lack of data on the time course of ISF and mechanical failure [11], and reliance on plain radiographs for fusion evaluation [5,13].

Therefore, we investigated the impact of ISF on fusion at the lumbosacral junction in patients undergoing long-segmental fusion coupled with L5–S1 interbody fusion for ASD. Computed tomography (CT) images obtained at 2-year post-surgery were used to evaluate the fusion status at L5–S1. Additionally, we assessed the effects of ISF on mechanical failures, as well as long-term radiographic and clinical outcomes. We also examined the temporal relationship between ISF and mechanical failure to better understand how ISF contributes to the development of mechanical failure.

This study was approved by This study was approved by the Institutional Review Board (IRB) of Samsung Medical Center (IRB no., 2024-07-144). The requirement for informed consent was waived owing to the retrospective nature of this study.

A total of 192 patients were included in this study. The mean age, T-score, and BMI were 69.2 years, −1.40, and 25.9 kg/m², respectively (Table 1). Revision cases accounted for 25.0% of the total, and none of these patients had a previous lumbosacral fusion. Regarding the fusion methods at L5–S1, ALIF and TLIF were performed for 106 (55.2%) and 86 (44.8%) patients, respectively. LLIF and PSO were performed in 138 (71.9%) and 16 (13.8%) patients, respectively. The mean fusion length was 7.9 levels, including the pelvis. The mean follow-up duration was 57.8 months. Postoperative sagittal parameters, including LL, PI–LL, SS, PT, TK, TPA, and C7–SVA, showed significant improvement, except for PI (Table 1).

Sixty-six patients (34.4%) developed bilateral ISFs (ISF group) (Table 2). The mean time to ISF development was 13.4 months. Among the ISF types, iliac screw loosening was the most common (80.3%), followed by rod fracture between the S1 pedicle screws and iliac screws (15.2%) and iliac screw fracture (4.5%).

Evaluation of the fusion status using BSF criteria based on 2-year CT images revealed that 10 patients had grade 1, 47 had grade 2, and 91 had grade 3 fusion (Table 3). At the time of CT acquisition, 64 patients had ISF. No significant association was noted between the presence of ISF and fusion grades (p=0.174). Specifically, in the no-ISF group, four patients (3.1%) had grade 1, 33 (25.8%) had grade 2, and 91 (71.1%) had grade 3 fusion. In the ISF group, six patients (9.4%) had grade 1, 14 (21.9%) had grade 2, and 44 (68.8%) had grade 3 fusion.

Mechanical failure rates were compared according to the development of ISF. No significant differences were found in terms of rod fracture at L5–S1 (p=0.273), revision surgery at L5–S1 (p=0.144), rod fracture at ≥L4–5 (p=0.432), revision surgery at ≥L4–5 (p=0.886), and PJF (p=0.700) (Table 4). However, a significant difference was observed in PJK rates according to ISF development. Specifically, PJK developed in 57 patients (40.7%) without ISF, whereas only seven patients (13.5%) with ISF developed PJK (p<0.001).

In the comparison of radiographic results, no significant differences in any preoperative sagittal parameters were noted between the ISF and no-ISF groups (Table 5). Similarly, at 6 weeks postoperatively, no significant between-group differences were observed regarding any of the sagittal parameters. However, at the final follow-up, the ISF group had a significantly greater PI than the no-ISF group (57.1° vs. 52.8°, p=0.034). Additionally, the ISF group showed greater PI–LL mismatch, PT, and TPA than the no-ISF group (15.9° vs. 11.3°, p=0.029 for PI–LL; 25.3° vs. 21.5°, p=0.046 for TK; and 23.5° vs. 20.2°, p=0.042 for TPA). Although not statistically significant, the ISF group tended to have a greater C7–SVA than the no-ISF group (43.2 mm vs. 36.2 mm, p=0.061).

A comparison of preoperative PROMs revealed no significant between-group differences in terms of VAS scores for back pain, ODI, and SRS-22 score (Table 6). However, at the last follow-up, the ISF group demonstrated significantly poorer clinical outcomes compared to the no-ISF group, including higher VAS scores for back pain (42.9 vs. 33.5, p=0.021). Additionally, the ISF group had lower scores in several SRS-22 subdomains, including pain (3.43 vs. 3.78, p=0.020), appearance (3.09 vs. 3.49, p=0.020), mental health (3.13 vs. 3.55, p=0.015), and the total SRS-22 score (3.22 vs. 3.53, p=0.023).

The lumbosacral junction remains a high-risk region for pseudoarthrosis or fixation failure after ASD surgery. This vulnerability is attributed to the unique anatomic features of this region, including the poor sacral bone quality, complex sacral anatomy, and high biomechanical stresses at the lumbosacral junction [21,22]. To mitigate these risks, surgeons commonly employ pelvic fixation in conjunction with bicortical S1 screws and anterior column support at L5–S1. This approach has been widely adopted for long-segment fusion surgeries that extends to the sacrum, aiming to reduce mechanical complications at the lumbosacral junction [3,4,23]. The increasing use of pelvic fixation in ASD surgery has led to a rise in the incidence of ISF, with reported rates ranging from 12% to 57% [2,5,8,11–13,15–17]. In the present study, the incidence of bilateral ISF was 34.4%, which is comparable to previously reported results. Notably, when including the 18 patients with unilateral ISF who were excluded from our analysis, the overall incidence of ISF was 43.8%.

Pelvic fixation is primarily used to reduce stress on S1 screws and promote fusion at the lumbosacral joint. Therefore, it can be assumed that ISF may hinder the achievement of solid fusion at L5–S1. Although numerous studies have investigated the risk factors for ISF [5–7,11,14], few have examined the effects of ISF on fusion at L5–S1. Eastlack et al. [5] evaluated the rates of ISF and its impact on surgical outcomes, including pseudoarthrosis at L5–S1. They found that 10.8% and 24.5% of patients with traditional iliac screws and S2AI screws, respectively, developed loosening or fracture of pelvic fixation devices. They demonstrated that the pseudoarthrosis rate at L5–S1 was significantly greater in the pelvic fixation failure group than in the no-failure group (0.5% vs. 25.8, p<0.001) in the overall study cohort. However, their study included a considerable number of patients undergoing only posterior fusion at L5–S1, rather than interbody fusion. Moreover, their fusion analysis relied solely on plain radiographs. Nguyen et al. reported a lower rate of 7.7% (20 of 260 patients) and an L5–S1 pseudoarthrosis of 8.8% (23 patients), defined as the need for revision surgery [8]. They emphasized that solid fusion at L5–S1 was visible on plain radiographs in 66.7% and 55.6% of patients, even with iliac screw fracture or loosening, respectively. However, the temporal relationship between pseudoarthrosis or solid fusion at L5–S1 and ISF remained unclear, as the study did not evaluate the time course of radiographic results and ISF.

Therefore, we evaluated fusion at L5–S1 using CT scans, taking into account the timing of radiographic examinations and ISF detection. At the 2-year follow-up CT examination, failed iliac screws were identified in 64 (33.3%) of 192 patients, indicating that most ISFs (97.0%, 64/66) occurred within the first 2 years postoperatively. We found no significant differences in fusion status on 2-year CT images between the ISF and no-ISF groups (Table 3). Solid fusion (grade 3) was achieved in 68.8% of patients with ISF (Fig. 3), which was comparable to the 71.1% rate observed in patients without ISF (Fig. 4). Furthermore, 18 patients experienced rod fractures at L5–S1, and four patients required revision surgery related to this complication (Table 4). Similar to the CT results, no significant differences in the rates of these mechanical failures were observed in relation to the presence of ISF. These findings suggest that ISF did not increase the risk of radiographic and clinical pseudoarthrosis at L5–S1. Although our results may contradict the assumption that ISF can interfere with the solid fusion at L5–S1, the lack of effect of ISF on fusion at L5–S1 can be attributed to the difference between the time to ISF development and the time required for fusion. Bony arthrodesis following interbody fusion is reportedly complete within 6–12 months postoperatively [19,24,25]. Given that ISF developed at a mean of 13.4 months in the present study, it can be inferred that fusion was likely completed in most cases before ISF occurred, although the exact timing of fusion was not evaluated by CT.

We also examined mechanical failures other than at L5–S1. Similar to the mechanical failure at L5–S1, the rates of rod fracture, subsequent revision surgery at ≥L4–5 levels, and PJF did not differ significantly in relation the presence of ISF (Table 4). However, the PJK group had a significantly smaller number of patients with ISF compared to the no-PJK group (p<0.001). This finding suggests that ISF may have a preventive effect against PJK development. Previous studies have reported similar results, indicating that increased distal stability provided by secure pelvic fixation can reciprocally increase stress in the proximal junction, leading to PJK development, and vice versa [12,26,27].

Although ISF did not negatively impact L5–S1 fusion, the final radiographic and clinical outcomes were significantly poorer in patients with ISF compared to those without ISF. Regarding the final radiographic results, sagittal parameters, including PT and TPA, were significantly greater in the ISF than in the no-ISF group (Table 5). When iliac screws are securely fixed to the pelvis, the spinal column and pelvis move as a single unit. However, bilateral ISFs cause dissociation between the spinal column and pelvis. As ISF develops, the fused spinal column tends to shift anteriorly due to gravity coupled with the loss of secure pelvic fixation. To compensate for this sagittal decompensation, the pelvis rotates posteriorly using hip extensor muscles, resulting in increased PT and TPA (Fig. 3). This posterior pelvic rotation (represented by increased PT) compensates for the increased C7–SVA. Consequently, the final C7–SVA was not significantly different between the groups, although it was greater in the ISF than in the no-ISF group. Notably, the PI was significantly greater in the ISF than in the no-ISF group (57.1° vs 52.8°, p=0.034), despite no preoperative or 6-week postoperative differences between the groups. The SI joint normally allows for rotational movement with up to 6° of freedom [28,29]. The two opposing forces at the SI joint—anterior shifting of the fused spinal column and posterior rotation of the pelvis—induce a twisting motion, known as nutation. The increased PI in the ISF can be attributed to the nutating motion of the SI joint, as nutation has been reported to increase PI [14]. Additionally, we observed no significant differences in the final LL between the no-ISF and ISF groups (41.5° vs. 41.2°, p=0.887). Given the comparable final LL between the groups and the increased PI in the ISF group, PI–LL mismatch at the final follow-up was bound to be significantly greater in the ISF than in the no-ISF group (15.9° vs. 11.3°, p=0.029).

The final clinical outcomes were significantly poorer in the ISF group as compared to the no-ISF group in terms of back pain VAS score, the pain, appearance, and mental health subdomains, as well as the total SRS-22 score. These inferior clinical outcomes in the ISF group can be attributed to the worse sagittal parameters, including greater PI–LL, PT, and TPA, observed in the ISF group.

Some limitations of this study should be considered while interpreting the results. This was a single-center retrospective study, which may have introduced an element of bias. Additionally, we only used traditional iliac screws, not S2AI screws, which may further limit the generalizability of our results. However, the relative merits of each pelvic fixation method are debated in the literature. While some studies suggest that S2AI screws reduce mechanical complications more effectively than traditional iliac screws [5–7], others advocate for the use of traditional iliac screws over the S2AI screws due to their ease of insertion, rod assembly, and lower overall complication rates, including screw loosening [2,5,8–10]. Therefore, it remains unclear which method is superior. The choice between traditional iliac screws and S2AI screws is currently based largely on the surgeon’s skill and preference. Lastly, since ISF is believed to be related to SI joint motion, the preoperative status of the SI joint can potentially influence the development of ISF. Although a slight motion of SI joint motion exists in normal individuals, it is generally considered negligible. Moreover, after thoracolumbar long fusion including the pelvis, the load on the SI joint substantially increases, potentially leading to greater stress on the SI joint. Consequently, the biomechanics of the SI joints are expected to differ before and after surgery. Therefore, we assumed that the impact of preoperative SI joint mobility is likely to be minimal.

Pelvic fixation plays a vital role in preventing mechanical failure at the lumbosacral junction in long-segment fusion for ASD. However, unique to pelvic fixation, delayed failures may be inevitable due to the ongoing motion at the SI joint postoperatively, regardless of the type of fixation used. Based on our understanding of the nature and clinical impact of these failures, further research is warranted to develop preventive strategies or solutions that enhance the durability of the index surgery.

A substantial proportion of patients (34.4%) in this study developed bilateral ISF after long-segment fusion for ASD. Notably, ISF did not adversely affect radiographic and clinical fusion at the lumbosacral junction. However, ISF was associated with inferior long-term radiographic and clinical outcomes.