Introduction

As the proportion of aging adults increases, the incidence of adult spinal deformity (ASD) progressively rises [1]. Compared with medical management, surgical intervention displays an enormous advantage in improving the quality of life of patients [1]. The main objective of ASD surgery is to achieve physiological three-dimensional alignment, especially in the sagittal plane [2,3,4,5]. However, ASD surgery has been also drawing increasing attention due to its host of various post-operative complications, especially mechanical complications (MCs) [6, 7].

MCs are a series of complications that primarily occur in implants and adjacent uninstrumented segments due to abnormal biomechanical distribution of loads, including rod fracture (RF), screw breakage (SB), screw dislodgement (SD) and proximal or distal junctional kyphosis (PJK or DJK) [7]. According to the reports in the relevant literature, the prevalence of MCs after ASD surgery is approximately 20–50% [6, 7]. In addition to impaired health that is often referred to in prior papers [7, 8], MCs are still serious economic burdens on patients [9]; Williamson et al. [9] highlighted that ASD patients suffering MCs bore the highest overall cost ($130,482.22) in comparison to other complications. Sagittal malalignment has been also proved to be strongly related to the occurrence and progression of MCs [3, 6]. Prior studies usually viewed MCs as a single entity and overlooked the differences among different subtypes [3, 6, 10]; to our knowledge, a study that has comprehensively investigated the sagittal characteristics of each MC subtype has not yet been performed.

The main objectives of this paper were to explore the sagittal alignment associated with different subtypes of MCs at different time points following ASD surgery and to assess the correlations of each MC subtype with the global alignment and proportion (GAP) score [3], in order to understand the pathological alignment and evolution process of various MCs, which could provide a more adapted revision scheme.

Material and methods

Patient population

This retrospective analysis of a prospective multicentre ASD database was conducted with the dates of surgeries between 2009 and 2020, and patients older than 18 years who underwent an index correction procedure with posterior instrumented spinal fusion (≥ 5 levels) and sacral or pelvic fixation were enrolled. The diagnosis of an ASD was based on meeting at least one of the following criteria: a coronal Cobb angle ≥ 20°, a pelvic tilt (PT) ≥ 25°, a sagittal vertical axis (SVA) ≥ 5 cm or thoracic kyphosis (TK) ≥ 60° [4]. The terminal event was defined as either the first reoperation for the treatment of MCs or a minimum of 2 years of follow-up. Demographic data, including age, sex, height, weight and body mass index (BMI) at the time of surgery, and operative information, including the upper and lower instrumented vertebrae (UIV and LIV) and number of fusion levels, were collected. Informed consent was obtained from each subject, and the Institutional Review Board approved the protocol at their respective sites.

Radiographic measurements

Standing full-spine X-ray examination was conducted for all participants preoperatively and at the 6-month and last follow-up after ASD surgery. The following parameters were measured in the sagittal plane: TK (T2–T12), thoracolumbar kyphosis (TLK, T10-–L2), lumbar lordosis (LL, L1–S1), upper LL (ULL, L1–L4), lower LL (LLL, L4–S1), lordosis distribution index (LDI, LLL divided by LL), pelvic incidence (PI), PI minus LL (PI-LL), PT, sacral slope (SS), C7-SVA and global tilt (GT) [2]; of note, positive values denoted kyphosis, while negative values denoted lordosis. Since Pesenti et al. [11]. reported that PI accounted for 43% of the ULL (R2 = 0.879, P < 0.001), the difference between the measured and ideal ULL (0.43 \(\times\) PI) was also recorded to reflect the relative ULL (RULL). In addition, according to the sagittal parameters at the last follow-up, the GAP score and its components, the relative pelvic version (RPV), relative lumbar lordosis (RLL), LDI, relative spinopelvic alignment (RSA) and age subscores, were computed to assess the extent of sagittal malalignment [3]. Meanwhile, health-related quality of life (HRQOL) scores, such as the visual analogue scale (VAS)-back score and the Oswestry Disability Index (ODI), were also documented.

Mechanical complications

According to the ultimate outcomes, the subjects were divided into a non-MC or MCs group, and the MCs group was composed of four subgroups, the RF, SB, SD and PJK groups; in each subgroup, there was a single type of major MC (the primary reason of patient complaint or revision surgery). Pseudarthrosis can result in intervertebral instability, which is one of important reasons of implant breakage; in this study, pseudarthrosis and RF were regarded as one composite category on account of a stronger association of pseudarthrosis with RF than SB [12,13,14]. The SD group involved screw loosening or pullout and hook dislodgement. PJK was defined as a kyphosis increase > 10° between the UIV and two levels above [15], and proximal junctional failure (PJF) [16] also belonged to the PJK group; DJK was not explored, given that it occurred less frequently when the LIV was chosen at the sacrum or ilium [17]. Moreover, the reoperation rate of each MC subtype was separately registered.

Statistical analysis

The data were analysed using SPSS 17.0 statistical software (SPSS Inc., Chicago, IL, USA). Descriptive statistics were expressed as the mean and standard deviation. The t or Mann–Whitney test (numerical variables) and χ2 or Fisher’s exact test (qualitative variables) were used to compare the differences between the non-MC and MCs groups. Subsequently, binary logistic regression was fit to examine the correlations and odds ratios (ORs) between the GAP subscores and diverse MCs. Statistical significance was defined as a P value < 0.05.

Results

A series of 371 patients (F/M: 4.9/1) who met the inclusion criteria were extracted from the dataset. The average age was 64.5 ± 10.5 (range 22–83) years at the time of operation, and the mean follow-up period was 47.9 ± 22.2 (range 18–103) months. In this cohort, 187 patients (50.4% of all patients) sustained post-operative MCs, including 106 patients (28.6%) with RF, 24 (6.5%) with SB, 18 (4.9%) with SD and 39 (10.5%) with PJK.

At baseline, there were no substantial differences in age, sex ratio or height between the non-MC and MCs groups except the PJK group had an older age (68.0 ± 6.6 years). The total MCs (69.7 ± 12.6 kg) and SB (73.4 ± 13.6 kg) groups had a larger weight, and the total MCs (27.1 ± 4.7 kg/m2), RF (27.3 ± 4.9 kg/m2) and SB (27.8 ± 4.9 kg/m2) groups had a higher BMI. Additionally, all the MCs groups had few discrepancies in their VAS-back and ODI scores. In terms of pre-operative sagittal alignment, TLK and LLL were greater in the total MCs (13.5 ± 21.0° and − 32.1 ± 15.0°) and RF (13.6 ± 21.9° and − 32.3 ± 15.5°) groups, and a comparatively flat ULL or larger RULL was in the total MCs (− 0.3 ± 17.8° or 24.7 ± 16.4°) and RF (0.6 ± 18.8° or 25.9 ± 18.1°) groups. In addition, the SB group had the greatest LL (− 42.1 ± 18.0°) and PI (63.6 ± 16.8°); the SD group had the least SVA (40.3 ± 41.8 mm) and GT (28.3 ± 14.9°); and the PJK group had the maximum TK (40.6 ± 16.2°), TLK (18.5 ± 16.6°), ULL (1.8 ± 15.0°) and RULL (26.0 ± 12.8°).

Regarding the fusion levels, the mean UIV was higher in the total MCs (approximately T9) and RF (approximately T8) groups and lower in the SD (approximately T11) group than in the non-MC group (approximately T10), and there was no significant difference in the LIV between the non-MC and MCs groups. Furthermore, the average fusion levels were longer in the total MCs (11.2 ± 3.3) and RF (11.7 ± 3.5) groups and shorter in the SD (9.0 ± 1.3) group than in the non-MC group (10.3 ± 3.6).

At 6 months after the operation, the VAS-back score was higher in the total MCs (3.7 ± 2.7), RF (3.9 ± 2.7) and PJK (4.0 ± 2.7) groups, and the ODI was larger in the PJK (40.1 ± 15.9) group. Both total MCs and SB groups had greater PT (25.2 ± 8.1° and 27.7 ± 8.5°) and GT (28.5 ± 11.8° and 31.6 ± 11.4°), while the PJK group had greater TK (58.8 ± 17.0°), TLK (18.9 ± 13.6°), PT (26.5 ± 7.8°), SVA (48.6 ± 54.5 mm) and GT (31.5 ± 12.7°).

At the last follow-up, back pain was more severe in the total MCs (4.9 ± 3.4), RF (5.1 ± 3.5) and PJK (5.2 ± 3.1) groups; physical dysfunction was widely aggravated in all the MCs groups. Local sagittal abnormity was principally reflected in two forms: a large PI-LL or namely a comparatively small LL, which was observed in the total MCs (11.2 ± 14.2° or − 46.8 ± 14.3°), RF (12.3 ± 14.1° or − 44.9 ± 13.5°) and SB (19.5 ± 12.4° or − 42.5 ± 14.0°) groups, and a pronounced TK, which was observed in the total MCs (55.2 ± 17.7°) and PJK (65.7 ± 17.6°) groups. In addition, all the MCs groups had a certain sagittal imbalance, and the SB group had the most severe cases, with the maximum PT (31.4 ± 7.7°), SVA (80.8 ± 59.5 mm) and GT (41.4 ± 14.3°). The reoperation rate of the total MCs group was 59.9%, which was relatively higher in the SB (75.0%) and RF (67.0%) groups, followed by the PJK (41.0%) and SD (38.9%) groups. The descriptive statistics of the above data are detailed in Table 1 and Fig. 1.

Table 1 Comparisons of clinical data between the non-mechanical complication and mechanical complications groups
Full size table
Fig. 1
figure 1

Comparisons of sagittal parameters between the non-mechanical complication (MC) and MCs groups (mean ± standard error of mean); *P < 0.05, and **P < 0.01

Full size image

The GAP score of all the MCs groups was distinctly higher than that of the non-MC group, and the highest score was in the SB group (8.9 ± 2.9); however, with regard to the GAP subscores, there were numerous discrepancies in various MCs subgroups. Logistic regressions revealed that the RSA score was statistically correlated with the total MCs (OR = 1.950, P < 0.001), RF (OR = 1.676, P = 0.007), SB (OR = 2.929, P = 0.016) and SD (OR = 3.212, P = 0.013) groups, while the RSA (OR = 2.199, P = 0.007) and age (OR = 4.758, P = 0.049) scores were statistically associated with PJK (Table 2, 3, 4, 5, 6).

Table 2 Logistic regressions for total mechanical complications based on the GAP subscores
Full size table
Table 3 Logistic regressions for rod fracture based on the GAP subscores
Full size table
Table 4 Logistic regressions for screw breakage based on the GAP subscores
Full size table
Table 5 Logistic regressions for screw dislodgement based on the GAP subscores
Full size table
Table 6 Logistic regressions for proximal junctional kyphosis based on the GAP subscores
Full size table

Discussion

In prior studies, diverse MCs were integrated to represent a single research object, and these studies reported several differences in the sagittal profiles between patients with and without MCs [3, 6]. Although some papers have also investigated the sagittal alignment of some types of MCs, such as PJK and RF [12, 15, 16], there is still an absence of relevant works summarizing the sagittal features of various MCs to date. Chanbour et al. [7] compared radiographic measurements among each MC subgroup, and they detected that PJK had higher post-operative TK than RF; nevertheless, this was only a single-institution study, and the sample size (n = 145) was comparatively low, which could beget a certain bias and may restrict the accuracy of conclusions. Therefore, multicentre and large-scale research would be more appropriate to avoid the aforementioned drawbacks.

In this cohort, a total of 371 eligible patients from multiple institutions were included to analyse the characteristics of diverse MCs. Some differences in demographic, operative and functional data should be noted. For example, BMI was generally larger in patients with MCs, illustrating that obesity was an imperative factor of MCs, which was also found in several previous studies [18, 19]. Noh et al. [19]. Realized the importance of obesity, and hence, sought to add BMI to better predict the occurrence of MCs after ASD surgery. In addition, the RF group had a higher UIV (approximately T8) and more instrumented segments, and Yasuda et al. [20]. Stated that more fusion levels could increase the probability of RF revision. Contrarily, the SD group had a lower UIV (approximately T11), which could be explained by that the upmost screws had more shear force when the UIV ending at the thoracolumbar inflection point area [21, 22]. Concerning the HRQOL scores, the VAS-back and ODI scores gradually increased in all the MCs groups during follow-up, and Bourghli et al. [8]. Indicated that a lack of improvement in the HRQOL scores portended a high revision rate after ASD surgery.

In addition, each MC had disparate characteristics in the sagittal plane at different time points. Preoperatively, the RF group had an evidently large TLK and RULL, and concomitant compensatory hyperlordosis in the lower lumbar spine. After sufficient correction of TLK and ULL, there was no aberrant sagittal alignment in the RF or non-MC group at 6 months postoperatively; nevertheless, previous papers found that a greater correction degree could cause more strain concentration at the rods, which might be an underlying reason for RF [12, 20]. At the last follow-up, RF was characterized by a reduced LL (or increased PI-LL), including ULL and LLL, which further led to sagittal malalignment, as reflected by pelvic compensatory retroversion and global sagittal imbalance (Fig. 2); on the other hand, post-operative sagittal malalignment altered the normal biomechanical distribution and then resulted in rod fatigue as well as a high likelihood of RF [12, 20].

Fig. 2
figure 2

A 50-year-old female. a Preoperatively, she had marked kyphosis in the thoracolumbar junctional (12°) and upper lumbar (23°) spine. b 6 months after correction surgery (T11-pelvis). c 2 years postoperatively, two rods fracture at the L5/S1, with a relatively small lumbar lordosis (LL, − 28°), large mismatch between pelvic incidence and LL (16°) and global sagittal imbalance (global tilt, 29°); the global alignment and proportion score was 8

Full size image

Although screw-related issues were less frequently mentioned in the published literature, the pre- and post-operative sagittal alignment had some special features. For instance, the SB group had the largest pre-operative PI and LL, confirming the close interactions between both parameters. After surgical treatment, the restoration of sagittal alignment remained unsatisfactory, represented by the highest GT and PT. At the last follow-up, the SB group featured a reduced LL (or increased PI-LL), but mainly LLL, which was not exactly similar to the featured findings in the RF group. The magnitude of sagittal imbalance was the worst among the four subgroups, as demonstrated by the largest PT, GT and SVA (Fig. 3); it is speculated that severe sagittal imbalance may create more mechanical loads on the pedicle screws and exacerbate the odds of SB [23]. In addition, the SD group had the least sagittal imbalance with the minimum SVA and GT preoperatively, and there were few differences compared with the non-MC group in the post-operative sagittal alignment except a forward truncal inclination at the last follow-up (Fig. 4).

Fig. 3
figure 3

A 76-year-old female. a Preoperatively, she had a large pelvic incidence (PI, 85°) and lumbar lordosis (LL, − 39°). b 6 months after correction surgery (T8-pelvis). c 4 years postoperatively, right iliac screw breakage, with a relatively small LL (− 44°), large mismatch between PI and LL (40°) and global sagittal imbalance (global tilt, 53°); the global alignment and proportion score was 10

Full size image
Fig. 4
figure 4

A 68-year-old female. a Preoperatively, she had small sagittal vertical axis (SVA, − 16 mm) and global tilt (13°) b 6 months after correction surgery (T12-pelvis). c 5 years postoperatively, T12 pedicle screws pullout, with a large SVA (86 mm); the global alignment and proportion score was 8

Full size image

PJK is a type of common MC that occurs after ASD surgery, and it has been widely researched in the literature as well [15, 16, 24, 25]. In the present study, the authors observed that the PJK group had the largest TK, TLK, ULL and RULL preoperatively, which was likely explained by two reasons. First, poor back musculature made it difficult to maintain sufficient extension of the thoracic spine, and Jeon et al. [24]. found that a smaller quantity (OR = 1.03) and more severe degeneration (OR = 1.46) of the paraspinal muscles increased the incidence of post-operative PJK; second, stiffness of the thoracic spine can decrease back extension, and Decker et al. [25]. mentioned that less TK flexibility could predispose patients to PJK after ASD surgery. At 6 months postoperatively, the kyphotic angle of the thoracic and thoracolumbar junctional spine was further aggravated, which induced the appearance of pelvic compensatory retroversion and global sagittal imbalance. At the last follow-up, the residual sagittal malalignment continued to deteriorate (Fig. 5).

Fig. 5
figure 5

A 73-year-old female. a Preoperatively, she had a relatively large thoracic kyphosis (TK, 46°). b 6 months after correction surgery (T11-pelvis). c 2 years postoperatively, proximal junctional kyphosis occurred with a large TK (66°) and global sagittal imbalance (global tilt, 36°); the global alignment and proportion score was 8

Full size image

Furthermore, the GAP score, proposed by Yilgor and colleagues [3], was utilized as an instrument to quantify sagittal malalignment. In this study, the total GAP score was distinctly elevated across all the MCs groups, suggesting that all patients with MCs had a certain sagittal imbalance, and the SB group had the highest score, which again certified that the sagittal imbalance of the SB group was the most severe. However, regarding the GAP subscores, there were many differences among several MCs subgroups. To further determine the reciprocal relationships of the GAP subscores or namely the specific aspects of sagittal malalignment with each MC, which has never been done before in the literature, logistic regressions were applied in the current study. The RSA score, meaning the deviation between the measured and estimated GT values [3], was significantly associated with all MCs. The results suggested that global sagittal imbalance was the main characteristic of MCs, and progressive deterioration of spinopelvic alignment, especially abnormal changes in GT parameters, might imply the emergence of MCs during the follow-up period. Additionally, an advanced age was a risk factor for PJK because geriatric patients usually have osteoporosis, paraspinal muscle degeneration and body compensatory capacity decline, and these pathological factors evidently contribute to PJK development [24].

In this study, the authors explored the clinical data of patients with diverse MCs, including their demographic information, functional scores, operative factors, reoperation rate and sagittal parameters, to better understand MCs and fill the gaps in prior relevant papers. Notably, each MC had its own characteristics of sagittal alignment at different time points, thereby providing a novel perspective to recognize the pathological alignment and evolution process of diverse MCs; meanwhile, the findings can also be conducive to tailoring a revision proposal to restore normal sagittal alignment. For example, for patients with RF and SB who require reoperation, an emphasis should be placed on LL reconstruction; specifically, the objective of RF revision is to augment both ULL and LLL, while the objective of SB revision should be enlarging LLL. For patients with PJK, revision surgery requires fusion extending to the upper thoracic spine to treat prominent TK. In addition, deterioration of sagittal spinopelvic alignment, particularly unexpected variation in GT parameters, was the common manifestation of MCs, and should therefore be closely monitored as an important predictor of MCs during the follow-up period; on the other hand, restoration of global sagittal morphology should become a key point when performing revision surgery [2].

Some limitations must be discussed as well. First, only VAS and ODI scores were employed due to limited space, and other HRQOL scores, such as SRS-22 and Short-Form Health Survey (SF)-36, will be necessary to ensure a more holistic evaluation of clinical outcomes in future research. Second, although this was a multicentre study, the sample of patients with MCs, such as the SD group (n = 18), was comparatively small; so a larger sample size is needed, and more characteristics of these infrequent MCs will need to be defined in future studies. Third, this study did not distinguish the patients instrumented with 2-rod or multiple-rod constructs, which might become a certain deficiency to some degree, although Bourghli et al. [26]. Reported that multiple-rod constructs cannot reduce pseudarthrosis and RF after pedicle subtraction osteotomy for ASD correction; in future, the relationship between rod fracture and sagittal alignment will be respectively explored in patients with 2-rod or multiple-rod constructs. Forth, the sagittal alignment assessment did not include correlations with the different types of the Roussouly classification [5]; however, the use of the GAP score enabled an accurate analysis tailored to each individual according to their specific PI. Finally, this paper provided a more exhaustive description of the sagittal alignment of disparate MCs, and the potential mechanisms need to be further uncovered via biomechanical experiments or finite element analysis in future studies.

Conclusion

In conclusion, compared with the non-MC group, each MC group possessed its own characteristics of sagittal alignment at different time points following ASD surgery. These differences can be helpful to comprehend the pathological alignment and evolution course of diverse MCs to guide corresponding revision planning.