Introduction

The objective of spinal fusion in adolescent idiopathic scoliosis (AIS) is to restore both coronal and sagittal alignments, in order to provide satisfactory long-term functional outcomes [1,2,3]. Sagittal balance is to date mainly assessed by long-length standing radiographs, on which numerous spino-pelvic parameters can be measured [4]. There is currently a growing concern regarding radiation exposure in immature patients, and practitioners now have a tendency to limit imaging prescriptions in children and adolescents during follow-up [5,6,7]. In our current practice, scoliotic patients are evaluated twice a year using low-dose biplanar stereoradiographs (EOS Imaging, Paris, France) and usually undergo supine bending ± traction films if surgery is considered [8]. During the spinal fusion procedure, two additional posteroanterior radiographs are performed on a radiolucent operating table: one at the end of exposure to analyze the frontal tilt of the planned lower instrumented vertebra (LIV) and confirm the instrumentation level, and a second one at the end of correction to assess T1, upper instrumented vertebra (UIV) and LIV residual tilts, but also to detect any implant misplacement [9].

Hospital stays are now reduced to 4–5 days in AIS, but standing radiographs are still traditionally performed before discharge. Most patients are still painful at that time, and the authors have noticed that analyzing postoperative alignment on such early radiographs was often challenging. The goal of this study was therefore to investigate the relevance of early standing postoperative radiographs after AIS posterior fusion.

Materials and methods

Patients

After IRB approval, all consecutive AIS patients (Lenke 1 to 6) who underwent posterior fusion between January 2015 and December 2015 were prospectively included. A minimum 2-year follow-up was required, and patients with previous spine or lower limb surgery were excluded. Patients were operated using hybrid constructs, combining lumbar pedicle screws (up to T11), thoracic sublaminar bands (from T4 to T10) and autostable claws for the proximal fixation (Zimmer Biomet, Warsaw, Indiana, USA, or Implanet, Bordeaux, France). No selective thoracic fusion was performed, and all LIV were located between L2 and L4.

Radiological measurements and analysis

Low-dose standing stereoradiographs were performed in all cases before hospital discharge (early X-ray), between the 4th and the 6th postoperative day, according to a previously reported protocol [10]. Subsequent clinical and radiological follow-up visits were organized at 4 months postoperative, 1 year and 2 years after surgery. Frontal and sagittal alignments were evaluated before discharge, on the first postoperative evaluation and at latest follow-up.

A validated visualization tool (Carestream Health, Rochester, NY, USA) was used to measure angles and distances on digitalized radiographs from the picture archiving and communication system (PACS) [11]. Fifteen parameters assessing spinal alignment (eight coronal and seven sagittal, summarized in Table 1) were measured two times by the same experienced observer on each radiograph, and the mean value was kept for analysis. In addition, any case of implant misplacement or migration was recorded.

Table 1 Coronal and sagittal radiological measurements
Full size table

Statistical analysis

Radiological parameters were compared between the three evaluation periods using paired student t-tests (Excel, Addinsoft SARL, Paris, France). P < 0.05 was considered significant.

Results

Patients

One hundred AIS patients (89 girls and 11 boys) were included. Mean age at surgery was 16 years (± 2). According to Lenke’s classification, there were 47 Type 1, 16 type 2, 15 type 3, 2 type 4, 17 type 5 and 3 type 6 [12]. The UIV was located on T2 or T3 in 88% (Fig. 1). The LIV was L2 for 27 patients, L3 for 50 patients and L4 for the remaining 23 cases. The mean length of hospitalization was 5 (± 1) days.

Fig. 1
figure 1

Distribution of upper instrumentation level in the cohort

Full size image

Radiological analysis

Significant differences between the early X-ray and the 4-month control were found for 12 out of the 15 radiological parameters measured (80%) (Table 2). In opposition, only three measurements were significantly modified between the 4-month radiograph and latest follow-up (Table 3). The most significant change occurred in the distal non-instrumented lordosis, but the difference only averaged 2.8° (p < 0.05).

Table 2 Comparison between early X-ray and 4-month control measurements (mean ± SD)
Full size table
Table 3 Comparison between 4-month control and latest follow-up measurements (mean ± SD)
Full size table

Comparison between early X-ray and 4-month control showed that parameters analyzing global spinal balance in both coronal and sagittal planes (CVA, SVA and CAEPL) improved significantly over time (Table 3) (Figs. 2, 3). The main coronal parameters investigated (CVA and clavicle tilt) also significantly improved during the 4 months postoperative (Fig. 2). In opposition, all sagittal parameters reflecting the non-instrumented spine and the adjacent compensation above or below the fusion mass significantly increased (Fig. 3).

Fig. 2
figure 2

Preoperative (a), early X-ray (b) and control at 4 months (c) after surgical correction of a Lenke 1 AIS, showing the spontaneous readjustment of CVA (from 14 to 4 mm) and clavicle tilt (from 18° to 3°)

Full size image
Fig. 3
figure 3

Preoperative (a), early X-ray (b) and control at 4 months (c) after surgical correction of a Lenke 5 AIS, showing the spontaneous readjustment of CVA and clavicle tilt

Full size image

Results of the comparison between the 4-month and last follow-up examinations are summarized in Table 2. Parameters analyzing the global spinal balance in coronal and sagittal plane (CVA, SVA and CAEPL) were not significantly modified. The only three parameters that significantly increased were: L1–L5 sagittal Cobb, the non-instrumented lordosis and the sagittal angle of the first non-instrumented lumbar intervertebral disk.

Implant misplacements

An implant misposition was reported in nine patients on the early X-rays (9%). Lumbar pedicle screws were involved in all cases, either too cephalad or too lateral, but no symptom was reported and none of them was changed because the LIV was properly instrumented in all cases. No modification of the intraoperative somatosensory/motor-evoked potentials occurred, and no neurological complication was recorded. No patient underwent revision surgery during the follow-up period.

Discussion

Goals of surgery

Most AIS patients are asymptomatic, and the surgical indication is often driven by patients’ cosmetic concerns. This surgery remains at risk and many complications, such as proximal junctional kyphosis, distal junctional kyphosis and adding-on are partly iatrogenic, due to a wrong fusion level or unbalanced arthrodesis. The aim of surgery is therefore to obtain a mass of fusion centered over the pelvis, while restoring the cosmetic aspect of the trunk and shoulders balance, and maintain or restore sagittal alignment, which has been correlated with long-term quality of life [13,14,15,16]. Despite the recent emphasis on sagittal analysis, respecting the frontal alignment is also of great importance, since residual LIV tilts > 5° or LIV translation > 2 cm have been associated with higher risk of adding-on and distal disk degeneration [3, 16].

Unfortunately, the intraoperative radiological control of sagittal alignment remains currently insufficient, mainly limited by the feasibility of a long-length cassette X-ray, as well as the poor visibility of the upper thoracic spine, hidden by the shoulders and the arms. In opposition, posteroanterior radiographs during surgery have proved to be helpful during surgery to adjust shoulders levels and residual frontal tilts at both ends of the construct [9]. They can also detect early implant misplacement, even without significant change in neuromonitoring that can lead to screw trajectory modification. This intraoperative frontal X-ray continues to be for us necessary and of great importance, but it clearly does not reflect the future global standing alignment and subsequent radiographs need to be considered to analyze postoperative balance. For that reason, and because we have since 2007 a low-dose stereoradiographic system, EOS images were systematically acquired before hospital discharge. However, since the recent development of rapid recovery pathway (RRP) protocols, the average length of stay after AIS surgery has decreased in our department from 7–10 days to 4–5 days. Even though patients are capable to walk and stand still for couple minutes in the EOS system after 3 or 4 days, they are still under pain medication, and have not recovered yet their physiological balance. For that reason, and because high concerns have been expressed in young patients regarding radiation exposure and the risk of cancer, the relevance of such early radiographs has been progressively questioned [5,6,7, 17].

Relevance of radiological analysis before discharge

Results of the current study confirm that early X-rays performed before discharge are not relevant to analyze postoperative global spinal alignment. There is therefore no need for additional immediate postoperative radiographs, if an intraoperative X-ray has already been taken. Indeed, important parameters in both frontal and sagittal planes, such as CVA, SVA and CAEPL, were significantly modified between the first radiograph and the control at 4 months postoperative (Table 3). In opposition, these parameters remained stable over time after that period, without significant change at final follow-up.

The main finding was that patients initially had a tendency to reduce their lumbar lordosis, probably due to muscle pain, with subsequent anterior shift of both SVA and CAEPL. Patients progressively increased their lordosis in the unfused lumbar segments (5° on average) and reached their final alignment at 4 months (Tables 2 and 3, Fig. 3). Similarly, T1–T12 kyphosis significantly increased during follow-up, mainly due to a 2.5° increase in the two segments above the UIV. This finding confirms that rates of PJK can efficiently be assessed at 3–4 months postoperative, when patients have found their final position above and below the fusion mass.

In the frontal plane, both UIV and LIV residual tilts significantly decreased during the first 4 months (Table 2), and the clavicle tilt reflecting shoulder balance also significantly improved. These parameters remained stable at latest examination. Despite the current findings, we still believe that the radiological assessment at 4 months remains necessary. First, the patient and his caregivers can visualize the spinal correction, which is a source of satisfaction. Second, some early radiological complications can be detected, such as PJK or distal adding-on, while they are still asymptomatic. Finally, it is important to be able to measure alignment parameters and therefore quantify postoperative outcomes (Fig. 4).

Fig. 4
figure 4

Preoperative (a), early X-ray (b), control at 4 months (c) and 2 years (d) after surgical correction of a Lenke 1 AIS. It illustrates the spontaneous rebalancing backward of SVA, from 28 mm (b) to 0 (c) then to − 15 mm (d) with a concomitant increase in non-instrumented lordosis, from 30° (b) to 42 (c) then to 48° (d)

Full size image

Limitations

The main limitation of this study is the lack of intra- and interobserver reliability assessment on radiological measurements. However, measurements were taken twice by an experienced spine surgeon, using a validated tool with image enhancement. In addition, the delay for postoperative X-rays was not standardized, and they were performed between the third and fifth day postoperative. For practical reason, the first radiological control after discharge was only performed at 4 months, so we were not able to determine the exact delay to recover a global alignment close to the definitive one. As a matter of fact, the unfused lumbar lordosis continued to increase until final follow-up, and shoulders are known to readjust until 2 years postoperative. The gravity line, now assessable from EOS 3D reconstructions, was not used in the current study, but this parameter associated with a forces platform will be further studied [18,19,20].

In conclusion, intraoperative radiographs remain necessary to detect implant misplacements and verify that the objectives of surgical planning had been reached. However, early standing X-rays performed before discharge are not relevant to assess postoperative alignment in AIS and could be postponed to the first postoperative visit to reduce radiation exposure. The current findings are only applicable to idiopathic scoliosis but should be studied in other spinal deformities.