Carlo Peeters

Introduction and of the major findings of this thesis
An important clinical study about the non-operative management of adolescent idiopathic scoliosis (AIS) was a randomized and preference cohort trial, published in 2013, which has shown us that brace therapy can significantly decrease the progression risk and subsequent risk for surgical treatment[1]. This study was stopped early owing to the efficacy of bracing and has contributed to the change of view of many doubting physicians who previously saw brace therapy as ineffective. Building upon this evidence, this thesis aimed to expand the knowledge about factors associated with brace treatment success in AIS as bracing is not successful in every patient. Strong evidence has been reported for the association between lack of initial in-brace correction and brace treatment failure, which is the reason that Chapter 4 and 5 focused on influencing factors on this in-brace correction[2].

Increased curve flexibility and thoracolumbar or lumbar curve pattern were found to be favourable predictive factors, and a double major curve pattern was found to be an unfavourable predictive factor for initial in-brace correction (Chapter 4). The degree of torso asymmetry and segmental peak positive and negative torso displacements in brace can be analyzed with the use of patients’ 3D surface and brace models, but unfortunately these parameters are not clearly associated with initial in-brace correction (Chapter 5). Besides initial in-brace correction, also compliance plays an essential role in the success of brace therapy[1-3]. Knowledge about the best way to monitor progression, and about the impact of brace wear and the effect of brace-related interventions on health-related quality of life (HRQOL) of scoliosis patients undergoing brace treatment could be important for motivational reasons and new insights for better brace compliance, respectively.

Chapter 6 shows that a follow-up protocol for AIS patients with in-brace radiographs is non-inferior regarding curve progression rate over time compared to a follow-up protocol with out-of-brace radiographs. In Chapter 7 a culturally adapted Dutch version of the Brace Questionnaire (BrQ) is presented, which is found valid for clinical and research purposes. This thesis also aimed to explore the possibilities of using a biplanar low-dose X-ray device (EOS imaging, Paris, France) as a tool for spinal length and pedicle size measurements. Knowledge about spinal length and subsequently growth of each individual AIS patient helps with accurate timing of both conservative and surgical treatment, and preoperative knowledge about pedicle sizes could contribute to the placement of an adequate amount of well-sized pedicle screws. Chapter 2 and 3 shows that there is a good validity and reliability for both spinal length and pedicle size measurements on EOS radiographs. This chapter discusses the interpretations and implications of these major findings of this thesis, limitations of the studies, and future perspectives.

The value of EOS radiographs for spinal length and pedicle size measurements
The clinical applications of the EOS imaging system in orthopaedics are expanding rapidly owing to several advantages. The system can, for example, provide standing low-dose radiographs of the whole spine at once, which reduces the amount of radiation substantially in comparison with conventional radiographs[4, 5]. Furthermore, it uses biplanar perpendicular radiographs with the result that images have no divergence in the vertical plane allowing more accurate two-dimensional (2D) and 3D measurements. Previous studies have shown that EOS has value for clinical practice and can be used for reliable measurements of spinal curvature and sagittal balance, but also for non-spine related measurements like pelvic tilt and acetabular cup orientation, femoral offset, and lower limb measurements[4-9]. Chapter 2 and 3 shows that EOS can also be used for reliable spinal length and pedicle size measurements. Within the management of AIS, this may help with accurate timing of both conservative and surgical treatment, and providing a preoperative indication of needed pedicle screw diameters.

Spinal length measurements
In literature, spinal length measurements are often performed on coronal radiographs, which have the disadvantage of X-ray beam divergence and not including deviations in the sagittal plane[10-13]. Length assessments of complex 3D deformities such as a scoliosis should therefore not be performed with 2D measure methods. As the EOS 3D spinal length measurements resulted in the best representation of the true spinal length, the 3D length measure method should be preferred above spinal length measurements on individual coronal or sagittal images (Chapter 2). The EOS imaging system should particularly be considered in scoliosis clinics where growth-friendly implants are used. Reliable 3D spinal length measurements and subsequently knowledge about the growth of each treated patient is essential here, since the patient’s ability to grow with these implants is limited. But monitoring of the spinal growth could also be useful in the treatment of the other patients with idiopathic scoliosis. Not only for determining the duration of brace treatment and timing the potential surgery, as surgery should be postponed until the peak growth velocity of the spine has passed to prevent complications like the crankshaft phenomenon, but also for further research purposes[14].

There exists a well-known relationship between the patient’s growth and development of the spinal deformity, and high spinal growth velocity during the early pubertal growth spurt is a predisposing factor for a rapid increase of the deformity[13, 15]. Monitoring of the patients’ individual spinal growth spurt and its velocity by using reliable 3D spinal length measurements could therefore contribute in predicting curve progression. It would be interesting to investigate the potential predictive value of spinal growth velocity, amongst other maturity indicators that reflect growth or remaining growth potential, for the prediction of the timing of the peak growth velocity of total body height and subsequently the curve progression in the individual AIS patient[15]. A study protocol for a prospective, longitudinal cohort study to answer this research question has already been developed, with sitting height velocity, leg length velocity, shoe size velocity, foot length velocity, skeletal age, Risser sign, triradiate cartilage, menarche, Tanner stage, and EMG ratios of the paraspinal muscles as the other maturity indicators[15].

The EOS 3D spinal length measurement method would be of value to this study protocol as it represents the true spinal length better than 2D measurements on the coronal image that originally was proposed in the protocol. Aside from a cohort study with limited amount of patients and study duration, the use of a multi-center prospective longitudinal database would be even more valuable. Such large database should preferably also include other countries to discover population group differences. Ultimately, the development of an artificial intelligence algorithm based on patient-specific factors and radiological parameters calculating the individual risk of curve progression would be ‘the icing on the cake’[16]. Because knowing if and when a curve will progress would prevent unnecessary brace treatments and reduce the bracing period and the number needed to treat to prevent one case of curve progression requiring surgery.

A limitation of the EOS 3D spinal length measure method is that despite good intra and interobserver reliability was observed, manual placement of measurement points may possibly be suboptimal because the visualization of vertebral endplates is not always good in the upper thoracic region due to overprojection of the shoulders. Secondly, this method is not standardized and therefore labor-intensive. When considering implementing for regular follow-up moments in standard practices and for large multi-center prospective longitudinal databases, it not realistic to expect that for every AIS patient the total spine length can be assessed and monitored easily. Ideally, this spinal length measurement would be captured in a 3D machine learning system in order to be less time-consuming. But as long as there is no automatic or easier way to achieve reliable 3D spinal lengths, this spinal length measure method would probably be used for research purposes only.

Pedicle size measurements
Prior to scoliosis surgery, preoperative knowledge about the pedicle size helps to maximize screw containment and minimize the risk of pedicle breach. Pedicle sizes should ideally be measured on preoperative computed tomography (CT) for the most reliable measurements. This is, however, not done routinely due to the exposure of this young population to high levels of radiation. Preoperative EOS images were suggested in Chapter 3 as potential alternative for pedicle size measurements, as these images are generated with much lower levels of radiation and have no divergence in the vertical plane. The results have shown that visible pedicle sizes can reliable be measured on coronal EOS radiographs. In daily practice, surgeons using free-hand pedicle screw insertion methods can therefore preoperatively measure intra- and extracortical pedicle widths for an indication of the needed pedicle screw diameters for those individual pedicles. When using intra-operative 3D imaging and a pedicle screw navigation system, preoperative knowledge of pedicle sizes could reduce the intra-operative dose, as for determining the optimal screw trajectory less resolution and therefore less radiation is needed.

It would be interesting to investigate whether pedicle size dimensions measured on EOS radiographs combined with a simpler intra-operative navigation system could result in a significant reduction in radiation dose without compromising accurate placement of well-sized pedicle screws. However, there is one major limitation of using the coronal EOS radiograph for pedicle size measurements in idiopathic scoliosis. Owing to vertebral rotation, pedicles on the concave side with a Nash-Moe grade score of 2-3 cannot be measured[17]. So not every pedicle can be provided with an indication of the needed pedicle screw diameter, as axial rotation can almost always be recognized in the spinal deformity of surgery candidates, especially near the apex. The pedicle size of the convex pedicle of the rotated vertebra is, unfortunately, not representative for the contralateral concave pedicle due to the asymmetry in idiopathic scoliosis[18, 19]. Furthermore, there is a systematic, small underestimation of the pedicle width measurements on EOS images for these convex pedicles with a Nash-Moe grade score of 2-3. But although surgeons should be aware of the small underestimations, these are likely clinically irrelevant, as pedicle screws generally differ 1mm in diameter sizes, while the mean differences of intra- and extracortical pedicle width measurements of Nash Moe 2-3 pedicles between EOS radiographs and intra-operative 3D images were only -0.47mm and -0.51mm, respectively.

Future studies should focus on measure instruments that can reliable measure pedicle sizes of all pedicles with less radiation than a CT-scan. Magnetic resonance imaging (MRI) has been proposed as alternative, but it was found inferior to CT for scoliosis patients, because it has poor accuracy to properly detect pedicle abnormalities[20]. The more severe the pedicle abnormality, the less diagnostic value the MRI had[20]. Particularly in spinal deformity surgery, preoperative knowledge about pedicle sizes is warranted due to the different morphometric characteristics of the pedicle dimensions[18]. Under- or oversizing of pedicle screws increases the risk of pedicle fracture and screw loosening[21]. Furthermore, in scoliotic spines, up to one third of the mid-thoracic pedicles are not appropriate for a safe intrapedicle screw placement[22]. Although a commonly accepted criteria for pedicle screw diameter selection has not been proposed in literature yet, the systematic review of studies with recommendations reported a screw diameter ranging from 80% to a maximum value of 125% of the pedicle width[21]. This is a wide range, in which higher values for maximum screw diameter / pedicle width ratio were described for pediatric populations, owing to the relative plasticity of the pedicle cortex in the pediatric spine. Since insertional torque is useful to predict screw fixation strength, future studies investigating screw diameter / pedicle width ratio in the scoliotic spine, should also include the peak insertion torque as determining factor[23].

The black box of brace manufacturing technology
In Chapter 4, moderate evidence is found that braces designed with computer-aided design (CAD) and manufacturing systems with or without finite element models simulation do not significantly improve initial in-brace correction, compared to braces fabricated using the conventional plaster-cast method. So far, our knowledge on working mechanisms of braces is limited and most braces are still hand-crafted by the orthotist. The introduction of new brace designing and manufacturing technologies in clinical practise allows further research in this field to obtain a better insight in the correction mechanisms of the brace. The results of the pilot study, presented in Chapter 5, shows that the degree of torso asymmetry and segmental peak positive and negative torso displacements in brace can be analyzed with the use of patients’ 3D surface and brace models, but that unfortunately these parameters are not clearly associated with initial in-brace correction of AIS patients with Lenke 1 or 5 curves. Although the patient sample and mean initial in-brace correction compared to literature were relatively small, it is very likely that these two measurable factors in the brace model alone are not helpful in predicting the radiographic initial in-brace correction. A possible explanation for the weak and negligible correlations is that peak positive displacement does not correlate with amount of applied pressure. A comparable amount of displacement directly applied on bones, for instance, would result in a larger pressure and torso displacement compared to the same displacement on fat tissue.

In literature, there is insufficient evidence for other potential brace related factors influencing the radiographic initial in-brace correction, such as the magnitude of the corrective force over brace pads (Chapter 4). Since there is no evidence based consensus on the best possible manner to achieve curve correction with bracing, the experience and even intuition of the orthotist play an essential role, representing more the art than the science of medicine[24]. Future studies should try to enlighten the black box of brace manufacturing technology as the exact mechanisms to achieve curve correction remains obscure. Starting with finding brace related parameters influencing the initial in-brace correction, would be a good strategy in order to obtain knowledge where new brace design and manufacturing methods should be focusing on. It is, for example, not excluded that when segmental peak torso displacements in-brace are combined with other brace related factors like the pad location and pad pressure or patient factors like curve magnitude or curve flexibility, they could be of added value in predicting initial in-brace correction and/or improving brace comfort. Future studies on CAD brace related factors that influences initial in-brace correction should include quantifiable parameters obtained from 3D scans and models in combination with in-brace pad pressure parameters obtained with electronic pressure sensors[25, 26]. In addition, the use of 3D ultrasound system could help with the determination of the optimum pressure level and location to assist the brace design[27].

In-brace or out-of-brace protocol for radiographic follow-up of patients with scoliosis?
In clinical practice, regular follow-up radiographs are usually made at 6 month intervals to detect curve progression during the brace treatment. Early detection of curve progression is important for motivational reasons as the most important positive factors influencing brace compliance are the patient’s desire to avoid surgery and to prevent curve progression[28]. According to the SOSORT bracing protocol, these follow-up radiographs should be taken out-of-brace to examine the effectiveness of treatment (level V of evidence)[29]. It allows visualization of progression above the pre-brace curve magnitude. However, Chapter 6 shows that a similar curve progression rate over time can be expected with both in-brace and out-of-brace protocols for radiographic follow-up in patients with idiopathic scoliosis. This is interesting, because it has generally been assumed that there is decreased detectability of curve progression owing to the partial curve correction by the brace. The major advantage of in-brace radiographic follow-up is that proper curve correction can be evaluated and brace corrections can be made if necessary. The total spine length of the patient is growing during the treatment, so periodically brace adjustments are necessary, and therefore also in-brace evaluations. When protocollary adjusting and optimizing the brace position by the orthotist before the radiograph is taken, this would probably result in better in-brace corrections. To investigate the ‘at home’ correction, the radiographs of the in brace group were made before interventions or adjustments by the orthotist or physician.

Therefore, clinicians should be careful with the interpretation of the in brace correction values, since adjustments were made after the radiograph. Ideally, the in-brace radiograph should be repeated after the brace adjustments by the orthopaedic surgeon and orthotist to evaluate its effectiveness. It has strongly proven that lack of initial in-brace correction is associated with brace treatment failure. But the influence of periodic checking of the in-brace correction during the bracing period on long term brace success has not yet been clarified[2]. For this latter research question, it would be rather useful to make an in-brace radiograph after the brace adjustments in order to evaluate the intervention. It might be interesting to know if not only the initial in-brace correction, but also the following periodically in-brace corrections contributes to better brace treatment success. As discussed in the previous paragraph, this also helps to better understand the correction mechanisms of the brace. Although monitoring of the curve remains essential during brace treatment, future follow-up methods should exclude radiographs in order to reduce the amount of radiation to this young population group. The use of low-dose radiographs of the whole spine at once (EOS) reduces the amount of radiation substantially in comparison with conventional radiographs and could therefore be an intermediate step before new radiation free imaging techniques have been developed and validated for the evaluation of curve magnitude and in-brace correction.

So, which protocol should we use for the time being?
It is hard to identify one best follow-up strategy, as both in-brace and out-of-brace protocols has its advantages and disadvantages. The protocol with in-brace radiographs was also non-inferior regarding curve progression rate over time. First, switching between protocols results in a temporary inability to detect curve progression, so this would not be recommended. An one-time switch from the protocol with in-brace radiographs to the protocol with out-of-brace radiographs by exception could be considered, when progression is demonstrated on subsequent follow-up radiographs and the major curve Cobb angle is exceeding 40 degrees. This is because curves exceeding 45-50 degrees are usually treated surgically, and out-of-brace radiographs can provide more useful information for clinical decision making in these severe curves[30]. Despite the similar curve progression rate over time between the two protocols, the severity of the major curve Cobb angle is still underestimated on an in-brace radiograph.

A potential delay in surgical treatment could occur, making the out-of-brace protocol preferable for potential future surgery candidates. For non-potential surgery candidates, for example AIS patients with a major curve below 40 degrees, a clinician might consider using the protocol with in-brace radiographs in order to evaluate the curve correction at each follow-up moment so that the brace can be adjusted if necessary. Since only surgical treated patients with failed brace treatment were included in the presented study, these recommendations should be interpreted with some caution for the non-surgical treated patients. However, the included patients with proven curve progression was considered as the most relevant group for the study’s research question. Furthermore, a possible lack of compliance to the brace treatment was not monitored and this is an important factor for treatment failure[1-3]. It is not known whether insufficient in-brace corrections, compliance or both were the reason for brace treatment failure in patients in Chapter 6. However, Chapter 4 shows insufficient evidence for compliance as influencing factor for initial in-brace correction. This was based on one study, in which brace wearing hours were recorded on a log sheet and by an orthosis monitoring system, that reported no significant difference on in-brace correction (<40% and ≥40% correction) after 4-6 months between three groups of different hours of brace wear (0–8 hours, 9–16 hours, and 17–23 hours)[31]. Why should we use the Brace Questionnaire? As the generally low compliance rates remains a challenge for healthcare professionals, further knowledge about the impact of brace wear and the effect of new brace modifications or brace-related interventions on different HRQOL domains could lead to new insights for better brace compliance. The revised Scoliosis Research Society 22-item questionnaire (SRS-22r) assesses the overall HRQOL of AIS patients, but does not contain a specific item on the influence of brace therapy on HRQOL [32]. For this reason, the BrQ was translated into the Dutch language, as presented in Chapter 7. The translated and culturally adapted Dutch version proved to be valid and reliable. The overall BrQ score, its floor and ceiling effects, internal consistency and reproducibility were comparable to previous BrQ validating studies [26, 33-38]. This suggests that the BrQ can be reliable used in the Dutch population group for AIS patients undergoing brace treatment. A recently published systematic review, including 60 articles of which 12 used the BrQ as HRQOL instrument, discovered that self-image, mental health, and vitality are the three most frequently reported domains in scoliosis patients undergoing brace treatment[39]. But the authors mentioned in their limitation section that the influence of factors such as curve magnitude on these three domains have not been clarified yet[39]. Future studies should therefore identify patient characteristics influencing these domains in order to provide more specific information on which patient group we should pay extra attention. A long-term longitudinal follow-up study with biannual time-intervals during the whole bracing period would be preferred, since the impact of brace wear for the individual AIS patient can change over time. In addition, the BrQ could help with monitoring the effect of new brace modifications or brace-related interventions on different HRQOL domains in future brace studies. Therefore, it is warranted that the HRQOL questionnaire also has specific items on the influence of the brace treatment. Besides for research purposes, the BrQ could be used for clinical applications as well. In daily practice, it is important to identify the patients undergoing active brace treatment who are scoring below the norm, so that additional brace adjustments, extra monitoring, and proper support of the physician, the parents, and/or a psychologist in the form of individual sessions or group sessions, can be provided[40].