Study aims
The primary aim of the BiPOWR trial is to prospectively describe and compare efficacy (coronal and sagittal curve maintenance) and –safety (occurrence of Serious Adverse Events; SAEs) of two innovative implant systems in neuromuscular EOS patients. Due to the relatively short follow-up of participants (1 year), with intermediate instead of final outcomes, the study is regarded as a limited-efficacy and -safety study [21]. A secondary aim is to describe and compare spinal- and implant growth, peri-operative parameters and Health-Related Quality of Life (HRQoL).
Study design
The BiPOWR trial is a prospective, multicenter, randomized, surgical trial with two parallel groups and a 1:1 allocation ratio. Ethical approval was provided by the Institutional Review Board of the UMC Utrecht (METC 18–179). Following inclusion, patients undergo “growth-friendly” scoliosis surgery with one of two implants, the SDS or the OWSER. Which implant they receive is decided by randomization. Following implantation, the patients are followed until 12 months post-operatively, during which their radiographic- and clinical outcomes will be described and compared. The analysis at 12 months provides a point for interim analysis in the short follow-up, to determine whether one or both implants show short-term technical malfunctions, and to ascertain whether early detectable differences are present in other outcome domains. After the 1 year analysis, both patient cohorts are followed bi-yearly for several more years, until after skeletal maturity, to show efficacy and safety in the long-term. A SPIRIT table showing a detailed follow-up timeline for the first year of follow-up is shown in Table 1.
Eligibility criteria
Inclusion criteria are:
Exclusion criteria are:
-
Closed triradiate cartilage.
-
Main curve proximal end vertebra at or above T3.
-
Presence of skeletal dysplasia affecting growth.
-
Presence of disease that severely influences bone quality or is associated with soft tissue weakness.
-
Presence of active systemic disease (other than neuromuscular disease).
-
Congenital spinal anomaly of > 5 vertebrae.
-
Previous instrumented spinal surgery.
-
Patients who cannot be followed for 1 year post-operatively.
Recruitment and informed consent
The treating physician identifies potential eligible patients, and asks the patient and their caregiver(s) whether they would like to receive additional study information. Upon agreement, the principal investigator is consulted for approval and the research team subsequently sends the study information. After at least two weeks, a consultation with the research team takes place where the in- and exclusion criteria are assessed, the study is explained in detail and patients’ questions are answered. Care is taken to ensure that the study design (in particular blinding and randomization) and the similarities and differences of the two treatment arms are understood. If willing to participate, the informed consent form is signed in duplicate and the consent is noted in the electronic patient record.
Randomization and blinding
After inclusion, the patient is randomized into one of two treatment groups. Before the start of the trial, a randomization sequence was created using a computer-generated random number sequence, which was used to create a permuted block design with random block sizes. A random number of blocks with random block sizes of 2, 4, 6 and 8 created the final sequence containing 28 randomized patients. The sequence was converted into allocation notes, which were stored in opaque, sealed envelopes. Each allocation envelope was sealed into a second, sequentially numbered, opaque envelope. The use of two envelopes prevents the use of trans-illumination to unblind allocation. Two weeks before surgery, the next numbered study envelope is retrieved by two members of the study team. The envelopes, the seals in particular, are visibly examined for tampering. The allocation is confirmed by both research team members, and the pseudonymized patient number is written on the allocation note and the study envelopes. The treating physician is then notified of the outcome of the allocation, to order the required implants.
The patient and caregiver(s) remain blinded to the result of allocation until after surgery to prevent selection bias through termination from the study before surgery commences, which is a possibility if they have a strong preference for one treatment arm [22]. After surgery has taken place, the allocation outcome is shared with the patient and caregiver(s). While blinding patients during the entire study period is theoretically possible, enforcing this would require extreme effort, as it would preclude the treating physician of showing any of the follow-up radiographs to the patient and caregiver(s) as these clearly show the different implants. Since these radiographs often provide a wealth of information to the patient and caregiver(s), denying them these raises ethical concerns and would likely decrease enrolment rate.
Surgical procedure and treatment arms
All surgical procedures will be performed using a less invasive bipolar posterior approach with somatosensory and motor-evoked potential monitoring [23]. The distal anchor is created with iliosacral screws (Tanit®; Euros, SAS, La Ciotat, France). The proximal anchor for SDS consists of bilateral pedicle screws at 3 consecutive levels, typically T2-T4. For the OWSER, a series of laminar- and pedicle hooks spanning 5 vertebrae is applied (T2-T6).
The SDS (Fig. 1) is an adjunct to standard 5.5 mm cobalt-chromium (CoCr) rods. It consists of three components: (1) a titanium (Ti6Al4V) spring placed over the rod, (2) two stacked oversized parallel connectors and (3) a buttress used to tension the springs (Stryker, Leesburg, VI, USA). In neuromuscular EOS, the SDS is placed bilaterally. The concave rod receives a strong 100 N spring (spring constant (k) = 1.32 N/mm), the convex rod receives a weaker 50 N spring (k = 0.68 N/mm). Around 6–7 cm of residual rod length is left for growth. The parallel connectors are secured to the distal anchor rods and the rod containing the spring is allowed to slide in the connector.
The OWSER (Fig. 2) consists of two titanium components: (1) a 5.5 mm titanium long rod with a notched end, and (2) a sliding domino that passively migrates only one way across the notched rod segment. The long rod is connected to the proximal anchor with the notched end facing distally. The connector is fixated with a short rod to the distal anchor. Implant growth potential (i.e. the notched segment of the long rod) can be 50 or 80 mm. In neuromuscular EOS, the OWSER is also placed bilaterally.
The implant configurations are shown radiographically in Fig. 3. All patients treated by either SDS or OWSER are allowed unrestricted physical activities postoperatively. We do not routinely perform manual axial trunk traction in the outpatient clinic for these most vulnerable of OWSER patients. For OWSER implant lengthening, we instead rely only on passive lengthening due to spinal growth, in combination with normal traction and bending generated through daily activities.
Outcomes
The primary outcomes include coronal Cobb angle – measured on radiographs – and occurrence of SAEs (defined as unplanned medical events which (could) result in permanent disability/damage, or which requires (lengthening of) hospitalization, a re-operation, or inpatient medical managing). SAEs will be categorized as either device/surgery related (e.g. rod fracture, post-operative neurological deficit), or as disease-related (e.g. pneumonia, pain) according to a previously created complication classification system for use in EOS treatment [24]. In addition to SAEs, complications that do not meet the criteria for an SAE, but which are device/surgery related (e.g. superficial infections, proximal junctional kyphosis) will also be recorded.
Secondary radiographic measurements include parameters such as T1-T12, T1-S1 and instrumented height, T5-T12 kyphosis, L1-S1 lordosis, and pelvic obliquity. In addition, implant length will be calculated at each timepoint for both groups (SDS: Spring length, OWSER: Notched rod length) so that implant growth velocity over time can be graphed and calculated. The number of outpatient axial traction events of OWSER patients, if these are performed, will also be reported. All radiographic measurements will be performed by a trained assessor. To ensure data quality, radiographic measurements are performed semi-automatically in Surgimap v2.3.2.1 software (Nemaris, New York, USA), after which all measurements are audited by the assessor. At the end of the data collection phase, all radiographic measurements are re-assessed by a second trained assessor who is blinded to the first assessors’ measurements. Measurements are compared and disputed measurements are discussed until consensus is reached. If differences are believed to be caused by random measurement variation, the arithmetic mean of both assessors’ measurement is taken.
HRQoL will be measured with the validated Dutch version of the EOSQ-24 which consists of 24 questions in 10 domains, from which a domain- and total score is calculated [25, 26]. All outcome parameters are measured pre-operatively, immediately post-operatively, and at 1, 3, 6, and 12 months follow-up.
Sample size calculation and statistical methods
The sample size calculation is based on potential differences between groups in maintenance of the coronal curve between the post-operative situation and 1-year follow-up. A difference in Cobb angle ≥ 5˚ is considered clinically relevant and can be reliably measured [27]. In our previous MCGR cohort studies, we found a change in Cobb angle between post-operatively and 1 year follow-up of 1˚ (SD 5.1˚) [28]. To determine differences in curve maintenance between groups ≥ 5˚ with a power (1-β) of 0.8, α = 0.05, and SD = 5.1˚, we calculated a sample size of n = 28, with a repeated measures ANOVA between factors design (comparing 2 timepoints). With an allocation ratio of 1:1, this means that 14 patients have to be included in each experimental group.
If all assumptions for performing repeated measures ANOVA are met, both systems will be compared with respect to coronal curve maintenance; age and sex will be included as covariates. For the other continuous variables, repeated measures ANOVA will be used, with the same covariates, but including all evaluated time points. If the assumptions for performing repeated measures ANOVA are not met (e.g. due to missing data), repeated measurement multilevel modelling will instead be performed, with patient, age, sex and treatment group as level 2 variables, and follow-up time as a level 1 variable.
A Kaplan–Meier survival analysis will be performed comparing both groups. The survival curves showing SAE-free survival of both groups will be statistically compared with the Log-Rank test. Analysis of EOSQ-24 scores will be performed using repeated measures ANOVA. In case there are missing EOSQ-24 item scores or questionnaires, multiple imputation with parcel summary scores will be performed using a previously published statistical method [29]. For all statistical tests, statistical significance will be set at p < 0.05. All statistical analyses will be performed in IBM SPSS statistics v.26.0.0.1 (Chicago, USA).