Comparison of 3 Different Minimally Invasive Surgical Techniques for Lumbar Spinal Stenosis

This randomized clinical trial compares the effectiveness of 3 minimally invasive posterior decompression techniques for lumbar spinal stenosis at 16 public hospitals in Norway.


Trial design 90
The NORDSTEN-SST is an open label, multicentre randomized controlled trial, with a 91

Randomization 116
Eligible patients are randomised into one of three arms, unilateral laminotomy with crossover, 117 bilateral laminotomy and spinous process osteotomy with a 1:1:1 allocation. A randomized 118 block design, stratified by hospital and the blocks made as small as possible, is used to ensure 119 equal distribution of all three treatments (randomly selected block size 4 and 6). 120 Randomisation is performed in the 6 weeks before surgery. The randomisation procedure is 121 concealed and administered by a study coordination centre at a university hospital and 122 communicated to a local research coordinator who is not involved in the treatment of the 123 patients. Randomisation is performed after the patient has signed the informed consent form. 124 The result of the randomisation is documented in the patient record. Hence, neither the 125 patients nor the surgeons can influence the type of intervention. 126 127

Sample size 128 129
The trial is planned with a superiority design. The study is designed to detect a difference of 7 130 ODI-points between the groups. With a standard deviation of 18, a significance level of 0.02, 131 80% power and a drop-out rate of 15%, we need to include 155 patients in each group. Thus, 132 we plan to include 465 patients over a 4-5 year period. The analysis will be performed 133 according to the intention to treat principles. If the number of drop-out or missing data 134 exceeds 15%, imputation will be performed 135

Outcome assessment 136
In each hospital, non-blinded coordinators (not surgeons) will ensure that the questionnaires 137 are completed at baseline, 3 months, 12 months, 2 years, 5 years and 10 years. (ZCQ-score), a ten point Numeric Rating Scale (NRS) for low back pain and for leg-pain, and 153 a global perceived effect scale. Here changes at 2, 5 and 10 years will be analysed. 154 The EQ-5D is a generic measure of health-related quality of life. Five domains are rated: 155 mobility, self-care, activity, pain and anxiety, each by three response categories to provide a 156 utility index ranging from -0.59 (worst possible) to 1.0 (best possible). In addition EQ-VAS 157 provides a single score of the patient`s health condition. EQ-5D is validated for the 158 Norwegian population. Despite its large measurement error it is often used in research for 159 spinal conditions [9, 10]. The ZCQ is a disease specific questionnaire for lumbar spinal 160 stenosis [11]. It includes symptom severity, physical activity and patient satisfaction during 161 follow-up. The global perceived effect scale is a seven point scale, which is recommended for 162 clinical trials of chronic pain conditions [12]. It has six response categories: 1 = completely 163 recovered, 2 = much improved, 3 = slightly improved, 4 = no change, 5 = slightly worse, 6 = 164 much worse and 7 = worse than ever. All questionnaires are validated for lumbar spinal 165 stenosis patients [10,11,13] and are in close accordance with recommended PROMs for the 166 study of low back pain conditions [14]. 167 We will also compare the proportion of patients classified as success, between the groups. 168 Based on change in ODI score after the operation, the patients will be dichotomized into 169 success and non-success groups. A success is defined as a patient with an improvement in 170 ODI score of at least 30%. This value is based on a national register study from the 171 Norwegian Registry for Spine Surgery (NORSpine), and is also in accordance with 172 recommendations from the IMMPACT group, when comparing clinical effect between groups 173 [15]. 174 The local study coordinator will record complications and adverse events, length of hospital 176 stay, duration of surgery, blood loss and the need for blood transfusion. 177 178 Radiological evaluations: 179 Radiological evaluations will be performed by independent investigators. 180 Preoperative and 3 months postoperative MRI scans will be evaluated to measure the extent 181 of decompression by calculating the change in dural sac cross-sectional area at the most 182 stenotic level (square millimetres [16]. The association between the increase in dural sac 183 cross-sectional area and clinical improvement (primary outcome) will be assessed. We will 184 compare the increase in dural sac cross-sectional area in patients who achieve a minimal 185 clinically important change of the ODI with those who do not. monitor, without influence on the scientific work, will be responsible for the monitoring. Due 194 to the nonregulated ICH GCP guideline for this trial (not including drug intervention) the risk 195 and safety will be safeguarded at the same level as data quality. All informed consent forms 196 will be checked, and all registrations of serious events will be monitored. According to the 197 monitoring plan selected variables will be checked. All hospitals will be visited regularly. 198 Adapted versions of the 'Investigator's Site File (ISF)' and the 'Trial Master File (TMF)' will 199 be checked for essential documents during the trial. Queries and deviations will be recorded 200 and reported, and the coordinators at responsible hospitals have two months to send a written 201 report with the required corrections to the monitor. All deviations from the protocol will 202 subsequently be recorded at the 'Note to file form'. Recorded deviations will be presented in All statistical analyses will be performed by a statistician blinded to the treatment given. 214 Since this is a large randomized controlled trial, we will likely not need to adjust for potential 215 confounding factors. However, if descriptive statistics reveal that there are imbalances in the 216 study groups we will make comparisons with proper adjustments for such imbalances. 217 218 When comparing changes in outcomes from baseline to a fixed time point (e.g. 2, 5 or 10 219 years) we will use ANOVA combined with post-hoc testing for continuous outcomes. For 220 categorical outcomes we will use standard chi-square analyses. If there is a need for 221 confounder adjustment we will estimate the proper regression models dependent on the level 222 of measurement of the dependent variable. Multilevel models might be used to analyse 223 repeated measures data. 224 225 226 When trying to identify factors that could serve as predictors for outcome we will estimate 227 multivariable regression models. The following factors are a priori considered to be the most 228 relevant: treatment group, age, sex, BMI, smoking, preoperative dural sac cross-sectional area 229 (in mm²), percentage achieved decompression, preoperative ODI-score and preoperative 230 NRS-score for leg pain and back pain. Bayesian information criterion (BIC) might be used as 231 a statistical tool to select the best performing model.        Comparison of different surgical treatments for Lumbar Spinal Stenosis. 5 A randomized controlled trial comparing the clinical and radiological 6 results using "spinous process osteotomi", "bilateral laminotomy" and 7 "unilateral laminotomy with crossover".
This is planned to be a multicenter study with 15 participating Norwegian spine centers. The 10 departments of orthopedics at Oslo University hospital and Kysthospitalet i Hagevik, 11 Haukeland University hospital, will be the principal investigators. Two MDs are connected to 12 the study as doctoral students. 13 14 involves a bulging disc, hypertrophy of the ligamenti flavi and hypertrophy of the facet-joints. 20

Background, introduction and hypothesis
The symptoms are most commonly neurogenic claudication, or sciatic pain, and Low Back 21 Pain. The most stenotic part is at the disc-level. Several studies with long-term follow up have 22 been performed to evaluate surgical treatment versus non-surgical treatment [1,20,36], and 23 they conclude that surgical intervention gives superior results. The surgical procedure 24 performed in these studies is a laminectomy. A Cochrane review article also concludes that 25 surgery is superior to non-surgical treatment [12]. 26 The surgical treatment of LSS is to perform a decompression. To achieve relief of the 27 stenosis, one must remove the flaval ligaments, and the lamina (whole, or parts of it) at the 28 disc level, and carry out a partial medial facetectomy to decompress the nerve-root in the 29 lateral recess. The traditional operation is a laminectomy, which involves the total removal of 30 the lamina and the flaval ligaments, with a medial facetectomy. This procedure also removes 31 much of the posterior ligamenture of the spine, and is considered to be a relatively invasive 32 procedure; it is also prone to postoperative instability of the operated spinal segment. In recent 33 decades bilateral laminotomy has become an alternative to a laminectomy [2,19]. This is a 34 less invasive procedure, in which parts of the lamina are retained, and the interspinal and 35 supraspinal ligaments are mostly kept intact. This procedure is therefore considered to be 36 better at preserving stability. More recently a unilateral decompression has been described 37 [21], referred to as a "bilateral laminotomy with crossover technique". The access to the 38 spinal canal is unilateral, and the ipsilateral side is decompressed as usual, whereas the 39 contralateral side is decompressed using a crossover technique. This has theoretical 40 advantages by preservating more of the paraspinal muscles, ligaments and bony structures. A 41 third alternative is to do a spinous process osteotomy [13]. This procedure also involves 42 unilateral release of the paraspinal mucles, and an osteotomy of the spinous process at the 43 Therefore, we plan to perform a multicentre, randomized trial to compare the clinical and 54 radiological results of "spinous process osteotomy" (SPO), "bilateral laminotomy" (BL), and 55 "unilateral laminotomy with crossover" (UL). 56

Clinical results 57
The main goal of this study is to identify which method gives the best long-term clinical 58 results, including the most pain reduction and the best functional results. The patient's own, 59 self-evaluated assessment of the surgical procedure is the most important outcome. Today, 60 there are several questionnaires that are used to map the clinical status of the patient. The 61 patient answers a questionnaire before the surgical procedure, and the same questionnaire is 62 repeated after surgery, to monitor the effect of the surgical treatment. We have decided to use 63 the same questionnaires that the Norwegian Quality Register for Spinal Surgery (NORSPine) 64 uses, and thus cooperate with the register when collecting data postoperatively. These considered more specific to spinal stenosis, and has also been evaluated in several studies 72 [25,33]. 73 In recent decades the ODI has had a special status within spinal care. There are hardly any 74 studies where clinical results are monitored without this questionnaire. In this questionnaire 75 '0' is the best status, and '100' is the worst status. The desired effect of surgical treatment is a 76 significant decrease in the patient's ODI-score. We have therefore chosen the ODI as a 77 primary outcome of this study. This application aims to reveal the 10 year improvement in 78 ODI related to the three different surgical methods. When using the ODI as a primary 79 outcome, it is important to decide beforehand what is to be regarded as a significant 80 improvement in the score following surgical treatment. This is now discussed as the Minimal 81 Clinically Important Difference (MCID). MCID is measured as the difference in the mean 82 ODI-scores of patients that score "some improvement" compared with those who score "no 83 improvement". This value has been found to be 10-12 ODI-points [5,14,23 mm². In the studies quoted, moderate stenosis was defined as a DSCSA between 70 and 100 92 mm², and significant stenosis as an area below 70 mm². One study found a significant 93 correlation between DSCSA and clinical symptoms, but one other study did not confirm this. 94 Both of these studies included a relatively small number of patients (n=109) [29]. Our 95 previous study [16], has shown a positive correlation between an increase in the DSCSA and 96 improvement in clinical symptoms. To our knowledge this has not been documented before. 97 At present, we cannot identify a threshold value in the spinal canal area that gives clinical 98 symptoms. Neither can we identify a minimal increase in area that is needed to give long-term 99 relief of symptoms. We hope to be able to answer these questions. Our previous study showed 100 an increase in DSCSA using SPO of 101%, up to an average area of 161 mm² postoperatively 101 [16]. Dalgic et al compared "bilateral laminotomy" and unilateral laminotomy with 102 crossover", and found a postoperative area of 126 mm². This is significantly lower than the 103 area found when SPO is used. We are therefore planning to measure DSCSA before and after 104 surgery, to see whether there are differences between the three different surgical methods. We 105 also want to investigate whether there is a correlation between achieved increase in area and 106 the patients' self-evaluated outcome of surgery. 107 108

Operative differences 109
As mentioned earlier, there are differences between the three surgical methods. We are 110 planning to observe whether there are significant differences between the methods as regards 111 operation time, amount of peri-operative bleeding, peri-operative complications and length of 112 hospital stay. And, we will also examine whether one of the three surgical procedures have an 113 increased rate of reoperations in the observation period. Thereby we are able to monitor costs 114 of each of the three methods. 115 We also want to find out whether any of the surgical methods preserves stability better than 116 the others. Preoperative standing x-rays, with lateral view of flexion-extension of the 117 lumbosacral spine are therefore important as a baseline. In the follow up period we will repeat 118 these x-ray investigations and thus find out if any of the methods is more prone to 119 postoperative instability. University Hospital as a consultant surgeon. Hellum is involved in the preparation of the 131 study, and will be a co-author of any publications. He is responsible for the Norwegian Disc-132 prosthesis study, and is the corresponding author of these publications. Aim of the study 159 We are planning a multicentre, randomized controlled trial to compare the clinical and 160 radiological results of three different surgical methods for surgical treatment of LSS. 161 The main objective of the study is to compare the clinical results: 162

Primary outcome: 163
The primary outcome is a measurement of the decrease of ODI (version 2.0) in the three 164 different surgical groups. This will be done by calculating the mean ODI improvement in each 165 group after two years, compared to the two other groups. 166 As a supplement to the primary outcome we will also compare the proportion of responders 167 between the SPO group, the BL group and the UL group, 24 months after the operation. 168 Based on the size of change in ODI score from before to after the operation, we will 169 dichotomize the patients, in a responder group and in a non-responder group. The threshold 170 value for being categorized as a responder will be computed in a planned study based on data 171 from The Norwegian Registry for Spine Surgery (NORSpine). 172

173
We are planning to follow the patients for 10 years. 174

Secondary outcomes: 175
The secondary outcomes are measurements of change in EQ-5D, SSS-score, NRS for LBP, 176 NRS for leg-pain, and self-evaluated effect of surgery in the three different surgical groups. 177 This will be done by calculating the average values after two years, in each group, and 178 comparing them to the two other groups. 179

Radiological evaluations: 180
There are several radiological questions we want to investigate.  We have prepared a separate patient information protocol, which is not presented here. 225 226 Research methods and analysis: 227 We have prepared a separate operation protocol, which is not presented here. 228 Statistics: 229 We have previously discussed that MCID is around 10-12 ODI-points for this group of 230 patients. This value indicates the individual effect of treatment versus no-treatment. There is 231 today no consensus in the literature of a clinically important difference between treatment 232 groups. We have therefore decided that we want to reveal a difference of 7 ODI-points 233 between the groups. A statistician has calculated a difference of 7 ODI-points between the 234 treatment groups, and in a regular superiority design. Standard deviation is set to 18. Because 235 we have three groups we want to compare the significans-level is lowered from 5% to 2% (p 236 = 2). The analyzes in the treatment group is done in the following way: -A compares with B, -237 A compares with C, -B compares with C.). With 80% power , significans level of 2%, and 238 dropout of 30% n in each group is estimated to 187 (131 + 56). 30% dropout is a high 239 number, but we are planning to monitor the patients for 10 years, so we must anticipate a high 240 dropout number. If we have problems to achieve 187 in each group, we will settle with 15% 241 dropout. This gives 155 patients in each group. 242 We will also perform descriptive analyzes. The superiority analyzes is conducted in a One-243 way Anova system and Post Hoc. 244 Therefore we are planning to include 561 patients over a 2-3 year period. The analysis is 245 conducted after the Intention To Treat Principle (ITT). Crossovers will be analyzed in the 246 group they were originally randomized into. A sensitivity analysis will be conducted where 247 patient crossing over from one treatment to another will receive the last score before 248 crossover. Dropouts will be registered with the last value before dropout (last value carried 249 forward). 250 Department of Clinical Research Support, at Oslo University Hospital will provide the block 251 randomisation, and is performed within a computer program called Medinsight. The 252 randomization process cannot be influenced by participating hospitals, and will be performed 253 within 1 month before treatment. 254 Patient-data, and registration will also be handled by the computer programme called 255 Medinsight. 256 The research protocol will be registered in www.clinicaltrials.gov. 257 The reporting of the trial will be bases on an adapted CONSORT (Consolidated Standards of 258 Reporting Trials) checklist for reporting in randomised trials [18]. 259 260 261 262

Supplementary analysis I 263
In addition to the main analysis we will investigate whether following predictors will be 264 associated with the success-rate after 24 months: of patients will be registered in an observation cohort. We will collect base-line 277 questionnaires of these patients, and perform radiological and clinical assessment. We plan to 278 follow this observational cohort over 10 years. Thereby we will be able to perform a predictor 279 analysis of the patients that experiences deterioration of the clinical condition. They will 280 receive and sign an informed consent form. 281 282 283 284

Supplementary analysis III 285
We have previously conducted a study that showed 44% union-rate of the SPO (accepted for 286 publication in Asian Spine Journal). We are planning to repeat this investigation, and evaluate 287 whether the osteotomy of the spinous process has negative implications for the clinical 288 results. A CT scan will therefore be performed two years postoperatively to evaluate the union 289 rate at the operated level. 290 Ethical considerations: 291 The patients will receive thorough information both orally and in writing, so that they can 292 give informed consent to the study. All three surgical procedures are well established 293 treatments in Norway today. There is no currently available evidence to suggest that one 294 method is significantly better than the others. In our opinion there are no ethical problems in 295 relation to this study. 296 We will follow the Helsinki-declaration, ICH Guideline for Good Clinical Practice (ICH 297 GCP). 298 We have recived an approvel from the Norwegian Commitees for Medical and Health 299 Research Ethics (2011/2034). 300

305
-MRI-scan for measuring preoperatively DSCSA. 306 -Following completion of the consent form the following PRO-data will be collected: ODI, 307 ZCQ, EQ-5D, VAS for back-and leg-pain demographics, work status, smoking status and 308 use of analgesics. 309 310 During the hospital stay: 311 312 -The surgeon will register radiological findings, ASA-classification, operation method, 313 operation time, level and blood loss, and per-operative complications. 314 -Complications will be registered during follow-up by an independent nurse or physician. 315 316 3 months postoperatively (± 2 weeks): 317 -The following PRO-data will be collected: Working status, ODI, ZCQ, EQ-5D, VAS for 318 back-and leg-pain, patient satisfaction with treatment outcome, and patient satisfaction with 319 treatment care. 320 -MRI scan to measure DSCSA postoperatively. 321 322 12 months postoperatively (±1 month): 323 324 -The same PRO-data as at 3 months will be collected. 325 326 2 years postoperatively (± 2months): 327 328 -The same PRO-data as at 3 months will be collected. 329 Skeletal x-rays: standard images will be performed. 330 CT-scanning for examining the union of the spinous process (only the SPO-group). 331 332 5 years postoperatively (± 3 months): 333 334 -The same PRO-data as at 3 months will be collected.