How to Best Treat Anterior Cruciate Ligament Injuries

Despite years of research on treating ACL injuries, it remains uncertain who should undergo reconstructive surgery and who should not.

Our project aims to provide definitive answers to these important clinical questions affecting more than 4,000 patients each year in Norway. We will provide this information by conducting a large comparative effectiveness trial of standard treatments for ACL deficiency. This RCT exploit the infrastructure of the Norwegian National Knee Ligament Registry (NNKLR), which uniquely include non-surgically treated ACL-tears and has almost complete coverage of surgical reconstructions. Our multi-center register-RCT design considerably reduces costs and enables us to more rapidly include many more patients. Consequently, the design and study size will give the results excellent external validity.

Surgery or not? It is estimated that 50% of patients with an ACL tear are treated surgically with a reconstruction.[1, 2] The majority of patients are young and active.[3] The general consensus in Scandinavia is that patients who manage to regain acceptable knee function after active rehabilitation alone should not initially have surgery. This treatment algorithm is supported mainly by only one high-quality randomized controlled trial.[4, 5] Prospective studies have reported that the functional recovery, the long term patient outcome and the radiographic sign of osteoarthritis are similar between patients treated surgically or non-surgically.[6-8] However, conclusion about cause and effect cannot be made due to the study designs. A recent Cochrane review concluded that there is a low-quality evidence that there was no difference between surgical and non-surgical treatment, and studies with high external validity is lacking.[9] In the debate of early surgery or not, the risk of subsequent meniscal and chondral injury in an unstable knee is often highlighted by promoters for early surgery. However, the studies behind this argument reports on the correlation between the timing of surgery and concomitant meniscal and cartilage injuries rather than on whether the initial treatment chosen was surgical or non-surgical.[10] This introduces a selection bias as a patients having surgery late may not be representative of all non-surgically treated patients. The findings in a systematic review on this topic by authors in the project group high-light that the uncertain estimate of new meniscal tears after ACL injury should not guide clinical practice for all patients.[11] Even though joint stability often is achieved with surgery, the rate of return to pre-injury activity level after surgery is reported less than 50%, and no consistent results in reduction of the risk of later osteoarthritis have been shown.[12] Admittedly, the risk of serious complications after ACL reconstruction is in general low, however still potentially devastating to the knee function.[13, 14]

About the Norwegian National Knee Ligament Register

The NNKLR was established in 2004 as the world´s first national cruciate ligament register[15]. The NNKLR assesses the outcome based on the Knee Injury and Osteoarthritis Outcome Score (KOOS) which is the gold standard validated patient reported outcome measure (PROM) [16] and by recording revision surgeries or other surgeries to the index knee. Since 2004 cruciate ligament reconstructions performed in Norway have been reported to the register with subsequent registration of revision surgeries and with input of PROM at 2, 5 and 10 years postoperatively. Just below 3000 patients are included yearly in the register. This has provided a unique resource for research and quality assessment, detecting inferior surgical procedures and providing prognostic information to guide practice.[17] Registries have had two major weaknesses limiting overall research impact:

1. Non-operatively treated injuries have not been included. This selection bias severely skews the evidence base for treatment guidelines.[17, 18,19] The NNKLR has recently piloted a project trying to mend this weakness.[20] The license for NNKLR issued by the Norwegian Data Inspectorate has been amended to allow for inclusion of patient at the time of ACL injury, allowing for follow-up within the register for the patients not undergoing surgery.
2. The design is fundamentally that of a prospective cohort. This does not allow researchers to make firm conclusions about differences in outcome between treatments. This major weakness of all national ACL registers is due to the inherent bias of being mainly observational.

Register-RCT The classic RCT remedies confounding bias by randomly allocating patients to each treatment ensuring equal distribution of measured and, importantly, unmeasured characteristics. However, RCTs often have strict inclusion and exclusion criteria leading to diminished generalizability to the general population. One may ask what good are RCTs if they can answer questions regarding only a tiny fraction of the population.[21] They are often expensive and time-consuming, limiting the possible amount of patients included. Recently, as an answer to these limitations, the implementation of randomization into registers has been proposed, and termed "the register-based RCT" (R-RCT).[22] The promise of this approach has been highlighted in high-ranking journals as the "next disruptive technology in clinical research".[21] Simply put, patients are included in a quality register, and the treatment is randomized. The outcome data is collected automatically through the register, cutting the costs and simplifying the follow-up. Analysis of the subsequent register data can conclude on effect estimates between treatments with the same level of evidence as those from regular RCTs, only with a greater external validity. The feasibility R-RCTs has been shown mainly in trials relating to interventions in cardiology and has only very recently been used in orthopedic trauma surgery[23-25].

Recruitment problems has been highlighted as a challenge in a previous RCT on ACL treatment.[26] The R-RCT mediates this by facilitating multi-center collaboration to ensure rapid inclusion of patients in the proposed trials.

Hypotheses, aims and objectives The overall aim of this project is to assess the effect of surgical treatment and non-surgical treatment of patients with a complete tear of the anterior cruciate ligament as assessed with the KOOS which is the commonly accepted gold standard patient reported outcome measure for knee injuries [16].

Primary objectives Investigate the outcome of surgical and non-surgical treatment of ACL-injury through a multi-center register-based RCT with outcome collection of primary endpoints solely performed by the NNKLR

Project methodology The NNKLR follows the patients until 10 years and this is the minimum time the project will follow the included patients. We include in the consent the possibility to contact all patients after the regular follow-up by the NNKLR as we plan to follow-up patients until 20 years after the injury and we also include in the consent the possibility of linking to for example the Norwegian Arthroplasty Register.

Project arrangements, method selection and analyses All participating centers already care for patients with cruciate ligament injuries and have all necessary equipment for this. The data collection at inclusion will be solely through a secure web-based system (Medisinsk Registreringssystem (MRS)- Helse Midt-Norge IT). During the planning and running of the trial the research group will work closely with the NNKLR.

Methods The NNKLR is a national quality registry. Reporting surgically treated patients to the register is mandatory for the surgeon. The surgeon report patient-specific data and surgical details on either a paper form or through a secure web-based interface at the time of inclusion. Information on implants used at surgery is registered with stickers provided by the manufacturers attached on the paper form, and by scanning the barcodes providing the same information when reporting electronically. Identical patient-specific data is collected at inclusion for the patients treated non-surgically. The web-based interface is maintained by the Medical Register System in the IT department of Central Norway Regional Health Authority. The NNKLR will be the main source of outcome collection in the planned trials. The patient reports PROMs at the time of inclusion and at 2-, -5- and -10-year follow-up, either through a paper-form or electronically. Clinical follow-up with subsequent primary or revision ACL reconstructions or other knee surgery to the index or contralateral knee are to be reported to the register by the surgeon.

Study design The study design is a register-based multi-center RCT. The structured rehabilitation will be identical in both groups and follow standardized goal-oriented guidelines.[27, 28] Randomization strategy: Randomization will be performed when the patients are included at first visit through the NNKLR embedded randomization module developed and implemented in this project. We will utilize centralized randomization stratified by participating centers[29, 30] Allocation concealment will be secured by requiring registration into the NNKLR and the R-RCT before allocation is revealed.

Sample size calculation: The sample size calculation will be based on a standard deviation of 20 for KOOS Quality of Life, a 90 % statistical power to detect a minimally clinically relevant difference of 10 between the treatment groups, an alpha level of 0.05 and allowing for a 20% drop-out rate.[5] In the non-surgically vs. surgical treatment RCT without stratification for age and activity level this will require at least 230 patients. Allowing for sub-group analysis on activity levels based on best estimates will increase sample-size to approximately 328 patients.

Statistical analyses: All subjects randomized will be included in the analysis. Investigation of the primary endpoint will be performed using analysis of covariance for the different treatment groups, stratified by site and adjusted and unadjusted for baseline KOOS-scores.[31] Both intention-to-treat and as-treated analyses will be performed. An absence of a clinically relevant difference will be assumed if the confidence interval of the difference includes differences of less than 10 units between groups. Secondary endpoints will be analyzed by analysis of covariance (KOOS) and chi-square tests (return to pre-injury activity level). Additional data collected outside NNKLR is clinical imaging (MRI/x-rays) taken in the course of treatment. Resource use (health economy) will be analysed through linkage to other registers (eg. NPR, IPLOS, Reseptregister) as stated in the consent and if needed supplemented with data from the electronic patient record.

Funding The project is fully funded through Program for klinisk behandlingsforskning (KLINBEFORSK) with a grant for the period of 2020 through 2024. There are no financial interests involved for neither the researchers nor the patients in this RCT.

User involvement We have teamed up with a named user.

Ethical considerations and data handling All participants must give written informed consent to be included into both the NNKLR an the register-based RCT. The NNKLR has a standing license from the Norwegian Data Protection Authority ref.nr. 16/01130-3/CDG covering both non-operatively treated and surgically treated patients included in the registry. The protocol for the register-based RCT and user-approved consent have been reviewed and approved the Regional Committees on Medical and Health Research Ethics (REC SØ).Before patient inclusion, the Ethical Boards at the including centres will need to review and approve the protocol (currently approved at Akerhus University Hospital).