WHAT IS ALREADY KNOWN ON THIS TOPIC

Patients with a meniscal tear are at a higher risk of developing knee osteoarthritis (OA), and surgery to the meniscus might increase the risk of knee OA in patients with degenerative tears. However, the influence of treatment strategy (surgical or non-surgical) on structural knee joint changes and the later risk of knee OA in young patients with meniscal tears is not known.

WHAT THIS STUDY ADDS

Initial treatment strategy (meniscus surgery or supervised exercise and education) did not influence short-term structural knee joint worsening in young adults with meniscal tears, as worsening of structural knee damage on MRI at the 2-year follow-up was limited and similar between treatment groups. In addition, early meniscal surgery was not superior to exercise therapy and education with optional delayed surgery in improving patient-reported outcomes (PROMs).

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

The findings suggest that both surgical and non-surgical treatment strategies yield similar two-year PROMs and structural knee joint changes in young adults with a meniscal tear. Studies on long-term structural damage are needed to investigate long-term knee-joint changes and possible approaches to prevent OA development.

Introduction

Meniscal tears in young adults have typically been treated with arthroscopic surgery. Recently, two randomised trials comparing early meniscal surgery to exercise therapy and education with optional delayed surgery (if needed) reported clinically relevant improvements in patient-reported outcomes (PROMs) for both treatment strategies, with negligible between-group differences.1 2 Furthermore, in the two trials, many patients randomised to exercise (84% at 12 months and 59% at 24 months) did not undergo surgery during follow-up.1 2

The risk of developing knee osteoarthritis (OA) is considerably elevated (up to sixfold higher) for knee injury and meniscal tear patients.3 4 However, whether the development of structural knee joint changes and the later onset of knee OA is affected by the initial treatment strategy (surgical or non-surgical) in young patients with meniscal tears is unknown. On the one hand, exercise therapy could potentially cause further damage to an already injured joint. On the other hand, arthroscopic partial meniscectomy (APM) has been reported to be associated with an increased risk of MRI-defined cartilage damage,5 a higher risk of developing radiographic knee OA,6 and progression of MRI-defined OA features7 in middle-aged and older patients with degenerative meniscal tears.

Since OA is the most common joint condition, affecting more than 500 million people worldwide,8 prevention is important to reduce the burden of OA.9 10 Thus, knowledge of how treatment strategy impacts structural knee damage and long-term pain and function in young patients is essential as it may help prevent the development of OA.

Therefore, in this secondary analysis of the ‘Danish RCT on Exercise vs Arthroscopic Meniscal surgery for young adults (DREAM) trial’—a multicentre randomised controlled trial (RCT) comparing two treatment strategies for meniscal tears in young adults,1 we aimed to investigate potential differences in MRI-defined structural knee damage and PROMs from baseline to 2 years in young adults with meniscal tears treated with either early meniscal surgery or exercise therapy and patient education with optional delayed surgery.

MethodsEquity, diversity and inclusion statement

Our author team included senior and less-experienced investigators of different genders from various health disciplines. The study population was recruited from public hospitals across Denmark, increasing the results’ diversity and generalisability.

Study design and participants

This study was a secondary analysis of the ‘DREAM trial’.1 Participants were recruited from seven orthopaedic departments across Denmark from January 2017 to December 2019. Inclusion criteria were 18–40 years of age, knee pain, clinical history and symptoms consistent with a meniscal tear (confirmed on MRI) and deemed eligible for meniscal surgery (APM or repair) by an orthopaedic surgeon. Inclusion was not limited to traumatic symptom onset, a specific symptom duration or specific types of tears. However, patients with a congenital discoid meniscus or clinical suspicion of displaced bucket handle tear (acute locking of the knee or extension deficit) confirmed on MRI were excluded. If there were no symptoms from acute locking of the knee, all types of tears were included. Other exclusion criteria were prior surgery of the affected knee, fracture of the affected extremity in the previous 12 months, complete rupture of any knee ligament or participation in supervised exercise therapy within the last 3 months. The participants were followed up at 3, 6 and 12 months with online questionnaires. Patients consenting to participate in the 2-year MRI follow-up assessment also received a 2-year follow-up questionnaire. In this study, we only included participants with 2-year follow-ups.

Interventions

Patients randomly assigned to receive meniscal surgery underwent APM or meniscal repair. The operating surgeon determined the type of surgery to increase generalisability to clinical practice. Patients allocated to exercise therapy and patient education participated in a 12-week programme consisting of two times a week supervised exercise sessions. Patient education was delivered at the beginning and the end of the programme by trained physical therapists.11 Detail of the programme is presented in the study protocol.12

Patients and public involvement

Patients and clinicians were involved in the development of the design of the intervention as described in the pilot paper11 . No patients or the public were involved in the planning of this secondary analysis.

OutcomesMRI

Baseline and 2-year follow-up MRIs were performed with a minimum of a 1.5-T scanner using the individual radiology departments’ protocol for suspected meniscal tears. All protocols included sagittal, axial and coronal sequences with and without fat suppression. The main outcomes were the between-group difference in worsening in MRI-defined structural damage (ie, new or progressed cartilage damage, osteochondral damage and osteophytes) from baseline to 24 months, assessed using a slightly modified Anterior Cruciate Ligament OsteoArthritis Score (ACLOAS).13 ACLOAS is a semiquantitative MRI-based scoring system that includes structural features relevant to acute injury, degenerative incident features and longitudinal follow-up of structural OA features. Cartilage damage is scored from 0 to 6 in 14 subregions. Osteochondral damage is scored from 0 to 4 in 14 subregions, and osteophytes are scored from 0 to 7 in 12 locations. The scores for cartilage damage, osteochondral damage and osteophytes were summed on the knee level. Thus, the worsening of individual MRI features includes new or progressed damage in one subregion and progression in the number of subregions affected. We did not include meniscal damage worsening in this study since one of the interventions (APM) specifically altered the meniscus morphology.

We also assessed for any bone marrow lesion (BML), knee joint effusion/synovitis, and anterior cruciate ligament (ACL) and collateral ligament status. A BML was defined as a reticular ill-defined hyperintense lesion on water sensitive fat suppressed sequences and was scored in 15 subregions (present/absent), and summed on the knee level. Effusion/synovitis or surrogates for this were scored in two locations: capsular distension in the suprapatellar recess (0–3) and signal alterations in Hoffa’s fat pad (0–3). Since effusion is often located in other recesses besides the suprapatellar recess, we added an assessment of overall effusion/synovitis, including all recesses. We used it as the primary effusion/synovitis score. Furthermore, we recorded the presence of Baker’s cysts and any possible change for overall effusion/synovitis and Baker’s cysts (decreased, unchanged, increased). An experienced musculoskeletal (MSK) radiologist (DIR) scored all the MRI scans, blinded to clinical information. The baseline and follow-up MRIs were assessed pairwise and unblinded to the sequence to maximise sensitivity to detect change.14 We assessed inter-rater reliability by another experienced MSK radiologist (EYK) independently reading 20% of the scans. Both MRI assessors were trained and supervised by a third experienced professor in MSK radiology (MPB) to reach a consensus before the scoring. The overall percentage of agreement and prevalence and bias adjusted kappa were calculated, except for the meniscal damage, where we used weighted kappa due to the high prevalence of lesions (all had lesions). Most scores had an inter-rater agreement of >95%, and the reliability was substantial (range, 0.61–0.8) or almost perfect (range, 0.81–1.0) according to the interpretation by Landis and Koch.15

Patient-reported outcomes

We assessed the between-group difference in change in the mean score of the Knee Injury and Osteoarthritis Outcome Scores (KOOS4) from baseline to 2 years (ie, primary outcome in the main study). The KOOS4 is the mean of four of five subscale scores, including pain, symptoms, function in sport and recreation, and quality of life, and ranges from 0 to 100, with lower scores indicating worse symptoms, function and quality of life.16 17 Additional PROMs were the between-group difference in change in the individual KOOS subscales and the Western Ontario Meniscal Evaluation Tool (WOMET), a meniscus-specific, valid and reliable PROM measure (converted to scores from 0 to 100, with lower scores indicating a worse quality of life).18 19

Statistical analysis

We used descriptive statistics to describe baseline characteristics and frequencies of MRI features.

We conducted all analyses according to the intention-to-treat principle, with patients distributed according to the treatment arm they were randomised to (irrespective of the treatment received). In addition, an as-treated analysis was performed. In this analysis, the patients undergoing meniscal surgery up until 2-year follow-up were included in one group. Those undergoing exercise and education (irrespective of their compliance with the exercise), who had not received delayed surgery, were included in the other group. Chi-square or Fisher’s exact test, as appropriate, was used to compare the frequency of participants with MRI-worsening between groups. A detailed analysis of the distribution of ACLOAS grades between the groups was irrelevant since the number and severity of the findings were low.

For the PROMs, we used the same analytical approach as in the primary reporting of the DREAM study1: a linear mixed model with time (baseline, 3, 6, 12 and 24 months as discrete variables), treatment group (surgery or exercise), and the interaction between time and treatment group as fixed effects constraining the difference between the arms to 0 at baseline (ie, adjusting for baseline imbalance). The model was adjusted for the randomisation stratification factors (centre and sex) and age. A patient-specific intercept and slope were added as random effects to accommodate within-person measurement dependence. A common error variance was assumed for all follow-up time points and treatment arms, although error variance can differ at baseline. The assumptions for model validity were checked using scatter plots of the residuals versus time and two-dimensional scatterplots of the best linear unbiased prediction of the random effects. A 95% CI excluding 10 points or more in KOOS4 was interpreted as no clinically meaningful difference. No imputation was performed as the mixed model included all patients.20 All statistical analyses were performed using Stata V.17.0 (StataCorp).

Results

Of 121 patients randomised, 82 (68%) consented to participate in the 2-year follow-up (39 from the surgical group and 43 from the exercise group). After several attempts to contact patients or reschedule appointments for MRI scans, 78 provided PROMs, and 77 provided MRI data. Thus, nine patients only had either the MRI or the PROMs data. Baseline characteristics were similar between patients participating in the follow-up and those who did not (online supplemental table S1). The mean time (SD) from baseline to 2-year follow-up was 28 months (4.4). The patients’ mean age (SD) at baseline was 29.5 (6.6). About a quarter were female; most patients were active and had a symptom duration of 0–12 months and a traumatic or semi-traumatic symptom onset. Overall, baseline characteristics in the two groups were similar, except for age and slightly better KOOS at baseline in the surgery group compared with the exercise and education group (table 1). This imbalance was consistent with the primary report of the DREAM study.1

Table 1

Baseline characteristics of the DREAM-participants with 2-year follow-up

At 12-month follow-up, 16 of the 61 patients (26%) randomly assigned to exercise and education had crossed over to surgery. Between 12 and 24 months, two additional patients from the exercise and education group had knee surgery, while four in the surgical group had a second knee surgery (figure 1). None of the participants reported any contact with general practitioners or hospitals due to a new knee injury between 12-month and 2-year follow-ups.

Figure 1

Flow of study participants. PROMs, patient-reported outcomes.

Baseline MRI findings

Baseline findings are summarised in table 1. Of the meniscal tears, 65% (n=53) were isolated medial, 30% (n=25) isolated lateral and 4% (n=3) had both medial and lateral tears. For medial meniscal tears, 34% (n=28) were bucket-handle or complex followed by 26% (n=21) horizontal, whereas radial and vertical tears were the most frequent lateral tears, 15% (n=12). There was no essential difference in the type of tears between the groups assessed with ACLOAS. In one participant (1%), the baseline scan was described without a meniscal tear. However, since the patient was included in the study, the clinical MRI report and the including surgeon must have assessed the scan differently, and thus we decided to also include the patient in this secondary analysis. Besides meniscal tears, there were few MRI-defined baseline changes with a similar distribution between groups. One patient fulfilled the suggested criteria for MRI-defined OA.21 22 Seven patients had ACL or medial collateral ligament changes, all related to mild sprains. The most common baseline MRI findings were knee joint effusion/synovitis (n=39), Hoffa synovitis (n=21) and BMLs (oedema) (n=16) (table 1).

MRI-defined damage worsening

Cartilage damage worsening was observed in 9% (n=7) of all patients with MRI and osteophytes in 3% (n=2) of patients. Details of the type of worsening (progression in one subregion or in regions affected are listed in online supplemental table S2). Still, only one patient had MRI-defined OA.21 22 We observed similar worsening of cartilage damage (p=1.000) and osteophytes (p=0.203) between treatment arms, and in the as-treated analyses. BML, knee effusion/synovitis, Hoffa synovitis and Baker’s cysts were unchanged in 73%–90% of the patients and here too we found no essential differences between the groups for these findings (table 2).

Table 2

MRI-defined* changes from baseline to 2-year follow-up

Patient-reported outcomes

We observed no essential difference in change between groups from baseline to 2 years in KOOS4 (table 3). The mean crude and adjusted between-group differences in change were −5.1 (95% CI −13.8, 3.7) and −1.4 (95% CI −9.1, 6.2) points (in favour of the exercise group), respectively. On average, most of the improvement was observed during the first 6 months of the trial. At 2 years the surgical group had improved by 16.4 (95% CI 10.4, 22.0), while the exercise group had improved by 21.5 (95% CI 15.0, 28.0). The individual KOOS subscales yielded similar results (table 3). For the WOMET, these improvements were 20.2 (95% CI 12.9, 27.5) for the surgery group and 26.7 (95% CI 18.6, 34.8) for the exercise and education group, with an adjusted between-group difference in change of −2.4 (95% CI −11.8, 6.9) from baseline to 2 years (table 3). Likewise, the as-treated analyses yielded no essential between-group differences in change (figure 2B; online supplemental table S3).

Figure 2

Mean unadjusted Knee Injury and Osteoarthritis Outcome Scores (KOOS4). (A) Patients distributed according to the treatment arm they were randomised to (irrespective of the treatment received). (B) In this analysis, the patients undergoing surgery up until 2-year follow-up were included in the surgery group. Those undergoing exercise and education (irrespective of their compliance with the exercise) who had not received delayed surgery constituted the other group. The error bars indicate 95% CIs.

Table 3

Patientreported outcomes at 2-year follow-up (main analysis)

Discussion

We found that the risk of 2-year worsening of structural damage was low and similar in patients undergoing early surgery and patients undergoing exercise and education with optional delayed surgery. Likewise, the as-treated analysis comparing early or delayed surgery to exercise and education did not reveal any essential differences. Moreover, we found that early surgery was not superior to a strategy of exercise and education with optional delayed surgery in improving pain, function and quality of life at 2 years, similar to the 12-month results.1

Our study, including young patients, found less structural worsening over 2 years than in previous studies, including older patients with degenerative meniscal tears.6 7 23 Moreover, in contrast to an earlier trial comparing APM with physiotherapy for degenerative meniscal tears in older patients with radiographic knee OA that reported significantly greater MRI-defined worsening of cartilage damage and osteophytes in the APM group compared with the exercise group,23 we did not detect a difference between groups. Whether the previous finding of more severe structural worsening in knees with OA relates to patients’ age, the degree of structural damage at baseline or other factors, like the type of meniscal repair or resection, tear/injury or symptom onset, is unknown. Most patients (80%) in our study had traumatic or semi-traumatic symptom onset. However, subgroup differences can exist for those with gradual onset that are most likely degenerative tears. Regarding the influence of age, Roos et al 24 reported that for patients who sustained an isolated meniscus injury between the ages of 17 and 30, the average time until the development of radiological signs of OA (on radiographs) was about 15 years. In contrast, the corresponding time interval was only about 2 years for those over age 30 with the same injury, indicating a strong association between age and the worsening of structural damage.

Comparing the structural worsening in our study to studies of young patients with ACL injuries, we find differences that indicate an association between the baseline damage and worsening (short-term or long-term). The ACL studies25 26 report more severe structural baseline knee damage (eg, cartilage and osteochondral lesions) and more structural worsening over time than our study, where the participants had isolated meniscal tears.

Recent systematic reviews report no additional clinically relevant benefit of APM over placebo surgery or exercise therapy in middle-aged and older adults with degenerative meniscal tears.27–29 Based on this evidence, clinical guidelines generally recommend against arthroscopic surgery and recommend non-surgical treatment for older patients with degenerative tears.28 However, young patients with meniscal tears are usually offered surgery.30 This secondary analysis of the 2-year outcome from the DREAM trial1 confirms the primary 12-month reporting and yields similar results as the 2-year reporting from the The Study of Traumatic meniscal tears: Arthroscopic Resection vs Rehabilitation (STARR) trial.2 Early surgery was not superior to exercise and education, with optional delayed surgery for treating isolated meniscal tears in young adults.

Limitations

Thirty-two per cent of patients were lost to follow-up. Although we performed analyses according to the intention-to-treat principle, the assumption of including all patients was violated. The resulting direction of this bias is unknown. Nevertheless, we observed no difference in baseline characteristics between patients participating and those lost to the 2-year follow-up. Also, the as-treated analyses should be interpreted with caution due to the low number of patients in each group in these analyses. The MRI scans were conducted at seven different departments, which may result in a difference in the visualisation of the findings between the departments. Since our primary MRI outcome was worsening structural damage, the scans were assessed pairwise and unblinded to the sequence to maximise sensitivity for change,15 mitigating some variations caused by differences between departments.

Clinical and research implications

These findings suggest that both treatment strategies are equally effective in relieving symptoms in young patients and highlight the importance of including the patient’s treatment preferences when deciding on a treatment strategy. Specific tear types may benefit more from one treatment than the other, but future studies on effect modification are needed to provide more insight into this important clinical issue. Moreover, studies with longer follow-ups are needed to investigate long-term knee joint changes and possible approaches to prevent OA development.

Conclusion

Our results suggest that in young adults treated for isolated meniscal tears the 2-year worsening of MRI-defined structural damage indicative of knee OA is low and similar between treatment strategies (early surgery vs exercise with the option of later surgery). In addition, early meniscal surgery is not superior to exercise and education in improving2-year PROMs. These findings are important in the decision-making between patients and clinicians on the treatment choice, as both strategies appear viable.

Data availability statement

Data are available upon reasonable request.

Ethics statementsPatient consent for publication

Consent obtained directly from patient(s).

Ethics approval

This study involves human participants and the study was approved by the Regional Committees on Health Research Ethics for Southern Denmark (S-20160151) and the Danish Data Protection Agency (University of Southern Denmark, 16/45314). Participants gave informed consent to participate in the study before taking part.

Acknowledgments

The DREAM study group (a complete list of contributors to the DREAM Study Group is provided in online supplemental file).