Arthroscopic partial meniscectomy for degenerative tears: Where do we stand?

In this issues of Osteoarthritis and Cartilage, Gauffin and colleagues report the findings of a randomized controlled trial of arthroscopic partial meniscectomy (APM) vs. a physical therapy regimen in subjects with degenerative meniscal tears. This carefully executed study is an important addition to an evolving body of research on the appropriate role of APM and other arthroscopic procedures in patients with osteoarthritis. We review research in this area over the last two decades to put this trial into context.

The menisci are cartilaginous, semi-lunar shaped knee structures that confer important load-bearing and stabilizing functions. Like other cartilaginous intraarticular structures such as the labra of the shoulder and hip, the menisci are vulnerable to degenerative tears. The introduction of magnetic resonance imaging in the latter decades of the 20^th century facilitated non-invasive recognition of meniscal tear, revealing that over one-third of individuals greater than 50 years old and over three-quarters of persons with knee osteoarthritis have degenerative meniscal tears^1,2. The introduction of arthroscopic surgical techniques in the same era provided a minimally invasive means of performing surgery to resect meniscal tears. Indeed, by the turn of the century, arthroscopic partial meniscectomy (APM) for degenerative meniscal tear had become among the most commonly performed orthopedic procedures, with over 400,000 performed in the US in 2006³.

The rapid growth in utilization of APM occurred without rigorous evaluation. Over the last twelve years several observations have called the effectiveness of arthroscopic treatment of osteoarthritis and degenerative meniscal tears into question. First, a randomized controlled trial of arthroscopic lavage and debridement for osteoarthritis showed considerable benefit in both arms with no additional benefit in surgically treated patients compared to those receiving sham surgery⁴. Similarly, a randomized trial showed that patients with osteoarthritis randomized to arthroscopy and those randomized to a physical therapy regimen both showed substantial improvement with no difference between the two treatment arms. ⁵ On the basis of these trials most treatment guidelines suggest that arthroscopic surgery is not indicated for OA. But most arthroscopies in patients with knee OA are performed to address symptoms clinicians attribute to meniscal tear, rather than osteoarthritis per se.

Several epidemiologic studies reported in the last decade further complicated the question of when APM is appropriate by showing that meniscal tear is often an asymptomatic finding. For example, subjects in a large community based sample found to have meniscal tear on MRI were no more likely to have knee pain than subjects without meniscal tear.¹ Similarly, subjects with meniscal tear and osteoarthritis had no more severe or frequent symptoms than subjects with OA and no tear.^2,3 These observations challenged the clinical rationale for arthroscopic partial meniscectomy and set the stage for more rigorous evaluation of APM.

Herrlin and colleagues published the first trial to evaluate the efficacy of APM in patients with degenerative meniscal tear in 2007. These investigators enrolled 97 subjects and randomized half to a rigorous exercise program that emphasized lower extremity strengthening as well as stretching, conditioning and balance. The other half received the same exercise program as well as APM. 28% of subjects randomized to the exercise arm crossed over to the surgical arm in the first fourteen months, generally because of persistent symptoms. In intention to treat analyses (in which patients are analyzed in the groups they were randomized to, irrespective of the treatment they received), the APM group had, on average, a nine point greater change in KOOS Pain score over six months than the exercise-alone group. This difference was not statistically significant, though 9 points on the KOOS Pain Scale is generally considered clinically meaningful⁶.

In the MeTeOR Trial (Meniscal Tear in Osteoarthritis Research),⁷ 351 subjects from seven US referral centers were randomized to APM with an exercise regimen or to the exercise regimen alone. The exercise regimen was similar to that used by Herrlin et al. By six months, 30% of subjects randomized to the PT arm had crossed over to receive APM. The ITT analysis showed that by six months, the group randomized to APM improved 2.7 points more on the 100-point WOMAC Function Scale than the group randomized to PT. This difference was neither clinically meaningfully nor statistically significant. Yim and colleagues published a similar trial to the MeTeOR and the Herrlin trials in 2013. This South Korean study did not document clinically meaningful or statistically significant differences between arms in VAS and Lysholm Scale after one year. There were also no crossovers.

Sihvonen and colleagues published the first sham controlled trial of APM in 2013. They included patients with medial meniscal degenerative tear and no radiographic evidence of OA. The sham group received arthroscopic lavage but no partial meniscal resection. The investigators documented marked improvements in both groups with no clinically important or statistically significant difference between arms in the improvement from baseline.

Against this backdrop of high quality trials that failed to demonstrate superiority of APM over either a PT regimen or sham surgery, Gauffin and colleagues present in this issue of Osteoarthritis and Cartilage a well-done RCT of APM vs. a standardized PT regimen. The APM group improved by 10.6 more points on the KOOS Pain score than the PT group. This difference was statistically significant and clinically meaningful.

Clinicians, investigators, policy makers and payers are challenged to interpret these five trials, four of which did not find APM to be statistically significantly superior to the comparator and one which did find APM to be associated with a statistically significant, clinically important advantage. Why the discrepant findings?

One possibility is that the prior trials missed an important effect because they were too small. Indeed, the Herrlin trial demonstrated a 9.6 point difference between APM and PT groups in improvement in KOOS Pain but the difference was not statistically significant. However the other trials did not demonstrate clinically important differences between study arms (Table).

Table.

Summary of five randomized controlled trials of arthroscopic partial meniscectomy vs. comparator in patients with degenerative meniscal tear

Author, year, country	Sample age, tear, OA status	# Eligible/ randomized	Intervention	Comparator	Primary outcome	Results (ITT analysis)	Change in APM Arm	Crossover
Herrlin, 2007, Sweden	45–64 yo, medial MT, KL 0–2	97/180	Exercise + APM	Exercise	Change in KOOS Pain	9 point difference favoring APM (NS)	33 pts on KOOS Pain	28% from PT to APM by 14 mo
Katz, 2013, US	45+ yo. MT, KLD0–3	331/1330	Exercise + APM	Exercise	Change in WOMAC function	2.4 point difference favoring APM (NS)	20.9 on WOMAC Fxn, 24.2 on KOOS Pain	30% from PT to APM by 6 mo
Yim, 2013, South Korea	43–62 yo, Horizontal medial MT, KL0–1	108/162	APM	Strength exercises	Change in VAS Pain, Lysholm Score	2.0 difference in VAS and 0.1 in Lysholm, favoring APM (NS)	34 on 100 mm VAS Pain	None
Sihvonen, 2013, Finland	35–65 yo, MT, KL0– 1	160/205	APM	Sham APM	Change in pain w exercise (VAS), Lysholm	1.6 pts on Lysholm, 1.0 on VAS, favoring APM (NS)	21.7 pts on Lysholm, 31 on 100 mm VAS pain	None
Gauffin, 2014, Sweden	45–64 yo, MT, KL 0– 2 (93% KL0–1)	150/179	Exercise + APM	Exercise	Change in KOOS Pain	10.6 points favoring APM (p<0.01)	17 pts on KOOS Pain	21% from PT to APM by 12 mo

VAS, Lysholm, KOOS scores transformed to 0–100, 100=best; ITT =intention to treat

Another possibility is that substantial cross over from nonoperative to surgical therapy makes the intention to treat analyses in these trials difficult to interpret. Indeed, in the Herrlin, MeTeOR and Gauffin trials, 20–30% of subjects assigned to the PT arm crossed over to surgery, potentially obscuring differences in outcome between PT and surgery in the intention to treat analyses. In fact the MeTeOR Trial provided a secondary ITT analysis that defined ‘failure’ as either a crossover or failure to improve by 8 points on the WOMAC Function scale. Surgery had significantly fewer failures, defined in this fashion, than nonoperative therapy (33% vs. 56%, p< 0.001)⁷. Thus, the similar outcome documented in intention to treat analyses in these trials may have arisen in part from crossover from nonoperative therapy to APM. That explanation does not explain the findings of the trial of Yim et al, which had no crossovers.

Third, as noted in the Table, APM was associated with substantial improvement in pain and function in each trial, on the order to 20 to 30 points on the 100 point outcome scales (KOOS Pain in three instances). Thus, the trials should not be interpreted as showing that ‘surgery didn’t work,’ but rather that in general the trials showed similar, clinically important improvements in both the surgical arms and the comparator arms.

The findings of the sham-controlled trial of Sihvonen and colleagues merits consideration as well. The trial shows similar outcome following APM and sham, suggesting that the effect of APM may be largely attributable to a placebo, or sham effect. The potency of sham surgery may relate to the substantial investment patients make into preparing for surgery and recovery, and the attendant raising of expectations^8,9. Does the similar outcome of APM and sham surgery vitiate the value of APM as a clinical intervention? We think not. Subjects in both the APM and sham arms improved by a clinically important amount on standardized measures of pain and function and over 90% of subjects in each arm were satisfied with the outcome of surgery¹⁰. If two active hypothetical pharmacologic treatments (e.g. medication A and medication B) were both associated with clinically important improvement and 90% patient satisfaction, we would present both choices to patients as options for a preference based decision. These data should spur further research into approaches for harnessing the benefits of sham and placebo interventions, with less costly and risky treatments than surgery. The results of Sihvonen and colleagues also raise the question of whether the substantial benefits associated with PT regimens in these trials may also be due to placebo effect. This too requires further research, as PT regimens are costly.

For the middle age patient with knee pain and degenerative meniscal tear, the preponderance of evidence today suggests that an initial regimen of strengthening based physical therapy is a reasonable initial step. Patients who fail to improve following nonoperative therapy can be referred for APM. Whether the benefits of surgery (or PT for that matter) are physiologic or due to placebo effects remains an open question, ripe for further research. But we must work with the evidence available and the data summarized in the Table, gathered from five rigorous RCTs, suggest that APM has a role in the management of this prevalent, disabling condition.