Arthroscopy

Arthroscopy is a minimally invasive surgical procedure in which a fibre-optic endoscope is inserted into the joint through a small incision. The surgeon makes a second incision through which to insert the surgical instruments to debride or resect areas within the knee by visualising under the scope. Various treatments can be delivered by arthroscopy, and different elements of treatment might well determine the efficacy of the arthroscopy in osteoarthritis. The possibilities include the following: (1) Washing the joint with saline to remove debris and crystals, which may be inducing pain and inflammation. (2) Debridement of torn menisci and removal of fragments of menisci or of other structures such as torn ligaments. (3) Resection of a proliferative synovium. (4) Excision and removal of loose articular cartilage fragments and smoothing over of cartilage lesions. (5) Removal or grinding down osteophytes that block full extension of the joint – an intervention rarely used in osteoarthritis studies. (6) Drilling of osteochondral lesions.

Evidence on efficacy of arthroscopy in osteoarthritis

For many years, arthroscopy was commonly performed to treat knee osteoarthritis – a process including lavage of the joint and debridement of roughened surfaces with removal of loose debris. Meniscal tears were often resected or repaired as part of this process. Uncontrolled case series suggested that arthroscopy was effective, with improvements lasting 1 year or more. In the largest and perhaps the most comprehensive of these studies, Aaron and colleagues studied 122 patients, of whom 110 were followed up. The Knee Society pain score, a global measure of knee pain, improved by 11.9 (out of a total of 50 points) on average. When defined as a postoperative pain score of greater than 30 points (higher scores connote better status), 65% of the patients had substantial pain relief. Improvement extended up to 36 months and was far more likely if the patients presented with relatively mild osteoarthritis (Kellgren and Lawrence Grade 2 versus Grade 3, for example) , had normal alignment and had preserved joint space. Those factors that tended to weigh against improvement with arthroscopy were the opposite of those that portended a good result – severe arthritis, malalignment and advanced joint narrowing. These findings and those of other case series, which reported, if anything, better results than this study suggested that arthroscopy had a therapeutic role in treating patients with knee osteoarthritis.

The first randomised trial evaluating arthroscopy was carried out as a comparative trial versus lavage, and its main goal was, in fact, to evaluate tidal lavage (large-volume lavage of the knee) as treatment for knee osteoarthritis comparing it with arthroscopy, which was thought to be a ‘gold standard’ treatment. There were no significant differences between the lavage group and the arthroscopy group at 3 or 12 months follow-up. However, the trial was small (only 32 patients in total) and may have been underpowered to detect differences suggested by the data; for example, assessments by patients suggested that there was more improvement in the arthroscopy than in the lavage group (e.g., 56% had improvement in pain score in the arthroscopy group at 12 months vs. only 43% in the lavage group). Physician-designated improvement was also more common in the arthroscopy group although not significantly so, but some of the outcomes were actually in contrast, with patient global assessment suggesting a better result for lavage than for arthroscopy at 12 months. Thus, this trial comparing lavage and arthroscopy arrived at indeterminate results with no significant differences between the two modalities of therapy. The small sample size precluded the detection of small and potentially clinically important differences .

These uncontrolled and early randomised trials formed the basis for a milestone randomised trial carried out, with arthroscopic surgery being compared with sham surgery . In this trial, 180 patients with osteoarthritis were randomly assigned to receive one of three treatments: arthroscopic debridement (which consisted of all various elements of arthroscopy noted earlier, in addition to lavage); arthroscopic lavage alone (in this treatment group, unstable meniscal tears were resected but no other debridement was done); or sham surgery. This last treatment arm was novel and included three 1-cm incisions made to mimic the arthroscopic portals, noises and instrumentation that suggested arthroscopy was being performed, along with a stay overnight in the hospital. Patients were later found to be blinded with respect to whether they had undergone the procedure or not.

The Moseley trial results showed absolutely no efficacy of arthroscopic debridement or lavage versus sham surgery. Neither the unique outcome measure that was the primary measure in this study, the knee-specific pain scale score, nor the other, more widely accepted, secondary outcome measures showed any difference among the debridement, lavage or placebo surgery groups. No differences were seen even as much as 3 years after surgery, with follow-up being excellent. While small differences between debridement and placebo might not have been detected, given the modest sample size of the study (approximately 60 in each group), at no point in the follow-up did the debridement group do better than the placebo group on average.

There were a number of features unique to the Moseley trial that disconcerted readers about the generalisability of its results. First, the trial was performed at the U.S. Veterans Administration primarily in men. Second, all arthroscopies were performed by only one surgeon whose own practice techniques may have been relevant to the success or failure of the procedure. Third, prominent mechanical symptoms of osteoarthritis such as catching or locking were not ascertained and were not used as eligibility criteria for this trial, so that potential effects of arthroscopy on these symptoms were not characterised. Lastly, an unusual and novel pain evaluation was carried out with a newly developed knee-specific pain scale.

Because of the concerns about generalisability that were raised in the Moseley trial, a group of Canadian investigators, led by Kirkley et al. , undertook another randomised trial considering the efficacy of arthroscopy in a broader, more generalisable sample of patients with knee osteoarthritis. The study design varied from that of Moseley in that no sham surgery was included and the only active treatment that was evaluated was debridement with lavage. It was compared with no surgery, with all patients in both groups given conventional therapy, including optimised physical therapy. Many of the peculiarities of the Moseley study were resolved in this study by Kirkley et al. The specificities included: the predominance of women (63% of subjects) in the Kirkley study; the use of a well-validated, widely used outcome measure in osteoarthritis trials, the Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC); the characterisation and inclusion of mechanical symptoms in the knee; and the involvement of multiple surgeons, with an agreed upon treatment protocol that included lavage, debridement, synovectomy, excision of degenerative tears in the meniscus and of fragments of cartilage and excision of osteophytes that prevented full extension. Similar to the other published trials of arthroscopy, micro-fracture of chondral defects was not performed. Physical therapy was comprehensive and standardised.

The results showed no difference at 1 or 2 years between the group that was randomised to arthroscopy and that randomised to control. Not only was pain not improved by arthroscopy versus the control, but mechanical symptoms were not affected by treatment as well. Even though uncontrolled studies had suggested and even orthopaedic clinicians had suggested in letters after the Moseley study that the patients who benefited most were those with milder osteoarthritis, the Kirkley trial reported that in those with milder grade 2 disease, there was also no effect of arthroscopy on symptoms. A variety of secondary outcomes were evaluated, and these also showed no difference between arthroscopic treatment and placebo. In general, the Kirkley trial resolved any remaining issues about the efficacy of arthroscopy that had not been answered by the Moseley trial. It showed that, in general, there is no role for arthroscopy as a treatment of osteoarthritis.

The results of these well-done, randomised trials evaluating arthroscopy provide important examples of how randomised trial data are sometimes necessary to evaluate treatments that we trust to be effective. It is noteworthy that in both trials there was improvement over 1–2 years in all patients treated not just in those undergoing arthroscopy, and that may have been why the uncontrolled studies of arthroscopy were so favourable. The failure of arthroscopy to be effective suggests that all elements of that treatment are ineffective. Specifically, lavage, the washing out of the knee to get rid of debris and crystals, while it may have a temporary beneficial effect, clearly has no long-term effect. (This null result has been suggested by trials of lavage itself). Furthermore, removing debris at the surface of cartilage or even in the meniscus is unlikely to have any beneficial effect as is the surgical treatment of degenerative tears, especially those that are not unstable (unstable tears in older osteoarthritis patients are quite uncommon).

In an editorial accompanying the trial by Kirkley, a prominent orthopaedic surgeon (Marx) , asked whether arthroscopy should ever be done in patients with osteoarthritis. He suggested that in a patient with mild or no osteoarthritis and who has a clear-cut, acute injury involving twisting or other knee trauma and whose symptoms date clearly to that injury where a magnetic resonance imaging (MRI) demonstrates a remediable lesion likely due to the injury, arthroscopy is indicated and might be of great help to the patient. It is the more common scenario in patients with chronic osteoarthritis who usually have co-existent meniscal tears that arthroscopy is of no value.

In summary, arthroscopy, while shown to be promising in uncontrolled studies, has now been convincingly demonstrated to be inefficacious for the treatment of osteoarthritis. It should not be carried out to help patients with osteoarthritis, except perhaps if there is evidence of recent trauma and a symptomatic meniscal tear.