A RETROSPECTIVE REVIEW OF YTTRIUM-90 SYNOVECTOMY IN THE TREATMENT OF KNEE ARTHRITIS

Transcription

1 British Journal of Rheumatology 1997;36:1100±1105 A RETROSPECTIVE REVIEW OF YTTRIUM-90 SYNOVECTOMY IN THE TREATMENT OF KNEE ARTHRITIS W. J. TAYLOR, M. M. CORKILL and C. N. A. RAJAPASKE Wellington Regional Rheumatology Unit, Hutt Hospital, Hutt Valley Health Corporation Ltd, Private Bag 31907, Lower Hutt, New Zealand SUMMARY We reviewed the case notes and X-rays of all patients with knee arthritis treated with yttrium-90 for the rst time at a single institution from November 1981 to November Outcomes were assessed as `improved' or `not improved' by review of the case notes at 3, 6 and 12 months, and by the absence of further intra-articular (IA) steroid injections. Of the 121 knees treated, 87 had adequate follow-up to allow an assessment of outcomes. Overall, 46% (95% CI 36±57) were improved at 12 months and 37% (95% CI 27±47) had no further IA injections (mean follow-up of 3.5 yr). Knees with osteoarthritis (OA) fared signi cantly worse with 10% (95% CI 0±29) vs 51% (95% CI 39±63) improved at 12 months (P < 0.05). Knees younger than 30 appeared to do better with 78% (95% CI 51±100) vs 28% (95% CI 17±45) having no further IA injections (P < 0.02). Knees with normal X-rays (Kellgren grade 0±1) did signi cantly better than those with more severe radiographic abnormalities (Kellgren grade 3±4), with 56% (95% CI 40±73) vs 24% (95% CI 8±40) improved (P < 0.01). Radiosynovectomy with yttrium-90 for knee arthritis appears to be of less value for patients with OA or with secondary OA changes on X-ray, and may be of more value for younger patients and those with spondyloarthropathies. KEY WORDS: Radiosynovectomy, Yttrium-90, Knee arthritis, Retrospective analysis. RADIOSYNOVECTOMY for treatment-resistant synovitis of individual joints, i.e. despite systemic pharmacotherapy and intra-articular (IA) steroid injections, has been used since 1963 [1]. Yttrium-90 (Y90) became preferentially used in the 1970 s because of improved b radiation and little g radiation [2]. The reported success rate varies substantially from a 15% response rate at 1 yr [3] to 69.5% at 3 yr [2]. Of the randomized controlled trials [3±13], only two demonstrate an advantage of Y90 over IA triamcinolone [10] or methylprednisolone acetate [11]. Despite this, it seems to have established a place in the therapeutic armamentarium of rheumatologists, as judged by its continued usage; perhaps because most controlled studies su er from small numbers and lack statistical power to show that yttrium radiosynovectomy is actually ine ective. Initially used to control synovitis due to rheumatoid arthritis (RA), radiosynovectomy has more recently been used for other in ammatory arthropathies, osteoarthritis (OA) [12] and haemophiliac arthropathy [14]. It is unclear from the available evidence whether patients with particular arthropathies respond di erently to radiosynovectomy. For instance, a large retrospective review found a success rate of 71% at 12 months in patients with OA [15], which contrasts with the ndings of a prospective study showing only a 41% success rate in a similar Submitted 5 August 1996; revised version accepted 21 March Correspondence to: W. J. Taylor, Spinal/Rehabilitation Unit, South Auckland Health, Private Bag 93319, Otahuhu, Auckland, New Zealand. group of patients [3]. It is the loss of articular cartilage and changes in adjacent bone that are characteristic of OA, and although synovial in ammation may be present, this is considered to be a secondary response [16]. It might, therefore, be expected that radiosynovectomy would be ine ective or only temporarily e ective in OA. We wished to examine the experience of a single institution using Y90 synovectomy over a period of 14 yr to assess the overall e ectiveness and safety, and to de ne the characteristics of responders. METHODS Yttrium synovectomy has been used only for knee joints in Wellington. A record of all patients receiving this treatment has been maintained since The procedure has followed a de ned protocol: Y90 silicate is prepared to order in Amersham and own to Wellington to be used within 24 h of arrival; the patient is admitted and informed consent is obtained; the knee is aspirated and 40 mg of triamcinolone acetonide are injected; 5 mci of Y90 silicate are injected through the same needle; the knee is splinted and the patient is bed-rested for 2 days. The case notes and X-rays of all patients who received Y90 radiosynovectomy for the rst time between November 1981 and November 1995 were reviewed. Repeat synovectomies were not analysed. Outcomes were assessed (a) by a statement of bene t at 3, 6 and 12 months as `improved' or `not improved' as recorded in the case notes and (b) by a surrogate marker of improvementðabsence of further IA steroid injections. The most consistently recorded response to treatment in the case notes was 1100 # 1997 British Society for Rheumatology

2 TAYLOR ET AL.: YTTRIUM-90 SYNOVECTOMY 1101 TABLE I Patient characteristics Inadequate outcome (n = 34) found to be a global opinion from the physician or patient. An unequivocally positive or negative statement regarding response was taken to re ect outcome. If this was not su ciently explicit in the case notes, then the patient was excluded from analysis. X-rays were graded using the method of Kellgren and Lawrence [17]. Follow-up extended to the most recent clinic visit. Adverse e ects, repeat procedures and subsequent knee arthroplasty were noted. Possible factors that might have in uenced responsiveness that were noted included primary disease, age, duration of disease, time since rst IA injection to the treated knee, number of prior IA steroid injections and X-ray changes. Patients with inadequate documentation or follow-up data were excluded from subsequent analysis. Signi cance testing for di erences between subgroups used the w 2 distribution in a series of 2 2 table analyses. Statistical signi cance was taken to be P < Con dence intervals for proportions were calculated using standard methods. Agreement between the two outcome methods used the kappa statistic. RESULTS One hundred and twenty-one knees in 88 patients were treated over the time period. Of these, 87 (72%) knees had adequate documentation or follow-up to allow an assessment of outcome. TABLE II Distribution of primary disease Inadequate outcome (n = 34) Adequate outcome (n = 87) Age (range), yr 51 (25±77) 55 (20±84) M:F ratio 15:19 35:52 Duration of disease (range), yr 12.8 (0.6±45) 14.9 (0.25±45) Time since rst IA steroid injection into treated knee (range), yr not known 3.9 (0.25±27) Number of previous IA steroid injections (range), yr not known 3.9 (0±14) Adequate outcome (n = 87) Rheumatoid arthritis 26.5% 39.1% Osteoarthritis 20.6% 11.5% Ankylosing spondylitis 0 5.7% Psoriatic arthritis 32.0% 26.4% Undi erentiated spondarthritis 11.8% 9.2% Other 8.8% 9.2% Patients being followed up in private practice was the most common reason for missing data from the hospital case notes. Mean follow-up was 3.5 yr (range 0.3±13). The characteristics of these knees are shown in Table I. Most patients had RA, psoriatic arthritis (PsA) or OA (Table II). There were no serious adverse e ects in this series of 121 knees, but a subsequent repeat radiosynovectomy in one of these patients was complicated by septic arthritis. There were no instances of needletrack burns. Patients received a mean of 3.9 IA steroid injections over a period of 3.7 yr prior to the yttrium synovectomy. Overall, 46% (95% CI 36±57) of knees improved at 12 months. Slightly less (37%, 95% CI 27±47) had no further IA injections after a mean follow-up of 3.5 yr. Sixteen per cent went on to knee arthroplasty after a mean of 3.0 yr and 22% had a further yttrium synovectomy after a mean of 2.9 yr. There were no signi cant di erences in terms of improvement rate or absence of further IA steroid injections with respect to sex, duration of disease, time since rst IA injection into the treated joint or the number of prior steroid injections into the treated joint (Table III). Patients with OA had a signi cantly poorer response with 10% (95% CI 0±29) vs 51% (95% CI 39±63) (P < 0.01) improving at 12 months. The spondyloarthritis group exhibited a trend towards better outcome with 47% (95% CI 30±64) vs 24% (95% CI 12±36) not signi cant (NS) not requiring a further IA injection, having previously had 3.8 IA steroid injections over the preceding 4.0 yr (Fig. 1). There were no signi cant di erences between the disease groups in terms of number or frequency of prior IA steroid injections (Table IV). Signi cantly more patients aged less than 30 had no subsequent IA steroids (78%, 95% CI 51±100 vs 28%, 95% CI 17±45, P < 0.02), but there were no di erences between the age groups with respect to improvement at 12 months (Fig. 2). Knees with no or minimal OA radiographic changes (Kellgren grade 0±1) responded better than those with more advanced changes (Kellgren grade 3±4) in terms of improvement at 12 months (56%, 95% CI 40±73 vs 24%, 95% CI 8±40, P < 0.01) (Fig. 3). Overall agreement between `improved at 12 months' and `no further IA steroid injections' was moderately good with kappa=0.46. DISCUSSION Overall, we found that Y90 radiosynovectomy was moderately e ective, with 46% of knees improved from baseline after 12 months and 37% of knees having no further IA steroid injections at a mean of 3.9 yr following the procedure. This is similar to the overall ndings by other investigators. Knees with primary OA or radiographic OA changes

3 1102 BRITISH JOURNAL OF RHEUMATOLOGY VOL. 36 NO. 10 FIG. 1.ÐOutcomes expressed as the percentage of knees improved at 12 months or not requiring further IA steroid injections, by primary disease. responded poorly compared to other arthropathies. There was a trend to a greater response in patients younger than 30 yr and in those with spondyloarthropathies. Conclusions must be guarded due to the FIG. 2.ÐOutcomes expressed as the percentage of knees improved at 12 months or not requiring further IA steroid injections, by age. limitations of retrospective reviews, particularly selection bias, outcome criteria and loss to follow-up. Chemical synovectomy has been reviewed recently [18]. The authors interpret the literature as showing TABLE III Outcomes by sex, duration of disease, time since rst IA injection and number of prior IA injections N Improved at 12 months (%) No further IA steroid injections (%) Sex Male Female Duration of disease (yr) < ± ± ± ± > Time since rst IA steroid injection into treated knee (yr) < ± ± ± ± > Number of previous IA steroid injections into treated knee 0± ± ± > TABLE IV Number and frequency of intra-articular steroid injections prior to yttrium radiosynovectomy Number of prior IA steroid injections (mean) Over the period prior to yttrium, yr (mean) Frequency of injections prior to yttrium, per year Osteoarthritis Rheumatoid arthritis Psoriatic arthritis Ankylosing spondylitis Other spondarthritis All spondarthritis Other Total

4 TAYLOR ET AL.: YTTRIUM-90 SYNOVECTOMY 1103 FIG. 3.ÐOutcomes expressed as the percentage of knees improved at 12 months or not requiring further IA steroid injections, by X-ray changes. `both controlled and uncontrolled studies show a greater than 50% improvement with yttrium synovectomy'. However, this must be tempered by a success rate of 38% at 1 yr with IA saline [13] or 52% with triamcinolone hexacetonide [10]. Furthermore, the single placebo-controlled trial that demonstrates an advantage to yttrium showed a statistically signi- cant improvement in knee circumference (35% vs 10%, P < 0.05), but not e usion (65% vs 48%, NS), although the trend was similar. There was also no statistically signi cant change in pain (57% vs 38%, NS) or `subjective improvement' (61% vs 38%, NS) [11]. A trial comparing IA Y90 with non-radioactive yttrium found no di erence in the e cacy of treating rheumatoid knees [5]. In this study, arthroscopic ndings appeared to be more marked in the patients treated with Y90, but clinical improvement seemed to be more related to the removal of large quantities of IA brin. There is thus a lack of satisfactory controlled trials of yttrium radiosynovectomy. This has led a meta-analysis to conclude that yttrium was superior to placebo, but not to IA triamcinolone hexacetonide. Furthermore, it has been shown that a single unpublished, negative placebo-controlled trial would invalidate the meta-analysis superiority of yttrium compared to placebo [19]. To complicate matters further, one of the trials incorrectly included in the meta-analysis of placebo vs yttrium actually compared methylprednisolone to yttrium [11]. This may mean that yttrium is more e ective than placebo to a greater extent than the meta-analysis would suggest. There are three controlled studies comparing yttrium synovectomy to IA steroids in the English literature [8, 10, 11]. The study by Szanto [11] demonstrated a signi cant bene t of yttrium compared to IA methylprednisolone (59% vs 14% improved at 12 months, P < 0.01), but su ered methodological faults. The control knees were contralateral knees that also had e usions, but no mention is made of randomization or blinded clinical assessments. The ARC trial [8], comparing yttrium with IA triamcinolone hexacetonide, was terminated prematurely because of recruitment problems. In the patients who were analysed, there was no di erence between the two treatments (42% vs 47%, NS). The other study comparing triamcinolone hexacetonide with yttrium [10] (only in abstract form) found a signi cant di erence between the two treatments (72% vs 52%, P < 0.05). The improvement rates are somewhat higher than those found in other studies. There is thus only limited evidence from randomized controlled trials that yttrium is superior to IA steroids. Studies comparing Y90 synovectomy to surgical synovectomy or chemical synovectomy with osmic acid show no di erences between these forms of treatment [4, 6]. There are several limitations to retrospective studies. Importantly, how are the criteria on which an assessment of `improvement' is made derived? It is preferable for these to be based on explicit, validated outcome measures such as that proposed by the American College of Rheumatology [20]. Such is not necessarily available in retrospective reviews and is a major limitation of this methodology. Our assessment of improvement was based on the clinical judgement of the treating physician and is, therefore, subject to bias. We therefore also looked at `absence of further IA steroid injections', which would be less subject to bias. This gave a less optimistic estimate of the overall response rate and differed from the `12 month improvement' result in a number of other instances. However, the overall agreement between `no further IA injection' and `12 month improvement' was moderately good. Although the short-term e ectiveness of a therapy is best evaluated by means of a controlled randomized trial [21], we believe that valid conclusions can be drawn from this study for a number of reasons. Firstly, all patients treated with yttrium in the Wellington region were identi ed by a prospectively maintained log book. This would tend to obviate selection bias. Secondly, the characteristics of patients with adequate outcome data were not signi cantly di erent from those with inadequate outcome data. Thirdly, the inadequate response of these patients to previous IA steroids (mean number of previous IA steroid injections was 3.9) would suggest that a response rate of 46% at 12 months would be signi cantly better to further IA steroids or placebo. Without a control group, however, the issue of e ectiveness cannot ever be adequately addressed. With similar biases a ecting all patients, it is valid to compare the relative di erences in e ectiveness amongst the various groups to identify characteristics of responders vs non-responders. It is important to target a relatively expensive procedure (NZ$800 + three hospital bed days) to patients who will bene t the most. We found that patients with OA responded poorly. This is perhaps not unexpected for reasons suggested above, but dif-

5 1104 BRITISH JOURNAL OF RHEUMATOLOGY VOL. 36 NO. 10 fers substantially from another large retrospective review [15] which found an encouraging response rate in patients with OA. Possible reasons for the di erence in results include a slightly lower dose of Y90 in our practice (5 mci vs 6 mci), the use of an IA steroid 3 days prior to yttrium rather than immediately prior to yttrium as in our protocol, older patients in our study (51 yr vs 40.5 yr), and longer duration of disease in our patients (12.8 yr vs 7.8 yr). It seems unlikely that the di erences in technique would account for the di erence in results for the OA group considering that the results for RA patients were very similar as in the present study (46% success rate vs 53%). Possibly related ndings in our study were of better response rates in young patients (<30 yr) and patients with normal X-rays. Other studies have noted similar e ects of abnormal X-rays [9]. It would appear that damaged or OA joints respond less well to yttrium radiosynovectomy. It might be imagined that joints with a thinner synovium might respond better to radiosynovectomy because there will be a greater percentage ablation of the synovium in these patients than in those with a more thickened synovium. There is some support for this notion with the nding that patients with spondyloarthropathies had a better response (although not signi cantly) than patients with RA or other arthropathies. The e ect of prior treatment with IA steroids was not marked. Neither the time elapsed since the rst IA steroid injection nor the number of prior steroid injections appeared to in uence the response rate to yttrium. It may be that pre-injection of steroids 2 or 3 days prior to radiosynovectomy would enhance the e ectiveness of yttrium by decreasing the volume of synovium for the yttrium to ablate. This study is unable to determine the validity of this concept. We suggest that yttrium radiosynovectomy is safe and moderately e ective for in ammatory arthritides of the knee. It was of little bene t for primary OA or knees with secondary OA changes on X-ray. A more de nitive conclusion regarding the e ectiveness of yttrium radiosynovectomy awaits the results of a su ciently large controlled trial. In the absence of such a study, and considering our results and a review of the evidence to date, our unit has formulated the following guidelines for the use of yttrium synovectomy for knee arthritis. These are intended to re ect a pragmatic approach to clinical practice in an area where evidence is less than ideal. WELLINGTON REGIONAL RHEUMATOLOGY UNIT GUIDELINES FOR YTTRIUM SYNOVECTOMY A. Criteria for consideration of treatment 1. In ammatory synovitis of the knee. 2. Radiographic changes no worse than Kellgren grade 2. B. Adequate trial of alternative therapy 1. Generalized disease activity should be controlled with systemic pharmacotherapy. 2. Individual knee joints should receive at least two intra-articular injections of 20 mg triamcinolone hexacetonide systematic pharmacotherapy before judging this therapy as ine ective. C. Use a protocol demonstrated to be safe 1. Admit patients to hospital for 3 days. 2. Aspirate the knee joint, inject 50 mg of triamcinolone acetonide and then 5 mci of yttrium-90 silicate through the same needle. 3. Nuclear Medicine technical sta to be present for assistance with disposal of syringes and needles and monitoring of the procedure. 4. Splint the knee and bed rest the patient for at least 2 days. ACKNOWLEDGEMENT We are grateful for the advice of Dr Roger Marshall of the Auckland University Medical School with regard to statistical analysis. REFERENCES 1. Ansell BM, Crook A, Mallard JR et al. Evaluation of intra-articular gold Au-198 in the treatment of persistent knee e usions. Ann Rheum Dis 1963;22:435±9. 2. Gumpel JM. 90Y colloids in chronic synovitis of the knee: a review 1970±1977. Rheumatol Rehabil 1979; 18(suppl.):38± Edmonds J, Smart R, Laurent R et al. A comparison study of the safety and e cacy of dysprosium-165 hydroxide macro-aggregate and yttrium-90 silicate colloid in radiation synovectomyða multicentre double blind clinical trial. Br J Rheumatol 1994;33:947± Gumpel JM, Roles NC. A controlled trial of intraarticular radiocolloids versus surgical synovectomy in persistent synovitis. Lancet 1975;i:488±9. 5. Yates DB, Scott JT, Ramsey N. Double blind trial of yttrium-90 for chronic in ammatory arthritis of the knee (abstract). Ann Rheum Dis 1977;36: Nissila M, Anttila P, Hamalieun M, Delbarre F. Comparison of chemical, radiation and surgical synovectomy for knee joint synovitis. Scand J Rheumatol 1978;7:225±8. 7. Sheppeard H, Aldin A, Ward DJ. Osmic acid versus yttrium-90 in rheumatoid synovitis of the knee. Scand J Rheumatol 1981;10:234±6. 8. ARC Multicentre Radiosynoviothesis Trial Group. Intra-articular radioactive yttrium and triamcinolone hexacetonide: an inconclusive trial. Ann Rheum Dis 1984;43:620±3. 9. Boerbooms A, Buijs W, Danen M, Van de Putte L, Vanderbrouke J. Radiosynovectomy in chronic synovitis of the knee joint in patients with rheumatoid arthritis. Eur J Med 1985;10:446± Menkes C, Le Go A, Verrier P, Delbarre F. A controlled trial of intra-articular yttrium-90, osmic acid and triamcinolone hexacetonide in rheumatoid arthritis.

6 TAYLOR ET AL.: YTTRIUM-90 SYNOVECTOMY 1105 In: XIV Int Congr Rheum San Francisco 1977:614 (Abstract). 11. Szanto E. Longterm followup of yttrium-90 treated knee joints. Scand J Rheumatol 1978;6:209± Will R, Laing B, Edelman J et al. A comparison of two yttrium-90 regimens in in ammatory and osteoarthropathies. Ann Rheum Dis 1992;51:262± Bridgeman JF, Brucknew F, Eisen V et al. Irradiation of the synovium in the treatment of rheumatoid arthritis. Q J Med 1972;42:357± Dawson TM, Ryan PFJ, Street AM et al. Yttrium synovectomy in haemophiliac arthropathy. Br J Rheumatol 1994;33:351± Stucki G, Bozzone P, Treur E et al. E cacy and safety of radiation synovectomy with yttrium-90: a retrospective long term analysis of 164 applications in 82 patients. Br J Rheumatol 1993;32:383± Dieppe PA. OsteoarthritisÐintroduction. In: Klippel JH, Dieppe PA, eds. Rheumatology. London: Mosby, 1994: Kellgren JH, Lawrence JS. Radiological assessment of osteoarthritis. Ann Rheum Dis 1957;16:494± Cruz±Esteban C, Wilke WS. Non-surgical synovectomy. Baillie(c)re's Clin Rheumatol 1995;9:787± Jones G. Yttrium synovectomy: a meta-analysis of the literature. Aust NZ J Med 1993;23:272± Felson DT, Anderson JJ, Boers M et al. American College of Rheumatology preliminary de nition of improvement in rheumatoid arthritis. Arthritis Rheum 1995;38:727± Sackett DL, Haynes RB, Guyatt GH, Tugwell P. Clinical epidemiology: a basic science for clinical medicine, 2nd edn. Boston, MA: Little Brown and Co. 1991:191.