Treatment of Rheumatoid Arthritis

Much remains to be learned about the cause and pathogenesis of rheumatoid arthritis, but we have learned enough to begin to develop rational treatment strategies. Developments in 1997 included the use of tetracyclines to affect collagenase activity or immune function and aggressive early multidrug therapy to "turn off" early rheumatoid arthritis. Another approach—use of soluble recombinant human tumor necrosis factor (TNF) receptor—is predicated on the view that immunologic factors and cytokines contribute to the pathogenesis of rheumatoid arthritis.

Minocycline Significantly Suppressed Early Rheumatoid Arthritis in a Majority of Patients

O'Dell JR, Haire CE, Palmer W, et al. Treatment of early rheumatoid arthritis with minocycline or placebo: results of a randomized, double-blind, placebo-controlled trial. Arthritis Rheum. 1997; 40:842-8.

On the basis of speculation that rheumatoid arthritis is an infectious disease, tetracycline was used in some parts of the United States from the 1940s through the 1960s. A negative study from Massachusetts General Hospital published in 1971 reported no benefit from this therapy (at least from low-dose tetracycline). Nevertheless, reports of clinical benefit continued to appear; these reports were supported by research findings showing that, quite apart from their antimicrobial activity, tetracyclines directly influence polymorph function and may interfere with collagenases (and possibly other proteolytic enzymes) in the joints. Two recent placebo-controlled studies suggested that tetracycline results in some improvement in clinical signs and symptoms of rheumatoid arthritis (1, 2).

O'Dell and colleagues' multicenter trial included 46 patients who had had rheumatoid arthritis for no longer than 1 year and had not received systemic steroid therapy or disease-modifying antirheumatic drugs. Patients were randomly assigned to receive 100 mg of minocycline orally or a matching placebo twice daily. Nonsteroidal drugs were allowed. Assessors, blinded for the first 6 months, examined outcomes using as their primary measure a 50% improvement in a modified Paulus composite score that was based on the numbers of tender and swollen joints, the duration of morning stiffness, and the erythrocyte sedimentation rate. Sixty-five percent of minocycline recipients but only 13% of placebo recipients achieved this level of improvement for 6 months or more. Secondary outcome measures also tended to favor active treatment. Minocycline continued to be advantageous when the study was unblinded for the final 6 months, and it reduced the need for disease-modifying antirheumatic drugs, such as methotrexate.

Whereas patients in earlier studies had long-standing rheumatoid arthritis, patients in O'Dell and colleagues' study had been ill for an average of less than 6 months. The trial has two major limitations: its short duration and small size. As a result, important side effects might have been missed. Moreover, an important goal of rheumatoid arthritis treatment—preventing radiographic bone erosions—was not determined because of the short duration of the study. Such treatment, continued for 2 to 3 years, could preclude the use of other aggressive treatment that might more effectively prevent erosive disease when used early. On balance, though, minocycline seems to be useful for early rheumatoid arthritis. The manner in which minocycline might suppress rheumatoid arthritis is not clear; the drug may affect immune modulation or interfere with inflammatory mediators or joint-damaging enzymes.

Combined Regimen Enhanced Symptom Resolution in Early Rheumatoid Arthritis Compared with Sulfasalazine Alone

Boers M, Verhoeven AC, Markusse HM, et al. Randomised comparison of combined step-down prednisolone, methotrexate and sulphasalazine with sulphasalazine alone in early rheumatoid arthritis. Lancet. 1997; 350:309-18.

Most past studies of combination treatment for rheumatoid arthritis have achieved outcomes only minimally better than those seen with monotherapy. One recent study, however, did strongly favor a combination of hydroxychloroquine, sulfasalazine, and methotrexate (3). Most rheumatologists now agree that second-line therapy should be instituted at the time of diagnosis and continued as long as any evidence suggests ongoing active disease. Much interest has been expressed in the "step-down" approach: aggressively treat rheumatoid arthritis very early in its course and, after clinical control is achieved, withdraw some drugs and continue therapy with those that are the least toxic. Treatment should focus on suppressing inflammation and pannus development and limiting erosive disease in the first year or two.

In a large multicenter trial from the Netherlands, Boers and colleagues compared a step-down regimen of prednisolone (initially 60 mg/d, tapered to 7.5 mg/d), sulfasalazine (500 mg twice daily and then 2000 mg once daily), and methotrexate (7.5 mg weekly) with sulfasalazine alone in 155 patients in whom rheumatoid arthritis had been diagnosed in the past 2 years. None of the patients had received a disease-modifying antirheumatic drug other than hydroxychloroquine. Prednisolone therapy was stopped by week 34, and methotrexate treatment was withdrawn at week 46.

After 28 weeks of treatment, a pooled outcome measure comprising the number of tender joints, grip strength, erythrocyte sedimentation rate, a global assessment, and McMaster questionnaire responses revealed that the combination regimen was associated with rapid improvement. Almost half of the patients receiving this therapy achieved 50% improvement according to American College of Rheumatology response criteria; in contrast, just more than one fourth of patients given monotherapy did so. The rates of remission (definite or probable) were 28% for the combination therapy group and 16% for the monotherapy group. The differential clinical response rapidly disappeared, however, after steroid and methotrexate treatments were withdrawn, and the difference between groups was not significant after 12 months. One major effect did persist: Patients given combination therapy had much less erosive disease after 80 weeks of therapy. Bone mineral density tended to decline in this group, although not to a statistically significant degree.

The enhanced early response to combination therapy is probably caused by the steroid. Unfortunately, the three-drug regimen studied does not seem to "turn off" the disease process. It may, however, slow the progression of erosive lesions even after therapy with the added drugs is withdrawn and thus may enhance long-term outcome. Apart from a slight decline in bone mineral density, the combined regimen had no additional short-term toxic effects. Further experience with the step-down approach may show that aggressive management must continue for some time to sustain clinical benefit.

Soluble Tumor Necrosis Factor Receptor Is a Novel and Effective Way To Suppress Active Rheumatoid Arthritis

Moreland LW, Baumgartner SW, Schiff MH, et al. Treatment of rheumatoid arthritis with a recombinant human tumor necrosis factor receptor (p75)-Fc fusion protein. N Engl J Med. 1997; 337:141-7.

The fact that even indisputably effective drugs often fail to suppress rheumatoid arthritis completely has prompted interest in "biological" agents. These agents might include antibodies against cells participating in the immune response; drugs that inhibit specific immune responses; or agents that interfere with the cytokines, growth factors, and neuropeptides that help perpetuate joint inflammation and destruction. One of the key cytokines in this process is TNF-, which is primarily made by mononuclear cells and binds to cell-surface receptors on lymphocytes, monocytes, polymorphs, and endothelial cells. Soluble TNF receptors (TNFRs) in tissue fluids normally "sop up" the cytokine before it can reach receptors on the cell surface; by doing so, they effectively down-regulate the inflammatory response.

Moreland and colleagues conducted a multicenter, double-blind, randomized trial to study the effects of directly administered, soluble recombinant TNFR:Fc. The human Fc component extends the receptor's duration of action. One hundred eighty patients who had not responded to disease-modifying antirheumatic drugs were randomly assigned to receive subcutaneous injections of 0.25, 2, or 16 mg of TNFR:Fc per m² of body surface area or placebo twice a week for 3 months. After 3 months, the number of swollen or tender joints was reduced by 61% in patients given the highest TNFR:Fc dose and by 25% in those given placebo. A 50% improvement in American College of Rheumatology response criteria was achieved by 57% of patients receiving the highest TNFR:Fc dose and 7% of placebo recipients. The effects were dose-dependent but slowly disappeared after treatment was stopped.

This study hints at a potentially promising method of treating rheumatoid arthritis. Most patients were better within the first month, and toxicity was minimal; however, a few local reactions and some upper respiratory tract symptoms did occur. A negative aspect of the study was the deterioration of the patients' condition as soon as treatment ceased. Of note, patients did not develop antibodies to TNFR. The study's duration was short, and the extent to which TNFR slows the progression of erosive disease remains unknown. Even if these results are replicated, TNFR:Fc treatment will be very costly. There is also theoretical concern that this therapy could predispose patients to infection, other autoimmune disorders, or even neoplastic disease. However, the amount of improvement was impressive; TNFR:Fc and related drugs may represent a breakthrough in the treatment of rheumatoid arthritis.

Modifying the Toxicity of Antirheumatic Drugs

As antirheumatic drugs are developed and refined, their safety remains a paramount consideration, particularly because they are often administered for many years. Three important safety issues are 1) avoiding peptic ulcer disease caused by nonsteroidal anti-inflammatory drugs [NSAIDs], 2) using etidronate to prevent steroid-induced osteoporosis, and 3) preventing hydroxychloroquine-related retinopathy.

Eradicating Helicobacter pylori before Nonsteroidal Anti-Inflammatory Treatment Reduced Risk for Peptic Ulcer

Chan FK, Sung JJ, Chung SC, et al. Randomised trial of eradication of Helicobacter pylori before non-steroidal anti-inflammatory drug therapy to prevent peptic ulcers. Lancet. 1997; 350:975-9.

If it is true that almost all peptic ulcers are caused by Helicobacter pylori or NSAIDs, it would seem obvious that it's not good to be exposed to both. Chan and colleagues sought to determine whether eradicating H. pylori would prevent ulcers in patients who were to begin taking NSAIDs, usually for osteoarthritis or rheumatoid arthritis. For all patients, infection with H. pylori at baseline was documented by a urease test and histologic examination, and peptic ulcer disease was ruled out. Patients were randomly assigned to receive naproxen alone or to have H. pylori eradicated by a triple-drug regimen consisting of bismuth subcitrate, tetracycline, and metronidazole before they received naproxen. Slightly more than one fourth of 47 patients given naproxen alone had peptic ulcer disease after 8 weeks of treatment. In the antibiotic treatment group, peptic ulcer was seen at follow-up in only 1 of the 40 patients in whom H. pylori had been eradicated but in 2 of the 5 patients who remained infected.

The study patients were a selected group of patients who lacked peptic ulcer at baseline despite the presence of H. pylori and who had not been taking NSAIDs. The study was short term and focused on endoscopic ulceration rather than clinically important outcomes, such as gastrointestinal bleeding, perforation, or hospital admission. Although it might seem eminently reasonable to eradicate H. pylori when NSAID therapy is imminent, the situation remains confusing. Paradoxically, other recent evidence suggests that the presence of H. pylori may actually promote the healing of NSAID-induced ulcers and reduce the risk for ulcer recurrence.

Etidronate Preserved Vertebral Bone Integrity in Patients Given Steroids

Adachi JD, Bensen WG, Brown J, et al. Intermittent etidronate therapy to prevent corticosteroid-induced osteoporosis. N Engl J Med. 1997; 337:382-7.

Prednisone is probably the most effective drug for treating rheumatoid arthritis, but it has many side effects, some serious. A recent review from the Mayo Clinic (4), covering two decades, reported at least one side effect in two thirds of patients with polymyalgia rheumatica who received either corticosteroids or NSAIDs and in 80% of patients given both drugs. On average, these patients received 9.6 mg of prednisone daily for about 2.5 years. They were two to five times more likely than matched controls to develop diabetes or fractures of the vertebrae or femoral neck (4).

Adachi and colleagues studied the use of a bisphosphonate, etidronate, for preventing steroid-induced bone loss. The study sample was 141 patients who had begun taking high-dose prednisone within the past 100 days and were expected to continue this therapy for at least 1 year. The chief indications for steroid therapy were polymyalgia rheumatica and rheumatoid arthritis, and the average daily dosage of prednisone (or equivalent) at study entry was 21 to 23 mg. Patients were randomly assigned to receive 400 mg of etidronate daily or a matching placebo for 2 weeks and then 500 mg of calcium carbonate daily for 11 weeks. The cycle was repeated three times, resulting in 1 year of treatment. The study patients were not systematically given vitamin D or hormone replacement therapy.

Etidronate protected against osteopenia, most impressively in postmenopausal women but also in younger women and men (Table 1). Clinically, the key is whether vertebral compression fractures can be prevented; in this study, etidronate made no substantial difference in premenopausal women or men and had only an equivocal effect in postmenopausal women. Nevertheless, intermittent etidronate therapy seems to be one way to blunt the osteopenic effect of steroid therapy.

Routine Screening for Hydroxychloroquine-Related Retinopathy Had Low Yield

Levy GD, Munz SJ, Paschal J, et al. Incidence of hydroxychloroquine retinopathy in 1,207 patients in a large multicenter outpatient practice. Arthritis Rheum. 1997; 40:1482-6.

Because current screening recommendations vary and none are evidence-based, Levy and colleagues reviewed 1207 patients who received hydroxychloroquine from 1991 to 1993 and had undergone a charted eye examination. Only 1 patient (0.08% of the study group) had definite retinal toxicity. Five others (0.4%) had probable retinopathy. No toxicity was found in the approximately 80% of patients who received less than 6.5 mg of hydroxychloroquine per kg of body weight daily. Therefore, routine ophthalmologic screening may not be warranted for these patients, especially in the era of managed care. Exceptions to this policy are patients who have received antimalarial agents for a prolonged period and those with renal insufficiency.

State-of-the-Art Laboratory Testing in Rheumatology

A survey of laboratory tests used to diagnose rheumatologic disorders yielded an article on antineutrophil cytoplasmic antibodies (ANCA) in patients with connective tissue disease, two studies on antinuclear antibodies in ostensibly healthy persons and in patients with lupus and other connective tissue diseases, and a look at the role of serologic testing in diagnosing Lyme disease.

Positive Result on Test for Antineutrophil Cytoplasmic Antibodies or Antiproteinase 3 Strongly Favors Systemic Vasculitis over Connective Tissue Disease

Merkel PA, Polison RP, Chang Y, et al. Prevalence of antineutrophil cytoplasmic antibodies in a large inception cohort of patients with connective tissue disease. Ann Intern Med. 1997; 126:866-73.

Two types of ANCA—antiproteinase 3 (anti-PR3) and antimyeloperoxidase (anti-MPO)—are associated with vasculitis. Immunofluorescence testing has disclosed neutrophilic cytoplasmic staining, the so-called c-ANCA pattern, in 91% of patients with active Wegener granulomatosis (5). This staining almost always reflects anti-PR3 as detected by enzyme-linked immunosorbent assay (ELISA). In contrast, anti-MPO is present in most patients with microscopic polyangiitis ("polyarteritis") or crescentic pauci-immune glomerulonephritis. It produces a perinuclear ANCA (p-ANCA) pattern. However, other autoantibodies also produce a p-ANCA pattern. Thus, a p-ANCA result is nonspecific; a specific ELISA is needed in this setting to detect anti-MPO.

Merkel and colleagues sought to determine how often ANCA testing yields false-positive results in patients with connective tissue diseases, which are part of the differential diagnosis of vasculitis. Within the first year of illness, serum samples were taken from 70 patients with systemic lupus erythematosus, 70 patients with rheumatoid arthritis, 45 patients with scleroderma, 36 patients with polymyositis, 165 patients with undifferentiated connective tissue disease, and 200 random blood donors. The study also included 26 controls with anti-PR3-positive vasculitis and 26 persons positive for anti-MPO. Immunofluorescent slides were interpreted by two experienced readers who were blinded to the clinical findings. Titers of both anti-PR3 and anti-MPO were measured by ELISA. The c-ANCA pattern was 100% specific and 89% sensitive for anti-PR3-positive vasculitis (essentially Wegener granulomatosis). The p-ANCA pattern was noted in about 80% of controls with anti-MPO-positive vasculitis but also in almost one third of patients with lupus (specificity, 69%). Antiproteinase 3 was found in no more than 3% of patients with other diseases and in none of the random blood donors. Antimyeloperoxidase was even less prevalent (1% of patients with rheumatoid arthritis or lupus) and thus was very specific. Atypical ANCA patterns were found in up to 40% of the various connective tissue disease subgroups. A cautionary note is that even among these highly trained readers, interpretation of ANCA immunofluorescence results differed greatly.

A major conclusion from this study is that identifying anti-PR3 or a c-ANCA pattern markedly increases the probability of systemic vasculitis (such as Wegener granulomatosis). A meta-analysis of four studies indicates that the positive likelihood ratio (the true-positive rate divided by the false-positive rate) of these two tests for Wegener granulomatosis is 65 (5). Another implication of Merkel and colleagues' study is that a p-ANCA pattern does not assure that anti-MPO is present; ELISA is required. Testing for ANCA by immunofluorescence requires experienced personnel.

Proper Cut-Off Antinuclear Antibody Titer Is Important in Diagnosing Systemic Lupus Erythematosus

Tan EM, Feltkamp TE, Smolen JS, et al. Range of antinuclear antibodies in "healthy" individuals. Arthritis Rheum. 1997;40:1601-11.

The traditional immunofluorescence assay for antinuclear antibody is considered 95% to 98% sensitive for detecting lupus, with reported false-positive rates of 2% to 5%. Up to one third of patients with rheumatoid arthritis have titers exceeding the cut-off point even though the titers are lower, on average, than those in patients with lupus.

Tan and colleagues asked 15 experienced laboratories throughout the world to test serum samples from 41 patients with lupus, 40 patients with rheumatoid arthritis, 37 patients with scleroderma, and 125 ostensibly healthy controls. Tests were done in the usual way for each laboratory except that all of them used HEP-2 cells as substrate. Each sample was tested blindly, twice in one laboratory and once in another. A positive result was recorded when both samples sent to laboratory 1 or the one tested in laboratory 2 were read as positive. This design may have generated slightly inflated true- and false-positive values.

At an antinuclear antibody titer of 1:40, the positive likelihood ratio for systemic lupus was only 3; at a titer of 1:320, it was 29 (Table 2). A negative result was more helpful in ruling out lupus than a low-titer positive result was in ruling in lupus; this occurred because the sensitivity is higher than the specificity. For determining how well antinuclear antibody testing distinguished between lupus and rheumatoid arthritis (rather than between lupus and no connective tissue disease), a positive test result of l:80 yielded a positive likelihood ratio of only 3. A higher titer of 1:320 increased the ratio to 29; only 3% of patients with rheumatoid arthritis tested positive. All serum samples from patients with scleroderma were positive at a titer of 1:40; 87% were positive at a titer of 1:160.

In this study, 1:40 as the cut-off titer for a positive antinuclear antibody result was very nonspecific. However, because of variation among laboratories, each laboratory must determine its own best cut-off point.

New Enzyme-Linked Immunosorbent Assays for Antineutrophil Cytoplasmic Antibodies Varied Widely in Sensitivity and Specificity

Emlen W, O'Neill L. Clinical significance of antinuclear antibodies: comparison of detection with immunofluorescence and enzyme-linked immunosorbent assays. Arthritis Rheum. 1997; 40:1612-8.

Given how time- and labor-intensive the immunofluorescence assay has proven to be and the problems of reproducibility among laboratories, ELISA is of interest because it is less costly and can be automated. Emlen and O'Neill contrasted the performance of commercial ELISA kits for antinuclear antibody with the performance of immunofluorescence testing. They used serum samples from 50 patients who met antinuclear antibody-independent criteria for lupus and 200 consecutive patients who were referred for antinuclear antibody testing. According to chart review, 9 of the latter patients had lupus and 12 had other forms of connective tissue disease. In addition to standard immunofluorescence assay using HEP-2 cells, specimens positive at borderline values were tested by immunodiffusion for anti-SS-A antibody. Serum samples were also tested by six commercial ELISA kits. Immunofluorescence testing identified antinuclear antibody in 88% of 50 patients with lupus. The sensitivity of ELISA ranged from 62% (GenBio, San Diego, California) to 90% (Helix Diagnostics, Sacramento, California). Among 156 patients with no evidence of connective tissue disease, false-positive antinuclear antibody test results were obtained by immunofluorescence assay in 33% of cases, by the Helix ELISA in 46%, by the Sanofi (New York, New York) ELISA in 18%, and by the GenBio ELISA in only 3%.

In general, agreement among the different ELISA kits was poor. Furthermore, test sensitivity varied inversely with specificity. Methods for ELISA that used whole nuclear extract were the most sensitive but the least specific. Clinicians must know which assay their laboratory is using; a sensitive ELISA is a reasonable substitute for immunofluorescence testing. However, because the positive predictive value of low-titer positive antinuclear antibody results on immunofluorescence assay or a sensitive ELISA is low, the antinuclear antibody test is more helpful in ruling out than ruling in lupus. The overall conclusion from this study is that even today, lupus remains essentially a clinical diagnosis.

Standardized Enzyme-Linked Immunosorbent Assay Accurately Detected Lyme Disease in Medium-Probability Settings

Tugwell P, Dennis DT, Weinstein A, et al. Laboratory evaluation in the diagnosis of Lyme disease. Ann Intern Med. 1997; 127:1109-23.

A lack of standardized ELISAs for Lyme disease has severely restricted the reliability of these tests. Early antibiotic treatment is known to limit late illness. As a result, 3 million serologic tests are done each year in the United States; for the most part, however, they are ordered when the probability of Lyme disease is low. For Tugwell and colleagues' clinical guideline on laboratory testing for Lyme disease, the authors performed a MEDLINE search, covering the years 1982 to 1996, that yielded 35 articles on ELISA for Lyme disease. Only 2 of these articles permitted calculations of sensitivity and specificity and provided reproducible information on sampling, clinical details, and the reference standard (6, 7). Only 4 of the 17 articles on Western blot testing met criteria for inclusion.

The implications of a positive or negative ELISA result, depending on the pretest likelihood of Lyme disease, are shown in Table 3. When the pretest probability of Lyme disease is low, most positive ELISA results are false and testing to establish the diagnosis is not indicated. When the pretest probability is high, patients should be treated without testing because even a negative ELISA result does not exclude disease. Testing is most helpful if the pretest probability of disease is intermediate. Western blot testing is indicated only if ELISA results are equivocal. The immediate challenge is for each laboratory to know the accuracy of the ELISA it uses and for clinicians to estimate the probability of disease on the basis of clinical findings (that is, exposure risk and objective disease manifestations).

Low Back Pain

Three articles on low back pain deal with the diagnostic value of lumbar radiographs, the efficacy of epidural steroid injections for sciatica caused by disk herniation, and the issue of whether employee education programs prevent low back injuries.

Early Lumbar Radiography for Low Back Pain Is a Low-Yield Procedure

Suarez-Almazor ME, Belseck E, Russell AS, et al. Use of lumbar radiographs for the early diagnosis of low back pain. Proposed guidelines would increase utilization. JAMA. 1997; 277:1782-6.

A basic reason why plain lumbar radiographs have not been more productive in patients presenting with acute low back pain is that the prevalence of specifically treatable disease—at least as encountered in the primary care setting—is low. Compression fracture, the most common such lesion, may be found in 1 in 25 patients; malignant disease, ankylosing spondylitis, and infection are far less common. Plain films miss one third of cancers and most epidural infections. Most identified abnormalities are frequently seen in persons who are not in pain and thus are nonspecific. Experts may differ considerably in the manner in which they read plain back films. Finally, there is the problem of gonadal radiation, much of which comes from oblique views that are rarely valuable (anteroposterior and lateral films suffice for routine purposes). All of these considerations led the Agency for Health Care Policy and Research [AHCPR] to propose in 1994 that lumbar radiographs be obtained during the first month of low back pain only when a "red flag" is present. These red flags are 1) for fracture: major trauma, minor trauma in older or potentially osteoporotic patients, long-term steroid use, age older than 70 years, or known osteoporosis and 2) for tumor or infection: age older than 50 years or younger than 20 years, a history of cancer, recent bacterial infection, constitutional symptoms, intravenous drug use, or pain that is severe at night or worse when supine.

In a retrospective study, Suarez-Almazor and colleagues evaluated the impact that these guidelines would have had on use of lumbar radiography for low back pain. Records of 963 patients were reviewed. The patients (mean age, 42 years) had experienced pain for less than 3 months and were seen in a primary care setting in Alberta, Canada, in 1992-1993. Minimum duration of follow-up was 2 years. Although 44% of patients had a red flag (usually age > 50 years), only 13% of all patients (and 19% of those with a red flag) had radiographs ordered at the first visit (more than half of the 13% had oblique-view radiographs).

In this cohort, only five fractures and three tumors were found, all detected by radiography and all in patients with a red flag; no infections were found. Had the AHCPR guidelines been followed, use would have increased by 238%, largely because of the nonspecific red flag of age older than 50 years. Thus, the AHCPR guidelines are sensitive but nonspecific. Age older than 50 years merely doubles the generally low pretest odds that a fracture or neoplasm is causing acute low back pain. A more cost-effective modification of the AHCPR guidelines would be to eliminate age older than 50 years as a red flag.

Epidural Steroids Are Not the Answer for Sciatica Caused by Disk Herniation

Carette S, Leclaire R, Marcoux S, et al. Epidural corticosteroid injections for sciatica due to herniated nucleus pulposus. N Engl J Med. 1997; 336:1634-40.

Much of the current practice in treating low back pain or sciatica conflicts with the evidence showing that most therapies are unproven or have been disproved. Disk surgery is reserved for the approximately 10% of patients with disk herniation whose disk-induced symptoms and signs do not resolve after 4 to 6 weeks. According to a recent meta-analysis of 12 controlled trials of epidural steroid injection for sciatica, only half of these trials demonstrated benefit. Most of these trials were not well designed (8).

In a double-blind study, Carette and colleagues randomly assigned 156 patients whose sciatica had been present for 4 to 12 months to receive, at 3-week intervals, up to three injections (done without radiographic guidance) of methylprednisolone acetate or saline. All patients had a positive result on a straight-leg-raising test or had a neurologic deficit, and herniated nucleus pulposus at the appropriate level was confirmed by imaging. Approximately one third of patients in each group markedly improved after 3 weeks, and 55% in both groups were much better at 3 months. More saline recipients than methylprednisolone recipients (44% compared with 33%) had returned to work by this time. Patients in each group received an average of two injections.

Of all studies done to date, this study is probably the strongest in terms of methodology. Although injections done without radiographic guidance may miss the epidural space 30% of the time, most patients in this study probably had at least one correctly placed injection of steroid. The conclusion, which seems robust, is that epidural steroids did not provide significant benefit for sciatica caused by herniated nucleus pulposus.

Back School Did Not Prevent Industrial Back Injury

Daltroy LH, Iversen MD, Larson MG, et al. A controlled trial of an educational program to prevent low back injuries. N Engl J Med. 1997; 337:322-8.

This randomized, controlled trial of "back school" was conducted in 4000 postal workers. Half of the participants were taught how to lift and handle heavy objects safely. During the 5-year follow-up, unschooled employees had the expected rate of low back injury: 2% per year. The employees who attended back school actually had a 10% higher rate and a median total cost per back injury of about $300; this was about threefold higher than the cost in the control group. The authors concluded that back school did not alter behavior. Of course, the problem could have been that it is not clear just what should be taught.

A Rheumatology "Grab Bag"

Doxycycline Was as Effective as and Safer Than Ceftriaxone for Treating Nonmeningitic Lyme Disease

Dattwyler RJ, Luft BJ, Kunkel MJ, et al. Ceftriaxone compared with doxycycline for the treatment of acute disseminated Lyme disease. N Engl J Med. 1997; 337:289-94.

In this nonblinded study, sponsored by the manufacturers of ceftriaxone, Dattwyler and colleagues compared ceftriaxone with doxycycline for the treatment of acute disseminated but nonmeningitic Lyme disease. The 140 patients, who had documented organ involvement or multiple erythema migrans lesions, were randomly assigned to receive 2 g of ceftriaxone daily for 2 weeks or 100 mg of doxycycline twice daily for 3 weeks. Respective clinical cure rates were almost identical: 85% and 88%. Only one documented treatment failure occurred in each group. Diarrhea was the most common adverse effect, noted in 37% of the ceftriaxone group but only 6% of the doxycycline group. Oral doxycycline, which is easily administered, less costly, effective, and relatively nontoxic, would seem to be the best choice for acute nonmeningitic Lyme disease. Patients with cardiac involvement may be an exception, but the study could not determine this because few such patients participated.

Color Duplex Ultrasonography May Help Diagnose Temporal Arteritis

Schmidt WA, Kraft HE, Vorpahl K, et al. Color duplex ultrasonography in the diagnosis of temporal arteritis. N Engl J Med. 1997; 337:1336-42.

This study from Germany addressed the accepted fact that, until recently, the only way to confirm temporal (or giant-cell) arteritis had been to obtain a biopsy specimen from the vessel. Even biopsy may yield a false-negative result, however, if a "skip area" is sampled. In the hope of finding an accurate noninvasive measure, color duplex ultrasonography was done by two physician-ultrasonographers in 30 patients with a diagnosis of temporal arteritis (21 with biopsy confirmation of the disease) and 82 controls without temporal arteritis. A hypoechoic halo surrounded the temporal artery lumen in 22 of the 30 patients with temporal arteritis (73%) and in 16 of the 21 patients with biopsy confirmation. Of note, none of the controls had a halo (that is, no false-positive results occurred). If this degree of specificity can be replicated elsewhere, temporal arteritis could be definitively diagnosed without biopsy.

Regular Exercise Reduced Pain and Disability from Osteoarthritis of the Knee

Ettinger WH Jr, Burns R, Messier SP, et al. A randomized trial comparing aerobic exercise and resistance exercise with a health education program in older adults with knee osteoarthritis. The Fitness Arthritis and Seniors Trial (FAST). JAMA. 1997; 277:25-31.

Because short-term studies suggest that regular exercise helps to relieve symptoms of osteoarthritis of the knee, Ettinger and colleagues randomly assigned more than 400 patients to one of three measures—aerobic exercise, resistance exercise, or a health education program—for 18 months. Both exercise regimens led to modest improvement in joint pain and disability. Patients with osteoarthritis of the knee should be encouraged to walk regularly; they will probably feel better, and their pain will not worsen.

A Quinine-Like Drug Was Active against Idiopathic Muscle Cramps

Jensen PH, Veenhuizen KC, Wesseling AI, et al. Randomised controlled trial of hydroquinine in muscle cramps. Lancet. 1997; 349:528-32.

In this blinded, placebo-controlled trial, Jensen and colleagues sought to determine whether the quinine-like drug hydroquinine, given in a dosage of 300 mg/d for 2 weeks, is an effective treatment for muscle cramps of unknown origin (including nighttime cramps). Muscle cramps were reduced by 50% in two thirds of patients given the active drug and only one fifth of placebo recipients. The response continued during a 2-week washout period. This was a small-scale and short-term study, and it may have missed potential side effects. Nevertheless, a trial of hydroquinine may be worthwhile in patients who frequently experience this distressing symptom.