ANTIBIOTIC THERAPY OF RHEUMATOID ARTHRITIS:
AN OBSERVATIONAL COHORT STUDY OF 98 PATIENTS WITH 451 PATIENT-YEARS OF FOLLOW-UP

Congress of Rheumatology. 1985:S85

Thomas McP. Brown, M.D. *. Joel W. Novak, M.S **; Marc C Hochberg, M.D., M.P.H.***; Anne S. Lindblad; M.S.**; Catherine A. Novak, M.S.N.**; Norma Morin, Ph.D.*** and Harold W. Clark, Ph.D.*

from
*The Arthritis Institute of the National Hospital, Arlington, VA
**The EMMES Corporation, Potomac, MD
***The Johns Hopkins Medical Institutions, Baltimore, MD

ABSTRACT

This is a retrospective observational study of 98 patients with definite or classical RA who began treatment with antibiotic therapy during the five-year period 1978-1982. The report details patient characteristics, patterns of treatment with antibiotics, NSAIDs, steroids, and SAARDs where applicable. Outcomes to treatment, including toxicity, symptomatic relief, activity level, and duration of treatment are described. The probability of remaining on antibiotic therapy, failure-free, at five years is 70%, based on 451 patient years of follow-up. Symptomatic improvement is reported. Treatment-related complications with long-term antibiotic therapy were acceptable.

Current concepts regarding the management of rheumatoid arthritis (RA) emphasize medical therapy, physical and occupational therapy, and patient education. The first-line agents used for medical therapy are the non-steroidal anti-inflammatory drugs (NSAIDs). Patients who are not adequately controlled by NSAIDs are usually treated with one or another of the slow-acting anti-rheumatic drugs (SAARDs) such as hydroxychloroqulne, gold, or penicillamine. These drugs are thought to modify the course of RA--as evidenced by a decrease in the signs and symptoms of inflammatory synovitis, a decrease in rheumatoid factor titers, and a decrease in the rate of formation and/or progression of radiographic joint erosions.1,2 Unfortunately, the efficacy of the SAARDs is not uniform and termination of treatment due to toxicity is common--with the result that many patients continue to demonstrate progressive, deforming and disabling disease requiring reconstructive orthopedic surgery.

Patients who fail on SAARDs and/or who develop toxicity to these compounds become candidates for treatment with cytotoxic agents such as azathioprine, methotrexate, and cyclophosphamide. Unfortunately, those alternatives also place the patient at unusual risk of severe and even life-threatening complications including leukopenia, thrombocytopenia, opportunistic infections, and long-term oncogenesis.1,3,4 Clearly, other therapeutic regimens are needed which are capable of producing long-term disease control without the undue risk of serious treatment-related morbidity.

Recent studies in both the pathogenesis of RA and in the treatment of a variety of arthritic disorders provide some basis for considering the use of antibiotic therapy in the treatment of RA. Consider the model of pathogenesis suggested by Harris: the genetically predisposed host develops a complex cellular and humoral immune response to an unknown antigenic stimulus which localizes in the joints, where it leads to the development of proliferative synovitis and joint destruction.5 The nature of the genetic predisposition appears, at least in most individuals with seropositive RA, to be related to the major histocompatibility gene products HLA-DR4, MB3, and MT3.6,7 While the nature of the antigenic stimulus remains undefined, an exogenous infectious agent(s) has been suspected'for many years. Bennett, in reviewing the role of infectious agents as etiologic factors in RA, has noted "...that a whole variety of organisms and/or their products may be responsible for initiating..." rheumatoid arthritis and, furthermore, that "when the scope of processes that can be associated with infection and arthritis is realized, one becomes aware of the extraordinary difficulties involved in evaluating the possible etiology of rheumatoid arthritis from this standpoint."8

One class of infectious organisms in particular, mycoplasma, has long been suspected of having a role in the etiology of RA12. These organisms have been associated with arthritis in numerous animal species including the pig,13 turkey,14 rat,15 mouse,115 and gorilla.16 Furthermore, they have been shown to initiate an arthritic process of long duration that is typified by a chronic, progressive course. However, because it has been difficult to isolate the organisms from joints, except during the early phases of the disease,13,17,18 acceptance of the mycoplasma theory of etiology has been limited.

Clinical applications of the microbial theories of pathogenesis to the treatment of a variety of arthritic conditions likewise have received attention. Recent reports by Steere and colleagues on the efficacy of antibiotic therapy in treating a chronic Lyme arthritis--which demonstrates an articular course similar to that seen in RA including pannus formation, and cartilage and bone destruction,19,20--lend support to the idea of treating arthritis

-2-

with antibiotics. Another disorder in which a bacterial infection has been related to an arthritis that is similar pathophyslologically to RA is bypass arthritis,21 wherein some patients who have undergone jejuno-ileal bypass for morbid obesity develop an immune-complex-mediated polyarthritis that clinically resembles RA. Utsinger has shown that circulating immune complexes containing bacterial antigens, antibodies and complement are present in the sera of almost all of those patients and can be identified in cryoprecipitates from synovial fluid specimens.22 Invariably, these patients have responded to treatment with Tetracycline used in conjunction with anti-inflammatory agents.21

Tetracycline and its derivatives have also been used for over 35 years by Brown, of the Arthritis Institute, National Hospital (AINH), to treat patients with RA.19 The scientific underpinning of Brown's work is rooted in a theory of causation which involves infection by inycoplasma.9-11 The only randomized, controlled clinical trial of Tetracycline in RA (Skinner et al) concluded that there was "... no significant benefit from long-term Tetracycline therapy in patients with classic and definite rheumatoid arthritls."24 However, given its small sample size (n = 30) and the relatively brief observational period (1 year), Skinner's study provides weak support for what must be considered a far-reaching conclusion.

The purpose of this study was to provide an assessment of antibiotic therapy in recently treated patients with RA; more specifically, the study sought to characterize the course, duration and complications of antibiotic therapy used by Dr. Thomas McP. Brown in patients with RA--and to assess available patient outcomes such as symptomatic improvement and ability to tolerate and continue therapy. Toward that end, a retrospective assessment of all of Brown's patients with RA who began antibiotic treatment in the five-year period January 1, 1978-December 31, 1982, was undertaken.

MATERIALS AND METHODS

The study was requested by Brown and funded by the AINH; it was designed and carried out by an independent statistical organization, The EMMES Corporation, located in Potomac, Maryland. The fim was solely responsible for the study design, the abstraction of primary medical records and the summarization and analysis of patient data reported in this manuscript. J.N, the firm's president, served as project director and statistician; N.H., Associate Professor of Medicine (rheumatology) and Epidemiology at the Johns Hopkins Medical Institutions, served as an independent clinical and epidemiologic consultant.

A three-phase study plan was executed. During the first phase, the series of patients to be evaluated was identified. In the second phase, hospital and clinic records for those patients were abstracted; and information pertaining to selected patient characteristics, course of treatment, current treatment status, and response to treatment were captured. A third phase was introduced after study commencement to obtain more current patient information than was available from the charts reviewed.

Phase I. Hospital and clinic records for all patients hospitalized by Dr. Brown between January 1, 1978 and December 31, 1982, at the National Hospital were reviewed. The abstractor/reviewers were registered nurses with a minimum of one year of experience in rheumatology. All abstractors attended a training session which covered study objectives, search procedures and the proper use of precoded forms which had been prepared for the study by The EMMES Corporation. The trainer and data coordinator (C.N.) is a nurse specialist in rheumatology and is a member of the Arthritis Health Professions Association section of the Arthritis Foundation.

To have been eligible for inclusion in the study series, patients must have begun antibiotic therapy during the designated period and, based on available information, must have been classifiable as a definite or classical case of RA per the revised criteria of the American Rheumatism Association (ARA).21 Patients who met fewer than five of the ARA criteria were excluded from this study. In addition, patients who had a history or current evidence of inflammatory rheumatic disease other than RA, or who were less than 16 years of age at the time of disease onset were disqualified from study.

Of the 539 cases reviewed in Phase 1, 441 were disqualified from study: of those, 169 were found to have begun treatment before January 1, 1978; available records for 267 failed to indicate that the patient met five or more of the ARA criteria; and five additional patients were found to have a disqualifying coexistent disease: Reiter's disease (3), systemic lupus (1), juvenile RA (1). Ninety-eight patients met all criteria for inclusion in the study; they form the basis of this report. (See Table I.)

Phase II. All available hospital and clinic records were abstracted for the 98 patients included in the study. Patient characteristics and the course of all medical treatment received during the study period were noted. Special attention was given to the administration of antibiotics, NSAIOs, steroids, and SAARDs. All complications of treatment were recorded, as were available baseline and follow-up measures of disease status. Of particular interest were patient-reported counts of the number of tender joints, the number of painful joints, and the duration of morning stiffness. Also included were patient reports of amount of weakness, fatigue, endurance, feeling of well being, mental acuity, and periodic depression which had been collected by Dr. Brown using the familiar zero-to-four scale.

Phase III. In order to obtain current information concerning treatment and and disease status, attempts were made to contact all 98 patients by questionnaire. At that time, all patients were also asked to complete and return the Stanford Health Assessment Questionnaire (HAQ).22 Forty-eight of the 98 patients returned the two questionnaires within four weeks of solicitation; 43 of the remaining 50 patients were contacted by phone and interviewed. Follow-up on six patients who died during the study period and one patient who resided in a nursing home was not pursued. For those seven patients, information obtained at the last recorded visit was compared to baseline measures to obtain outcome values. All telephone interviews were performed by Dr. N. Morin, an epidemiologist with experience in rheumatology and in the use of the HAQ.

Criteria. The outcomes of particular interest in this study were treatment failure, time to treatment failure, patient reports of symptomatic relief, and toxicity. Patients were considered to be treatment "failures" for the following reasons: (a) the termination of antibiotic treatment due to lack of efficacy, toxicity or for unknown reasons, (b) the addition of SARDs to the treatment regimen, or (c) any increase in the dose of steroids because of worsening disease. Treatinent duration was calculated from date of treatment start to the last known date of treatment with antibiotics without the use of SAARDs or increased use of steroids.

Patients reporting a 50% decrease in the number of tender joints, painful joints, or duration of morning stiffness were considered to have demonstrated a meaningful improvement in clinical status.23 Similarly, patients who demonstrated at least a two-grade improvement on a four-point scale in the amount of weakness, fatigue, endurance, feelings of well being, mental accuity or periodic depression were considered to have exhibited a clinically meaningful improvement in those parameters.

Statistical Methods

All data were double-keyed and stored in an IBM mainframe. Tailored computer edit routines were used to assure data consistency. Kaplan-Meier estimates of the time to treatment failure were calculated and a chi-square statistic was used to evaluate time trends in response to treatment.24 The proportional hazards methods of Cox, as implemented by SAS, were used to evaluate the significance of prognostic variables.25 Patients who died of unrelated causes or who terminated treatment because of remission of symptoms or unavailability of drug supply were censored in the analysis.

RESULTS

Patient Characteristics. Characteristics of the 98 study patients are summarized in Tables 2 and 3. Mean age of the patients at commencement of antibiotic therapy was 52 years; only 3% of patients were 70 years or older at the time antibiotic therapy began. Sixty-two percent of patients were female, and 86% were married at the time antibiotic therapy began. Almost all patients were white; most had been diagnosed as having RA for five or more years prior to beginning antibiotic therapy.

Eighty-seven of the 98 patients studied (88%) began their antibiotic therapy during the first three years of the five-year study period. The sharp decline observed in the number of patients beginning treatment during 1981 and 1982 has been attributed to Dr. Brown's full schedule during that time. Two-thirds of the patients studied were classified as having definite RA; while one-third were classified as having classical RA. Morning stiffness, joint pain and swelling had been experienced by 94% of the patients, respectively; all three symptoms were reported for more than 90% of the group. The frequency with which other ARA criteria and symptoms were found in the study group are summarized in Tables 3 and 4. Forty percent of patients had no previous therapy with SAAROs, e.g., gold, penicillamine, and antimalarials; no patient had received cytotoxic agents prior beginning antibiotic therapy. Fifty-five of the patients (56%) had received gold before beginning antibiotic therapy; 41 patients (42%) were receiving corticosteroid therapy with 12 of those receiving more than 10 mg per day.

Patterns of Treatment. During the course of treatment, a variety of antibiotics, doses, schedules, and routes of administration were utilized. (See Table 5.) Patterns of treatment varied widely among patients as well as within patients. Almost all (98%) received both oral and intravenous (IV) antibiotics. During the course of their therapy, 71% of patients received three or more agents orally and 53% received three or more agents via the intravenous route. Vibramycin (61%), Minocin (67%), and Ampicillin (55%) were used most commonly for oral administration, while Lincocin (80%), Cleocin (60%) and Vibramycin (55%) were most commonly used for IV administration. Erythromycin, Tetracycline and Amoxicillin were also used with notable frequency.

While specific regimens used may have varied, the general pattern of treatment tended to be consistent. Patients were generally given intravenous Tetracycline (or one of its derivatives) daily for 10 days and this was repeated at six-month intervals; most patients also received small doses of oral antibiotics such as Vibramycin 50 mg or Minocin 100 mg three times per week between IV administrations. Logistics (distance from hospital and clinic) and disease activity were found to influence schedule, dose, and and drug selection. During periods of extreme disease activity, the frequency of intravenous treatments or dose of oral antibiotics was increased; and, prescription of an alternate Tetracycline derivative or modification of dose was used to manage side effects such as diarrhea or vaginitis, resistance to the antibiotic or poor absorption of drug. Intramuscular injections were occasionally substituted for intravenous treatments. Thirty-nine percent of patients also received intraarticular antibiotic therapy, which was given in conjunction with steroids to 33 of the 38 patients.

No patient was treated with antibiotics alone; one or more NSAIDs were used in all patients at some time in the course of their treatment with antibiotics. Eighty-four percent of patients received one or another of the non-salicylate NSAIDS, 51% received aspirin, and 69% received some other medication for pain, usually acetominophene or propoxyphene.

Of the 41 patients who were using steroids at the commencement of antibiotic therapy, 17 were able to end or decrease by 50% the amount of steroids used, and were continuing antibiotic therapy as of last contact. Two patients, initially on steroids, increased their steroid dosage without a later dose reduction; and six patients not originally on steroids commenced their use after beginning antibiotic therapy. Of the eight patients who increased their use of steroids, five experienced a worsening in their disease status while on antibiotics, two experienced a worsening of their disease after they were unable to obtain further treatment with antibiotics, and one patient began taking steroids to control allergies.

Complications of Treatment. The antibiotic therapy administered was generally tolerated quite well. Thirty-eight percent of patients complained of one or another complication of treatment: nausea 14%, diarrhea 13%, rash 9%, and vaginitis 7%. Multiple complications, in either time or kind, were uncommon. While severity of these complications could not be meaningfully graded from the available records, there was no indication that they were particularly serious. Four patients indicated that they terminated treatment because of adverse reactions and have been counted as treatment failures: diarrhea (1), stomach upsets (1), nausea (1), and severe headaches (1). It is noteworthy that all four of those patients were advised to terminate antibiotic treatment by a physician other than Dr. Brown. Neither hematologic complications nor proteinuria were noted.

Treatment Outcomes. Twenty-seven of the 98 patients (28%) who began antibiotic therapy are considered to have failed treatment. (See Table 6.) Nineteen of those patients terminated antibiotic treatment because of complications (4), lack of treatment efficacy (13), or for unknown reasons (2). The remaining eight patients, though they were last known to be continuing antibiotic therapy, are considered to be treatment failures because of: increased use of steroids (4) or initiation of adjunctive treatment with SAARDs (penicillamine (2) and antimalarials (2)). The duration of therapy for the 27 treatment failures ranged from 9 days to 4 years; the median time to treatment failure for those patients is estimated to be 16 months.

Results of a life table analysis of times to treatment failure for the 98 study patients are summarized in Table 7, and they are graphically presented in Figure 1. As can be seen, our best estimate of the chance that a similar group of patients would remain on antibiotic therapy failure free for one year is 90%; for two years It is 84%; for three, four and five years it is 77%, 71% and 70% respectively. Taking sampling error into account, we are 95% confident that > 59% of patients would be continuing antibiotic therapy failure free as defined above after four years of observation. Time to treatment failure was not found to be statistically related to sex, duration of disease, ARA classification (viz classical or definite RA) or prior treatment with gold or steroids (p > .1)

Symptomatic Relief. Patient reports of symptomatic relief were reviewed and summaries are provided in Tables 8 and 9. Eighty-four percent of patients for whom baseline and follow-up information were available reported a 50% or greater decrease in the number of tender joints, the number of painful joints, or the duration of morning stiffness. Forty-nine percent of such patients experienced > 50% improvement in at least two of the three symptoms and 31% of evaluable patients reported > 50% Improvement in all three symptoms. A trend analysis of the data indicates a statistically significant association between year in which therapy was initiated and response to treatment (p = .001). Symptomatic improvement with respect to weakness, fatigue, endurance, feelings of well-being, mental accuity, or depression was observed in three quarters of all patients; improvement on multiple criteria was common. Here again, rate of improvement on these criteria was associated with year of treatment start (p = .05).

Stanford Health Assessment Questionnaire (HAQ). Mean values were calculated for the three parameters measured by the HAQ (viz pain, global assessment of disease activity, and ability to function), as were bounds for the 95% confidence intervals around the three means (Table 10). Similar calculations are presented for two subgroups: (a) treatment failures and (b) patients who cannot be considered as failures at least at this time. Patients who have been identified as failures consistently demonstrated scores suggesting increased levels of pain, worsened global assessment of disease activity, and reduced ability to function independently; there is no overlap among any of the three sets of 95% confidence intervals.

DISCUSSION

Ultimately, the efficacy of any treatment can only be established through well-designed and well-executed controlled clinical trials. The findings of this uncontrolled retrospective study, however, do suggest that antibiotic therapy, e.g., Tetracycline and its derivatives, has been of value in managing these 98 patients with classical or definite rheumatoid arthritis--60% of whom had received one or another of the slow-acting anti-rheumatic drugs prior to starting on antibiotic therapy, three-quarters of whom were seropositive for rheumatoid factor, and all of whom had clinically active disease.

Antibiotic therapy was generally well tolerated; although 38% of patients reported an adverse reaction, only four patients discontinued antibiotic therapy for toxicity. Treatment failure, which was defined here as (a) the termination of antibiotic therapy for lack of efficacy, toxicity or unknown reasons, (b) the increased use of adjuvant therapy with steroids, and/or (c) the initiation of adjunctive therapy with any of the slow-acting antirheumatic drugs was seen in only 27 patients (28%) after a total of 451 patient-years of observation. Life table analyses indicate that the probability of remaining on treatment and free of failure as defined for five years is 70%. In addition, the majority of patients studied reported meaningful improvement in the number of painful and tender joints and in the duration of morning stiffness; meaningful improvement was also reported on all of the subjective symptoms evaluated, viz weakness, fatigue, endurance, feelings of well-being, mental accuity, and depression. And, while the positive association observed between time on treatment and the extent of symptomatic relief may be explained in a variety of ways, clearly one possibility is that longer periods of therapy lead to greater therapeutic benefit. Thus, antibiotic therapy appears to be associated with an ability to maintain a majority of patients with RA in either a stable or improved clinical state over a long period of time with a low degree of toxicity and no serious life-threatening adverse effects.

As with all retrospective and/or observational studies, these findings must be evaluated with great caution; and, interpretations must take into account the risk of bias associated with patient selection and evaluation. In the current study, efforts were made to minimize some of the more obvious sources of bias that might have influenced study findings: an independent statistical center and rheumatologist were used to design and carry out the study; to the extent possible in a retrospective study, the patients form a series--all of whom met ARA criteria for definite and classical RA; follow-up for almost all patients was successfully pursued; and criteria for treatment failure and clinical improvement were set forth before the study effort began and not afterward. It is noteworthy that data gathered using the Stanford Health Activity Questionnaire support the idea that the criteria used to determine "treatment fallure" were clinically meaningful.

Alternate Therapies. The findings reported here also compare favorably to the historical experience of other investigators who used either placebo or SAARDs although given the lack of standardization in reporting methods26 and the possibility that significant differences existed in patient populations, drug regimens and outcome measures, such comparisons must be

interpreted with great caution. As noted above, the data reported here show that Brown maintained 84% of patients on antibiotic therapy, failure-free, at two years; in comparison, Sigler, using gold, was able to maintain 13 of 16 patients (81%).on therapy for 24 months.27 However, the findings of several prospective observational studies which reported on the long-term efficacy of gold therapy do not compare well to Brown's experience of 71% and 70% treatment survival at four and five years, respectively. In particular, Rothermich and colleagues reported that 40 of 97 patients (41%) with definite or classical RA were improved and still on therapy after one year of gold therapy--but that only 15% of those patients were receiving gold at four years.28 In another report by Rothermich et al, 91 of 171 patients (53%) who received gold therapy had terminated treatment because of lack of response or toxicity by 12 months; and that at the end of four years of observation, only 11% of patients remained on therapy.29 Kean and Anastassiades30 and Sharp and colleagues similarly noted the lack of sustained benefit from chrysotherapy.31,32

Long-term results with hydroxychloroquine are hardly better. Runge reported on a life-table analysis of treatment failure for 101 patients with RA treated between 1970 and 197733 indicating that median time to treatment failure was 13 months--and that the major cause of termination was lack of efficacy. In that study, only 14 patients (14%) were still receiving hydrochloroquine and under observation at 50 months--at which time the rate of treatment failure had reached 70% with over 45% of failures due to lack of efficacy.

While only a few long-term observational cohort studies of penicillamine therapy in RA have been reported,34-36 these reports, too, are not favorable.

Hill34 has noted that about 15% of patients discontinued therapy during the first six months and that an additional third of the patients studied discontinued during the second six months. Munthe and Kass,35 reporting on 266 patients with definite or classical RA treated with penicillamine over an eight-year period have indicated that 40% of patients discontinued therapy with penicillamine because of adverse effects or lack of response; and Runge, reporting on 40 patients who received penicillamine indicated that the treatment termination rate was 50% at 18 months.33 Thus, antibiotic therapy, as received by the patient cohort reported herein, appears to compare favorably with the reported experience with currently prescribed SAARDs--particularly in terms of complications and treatment duration.

Several recent studies which used placebo-treated control groups also offer interesting reference points for assessing antibiotic therapy. In a study which was both too small and too short to be informative, Skinner et al found that at one year 14 of 16 (88%) patients were still on placebo therapy, while 13 of 14 (92%) were still being treated with Tetracycline.20 In contrast, Ward et al reported that 86% of patients assigned to a placebo group remained on therapy at 21 weeks,23 and Wenger and associates noted that only 54% of patients assigned to a placebo control group remained on therapy at 6 months.37 While none of the three studies unequivocally establishes the short-tern failure rate for placebo therapy in patients with definite or classical RA, the Ward and Wenger reports are certainly more in keeping with the common clinical experience than that reported by Skinner et al. And, while the ability to maintain patients with definite and classical RA on placebo therapy for four and five years remains completely unevaluated, the common experience once again suggests that it would not be close to the 50-70% rate being reported here for patients who received antibiotic therapy.

An Hypothesis. If antibiotic therapy is effective in RA, as is suggested by these data, the mechanism of action may be similar to that observed in the treatment of Lyme disease38 with antibiotics or of Tetracycline in the treatment of bypass arthrltis.39 That is, if mycoplasmas are an inciting agent in RA, then either or both of the mechanisms hypothesized by Bennett may be operative: (a) the mycoplasma may localize within the joint space and trigger an immune response, and/or (b) the mycoplasma may localize in a distant site and trigger a systemic immune response which produces synovitis. Either hypothesis is consistent with the apparent beneficial effects observed in these 98 study patients.


Click here to go back to the Studies page.