Recent investigations have provided evidence that tetracyclines are moderately effective drugs in the treatment of early RA [7–12, 15, 16]. Their efficacy appears to be much less robust in long-standing disease. Despite this, the mean duration of RA for NDB participants initiating minocycline or doxycycline was 14.8 years; this is longer than previous studies by Kloppenburg and colleagues (12 ± 10 years) and Tilley and colleagues (8.4 ± 8.6 years), in which long-standing disease showed only modest improvement with minocycline treatment [8, 9]. Additionally, minocycline and doxycycline users in the current study probably had refractory RA, as indicated by more frequent prednisone use and increased lifetime exposure to DMARDs and biologic agents. This was supported by the Cox regression data, which showed that patients initiating minocycline and/or doxycycline had increased disease activity and greater cumulative exposure to nonbiologic DMARDs. These findings are similar to a small retrospective analysis of minocycline at a community hospital, in which the mean duration of RA was 18 years and patients had failed treatment with two to eight other DMARDs . Although statistically significant, it is unlikely that small differences in age, race, physical component summary score, and insurance type are clinically relevant. These findings suggest that providers have not embraced tetracyclines as primary treatment options for RA and reserve their use primarily as salvage therapy in patients with long-standing disease that have failed other agents. These results stand in contrast to double-blind, placebo-controlled trials in early, DMARD-naïve RA patients in which minocycline resulted in significant short-term disease improvement  and long-term remission , and proved more efficacious than hydroxychloroqine . Likewise, doxycycline in combination with methotrexate has proven more efficacious than methotrexate monotherapy . Moreover, results of a meta-analysis of tetracyclines found meaningful effects on tender and swollen joint counts, erythrocyte sedimentation rate, and patient-reported pain , comparable with hydroxychloroquine , sulfasalazine , and gold .
Contemporary studies of minocycline and doxycycline in RA coincided with the emergence of biologics. Rheumatologists now have an expanding arsenal of these potent agents available, in addition to conventional DMARDs. Antibiotics are generally considered late in the disease process after other standard treatments have failed. While we do not argue that minocycline and doxycycline should be used as first-line agents, they should be considered with other second-line DMARDs as options for combination therapy with methotrexate and for patients who are reticent to try conventional therapies or cannot afford them. Additionally, recent research into the possible role of oral filamentous bacteria in the pathogenesis of RA may renew interest in the therapeutic role of antibiotics [30–34].
Despite concerns about drug toxicity, the current study found that minocycline and doxycycline are generally well tolerated. Side effects were reported by only 17.8% of minocycline users and 11.8% of doxycycline users, most of which were of moderate severity. Less than one-half of these patients discontinued minocycline or doxycycline because of side effects, suggesting that they are at least as well tolerated as other second-line RA medications, for which drug toxicity causes discontinuation in 15% of all patients . For minocycline, cutaneous side effects were the most common, accounting for over one-half of all patient reports. This is consistent with previous research, in which hyperpigmentation has been reported in 40% of chronic minocycline users [36, 37]. Since we did not have dedicated coding for hyperpigmentation, we speculate that this side effect was variously reported as skin, other, photosensitivity, purpura, and rash. Dizziness was reported less frequently than in previous reports, accounting for only about 10% of all reported side effects [8, 9, 38]. Doxycycline side effects were also similar to other studies, with gastrointestinal and skin manifestations being most common . Interestingly, dizziness was nearly as common with doxycycline use (8.2%) as with minocycline (9.5%).
Nearly one in five rheumatologists had prescribed minocycline or doxycycline, and about 10% of patients had used them at some point in their disease course. However, most doctors had only prescribed these treatments for one or two patients. Although we do not have data to explain this pattern, it is possible that tetracyclines are used to treat a niche population of RA patients, such as those who are reticent to try more conventional DMARDs for fear of potential toxicities or those who may have requested from their physician an antibiotic-based treatment. The median duration of minocycline (6 months) and doxycycline (3 months) treatment was shorter than for other second-line DMARDs . This is not surprising given that our cohort had long-standing, refractory disease, thus limiting the efficacy of any treatment option. Additionally, short-term use of minocycline and doxycycline for treatment of infectious processes may have skewed our results. For those who initiated minocycline or doxycycline during NDB observation, 28% were stopped following the first month of treatment, with side effects accounting for 17% of these discontinuations. Since it is unlikely that minocycline and doxycycline would be stopped so quickly if initiated as DMARDs, some patients probably received these drugs for short-term antibiotic use. When patients with courses of minocycline and doxycycline equal to or less than 30 days were eliminated via a sensitivity analysis (n = 133), however, the sociodemographic and disease characteristic results were not meaningfully changed (data not shown). Excluding patients with 1 month or less of treatment had only modest effects on the median duration of minocycline (7 months) and doxycycline (6 months) therapy.
The present study had several limitations. In the analysis of socioeconomic and disease characteristics, minocycline and doxycycline users were combined to improve power. These two groups possibly have distinctive qualities, although there were no significant differences when we compared minocycline and doxycycline users with one another (data not shown). The current analysis also did not have access to laboratory markers such as the rheumatoid factor status of patients; seropositivity may be predictive of minocycline and doxycycline efficacy in early RA [10–12, 15], thus impacting timing of drug initiation. Finally, our findings may not be generalizable to racial/ethnic minorities as Caucasian patients made up about 90% of our cohort, with African Americans and Hispanic patients accounting for 4.7% and 2.9% of the sample, respectively. Despite these limitations, this is to our knowledge the first study to examine minocycline and doxycycline use in a large number of community-dwelling RA patients. Most clinical trials of minocycline and doxycycline in RA have been fairly small, and systematic reviews of drug toxicities have relied upon case reports and trials with diverse dosing strategies, patient populations, and drug indications. By focusing on community-dwelling patients, we were able to investigate the real-world application of tetracycline therapy in RA.