Acne, Antibiotics, and Upper Respiratory Tract Infections

In this issue of the ARCHIVES, Margolis et al¹ report for the first time that there is a statistically significant association between antibiotic use in acne and the incidence of upper respiratory tract infections (URTIs). Their findings will likely stimulate considerable discussion, especially given the increasing public awareness and concern about antibiotic use in general.

Although the authors are careful to point out that their study does not prove causation, their finding of an approximately 2-fold increased odds of URTIs associated with antibiotic use in acne is important and deserves further examination. If subsequent studies also confirm this association, current treatment of acne may change. In this editorial, we address the methodology of the study and the plausibility of its results.

Overall, Margolis and colleagues report a methodologically sound cohort study using a reliable database and a large sample of more than 118 000 patients. They control for several potential confounders in their analysis, thus reducing selection and ascertainment biases. However, as recognized by the study authors, an inherent limitation of cohort studies involves selection bias owing to unmeasured confounders—either known or unknown—that can only be adequately addressed through randomization. Examples of such covariates would include acne severity, socioeconomic status, smoking, comorbidities, and other potential differences between patients who are and are not prescribed or willing to use antibiotics over the long term.

The authors provide an insightful appraisal of their methodology and are careful to acknowledge that their study does not establish a causal association between antibiotic use and URTIs. The absence of a dose-response relationship when the analyses are stratified by route of antibiotic administration suggests a lack of causation. However, the observed association and its clinical importance might be strengthened if further exploratory analyses are conducted to characterize the findings in more detail.

First, to differentiate between bacterial and viral causes of URTIs, analyses assuming a bacterial cause could be performed in cases in which an antibiotic is prescribed to treat the infection. Second, the frequency distribution of URTIs per patient in each group would also help to ascertain the clinical significance of the results. For example, it would be less important clinically if all of the antibiotic users experienced only a single URTI over 12 months than if they each had multiple recurrent infections compared with nonusers. Third, it would be useful to establish more clearly whether a dose-response relationship exists between antibiotic use and URTIs, using both the duration of antibiotic usage and the route of administration as measures of cumulative exposure. Fourth, the incidence of URTIs in the 12 months preceding the study would help to demonstrate whether the 2 comparison groups were truly comparable at baseline, before the use of antibiotics for acne.

Furthermore, the rationale is unclear for using a 6-week threshold to define an antibiotic user, particularly given the fact that a 3-month threshold was used in a previous study of antibiotic use and oropharyngeal flora that was conducted by Levy et al.² The robustness of Margolis and colleagues’ results should thus be investigated with sensitivity analyses using thresholds of a longer duration.

Beyond the methodological considerations, what is the biologic plausibility of a 2-fold increase in the odds of URTIs developing as a result of the use of topical or oral antibiotics? The answer is not clear from current knowledge, but to shed light on plausibility there are several lines of evidence that might be considered. The first is antibiotic-induced bacterial resistance. Methicillin-resistant Staphylococcus aureus and vancomycin-resistant Enterococcus are well-known examples outside the realm of acne treatment, and resistant Propionibacterium acnes associated with the use of oral and especially topical antibiotics has been reported for years.^{3 - 5} Prior work by Levy et al² has demonstrated that Streptococcus pyogenes colonization and resistance in the oropharynx are associated with the use of oral and topical antibiotics in patients with acne. Nevertheless, bacterial resistance does not seem likely as an explanation for increased URTIs, especially since—as Margolis and coauthors note—“so few URTIs are attributable to bacterial causes.” Although the present study did not differentiate viral from bacterial URTIs, the lack of increased risk for urinary tract infections¹ in the same cohort is supportive of noncontribution from resistant bacteria.

Second, antibiotic-induced alteration of resident microflora could potentially influence the risk for URTIs. Antibiotic-induced diarrhea and oral or vulvovaginal candidiasis commonly occur with the use of tetracylines and erythromycin. Diarrhea due to Clostridium difficile, although usually associated with oral antibiotic therapy, has been reported after topical clindamycin therapy.^{6 - 7} Furthermore, otomycosis due to Candida or Aspergillus has been reported in association with the use of topical fluoroquinolones in pediatric patients.⁸

Despite the above-mentioned examples, it is uncertain whether antibiotic use in patients with acne causes an alteration of resident oropharyngeal microflora that could predispose to URTIs. However, Margolis et al¹ discuss the concept of colonization of one organism influencing the infectivity of another, including infections with viruses. There is a growing body of literature that supports the concept of polymicrobial infections being caused by synergism, additive effects, or other predisposing mechanisms.⁹

The third line of evidence concerns the immunomodulatory effects of antibiotics.^{10 - 11} Anti-inflammatory effects of tetracyclines and macrolides have been established in vitro^{12 - 13} and in mucosae in both human¹¹ and animal models.^{14 - 15} Tetracyclines have been shown to suppress neutrophil chemotaxis in vitro,¹² an effect that has also been demonstrated in patients with acne.¹⁶ Also, tetracyclines can suppress matrix metalloproteinases,^{15 ,17 - 18} up-regulate mitochondrial genes that protect against apoptosis,¹⁹ and induce neuroprotective anti-inflammatory effects.^{20 - 21} Immunomodulatory effects of macrolide antibiotics are thought to occur through several mechanisms, including the down-regulation of cell adhesion molecule expression,²² acceleration of apoptosis of activated lymphocytes,²³ and suppression of matrix metalloproteinases.²⁴ However, given the complexity of mucosal protection against infection, it remains unknown whether the immunomodulatory effects of these antibiotics could help explain the increased incidence of URTIs.

Finally, what are the implications of Margolis and colleagues’ study? For clinicians, the findings do not yet justify a change in current practice, and physicians still have the responsibility of explaining to patients that the observed association does not necessarily imply causation. However, the public response, whether justified or not, may influence how physicians treat acne. Margolis and coworkers should be congratulated on a well-designed study, albeit with stated limitations and the need for further analyses. We can all hope that their findings will stimulate further research to strengthen or dispel the observed association between URTIs and antibiotic use in patients with acne. Large prospective clinical studies will be required, ideally as part of a randomized trial. If an association is confirmed, additional research will also be needed to discover the precise mechanism.

Correspondence: Dr Shaw, Division of Dermatology, University of Toronto, 399 Bathurst St, Toronto, Ontario, Canada M5T2S8 (jc.shaw@utoronto.ca).

Financial Disclosure: None.