Granuloma Annulare: A Retrospective Series of 133 Patients
Although granuloma annulare (GA) is a relatively common dermatologic condition, little is known about its epidemiology and pathogenesis. Additionally, there is conflicting evidence regarding its association with other diseases. In our retrospective study, we attempted to determine the number of cases of GA seen annually at a large academic center and clarify disease associations with GA. We utilized PennSeek, a tool developed by the Penn Medicine Data Analytics Center to search electronic medical records (EMRs), to identify all cases of GA seen at the University of Pennsylvania (Philadelphia, Pennsylvania) from 2008 to 2014. All biopsy-proven cases of GA from 2010 to 2014 were reviewed for medical comorbidities, medications, treatments, and outcomes. On average, the percentage of patients given a diagnosis of GA annually was 0.22%. There were 133 cases of biopsy-proven GA that were reviewed. The findings solidify our understanding of the epidemiology of GA and diseases that can be associated with GA.
Practice Points
- Although the pathogenesis of granuloma annulare (GA) is unknown, associations between the disorder and underlying systemic processes (eg, diabetes mellitus, hyperlipidemia, thyroid disease, human immunodeficiency virus) have been proposed.
- This study elicited a period prevalence of GA of 0.22% to 0.27%.
- The most commonly used treatments of GA were topical steroids and intralesional triamcinolone, followed by hydroxychloroquine.
Concomitant Disorders and Unrelated Medical Therapy
Hypothyroidism is statistically significantly overrepresented in our patient population (30.1%) compared with an average prevalence of 1% to 2% in iodine-replete populations (Fisher exact test, P<.001).10 This finding is consistent with prior small studies and cases series, which have suggested an association between autoimmune thyroiditis and GA.11-14
Despite prior reports of a possible association between HIV and GA,15-24 none of our patients had a diagnosis of HIV. However, many of our patients were not tested for HIV, which confounds our results and may represent a practice gap in the field.
At 1.5%, the prevalence of type 1 diabetes mellitus in our patients is slightly higher than the national average of 0.3%.25 However, based on a Fisher exact test of analysis of proportions, this difference is not statistically significant (P=.106).
At 1.5% and 2.3%, the prevalence of hepatitis B and hepatitis C, respectively, in our patients is slightly higher than the national average of 0.5% and 1%, respectively.26 However, based on a Fisher exact test of analysis of proportions, these differences are not statistically significant (P=.142 and P=.146, respectively).
,Given the high prevalence of hyperlipidemia in the United States (31.7%), this disease is not overrepresented in our sample (30.1%), though others have suggested there may be a connection.27,28 Based on a Fisher exact test, this difference of proportions is not statistically significant (P=.780).
Selective serotonin reuptake inhibitor use is common in the United States; approximately 11% of Americans older than 12 years use an SSRI.29 At 18.1%, the use of SSRIs in our patient group was statistically significantly higher than the national average (Fisher exact test, P=.017), suggesting a possible association between SSRI use and development of GA, warranting further investigation.
The use of calcium channel blockers, interferon, and tumor necrosis factor inhibitors was not significantly associated with GA in our series.
GA Therapy
The most commonly used treatments for GA in our study were topical steroids and intralesional triamcinolone, followed by hydroxychloroquine; all treatments employed exhibited a widely variable response. Assessing treatment response via retrospective chart review is challenging and response rates may not be accurately captured.
Study Limitations
Our study had several limitations. In calculating the period prevalence of GA, our query was limited by the number of years that the EMR has been in place. The number of cases we reviewed for clinical characteristics was limited to 133, as many cases with the ICD-9 diagnosis of GA were not biopsy proven and therefore were not included in our review. Many of the cases we reviewed were lost to follow-up, which prevented us from determining treatment outcomes.
Another weakness of our study was that our query did not provide an estimate of incidence or prevalence of GA overall, as this analysis was not a population-based study. The power of our study was limited by the number of cases of GA seen annually and the number of patients lost to follow-up. Additionally, our study population may only be generalizable to other large academic centers.
Conclusion
This study further solidifies our understanding of the epidemiology of GA and diseases that can be associated with GA. We identified a higher female to male ratio than previous reports, and consistent with prior reports, we noted potential associations with conditions such as thyroid disease and hyperlipidemia. Our population demonstrated higher rates of SSRI use than expected, warranting further investigation. Dermatologists should be aware of potential disease associations with GA, but as a whole we need better data and larger studies to determine the appropriate evaluation and treatment for patients with GA.
