S. P. Olsen1, M. C. Perez2, A. M. Priddy1, E. S. Armbrecht1, A. K. Behera1, S. W. Fosko1, N. G. Zeitouni3, D. E. Winstead4, F. E. Johnson1 1Saint Louis University School Of Medicine,St. Louis, MO, USA 2Moffitt Cancer Center And Research Institute,Surgery,Tampa, FL, USA 3Roswell Park Cancer Institute,Dermatology,Buffalo, NY, USA 4Sarcoma Foundation Of America,Burlington, NC, USA
Introduction: Rosacea-affected tissue has increased angiogenesis/lymphangiogenesis. Isolated case reports have intimated that rosacea, and particularly rhinophyma (a late manifestation of rosacea), might cause angiosarcoma (AS) or lymphangiosarcoma (LAS) in rosacea-affected tissue. As in Stewart-Treves syndrome, AS/LAS feature prominent varicose veins, lymphatic obstruction, edema, and other features of chronic inflammation. No reports aimed at quantitatively establishing a causal relationship have been published, as far as we are aware. Recent evidence indicates that an aberrant innate immune response, mediated by altered endogenous polypeptides known as cathelicidins, is a primary pathologic event in rosacea. They are pro-inflammatory and vasoactive agents that cause the excessive inflammation in rosacea. Thus it is plausible that rosacea may be a risk factor for AS/LAS. We carried out an observational study to confirm or refute this possibility.
Methods: IRB approval was obtained. We developed a simple one-page data sheet to abstract data from pathology reports of patients with head and neck sarcomas of all sub-types. We excluded Kaposi sarcoma, Ewing sarcoma, carcinoma, and sarcoma not in the head and neck region. There were entries for other possible causes of sarcoma (prior radiation exposure, genetic predisposition, etc.). We obtained data from several large cancer centers and calculated the crude odds ratio in this classic retrospective case-control study.
Results: Data were analyzed in “waves” corresponding to the data sources. Wave 1 had 228 cases from 3 centers, of which 198 met inclusion criteria. The odds ratio for the association between rosacea and AS/LAS was 11.9 (95% CI: 1.05, 136.0). This is statistically significant. The confidence interval is wide. Wave 2 included 53 cases from one cancer center, of which all met the inclusion criteria. The odds ratio for the association between rosacea and AS/LAS was 0.56 (95% CI: 0.08, 3.72). This is not statistically significant. Combining wave 1 and wave 2, we had 251 evaluable cases. The odds ratio for the association between rosacea and AS/LAS was 4.76 (95% CI: 1.11, 20.50). This is statistically significant but the confidence interval is wide.
Conclusion: This is the first quantitative analysis designed to measure the association between rosacea and head and neck angiosarcoma/lymphangiosarcoma. A statistically significant association was observed but the confidence interval surrounding the odds ratio is wide. We seek access to other large data sets for additional analyses.
