However, the Expert Committee still did not define either the length of time of flushing or nontransient erythema. It also did not specify convex surfaces of the face with periocular sparing as the characteristic pattern or provide additional information on how photoaging fits into the definition. The updated classification stated that centrofacial erythema must not be from cutaneous lupus or seborrheic eczema, and steroid-induced rosacea was still excluded.12 However, there is still the need to exclude other systemic conditions, such as mastocytosis, carcinoid, polycythemia vera, and dermatomyositis. Therefore, the potential for subjective error and inclusion of nonrosacea diseases persists.
A critical change was elimination of the granulomatous rosacea variant. In 2002, this variant was defined by monomorphic, yellow-brown to red papules and nodules that led to scarring. This variant, however, did not share the commonalities of the other subtypes, including persistent facial erythema, limitation to convex surfaces, periocular sparing, and transient flushing.3,13 At the time, Crawford et al6 proposed that the variant be recategorized as granulomatous facial dermatitis. In the updated NRS classification, this variant and phenotypic description was eliminated from the schema.12 It is unclear if it was removed because of these discrepancies or if the NRS panel felt it had a distinct pathogenesis from the proposed rosacea pathophysiology; however, we applaud this change.
Both the ROSCO and NRS classification schemes mention progression between the various phenotypes,10,12 suggesting that rosacea phenotypes exist along a continuum, progressing and regressing with disease severity. The main study addressing this point was based on the self-reported retrospective patient memory of disease features in rosacea patients. The authors used a modified criterion of centrofacial erythema alone to define ETR; therefore, a person who began their disease with this finding but then acquired inflammatory lesions or phymas was defined as progressing along a spectrum.14 Given that persistent erythema of convex surfaces of the face is common in all subtypes, we do not find it surprising that the authors found (using their modified criteria) that ETR appeared to progress to papulopustular and phymatous subtypes in a small number of patients. We strongly disagree with their interpretation and conclusion.
In our experience, ETR patients have fine textured skin without sebaceous quality or a history of extensive acne (Figure 1). Flushing is common and usually lasts 10 minutes to 1 hour. There might be concurrent burning or stinging; however, there is no associated sweating, lightheadedness, palpitations, or diagnostic laboratory findings, which distinguishes ETR from other common causes of flushing. The persistent centrofacial erythema involves convex surfaces, spares periocular skin, and can be best defined as present for longer than 3 months.
In contrast, phymas occur commonly in patients with thick and sebaceous (glandular) skin (Figure 2).6,15-17 Men are most often affected and usually have a history of moderate to severe acne. It is not uncommon to observe nodules, cysts, and scarring in addition to papules and pustules. These eruptions primarily cluster on the central face and present in areas of nontransient erythema. Flushing, although less prominent than in other phenotypes, also can be seen.
Taken together, we find no convincing evidence from published studies or extensive experience caring for rosacea patients that classic ETR progresses to phymatous rosacea, or the other way around, as displayed in the ROSCO panel report.8 The type of skin seen in Figure 1 will not “progress” to the type seen in Figure 2. Furthermore, treatment will not “reverse” the phymatous skin into thin, ETR-type skin. The implications are important: If a female patient is given a diagnosis of ETR, she will not develop an enlarged phymatous nose. Patients with thick sebaceous skin, as in Figure 2, usually tolerate treatments such as benzoyl peroxide that other rosacea patients do not and frequently respond well to such intervention.
Implications and Future Directions
We present an overview of 2 rosacea classification systems, hoping to stimulate further refinement. Looking forward, there are many directions for further investigation into the pathophysiology of rosacea. From a genetic standpoint, there needs to be continued molecular and epidemiologic data to determine the underlying genetic contributions to disease.
There has been some progress in the realm of understanding the mechanisms of inflammation; we urge further investigation to elucidate how “subclinical neuroinflammation” might lead to glandular hyperplasia.12 We also see value in examining the genetic and hormonal contributions to phymas, as they may be different than those seen in the ETR-type patients. Last, more studies focusing on comorbidities that contribute to or arise from rosacea are welcomed.
The ultimate goal is to develop a classification system that integrates clinical descriptions, pathophysiologic mechanisms, and benchmark indicators of disease. Only then can we have a true gold standard for the diagnosis of rosacea, one that allows for improved personalized treatment and better outcomes.