A Predictive Model for Differential Diagnosis Between Rosacea and Sensitive Skin: A Cross-Sectional Study
Rosacea and sensitive skin (SS) are two common dermatological conditions that share overlapping clinical features, particularly facial erythema, making their differential diagnosis challenging. However, the pathogenesis, treatment, and prognosis of these conditions differ significantly, necessitating reliable non-invasive methods to distinguish between them. This study explores the use of clinical information combined with advanced imaging technology to develop a predictive model for differentiating rosacea from SS.
Background and Significance
Rosacea is a chronic inflammatory skin disease primarily affecting the central facial region. Its hallmark manifestations include flushing, persistent erythema, papules, pustules, and phymatous changes. Among these, erythema is the most prevalent feature across all subtypes of rosacea. Sensitive skin, on the other hand, is characterized by unpleasant sensations such as itching, burning, pain, and tingling in response to physical, thermal, or chemical stimuli that typically do not affect healthy skin. Notably, erythema is also reported in 74% of SS patients, further complicating the distinction between the two conditions.
Given the differences in their underlying mechanisms and management strategies, accurate differentiation is crucial. Traditional diagnostic methods, such as histopathological examination, reflectance confocal microscopy, and dermoscopy, focus on visualizing microstructures but are limited by sampling variability. In contrast, color imaging and analysis systems, such as the VISIA® complexion analysis system, offer a non-invasive approach to mapping facial skin conditions with high resolution. This study leverages the Red/Brown/X (RBX) algorithm of the VISIA® system to analyze red area distribution patterns and their diagnostic utility.
Study Design and Methodology
This cross-sectional study enrolled 275 patients (201 with rosacea and 74 with SS) with Fitzpatrick skin types III to IV from the Dermatology Department of the First Affiliated Hospital of Nanjing Medical University between April 2019 and March 2020. The diagnosis of rosacea was based on the 2017 update by the National Rosacea Society Expert Committee, while SS was diagnosed using a self-perception sensitive skin questionnaire and a positive lactic acid sting test, with the exclusion of other inflammatory skin diseases. All patients provided written informed consent, and the study was approved by the Institutional Research Committee.
Clinical characteristics and medical histories were recorded for all participants. Facial images were captured using the VISIA® 6.0 system, and red area images were analyzed using the RBX algorithm. The distribution of red areas was classified into four specific patterns: peace sign, wing shape, earlobe, and dots/globular patterns. Statistical analyses were performed using STATA 14.0, including univariate and multivariate regression analyses, to identify factors associated with the diagnosis of rosacea. Decision curve analysis (DCA) was used to evaluate the predictive performance of the models.
Key Findings
Clinical Characteristics and Medical History
Significant differences were observed in the clinical characteristics and medical histories of rosacea and SS patients. The majority of rosacea patients (64.2%) had a disease course exceeding 20 months, compared to 60.8% of SS patients with a course under 20 months. A history of inappropriate skincare was reported by 83.8% of SS patients, while only 41.3% of rosacea patients reported such a history. Facial erythema was present in all rosacea patients and 95.9% of SS patients.
Red Area Distribution Patterns
The analysis of red area distribution patterns revealed distinct differences between the two conditions. The peace sign pattern was observed in 54.7% of rosacea patients but only 14.9% of SS patients. The wing shape pattern was exclusive to rosacea, appearing in 19.9% of cases. The earlobe pattern was present in 50.7% of rosacea patients and 12.2% of SS patients. The dots/globular pattern, corresponding to papules, pustules, and nodules, was more common in rosacea but not observed in SS patients.
Predictive Factors for Rosacea
Univariate regression analysis identified the course of the disease at the time of the visit, inappropriate skincare, and the presence of peace sign and earlobe patterns as significant predictors of rosacea. Multivariate regression analysis confirmed these factors as independently associated with rosacea diagnosis. The peace sign pattern had the highest positive likelihood ratio (4.547), followed by the earlobe pattern (4.198). Inappropriate skincare had a negative likelihood ratio of 3.143 for rosacea.
Model Performance
Two predictive models were developed. Model 1 incorporated the peace sign and earlobe patterns, along with the course of the disease and inappropriate skincare, achieving an area under the curve (AUC) of 0.861 (95% CI: 0.818–0.904). Model 2, which included only the peace sign and earlobe patterns, had an AUC of 0.788 (95% CI: 0.740–0.835). Decision curve analysis demonstrated that Model 1 outperformed Model 2 across threshold probabilities of 20%–80%.
Discussion
This study highlights the utility of combining clinical variables with advanced imaging technology for the differential diagnosis of rosacea and SS. The peace sign and earlobe patterns of erythema distribution, detected using the RBX algorithm, were found to be highly predictive of rosacea. These patterns likely reflect underlying vascular abnormalities, such as the dilation of specific facial arteries, which are characteristic of rosacea but not SS.
The peace sign pattern, characterized by erythema on the forehead, nose, chin, and cheeks, may result from abnormal dilation of the arteria supratrochlearis, arteria angularis, and arteria infraorbitali. This pattern was predominant in rosacea patients but also appeared in a subset of SS patients, potentially due to the overlap between SS and rosacea-prone skin types. The wing shape pattern, exclusive to rosacea, may be associated with the zygomatic-facial and facial arteries. The earlobe pattern, observed in half of the rosacea patients, is likely due to inadequate venous drainage in this anatomically suspended region.
The study also underscores the importance of clinical history in the diagnostic process. The duration of the disease and a history of inappropriate skincare were significant predictors, with the latter being more common in SS patients. These findings align with the understanding that SS is often triggered or exacerbated by external factors, whereas rosacea has a more intrinsic pathophysiology.
Limitations and Future Directions
While this study provides valuable insights, it has certain limitations. The sample size, particularly for SS, was relatively small, and the clinical subtypes of rosacea were not stratified, which may have influenced the results. Future studies with larger cohorts and subtype-specific analyses are needed to validate these findings and further refine the predictive model.
Additionally, the study focused on Fitzpatrick skin types III to IV, limiting the generalizability of the results to other skin types. Expanding the research to include a broader range of skin types would enhance the applicability of the model. Further exploration of the underlying vascular and inflammatory mechanisms driving the observed patterns could also provide deeper insights into the pathophysiology of rosacea and SS.
Conclusion
This study demonstrates that the peace sign and earlobe patterns of facial erythema, detected using the RBX technology of the VISIA® system, are effective predictors for distinguishing rosacea from SS. When combined with clinical factors such as the course of the disease and a history of inappropriate skincare, these patterns significantly enhance diagnostic accuracy. This non-invasive approach offers a valuable tool for clinicians in the differential diagnosis of these two conditions, ultimately improving patient care and outcomes.
doi.org/10.1097/CM9.0000000000001001
Was this helpful?
0 / 0