Introduction

Rosacea is a chronic inflammatory skin disease that preferably affects the cheeks, nose, chin and forehead1. It mostly affects patients at an age of 30–50 years and is characterized frequently by erythema of affected facial regions, papules, pustules, telangiectasia and flushing2. It is defined by episodes of exacerbation and remission, often triggered by factors such as heat, stress, UV-light, smoking and alcohol2. Rosacea particularly affects fair-skinned people of Celtic origin or northern European descent, with an estimated prevalence of 5–10% in these populations3,4. Though such a common disease, the pathophysiology underlying rosacea remains unclear. While an infestation with Demodex species has been associated with rosacea, vascular and neuronal dysfunction have also been proposed as contributors to rosacea pathophysiology5,6,7. Additionally, the dysregulation of the innate and adaptive immune system has been shown to be relevant in rosacea pathogenesis2. These findings are especially interesting, as rosacea has been shown to be associated with chronic inflammatory comorbidities such as inflammatory bowel disease, coronary artery disease and autoimmune disorders8,9,10,11.

Carcinogenesis is often associated with an imbalanced inflammatory and immune response. As such, a possible relationship between cancer development and rosacea should be investigated. Skin cancer is the most common form of cancer and while malignant melanoma makes up only 4% of skin cancers, it is cause of 50% of skin cancer related deaths, making it the most aggressive and lethal form of skin cancer12,13. The most common cause for malignant melanoma lies in the cumulative exposure to UVA and UVB radiation14. UVA exposure has been shown to lead to oxidative-stress induced DNA damage15. UVB, on the other hand, induces the formation of photoproducts, cyclobutene pyrimidine dimers especially, and the accumulation of DNA mutations16,17. Additionally, it regulates the recruitment and activation of inflammatory cells such as macrophages and neutrophils into the skin, which is related to the malignant switch of melanocytes18,19. Due to the role of chronic inflammation and the immune system in rosacea pathophysiology, an association of melanoma and rosacea seems possible.

“Real-world” databases provide access to electronic medical records of large healthcare networks, enabling a population wide, unbiased study of diseases and risk factors. These databases have become increasingly important in clinical practice, as they allow for large-scale observational studies not otherwise feasible20,21. TriNetX is a global federated health research network that provides access to electronic medical health records of large healthcare organizations (HCOs) worldwide. Using this analytics platform to perform a single-centre study in the Mount Sinai Health System network, an American working group was able to identify an association of rosacea and diseases of the circulatory system, cerebrovascular diseases, hypertension and arterial disease22. The aim of this study was to use data from the real-world database TriNetX to investigate a possible correlation between rosacea and systemic diseases on a global level and to explore potential associations of rosacea and malignant melanoma.

Methods

Ethics approval

This study was reviewed and approved by Ethikkommission der Charité—Universitätsmedizin Berlin. This retrospective study is exempt from informed consent. The data reviewed is a secondary analysis of existing data, does not involve intervention or interaction with human subjects, and is de-identified per the de-identification standard defined in Section §164.514(a) of the HIPAA Privacy Rule. The process by which the data is de-identified is attested to through a formal determination by a qualified expert as defined in Section §164.514(b)1 of the HIPAA Privacy Rule. This formal determination by a qualified expert was refreshed in December 2020.

Cohort definition

The data used in this study was obtained on June 5th and July 24th, 2023, from the TriNetX platform, which provided access to electronic medical records (diagnoses, procedures, medications, laboratory values, genomic information) from approximately 21,913,235 patients from 74 HCOs. Cohorts were defined by an inpatient encounter during the last 20 years and the presence or absence of the ICD-10 code L71 [Rosacea] as the index event. No patients were excluded due to a surpassing of this timeframe for the risk analyses. A subset analysis was carried out on April 2nd, 2024 for rosacea patient with ICD-10 codes C43 [malignant melanoma of the skin] and C44 [other and unspecified malignant neoplasm of the skin] separately. 86% of the data is derived from US patients and ethnicity is routinely queried upon inclusion in the database. The composition of our study cohort is as follows: 82% Caucasian, 3% Black/African American, 3% other, 1.6% Asian, 0.3% Indian or Alaska Natives, 0.1% Hawaiian or Pacific Islanders, 10% unknown (non-US data). In Asian cohorts, in addition to the defining factor of the ICD-10 code L71 [Rosacea], we added demographic as a defining criterion. During our query, we received the data of around 493,154 Asian patients from 61 HCOs. Of these, none were excluded due to the surpassing of the previously mentioned timeframe of 20 years for risk analysis.

Statistics

We used TriNetX analytics tools for propensity score matching, compare outcome analysis and Kaplan–Meier survival analysis. After obtaining the baseline medial records including age, sex and diagnoses, we filtered for patients with and without rosacea (L71) using the ICD-10 code based system. To allow comparison between the two disparaging groups, we used propensity score matching for 1:1 matching of patients based on age and sex. After matching, we investigated comorbidities previously linked to rosacea (Table 1). We calculated the Risk Difference, Risk Ratio and Odds Ratio of these comorbidities in our cohorts with and without rosacea. Furthermore, we executed Kaplan–Meier analysis specifically for malignant melanoma. The primary outcome in this analysis was median survival, which was defined as the number of days when survival dropped below 50%. Additionally, we carried out Log-Rank and Proportionality testing and calculated the Hazard Ratio. Risk analysis included outcomes and comorbidities prior and after the index event, while Kaplan–Meier survival analysis excluded patients with an outcome prior to the time window starting one day after the index event and ending five years after the index event.

Table 1 Comorbidities analyzed in patients with or without rosacea.
Full size table

Results

We received the medical health record of 132,388 patients with a L71-diagnosis (+ rosacea) and 21,780,847 patients without a L71-diagnosis (-rosacea). Data inquiry, allocation and propensity score matching using the TriNetX database are shown in a consolidated standard of reporting trial (CONSORT) flow (inquiry on June 5th, 2023) (see Fig. 1).

Figure 1
figure 1

CONSORT flow diagram of data inquiry for patients with and without rosacea (L71). Major stages indicated in blue boxes. ICD-10: International Classification of Diseases 10, L71: rosacea.

Full size image

After propensity score matching, both cohorts consisted of 122,444 patients. Each cohort consisted of 84,752 female subjects (69.2% of study population) and 37,692 male subjects (30.8% of study population). Due to matching, age did not vary significantly in the cohorts, with a mean age of 55.5 at index event (see Table 2).

Table 2 Inquiry results and propensity score matching of patients with and without rosacea.
Full size table

Rosacea is associated with an increased risk of malignant melanoma and systemic diseases

After propensity score matching, we determined the risk of multiple systemic diseases in both our cohorts, that have previously been associated to rosacea8 (see Table 3). While the risk of being diagnosed with a vascular disease was at 0.185 in patients without rosacea, this risk increased to 0.336 in patients with rosacea [OR 2.234 (2.192, 2.276)]. The Risk Difference for heart disease lay at 0.108, with an OR of 1.649 (1.621, 1.677) for patients with rosacea. Similarly, the OR of type 2 diabetes was 1.618 (1.584, 1.652), with an increased risk in rosacea patients. The risk of metabolic diseases was severely increased in patients with rosacea, with 56,082 rosacea patients of the 122,444 cohort having been diagnosed with metabolic diseases, while only 25,801 patients without rosacea had diagnosed metabolic diseases [OR 3.165 (3.110, 3.222)]. Joint disease and ophthalmologic disease as well were more prevalent in our rosacea cohort [joint disease: OR 4.164 (4.083, 4.246), ophthalmologic disease: OR 4.801 (4.681, 4.924)]. Surprisingly, malignant melanoma and other skin neoplasms were the comorbidity most strongly associated with rosacea. Though 2192 patients without rosacea had been diagnosed with a malignant skin cancer, 12,128 of the 122,444 rosacea patients had concomitant skin cancer, making up approximately 10% of the study population [OR 6.031 (5.759, 6.316)]. In a subset analysis of rosacea patients with neoplasms of the skin, we were able to determine not only an increased risk of non-melanoma skin cancer [C44; OR 5.550 (5.345, 5.763)], but of malignant melanoma (C43) as well [OR 4.468 (4.144, 4.818)]. With the starkly increased risk for malignant melanoma in our rosacea population, we performed a Kaplan–Meier analysis of this subset of patients. 15,056 patients in cohort 1 (w/rosacea) and 1752 patients in cohort 2 w/o rosacea were excluded from this analysis, as they were diagnosed with malignant melanoma before their rosacea diagnosis. The survival probability at the end of the time window was 92.51% and 97.71% for the cohort with or without rosacea, respectively. At an HR of 3.286 (95% CI 3.101, 3.481), the mortality of malignant melanoma patients was higher if they also suffered from rosacea (p = 0.059).

Table 3 Summary of rosacea comorbidities in cohort 1 and cohort 2.
Full size table

Rosacea is not a risk factor for malignant melanoma in Asian populations

As previous studies in Asian populations have shown no association of rosacea and malignant melanoma23, we analyzed this sub-cohort separately. After adding the demographic parameter, we received the medical records of 470,834 patients. Of these, 1494 (0.31%) had received the diagnosis of rosacea, while 469,340 (99.68%) had not. Data inquiry, allocation and propensity score matching using the TriNetX database are shown in a consolidated standard of reporting trial (CONSORT) flow (inquiry on July 24th, 2023) (see Fig. 2).

Figure 2
figure 2

CONSORT flow diagram of data inquiry for patients with and without rosacea (L71) in Asian sub-cohort. Major stages indicated in blue boxes. ICD-10: International Classification of Diseases 10, L71: rosacea.

Full size image

After propensity score matching, both cohorts consisted of 1494 patients. The cohorts then contained 1109 (74.23%) female and 475 (31.79%) male patients with an average age of 45.9 years (see Table 4). In the Asian subpopulation, both cohorts had 10 events of malignant melanoma, leading to a risk of 0.007 of being diagnosed with malignant melanoma both with and without rosacea. Hence, there was no risk difference and an OR of 1 (0.415, 2.410) (see Table 5).

Table 4 Inquiry results of Asian sub-cohort with and w/o rosacea.
Full size table
Table 5 Kaplan–Meier survival analysis of Asian sub-cohort.
Full size table

Discussion

The aim of this study was to gain unbiased insight into a possible association of rosacea and systemic diseases on a global level using the rea-world database TriNetX. Here, we were able to show an association not only to metabolic disease, type 2 diabetes, joint problems and ophthalmologic disease, but were able to identify a strong association of rosacea and malignant melanoma as well. When viewed separately from other demographics, our Asian sub-cohort showed no association of rosacea and malignant melanoma, which is in accordance with previous data from Asian working groups23,24.

In their 2022 study, Cho et al. investigated the association between rosacea and different subtypes of skin cancer in a nationwide cohort study23. Using the data of 11,420 patients collected over nine years in Korea, they were able to show an increased risk of actinic keratosis and keratinocytic carcinoma in rosacea patients. While the Incidence Risk Ratio (IRR) for skin cancer in general was statistically significant in rosacea patients at 2.61 (95% CI 1.60–4.28), this was not the case when regarding malignant melanoma separately [IRR 1.19 (0.28–4.97)]. A possible explanation for this lack of association of malignant melanoma and rosacea in this Korean population could lie in the differing distribution pattern of malignant melanoma in Korean patients. While superficial spreading melanoma and nodular melanoma make up 70% and 20% of diagnosed melanoma cases in patients of European descent respectively25, acral melanoma has been shown to be the most commonly diagnosed form of melanoma for Korean, Taiwanese and Chinese patients26,27,28. Not only is this subtype of melanoma not characterized by UV-radiation induced mutations as commonly found in other subtypes, but is it commonly found in the lower regions of the body, nailbeds and palms, regions not commonly affected by rosacea29,30. Additionally, rosacea is often underdiagnosed in patients with non-white skin, as the discerning of efflorescences such erythema and teleangiectasias often proves to be difficult31. This highlights the need for a more differentiated regard of ethnicity as a factor in medical research.

In this study, we were able to identify a strong association of rosacea and malignant melanoma in Caucasian patients. Our findings stand in contrast to those gathered in Denmark by Egeberg et al.4. In their Danish nationwide cohort study of 49,475 rosacea patients and 4,312,213 subjects of the general Danish population, Egeberg et al. analyzed the risk for cancer in patients with rosacea. Here, they discovered a significant association between rosacea and non-melanoma skin cancer, though not for malignant melanoma. Additionally, rosacea was associated with a higher risk of basal cell carcinoma, but not malignant melanoma in the American Nurses’ Health Study II carried out from 1989 to 201132. In their 2023 systemic review on the association of rosacea and malignancies, Thapa et al.33 were able to determine an increased risk of non-melanoma skin cancer in rosacea patient, though no significant association with melanoma could be found. Taken together, the contrasts in our findings stress the limitations of observational studies and the need for further investigations into the association of rosacea and all subtypes of skin cancers.

Observational studies have identified a plethora of comorbidities linked to rosacea. Not only is there evidence suggesting increased incidence of inflammatory bowel disease and cardiovascular diseases, but also metabolic diseases such as dyslipidemia and diabetes8. In our study, we were able to further validate these association. Additionally, we were able to determine an increased risk of ophthalmologic disease including blepharitis and disease of the lacrimal duct. As up to 72% of rosacea patients are estimated to suffer from ocular symptoms and corneal ulcers due to ocular rosacea can deeply impact vision34, we hypothesize that the increased number of ophthalmologic diseases in our cohort could be a reflection of the high percentage of ocular rosacea. These findings underline the importance of interdisciplinary cooperation in the treatment rosacea.

Due to the retrospective nature of our study, certain limitations apply to its interpretation. As the TriNetX database allows filtering based on ICD codes provided in medical health records, this does not directly imply a medically confirmed diagnosis. This filtering disregards different subtypes and severities of rosacea as well as important prognostic factors in malignant melanoma, such as Breslow’s depth or melanoma subtypes. Moreover, there was no given information on lifestyle factors such as UV exposure, both a trigger for rosacea exacerbation and the development of skin cancers such as malignant melanoma2,16. Even though this study included a global patient population, the data requires a cautious interpretation, as it allows only for the detection of correlations, not causative relationships. While the database incorporates different ethnicities, it is dominated by a Caucasian population, a population commonly affected by rosacea. This limits the transferability of this data on a global level. Nevertheless, the novel association of rosacea and malignant melanoma warrants further investigation.

In the last decades, screenings for early detection of malignant melanoma have been controversially discussed. While there have been differing opinions on the socio-economic cost of skin cancer screenings and overtreatment of malignant melanoma, studies support the potential benefit of screening35. Not only is an early diagnosis of melanoma crucial for therapy as it is decisive for TNM staging and corresponding survival rates36, but the rising cost of skin cancer treatment is an economic burden that highlights the need for early detection and skin cancer prevention13. Additionally, UV radiation is not only a main contributor in the development of melanoma and non melanoma skin cancer, but also a known trigger and hypothesized to be crucial to rosacea pathogenesis37,38. Given the correlation of malignant melanoma and rosacea as well as the increased mortality of rosacea patients with concomitant melanoma in our study, the importance of UV avoidance or protection should not be underrepresented in consultation of rosacea patients.

In conclusion, we were able to determine an increased risk of joint problems, metabolic disease, visual disturbances, type 2 diabetes and malignant melanoma in rosacea patients. While rosacea has been previously linked to an increased risk of systemic diseases, our findings regarding malignant melanoma contrast with the findings of past studies. These differences highlight the need for further investigation of the possible connection of these two dermatologic diseases.