|Year : 2020 | Volume
| Issue : 1 | Page : 15-21
Is rosacea a risk factor for cancer: A population-based cohort study in Taiwan
Tsung-Hsien Chang1, Hsiu J Ho2, Yun-Ting Chang1, Chung-Pin Li3, Chun-Ying Wu4, Chen-Yi Wu5
1 Department of Dermatology, Taipei Veterans General Hospital; Department of Dermatology, National Yang-Ming University, Taipei, Taiwan
2 Division of Translational Research, Taipei Veterans General Hospital, Taipei, Taiwan
3 Department of Medicine, Division of Gastroenterology and Hepatology, Taipei Veterans General Hospital; National Yang-Ming University School of Medicine, Taipei, Taiwan
4 Division of Translational Research, Taipei Veterans General Hospital; National Yang-Ming University School of Medicine, Taipei; Graduate Institute of Clinical Medicine, China Medical University, Taichung; Department of Public Health, Institute of Public Health, National Yang-Ming University, Taipei, Taiwan
5 Department of Dermatology, Taipei Veterans General Hospital; Department of Dermatology, National Yang-Ming University; Department of Public Health, Institute of Public Health, National Yang-Ming University, Taipei, Taiwan
|Date of Submission||30-May-2019|
|Date of Decision||13-Jul-2019|
|Date of Acceptance||02-Aug-2019|
|Date of Web Publication||27-Feb-2020|
Department of Dermatology, Taipei Veterans General Hospital, No. 201, Section 2, Shih-Pai Road, Taipei
Source of Support: None, Conflict of Interest: None
Background: Rosacea is a chronic inflammatory skin disease with mounting evidence associating it with systemic disorders. Cancer, induced or facilitated by chronic inflammatory microenvironment, shares common pathogenic mechanisms with rosacea. Objectives: We performed a population-based cohort study to investigate the risk of developing cancer among people with rosacea in Taiwan. Methods: A total of 65,526 patients with rosacea and 262,104 age-, sex-, and comorbidity-matched controls were identified from the Taiwan's National Health Insurance Research Database between 1997 and 2013. All participants were followed up for 2–12 years. Incidence rates (IRs) of overall and specific types of cancer were calculated. Cumulative incidences of cancer were compared between the two cohorts by Kaplan–Meier method and modified log-rank test. Hazard ratios (HRs) adjusted for age, sex, and comorbidities for overall and specific malignancies were estimated using subdistribution proportional hazard models. Results: The IR (per 1000 person-years) of cancer was 2.83 in patients with rosacea and 3.00 in controls. There was no difference in cumulative incidence of cancer between patients with or without rosacea (P = 0.109). The risk of developing cancer did not increase among patients with rosacea (HR = 1.04; 95% confidence interval = 0.98–1.11). In addition, patients with rosacea did not have a significantly increased risk of developing any specific type of cancer. Conclusion: We found no association between rosacea and malignancy. These results did not agree with those reported in previous studies. Further research should be conducted to clarify the association between rosacea and cancer, especially focusing on the pathophysiology.
Keywords: Cohort studies, comorbidity, epidemiology, neoplasms, rosacea, skin diseases
|How to cite this article:|
Chang TH, Ho HJ, Chang YT, Li CP, Wu CY, Wu CY. Is rosacea a risk factor for cancer: A population-based cohort study in Taiwan. Dermatol Sin 2020;38:15-21
|How to cite this URL:|
Chang TH, Ho HJ, Chang YT, Li CP, Wu CY, Wu CY. Is rosacea a risk factor for cancer: A population-based cohort study in Taiwan. Dermatol Sin [serial online] 2020 [cited 2020 May 28];38:15-21. Available from: http://www.dermsinica.org/text.asp?2020/38/1/15/279601
| Introduction|| |
Rosacea is a chronic inflammatory skin disease characterized by transient or persistent erythema, edema, telangiectasia, and papulopustules or phymatous change in the central face. Women are more frequently affected than men, and most of the patients develop their symptoms in their middle age. Rosacea is more prevalent among people with light skin colors (Fitzpatrick skin phototypes I and II), with the reported prevalence in western countries ranging from 1.3% to 22%.,,, People of color, on the contrary, appear to have a lower risk of developing rosacea, whether it is because of genetic difference, underdiagnosis due to a masking effect of dark skin, or protective effect of melanin from actinic damage, which is a well-known exacerbating factor of rosacea.,,
The pathophysiology of rosacea remains elusive. Several mechanisms have been proposed, including anomalous innate and adaptive immune response, vascular hyperreactivity, neuronal dysregulation, and micropathogens such as Demodex folliculorum or Helicobacter pylori., Previously believed to be a skin-limited condition, rosacea has been recently found to be associated with various systemic disorders such as cardiovascular, gastrointestinal, metabolic, autoimmune, neurologic, and psychiatric diseases., Common etiopathogenesis has been proposed to explain the observed association. Cancer, induced or facilitated by chronic inflammatory microenvironment, also shares common pathogenic mechanisms with rosacea., However, studies investigating the association between rosacea and cancer were relatively sparse.,,, Although certain types of cancer, including cutaneous and internal malignancies, were found to be associated with rosacea in these studies, no solid conclusion could be drawn due to differences in study design, size of study samples, methods of identifying rosacea cases, and controlled covariate factors. In addition, none of these studies focused on people of color. The incidence of rosacea and various types of cancer in people of color is different from those in fair-skinned people. Therefore, we performed a population-based cohort study to examine the risk of cancer among people with rosacea in Taiwan.
| Materials and Methods|| |
Data analyzed in this study were retrieved from the Taiwan's National Health Insurance Research Database (NHIRD), which enrolls >99% of Taiwanese civilians (approximately 23 million people) and documents registration information and original claims data from 93% of medical care institutions in Taiwan. The reliability of NHIRD has been demonstrated in several validation studies, and it has been widely utilized in epidemiological studies in Taiwan.,,,
Scrambled information of patients' age, sex, date of inpatient and outpatient visits, diagnosis based on the Ninth Edition of the International Classification of Disease (ICD-9) coding, and medical specialty of the health-care providers are recorded in the NHIRD, while personal information, such as body mass index, socioeconomic status, lifestyle, family history, and habit of drinking or smoking, are unavailable in the database. The study protocol was approved by the Institutional Review Board of Taipei Veterans General Hospital (IRB number: 2017-08-005CC), who granted a waiver of informed consent for retrospective studies utilizing data in the NHIRD.
We identified all patients with diagnoses of rosacea (ICD-9-CM code 695.3) between 1997 and 2013 from NHIRD. To enhance the diagnostic validity, only patients with at least three outpatient diagnoses of rosacea from a dermatologist or at least one inpatient diagnosis of rosacea from any medical specialty were enrolled. The date of the first diagnosis of rosacea was used as an index date of entering the rosacea cohort. In addition, as lesions of cutaneous lupus erythematosus (ICD-9-CM code 695.4), seborrheic dermatitis (ICD-9-CM code 690.1), and acne vulgaris (ICD-9-CM code 706.1) are easily confused with those of rosacea, we further excluded individuals with any two of these diseases between 1997 and 2013.
A control cohort without diagnosis of rosacea was established from the Longitudinal National Health Insurance Research Database (LHID) 2000. LHID 2000 is a subset of NHIRD that comprises the entire original claims data of one million representative patients randomly sampled from NHIRD in 2000. To choose more compatible controls, patients without medical visit in any 1 year of time during the study period were excluded. Individuals in both cohorts with conflicting sex, unknown birthday, and diagnosis of cancer before the index date were not included.
Every patient with rosacea was then matched with controls at a ratio of 1:4 by age, sex, and propensity score on the index enrollment day. The propensity score was calculated by comorbidities including hypertension (ICD-9-CM codes 401–405), diabetes mellitus (ICD-9-CM code 250), hyperlipidemia (ICD-9-CM codes 272.0–272.2), peptic ulcer (ICD-9-CM codes 531–533), chronic obstructive pulmonary disease (ICD-9-CM codes 491, 492, and 496), cerebrovascular disease (ICD-9-CM codes 430–431 and 433–436), coronary artery disease (ICD-9-CM codes 410–414), and chronic kidney disease (ICD-9-CM codes 584–586). All enrolled individuals were followed up for at most 12 years from the index date until occurrence of cancer or death or until December 31, 2013. Patients who were followed up for <2 years were excluded from the analysis.
The primary outcome was the first diagnosis of any type of cancer (ICD-9-CM codes 140–208). We identified patients diagnosed with cancer from the Registry for Catastrophic Illness Patients Database, a subset of NHIRD. Patients with all types of cancer can be registered in this database and granted exemption from all medical copayments. Stringent criteria were set up for registration, which necessitated cautious review of patients' medical history, laboratory data, and imaging discovery by at least two medical specialists. Confirmation of diagnosis of cancer further needs cytological or pathological evidence. Therefore, diagnosis of cancer in this study is highly reliable.
To further examine the association of rosacea with different types of cancer, the occurrence of specific types of cancer was identified, including cancer of the oral cavity and pharynx (ICD-9-CM codes 140–149), esophageal cancer (ICD-9-CM code 150), gastric cancer (ICD-9-CM code 151), colorectal cancer (ICD-9-CM codes 153, 154.0, and 154.1), hepatic cancer (ICD-9-CM code 155), pancreatic cancer (ICD-9-CM code 157), lung cancer (ICD-9-CM code 162), melanoma (ICD-9-CM code 172), nonmelanoma skin cancer (NMSC) (ICD-9-CM code 173), female breast cancer (ICD-9-CM code 174), cervical cancer (ICD-9-CM code 180), endometrial cancer (ICD-9-CM codes 179 and 182), ovarian cancer (ICD-9-CM code 183), prostate cancer (ICD-9-CM code 185), kidney cancer (ICD-9-CM code 189.0), bladder cancer (ICD-9-CM code 188), brain cancer (ICD-9-CM code 191), thyroid cancer (ICD-9-CM code 193), and hematologic cancer (ICD-9-CM codes 200–208). These specific types of cancer were chosen before data analysis. When the association between rosacea and certain type of cancer was analyzed, individuals in both cohorts were followed up until occurrence of that certain type of cancer or death or until December 31, 2013.
Demographic characteristics and selected comorbidities were expressed as means and standard deviations (SDs) for continuous variables and frequencies and percentages for categorical variables. Two-sample t-test and Pearson's Chi-square test were used for analyzing between-group differences for continuous variables and categorical variables, respectively. The cumulative incidences of cancer between the rosacea and control cohorts were compared using the Kaplan–Meier method and modified log-rank test proposed by Gray. Death before the occurrence of cancer was regarded as the competing risk event. Incidence rates (IRs) of cancer in both the groups were calculated by dividing the number of study outcomes by total person-time at risk, presented by unit of case numbers per 1000 person-years (PYs). The effects of rosacea on all and specific types of cancer were measured by calculating the hazard ratios (HRs) and 95% confidence interval (CI) using subdistribution proportional hazard models. Covariates adjusted in the multivariable hazard regression analysis included age, sex, hypertension, diabetes, hyperlipidemia, coronary artery disease, cerebrovascular disease, chronic obstructive pulmonary disease, peptic ulcer, and chronic kidney disease. Age-stratified subdistribution proportional hazard models were applied to evaluate the effect of rosacea on development of cancer in patients with different ages (age <40 years, age between 40 and 59 years, and age ≥60 years). A sensitivity analysis not excluding patients with any two diagnoses of cutaneous lupus erythematosus, seborrheic dermatitis, and acne vulgaris was performed.
SAS version 9.4 (SAS Institute, Cary, NC, USA) was used to manage study data, whereas R version 3.3.3 was utilized for testing and modeling. A two-sided P < 0.05 was considered statistically significant.
| Results|| |
A total of 65,526 patients with rosacea and 262,104 age-, sex-, and comorbidity-matched controls were identified. The distribution of background characteristics and selected comorbidities is shown in [Table 1]. There were no differences in age, sex, and comorbidities between the two groups after matching. Approximately 68% of patients in both the groups were female and 32% were male. Most individuals were aged between 20 and 60 years, with a mean age of 37 years (SD = 16 years). The mean follow-up period of both the groups was around 7 years.
|Table 1: Demographic characteristics and comorbidities of patients in both the study groups|
Click here to view
[Figure 1] shows the Kaplan–Meier curves of the cumulative incidence of cancer in patients with or without rosacea. Modified log-rank test revealed no significant difference in cumulative incidence between the two groups (P = 0.109). [Table 2] shows the IRs of cancer in patients with or without rosacea. The IR of cancer was slightly lower in patients with rosacea (2.83 per 1,000 PYs) than in those without rosacea (3.00 per 1000 PYs). [Table 3] shows the HRs of patients with rosacea relative to the controls. The overall cancer risk (HR = 1.04; 95% CI = 0.98–1.11) did not significantly increase in patients with rosacea nor did the risk for any specific type of cancer. Age-stratified analyses revealed increased risks for lung cancer (HR = 1.32; 95% CI = 1.01–1.74) and melanoma (HR = 4.62; 95% CI = 1.15–18.48) in rosacea patients with the age between 40 and 59 years. The sensitivity analysis not excluding patients with multiple diagnoses of cutaneous lupus erythematosus, seborrheic dermatitis, and acne vulgaris showed no significant increase in the overall cancer risk (HR = 1.01; 95% CI = 0.95–1.08) and any specific cancer risk [Supplementary Table 1].
|Figure 1: Kaplan–Meier curves. Cumulative incidence of overall cancer in patients with or without rosacea. The difference between the two study groups was calculated using modified log-rank test|
Click here to view
|Table 3: Hazard ratios of cancer in patients with rosacea relative to the controls|
Click here to view
| Discussion|| |
In this nationwide cohort study, we examined the risk of cancer in patients with rosacea. We did not detect an increased risk for overall or any specific type of cancer in patients with rosacea after adjustment of potential confounders, including sex, age, and comorbidities. The null association between rosacea and cancer remained unchanged in the sensitivity analysis.
To date, there have been only a few studies exploring the association between rosacea and cancer. With respect to skin malignancy, a previous study conducted by Dupont with a small sample size revealed no association between rosacea and skin cancer, while recent large-scale studies yielded positive finding – the questionnaire-based American study conducted by Li identified an increased risk of basal cell carcinoma (HR = 1.50; 95% CI = 1.35–1.67), and the Danish nationwide cohort study performed by Egeberg et al. detected an increased risk of NMSC (HR = 1.36; 95% CI = 1.26–1.47). In our study, the risk of developing NMSC did not increase in patients with rosacea across all age groups. Increased susceptibility to actinic damage due to decreased levels of filaggrin-deduced photoprotectants after prolonged skin inflammation and the possibility that cases with rosacea tended to have more accumulative amounts of ultraviolet (UV) exposure in their early age were proposed to explain the observed association between rosacea and NMSC.,, With more abundant photoprotective eumelanin in the epidermis, Asians are inherently more unsusceptible to solar damage and thus more difficult to develop skin cancer, which can be further supported by the observation that the IR of NMSC in the control group of our study was 0.05 per 1,000 PYs, whereas that of the Danish study was 2.86/1000 PYs. Therefore, the null association between rosacea and NMSC found in our study could be explained both physiologically and statistically, i.e., a higher resistance to UV-induced carcinogenesis in people of color and a lack of power to detect statistically positive association due to rareness of NMSC in both the groups. Regarding the risk for melanoma, both Li et al. and Egeberg et al. found null association between rosacea and melanoma., In our study, although an increased risk for melanoma (HR = 4.62; 95% CI = 1.15–18.48) was observed in rosacea patients with the age between 40 and 59 years, there were only four incident cases in the rosacea cohort and four incident cases in the controls. Due to the small number of incident cases, the positive association between rosacea and melanoma might not be determined in our study.
With respect to internal malignancies, the American study showed a significant association between thyroid cancer and rosacea (HR = 1.59; 95 CI%=1.07–2.36), whereas the Danish study found hepatoma (HR = 1.42; 95 CI%=1.06–1.90) and breast cancer (HR = 1.25; 95 CI%=1.15–1.36) to be significantly related with rosacea. Aberrant inflammation was regarded as the shared pathogenesis underlying rosacea and these malignancies. In our study, however, the risks of developing the above-mentioned internal malignancies were not increased. Although there has been evidence linking chronic inflammation to thyroid cancer, hepatoma, and breast cancer,,, it may be arbitrary to ascribe the association between rosacea and these malignancies to inflammation in the absence of studies specifying common inflammatory markers. After all, too many conditions are involved with inflammation, which make it a vague term without further elaboration. Moreover, certain types of cancer with more compelling evidence of their association with rosacea-related chronic inflammation have not been found to be associated with rosacea per se. For instance, inflammatory bowel disease is a well-recognized risk factor of developing colorectal cancer, and its incidence was found to be increased among patients with rosacea., Nonetheless, no association was detected between rosacea and colorectal cancer. Another example is gastric cancer, which can be induced by long-term infection of H. pylori. This microorganism was believed to play a role in the development of the central symptoms of rosacea., However, no association between gastric cancer and rosacea was discovered. To interpret the association between rosacea and the above-mentioned malignancies by inflammation, more direct and clear evidence should be disclosed.
Another Danish nationwide study conducted by Egeberg et al. in an earlier period showed an increased incidence of brain tumor among patients with rosacea (IR ratio = 1.36; 95% CI = 1.18–1.58). Unlike the association between rosacea and the above-mentioned malignancies where no specific inflammatory pathways were identified, the enhanced expression of matrix metalloproteinase has been demonstrated to play a key role in the pathogenesis of both rosacea and glioma,,, which comprises >80% of the primary brain malignancy. Our failure to identify the positive association between rosacea and brain tumor could be explained by the smaller proportion of glioma among primary brain malignancies in Taiwan, as well as the different genetic pathways to glioma in Asians and Whites,,, which may influence the potential pathogenetic association between rosacea and glioma.
Egeberg et al. also attributed part of the association between cancer and rosacea to unadjusted personal factors. Alcohol has been proposed to be a risk factor of rosacea, as well as hepatoma and breast cancer.,, A falsely positive association between rosacea and these malignancies might be detected if drinking behavior was unadjusted. In our study, alcohol was not controlled either, but the positive association between rosacea and these malignancies did not exist. It could be partially explained by well-controlled metabolic comorbidities such as hypertension, diabetes, and hypercholesterolemia in our study. These metabolic disorders are more prevalent in individuals with drinking behavior, and adjusting these factors may hence correct the potentially imbalanced distribution of habitual drinkers in rosacea and control cohorts.
On the contrary, patients with rosacea were found to have a decreased risk of developing lung cancer as reported in the Danish study (HR = 0.78; 95% CI = 0.69–0.89). The uneven distribution of smoking population between the rosacea and control cohorts was thought to be the potential cause because of the socioeconomic differences in both cohorts. In our study, there was no increased risk for lung cancer in rosacea patients as a whole but a marginally increased risk (HR = 1.32; 95% CI = 1.01–1.74) for lung cancer in those aged between 40 and 59 years. Chronic obstructive pulmonary disease, a chronic illness closely related to smoking, was included in our adjusted covariate factors. This may make a compensation to our failure to control smoking behavior and explain the difference in risk of lung cancer in the Danish study and that in ours.
Overall, genetic difference, which contributes to the variant constitutive tendency of developing certain types of cancer and un-uniform resistance to environmental carcinogens, may be the root cause of the observed differences lying between the results of previous western studies and those of ours.
Our study has several strengths. This is the first large-scale study performed that mainly includes Asians to explore the association between rosacea and cancer. Using the nationwide NHIRD as the data source, a potential selection bias could be obviated, thereby making evaluation of a specific type of cancer with low incidence feasible. Moreover, we only included patients with at least three diagnoses of rosacea from outpatient services or one diagnosis from inpatient services and further excluded patients with any two diagnoses of cutaneous lupus erythematosus, seborrheic dermatitis, and acne vulgaris. This leads to a much higher specificity in including rosacea cases than previous epidemiological studies., Moreover, we incorporated several common comorbidities across different organ systems in our covariate factors. These are all potential confounders that may skew the association between rosacea and risk of cancer if uncontrolled.
However, certain limitations exist in the current study. First, differences in the severity of rosacea and heterogeneity of pathophysiology among different subtypes of rosacea may lead to distinct cancer risk. However, analyses of cancer risk according to different severity and subtypes of rosacea were not performed since data regarding subtypes and severity were not available in the NHIRD. Second, personal information such as height, weight, physical activity, dietary style, family history of cancer, and smoking and drinking habits may be important confounders for certain types of cancer. These data are unavailable in the NHIRD and hence not presented and controlled in our study. Third, our study population comprised Taiwanese individuals, whose skin phototypes are primarily III and IV. Therefore, care should be taken when extrapolating our study results to other races, especially results of the association between rosacea and skin cancer.
| Conclusion|| |
Our study revealed no association between rosacea and overall or specific types of cancer. However, these results did not agree with those reported in previous studies from western countries. Further studies should be performed to clarify the association between rosacea and cancer, especially focusing on pathophysiology.
Data in this study were retrieved from the NHIRD provided by the National Health Insurance Administration and Ministry of Health and Welfare and managed by the National Health Research Institutes.
Financial support and sponsorship
Conflicts of interest
There are no conflicts of interest.
| References|| |
van Zuuren EJ. Rosacea. N
Engl J Med 2017;377:1754-64.
Tan J, Berg M. Rosacea: Current state of epidemiology. J Am Acad Dermatol 2013;69:S27-35.
Spoendlin J, Voegel JJ, Jick SS, Meier CR. A study on the epidemiology of rosacea in the U.K. Br J Dermatol 2012;167:598-605.
Abram K, Silm H, Oona M. Prevalence of rosacea in an Estonian working population using a standard classification. Acta Derm Venereol 2010;90:269-73.
Tan J, Schöfer H, Araviiskaia E, Audibert F, Kerrouche N, Berg M, et al.
Prevalence of rosacea in the general population of Germany and Russia – The RISE study. J Eur Acad Dermatol Venereol 2016;30:428-34.
Rosen T, Stone MS. Acne rosacea in blacks. J Am Acad Dermatol 1987;17:70-3.
Alexis AF. Rosacea in patients with skin of color: Uncommon but not rare. Cutis 2010;86:60-2.
Bae YI, Yun SJ, Lee JB, Kim SJ, Won YH, Lee SC. Clinical evaluation of 168 Korean patients with rosacea: The sun exposure correlates with the erythematotelangiectatic subtype. Ann Dermatol 2009;21:243-9.
Two AM, Wu W, Gallo RL, Hata TR. Rosacea: Part I. Introduction, categorization, histology, pathogenesis, and risk factors. J Am Acad Dermatol 2015;72:749-58.
Holmes AD, Spoendlin J, Chien AL, Baldwin H, Chang AL. Evidence-based update on rosacea comorbidities and their common physiologic pathways. J Am Acad Dermatol 2018;78:156-66.
Haber R, El Gemayel M. Comorbidities in rosacea: A systematic review and update. J Am Acad Dermatol 2018;78:786-92. e8.
Coussens LM, Werb Z. Inflammation and cancer. Nature 2002;420:860-7.
Colotta F, Allavena P, Sica A, Garlanda C, Mantovani A. Cancer-related inflammation, the seventh hallmark of cancer: Links to genetic instability. Carcinogenesis 2009;30:1073-81.
Dupont C. Rosacea is not associated with skin cancer. Arch Dermatol 1986;122:1099.
Li WQ, Zhang M, Danby FW, Han J, Qureshi AA. Personal history of rosacea and risk of incident cancer among women in the US. Br J Cancer 2015;113:520-3.
Egeberg A, Hansen PR, Gislason GH, Thyssen JP. Association of rosacea with risk for glioma in a danish nationwide cohort study. JAMA Dermatol 2016;152:541-5.
Egeberg A, Fowler JF Jr., Gislason GH, Thyssen JP. Rosacea and risk of cancer in Denmark. Cancer Epidemiol 2017;47:76-80.
The National Health Insurance Statistics. National Health Insurance Administration. Ministry of Health and Welfare; 2017. p. 31-2.
Cheng CL, Lee CH, Chen PS, Li YH, Lin SJ, Yang YH. Validation of acute myocardial infarction cases in the national health insurance research database in taiwan. J Epidemiol 2014;24:500-7.
Dai YX, Chen TJ, Chang YT. Skin care services and disease prevalence in Taiwan: A nationwide study. Dermatol Sin 2018;36:124-30.
Dai YX, Wang SC, Chou YJ, Chang YT, Chen TJ, Li CP, et al.
Smoking, but not alcohol, is associated with risk of psoriasis in a Taiwanese population-based cohort study. J Am Acad Dermatol 2019;80:727-34.
Chang TH, Tai YH, Dai YX, Chang YT, Chen TJ, Chen MH. Risk of atopic diseases among siblings of patients with attention-deficit hyperactivity disorder: A nationwide population-based cohort study. Int Arch Allergy Immunol 2019;180:37-43.
Gray RJ. A class of K-sample tests for comparing the cumulative incidence of a competing risk. Ann Stat 1988;16:1141-54.
Fine JP, Gray RJ. A proportional hazards model for the subdistribution of a competing risk. J Am Stat Assoc 1999;94:496-509.
Mildner M, Jin J, Eckhart L, Kezic S, Gruber F, Barresi C, et al.
Knockdown of filaggrin impairs diffusion barrier function and increases UV sensitivity in a human skin model. J Invest Dermatol 2010;130:2286-94.
Hvid M, Johansen C, Deleuran B, Kemp K, Deleuran M, Vestergaard C. Regulation of caspase 14 expression in keratinocytes by inflammatory cytokines – A possible link between reduced skin barrier function and inflammation? Exp Dermatol 2011;20:633-6.
Vestergaard C, Hvid M, Johansen C, Kemp K, Deleuran B, Deleuran M. Inflammation-induced alterations in the skin barrier function: Implications in atopic dermatitis. Chem Immunol Allergy 2012;96:77-80.
Scherer D, Kumar R. Genetics of pigmentation in skin cancer – A review. Mutat Res 2010;705:141-53.
DeNardo DG, Coussens LM. Inflammation and breast cancer. Balancing immune response: Crosstalk between adaptive and innate immune cells during breast cancer progression. Breast Cancer Res 2007;9:212.
Liotti F, Visciano C, Melillo RM. Inflammation in thyroid oncogenesis. Am J Cancer Res 2012;2:286-97.
Bishayee A. The role of inflammation and liver cancer. Adv Exp Med Biol 2014;816:401-35.
Beaugerie L, Itzkowitz SH. Cancers complicating inflammatory bowel disease. N
Engl J Med 2015;372:1441-52.
Egeberg A, Weinstock LB, Thyssen EP, Gislason GH, Thyssen JP. Rosacea and gastrointestinal disorders: A population-based cohort study. Br J Dermatol 2017;176:100-6.
Wu CY, Chang YT, Juan CK, Shieh JJ, Lin YP, Liu HN, et al.
Risk of inflammatory bowel disease in patients with rosacea: Results from a nationwide cohort study in Taiwan. J Am Acad Dermatol 2017;76:911-7.
Uemura N, Okamoto S, Yamamoto S, Matsumura N, Yamaguchi S, Yamakido M, et al. Helicobacter pylori
infection and the development of gastric cancer. N
Engl J Med 2001;345:784-9.
Szlachcic A. The link between helicobacter pylori
infection and rosacea. J Eur Acad Dermatol Venereol 2002;16:328-33.
Nakada M, Okada Y, Yamashita J. The role of matrix metalloproteinases in glioma invasion. Front Biosci 2003;8:e261-9.
Ricard D, Idbaih A, Ducray F, Lahutte M, Hoang-Xuan K, Delattre JY. Primary brain tumours in adults. Lancet 2012;379:1984-96.
Chien LN, Gittleman H, Ostrom QT, Hung KS, Sloan AE, Hsieh YC, et al.
Comparative brain and central nervous system tumor incidence and survival between the united states and Taiwan based on population-based registry. Front Public Health 2016;4:151.
Das A, Tan WL, Teo J, Smith DR. Glioblastoma multiforme in an Asian population: Evidence for a distinct genetic pathway. J Neurooncol 2002;60:117-25.
Mochizuki S, Iwadate Y, Namba H, Yoshida Y, Yamaura A, Sakiyama S, et al.
Homozygous deletion of the p16/MTS-1/CDKN2 gene in malignant gliomas is infrequent among Japanese patients. Int J Oncol 1999;15:983-9.
Dubrow R, Darefsky AS. Demographic variation in incidence of adult glioma by subtype, United States, 1992-2007. BMC Cancer 2011;11:325.
Morgan TR, Mandayam S, Jamal MM. Alcohol and hepatocellular carcinoma. Gastroenterology 2004;127:S87-96.
Seitz HK, Pelucchi C, Bagnardi V, La Vecchia C. Epidemiology and pathophysiology of alcohol and breast cancer: Update 2012. Alcohol Alcohol 2012;47:204-12.
Li S, Cho E, Drucker AM, Qureshi AA, Li WQ. Alcohol intake and risk of rosacea in US women. J Am Acad Dermatol 2017;76:1061-7. e2.
Sun K, Ren M, Liu D, Wang C, Yang C, Yan L. Alcohol consumption and risk of metabolic syndrome: A meta-analysis of prospective studies. Clin Nutr 2014;33:596-602.
[Table 1], [Table 2], [Table 3]