Adjusted odds ratios for solar keratosis by hair color comparing psoriatic patients and controls.
Adjusted odds ratios for solarkeratosis by propensity to sunburn comparing psoriatic patients and controls.
Paltiel O, Adler B, Herschko K, Tsukrov B, David M. Are Patients With Psoriasis Susceptible to the Classic Risk Factors for Actinic Keratoses?. Arch Dermatol. 2004;140(7):805-810. doi:10.1001/archderm.140.7.805
Copyright 2004 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.2004
An increased prevalence of benign solar damage (eg, facial wrinkles) but not neoplastic lesions was observed among patients with psoriasis whowere exposed to Dead Sea climatotherapy compared with controls.
To compare the prevalence of actinic keratosis in psoriatic patients and controls and to assess whether known risk factors behave similarly inboth groups.
Multicenter cross-sectional study.
Dermatology clinics in 4 participating Israeli hospitals and at a Dead Sea clinic.
Adult subjects (n = 460) with plaque-type psoriasis were recruited from the Israel Psoriasis Association (volunteer sample) and from dermatology clinics(convenience sample). The control group (n = 738) consisted of nonimmunosuppressed patients attending these clinics for benign conditions unrelated to sun exposure,such as atopic or contact dermatitis.
Main Outcome Measures
Prevalence and distribution of actinic keratoses and odds ratios associated with skin, hair, and eye color and propensity or history of sunburn adjustedfor age, ethnicity, and sun exposure .
Actinic keratoses were observed in 200 controls (27%) and 51 subjects (11%) (P<.001). This increased prevalence occurred in bothsexes, participants aged 35 years or older, all ethnic groups, smokers, and nonsmokers. The anatomical distribution of lesions did not substantially differbetween subjects and controls. In multivariate analysis, psoriasis conferred a protective effect (odds ratio, <1), as did dark skin, dark eyes, anda history of severe sunburn in childhood. However, significant interactions were observed between psoriasis and hair color as well as psoriasis and propensityto sunburn, whereby a linear association was observed for controls but not for patients with psoriasis.
Psoriasis confers protection against actinic keratosis. Hair color and propensity to sunburn exert differential effects among psoriatic patientsand controls.
There is a consensus among the scientific community that the 3 major types of skin cancer—squamous cell carcinoma (SCC), basal cell carcinoma(BCC), and malignant melanoma—are caused by sun exposure.1,2 Furthermore, inherited characteristics, such as skin type and propensity to sunburn, mayhave a marked effect on the risk of skin cancer.3 It is not known whether the presence of other dermatologic conditions modifiesthe association between sun exposure, skin type, and actinic damage.
Psoriasis is a chronic skin condition that affects approximately 2%4 of the population, but with considerable ethnic and geographic variation.5 Manifestations of thedisease are often ameliorated by sun exposure.6,7 An increased incidence of nonmelanoma skin cancer (NMSC) (especially SCC) hasbeen reported in individuals with psoriasis who are exposed to high cumulative doses of psoralen–UV-A.8- 11 Inone study, climatotherapy at the Dead Sea among Danish patients was found to be associated with an increased risk of NMSC.12 Thestudy results, however, were possibly confounded by the fact that patients selected for Dead Sea climatotherapy were those whose psoriasis was improvedby sun exposure. Treatment with UV-B may also be associated with a mild increase (2% per year) of NMSC.13 High-dose psoralen–UV-A therapy (1000 J/cm2) has also been reported to be associatedwith an increased risk of actinic keratosis.14
A cross-sectional study was previously performed in Israel that compared actinic damage among patients with psoriasis (87% of whom had undergone climatotherapyat the Dead Sea Solarium Clinic, Ein Bokek, Israel) and controls (individuals without psoriasis) (M.D., B. T., B. A., et al, unpublished data, 2000). Inthat study, the control subjects had higher self-reported rates of previous skin biopsies, removal of benign growth, or previous malignant neoplasms.There was an association between extent of exposure to the Dead Sea and benign photodamage, such as facial wrinkles, elastosis, solar lentigo, and poikiloderma.However, solar keratosis was more prevalent among controls than among patients with psoriasis and showed no association with days of exposure at the DeadSea. This surprising finding prompted a detailed analysis of factors associated with the presence of solar keratoses, with a comparison of persons with andwithout psoriasis.
Subjects were patients with plaque-type psoriasis aged 20 to 70 years with a disease duration of at least 7 years. They were recruited from amongmembers of the Israel Psoriasis Association at a Dead Sea psoriasis clinic or attended dermatology clinics at 1 of the 4 participating hospitals. Controlswere patients aged 20 to 70 years who were attending dermatology clinics for benign skin conditions, eg, contact dermatitis and atopic dermatitis. Patientswith vitiligo, immunosuppression, autoimmunity, or suspected malignancy as the reason for the clinic visit were excluded as controls. We also excludedcontrols with skin types V or VI.
All participants provided signed informed consent, and the study protocol was approved by the institutional review boards of all participating hospitals.A questionnaire was administered to subjects and controls and included items concerning demographic characteristics, sun exposure, propensity to sunburn,and previously diagnosed benign and malignant neoplasms. A structured physical examination was performed by a qualified dermatologist who noted skin type,hair color, eye color, and the presence and location of suspected malignant skin lesions as well as solar keratosis and other signs of photodamage.
We compared characteristics of subjects with psoriasis and controls using the χ2 test for categorical variables and the Mann-Whitneytest for comparing medians of continuous variables. Variables included in the analysis were group (patients with psoriasis or controls); skin type (I-IVreclassified into light or dark); eye color (black/brown or blue/green); hair color (black, brown, or blonde/red); propensity to sunburn (often/always,sometimes, or never), and history of severe sunburn in childhood (yes or no). We also constructed a summary variable of "fairness," which took into accounteye color, hair color, and complexion. We constructed logistic regression models for the presence of solar keratosis on examination, controlling forage (continuous variable), country of origin (Israel, Asia, North Africa, or other), yearly hours of sun exposure (recreational and occupational categorizedinto quartiles), and smoking history (current smoker: yes or no). To these models we added psoriasis/control status and individual measures of sun sensitivity,such as skin color, hair color, eye color, and propensity to sunburn. The models were slightly modified in terms of the covariates entered accordingto goodness-of-fit criteria (Hosmer-Lemeshow test).
We then tested whether there were significant interactions between psoriasis/control status and these variables. We also tested whether solar elastosis and solarkeratosis appeared together in the same anatomical locations and measured agreement using the κ statistic. All analyses were performed using SPSSsoftware (Version 10; SPSS Inc, Chicago, III). For all tests of significance, a 2-sided P value of .05 was considered statisticallysignificant.
The study population consisted of 460 subjects with psoriasis and 738 controls (N = 1198). Patients with psoriasis were more likely to be currentsmokers (34% vs 27%), male (57% vs 40%), and of European origin (31% vs 22%) than the controls. Of the patients with psoriasis, 49 (12%) had received psoralen–UV-Atherapy and 109 (26%) had been treated with UV-B. Very few malignant neoplasmswere noted on examination. Six cases of BCC and 3 of SCC were suspected amongthe patients with psoriasis, whereas among the controls the corresponding numbers were 11 and 3. Only 5 cases (2 SCCs [1 each among patients with psoriasisand controls] and 3 BCCs [1 among patients with psoriasis and 2 among controls]) were confirmed histologically. Solar keratoses were present among 200 controls(27%) and 51 patients with psoriasis (11%) (P = .001). In both psoriatic patients and controls, the prevalence of solar keratosisincreased with age. At all ages, in both sexes, and in all ethnic groups,the prevalence of solar keratosis was higher among controls than among psoriaticpatients (Table 1). Among the controls, European-American origin was more common in those with solar keratoses,but this pattern was not seen among the psoriatic patients.
Among those with solar keratosis, the number of lesions per subject varied from 1 to 105, with 26% of psoriatic patients and 38% of controls having6 or more lesions. Table 2 shows the comparison of the distribution of actinic keratosis in patients with psoriasisand controls. In both groups, lesions were most commonly seen in the head, neck, and face areas. Solar keratoses and solar elastosis were found simultaneouslyin the face region in 92 cases (κ = 0.34; P<001, indicating fair agreement).
Table 3 shows the associations between sun sensitivity and the presence of any solar keratosis in psoriaticpatients and controls. Of note, blue or green eye color was associated with a higher frequency of solar keratosis in both psoriatic patients and controls.Light skin was associated with this lesion in controls but not in psoriatic patients. With regard to hair color, no association was noted among the psoriaticpatients, whereas this characteristic was highly associated with the presenceof keratosis amongthe controls. The composite variable fairness, which combined hair, skin, and eye color into a single profile,was associated with the presence of solar keratosis in both psoriatic patientsand controls. Finally, a monotonic relationship was noted between propensityto sunburn and keratoses among the controls but not among the psoriatic patients.Similarly, a weak association between a history of severe sunburn in childhoodand solar keratosis was noted in the psoriatic patients, whereas the association was strong in the controls.
We assessed the prevalence of solar keratosis according to quartiles of sun exposure, and no association was noted among the psoriatic patients,whereas this lesion was associated with degree of sun exposure in the controls.Of note, the association was nonmonotonic, with a positive relationship notedacross the first, second, and third quartiles and a decrease observed for the fourth (heaviest exposure) quartile.
The logistic regression analyses are shown in Table 4. In this multivariate analysis, we show the association between solar keratosis, patient characteristics (eg, skin color and eye color),group (patients with psoriasis or controls), and the interactions between them. The reported odds ratio are adjusted for covariates such as ethnic origin(ie, Asia-Africa, Europe-America, and Israel), sex, age, and degree of sunexposure. Smoking was removed because it did not contribute to the models.In these analyses, psoriasis exerted a protective effect (odds ratio, <1) on the presence of solar keratosis. Furthermore, dark skin, dark eyes, andblack hair were also protective against the presence of solar keratosis. Significant interactions were noted between black vs blonde hair color and psoriasis whenthe presence of solar keratosis was analyzed, and between propensity to sunburnand psoriasis, meaning that the relationship between these patient characteristicsand solar keratosis have differential effects in psoriatic patients and in nonpsoriatic patients (Figure 1 and Figure 2). This analysis shows a clear increasein the odds of solar keratosis with lighter hair color and with increased propensity for sunburn in the controls, whereas the relationship was flatamong the patients with psoriasis.
Traditional and well-established risk factors for NMSC and photodamage have included skin and hair characteristics as well as occupational and recreationalsun exposure.1 These factors,2,15,16 as well as a history of severe sunburn before the age of 20 years,17 arealso important in determining susceptibility to actinic keratoses. To our knowledge, there has not been a systematic comparison of the relative effectsof these risk factors in psoriatic and nonpsoriatic patients.
Solar or actinic keratoses are generally considered to be premalignant lesions,15,16,18 butsome authors19,20 consider them to be established SCCs. In the present cross-sectional study, their prevalencewas lower among psoriatic than nonpsoriatic patients attending dermatology clinics for problems unrelated to photodamage. Some of these differences maybe explained by differential distribution of age, ethnicity, and sun exposurebetween subjects and controls. However, on multivariate analysis, even afterthese factors and characteristics related to sun sensitivity, such as skincolor, eye color and hair color, were controlled for, the odds of solar keratosiswas strikingly lower for the patients with psoriasis than for the controls. Furthermore, in some instances (eg, hair color and propensity to sunburn),the classic risk factors appeared to have a differential effect in psoriatic patients compared with nonpsoriatic patients. This analysis provides preliminarydata that demonstrate a modifying effect of psoriasis on the classic factors that have been shown to affect susceptibility to neoplastic or preneoplasticphotodamage.
The first question that should be asked is whether our findings were the result of chance. Although the interactions we observed were statisticallysignificant, we did not perform the study with this hypothesis in mind. Selection bias may have played a role, as neither the psoriatic group nor the controlgroup was population based. However, given the high degree of sun exposureamong the patients with psoriasis, it would have been expected that they wouldbe at greater risk for the development of skin neoplasms and preneoplastic conditions than a population of patients with rashes and other dermatologiccomplaints. Since NMSC is not reported to the Israel Cancer Registry, we have no population-based estimate of actinic keratoses or NMSC in Israel; therefore,it is difficult for us to measure the degree of bias in the selection of the study population. Our analysis took into account confounders such as countryof origin, age, sex, and smoking. Indeed, the results were weakened when we adjusted for these factors, but some of the interactions remained. Unmeasuredconfounders may have played a role in these results.
Therefore, while these findings may be spurious, the observation that actinic keratosis is more prevalent among nonpsoriatic patients is consistentin all subgroups, and there is a suggestion that hair color and propensityto sunburn played a differential role in contributing to solar keratosis inthe 2 groups. Is there biological plausibility for this finding? A survey of psoriatic patients21 showed that skin type(based on ability to tan and susceptibility to sunburn) showed higher correlations with minimal erythema dose after UV irradiation than did hair or eye color.Interestingly, in 1984, Stern and Momtaz22 found that skin type (and not hair or eye color) predicted skin cancer risk in psoriaticpatients who were treated with psoralen–UV-A, providing additional evidence for a possible differential or inconsistent effect of the classic risk factorsfor NMSC among patients with psoriasis.
Apart from malignant changes ascribed to the treatment of psoriasis , little is known about actinic damage in this disease. Two studies performedin the 1980s suggested that persons with psoriasis have a similar risk of developing SCC compared with several other populations of patients.23,24 In a Danish cohort, the risk of NMSCamong patients with psoriasis was 2.5 times that of the general population.25 A more recent cohort study26 reported a 2- to 4-fold risk of NMSC developing among persons with psoriasis comparedwith hypertensive controls, with the risk varying with the severity of psoriasis. It is difficult to separate the risks associated with antipsoriatic treatmentfrom those inherently associated with the disease. Certain skin diseases, such as atopic dermatitis and urticaria, are substantially rarer among psoriaticpatients than among controls.27 In 1977, Kocsard28 reported that actinic keratoses are rare in patients with psoriasis compared with controls. His study, which was carried out amongAustralian military personnel, found relatively few cases of actinic keratosis, even among fair-skinned and blue-eyed persons with psoriasis, whereas 88%of the persons without psoriasis who were examined had evidence of these lesions.
The prevalence of psoriasis appears to vary by country and skin type,29 with a higher frequency noted in European populations compared with African or Asian populations and with specific HLA types.4 Furthermore, while actinic keratoses are thought tobe a "dose meter of chronic sun damage,"16 this does not seem to be the case in all populations.30 Toour knowledge, there is only 1 case report of multiple actinic keratoses that developed in sun-protected areas in a psoriatic patient who received a doseof 883 J/cm2 of UV-A,31 but little is known about the inherent (as opposed to treatment-related) risk of malignantand premalignant skin disorders in persons with psoriasis.
On a genetic level, the GSTM1 null phenotype32 has been shown to be a risk factor for solar keratosis in a white population, after skin type and ability to tan were controlledfor. On the other hand, a recent study33 has shown that GSTM1 variant alleles are associated withpsoriasis. Thus, alleles associated with an increased incidence of psoriasis may be associated with a decreased risk of actinic skin damage.
Furthermore, recent studies have shown that psoriatic patients with skin cancer (BCC) have evidence of increased DNA damage and defective DNArepair compared with those without BCC.34,35 Interestingly, Dybdahl et al35 reported that psoriatic patientswithout BCC had marginally higher repair than controls (although the results were not statistically significant), suggesting that skin cancers occur ina subgroup of psoriatic patients, while others may be protected. It is conceivable that in some patients with psoriasis, hyperproliferation of keratinocytesmay protect against the effects of sun damage when DNA repair mechanisms are intact, explaining the interactions we found in our study. Indeed, the occurrenceof SCC within psoriatic plaques is exceedingly rare.36 Our findings may serve as a foundation for future studies specifically designedto answer these questions. Notwithstanding the results of our study, it should be stressed that cutaneous neoplasia occurs in patients with psoriasis andthat efforts to prevent skin cancer in this population are still required.
Correspondence: Ora Paltiel, MDCM, MSc, School of Public Health, Hadassah-Hebrew University, POB 12000, Jerusalem, Israel 91120 (email@example.com).
Accepted for publication December 29, 2003.
This study was supported in part by the Dead Sea Medical Research and Development Center, Dead Sea, Israel.
Participating hospitals were Rabin Medical Center, Petah Tikva; Hadassah-Hebrew University Hospital, Ein-Karem, Jerusalem; Ha'emek Medical Center, Afula;Sheba-Tel-Hashomer Medical Center, Tel-Aviv; all in Israel.