Evaluation of Skin Microtopography as a Measure of Ultraviolet Exposure

Transcription

1 Investigative Ophthalmology & Visual Science, Vol. 33, No., May 99 Copyright Association for Research in Vision and Ophthalmology Evaluation of Skin Microtopography as a Measure of Ultraviolet Exposure Johanna M. 5eddon,*t Kathleen M. Egan,* Yuqing Zhang,* Ellen J. Gelles,* Robert J. Glynn,* Carole A. Tucker,* and Evangelos 5. Gragoudasf A pilot study was conducted to investigate the use of skin microtopography as a semiquantitative noninvasive method for estimating cumulative sun exposure in epidemiologic studies of eye disease. The subjects received a kit through the mail containing materials needed to make a replica of the skin texture of a sun-exposed area of the hand. Each subject previously had undergone a skin biopsy around the same site to evaluate elastotic degeneration, and all were interviewed about past sun exposures. A gradable skin impression was obtained from 9 of 5 (83%) participants after two mailings. The impressions were graded according to the degree of skin texture alteration using standard photographs; interobserver reliability was.73 using a weighted kappa statistic. The impression score was correlated most strongly with age (r =.53). Independent predictors of higher impression scores (more skin texture changes) were older age, cigar or pipe smoking, less education, lighter iris color, lighter skin color, male gender, and tendency to sunburn. After adjustment for age and the other predictor variables, the biopsy score was not correlated with the impression grade (r =.8, P =.3). Behaviors indexing sun exposure were not correlated with microtopography. These results suggest that skin microtopography as done in this study reflects aging from intrinsic parameters more than from actinic damage. Invest Ophthalmol Vis Sci 33:93-98,99 Sunlight has been implicated as a cause of several human neoplasms and degenerative conditions. Among these are the ocular diseases including cataract, macular degeneration, and melanoma of the uveal tract. " 5 Most previous epidemiologic investigations have relied on questionnaire data to quantify sunlight exposure, but this method has inherent limitations and may result in misclassification of exposure. In addition, it is not clear which questions provide valid estimates of past exposure. We recently reported an objective method for estimating cumulative exposure to the sun involving computer-based quantitation of dermal elastosis in skin biopsy specimens. Although the method is highly reproducible and has the advantage of providing a quantitative estimate of actinic damage, it is technically complex and involves an invasive procedure. Given the questionable feasibility of using skin biopsies in studies involving subjects sampled from the community, a more easily administered and widely applicable method of assessing ultraviolet exposure objectively would be beneficial From the *Epidemiology Unit and fretina Service, Massachusetts Eye and Ear Infirmary, and the Department of Ophthalmology, Harvard Medical School, Boston, Massachusetts. Submitted for publication: March, 99; accepted November, 99. Reprint requests: Johanna M. Seddon, MD, Epidemiology Unit, Massachusetts Eye and Ear Infirmary, 43 Charles Street, Boston, MA 4. for use in large-scale studies. We evaluated skin microtopography, a method which involves making a replica of the surface of the skin. Previous studies have suggested that skin impressions produced by trained personnel provide a valid estimate of actinic skin changes. 7 ' 8 In this report, we describe the results of a pilot study which examined the following: () the feasibility of obtaining skin impressions by mail for potential use in large surveys, () the reliability of grading skin impressions obtained in this manner, and (3) the validity of skin microtopography as a semiquantitative measure of actinic damage by comparison with skin biopsy scores. Subjects and Methods The subjects were patients with uveal melanoma, and their family members and friends, who had participated in a previous study assessing alternative methods for estimating past sun exposure. In that study, each participant was interviewed concerning measures of sun sensitivity and cumulative ultraviolet radiation exposure at the Massachusetts Eye and Ear Infirmary (MEEI). In addition, each participant underwent a skin biopsy taken from the dorsal surface (sun-exposed area) of the left hand. A method was developed for quantifying the degree of elastosis in the biopsy specimen objectively as an indicator of elastotic degeneration after sun exposure. The resulting biopsy score was found to be highly reproducible and 93 Downloaded From: on //8

2 94 INVESTIGATIVE OPHTHALMOLOGY VISUAL SCIENCE / May 99 Vol. 33 was correlated moderately with several questionnaire measures of cumulative sun exposure. 9 Skin Microtopography Each of the surviving 5 participants in the original study ( patients with uveal melanoma and 49 companions) received a kit through the mail containing the materials needed to make a replica of the skin (skin microtopography). The kit included a premeasured quantity of fast-setting liquid impression material (Optosil Flussig, Bayer, Leuerkusen, Germany), commercially available for dental impressions. Detailed instructions for making the impression at home were included in each mailed packet including diagrams depicting placement of the material on the dorsal surface of the left hand. Participants who had difficulty making the impression or who returned a poor quality impression received another kit. For any subject who did not complete the impression at home, the skin impression was made by study personnel during a follow-up appointment at MEEI. The impression was made within 8 months of the biopsy in all participants. The skin impressions were graded independently by two readers masked to the identity of the participant. A binocular dissecting microscope at X was used. The quality of each impression was scored as "adequate" or "poor." A specimen was considered poor in quality (and therefore not graded) if it lacked sufficient anatomic detail for valid grading. The remaining impressions were graded using an earlier method, as previously described, to assess skin surface alterations thought to reflect actinic damage. The degree of actinic damage was graded from (least damage) to (most damage) using photographic standards and detailed written descriptions of the six grades (Fig. ). The impressions graded differently by the two readers were reevaluated by the two readers jointly, and a consensus grade was assigned. When the degree of damage was judged during arbitration to lie approximately midway between two grades (n = ), the impression was scored as the average of the two grades (.5,.5, 3.5, 4.5, or 5.5). Structured Interview The participants were interviewed in a private office in the hospital outpatient area by a trained interviewer concerning demographic and constitutional factors, behaviors related to sunlight exposure, and smoking and alcohol consumption. The demographic information included age, gender, and education (scored as less than 4 years or 4 or more years). The subject's nationality was evaluated with the referent group comprised of subjects with both parents of Southern European or Mediterranean ancestry as described previously. 5 The skin color of the lower surface of the upper right forearm was graded visually against a ten-step complexion chart from darkest () to lightest () color. Eye color was assessed by comparing the subject's iris color to three standard iris color photographs and classified into four grades, from dark brown () to blue or gray (4). The subjects were asked to estimate the number of moles (none, one to nine, or ten or more) and number of freckles (none, -, or > ) on their skin after a description of these pigmented lesions was provided. The type of skin reaction after exposure to the sun in the summer for hour without sunscreen was reported as no reaction, tan, some redness only, sunburn, painful burn, or burn and blister. A complete residential history (from place of birth to current residence) was obtained. The respondent was asked to estimate the amount of time spent in the sun at each residence. From this information, the total number of years lived below the latitude 4 N latitude was calculated for each subject, as described previously. 5 In addition, a lifetime solar radiation value was calculated. The participants were asked to report their occupational history, including the number of years worked and the average number of hours spent outdoors in the sun between 9 AM and 5 PM on each job. Exposure during those hours also was obtained for intervals of work in the home or when unemployed. The number of hours per week outdoors then was evaluated as a continuous variable. A history of sunlight-related habits, including the number of times per year in which the subject participated in outdoor hobbies, the number of holidays in sunny climates, and the average number of hours spent sunbathing per week in the summer, during both teen-age years and the previous 5 years was obtained. Typical weekly alcohol consumption (the number of drinks of beer, wine, and distilled spirits) was assessed. Smoking was defined as the product of the number of years smoked times the number of packs of cigarettes, number of cigars, or number of bowls of pipe tobacco per day typically smoked. Skin Biopsy A quantitative measurement of elastic tissue degeneration was obtained using a computerized image analysis method. Details of this method have been previously reported. Briefly, a 4.-mm punch biopsy was obtained from each participant from the dorsum of the left hand. The specimens were embedded in paraffin, and 4-/um sections were stained with Verhoeff modified elastic stain. The degree of elastosis in each specimen was quantified using an image digitizer video monitored by a dermatopathologist. The computer was programmed to analyze the interfollicular Downloaded From: on //8

3 No. SKIN MICROTOPOGRAPHYIN UV EXPOSURE / Seddon er ol 95 Fig.. The Beagley-Gibson system of grading cutaneous microtopographs (reproduced by permission of the author). The system of grading cutaneous microtopographs described by Beagley and Gibson relates to changes in skin surface texture, which, in normal skin, is composed of a series of transverse and diagonal primary lines that intersect to form quadrilaterals and triangles. Within these primary lines are sets of smaller secondary lines, which often meet in the center of the figure, forming a star configuration. The six-step grading system devised by Beagley and Gibson, and based on alterations in these skin surface characteristics thought to represent actinic damage, is illustrated in the standard photographs. Grade : Primary lines are all of the same depth. Secondary lines are all clearly visible, are nearly the same depth as the primaries, and often meet to form an apex of triangles ("star formation"). Grade : Someflatteningand loss of clarity of the secondary lines. Star formations are still present, but often one or more of the secondary lines making up the configuration are unclear. Grade 3: Unevenness of the primary lines. Noticeable flattening of the secondaries with little or no star formation. Grade 4: Macroscopic deterioration in texture. Coarse, deep primary lines. Distoration and loss of secondary lines. Grade 5: Noticeable flat skin between the primary lines. Few or no secondary lines. Grade : Large, deep, and widely spaced primary lines. No secondary lines. tissue at a specified depth and area for all slides. The percentage of elastosis was quantified within three rectangles measured vertically from the dermoepidermal junction downward. A score was obtained for each participant based on the number of positive pixels per 8 X -Mm (vertical X horizontal) area with elastic tissue stain. Higher scores indicated greater skin damage. A validity test using independent physical measurements of elastosis showed differences of no more than.7%. Reliability tests for reading the same section on different days and adjacent sections showed good agreement between readings. Analysis The reliability of the grading system was evaluated using a weighted kappa statistic to give credit for near agreement between readers. This method weighs each instance of disagreement by the square of the deviation of the pair of observations from exact agreement. For example, perfect agreement is graded as, disagreement by one is assigned a, disagreement by two is assigned a 4, and so forth. The association between the impression grade and biopsy score was eval- uated using Pearson's correlation coefficient. Multiple linear-regression analysis was used to identify questionnaire items predicting the skin impression score. Associations between the impression grade and each potential predictor variable were computed first after adjusting for age and sex. Each constitutional or behavioral variable then was added into the regression model separately. Variables with partial correlation coefficients significant at the. level were considered for inclusion in further regression analyses. A multivariate regression model then was fitted. Variables entering this regression model were determined by a stepwise method with an F statistic significant at the. level.3 Results The mean age of the 5 participants was 54 yr (range, 8-77). Fifty percent were women. A skin impression of adequate quality was obtained from of these participants (89%). Only one mailing was necessary to obtain an adequate impression from 73 (7%) participants. A total of 47 kits were mailed to obtain 9 gradable skin impressions; six others were ob- Downloaded From: on //8

4 9 INVESTIGATIVE OPHTHALMOLOGY b VISUAL SCIENCE / May 99 Vol. 33 tained during follow-up examination. The quality of the skin impressions was evaluated in relation to the age, gender, and education of the subject (Table ). Older subjects had more difficulty producing a good quality skin impression (chi-square test for trend, P =.). No marked differences were evident by gender or educational level. Interreader Reliability There was exact agreement between the two readers on the skin impression grade for 7 (7%) impressions; the readers disagreed by more than one grade three times (Table ). The interobserver reliability for gradable skin impressions was good (weighted kappa,.73; 95% confidence interval,.4-.8). The consensus grade was 4 or more for 73 (7%) impressions, indicating that most subjects had macroscopic deterioration of their skin texture. Correlation Between Biopsy Score and Microtopography Grade There was moderate correlation between the microtopography grade and the biopsy score (Fig, ; r =.49, P <.) not considering other variables. The regression equation relating these two measures is: biopsy score = (microtopography grade). For the biopsy score, the mean positive pixels per unit area was 845, and the standard deviation was 95. Both measures were highly correlated with age (correlation with age for biopsy: r =., P <.). Predictors of Microtopography Grade Only 8 of subjects (38%) aged 4 yr or younger had macroscopic deterioration (grade 4 or higher) compared with 37 of 4 subjects (93%) aged yr or older. Age was correlated highly with the impression Table. Quality of skin impressions by gender, age, and education Gender Male Female Age <; >59 Education (years) <4 ;>4 Total Good N(%) 49 (48) 53 (5) () 4 (4) 4 (39) 3 (35) (5) Poor N(%) 8() 5(38) () 4(3) 9(9) 4(3) 9(9) 3 X =.,df= (P =.75) X = 5.38, df=(/> =.7) X for trend = 5.37, df= l(/> =.) x =.84, df = {P =.3) Table. Interreader reliability of grading good quality skin impressions Reader grade Total 34 5 Reader grade Weighted Kappa =.73 (95% confidence interval.4-.8) Total grade (r =.53, P <.). Men and women were similar in age (mean age of men, 55 yr; mean age of women, 53 yr; t =.8, P =.4), but men had higher grades than women (t =., P =.4). Stepwise multiple regression was used to identify questionnaire items that independently predicted the skin impression score. Of the factors evaluated, age, cigar or pipe smoking, educational level, iris color, gender, skin color, iris color, and initial skin reaction to summer sun remained in the final model (Table 3). Older age, cigar or pipe smoking, less education, lighter iris and skin color, male gender, and tendency to sunburn were associated with more skin texture changes based on microtopography. These variables together explained approximately % of the variability in the skin microtopography score. The biopsy score did not enter the model at the. level. Adjusting for age and the other predictor variables, the biopsy score explained only 3% of the variability in the impression score (r =.8, P =.3). Discussion Our results suggest that, over a wide range in age and educational level, the subjects were able to produce good quality skin impressions at home when they were sent detailed instructions and appropriate materials. Most subjects were willing to participate and able to produce impressions that could be graded reliably. However, participants in this study were a selected group of individuals who were patients of the investigators, or their family members or friends, and who had previously consented to undergo skin biopsy. Use of skin impressions received by mail in other types of populations who often are asked to participate in large epidemiologic studies may not be as successful. Overall, the strongest set of predictors of the skin microtopography grade were intrinsic, demographic, and constitutional factors, particularly age. This may indicate that such factors provide the most information concerning cumulative superficial skin damage associated with ultraviolet exposure and that behaviors are less important. These results are consistent Downloaded From: on //8

5 No. SKIN MICROTOPOGRAPHY IN UV EXPOSURE / Seddon er ol 97 Correlation Between Biopsy Score and Microtopography Grade Among Participants With Gradable Skin Impressions 55-5 (r =.49,P<.) Microtopography Grade Fig.. Correlation between biopsy score and microtopography grade among participants with gradable skin impressions (r =.49, P <.). Biopsy score based on number of positive pixels per 8 X ^m area with elastic tissue stain using a computerized image analysis. Microtopography grade (-) based on skin impression standards. with those of others who suggested that age is one of the best available estimates of total accumulated dose of ultraviolet radiation. 7 ' 8 It was found also that male sex was associated positively with a higher grade of skin damage, which was independent of constitutional factors and sunlight exposure. 8 It is possible that men are more likely to have greater actinic damage from behaviors that increase ultraviolet exposure, which were not captured by questionnaires used in either this previous study or in ours. With regard to other constitutional factors, lighter iris color, lighter skin color, and tendency to burn on exposure to sunlight were related independently to the skin impression score in our study, controlling for demographic and sunlight behavior variables. The tendency to sunburn was the only constitutional factor independently related to skin changes in the previous study. 8 Little of the variability in microtopography grade could be explained by the biopsy score, after adjusting for age. Furthermore, none of the other estimates of ultraviolet exposure (eg, years lived in the South, lifetime solar radiation exposure, and frequency of outdoor hobbies) were related to the skin impression grade. These findings suggest that skin microtopography as assessed in this study may be a relatively poor measurement of actinic damage. Although a skin impression assesses more superficial microarchitectural skin changes, while a biopsy specimen reflects deeper cutaneous elastotic degeneration related to ultraviolet exposure, we might expect these changes to be correlated. However, elastosis may be an earlier marker of ultraviolet light damage and can be found in relatively mild photodamaged skin. 4 Elastosis then may not necessarily correlate well with fine and coarse wrinkling of the skin. Previous studies suggest a stronger association between skin microtopography and measurements of cumulative sun exposure than we could demonstrate. One study reported a high correlation between "photoaging" of the skin (as assessed by a dermatologist) and independent grades obtained from skin impressions. 5 Skin impressions have been used to assess the effect of tretinoin on photodamaged skin. An association was found between impression grade and solar keratosis and nonmelanotic skin cancer and that Table 3. Independent predictors of skin impression grade based on stepwise multiple regression among good quality skin impressions Variable Category Coefficient Standard error Partial R Age Smoking cigar/pipe Education (years) Gender Iris color Skin color Skin reaction to sunlight Total R Continuous No* Yes :4* <4 Female* Male Brown-blue (ordinal) Dark-light (ordinal) Severe burn to tan (ordinal) Variables not entering at. level were cigarette smoking (pack-years), alcohol consumption, nationality, number of moles and freckles, years lived in the south, lifetime solar radiation exposure, lifetime occupational sun exposure, frequency of outdoor hobbies (current and during teen years), number of holidays in sunny climates (current and during teen years), and hours per week sunbathing in summer (current and during teen years). * Referent category. Downloaded From: on //8

6 98 INVESTIGATIVE OPHTHALMOLOGY & VISUAL SCIENCE / May 99 Vol. 33 study referred to a previous finding of an age-adjusted correlation between microtopography grade and a total lifetime ultraviolet radiation score. 7 In addition, an independent association was reported with primarily outdoor occupation. 8 However, these studies 7-8 did not control for education, which was independently correlated with skin changes in our study (more skin texture modifications with fewer years of education), and this may explain in part the differences in our results. Although these studies differed from ours in geographic site and probably occupational exposures, our study had an adequate range of ultraviolet exposures (as demonstrated by the biopsy scores and other related behaviors) and was sufficient to detect a stronger correlation between microtopography and ultraviolet light parameters if one were present. Another difference between this and previous studies, in addition to the different populations studied, was that skin impressions were obtained by mail. It is possible that such impressions were less representative of the skin's surface architecture than impressions made under controlled circumstances by trained personnel. A comparison between impressions obtained by mail and those obtained in person on the same individuals would be informative. It was interesting that cigar or pipe smoking was associated independently with microtopographic skin changes on the hand. Pack years of cigarette smoking alone, however, was not. A marked association has been reported between cigarette smoking and facial wrinkling, and the author suggested that wrinkled skin be added to the observations concerning the hazards of smoking. 7 Pipe and cigar smokers were not evaluated as smokers in that study. Impressions were obtained from a sun-exposed area of the hand. This site was chosen because biopsies had been taken previously from a similar site on the hand and a goal of the study was to compare biopsy results with microtopography. Although damage to the skin around the eye may be more relevant to the study of eye diseases, self-made paraocular impressions are probably not safe or practical and were therefore not included in this study to assess self-administered impressions obtained through the mail. Questionnaire methods aimed at assessing sunlight exposure are known to be problematic because weekly and seasonal variability make it difficult to assess current exposures, and past exposures (which may be more relevant) are even more difficult to measure reliably. The resulting misclassification or measurement error may explain in part the differences in results among studies. Furthermore, individuals who receive the same exposure will vary in their susceptibility to solar insult in ways that we may not be measuring (eg, immunogenetic characteristics). The use of skin biopsies to assess ultraviolet light damage appears reliable and valid, 9 but it is not practical in most circumstances. Skin microtopography is reliable and more easily administered. However, on the basis of this study, microtopography does not provide much additional information beyond knowledge of the subject's age and constitutional factors. Other techniques for efficient, reliable assessment of ultraviolet exposure are needed. Key words: epidemiology, microtopography, skin biopsy, skin impressions, ultraviolet exposure, UV measurement References. Collman GW, Shoe DL, Shy CM, Checkoway H, and Luria AS: Sunlight and other risk factors for cataract: An epidemiologic study. Am J Public Health 78:459, Taylor HR, West SK, Rosenthal FS, et al: Effect of ultraviolet radiation on cataract formation. N Engl J Med 39:49, Young RW: Solar radiation and age-related macular degeneration. Surv Ophthalmol 3:5, Tucker MA, Shields JA, Hartge P, Augsburger J, Hoover RN, and Fraumeni JF: Sunlight exposure as a risk factor for intraocular malignant melanoma. N Engl J Med 33:789, Seddon JM, Gragoudas E, Glynn RJ, Egan KM, Albert DM, and Blitzer PH: Host factors, ultraviolet radiation, and risk of uveal melanoma. A case-control study. Arch Ophthalmol 8:74, 99.. Flotte T, Seddon J, Zhang Y, Glynn RJ, Egan KM, and Gragoudas ES: A computerized image analysis method for measuring elastic tissue and solar elastosis. J Invest Dermatol 93:358, Holman CDJ, Armstrong BK, Evans PR, et al: Relationship of solar keratosis and history of skin cancer to objective measures of actinic skin damage. Br J Dermatol :9, Holman CDJ, Evans PR, Lumsden GJ, and Armstrong BK: The determinants of actinic skin damage: Problems of confounding among environmental and constitutional variables. Am J Epidemiol :44, Zhang Y, Glynn R, Seddon J, Egan K, and Gragoudas E: A validation study of questionnaire measures of cumulative sunlight exposure. Invest Ophthalmol Vis Sci 3:459, Holman CDJ: Risk factors in the causation of human malignant melanoma of the skin. Doctor of Philosophy thesis. University of Western Australia, Nedlands, Western Australia, 983, pp Scotto J and Fears JR: Skin cancer epidemiology: Research needs. NCI Monogr 5:9, Maclure M and Willett WC: Misinterpretation and misuse of the kappa statistic. Am J Epidemiol :, Kleinbaum DG, Kupper LL, and Muller KE: Applied Regression Analysis and Other Multivariable Methods, nd ed. Boston, PWS-Kent, 988, pp Gilchrest BA: Physiology and pathophysiology of aging skin. In Biochemistry and Physiology of the Skin, Goldsmith LA, editor. Oxford, UK, Oxford University Press, 99, pp Grove GL, Grove MJ, and Leyden JJ: Effects of topical retinoin on photoaged skin as measured by optical profilometry. In Methods in Enzymology, Packer L, editor. New York, Academic, 99, pp Weinstein GD, Nigra TP, Pochi PE, et al: Topical tretinoin for treatment of photodamaged skin. Arch Dermatol 7:59, Daniell HW: Smoker's wrinkles: A study in the epidemiology of "crow's feet." Ann Intern Med 75:873, 97. Downloaded From: on //8