- Email: [email protected]
Age, sunlight, and facial skin: A histologic and quantitative study
Clinical and laboratorv studies Age, sunlight, and facial skin: A histologic and quantitative study Raphael Warren, Phlr," Vladimir Gartstein, Phl)," Albert M. Kligman, MD, PhD,b William Montagna, Phfr,? Richard A. Allendorf," and Gregg M. Ridder, PhDa Cincinnati, Ohio; Philadelphia, Pennsylvania; and Beaverton, Oregon Quantitative methods were developed to assess theinterrelation between ageand sunlight on the facial skin of healthy women living in the same sunny area. Thewomen were grouped intothe following categories: young versus old andlow versus high solar exposure. The features evaluated were perceived age, amount of facial wrinkling, skin color, and skin elasticity. A punch biopsy specimen of cheek skin was obtained and prepared histologically for evaluation ofsolar elastosis. Thehistologic examination was complemented byquantification ofcollagen andelastin bycomputer-assessed image analysis. Perceived agewas estimated by untrained women viewing high quality photographs. As expected, those.with greatersun exposure looked older and hadmore wrinkles, more severe elastosis, increased elastin, and decreased collagen. (J AM ACAD DERMATOL 1991;25:751-60.) Age and sunlight leave their mark on exposed skin.I, 2 Photodamaged skinresults in a prematurely aged appearance with its well-known stigmata of yellowing, wrinkling, blotchiness, various growths, and leathery dryness. These changes are probably the consequence of at least two different processes, the intrinsic biologic clock and actinic damage. Chronic exposure to sunlight induces changes that are superimposed on age-related deviations, especially effacement of microtopography'; reduced microcirculatiorr'; lax, sagging skins, 6; changes in the quantity and integrity of dermal elastic tissue (elastosis7,8); and epidermal atypia and dysplasia." These distortions are almost always expressed in qualitative terms. The interaction between chronologic age and cumulative dosage assumed that the sum of solar exposure is a prominent, if not decisive, factor in the generation of dermatoheliosis. Another question that deserves a quantitative approach is how age and amount of sunlight cause changes in From the Procter & Gamble Company, Miami Valley Laboratories, Cincinnati"; the Department of Dermatology, University ofPennsylvania, Philadelphia''; and the Orgeon Regional Primate Research Center, Beaverton," Supported by The Procter & Gamble Company. Presented in part atthe Forty-sixth Annual Meeting of the American Academy ofDermatology, San Antonio, Tex., Dec. 5-I0, 1987. Accepted for publication May 1, 1991. Reprint requests: R. Warren, PhD, Miami Valley Laboratories, The Procter & Gamble Company, P.O. Box 398707, Cincinnati, OH 45239-8707.
16/1/30608
appearance that influence the way in which naive observers scoreperceived age. Because of the interest in strategies of protection against photoaging, it seemed timely to apply modern technology to secure quantitative informationon the clinical and histologic changes in actinically damaged facial skin. Accordingly, we selected ,women living in an area of high insolation who differed only in respect to age and solar exposure. A panelofwomen judges whowerenaive as to this procedure evaluatedperceived age on viewing highresolution photographs. The amount of elastin and collagen was quantified by image analysis of cheek biopsy specimens. Finally, the total length of wrinkles and furrows was measured. We were able to appraise the impact of age and radiation on the extent of actinic damage. . MATERIAL AND METHODS Subject selection
Westudied 41 caucasian women ofskin types I,II, and III'o who had lived in Tucson, Arizona for at least 10 years. Subjects were selected and placed intoone offour groups after(1) their determination of ageand solar exposure and(2) an independent examination by a dermatologist. These subjects were placed in two groups, onthe basis ofduration ofsunlight exposure during the previous year (in onegroup the women were exposed less than 2 hours a week, with no gross signs of solar damage, i.e., lentigines, keratoses, sunburn, or telangiectases; in the othergroup thewomen were exposed more than 12 hours a week; these had multiple signs of solar damage). The
751
Journal of the American Academy of Dermatology
752 Warren et al.
Fig. 1. Photographic processing for wrinkle measurements. A, Digitized and masked image. B, Wrinkles highlighted.
High-resolution facial photography
Fig. 2. Tonometer placed on temple to measure skin elasticity.
subjects were further assigned to two age groups: 19 young adults 25 to 31 years old, and 22 middle-aged adults 45 to 51 years old. Subjects were also categorized by smoking habits. Other factors such as occupation and ethnicity were not considered. The examination by the dermatologist excluded those subjects with precancerous lesions, skin cancer, or unrelated skin disease. This study followed the Good Clinical Practices guidelines, including written informed consent and review by the Institutional Review Board of Argus Research, Inc.
Before being photographed, each woman removed her make-up and washed her face. Two photographs were taken: a frontal view of the whole face and a close-up view of the left temple ("crow's feet") area. The full-face photographs were used for nonexpert grading of apparent age. For frontal face photography, the apparatus used was a Hasse1bladwith 120S planar lens. The head positioning system included a special chair that could be precisely aligned for front-view photographs. The lighting consisted of three sources providing soft directional illumination from the left, right, and above. The optical magnification was 1:5. The entire apparatus was mounted on a rigid superstructure to precisely control spatial dimensions. For 1:1 close-up photography, a Nikon FM-2 35mm camera fitted with a zoom micro Nikkor f4.0 lens and a single light source was used. Positioning was controlled as described earlier. The film used was Kodak Professional Print Negative; one batch was used for the entire study. Thirty-five millimeter positive transparencies were prepared from the negatives for assessment of perceived age and image analysis. The processing of photographic materials was tightly controlled to keep film density and color balance within ± 0.05 optical density units.
Image analysis
Measurement of furrows. The method to measure wrinkles is that described by Gartstein and Shaya.!' The photographs were digitized and a shading correction was used to compensate for possiblenonuniformity of the light source. The image was then manually masked with a graphics tablet to exclude features such as eyes, mouth,
Volume 25
Number 5, Part I November 1991
Quantitative measures ofskin aging 753
Fig. 3.. Four subjects representing the age and solar exposure groups. Younger low exposure (YL), younger high exposure (YH), older low exposure (OL), and older high exposure (OH). nose, and eyebrows (Fig. I). Measured areas included the forehead, cheek, chin, and nasolabial folds. Also included were regions medial and lateral to the eyes and under the eyebrows, but above the eyelids.The furrows were defined by their shadows by use of a fixed threshold intensity value. Reliability was improved by the use of a specially designed filter that enhanced high-frequency image components. This equipment permitted the detection of all furrows and lines wider than 0.5 mm. Their total length was calculated by summing all lines longer than 0.5 em. Only skin features with a length-to-width ratio greater than 10:1 were defined as wrinkles. No duplicate measures were made, that is, from confluences of bifurcations. Skin color. Skin color was measured on the forehead with the CIE 'L,a,b' color scale.'? The "L-value" (lightness) was used to evaluate the degree of skin darkness and the "a-value" (redness-greenness) was used to evaluate erythema.
Image analysis was performed with a commercially available system that included the VAX 8350 computer (Digital Equipment Inc.) and Gould IP8500 Image Processor (Viacom Inc.). An Eikonix 78/99 Digital Camera (Eiconic, Inc.) was used to convert photographic slides to digital images.
Skin elasticity Elasticity was measured with a tonometer placed over the left temple lateral to the commissure of the eye.11 This instrument consists of a Linear Variable Differential Transformer to measure displacement, an electric solenoid to create a lifting force, and a vacuum to ensure intimate attachment to the skin (Fig. 2). The subjects were supine with their heads resting securely on the padded instrument board. A force of 12.5 gm was applied for 0.25 seconds and abruptly removed. The computer automatically followed the skin deformation recovery period of similar length. The total cycle time was 0.50 seconds. Skin
Journal of the American Academy of Dermatology
754 Warren et al.
YH
Fig. 4. Hematoxylin-and-Lee staining of skin biopsy specimens. Skin biopsy specimens were prepared as described in Material and Methods. Photomicrographs show different degrees of elastosis as described in Table I. Younger low exposure (grade 2) (YL), younger high exposure (grade 4) (YH), older low exposure (grade 6) (OL), and older high exposure (grade 8) (OH). Der, Dermis; E, elastin; Ep, epidermis; Gr, grenz zone. Bar = 100 ~m. elasticity was defined as the percent of recovery after vertical deformation. (The percentage recovery of loose skin is lower than that of firm skin.) This arrangement yields information essentially comparable to the levarometry of Dikstein et al. l 3 Our method, however, used a computercontrolled force that provides fast and highly controlled skin lift/relaxation cycles. Reproducibility of measurements was within 4%.
Perceived age The high-resolution color photographs were examined as illustrated in Fig. 3. A panel of 24 women judges, 20 to 60 years old, evaluated 35 moo slides made from the 2 JA format negative film. The slides were projected into a high-resolution, life-size rear-projection screen. The judges were seated at a computer console 1.5 m from the screen and were asked: "How old does this woman look?" to the nearest year. The photographs were randomly presented to each judge.
Histologic examination A 4 mm punch biopsy specimen was obtained from the malar eminence of the face under local lidocaine anes-
thesia. Each specimen was fixed in phosphate-buffered formalin (pH 7.0) and stored at 4 C. Each specimen was coded to ensure anonymity. The specimens were embedded in glycol methacrylate with the JB-4 Embedding Kit (Polysciences, Inc.), sectioned at 2.0 J,Lm thickness, and stained with hematoxylin and Lee, a mixture of Harris' hematoxylin, Lee's methylene blue, and Lee's basic fuchsin.'! This technique enables visualization of elastin (red or pink) and keratinocyte cytology (Fig. 4). PicroSirius Red F3BA15 was also used for estimation of collagen (Fig. 5). Grading of actinic elastosis. Hematoxylin-and-Leestained sections were examined by one of us (W. M.) and graded on a 9-point scale (Table I). The scale was developed after examination of histologic slides without implicit knowledge of subjects and exposure groups. The scale reflects the degree of elastotic degeneration in the dermal matrix. It is based on multiple markers that occur concurrently during the process of elastosis. The slides were reviewed and graded multiple times with insignificant deviation in elastotic grade. A grade of I indicates absence of elastosis (normal skin) and grade 9 designates an end-stage or severe elastotic degeneration. Fig. 4 illustrates grades of 2, 4, 6, and 8. 0
Volume 25 Number 5, Part I November 1991
Quantitative measures ofskin aging 755
Fig. 5. Picro-Sirius Red-stained sections of skin biopsy specimens. Skin biopsy specimens
were prepared as described in Material and Methods. Photomicrographs represent the four age and solar exposure groups. Younger low exposure (YL), younger high exposure (YH), older low exposure (OL), and older high exposure (OH). C, Collagen. Der, dermis; Ep, epidermis. Bar = 100ILm.
Image analysis of actinic elastosis. We used the IBAS 2000 (Zeiss) to measure the area fraction of the elastotic material in hematoxylin-and-Lee-stained tissues; collagenous material was measured on Picro-Sirius Redstained samples. The thickness of the epidermis and of the subepidermal grenz zone was measured on the hematoxylin-and-Lee-stained tissues. Three microscopic fields, free of skin appendages, were selected for analysis in each stained specimen. The optical part of the IBAS analyzer consistsof a Universal Zeiss UEM microscope with a 16X Planapo objective.Microscopic images are converted into electronic images with a Sony CCD3000 color camera. The IBAS electronicimages contained 512 X 512 pixels that represented approximately 270 X 270 ILm of the original histologic section. The analyzed images were continuously displayed on a high-resolution IBAS color monitor to allow an operator to control the process of analysis. A brief description of algorithms used to measure hematoxylin-and-Lee- and Piero-Sirius Redstained samples follows. Evaluation ofhematoxylin-and-Lee-stainedsamples. The tissue sections were positioned so that the epidermis always occupied the left 25% of the image. The process
Table I. Grading scale for histologic elastosis Grade
Comments
I 2 3
No increase in elastin fiber width Increase in elastin fibers, which are thinner Increase in elastin fiber thickeness with greater amount of fibers more curled and branched. Some macrophages have phagocytosed elastin fibers. Above changes intensified Upper bed of amorphous fibers and lower one of thin, tangled, and dense fibers As above, intensified More disintegration of fibers Thiek amorphous mass in upper bed Mostly amorphous upper bed with nearly total elimination of collagen.
4 5 6 7 8 9
began with the digitization and shading correction of the microscopic color images. The algorithm consisted of two steps. (1) The width of the grenz zone was determined as the distance between the dermoepidermal junction and the edge of the densely stained elastotic material in the
Journal of the American Academy of Derm atology
756 Warren et al.
used to extract the epidermis. The remaining image area was considered to be the dermis and was measured. The green component was selected to identify collagen. The contrast of the stained material was enhanced by normalizing the gray levelhistogram to the full 0-255 range. A fixed threshold levelwas used to segment the gray level image into binary form and the area occupied by the collagen in the dermis was computed. Viable epidermal cell layers. Sections stained with Picro-Sirius-Red were used to visualize the viable epidermis and the granular cell layers (Fig. 5). The number of the viable celllayers in the epidermis (basal, spinous, and granular cell layers) was quantified with the stratum corneum used as the upper reference point and the dermis as the lower reference point. Four different sites per specimen were selected for counting.
Statistical analysis Data were analyzed statistically with analysis of variance (ANOVA and GLM SAS procedure); the expert grading was analyzed by means of Wilcoxon rank statistics (NPARIWAY SAS procedure) .
RESULTS
Fig. 6. Image analysis processing of hematoxylin-and-
Lee-stained skin sections. Skin biopsy specimens were prepared as described in Materials and Methods. A, Microscopic images were first digitized and grenz zone identified. B, Elastin component in the dermis is identified. dermis . A 5 pixel wide band parallel to the junction was sequentially shifted until the magenta-stained elastotic material comprised 20% of this band. The distance between the derrnoepidermal junction and the trailing edge was measured (Fig. 6, A). (2) To measure the elastotic material, the magenta-stained elastotic material in the dermis was enhanced by subtracting the green component from the blue. The resulting image was presented to the operator, who established the threshold level necessary to include all the elastotic material. The total area of the magenta-stained elastin was calculated (Fig. 6, B). Evaluation of'Picro-Sirius-Red-stained samples. The sample was positioned to maximize the visibility of the dermis. The operator oriented the field in such a way that the dermoepiderrnal junction was vertical in the left 25% of the field. The red component was used to identify the epidermis . A linear histogram extension and a directional lowpass filter was applied to the red image component to further enhance the contrast. A fixed threshold level was
Estimation of perceived age. The answers to the question: "How old does this woman look?" are presented in Table II, which shows that the amount of solar exposure sharply intensifiedthe appearance of aging in the older group. By contrast, in the younger group, the amount of sunlight had no effect on perceived age. The perceived age for both groups was older than the chronologie age. Quantification of facial furrows. Table II shows that the older group had significantly more wrinkles than the younger group (p < 0.005). Solar exposure greatly increased the total wrinkle length only in the older group (p < 0.05). The younger group was not affected by sunlight. When the crow's feet periorbital area was specifically examined, only the older group was affected by the amount of solar exposure (p < 0.05). Skin color. No differences in skin color with respect to skin redness and darkness were found among the four groups. Skin elasticity. Table III shows that the differences were age dependent. Solar exposure had no effect. Elastotic grading. Table IV shows that the older group had a significantly higher degree of elastosis than the younger one. Within the younger age group, the amount of sunlight had no impact on
Volume 25 Number 5, Part I November 1991
Quantitative measures ofskin aging 757
Table Il, Effect of sun exposure on perceived age and wrinkle length Exposure
Younger group Low (n = 10) High (n::::9) Older group Low (n =- 9) High (n:::: 13)
Perceived age* (yr) (mean ± SE)
Chronologie age (yr) (mean ± SE)
Total wrinkle lengtht (em ± SE)
Crow's feet areaf (em ± SE)
30.9 ± 1.3 30.8 ± 1.3
28.5 ± 0.6 27.1 ± 0.6
27.4 ± 2.8 22.6 ± 2.6
5.6 ± .0.9 4.9 ± 1.0
53.7 ± 1.3 58.2 ± 1.1
47.8 ± 0.7 47.6 ± 0.4
53.5 ± 5.8 75.7 ± 6.8
12.0 ± 1.3t 17.7 ± 2.0
*Perceived age in older high exposure group is greater than in low exposure group (p < 0.01). tOlder group had more wrinkles than younger group (p < 0.005). tWrinkle length in older high exposure group was greater than in low exposure group (p < 0.05).
elastosis. However, if exposure was tested within the older age groups, the difference was directional (p < 0.1). Assessment of elastosis and collagen. The results of Table IV show that elastin content increased with both age and solar exposure. However, the difference induced by solar exposure was much greater in the older group (p < 0.05). There was a progressive loss of collagen with increasing sun exposure and age. The difference from solar exposure was directionally greater in the older group (p < 0.1). The amount of elastin and collagen was inversely related and the correlation between the two was high (Fig. 7). Grenz zone and epidermal thickness. Most older subjects in either the low-exposure group (8 of 9) or the high-exposure group (12 of 13) had a grenz zone. Sun exposure induced a directional narrowing of the zone from 32.1 to 23.4 ,urn (p < 0.1). Only one subject in each of the younger groups had an identifiable grenz zone. There were no meaningful changes in epidermal thickness related to solar exposure. DISCUSSION Although numerous studies have shown that age can affect a variety of skin features, this study was specifically designed to see whether sunlight aggravated these features from a small but controlled sampling from the same sunny area, and focusing only on the face. Our highly sensitive photographic techniques confirmed what has been intuitively inferred for a long time, that sunlight is a prominent factor for the appearance of premature aging, independent of other factors. By middle age, the subjects who had received more sun exposure appeared older. The extent of dermal degenerative change in this
Table III. Effect of age and solar exposure on wrinkle length Group
Younger group Low (n = 10) High (n = 9) Older group Low (n = 9)
High (n = 13)
Total wrinkle length" (em ± SE)
Crow's feet area * (em ± SE)
27.4 ± 2.8 22.6 ± 2.6
5.6 ± 0.9 4.9 ± 1.0
53.5 ± 5.8 75.7 ± 6.8
12.0 ± L3t 17.7±2.0
*Older group had more wrinkles than younger group (p < 0.005). [Wrinkle length in older high exposure group was greater than in low exposure group (p < 0.05).
group also correlated with premature aging. Although occupations, other than those directly related to outdoor activities (which were included in the questionnaire), might playa role in wrinkle development (i.e., airline pilot exposure to gamma irradiation or baker's exposure to infrared), it was not apparent or a criterion for this study. Similarly, other experiential factors, such as wind exposure, were not discriminating criteria in this study. One benefit of our study design is that it allows a determination of the relation between subjective and objective measures of aging. For example, we found that there was a high correlation between perceived age and facial wrinkles (Fig. 8, A). The assessment of elastosis was also highly correlated with perceived age (Fig. 8, B). Next, elastin and, more significantly, the quantity of collagen, as measured by morphometric image analysis, especially showed tight correlations to elastosis and to perceived age (Fig. 8, C). This is intriguing because the histologic assessment was based on the hematoxylin-and-Lee-stained sections, whereas collagen quantity was assessed with
Journal of the American Academy of Derma tology
758 Warren et al.
70
....
• •
High Exposure Low Exposure
r::::- 0.66 p
•
•
10
,. ..
.. .
•
O.....- -......-----''------''---_--lL.-_~ o 15 30 45 60 Percent Elastin in Dermis
Fig. 7. Correlation plot between elastin and collagen. Relative abundanceof elastin and collagen was quantified by means of image analysis as described in Materialsand Methods. This plot is a correlation between the abundance of elastin in the hematoxylin-andLee-stained sections and theabundance of collagen in thePicro-Sirius Red-stainedsections.
Table IV. Effect of age and solar exposure on histologic elastosis, dermal elastin, and collagen Group
Younger group Low (n = 10) High (n =9) Older group" Low (n= 9) High en =13)
Elastosis grade( ± SE)*
% Elastin( ± SE)t
% Collagen (± SE)t
3.1 ± 0.3 3.4 ± 0.7
20,4 ± 3.3 23.5 ± 3.4
61.7 ± 4.2 58.0 ± 4.5
6.0 ± 0.5:1: 6.8 ± 0.4
31.7 ± 3.4§ 39.9 ± 2.9
40.2 ± 4.511 31.5 ± 3.7
'Older subjects had greaterelastosis than younger subjects (p < 0.001). tOlder subjects had more elastin and less collagen than younger subjects (p < 0.005). :t:Elastosis wasworse in olderhighexposure groupthan in olderlow exposure group (p < 0.1). §Older highexposure grouphadmoreelastin than the olderlowexposure group (p < 0.05). IIOlder highexposure group had less collagen than the olderlowexposuregroup (p < 0.1).
thePiero-Sirius-Red-stained sections. Thissuggests that instrument measurements of collagen quantity provide the strongest indicator of elastosis. Earlier studies by Uitto et al. 16 also demonstrated a direct relation between morphometric quantification of elastic fibers in the dennis andbiochemical quantification of elastin. The grenz zone refers to a subepidermal band of normal dermis. It isthought tobe a siteof continual
dermalrepair and consists of normal collagen fibers. With our histochemical methods, the grenz zone becomes visually apparent only after there is sufficientelastotic damage. Thisis strikingly clear in the oldergroup. With progressive elastosis, the zoneappears to become thinner. Clinicians have proposed that actinic elastosis can be prevented and possibly reversed if the solar assault is stopped (e.g., through the use of sunscreens). In their study conducted
Volume 25 Number 5, Part 1 November 1991
Quantitative measures ofskin aging 759
•
•
.... ..-:,
65
leA .
55
•
HighExposure
•
Law Exposure
r=O.85
p
A 70 A
• 60
r = 0.79 p<0.OO01
••
•
•
GI
ce
'e ...... GIll)
40
•
.
I 50
> ..
--
•
r=-O.90 p
•
22
.
E
A
~60
•
.1;
m
c:
~ 8....
t
~ 20
•
=40
40
. t
30
....
•
e55
•
•
.... 20 2
B
3
4
5
6
7
8
0
9
Histological Grade
Hlgh~"
A Low ExpoIlU ..
•
'Qj u>.
...GI D.
120
80
Facial Wrinkles (cent'meters)
• •
Low Exposure High Exposure
01
a a
C
1
3 5 7 Histological Grade
• • 9
Fig. 8. A, Correlationplot between perceived age and totalfacial wrinkles. B, Correlation plot betweenperceived age and elastosis. C, Correlation plotbetween collagen andelastosis.
more than 25 years ago, Gerstein and Freeman'? showed that actinically damaged neck skin would improve when transplanted to a solar protected area. The grenz zone became wider with the deposition of
normal histochemically stained collagen. The epidermis also appeared normal with the reappearance of rete ridges and restoration of more uniform epidermal differentiation.
760 Warren et al.
The treatment of sun-damaged skin with chemical peels or dermabrasion has been used for several years.IS-20 Topical treatmentwithtretinoinhas commandedmuchpublic attention andindicates the feasibility of a pharmacologic approach to partial reversalofphotodamage. 2.1, 22Asprevention andrepair of actinically damaged skin becomes evenmoreemphasized in the media and scientific literature the quantitative methods wehavedescribed willbe useful for measuring the benefits of strategiesaimed at avoiding the appearance of premature aging. We are indebted to F. E. Dunlop, MD, of Argus Re-
search,Inc., and R. P. Block, R. M. Cook, W. H. Elsnau, and C. R. Hawley for their expert technical assistance.
REFERENCES 1. Kligman LH, Kligman AM. Photoaging, In: Fitzpatrick TB, Eisen AZ, Wolff K, et al, eds, Dermatologyin general medicine. 3rd ed.New York: McGraw-Hill, 1986:1470-5. 2. Gilchrest BA.Skinand aging processes. BocaRaton: CRC Press, 1983. 3. Holman COl, Armstrong BK, Evans PR, et al. Relationship of solar keratosis and history of skincancer to objectivemeasures ofactinic skindamage. Br J Dermatol1984; 110:129-38. 4. MontagnaW, Carlisle K. Structural changes in aged human skin. J Invest DermatoI1979;73:47-53. 5. Leveque JL, CorcuffP,de Rigal J, et al. In vivostudies of the evolution ofphysical properties of the human skin with age. Int J DermatoI1984;23:322-9. 6. Daly CH, Odland GF. Age-related changes in the mechanical properties ofskin. J InvestDermatoI1979;73:84-7. 7. DittoJ, MatsuokaLY, Kornberg RL. Elastic fibersin cutaneous elastosis. In:Rudolph R, ed,The biological basisof cosmetic surgery. St. Louis: CV Mosby, 1986:307-38.
Journal of the American Academy of Dermatology 8. Mera SL, Lovell CR, Russell Jones R, et al. Elastic fibers in normaland sun-damaged skin:an immunohistochemical study. Br J DermatoI1987;1l7:21-7. 9. KligmanLH, KligmanAM. The nature of photoaging: its prevention and repair. Photodermatology 1986;3:215-27. 10. Cripps OJ. Natural and artificial photoprotection. J Invest Oermatol 1981;77:154-7. 11. Gartstein Y, Shaya SA. Image analysis of facial skin features. Proc Int Soc Optic Eng 1986;626:284-8. 12. Pratt WK. Digital image processing. New York: John Wiley & Sons, 1978:50-90. 13. DiksteinS, Hartzshtark A, Bercovici P. The dependenceof low-pressure indentation,slackness, and surface pH on age in forehead skin of women. J Soc Cosmet Chem 1984; 35:221-8. 14. Montagna W, Kirchner S, Carlisle K. Histology of sundamaged human skin. J AM ACAD OERMATOL 1989; 21:907-18. 15. Sweat F, Puchtler H, RosenthalSI. Sirius red F3BA as a stain for connective tissue. Arch Pathol 1964;78:69-72. 16. Uitto J, Paul L, Brockley K, et al. Elastin fibers in human skin: quantitation of elasticfibers by computerized digital image analyses and determination of elastin by radioimmunoassay of desmosine. Lab Invest 1983;49:499-505. 17. Gerstein W, Freeman RG. Transplantation of actinicalty damaged skin. J Invest OermatolI963;41:445-50. 18. Ayres S III, Wilson JW, Luikart R II. Dermal changes following dermabrasion. Arch Dermatol 1959;79:553-68. 19. Ayres S Ill. Dermal changes following application of chemical cauterants to aging skin: superficial chemosurgery. Arch Oermatol 1960;82:578-85. 20. Stegman S1.A comparative histological studyof the effects of three peeling agents and dermabrasion of normal and sundamagedskin. Aesth Plast Surg 1982;6:123-5. 21. KligmanAM, GroveGL, HiroseR, et al. Topicaltretinoin for photoaged skin. J AMACAD DERMATOL 1986;15:83659. 22. WeissJS, EllisCN, HeadingtonJT, et al. Topicaltretinoin improves photoaged skin. JAMA 1988;259:527-32.
16/1/30608
appearance that influence the way in which naive observers scoreperceived age. Because of the interest in strategies of protection against photoaging, it seemed timely to apply modern technology to secure quantitative informationon the clinical and histologic changes in actinically damaged facial skin. Accordingly, we selected ,women living in an area of high insolation who differed only in respect to age and solar exposure. A panelofwomen judges whowerenaive as to this procedure evaluatedperceived age on viewing highresolution photographs. The amount of elastin and collagen was quantified by image analysis of cheek biopsy specimens. Finally, the total length of wrinkles and furrows was measured. We were able to appraise the impact of age and radiation on the extent of actinic damage. . MATERIAL AND METHODS Subject selection
Westudied 41 caucasian women ofskin types I,II, and III'o who had lived in Tucson, Arizona for at least 10 years. Subjects were selected and placed intoone offour groups after(1) their determination of ageand solar exposure and(2) an independent examination by a dermatologist. These subjects were placed in two groups, onthe basis ofduration ofsunlight exposure during the previous year (in onegroup the women were exposed less than 2 hours a week, with no gross signs of solar damage, i.e., lentigines, keratoses, sunburn, or telangiectases; in the othergroup thewomen were exposed more than 12 hours a week; these had multiple signs of solar damage). The
751
Journal of the American Academy of Dermatology
752 Warren et al.
Fig. 1. Photographic processing for wrinkle measurements. A, Digitized and masked image. B, Wrinkles highlighted.
High-resolution facial photography
Fig. 2. Tonometer placed on temple to measure skin elasticity.
subjects were further assigned to two age groups: 19 young adults 25 to 31 years old, and 22 middle-aged adults 45 to 51 years old. Subjects were also categorized by smoking habits. Other factors such as occupation and ethnicity were not considered. The examination by the dermatologist excluded those subjects with precancerous lesions, skin cancer, or unrelated skin disease. This study followed the Good Clinical Practices guidelines, including written informed consent and review by the Institutional Review Board of Argus Research, Inc.
Before being photographed, each woman removed her make-up and washed her face. Two photographs were taken: a frontal view of the whole face and a close-up view of the left temple ("crow's feet") area. The full-face photographs were used for nonexpert grading of apparent age. For frontal face photography, the apparatus used was a Hasse1bladwith 120S planar lens. The head positioning system included a special chair that could be precisely aligned for front-view photographs. The lighting consisted of three sources providing soft directional illumination from the left, right, and above. The optical magnification was 1:5. The entire apparatus was mounted on a rigid superstructure to precisely control spatial dimensions. For 1:1 close-up photography, a Nikon FM-2 35mm camera fitted with a zoom micro Nikkor f4.0 lens and a single light source was used. Positioning was controlled as described earlier. The film used was Kodak Professional Print Negative; one batch was used for the entire study. Thirty-five millimeter positive transparencies were prepared from the negatives for assessment of perceived age and image analysis. The processing of photographic materials was tightly controlled to keep film density and color balance within ± 0.05 optical density units.
Image analysis
Measurement of furrows. The method to measure wrinkles is that described by Gartstein and Shaya.!' The photographs were digitized and a shading correction was used to compensate for possiblenonuniformity of the light source. The image was then manually masked with a graphics tablet to exclude features such as eyes, mouth,
Volume 25
Number 5, Part I November 1991
Quantitative measures ofskin aging 753
Fig. 3.. Four subjects representing the age and solar exposure groups. Younger low exposure (YL), younger high exposure (YH), older low exposure (OL), and older high exposure (OH). nose, and eyebrows (Fig. I). Measured areas included the forehead, cheek, chin, and nasolabial folds. Also included were regions medial and lateral to the eyes and under the eyebrows, but above the eyelids.The furrows were defined by their shadows by use of a fixed threshold intensity value. Reliability was improved by the use of a specially designed filter that enhanced high-frequency image components. This equipment permitted the detection of all furrows and lines wider than 0.5 mm. Their total length was calculated by summing all lines longer than 0.5 em. Only skin features with a length-to-width ratio greater than 10:1 were defined as wrinkles. No duplicate measures were made, that is, from confluences of bifurcations. Skin color. Skin color was measured on the forehead with the CIE 'L,a,b' color scale.'? The "L-value" (lightness) was used to evaluate the degree of skin darkness and the "a-value" (redness-greenness) was used to evaluate erythema.
Image analysis was performed with a commercially available system that included the VAX 8350 computer (Digital Equipment Inc.) and Gould IP8500 Image Processor (Viacom Inc.). An Eikonix 78/99 Digital Camera (Eiconic, Inc.) was used to convert photographic slides to digital images.
Skin elasticity Elasticity was measured with a tonometer placed over the left temple lateral to the commissure of the eye.11 This instrument consists of a Linear Variable Differential Transformer to measure displacement, an electric solenoid to create a lifting force, and a vacuum to ensure intimate attachment to the skin (Fig. 2). The subjects were supine with their heads resting securely on the padded instrument board. A force of 12.5 gm was applied for 0.25 seconds and abruptly removed. The computer automatically followed the skin deformation recovery period of similar length. The total cycle time was 0.50 seconds. Skin
Journal of the American Academy of Dermatology
754 Warren et al.
YH
Fig. 4. Hematoxylin-and-Lee staining of skin biopsy specimens. Skin biopsy specimens were prepared as described in Material and Methods. Photomicrographs show different degrees of elastosis as described in Table I. Younger low exposure (grade 2) (YL), younger high exposure (grade 4) (YH), older low exposure (grade 6) (OL), and older high exposure (grade 8) (OH). Der, Dermis; E, elastin; Ep, epidermis; Gr, grenz zone. Bar = 100 ~m. elasticity was defined as the percent of recovery after vertical deformation. (The percentage recovery of loose skin is lower than that of firm skin.) This arrangement yields information essentially comparable to the levarometry of Dikstein et al. l 3 Our method, however, used a computercontrolled force that provides fast and highly controlled skin lift/relaxation cycles. Reproducibility of measurements was within 4%.
Perceived age The high-resolution color photographs were examined as illustrated in Fig. 3. A panel of 24 women judges, 20 to 60 years old, evaluated 35 moo slides made from the 2 JA format negative film. The slides were projected into a high-resolution, life-size rear-projection screen. The judges were seated at a computer console 1.5 m from the screen and were asked: "How old does this woman look?" to the nearest year. The photographs were randomly presented to each judge.
Histologic examination A 4 mm punch biopsy specimen was obtained from the malar eminence of the face under local lidocaine anes-
thesia. Each specimen was fixed in phosphate-buffered formalin (pH 7.0) and stored at 4 C. Each specimen was coded to ensure anonymity. The specimens were embedded in glycol methacrylate with the JB-4 Embedding Kit (Polysciences, Inc.), sectioned at 2.0 J,Lm thickness, and stained with hematoxylin and Lee, a mixture of Harris' hematoxylin, Lee's methylene blue, and Lee's basic fuchsin.'! This technique enables visualization of elastin (red or pink) and keratinocyte cytology (Fig. 4). PicroSirius Red F3BA15 was also used for estimation of collagen (Fig. 5). Grading of actinic elastosis. Hematoxylin-and-Leestained sections were examined by one of us (W. M.) and graded on a 9-point scale (Table I). The scale was developed after examination of histologic slides without implicit knowledge of subjects and exposure groups. The scale reflects the degree of elastotic degeneration in the dermal matrix. It is based on multiple markers that occur concurrently during the process of elastosis. The slides were reviewed and graded multiple times with insignificant deviation in elastotic grade. A grade of I indicates absence of elastosis (normal skin) and grade 9 designates an end-stage or severe elastotic degeneration. Fig. 4 illustrates grades of 2, 4, 6, and 8. 0
Volume 25 Number 5, Part I November 1991
Quantitative measures ofskin aging 755
Fig. 5. Picro-Sirius Red-stained sections of skin biopsy specimens. Skin biopsy specimens
were prepared as described in Material and Methods. Photomicrographs represent the four age and solar exposure groups. Younger low exposure (YL), younger high exposure (YH), older low exposure (OL), and older high exposure (OH). C, Collagen. Der, dermis; Ep, epidermis. Bar = 100ILm.
Image analysis of actinic elastosis. We used the IBAS 2000 (Zeiss) to measure the area fraction of the elastotic material in hematoxylin-and-Lee-stained tissues; collagenous material was measured on Picro-Sirius Redstained samples. The thickness of the epidermis and of the subepidermal grenz zone was measured on the hematoxylin-and-Lee-stained tissues. Three microscopic fields, free of skin appendages, were selected for analysis in each stained specimen. The optical part of the IBAS analyzer consistsof a Universal Zeiss UEM microscope with a 16X Planapo objective.Microscopic images are converted into electronic images with a Sony CCD3000 color camera. The IBAS electronicimages contained 512 X 512 pixels that represented approximately 270 X 270 ILm of the original histologic section. The analyzed images were continuously displayed on a high-resolution IBAS color monitor to allow an operator to control the process of analysis. A brief description of algorithms used to measure hematoxylin-and-Lee- and Piero-Sirius Redstained samples follows. Evaluation ofhematoxylin-and-Lee-stainedsamples. The tissue sections were positioned so that the epidermis always occupied the left 25% of the image. The process
Table I. Grading scale for histologic elastosis Grade
Comments
I 2 3
No increase in elastin fiber width Increase in elastin fibers, which are thinner Increase in elastin fiber thickeness with greater amount of fibers more curled and branched. Some macrophages have phagocytosed elastin fibers. Above changes intensified Upper bed of amorphous fibers and lower one of thin, tangled, and dense fibers As above, intensified More disintegration of fibers Thiek amorphous mass in upper bed Mostly amorphous upper bed with nearly total elimination of collagen.
4 5 6 7 8 9
began with the digitization and shading correction of the microscopic color images. The algorithm consisted of two steps. (1) The width of the grenz zone was determined as the distance between the dermoepidermal junction and the edge of the densely stained elastotic material in the
Journal of the American Academy of Derm atology
756 Warren et al.
used to extract the epidermis. The remaining image area was considered to be the dermis and was measured. The green component was selected to identify collagen. The contrast of the stained material was enhanced by normalizing the gray levelhistogram to the full 0-255 range. A fixed threshold levelwas used to segment the gray level image into binary form and the area occupied by the collagen in the dermis was computed. Viable epidermal cell layers. Sections stained with Picro-Sirius-Red were used to visualize the viable epidermis and the granular cell layers (Fig. 5). The number of the viable celllayers in the epidermis (basal, spinous, and granular cell layers) was quantified with the stratum corneum used as the upper reference point and the dermis as the lower reference point. Four different sites per specimen were selected for counting.
Statistical analysis Data were analyzed statistically with analysis of variance (ANOVA and GLM SAS procedure); the expert grading was analyzed by means of Wilcoxon rank statistics (NPARIWAY SAS procedure) .
RESULTS
Fig. 6. Image analysis processing of hematoxylin-and-
Lee-stained skin sections. Skin biopsy specimens were prepared as described in Materials and Methods. A, Microscopic images were first digitized and grenz zone identified. B, Elastin component in the dermis is identified. dermis . A 5 pixel wide band parallel to the junction was sequentially shifted until the magenta-stained elastotic material comprised 20% of this band. The distance between the derrnoepidermal junction and the trailing edge was measured (Fig. 6, A). (2) To measure the elastotic material, the magenta-stained elastotic material in the dermis was enhanced by subtracting the green component from the blue. The resulting image was presented to the operator, who established the threshold level necessary to include all the elastotic material. The total area of the magenta-stained elastin was calculated (Fig. 6, B). Evaluation of'Picro-Sirius-Red-stained samples. The sample was positioned to maximize the visibility of the dermis. The operator oriented the field in such a way that the dermoepiderrnal junction was vertical in the left 25% of the field. The red component was used to identify the epidermis . A linear histogram extension and a directional lowpass filter was applied to the red image component to further enhance the contrast. A fixed threshold level was
Estimation of perceived age. The answers to the question: "How old does this woman look?" are presented in Table II, which shows that the amount of solar exposure sharply intensifiedthe appearance of aging in the older group. By contrast, in the younger group, the amount of sunlight had no effect on perceived age. The perceived age for both groups was older than the chronologie age. Quantification of facial furrows. Table II shows that the older group had significantly more wrinkles than the younger group (p < 0.005). Solar exposure greatly increased the total wrinkle length only in the older group (p < 0.05). The younger group was not affected by sunlight. When the crow's feet periorbital area was specifically examined, only the older group was affected by the amount of solar exposure (p < 0.05). Skin color. No differences in skin color with respect to skin redness and darkness were found among the four groups. Skin elasticity. Table III shows that the differences were age dependent. Solar exposure had no effect. Elastotic grading. Table IV shows that the older group had a significantly higher degree of elastosis than the younger one. Within the younger age group, the amount of sunlight had no impact on
Volume 25 Number 5, Part I November 1991
Quantitative measures ofskin aging 757
Table Il, Effect of sun exposure on perceived age and wrinkle length Exposure
Younger group Low (n = 10) High (n::::9) Older group Low (n =- 9) High (n:::: 13)
Perceived age* (yr) (mean ± SE)
Chronologie age (yr) (mean ± SE)
Total wrinkle lengtht (em ± SE)
Crow's feet areaf (em ± SE)
30.9 ± 1.3 30.8 ± 1.3
28.5 ± 0.6 27.1 ± 0.6
27.4 ± 2.8 22.6 ± 2.6
5.6 ± .0.9 4.9 ± 1.0
53.7 ± 1.3 58.2 ± 1.1
47.8 ± 0.7 47.6 ± 0.4
53.5 ± 5.8 75.7 ± 6.8
12.0 ± 1.3t 17.7 ± 2.0
*Perceived age in older high exposure group is greater than in low exposure group (p < 0.01). tOlder group had more wrinkles than younger group (p < 0.005). tWrinkle length in older high exposure group was greater than in low exposure group (p < 0.05).
elastosis. However, if exposure was tested within the older age groups, the difference was directional (p < 0.1). Assessment of elastosis and collagen. The results of Table IV show that elastin content increased with both age and solar exposure. However, the difference induced by solar exposure was much greater in the older group (p < 0.05). There was a progressive loss of collagen with increasing sun exposure and age. The difference from solar exposure was directionally greater in the older group (p < 0.1). The amount of elastin and collagen was inversely related and the correlation between the two was high (Fig. 7). Grenz zone and epidermal thickness. Most older subjects in either the low-exposure group (8 of 9) or the high-exposure group (12 of 13) had a grenz zone. Sun exposure induced a directional narrowing of the zone from 32.1 to 23.4 ,urn (p < 0.1). Only one subject in each of the younger groups had an identifiable grenz zone. There were no meaningful changes in epidermal thickness related to solar exposure. DISCUSSION Although numerous studies have shown that age can affect a variety of skin features, this study was specifically designed to see whether sunlight aggravated these features from a small but controlled sampling from the same sunny area, and focusing only on the face. Our highly sensitive photographic techniques confirmed what has been intuitively inferred for a long time, that sunlight is a prominent factor for the appearance of premature aging, independent of other factors. By middle age, the subjects who had received more sun exposure appeared older. The extent of dermal degenerative change in this
Table III. Effect of age and solar exposure on wrinkle length Group
Younger group Low (n = 10) High (n = 9) Older group Low (n = 9)
High (n = 13)
Total wrinkle length" (em ± SE)
Crow's feet area * (em ± SE)
27.4 ± 2.8 22.6 ± 2.6
5.6 ± 0.9 4.9 ± 1.0
53.5 ± 5.8 75.7 ± 6.8
12.0 ± L3t 17.7±2.0
*Older group had more wrinkles than younger group (p < 0.005). [Wrinkle length in older high exposure group was greater than in low exposure group (p < 0.05).
group also correlated with premature aging. Although occupations, other than those directly related to outdoor activities (which were included in the questionnaire), might playa role in wrinkle development (i.e., airline pilot exposure to gamma irradiation or baker's exposure to infrared), it was not apparent or a criterion for this study. Similarly, other experiential factors, such as wind exposure, were not discriminating criteria in this study. One benefit of our study design is that it allows a determination of the relation between subjective and objective measures of aging. For example, we found that there was a high correlation between perceived age and facial wrinkles (Fig. 8, A). The assessment of elastosis was also highly correlated with perceived age (Fig. 8, B). Next, elastin and, more significantly, the quantity of collagen, as measured by morphometric image analysis, especially showed tight correlations to elastosis and to perceived age (Fig. 8, C). This is intriguing because the histologic assessment was based on the hematoxylin-and-Lee-stained sections, whereas collagen quantity was assessed with
Journal of the American Academy of Derma tology
758 Warren et al.
70
....
• •
High Exposure Low Exposure
r::::- 0.66 p
•
•
10
,. ..
.. .
•
O.....- -......-----''------''---_--lL.-_~ o 15 30 45 60 Percent Elastin in Dermis
Fig. 7. Correlation plot between elastin and collagen. Relative abundanceof elastin and collagen was quantified by means of image analysis as described in Materialsand Methods. This plot is a correlation between the abundance of elastin in the hematoxylin-andLee-stained sections and theabundance of collagen in thePicro-Sirius Red-stainedsections.
Table IV. Effect of age and solar exposure on histologic elastosis, dermal elastin, and collagen Group
Younger group Low (n = 10) High (n =9) Older group" Low (n= 9) High en =13)
Elastosis grade( ± SE)*
% Elastin( ± SE)t
% Collagen (± SE)t
3.1 ± 0.3 3.4 ± 0.7
20,4 ± 3.3 23.5 ± 3.4
61.7 ± 4.2 58.0 ± 4.5
6.0 ± 0.5:1: 6.8 ± 0.4
31.7 ± 3.4§ 39.9 ± 2.9
40.2 ± 4.511 31.5 ± 3.7
'Older subjects had greaterelastosis than younger subjects (p < 0.001). tOlder subjects had more elastin and less collagen than younger subjects (p < 0.005). :t:Elastosis wasworse in olderhighexposure groupthan in olderlow exposure group (p < 0.1). §Older highexposure grouphadmoreelastin than the olderlowexposure group (p < 0.05). IIOlder highexposure group had less collagen than the olderlowexposuregroup (p < 0.1).
thePiero-Sirius-Red-stained sections. Thissuggests that instrument measurements of collagen quantity provide the strongest indicator of elastosis. Earlier studies by Uitto et al. 16 also demonstrated a direct relation between morphometric quantification of elastic fibers in the dennis andbiochemical quantification of elastin. The grenz zone refers to a subepidermal band of normal dermis. It isthought tobe a siteof continual
dermalrepair and consists of normal collagen fibers. With our histochemical methods, the grenz zone becomes visually apparent only after there is sufficientelastotic damage. Thisis strikingly clear in the oldergroup. With progressive elastosis, the zoneappears to become thinner. Clinicians have proposed that actinic elastosis can be prevented and possibly reversed if the solar assault is stopped (e.g., through the use of sunscreens). In their study conducted
Volume 25 Number 5, Part 1 November 1991
Quantitative measures ofskin aging 759
•
•
.... ..-:,
65
leA .
55
•
HighExposure
•
Law Exposure
r=O.85
p
A 70 A
• 60
r = 0.79 p<0.OO01
••
•
•
GI
ce
'e ...... GIll)
40
•
.
I 50
> ..
--
•
r=-O.90 p
•
22
.
E
A
~60
•
.1;
m
c:
~ 8....
t
~ 20
•
=40
40
. t
30
....
•
e55
•
•
.... 20 2
B
3
4
5
6
7
8
0
9
Histological Grade
Hlgh~"
A Low ExpoIlU ..
•
'Qj u>.
...GI D.
120
80
Facial Wrinkles (cent'meters)
• •
Low Exposure High Exposure
01
a a
C
1
3 5 7 Histological Grade
• • 9
Fig. 8. A, Correlationplot between perceived age and totalfacial wrinkles. B, Correlation plot betweenperceived age and elastosis. C, Correlation plotbetween collagen andelastosis.
more than 25 years ago, Gerstein and Freeman'? showed that actinically damaged neck skin would improve when transplanted to a solar protected area. The grenz zone became wider with the deposition of
normal histochemically stained collagen. The epidermis also appeared normal with the reappearance of rete ridges and restoration of more uniform epidermal differentiation.
760 Warren et al.
The treatment of sun-damaged skin with chemical peels or dermabrasion has been used for several years.IS-20 Topical treatmentwithtretinoinhas commandedmuchpublic attention andindicates the feasibility of a pharmacologic approach to partial reversalofphotodamage. 2.1, 22Asprevention andrepair of actinically damaged skin becomes evenmoreemphasized in the media and scientific literature the quantitative methods wehavedescribed willbe useful for measuring the benefits of strategiesaimed at avoiding the appearance of premature aging. We are indebted to F. E. Dunlop, MD, of Argus Re-
search,Inc., and R. P. Block, R. M. Cook, W. H. Elsnau, and C. R. Hawley for their expert technical assistance.
REFERENCES 1. Kligman LH, Kligman AM. Photoaging, In: Fitzpatrick TB, Eisen AZ, Wolff K, et al, eds, Dermatologyin general medicine. 3rd ed.New York: McGraw-Hill, 1986:1470-5. 2. Gilchrest BA.Skinand aging processes. BocaRaton: CRC Press, 1983. 3. Holman COl, Armstrong BK, Evans PR, et al. Relationship of solar keratosis and history of skincancer to objectivemeasures ofactinic skindamage. Br J Dermatol1984; 110:129-38. 4. MontagnaW, Carlisle K. Structural changes in aged human skin. J Invest DermatoI1979;73:47-53. 5. Leveque JL, CorcuffP,de Rigal J, et al. In vivostudies of the evolution ofphysical properties of the human skin with age. Int J DermatoI1984;23:322-9. 6. Daly CH, Odland GF. Age-related changes in the mechanical properties ofskin. J InvestDermatoI1979;73:84-7. 7. DittoJ, MatsuokaLY, Kornberg RL. Elastic fibersin cutaneous elastosis. In:Rudolph R, ed,The biological basisof cosmetic surgery. St. Louis: CV Mosby, 1986:307-38.
Journal of the American Academy of Dermatology 8. Mera SL, Lovell CR, Russell Jones R, et al. Elastic fibers in normaland sun-damaged skin:an immunohistochemical study. Br J DermatoI1987;1l7:21-7. 9. KligmanLH, KligmanAM. The nature of photoaging: its prevention and repair. Photodermatology 1986;3:215-27. 10. Cripps OJ. Natural and artificial photoprotection. J Invest Oermatol 1981;77:154-7. 11. Gartstein Y, Shaya SA. Image analysis of facial skin features. Proc Int Soc Optic Eng 1986;626:284-8. 12. Pratt WK. Digital image processing. New York: John Wiley & Sons, 1978:50-90. 13. DiksteinS, Hartzshtark A, Bercovici P. The dependenceof low-pressure indentation,slackness, and surface pH on age in forehead skin of women. J Soc Cosmet Chem 1984; 35:221-8. 14. Montagna W, Kirchner S, Carlisle K. Histology of sundamaged human skin. J AM ACAD OERMATOL 1989; 21:907-18. 15. Sweat F, Puchtler H, RosenthalSI. Sirius red F3BA as a stain for connective tissue. Arch Pathol 1964;78:69-72. 16. Uitto J, Paul L, Brockley K, et al. Elastin fibers in human skin: quantitation of elasticfibers by computerized digital image analyses and determination of elastin by radioimmunoassay of desmosine. Lab Invest 1983;49:499-505. 17. Gerstein W, Freeman RG. Transplantation of actinicalty damaged skin. J Invest OermatolI963;41:445-50. 18. Ayres S III, Wilson JW, Luikart R II. Dermal changes following dermabrasion. Arch Dermatol 1959;79:553-68. 19. Ayres S Ill. Dermal changes following application of chemical cauterants to aging skin: superficial chemosurgery. Arch Oermatol 1960;82:578-85. 20. Stegman S1.A comparative histological studyof the effects of three peeling agents and dermabrasion of normal and sundamagedskin. Aesth Plast Surg 1982;6:123-5. 21. KligmanAM, GroveGL, HiroseR, et al. Topicaltretinoin for photoaged skin. J AMACAD DERMATOL 1986;15:83659. 22. WeissJS, EllisCN, HeadingtonJT, et al. Topicaltretinoin improves photoaged skin. JAMA 1988;259:527-32.