2.1.

Study Design and Setting

The study was performed at the Department of Dermatology and Allergy, Charité-Universitätsmedizin Berlin (52.3°N, 13.2°E) in the Clinical Research Center for Hair and Skin Science. Female volunteers of three distinct age groups were recruited for this explorative study between January and April 2014. All subjects gave their written informed consent prior to the visit and the study was conducted in accordance with the principles of the current version of the declaration of Helsinki.

2.2.

Sample

Skin healthy volunteers were enrolled to have eight subjects in each of three age groups: 30 to 40, 50 to 60, and 70 to 80 years. The chosen sample was considered to display different signs of aging because chronological age is the main predictor for skin aging.²⁷ Inclusion criteria were among others no cosmetic or rejuvenation procedures 10 years prior to the study, no hormonal replacement therapy in the past two years, no extreme UV exposure eight weeks before the visit, and nonsmoking status. Subjects with a history of certain medications (e.g., photosensitizing agents, systemic, or topical corticosteroids) or a history of dermatologic diseases were not included in the study. Subjects were asked to avoid water contact at the measurement sites 12 h prior to the measurements and not to apply any cosmetic products on their arms. All the measurements were noninvasive and were performed in similar and standardized conditions.

2.3.

Measurement Sites

The measurement fields had a size of $10\mathrm{cm}\times 5\mathrm{cm}$ and were located parallel to the long bones of the right upper extremity. On the DFA and VFA, the measurement fields were located halfway between the wrist and the cubital lines. On the UIA, it was halfway between the cubital line and the anterior axillar line.

2.4.

Skin-Color Measurements

Skin color was measured by reflectance spectrophotometric devices. The melanin- and erythema indices were obtained using the Mexameter MX18 (Courage & Khazaka, Cologne, Germany). This instrument uses narrow-band illumination on the skin at 568, 660, and 880 nm and parameters are based on the measured reflections.²⁸ $L*a*$, and $b*$ from the CIELab color space were measured with the broad spectrum Chromameter CM-700d (Konica Minolta, Osaka, Japan)²⁹ at D65 lighting standard. All the color measurements were conducted in triplicate per measurement field. The devices were removed after every single measurement and placed again on the skin surface with a distance of 2 cm.

2.5.

Measurement of Skin Dyspigmentation

Skin dyspigmentation was quantified by applying an image-processing technique described previously.¹⁷ Standardized clinical images were taken of the measurement fields using a Lab Imaging System (Canfield Scientific, Parsippany, New Jersey) for consistent and standardized lighting; coupled with an arm positioning device. To enhance image quality, the room was completely darkened during photoacquisition. After color calibration, using seven standardized color chips inside the image frame, the “cross-polarized” images were transformed into RBX^®-Brown images.³⁰ The RBX^® transformation is based on a spectrocolorimetric light transport model of the skin to compute information of melanin (brown) and hemoglobin (red) concentration and distribution. The RBX^® transform is computed from a training database of color-calibrated crosspolarized images. This database is comprised of subjects with various skin types to obtain a unique comparative solution that is independent of the skin type. The RBX^®-Brown values are relative to the expected mean pigmentation of the training database population.

Fig. 1

Measurement of skin dyspigmentation on DFA, VFA, and the UIA by image processing. (a) As captured, crosspolarized clinical images, with the measurement area outlined in white. (b) Images cropped to the measurement area. (c) RBX^®-Brown images of the measurement area. (d) Hyperpigmentation (blue) and hypopigmentation (yellow) detection inside the arm measurement areas from RBX^®-Brown image analysis.

The three parameters to quantify skin melanin inhomogeneity based on the RBX^®-Brown transformation using image analysis are the hyperpigmentation index,³⁰ the hypopigmentation index, and the overall pigmentation intensity (Fig. 1). In a first step, the expected average pigmentation values of the areas of interest were calculated. Second, the thresholds were set to define individual areas of hyper- and hypopigmentation, i.e., areas where the melanin concentration is higher and thereby the skin is darker and vice versa. The threshold parameters for this were calculated using the RBX^® training database and were applied after normalizing the image with respect to the subject’s individual background pigmentation value. The hyper- and hypopigmentation indices represent the proportion of darker or lighter areas of the entire measurement field relative to the background skin pigmentation of the subject in the RBX^®-Brown image. Index values can theoretically range from 0 (no hyper- or hypopigmentation) to 1 (maximum hyper- or hypopigmentation) within the measurement area. The overall pigmentation intensity, negatively correlating with the average melanin concentration, was computed as the average RBX^®-Brown value within the measurement area. This value can theoretically range from $-25.5$ to 26.4, representing dark to fair skin, respectively, although the typical range for human skin tone is much narrower.

Standardized clinical images were taken twice from the same measurement fields with 3 h in between to be able to investigate the reliability of the dyspigmentation parameters. Statistical analysis was conducted using the means of both measurements.

2.6.

Statistical Analysis

Data were analyzed descriptively by calculating means and standard deviations for each measurement per skin area for the whole sample and each group. Comparison of means between the age groups was conducted by one-way analysis of variance (ANOVA). An alpha-value of 0.05 (two sided) was used. Pearson’s correlation coefficients were calculated to estimate the association between pigmentation parameters and age. An $r$ smaller than $-0.3$ or higher than 0.3 was regarded as minimal strength of association.

To compare skin-color measurements and dyspigmentation indices between the measurement fields, paired sample $t$-tests were calculated, using an alpha-value of 0.05 (two sided).

To investigate the reliability of repeated measurements, intraclass correlation coefficients [ICCs (1,1)] were calculated for each variable within repeated measurements per field. Indices greater than 0.7 were considered as good and above 0.9 as excellent test–retest reliability.^31,32 Statistical analysis was performed using SPSS Version 20.

3.1.

Study Sample

Sample characteristics are displayed in Table 1. Fitzpatrick photo types and body mass indices of the subjects were comparable. Mean age of subjects was 33.5 years in the young, 55.4 years in the midaged, and 76.6 years in the older group.

Table 1

Sample characteristics.

3.2.

Skin-Color Measurements

Results are presented in Table 2. Mean erythema indices ranged from 149.6 (SD 43.4) at the UIA of the young group to 233.7 (SD 55.4) at the DFA of the midaged subjects. Correlations with age ranged between 0.01 and 0.18. Average melanin indices were lowest at the UIA of the midaged group with 129.8 (SD 23.3) and highest at the DFA of the old group with 177.1 (SD 36.7). Correlation coefficients ranged between 0.14 and 0.27. Mean $L*$ decreased with increasing age at all areas, e.g., from 68.68 (SD 2.0), to 67.35 (SD 3.4), and to 65.59 (SD 1.74) at the VFA. The negative associations with age ranged from $-0.30$ at the UIA to $-0.44$ at the VFA. Mean redness ($a*$) varied between 4.95 (SD 1.84) and 7.58 (SD 1.57) as measured on the UIA of the midaged and DFA of the older subjects, respectively. Means among age groups were comparable and associations with age were between 0.25 and 0.28. Yellowness ($b*$) values did not differ among age groups and correlations with age were between 0.08 and 0.19. There were no statistically significant differences between age groups for any parameter.

Table 2

Means, standard deviations, associations, and reliability coefficients of skin color and dyspigmentation parameters.

Results of skin-color measurements were lowest at the UIA and highest at the DFA within the age groups for almost every parameter, e.g., $a*$ 4.95 (SD 1.84), 5.50 (SD 1.86), and 7.11 (SD 1.61) at the UIA, VFA, and DFA of the midaged subjects, respectively. The opposite was observed for $L*$, e.g., 69.21 (SD 2.17), 68.68 (SD 2.00), and 65.89 (SD 2.27) at the UIA, VFA, and DFA of the young group.

3.3.

Quantification of Dyspigmentation

Results are presented in Table 2. Hyperpigmentation indices increased with age, e.g., 0.06 (SD 0.01), 0.09 (SD 0.03, and 0.12 (SD 0.04) at the VFA of the young, midaged, and older subjects. Means were statistically significantly different at all skin areas. Correlation coefficients of hyperpigmentation indices and age were between 0.61 and 0.74. Hypopigmentation indices ranged from 0.00 to 0.14 and were significantly different between the age groups at all measurement fields. Correlations with age were between 0.676 and 0.740. Overall, pigmentation intensity decreased with increasing age, e.g., $-2.43$ (SD 0.50), $-2.73$ (SD 0.86), and $-3.27$ (SD 0.77) at the VFA. Differences between the age groups were statistically not significant. The strongest association with age was found on the VFA, $r=-0.48$.

3.4.

Differences Between Skin Areas

Results are presented in Table 3. Erythema indices differed significantly between all measurement fields ($p<0.001$). The mean difference was highest between the DFA and the UIA. Mean melanin indices differed statistically significantly between the DFA–UIA and the DFA–VFA ($p<0.001$). Luminance was significantly different between all the measurement fields and mean differences ranged from $-3.57$ to $-1.18$. Redness ($a*$) and yellowness ($b*$) were significantly different when comparing the DFA to the other measurement fields, $p<0.001$ in all cases. Hyper- and hypopigmentation indices were comparable between the VFA and the UIA and statistically significantly different when comparing DFA with the other areas ($p<0.001$). Overall pigmentation intensity was statistically different between the DFA and the UIA ($p<0.001$) and between the DFA with the VFA ($p<0.001$).

Table 3

Paired comparisons between skin areas.

3.5.

Reliability of Measurements

Results are shown in Table 2. Both erythema and melanin indices revealed good reliability between the three repeated measurements as indicated by the ICCs higher than 0.8. $L*a*$, and $b*$ values revealed ICCs of at least 0.85 at all skin areas and showed excellent reliability of $L*$ among the three repeated measurements. Dyspigmentation parameters based on image processing had good to excellent reliability. ICC values of the hyperpigmentation index were above 0.81 and reliability estimates of hypopigmentation index were greater than 0.91. Overall, pigmentation intensity at the DFA had the lowest ICC coefficients of 0.8.