2.1.

Subjects

The influence of age was investigated on 33 healthy volunteers aged between 19 and 74 years with skin type II, according to the Fitzpatrick classification.³⁰

For the investigation of the skin type dependency, 33 volunteers were recruited. The volunteers were aged between 25 and 35 years and of skin types I–III. For each skin type, 11 subjects were included.

To investigate the correlation between the different skin carotenoid sensors and kinetic changes due to the intake of a carotenoid-rich supplement, 30 healthy volunteers (13 male, 17 female, mean age 36.3 years, skin types I to III) were included in the study.

Excluded from these studies were with skin diseases, admission to an institution due to an administrative or judicial order (according to § 29 MPG), known drug addiction or alcoholism, and expectant or lactating women. In the uptake study, only subjects with skin carotenoid values below 6 were enrolled. The investigations were performed in accordance with the ethical guidelines of the Declaration of Helsinki and had been approved by the local Ethics Committee prior to starting the study. All volunteers gave their informed written consent.

2.2.

Resonance Raman Spectroscopy

The cutaneous carotenoid concentration was determined by RRS.¹⁸ This noninvasive method was applied on the palm of the right hand, since carotenoid concentrations are particularly high in this area due to the thick SC. The carotenoids were resonantly excited using an argon laser operating at 488 nm. The intensity of the strong carbon double-bond stretch vibration Raman line, measured at $1525{\mathrm{cm}}^{-1}$, was used for cutaneous carotenoid detection. Every measurement was conducted five times and the mean value was determined. The data are presented in arbitrary units giving values from 0 to 10. Zero corresponds to values below the threshold and 10 is the highest value observed, corresponding to $\sim 1\mathrm{nmol}\text{carotenoids}/\mathrm{g}$ skin.¹⁸

2.3.

CaroLED Carotenoid Concentration Sensor

The CaroLED sensor (Laser- und Medizin-Technologie Berlin, LMTB, Germany, Fig. 1) is intended to quantify the carotenoid concentration in human skin. It was specifically calibrated for measurements on the palm of the hand. In previous measurements with RRS and CaroLED on the same subjects, a calibration data set was acquired. After preprocessing of the raw CaroLED data, a multivariate calibration based on partial least squares regression was performed. This calibration with RRS as the reference method gives carotenoid levels on a relative scale. The data acquired in the study presented in this publication were not part of the calibration procedure.

Fig. 1

Image of CaroLED sensor.

The underlying measuring principle of CaroLED is based on SRRS.^31–34 Five light-emitting diodes (LED) of different wavelengths in the visible and near-infrared region are illuminating the skin through a measurement window. While light is propagating through the individual skin layers, the spectral signature of carotenoids and other skin constituents is obtained.³⁵ Four photodiodes, each of which is placed at a different lateral distance to the illumination site, measure the spectral signature at different depth sensitivities. Combining this information, the influence of skin variance regarding light scattering and disturbing chromophores such as hemoglobin or melanin on the carotenoid measurement is minimized. The measurements were performed five times on the palm of the right hand.

2.4.

Multiple Spatially Resolved Reflection Spectroscopy Sensor

The multiple spatially resolved reflection spectroscopy (MSRRS) sensor (Opsolution, Kassel, Germany) is an optical sensor based on MSRRS which is commercially available under the trade name biozoom (model Biozoom Portable). The MSRRS sensor uses several differently positioned light sources and light detectors (Fig. 2), exhibiting different distances between light source and detector and different angles for irradiation and detection. The measurement time for one measurement takes $\simeq 15\mathrm{s}$.

Fig. 2

Photograph of the MSRRS sensor.

With the MSRRS sensor, it is possible to get reliable information, even if the sample is inhomogeneous or shows a gradient in substance concentrations. This makes the MSRRS sensor less sensitive to errors caused by the position of the measurement, as it averages the properties of the sample over a larger area.

The LED light sources cover a spectral range from $\sim 350$ to 1000 nm wavelength in 16 steps providing 118 light emitters. The sensor picks up the backscattered light at a total number of 152 light-sensitive areas. This combination results in almost 18,000 raw data values that are picked up several times during one measurement at a sensor surface area of $20\mathrm{mm}\times 20\mathrm{mm}$. The sensor was calibrated to the RRS using more than 3000 measurements.³⁶ The calibration was computed by using a genetic algorithm on data pairs of preprocessed MSRRS raw data and RRS data to determine structure and parameters of the evaluation algorithm. Since the MSRR sensor is finally intended for use under everyday life conditions, the MSRRS data were collected in a previous calibration study on different volunteers, whereby the ambient temperatures and surface contact parameters were artificially varied. Therefore, the measurements in the present study are the blind test for the MSRRS sensor. The measurements in this study were performed five times on the palm of the right hand.

2.5.

Blood Carotenoids

For determining the total blood carotenoids, ethylenediaminetetraacetic acid (EDTA) blood was sampled and centrifuged to separate the cells from plasma. An amount of $400\text{-}\mu \mathrm{l}$ EDTA plasma was taken to analyze the blood carotenoid content using a mobile device iCheck™ (BioAnalyt, Germany).

2.6.

Skin Parameters Melanin and Redness

The skin parameters such as melanin and redness were measured using the Mexameter^® MX 18: (Courage + Khazaka electronic GmbH, Cologne, Germany). The device measures melanin and hemoglobin (erythema or redness) based on RS.

2.7.

Study Protocols

For the age- and skin-type studies, the volunteers were measured at every visit with all three skin sensors and blood was sampled afterward. Furthermore, a questionnaire was completed on lifestyle, nutrition, and stress conditions. Before the measurements started, the subjects were asked to acclimate for 5 to 10 min. In addition, they had been instructed not to use any skin care products 24 h before investigation.

For the uptake study, the volunteers were investigated at five visits. At every visit, the volunteers were measured five times with each of the three skin carotenoid sensors. Data analysis was based on the mean values of the repeated measurements. At the first visit, exclusively volunteers exhibiting values below 6 were screened using the RSS device as specifically subjects with a low to medium skin carotenoid status were in the focus of these investigations. Subsequently, the same volunteers were measured with all three cutaneous carotenoid sensors on the right palm. Afterward, their blood was sampled to determine the blood carotenoid concentrations and they completed the questionnaire. The same procedure was performed after 6, 12, and 14 weeks of intake as well as 3 weeks after discontinuance of the supplementation.

2.8.

Vegetable Extract

During the uptake study, the volunteers ingested three capsules of a curly kale/red pepper extract once a day (provided by BioActive Food, Bad Segeberg, Germany), hereinafter called verum. One capsule contained the following substances: $400\mu \mathrm{g}$ lutein, $70\mu \mathrm{g}$ beta-carotene, $30\mu \mathrm{g}$ lycopene, and $20\mu \mathrm{g}$ zeaxanthin. As a control, a placebo capsule containing no antioxidants was given.

2.9.

Questionnaire

In the study, a questionnaire was prepared and subsequently completed by the volunteers to monitor their nutritional habits and lifestyle patterns, such as alcohol consumption, smoking, sun exposure, and stress.³⁷

Dietary intakes of fruits and vegetables were examined by self-assessment based on a food frequency questionnaire with two separate food lists containing 13 common fruits and 13 common vegetables that had to be marked in a frequency category (“rarely,” “$\ge \text{once a month}$,” “$\ge \text{once a week}$,” and “$\ge 4$ times a week”).

2.10.

Statistical Analysis

A descriptive explorative data analysis was performed using SPSS statistics (IBM^® SPSS^® Statistics, version 19, Inc. Chicago, Illinois). Since the obtained data were not normally distributed, nonparametric tests were performed. For independent data, the Mann–Whitney U test was applied, and for related data, the Wilcoxon test was applied. Values of $p\le 0.05$ were considered to indicate a difference. To evaluate the significance over the entire measurement period, generalized estimating equation was performed; a $p$ value $\le 0.05$ was considered as significant.

Correlation data were determined using Excel by calculating the Pearson correlation coefficients.

3.1.

Influence of Age

To investigate the influence of the age, 33 volunteers of skin type II aged from 19 to 79 years, were investigated with all three skin sensors, whereas blood samples were taken to analyze the total amount of blood carotenoids. Three age groups of 11 subjects, each, were differentiated: group one consisting of volunteers below 30 years, group two comprised volunteers aged between 31 and 55 years, and the volunteers of the last group were older than 55 years. The mean values and standard deviations determined per group and detection method are shown in Fig. 3. No significant differences in skin or blood carotenoids were found among the three age groups.

Fig. 3

Mean values and SEM of skin and blood carotenoids for three different age groups, 19 to 30 years, 30 to 55 years, and 55 to 79 years, $n=11$ for each group, skin type II. Values for skin carotenoids are shown in a.u. for blood carotenoids in $\mathrm{mg}/\mathrm{L}$.

3.2.

Influence of Skin Types I to III

The influence of the skin type was investigated by including 33 volunteers of skin type I ($n=11$), skin type II ($n=11$), and skin type III ($n=11$), respectively. The volunteers were aged $30\pm 5$ years. Blood was measured as a reference because the blood values should not be influenced by the skin type. The mean values and standard deviations for the skin and blood carotenoids per group and detection method are shown in Fig. 4.

Fig. 4

Skin and blood carotenoids ($\text{mean}\pm \mathrm{SEM}$) for skin types I, II, and III and mean age $30\pm 5$. Skin carotenoids are shown in a.u., blood carotenoids in $\mathrm{mg}/\mathrm{L}$, $n=11$ for each skin type.

No statistically significant differences in skin and blood carotenoids were found for the different skin types.

Furthermore, the melanin content and the redness of the skin were measured for selected volunteers (Fig. 5).

Fig. 5

Melanin and redness values ($\text{mean}\pm \mathrm{SEM}$) on palm and inner forearm of the volunteers of different skin types. Number of volunteers, skin type $\mathrm{I}=7$, skin type $\mathrm{II}=5$, and skin type $\mathrm{III}=6$.

The melanin content of the palm increased with increasing skin type but this is more pronounced on the inner forearm. The redness at the palm shows higher values than on the arm. At the forearm, a significant difference between skin types I and II could be observed for redness ($p=0.003$), while for melanin a trend was observed ($p=0.072$). Regarding differences between skin types II and III, again only at the forearm the values showed a trend for melanin ($p=0.052$). However, for all data in Fig. 5, significant differences between skin types I and III appeared ($p<0.05$) with the exception of redness at the palm; here only a trend was visible ($p=0.051$).

3.3.

Kinetics of Cutaneous Carotenoids

The uptake study was performed on 30 volunteers, 13 male, 17 female, skin types I, II, and III, mean aged 36.3 years (15 verum and 15 placebo) from October 2014 to March 2015. After visit 1, the volunteers started to ingest the vegetable extract or the placebo sample for 14 weeks. The skin carotenoids were measured with all three skin sensors on the palm of the right hand, the blood carotenoids were determined and the questionnaire was completed before, 6, 12, and 14 weeks after supplementation and 3 weeks after the discontinuance of supplementation to see if either skin or blood carotenoids or both had decayed.

To demonstrate the kinetics of the values, the blood and skin carotenoid concentrations relative to the initial values are shown with their mean and standard deviations in Figs. 6 and 7.

Fig. 6

Mean and SEM of blood carotenoids relative to initial values of individual measurements.

Fig. 7

Mean and SEM of skin carotenoids relative to initial values of the placebo and verum groups measured by the two SRRS and the RRS systems.

The blood values increased already after 6 weeks of intake to the maximum level (55% increase) and remained almost stable until the intake was discontinued. After stopping the intake, the values decreased almost to the initial value. The placebo values also increased by 20% and decreased after discontinuing the intake.

In contrast to the blood values, the skin values still increased for a period between 6 and 12 weeks and remained almost stable for the next 2 weeks of intake. This study illustrated for the first time that the regular intake of a moderate dose of antioxidants leads to stable skin carotenoid values. When the intake was discontinued, the values decreased although at a smaller rate compared to the blood values. The maximum increase of the skin carotenoids was dependent on the method used. With 48%, RRS exhibits higher changes. Both reflectance measurements reached $\sim 26\%$ increase. As observed in blood, an increase in cutaneous carotenoid values was also obtained in the placebo group.

To investigate the significance of difference between placebo and verum, the Mann–Whitney U test was applied for all time points and sensors. The $p$ values are given in Table 1.

Table 1

Calculated p values for all time points and sensors between placebo and verum using the Mann–Whitney U test; in brackets, the significance level is clarified.

The questionnaires were analyzed for changes in nutrition or stress factors over the duration of the study. Significant changes could be found neither in the verum nor in the placebo group.

3.4.

Correlations Between the Skin and Blood Carotenoid Values

To validate the sensors, the correlation of the carotenoid values between the two reflectance-spectroscopy-based skin sensors and the RRS device, and the correlation to the blood values of all three noninvasive sensors were calculated using Excel (Table 2). As can be seen in Fig. 7, the absolute values given by the sensors are not identical. The correlation between the sensors was used for comparison purposes to neutralize effects caused by the sensors scaling the displayed output values differently. The MSRRS and CaroLed sensors are scaling the output so that an increase of the carotenoid levels results in a smaller increase of the displayed values compared to the RRS measurement. There is also an offset among the different sensors. The different scaling also explains the different absolute values seen in Fig. 6. Offsets and constant scaling factors are eliminated by using correlation values providing an objective comparison.

Table 2

Pearson correlation coefficient R for the data of the uptake study (n=150).

The comparison of the carotenoid values obtained using reflectance-based sensors and the RRS system demonstrate a good correlation. The MSRRS data from the first visit show significantly lower correlation than visits two to five, leading to the assumption that in this series an error occurred during the data collection. For the MSRRS and the CaroLED sensor, the study was a valid blind test, as the data of the study were not used for the calibration of the algorithms.

Based on the uptake study, the reproducibility of measurements conducted with the three carotenoid sensors was compared by calculating the mean standard deviation of five measurements performed at every visit and are shown in Table 3.

Table 3

Calculated mean standard deviations of all five repetition skin carotenoid measurements for the applied devices during the uptake study.

The results show that the reproducibility of RRS is slightly below that of the MSRRS and CaroLED sensors.