Abstract
Background
Skin carotenoid measurements are emerging as a valid and reliable indicator of fruit and vegetable intake and carotenoid intake. However, little is known about the extent to which skin carotenoid responsivity to dietary changes differs based on demographic and physiologic characteristics.
Objectives
This study examined potential effect modifiers of skin carotenoid and plasma carotenoid responses to a carotenoid-rich juice intervention.
Methods
We leveraged data from 2 arms of a 3-site randomized controlled trial of a carotenoid-containing juice intervention (moderate dose = 6 ounces juice, 4 mg total carotenoids/d, high dose = 12 ounces juice, 8 mg total carotenoids/d) (n = 106) to examine effect modification by age, self-categorized race/ethnicity, biological sex, baseline body fat, body mass index, skin melanin, skin hemoglobin, skin hemoglobin saturation, skin coloration, sun exposure, and baseline intake of carotenoids from foods. Skin carotenoid concentrations were assessed using pressure-mediated reflection spectroscopy (Veggie Meter), and plasma carotenoid concentrations were measured using high-performance liquid chromatography.
Results
In bivariate analyses, among the high-dose group (8 mg/d), those of older age had lower skin carotenoid responsiveness than their younger counterparts, and those with greater hemoglobin saturation and lighter skin had higher skin carotenoid score responsiveness. In the moderate-dose group (4 mg/d), participants from one site had greater plasma carotenoid responsiveness than those from other sites. In multivariate analyses, participants with higher baseline skin carotenoids had smaller skin carotenoid responses to both moderate and high doses.
Conclusions
Changes in skin carotenoid scores in response to interventions to increase fruit and vegetable intake should be interpreted in the context of baseline skin carotenoid scores, but other variables (e.g., self-categorized race/ethnicity, biological sex, baseline body fat, body mass index, skin melanin, and sun exposure) do not significantly modify the effect of carotenoid intake on changes in skin carotenoid scores.
This trial was registered at clinicaltrials.gov as NCT04056624.
Keywords: skin carotenoids, plasma carotenoids, dietary intake, fruits and vegetables
Introduction
Fruit and vegetable (FV) intake is important for optimizing health and reducing risk of chronic diseases [1]. However, few Americans consume recommended amounts of FVs [[2], [3], [4]], and low-FV intake is particularly problematic in populations experiencing a disproportionate burden of diet-related chronic diseases [3,4]. To rigorously evaluate scalable programs and policies to increase population-level FV intake, valid, reliable, and noninvasive FV intake assessment tools are needed [5,6]. Skin carotenoid concentrations (SCCs) measured using noninvasive reflection spectroscopy (as with the Veggie Meter) are emerging as valid and reliable methods to objectively assess intake [[7], [8], [9], [10]]. Importantly, SCCs can be used to assess population carotenoid intake both cross-sectionally [7,9] and prospectively in relationship to an intervention [8,11].
In best practice evaluations of validity and application of dietary biomarker measures, “robustness” is a key consideration, in addition to other factors, such as plausibility, dose-response, time-response, and reliability [12]. Robustness refers to the validation of the biomarker in different subjects and study settings, as well as information about individual-level effect modifiers [12]. Examining potential modifiers of the association between dietary intake and biomarker outcomes is important for understanding biomarker robustness so that intervention effects can be accurately interpreted.
There are potential individual-level effect modifiers that may influence the accurate measurement and interpretation of skin carotenoid biomarkers as a marker of dietary change. Previous studies have shown that biological factors, such as age [[13], [14], [15], [16]], self-reported race/ethnicity [17], biological sex [[18], [19], [20], [21]], baseline adiposity, BMI [19,[22], [23], [24]], and baseline intake of carotenoids from foods [18] are all cross-sectionally associated with skin carotenoids and thus could potentially influence responsiveness of an individual’s plasma or skin carotenoids to carotenoid supplementation. However, none of these studies were longitudinal feeding trials to examine SCC responsiveness or how SCC changes with dietary interventions in the presence or absence of these individual-level characteristics. Furthermore, underlying genetic differences [[25], [26], [27], [28]] related to genetic ancestry could be associated with differential carotenoid metabolism, uptake, and biodistribution kinetics. In addition, it is unknown how chromophores in the skin, such as melanin and hemoglobin, affect the SCC response to carotenoid supplementation.
A limited number of studies have examined potential effect modification in response to dietary interventions. Obana et al. [29] found the magnitude of change in skin carotenoids measured by reflection spectroscopy (using the Veggie Meter) following a brief dietary intervention was greater among those with lower baseline skin carotenoid scores than those with higher baseline scores; thus, baseline skin carotenoid score could influence responsiveness of individuals’ skin carotenoids to changes in dietary carotenoid intake. Finally, we found significant study site differences in responsiveness to a carotenoid juice [8], which persisted even when controlling for demographic factors, suggesting that site-related factors, such as sun exposure or device calibration, could be related to SCC responsiveness to dietary changes.
As more nutrition researchers consider using skin carotenoid measurements to approximate change in FV intake, particularly in community- and behaviorally-focused intervention studies, there is a need to understand whether this measurement may be less responsive or unresponsive to change in carotenoid intake in some population groups. Therefore, the purposes of this study were to examine potential modifiers of the following: 1) skin carotenoid response and the 2) plasma carotenoid response to a carotenoid-rich juice intervention. We examined plasma carotenoid responsiveness because plasma carotenoid concentrations are the current criterion standard biomarker for carotenoid and FV intake [[30], [31], [32]]. Furthermore, because plasma carotenoids also increase with carotenoid supplementation [33,34], we sought to compare plasma carotenoid responsiveness to skin carotenoid responsiveness.
Methods
Participant recruitment and randomization: In brief, participants were recruited from 28 June, 2021 to 14 April, 2022 at 1 of 3 study sites (East Carolina University, Baylor College of Medicine, and the University of Minnesota) and randomly assigned to one of the following 6-week interventions: 1) control, negligible carotenoid group (Mott’s Apple Juice, Mott’s LLP), 2) moderate-dose (V8 orange–carrot juice, Campbell Soup Company, providing 4 mg total carotenoids/d), and 3) high-dose (V8 orange–carrot juice/d, providing 8.184 mg total carotenoids/d (2.804 mg α-carotene, 5.350 mg β-carotene, 0.011 mg lutein, 0.001 mg zeaxanthin, 0.017 mg lycopene). Participants attended study visits at baseline, 3 wk, and 6 wk. The recruitment goal was n = 156, with approximately 38 to 40 per racial/ethnic group (African American/Black, Asian, Hispanic, and White). For the purposes of this paper, analyses excluded the control group because there was no intervention effect and therefore, no reason to test for effect measure modification, leaving the final sample size for the current analyses n = 106. Compliance was high, with participants in the high-dose group consuming 97.8% of the study juice during the first 3 wk and 95.9% of the study juice during the second 3 wk [8]. Compliance in the moderate-dose and control groups was similarly high or higher [8]. Study design and methods are described in detail elsewhere [8]. This research was approved by the East Carolina University Institutional Review Board and was carried out in accordance with the Declaration of Helsinki for experiments involving humans.
Skin carotenoid measurement: the Veggie Meter, which uses pressure-mediated reflection spectroscopy to quantify skin carotenoids on a scale from 0 to 800, was used to assess SCC. A higher score generally indicates higher carotenoid intake [7]. Participants washed and dried their hands before placing the right index finger into the Veggie Meter. Three readings were obtained, and the average was used in analyses.
Plasma carotenoid measurement: plasma carotenoids, glucose, and cholesterol were measured from a fasting blood draw. Plasma carotenoid concentrations were measured by HPLC photodiode array detection. Trans-lutein, trans-zeaxanthin, cis-lutein/zeaxanthin, α-cryptoxanthin, β-cryptoxanthin, trans-lycopene, cis-lycopene, α-carotene, trans-β-carotene, and cis-β-carotene were quantified and summed for total carotenoids.
Skin coloration, hemoglobin: the Konica Minolta CM-700d Spectrophotometer was used, with a skin analysis software package (CM-SA) that estimates melanin index, hemoglobin index, hemoglobin saturation index, and skin coloration (where L∗ represents skin lightness, with higher values representing more lightness, and positive values of a∗ and b∗ representing degrees of redness and yellowness, respectively). All are potential modifiers as they contribute to skin coloration [35] and may influence the sensitivity of the Veggie Meter to detect changes in skin carotenoids.
Dietary data: baseline intake of carotenoids from foods was obtained from the Diet History Questionnaire (DHQ)-III. The DHQ-III is a validated questionnaire including 135 foods/beverages and 26 dietary supplements. Participants completed the past month online version of the questionnaire to indicate diet the past 30 d, and DHQ output was used to approximate intake of carotenoids from foods.
Additional potential modifiers: self-categorized race/ethnicity, sex, income, and age were obtained from standard demographics questions. Sun exposure was assessed by asking validated questions about sunlight exposure [36]. At the first visit, participant height was measured using a calibrated height board, and at each visit, weight and body adiposity (body fat percentage) were measured using a combination scale and bioelectrical impedance device (Tanita DC-430U Body Composition Analyzer). Data collection season and study site were also tested as potential modifiers.
Statistical analysis: only the moderate-dose and the high-dose groups from the original study were included in this analysis. Pearson’s or point biserial correlations between the 6-wk changes in skin and plasma carotenoid levels and potential individual-level effect modifiers were examined [8]. (Correlations between changes in plasma carotenoids and optical measurements, such as skin coloration, were not examined because they do not have a hypothesized association with plasma carotenoid concentration detection or physiology.) Variable distributions were verified using graphic tools (e.g., histograms and normal QQ plots). Scatter plots were used to examine outcome variables compared with covariates to ensure the outcome, and covariates were linearly correlated. The base-2 log transformation was applied to the plasma carotenoids to correct for right-skewness in the data.
Then, general linear models were used to examine whether covariates might interact with the intervention group assignment. Following the research team’s prior published study [7], 2 base models (one for skin carotenoid changes and one for plasma carotenoid changes) were used, including study site, self-reported race/ethnicity, baseline BMI, baseline skin or plasma carotenoid levels, and the intervention group assignments as covariates. After that, each potential modifier and its interaction with the intervention group assignment were individually added into the base models and statistically tested. The following were examined as potential modifiers: data collection season, study site, self-reported race/ethnicity, biological sex, age, baseline percent body fat, BMI, melanin index, hemoglobin index, hemoglobin saturation index, skin coloration (redness and yellowness and skin lightness), blood total cholesterol, fasting blood glucose, sun exposure, and baseline intake of carotenoids from foods. Effect sizes were reported as estimated mean changes (95% confidence interval [CI]) or slopes (95% CI). SAS version 9.4 (SAS Institute Inc.) was used to analyze results, and a P value of 0.05 was considered statistically significant.
Results
The participant flow diagram was previously published [8]. A total of 323 potential participants were screened for eligibility; 169 were enrolled, and 158 completed the study. There were 106 participants in the high- and moderate-dose groups, which were included in the current analyses (Table 1). Participants were, on average 32.5 (11.2) y of age, with a mean BMI of 25.7 (4.00) kg/m2, mean skin carotenoid score of 302 (96.4), and mean plasma carotenoid level (sum of trans-lutein, trans-zeaxanthin, cis-lutein/zeaxanthin, α-cryptoxanthin, β-cryptoxanthin, trans-lycopene, cis-lycopene, α-carotene, trans-β-carotene, and cis-β-carotene) of 162 (72.0) μg/dL.
Table 1.
Summary of baseline statistics among n = 106 participants in a randomized controlled trial of a carotenoid-containing juice
| Variable | Total sample | Moderate dose | High dose | |
|---|---|---|---|---|
| (n = 106) |
(n = 52) |
(n = 54) |
||
| Mean (SD) | Mean (SD) | Mean (SD) | ||
| Skin carotenoid score | 302 (96.4) | 311 (103) | 292 (90.0) | |
| Plasma carotenoid (μg/dL) | 162 (72.0) | 163 (76.2) | 162 (68.5) | |
| Age (y) | 32.5 (11.2) | 32.9 (11.7) | 32.1 (10.9) | |
| % body fat | 24.0 (8.48) | 23.4 (8.35) | 24.7 (8.63) | |
| BMI (kg/m2) | 25.7 (4.00) | 25.7 (4.17) | 25.7 (3.86) | |
| Arm melanin index | 1.04 (0.52) | 1.03 (0.54) | 1.05 (0.50) | |
| Arm hemoglobin index | 1.10 (0.24) | 1.10 (0.22) | 1.11 (0.26) | |
| Arm hemoglobin saturation | 42.6 (15.1) | 42.9 (15.0) | 42.3 (15.3) | |
| a∗ (skin coloration – redness) | 6.45 (2.33) | 6.36 (2.51) | 6.54 (2.17) | |
| b∗ (skin coloration – yellowness) | 16.6 (2.77) | 16.6 (2.90) | 16.6 (2.66) | |
| L∗ (skin lightness) | 59.4 (8.84) | 59.7 (9.05) | 59.1 (8.71) | |
| Plasma cholesterol (mg/dL) | 294 (236) | 332 (283) | 258 (174) | |
| Plasma glucose level (mg/dL) | 80.8 (10.3) | 81.5 (9.52) | 80.2 (11.0) | |
| Sun exposure (h/d) | 2.32 (1.06) | 2.32 (1.14) | 2.33 (0.98) | |
| Carotenoid intake (μg/d) |
15,081 (11,756) |
14,750 (11,544) |
15,394 (12,052) |
|
| Variable |
Category |
n (%) |
n (%) |
n (%) |
| Season | 2021 Summer | 42 (39.6) | 20 (38.5) | 22 (40.7) |
| 2021 Fall | 44 (41.5) | 22 (42.3) | 22 (40.7) | |
| 2021 Winter–2022 Spring | 20 (18.9) | 10 (19.2) | 10 (18.5) | |
| Study site | ECU | 40 (37.7) | 19 (36.5) | 21 (38.9) |
| UMN | 26 (24.5) | 13 (25.0) | 13 (24.1) | |
| BCM | 40 (37.7) | 20 (38.5) | 20 (37.0) | |
| Sex | Male | 61 (57.5) | 33 (63.5) | 28 (51.8) |
| Female | 45 (42.5) | 19 (36.5) | 26 (48.2) | |
| Race/ethnicity | Non-Hispanic African American/Black | 25 (23.6) | 12 (23.1) | 13 (24.1) |
| Asian | 27 (25.5) | 14 (26.9) | 13 (24.1) | |
| Non-Hispanic White | 29 (27.4) | 14 (26.9) | 15 (27.8) | |
| Hispanic/Latinx | 25 (23.6) | 12 (23.1) | 13 (24.1) | |
BCM, Baylor College of Medicine; ECU, East Carolina University; UMN, University of Minnesota.
L∗ represents skin lightness, and positive values of a∗ and b∗ represent degrees of redness and yellowness.
Table 2 shows unadjusted Pearson’s or point biserial correlations between the 6-wk changes in skin and plasma carotenoid levels and each of the potential individual-level effect modifiers. In the high-dose group, age had a significantly negative correlation (r = −0.31), baseline arm hemoglobin saturation had a significantly positive correlation (r = 0.28), and skin lightness had a significantly positive correlation (r = 0.27) with the 6-wk changes in skin carotenoid level. This suggests that as age increases, skin carotenoids increase at a lower rate in response to dietary changes, and as hemoglobin saturation and skin lightness increase, skin carotenoid scores increase at a higher rate. In the moderate-dose group, baseline skin carotenoid level had a significantly inverse correlation (r = −0.35) with the 6-wk changes in skin carotenoid level, such that those with a higher baseline skin carotenoid level had a smaller increase in skin carotenoids over 6 wk. Being at the East Carolina University site had a significantly positive correlation (r = 0.28) with the 6-wk changes in plasma carotenoid level.
Table 2.
Unadjusted Pearson’s or point biserial correlations between the 6-wk changes in skin and plasma carotenoid levels and each of the baseline covariates/factors
| Baseline covariates/factors | Change in Veggie Meter-assessed skin carotenoid level |
Change in log2-transformed plasma carotenoid level |
||
|---|---|---|---|---|
| Moderate dose (n = 52) | High dose (n = 54) | Moderate dose (n = 52) | High dose (n = 54) | |
| Pearson’s correlation | ||||
| Skin carotenoid | −0.35∗ | −0.17 | ||
| Plasma carotenoid (μg/dL)1 | −0.25 | −0.15 | ||
| Age (y) | −0.03 | −0.31∗ | 0.06 | 0.08 |
| % body fat | −0.02 | −0.03 | −0.06 | −0.02 |
| BMI (kg/m2) | −0.18 | −0.20 | 0.01 | −0.17 |
| Arm melanin index | −0.06 | −0.22 | ||
| Arm hemoglobin index | −0.06 | −0.18 | ||
| Arm hemoglobin saturation | 0.06 | 0.28∗ | ||
| a∗ (Skin coloration – redness) | −0.06 | −0.25 | ||
| b∗ (Skin coloration – yellowness) | −0.03 | 0.07 | ||
| L∗ (Skin lightness) | 0.09 | 0.27∗ | ||
| Plasma cholesterol (mg/dL)1 | −0.06 | 0.14 | 0.16 | −0.02 |
| Plasma glucose level (mg/dL) | −0.04 | 0.17 | 0.13 | 0.02 |
| Sun exposure (h/d) | 0.11 | −0.06 | −0.19 | −0.05 |
| Carotenoid intake (μg/d)1 | −0.16 | −0.03 | 0.07 | −0.13 |
| Point biserial correlation | ||||
| Season | ||||
| 2021 Summer | 0.00 | 0.19 | 0.00 | 0.06 |
| 2021 Fall | 0.04 | −0.24 | −0.17 | −0.10 |
| 2021 Winter–2022 Spring | −0.05 | 0.06 | 0.21 | 0.05 |
| Study site | ||||
| ECU | −0.17 | 0.16 | 0.28∗ | 0.18 |
| UMN | 0.06 | −0.04 | −0.10 | −0.16 |
| BCM | 0.11 | −0.12 | −0.18 | −0.05 |
| Sex | ||||
| Male vs. female | −0.13 | −0.13 | 0.03 | −0.10 |
| Race/ethnicity | ||||
| Non-Hispanic African American/Black | −0.05 | −0.23 | 0.16 | −0.04 |
| Non-Hispanic Asian | 0.23 | 0.21 | −0.06 | 0.00 |
| Non-Hispanic White | 0.06 | −0.01 | −0.03 | 0.10 |
| Hispanic/Latinx | −0.26 | 0.04 | −0.07 | −0.06 |
BCM, Baylor College of Medicine; ECU, East Carolina University; UMN, University of Minnesota.
∗P < 0.05. L∗ represents skin lightness, and positive values of a∗ and b∗ represent degrees of redness and yellowness.
log2 transformation applied.
Table 3 summarizes the overall effects of the intervention arms and the effect modification of the selected baseline covariates on 6-wk changes in Veggie Meter-assessed skin carotenoid levels. The overall mean changes were estimated from the base model, and effect modification was estimated from the base model plus the covariate and its interaction with the intervention group assignments. The base model showed that the 2 intervention groups (i.e., moderate dose and high dose, respectively) achieved an overall 6-wk mean increase of 61.1 (95% CI: 44.6, 77.7), and 120 (95% CI: 104, 137) units in skin carotenoid score. These mean increases were significantly (P < 0.001) different across the 2 intervention groups. The models with the covariates and the interactions showed that these effect sizes appeared to vary across the 3 study sites, the 4 racial/ethnic groups, and the 2 sex groups. However, these variations were not statistically significant (P > 0.05) within each intervention group, nor were the interactions between the covariates and the intervention group assignments.
Table 3.
Overall effects of the interventions and effect modifications of selected baseline covariates/factors on 6-wk changes in Veggie Meter-assessed skin carotenoid level
| Group | Moderate dose (mean and 95% CI) | High dose (mean and 95% CI) | P1 |
|---|---|---|---|
| Overall mean change | 61.1 (44.6, 77.7) | 120 (104, 137) | <0.001 |
| Mean change by study site | 0.414 | ||
| ECU | 52.9 (25.3, 80.4) | 131 (105, 158) | |
| UMN | 70.3 (37.3, 103) | 125 (91.3, 159) | |
| BCM | 60.5 (33.9, 87.1) | 104 (76.8, 131) | |
| Mean change by race/ethnicity | 0.564 | ||
| Non-Hispanic African American/Black | 58 (22.5, 93.4) | 91 (57, 125) | |
| Non-Hispanic Asian | 86.9 (52.8, 121) | 149 (115, 183) | |
| Non-Hispanic White | 55.8 (23.6, 87.9) | 116 (85, 148) | |
| Hispanic/Latinx | 44.4 (8.78, 80) | 125 (91.7, 159) | |
| BMI slope | −1.76 (−6.23, 2.7) | −3.25 (−7.71, 1.2) | 0.621 |
| Skin carotenoid slope | −0.29 (−0.47, −0.12)∗ | −0.20 (−0.39, −0.01)∗ | 0.456 |
| Mean change by season | 0.417 | ||
| 2021 Summer | 55.7 (27.7, 83.6) | 134 (107, 160) | |
| 2021 Fall | 63.9 (38, 89.8) | 109 (82.8, 135) | |
| 2021 Winter–2022 Spring | 67.6 (27, 108) | 117 (76.5, 158) | |
| Mean change by sex | 0.530 | ||
| Male | 60.3 (39.1, 81.5) | 112 (89.2, 135) | |
| Female | 62.9 (34.9, 90.8) | 130 (106, 154) | |
| Age slope | 0.24 (−1.23, 1.71) | −1.14 (−2.75, 0.46) | 0.202 |
| % body fat slope | 0.23 (−2.02, 2.48) | 0.59 (−1.53, 2.71) | 0.796 |
| Arm melanin index slope | 29.7 (−20.0, 79.3) | 5.60 (−45.7, 56.9) | 0.285 |
| Arm hemoglobin index slope | 20.9 (−60.7, 102) | −23.1 (−88.5, 42.2) | 0.390 |
| Arm hemoglobin saturation index slope | −0.10 (−1.72, 1.51) | 0.85 (−0.73, 2.43) | 0.215 |
| a∗ (Skin coloration – redness) slope | 4.56 (−4.18, 13.3) | −1.26 (−10.7, 8.23) | 0.253 |
| b∗ (Skin coloration – yellowness) slope | 3.80 (−3.01, 10.6) | 3.94 (−3.15, 11.0) | 0.975 |
| L∗ (Skin lightness) slope | −0.71 (−3.84, 2.41) | 0.73 (−2.41, 3.88) | 0.274 |
| Cholesterol2 slope | 6.09 (−14.9, 27.0) | 11.5 (−12.6, 35.7) | 0.732 |
| Glucose slope | −1.00 (−2.81, 0.82) | 1.26 (−0.26, 2.79) | 0.055 |
| Sun exposure slope | 12.0 (−2.68, 26.7) | −3.78 (−21.3, 13.8) | 0.165 |
| Carotenoid intake2 slope | −2.64 (−23.4, 18.1) | −3.45 (−24.8, 17.9) | 0.958 |
∗P < 0.05.
BCM, Baylor College of Medicine; ECU, East Carolina University; UMN, University of Minnesota.
The overall mean changes were estimated from the base model, and the effect modifications were estimated from the base model plus the covariate/factor and its interaction with the intervention group assignment. L∗ represents skin lightness, and positive values of a∗ and b∗ represent degrees of redness and yellowness.
P value for difference between moderate and high-dose groups.
log2 transformation applied.
For continuous covariates, such as baseline BMI, baseline skin carotenoid level, age, hemoglobin index, hemoglobin saturation index, and skin coloration (redness), slope estimates were provided for both intervention groups. Among all covariates, only the baseline skin carotenoid level had significant (P < 0.05) slopes for both intervention groups, meaning that it had significant modifying effects for both the moderate- and high-dose interventions. However, the slopes were not significantly different (P = 0.456) between the 2 intervention groups.
Similar analyses for 6-wk changes in log2-transformed plasma carotenoid level as an outcome are provided in Table 4. The base model estimated a mean increase of 0.38 (95% CI: 0.30, 0.47) and 0.68 (95% CI: 0.58, 0.79) units in the outcome for the moderate-dose and high-dose interventions, respectively, and the 2 mean increases were significantly (P < 0.001) different between the 2 intervention groups. However, the models with the covariates and the interactions did not detect any significant effect modifications or interactions. Similar analyses for 3-wk changes in Veggie Meter-assessed skin carotenoid level and log2-transformed plasma carotenoid level are included in Supplemental Tables 1–3.
Table 4.
Overall effects of the interventions and effect modifications of selected baseline covariates/factors on 6-wk changes in log2-transformed plasma carotenoid level
| Group | Moderate dose (mean/slope and 95% CI) | High dose (mean/slope and 95% CI) | P |
|---|---|---|---|
| Overall mean change | 0.38 (0.30, 0.47) | 0.68 (0.58, 0.79) | <0.001 |
| Mean change by study site | 0.892 | ||
| ECU | 0.48 (0.33, 0.62) | 0.76 (0.59, 0.93) | |
| UMN | 0.35 (0.18, 0.52) | 0.61 (0.40, 0.83) | |
| BCM | 0.32 (0.19, 0.46) | 0.66 (0.49, 0.84) | |
| Mean change by race/ethnicity | 0.668 | ||
| Non-Hispanic African American/Black | 0.48 (0.30, 0.66) | 0.64 (0.42, 0.86) | |
| Non-Hispanic Asian | 0.35 (0.17, 0.53) | 0.69 (0.47, 0.91) | |
| Non-Hispanic White | 0.33 (0.17, 0.49) | 0.72 (0.52, 0.92) | |
| Hispanic/Latinx | 0.37 (0.19, 0.55) | 0.67 (0.46, 0.89) | |
| BMI slope | −0.01 (−0.03, 0.02) | −0.02 (−0.04, 0.01) | 0.549 |
| Plasma carotenoid1 slope | −0.11 (−0.27, 0.05) | −0.07 (−0.26, 0.11) | 0.744 |
| Mean change by season | 0.705 | ||
| 2021 Summer | 0.34 (0.21, 0.48) | 0.67 (0.50, 0.84) | |
| 2021 Fall | 0.34 (0.22, 0.47) | 0.67 (0.50, 0.84) | |
| 2021 Winter–2022 Spring | 0.56 (0.36, 0.76) | 0.74 (0.48, 1.00) | |
| Mean change by sex | 0.701 | ||
| Male | 0.38 (0.27, 0.48) | 0.65 (0.50, 0.80) | |
| Female | 0.39 (0.25, 0.53) | 0.72 (0.56, 0.87) | |
| Age slope (×10-3) | 3.36 (−4.12, 10.8) | 5.81 (−4.35, 16.0) | 0.692 |
| % body fat slope (×10-3) | 0.54 (−10.9, 12.0) | 1.65 (−11.7, 15.0) | 0.890 |
| Cholesterol1 slope | 0.05 (−0.05, 0.16) | −0.04 (−0.19, 0.11) | 0.292 |
| Glucose slope (×10-3) | 2.69 (−6.55, 11.9) | 0.31 (−9.71, 10.3) | 0.723 |
| Sun exposure slope | −0.05 (−0.12, 0.03) | −0.04 (−0.15, 0.07) | 0.871 |
| Carotenoid intake1 slope | 0.05 (−0.05, 0.15) | −0.07 (−0.20, 0.06) | 0.171 |
BCM, Baylor College of Medicine; ECU, East Carolina University; UMN, University of Minnesota.
The overall mean changes were estimated from the base model, and the effect modifications were estimated from the base model plus the covariate/factor and its interaction with the intervention group assignment.
log2 transformation applied.
Discussion
Findings suggest that reflection-spectroscopy-assessed skin carotenoids as measured by the Veggie Meter are a robust measure to examine changes in carotenoid intake and thus a potentially useful tool for evaluating policies and programs designed to increase FV intake. There is potential evidence of smaller changes in skin carotenoids in response to increased carotenoid intake among those with higher starting SCCs and among older populations. This effect modification suggests that when estimating sample sizes needed to detect a significant effect on Veggie Meter-assessed skin carotenoid scores, and when one needs to anticipate an effect size, the baseline carotenoid status of the population from which recruitment is occurring will be an important factor to consider. Further, researchers should ensure that baseline skin carotenoid measurements for individuals are obtained before the intervention or control treatment is applied so that findings can be interpreted in the context of baseline carotenoid status.
It will be important to consider baseline carotenoid status if the relative effectiveness of a program or policy is being evaluated via skin carotenoid measurement. For example, if those who appear to have greater responsiveness to a dietary intervention, as measured by skin carotenoids, are of lower income, this could be either due to a truly stronger effect of the intervention among those with low incomes; in contrast, it could also be that those with lower incomes have lower baseline carotenoid status compared with participants of higher incomes so that greater increases in skin carotenoids in this group are fully or partially attributable to this baseline difference rather than a greater invention effect.
Our finding that baseline skin carotenoid score is a potential modifier of skin carotenoid response is similar to that of Obana et al. [29] who found a negative correlation between baseline Veggie Meter-assessed SCC and the percentage increase in SCC at month 6 (r = −0.36, P < 0.001). In a previous analysis, Jilcott Pitts et al. [8] found the correlation between baseline Veggie Meter-assessed SCC and percentage increase in SCC at week 6 to be r = −0.51 (P < 0.001) and −0.54 (P < 0.001), in the moderate- and high-dose groups, respectively.
Notably, race/ethnicity, biological sex, melanin, body fat percentage, and BMI were not significant modifiers of the effect of the juice intervention on skin carotenoid levels. These variables have been suggested as important to measure when using skin carotenoid scores as outcome measures in nutrition interventions [[13], [14], [15]]. Our findings are important for researchers using the Veggie Meter to determine intervention effectiveness and provide support for the use of the device in future public health nutrition intervention evaluations among diverse populations.
In the bivariate analyses, age had a statistically significant correlation with 6-wk changes in skin carotenoid scores, suggesting that as age increases, skin carotenoids increase at a lower rate in response to dietary changes. However, age was no longer a statistically significant covariate when controlling for other factors. Further, in bivariate analyses, participants at the East Carolina University (ECU) site had a significantly positive correlation with the 6-wk changes in plasma carotenoid level. There are different site-specific factors that could influence this, such as experimental variables, such as juice storage and batching or interdevice calibration, or there could be other unaccounted for environmental or demographic variables. In the future, site should continue to be used as a factor when analyzing skin carotenoid data from multisite research. Additionally, it would be beneficial to have a method to cross-calibrate Veggie Meter devices across study sites to ensure consistency and accuracy in studies utilizing multiple devices.
Prior research found that skin yellowness modified responsiveness of skin carotenoid score to dietary increases in carotenoid intake [35]. However, in the current study, we did not find effect modification by skin yellowness or redness. We did find a positive correlation with hemoglobin saturation and skin lightness, such that as hemoglobin saturation and skin lightness increased, there were greater increases in skin carotenoid scores. It is unclear why there is a positive interaction between hemoglobin or melanin and changes in skin carotenoids with diet. Whether this is due to a physiologic effect, optical effect, or signal overlap should be investigated in future studies with additional hemoglobin and melanin measurement methodologies. For the skin lightness measure, a higher number means lighter skin tone, and thus, may be associated with greater responsiveness. This study is not without limitations. This analysis was exploratory in nature, and the original study was not designed to be fully powered to detect effect modification by baseline covariates. Hence, a larger study might be needed to accurately discern modifiers, and these results should be compared with those of future studies to assess consistency. Future studies should examine potential modifiers of skin carotenoid responses, including broader BMI ranges and other carotenoid food sources. In addition, we only tested one type of skin carotenoid measurement device, and our findings might not be applicable to other devices for skin carotenoid measurement. A further limitation is that many variables were examined as potential effect modifiers, and consequently, some of the significant findings may result from type 1 error. Study strengths include a diverse population and the examination of effect modification in the context of a randomized controlled feeding trial in which carotenoid intake was controlled.
Overall, the analyses presented here support work by others [8,11,37] suggesting that the Veggie Meter is a useful tool for researchers designing and evaluating public health interventions. Based on the current findings, changes in SCC in response to an intervention should be interpreted in the context of baseline skin carotenoid scores, but other variables did not significantly modify the effect of carotenoid intake on changes in Veggie Meter-assessed SCC.
Author contributions
The authors’ contributions were as follows – SJP, MNL, NEM, LH: designed the study; QW: conducted all statistical analyses; SJP: wrote initial first draft, with intellectual content added by all coauthors; and all authors: read and approved the final manuscript.
Funding
This study was funded by the National Institutes of Health (R01 HL142544) and the USDA/ARS (cooperative agreement 3092-51000-059-NEW2S, to NEM). The contents of this work are solely the responsibility of the authors and do not necessarily represent the official views of the NIH or USDA.
Data availability
Data described in the manuscript, code book, and analytic code will be made available upon request pending appropriate Institutional Review Board approval and review of and approval from the investigators on the current project.
Declaration of interests
The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: Stephanie Pitts reports a relationship with Gretchen Swanson Center for Nutrition that includes: consulting or advisory. Stephanie Pitts reports a relationship with University at Buffalo that includes: consulting or advisory.
Conflict of interest
The authors report no conflicts of interest.
Footnotes
Supplementary data to this article can be found online at https://doi.org/10.1016/j.tjnut.2023.09.014.
Appendix A. Supplementary data
The following is the Supplementary data to this article:
References
- 1.US Department of Agriculture, US Department of Health and Human Services . 9th edition [Internet] 2020. Dietary Guidelines for Americans, 2020-2025.https://www.dietaryguidelines.gov/ [cited 5 May, 2023]. Available from: [Google Scholar]
- 2.Krebs-Smith S.M., Guenther P.M., Subar A.F., Kirkpatrick S.I., Dodd K.W. Americans do not meet federal dietary recommendations. J. Nutr. 2010;140(10):1832–1838. doi: 10.3945/jn.110.124826. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Rehm C.D., Peñalvo J.L., Afshin A., Mozaffarian D. Dietary intake among US adults, 1999-2012. JAMA. 2016;315(23):2542–2553. doi: 10.1001/jama.2016.7491. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Lee-Kwan S.H., Moore L.V., Blanck H.M., Harris D.M., Galuska D. Disparities in state-specific adult fruit and vegetable consumption — United States, 2015. MMWR Morb. Mortal. Wkly. Rep. 2017;66(45):1241–1247. doi: 10.15585/mmwr.mm6645a1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Roark R.A., Niederhauser V.P. Fruit and vegetable intake: issues with definition and measurement. Public Health Nutr. 2013;16(1):2–7. doi: 10.1017/S1368980012000985. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Kirkpatrick S.I., Collins C.E., Keogh R.H., Krebs-Smith S.M., Neuhouser M.L., Wallace A. Assessing dietary outcomes in intervention studies: pitfalls, strategies, and research needs. Nutrients. 2018;10(8):1001. doi: 10.3390/nu10081001. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Jilcott Pitts S.B., Moran N.E., Wu Q., Harnack L., Craft N.E., Hanchard N., et al. Pressure-mediated reflection spectroscopy criterion validity as a biomarker of fruit and vegetable intake: a 2-site cross-sectional study of 4 racial or ethnic groups. J. Nutr. 2022;152(1):107–116. doi: 10.1093/jn/nxab349. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Jilcott Pitts S., Moran N.E., Laska M.N., Wu Q., Harnack L., Moe S., et al. Reflection spectroscopy-assessed skin carotenoids are sensitive to change in carotenoid intake in a 6-week randomized controlled feeding trial in a racially/ethnically diverse sample. J. Nutr. 2023;153(4):1133–1142. doi: 10.1016/j.tjnut.2023.02.017. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Jahns L., Johnson L.K., Conrad Z., Bukowski M., Raatz S.K., Jilcott Pitts S., et al. Concurrent validity of skin carotenoid status as a concentration biomarker of vegetable and fruit intake compared to multiple 24-h recalls and plasma carotenoid concentrations across one year: a cohort study. Nutr. J. 2019;18(1):78. doi: 10.1186/s12937-019-0500-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Radtke M.D., Pitts S.J., Jahns L., Firnhaber G.C., Loofbourrow B.M., Zeng A., et al. Criterion-related validity of spectroscopy-based skin carotenoid measurements as a proxy for fruit and vegetable intake: a systematic review. Adv. Nutr. 2020;11(5):1282–1299. doi: 10.1093/advances/nmaa054. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Casperson S.L., Roemmich J.N., Larson K.J., Hess J.M., Palmer D.G., Jahns L. Sensitivity of pressure-mediated reflection spectroscopy to detect changes in skin carotenoids in adults without obesity in response to increased carotenoid intake: a randomized controlled trial. J. Nutr. 2023;153(2):588–597. doi: 10.1016/j.tjnut.2023.01.002. [DOI] [PubMed] [Google Scholar]
- 12.Dragsted L.O., Gao Q., Scalbert A., Vergères G., Kolehmainen M., Manach C., et al. Validation of biomarkers of food intake-critical assessment of candidate biomarkers. Genes Nutr. 2018;13:14. doi: 10.1186/s12263-018-0603-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Burkholder S., Jilcott Pitts S., Wu Q., Bayles J., Baybutt R., Stage V.C. Skin carotenoid status over time and differences by age and sex among Head Start children (3–5 years) J. Nutr. Educ. Behav. 2021;53(2):103–109. doi: 10.1016/j.jneb.2020.10.019. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Rush E., Amoah I., Diep T., Jalili-Moghaddam S. Determinants and suitability of carotenoid reflection score as a measure of carotenoid status. Nutrients. 2020;12(1):113. doi: 10.3390/nu12010113. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Takayanagi Y., Obana A., Muto S., Asaoka R., Tanito M., Ermakov I.V., et al. Relationships between skin carotenoid levels and metabolic syndrome. Antioxidants (Basel) 2021;11(1):14. doi: 10.3390/antiox11010014. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Jung S., Darvin M.E., Chung H.S., Jung B., Lee S.H., Lenz K., et al. Antioxidants in Asian-Korean and Caucasian skin: the influence of nutrition and stress, Skin Pharmacol. Physiol. 2014;27(6):293–302. doi: 10.1159/000361053. [DOI] [PubMed] [Google Scholar]
- 17.Jilcott Pitts S.B., Jahns L., Wu Q., Moran N.E., Bell R.A., Truesdale K.P., et al. A non-invasive assessment of skin carotenoid status through reflection spectroscopy is a feasible, reliable and potentially valid measure of fruit and vegetable consumption in a diverse community sample. Public Health Nutr. 2018;21(9):1664–1670. doi: 10.1017/S136898001700430X. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Obana A., Gohto Y., Gellermann W., Ermakov I.V., Sasano H., Seto T., et al. Skin Carotenoid Index in a large Japanese population sample. Sci. Rep. 2019;9(1):9318. doi: 10.1038/s41598-019-45751-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Toh D.W.K., Loh W.W., Sutanto C.N., Yao Y., Kim J.E. Skin carotenoid status and plasma carotenoids: biomarkers of dietary carotenoids, fruits and vegetables for middle-aged and older Singaporean adults. Br. J. Nutr. 2021;126(9):1398–1407. doi: 10.1017/S0007114521000143. [DOI] [PubMed] [Google Scholar]
- 20.Ermakov I.V., Ermakova M., Sharifzadeh M., Gorusupudi A., Farnsworth K., Bernstein P.S., et al. Optical assessment of skin carotenoid status as a biomarker of vegetable and fruit intake. Arch. Biochem. Biophys. 2018;646:46–54. doi: 10.1016/j.abb.2018.03.033. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Matsumoto M., Suganuma H., Shimizu S., Hayashi H., Sawada K., Tokuda I., et al. Skin carotenoid level as an alternative marker of serum total carotenoid concentration and vegetable intake correlates with biomarkers of circulatory diseases and metabolic syndrome. Nutrients. 2020;12(6):1825. doi: 10.3390/nu12061825. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Keller J.E., Taylor M.K., Smith A.N., Littrell J., Spaeth K., Boeckman C.R., et al. Correlation of skin carotenoid content with 3-day dietary intake in community dwelling older adults. J. Food Compost Anal. 2022;105 doi: 10.1016/j.jfca.2021.104243. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Di Noia J., Gellermann W. Use of the spectroscopy-based Veggie Meter® to objectively assess fruit and vegetable intake in low-income adults. Nutrients. 2021;13(7):2270. doi: 10.3390/nu13072270. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Jones A., Radtke M., Chodur G., Scherr R. Assessing the relationship between nutrition knowledge and skin carotenoids in university students. Curr. Dev. Nutr. 2020;4:1313. doi: 10.1093/cdn/nzaa059_030. [DOI] [Google Scholar]
- 25.Borel P. Genetic variations involved in interindividual variability in carotenoid status. Mol. Nutr. Food Res. 2012;56(2):228–240. doi: 10.1002/mnfr.201100322. [DOI] [PubMed] [Google Scholar]
- 26.Borel P., De Edelenyi F.S., Vincent-Baudry S., Malezet-Desmoulin C., Margotat A., Lyan B., et al. Genetic variants in BCMO1 and CD36 are associated with plasma lutein concentrations and macular pigment optical density in humans. Ann. Med. 2011;43(1):47–59. doi: 10.3109/07853890.2010.531757. [DOI] [PubMed] [Google Scholar]
- 27.Zubair N., Kooperberg C., Liu J., Di C., Peters U., Neuhouser M.L. Genetic variation predicts serum lycopene concentrations in a multiethnic population of postmenopausal women. J. Nutr. 2015;145(2):187–192. doi: 10.3945/jn.114.202150. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Borel P., Desmarchelier C., Nowicki M., Bott R. Lycopene bioavailability is associated with a combination of genetic variants. Free Radic. Biol. Med. 2015;83:238–244. doi: 10.1016/j.freeradbiomed.2015.02.033. [DOI] [PubMed] [Google Scholar]
- 29.Obana A., Asaoka R., Miura A., Nozue M., Takayanagi Y., Nakamura M. Improving skin carotenoid levels in young students through brief dietary education using the Veggie Meter. Antioxidants (Basel) 2022;11(8):1570. doi: 10.3390/antiox11081570. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Burrows T.L., Williams R., Rollo M., Wood L., Garg M.L., Jensen M., et al. Plasma carotenoid levels as biomarkers of dietary carotenoid consumption: a systematic review of the validation studies. J. Nutr. Intermed. Metab. 2015;2(1–2):15–64. doi: 10.1016/j.jnim.2015.05.001. [DOI] [Google Scholar]
- 31.Martini M.C., Campbell D.R., Gross M.D., Grandits G.A., Potter J.D., Slavin J.L. Plasma carotenoids as biomarkers of vegetable intake: the University of Minnesota Cancer Prevention Research Unit feeding studies, Cancer Epidemiol. Biomarkers Prev. 1995;4(5):491–496. [PubMed] [Google Scholar]
- 32.Campbell D.R., Gross M.D., Martini M.C., Grandits G.A., Slavin J.L., Potter J.D. Plasma carotenoids as biomarkers of vegetable and fruit intake, Cancer Epidemiol. Biomarkers Prev. 1994;3(6):493–500. [PubMed] [Google Scholar]
- 33.Rock C.L., Flatt S.W., Wright F.A., Faerber S., Newman V., Kealey S., et al. Responsiveness of carotenoids to a high vegetable diet intervention designed to prevent breast cancer recurrence. Cancer Epidemiol. Biomarkers Prev. 1997;6(8):617–623. [PubMed] [Google Scholar]
- 34.Le Marchand L., Hankin J.H., Carter F.S., Essling C., Luffey D., Franke A.A., et al. A pilot study on the use of plasma carotenoids and ascorbic acid as markers of compliance to a high fruit and vegetable dietary intervention. Cancer Epidemiol. Biomarkers Prev. 1994;3(3):245–251. [PubMed] [Google Scholar]
- 35.Pezdirc K., Hutchesson M.J., Williams R.L., Rollo M.E., Burrows T.L., Wood L.G., et al. Consuming high-carotenoid fruit and vegetables influences skin yellowness and plasma carotenoids in young women: a single-blind randomized crossover trial. J. Acad. Nutr. Diet. 2016;116(8):1257–1265. doi: 10.1016/j.jand.2016.03.012. [DOI] [PubMed] [Google Scholar]
- 36.Glanz K., Yaroch A.L., Dancel M., Saraiya M., Crane L.A., Buller D.B., et al. Measures of sun exposure and sun protection practices for behavioral and epidemiologic research. Arch. Dermatol. 2008;144(2):217–222. doi: 10.1001/archdermatol.2007.46. [DOI] [PubMed] [Google Scholar]
- 37.Bayles J., Peterson A.D., Jilcott Pitts S., Bian H., Goodell L.S., Burkholder S., et al. Food-based Science, Technology, Engineering, Arts, and Mathematics (STEAM) learning activities may reduce decline in preschoolers’ skin carotenoid status. J. Nutr. Educ. Behav. 2021;53(4):343–351. doi: 10.1016/j.jneb.2020.10.017. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
Data described in the manuscript, code book, and analytic code will be made available upon request pending appropriate Institutional Review Board approval and review of and approval from the investigators on the current project.