Biomarkers of Tretinoin Precursors and Tretinoin Efficacy in Moderate to Severe Facial Photodamage (2024)

Key Points

Question Can a cosmeceutical formulation of triple tretinoin precursors (TTP) provide efficacy comparable with that of tretinoin (RA) for photodamage treatment, and what are the molecular targets that best correlate with retinoid efficacy?

Findings In this randomized clinical trial of 24 patients, at week 24, there was no significant difference in photoaging scores among those treated with TTP vs RA. Changes in fine wrinkles correlated with suppression of MMP2 messenger RNA, but not other evaluated matrix metalloproteinases, CRABP2, or procollagen I.

Meaning The gene MMP2, which encodes for a type IV collagenase, may represent a previously underappreciated mediator of retinoid efficacy.

Importance Topical formulations of tretinoin precursors (retinol and its ester derivatives) are widely available over the counter and may offer similar clinical benefits to those of tretinoin for treatment of photoaging. However, which of the many purported molecular effects of retinoids most strongly drives clinical improvements in tretinoin-treated skin remains unclear.

Objectives To evaluate the clinical efficacy of topical tretinoin precursors (TTP) vs tretinoin (RA) in treating moderate to severe facial photodamage and to identify potential biomarkers that correlate with clinical efficacy.

Design, Setting, and Participants This randomized, double-blind, single-center, parallel-arm study of 24 patients with moderate to severe facial photodamage was conducted at an academic referral center from November 2010 to December 2011, with data analysis performed from January 2012 to December 2021.

Interventions Daily topical application of 0.02% RA or 1.1% TTP formulation containing retinol, retinyl acetate, and retinyl palmitate for 24 weeks.

Main Outcomes and Measures Photoaging and tolerability were assessed by dermatologist evaluations and patient-reported outcomes. Target gene expression was assessed by real-time quantitative polymerase chain reaction of biopsied tissue from treated areas.

Results A total of 20 White women were ultimately analyzed (9 randomized to TTP, 11 randomized to RA). At week 24, there was no significant difference in Griffiths photoaging scores among patients receiving TTP vs RA (median, 4 vs 5) (TTP − RA difference: −1; 95% CI, −2 to 1; P = .27). Treatment with TTP was associated with erythema 6 times less frequently than RA (11% vs 64%) (TTP − RA difference: −0.53; 95% CI, −0.88 to −0.17; P = .01). Target gene analysis showed significant CRABP2 messenger RNA (mRNA) induction (confirming retinoic acid receptor signaling) but no significant changes in procollagen I or MMP1/3/9 mRNA in TTP-treated samples. Instead, MMP2 mRNA, which encodes a type IV collagenase, was significantly reduced in TTP-treated samples (week 24 − baseline mRNA difference: −5; 96% CI, −33 to 1.6; P = .02), and changes in MMP2 were strongly correlated with changes in fine wrinkles (r = 0.54; 95% CI, 0.12 to 0.80; P = .01). Interestingly, patients with severe baseline wrinkles exhibited greater improvements (r = −0.74; 95% CI, −0.89 to −0.43; P < .001). This trend was mirrored in MMP2 mRNA, with initial expression strongly predicting subsequent changes (r = −0.78; 95% CI, −0.89 to −0.43; P < .001).

Conclusions and Relevance In this randomized clinical trial, there was no significant difference in efficacy between this particular formulation of TTP and tretinoin 0.02%. However, the results of these mechanistic studies highlight MMP2 as a possible mediator of retinoid efficacy in photoaging.

Trial Registration ClinicalTrials.gov Identifier: NCT01283464

Retinoids are broadly accepted as the criterion standard for topical management of photodamage—improving the fine wrinkles and dyspigmentation characteristic of photoaged skin.^1-4 One of the major deterrents to their topical use, however, is the associated local skin irritation, such as erythema, pruritus, stinging, burning, and scaling.^5,6 In light of these toxic effects, efforts have been made to study the efficacy and tolerability of alternative forms of topical retinoids, particularly natural precursors of tretinoin, the bioactive metabolite of vitamin A. One such precursor is retinol, which generates tretinoin through sequential oxidation. Retinol can also be esterified with fatty acids and subsequently stored as retinyl esters like retinyl acetate and retinyl palmitate.⁷ Cosmeceutical products containing retinol and retinyl esters are widely available over the counter and may provide comparable efficacy as prescription-grade tretinoin.^8,9

However, it is still unclear which of the many purported effects of retinoids is directly responsible for the clinical efficacy of tretinoin or its precursors. For example, retinoids have been shown to both induce procollagen I synthesis and suppress expression of matrix metalloproteinases (MMPs; eg, MMP1, MMP3, and MMP9), ultimately culminating in the restoration of collagen.^9-12 From this molecular understanding, however, it has largely been assumed that these mechanisms are not only clinically relevant—directly contributing to the effacement of wrinkles—but also completely dependent on retinoic acid receptor (RAR) signaling. Indeed, to our knowledge, no study has directly assessed these assumptions by correlating the clinical improvements observed with retinoid therapy with parallel changes in molecular targets in the skin. Clearly addressing these questions could help identify specific molecular targets driving therapeutic benefits and guide future drug development.

In light of this knowledge gap, we designed an exploratory randomized clinical trial evaluating a topical formulation of 1.1% triple tretinoin precursors (TTP)—containing retinol, retinyl acetate, and retinyl palmitate—and topical 0.02% tretinoin (RA) for treating moderate to severe photodamage over 24 weeks. Through this study, we primarily sought to identify specific drivers of retinoid efficacy by analyzing target gene expression alongside clinical efficacy.

Patients with at least moderate facial photodamage (defined as a score of 4 or greater on the 9-point Griffiths scale for photoaged skin¹³) were enrolled in the single-center, randomized, parallel-arm study. Screening and enrollment were performed at the Johns Hopkins Outpatient Center in Baltimore, Maryland, with the following inclusion criteria: age 35 years and older, Fitzpatrick skin phototype I to V, good general health, ability to undergo study biopsies, and willingness to use only the provided facial skincare products and study drugs for the duration of the study. Exclusion criteria included preexisting or dormant dermatologic conditions; cosmetic procedures performed on the treatment areas; oral retinoid or steroid medication use within 6 months before randomization; use of topical retinoid, salicylic acid, lactic acid, or α-, β-, or poly-hydroxy acid product within 3 months before randomization; and sensitivity to cosmetic products or components of the study drugs. Dermatologic treatments, procedures on the face, and excessive sun exposure were not permitted during the study. Pregnant and lactating individuals were not enrolled. Self-reported race and ethnicity information was collected and used as part of the analysis to ascertain whether these factors played a role in the responses observed.

The protocol (Supplement 1) was approved by the Johns Hopkins Medicine Institutional Review Board, and written informed consent was obtained from all participants. The study was registered with ClinicalTrials.gov (NCT01283464) and was conducted from November 2010 to December 2011, with data analysis performed from January 2012 to December 2021. This study followed the Consolidated Standards of Reporting Trials (CONSORT) reporting guideline.

Participants were randomized with 1:1 allocation ratio by the Johns Hopkins Hospital Investigational Drug Service using a computerized pseudorandom number generator. A 1.1% TTP cosmetic product provided by SkinMedica (Tri-Retinol Complex ES containing 1% retinol, 0.05% retinyl acetate and 0.05% retinyl palmitate) and 0.02% tretinoin (Renova) were investigated in this study and dispensed by the Investigational Drug Service. In both treatment groups, the study drug was a white cream. Participants and research staff who evaluated participants were blinded to treatment allocation.

Patients applied the study medication to the entire face every other evening for the first 4 weeks and then every evening for weeks 5 through 24. If this dosage was not tolerated at any time during the study, the frequency of usage could be temporarily reduced to every other day and then subsequently increased. All patients were required to maintain a usage diary. They were also encouraged to use sunscreen daily and instructed to wash and dry their faces before applying a thin layer of the study medication. Facial cleanser, moisturizer, and sunscreen with a sun protective factor of 30 were provided to all patients. Moisturizer, if used, was applied at least 1 hour before or after application of the study medication.

At baseline and at 4, 12, 18, and 24 weeks of the study, digital photographs were taken, and a treatment-blinded dermatologist performed in-person evaluations, collecting the following data: degree of facial photodamage, signs of cutaneous irritation (ie, erythema, edema, dryness/scaling), global assessments of improvement, patient-reported outcomes (ie, self-assessments of efficacy, tolerability, quality of life, and satisfaction), and adverse events. All participants were monitored for serious adverse events.

A 9-point scale for photoaged facial skin was used to evaluate coarse and fine lines, pigmentation and yellowing, and overall clinical severity of photodamage at each study visit.¹³ Overall scores for coarse and fine wrinkles were calculated by totaling the individual scores for each region of the face (periocular, forehead, cheek, perioral). Additionally, a dermatologist assigned a Griffiths photodamage score and a global photodamage improvement score to each participant to evaluate overall improvement from baseline. To compare relative tolerability of TTP and RA, treatment-blinded investigators made objective assessments of irritation (erythema, edema, and dryness/scaling).

At baseline and 6 months after trial initiation, two 3-mm skin biopsies were obtained from the same sun-exposed periocular area of the face. Posttreatment skin samples were taken at least 2 cm away from the baseline biopsy site to avoid sampling scar tissue. Biopsies were embedded in Tissue-Tek OCT compound (Sakura Finetek), snap frozen in liquid nitrogen, and stored at −80 °C until being analyzed.

Total RNA was isolated from frozen OCT-embedded biopsy tissues with RNeasy Fibrous Tissue Mini Kits according to the manufacturer’s protocol (Qiagen) and transcribed to cDNA using iScript cDNA Synthesis Kit (BioRad). Real-time quantitative polymerase chain reaction was carried out with iTaq Universal SYBR Green Supermix (BioRad) using the following primers: HPRT-F: 5′-CCTGGCGTCGTGATTAGTGAT-3′, HPRT-R: 5′-AGACGTTCAGTCCTGTCCATAA-3′, CRABP2-F: 5′-CTCAAAGTGCTGGGGGTGAA-3′, CRABP2-R: 5′-TGATCTCCACTGCTGGCTTG-3′, COL1A1-F: 5′-GGTCAGATGGGCCCCCG-3′, COL1A1-R: 5′-GGCAGCACCAGTAGCACC-3′, MMP1-F: 5′-ATGCTGAAACCCTGAAGGTG-3′, MMP1-R: 5′-GAGCATCCCCTCCAATACCT-3′, MMP2-F: 5′-CGCTACGATGGAGGCGCTAA-3′, MMP2-R: 5′-AGAAGGTGTTCAGGTATTGCACTG-3′, MMP3-F: 5′-GTCTCTTTCACTCAGCCAAC-3′, MMP3-R: 5′-ATCAGGATTTCTCCCCTCAG-3′, MMP9-F: 5′-GGCAGCTGGCAGAGGAATAC-3′, MMP9-R: 5′-GGCCCCAGAGATTTCGACTC-3′.

Week 24 measurements of Griffiths photodamage score served as the primary outcome of the study. Secondary outcomes included all other clinical measurements of photoaging at week 24, frequency of adverse events, and target gene expression.

Patients with both baseline and posttreatment clinical measurements and biopsies were included in the analysis. Unpaired Mann-Whitney tests were conducted for clinical outcomes at week 24. For finding correlates of clinical efficacy, baseline and week 24 clinical outcomes were analyzed using Wilcoxon matched-pairs signed rank test, while clinical and gene expression changes were correlated with Spearman correlation. Analyses were performed with GraphPad Prism, version 9.0.0 (GraphPad Software), and statistical significance was taken at the 5% level. Exact CIs for nonparametric tests were calculated assuming symmetric distributions and may not directly reflect test statistics.

Twenty-four White participants (2 men and 22 women) aged 40 to 84 years (mean [SD], 62 [9.5] years) with Fitzpatrick skin phototypes I through III were included in the study and randomized to a treatment group (Figure 1). Twenty patients completed the entire course of treatment and contributed both baseline and posttreatment biopsies. Enrollment of 3 patients was discontinued owing to nonadherence with usage instructions. One patient withdrew consent after experiencing a non–study-related broken hip during the trial. There were no significant differences between treatment groups for age, sex, race and ethnicity, Fitzpatrick skin type, or photoaging severity (Table).

To assess clinical efficacy, we compared week 24 Griffiths measurements of photoaging and found that there were no statistically significant differences between the 2 treatment arms (TTP − RA median difference: −1; 95% CI, −2 to 1; P = .27) (Figure 2A-E), which is consistent with prior studies.^8,9 Next, to assess the tolerability of TTP vs RA, treatment-blinded investigators assessed patients’ skin for local skin irritation and collected subjective reports of burning, itching, tightness, or peeling. Patients treated with RA exhibited erythema 6 times more frequently during the course of treatment than did patients in the TTP group (TTP − RA difference: −0.53; 95% CI, −0.88 to −0.17; P = .01) (Figure 2F). No significant differences in other objective or subjective measures were found between the 2 treatment groups (edema TTP − RA difference: 0; 95% CI, 0 to 0; dryness: −0.08; 95% CI, −0.51 to 0.35; P = .71; burning: 0.32; 95% CI, −0.08 to 0.72; P = .14; itching: 0.15; 95% CI, −0.23 to 0.53; P = .44; tightness: 0.41; 95% CI, −0.02 to 0.81; P = .06; peeling: 0.05; 95% CI, −0.33 to 0.43; P = .79) (Figure 2F-G). At the end of the study, 8 of 9 (89%) respondents using TTP and 8 of 11 (73%) using RA indicated that they would be interested in continuing use of the product.

We were next interested in which molecular targets best correlate with improvements in photodamage. To this end, we first performed pairwise analyses comparing patients’ posttreatment evaluations of photodamage with their respective baseline measurements. We found that patients using TTP exhibited statistically significant decreases in Griffiths photodamage score (week 24 − baseline median difference: −1; 96% CI, −2 to 0; P = .04) (Figure 2E). Patients in the RA group did not show statistically significant differences in this metric (week 24 − baseline: 0; 99% CI, −3 to 0; P = .09) (Figure 2E). Based on this clinical analysis, we predicted that analogous pairwise comparisons of potential biomarkers should reveal significant and consistent changes in mRNA expression, at least with TTP treatment.

To first confirm engagement of RARs, we quantified mRNA expression of cellular retinoic acid binding protein 2 (CRABP2), which is induced with RAR activation in the regulation of skin growth and differentiation.¹⁴ As anticipated, CRABP2 mRNA was significantly induced across both groups compared with pretreatment levels (TTP week 24 − baseline median difference: 847; 96% CI, 394 to 2637; P = .004; RA week 24 − baseline: 399; 99% CI, −205 to 1162; P = .01) (Figure 3A), confirming that both retinoids functionally activate RAR signaling.

Given that wrinkles arise from the net loss of collagen, we evaluated whether TTP could either induce procollagen I (promoting collagen formation) or suppress MMPs (preventing collagen degradation). We were surprised to find that TTP did not significantly induce mRNA expression of procollagen I mRNA (week 24 − baseline: 9.4; 96% CI, −180 to 191; P = .82) (Figure 3B) or suppress MMP1 (week 24 − baseline: 3.6; 96% CI, −14 to 15; P = .50), MMP3 (week 24 − baseline: 0.79; 96% CI, −1.2 to 8.4; P = .43), or MMP9 (week 24 − baseline: −3.1; 96% CI, −11.3 to 2.8; P = .36) (Figure 3C-E), arguably 3 of the most well-studied MMPs involved in photoaging.¹⁵ Similarly, RA treatment did not significantly modulate the expression of these targets with the exception of MMP9 (COL1A1 week 24 − baseline: 19.6; 99% CI, −306 to 421; P = .64; MMP1: 16.8; 99% CI, −30.9 to 57.5; P = .41; MMP3: 2.8; 99% CI, −12.5 to 37.6; P = .76; MMP9: −6.1; 99% CI, −25.3 to −0.8; P = .003) (Figure 3B-E).

In the absence of a clear molecular explanation for the observed clinical efficacy of TTP, we extended our analyses to MMP2, which degrades type IV collagen present at the dermal-epidermal junction. Recently reported to be induced by UV-B–mediated DNA damage,¹⁶ MMP2 and type IV collagen degradation—but not type I collagen degradation—closely correlate with wrinkle status in chronically UV-exposed mouse skin.^17,18 This observation—alongside the fact that MMP2 is far more abundantly expressed than MMP1/3/9 and steadily increases with age in sun-exposed skin but not sun-protected skin¹⁶—led us to interrogate MMP2 as a possible target gene. Doing so, we found that MMP2 mRNA was significantly suppressed in TTP-treated samples (week 24 − baseline: −5; 96% CI, −33 to 1.6; P = .02) but remained unchanged in RA-treated samples (week 24 − baseline: 1.6; 99% CI, −27 to 11; P = .58) (Figure 3F), mirroring our prior clinical analysis (Figure 2E).

To evaluate whether MMP2 modulation might underlie the clinical effects of retinoids, we calculated Spearman coefficients between patients’ changes in clinical metrics and MMP2 mRNA. We found that changes in MMP2 expression strongly correlated with changes in fine wrinkles overall (r = 0.54; 95% CI, 0.12 to 0.80; P = .01) and in periocular fine wrinkles (r = 0.44; 95% CI, −0.01 to 0.75; P = .049) (Figure 4A-B). That is, patients with greater improvements in wrinkles had more dramatic suppression of MMP2 (Figure 4A-B). Furthermore, trends within each of the treatment groups (Figure 4A-B, orange and light blue lines) closely mirrored the overall trend (black line), suggesting that this relationship may broadly apply to retinoids. In contrast to MMP2, changes in CRABP2 did not share any meaningful correlations with changes in periocular fine wrinkles (r = −0.02; 95% CI, −0.47 to 0.44; P = .94) or any other clinical metrics (Figure 4C, data not shown).

Interestingly, we also found that the initial severity of wrinkles at baseline strongly predicted their ultimate response to retinoid therapy (fine wrinkles overall: r = −0.74; 95% CI, −0.89 to −0.43; P = .002; periocular fine wrinkles: r = −0.91; 95% CI, −0.97 to −0.78; P < .001) (Figure 4D). The resulting linear models’ 95% CIs for x-intercepts (x-int) unexpectedly did not include the origin (fine wrinkles overall x-int: 13.5; 95% CI, 10.1 to 15.5; periocular fine wrinkles x-int: 3.9; 95% CI, 3.5 to 4.3), suggesting that there may be a nonzero threshold of initial wrinkles below which retinoid therapy might not be beneficial. Repeating this analysis for mRNA expression, we found that MMP2 (r = −0.78; 95% CI, −0.91 to −0.50; P < .001; x-int = 11.4; 95% CI, 7.8 to 14.3), but not CRABP2 (r = −0.06; 95% CI, −0.50 to 0.41; P = .81), replicated these trends (Figure 4E-F). In summary, these data suggest that MMP2 may play a role in the therapeutic effect of retinoids.

In this exploratory study, we first evaluated the efficacy of TTP and RA in treating photodamaged skin and found no significant differences in week 24 measurements of photoaging between the 2 treatment arms, corroborating past studies.⁸ Given the main limitations of small sample size and a predominantly White population, TTP use would still benefit from further investigation in larger, more diverse cohorts where parametric methods can be used.

Interestingly, despite TTP having a higher concentration of tretinoin equivalents than RA (1.1% vs 0.02%), we observed that patients receiving TTP experienced erythema 6 times less frequently than patients treated with RA (Figure 2F), consistent with a prior study comparing 1.6% retinol and 0.025% RA.¹⁹ There are at least 2 possible explanations for this apparent discrepancy. First, given the study’s limitation of not being able to control for vehicle differences, it is very possible that TTP vehicle itself actively ameliorates the adverse effects of tretinoin. Another possibility, which is not necessarily mutually exclusive, is that precursors present in TTP (retinol, retinyl acetate, and retinyl palmitate) may have unique functions independent of tretinoin conversion that may be counteracting the toxic effects of RA. Future studies can help identify these novel properties of tretinoin precursors and vehicle ingredients that may minimize local toxic effects of retinoids.

In addition to assessing RAR activation in TTP-treated and RA-treated skin samples, we also analyzed which target gene most closely correlates with clinical improvements in photoaging. We found that expression levels of procollagen I and MMP1, MMP3, and MMP9 were not meaningfully altered (Figure 3A-E), despite the clinical efficacy of TTP treatment (Figure 2E). One possibility is that UV from continued sun exposure on the face could have reduced procollagen I levels,²⁰ negating the action of retinoids. Consistent with this idea, studies demonstrating strong induction of procollagen I have either occluded the skin after retinoid treatment or have involved skin that is arguably less sun-exposed than the face (for example, the forearm).^9,21 Regardless of the underlying reason, the fact that we observed significant clinical improvements despite unimpressive changes in procollagen I and MMP1, MMP3, and MMP9 suggests that a mechanism independent of collagen synthesis or suppression of characteristic MMP activity may be driving clinical efficacy.

Our analysis suggests that MMP2, a type IV collagenase, may represent one possible mechanism. We found that MMP2 mRNA expression was not only suppressed in TTP samples (Figure 3F) but also, at a more granular level, correlated with clinical improvements in fine wrinkles—a result not observed in any other evaluated gene, including CRABP2 (Figure 4A-C). These unexpected findings have important implications for understanding the therapeutic mechanism of retinoids. First, the fact that the semiquantitative clinical scores would meaningfully correlate with any genes bolsters the validity of these grading techniques. Second, these results suggest that retinoid efficacy may depend more heavily on preventing collagen degradation, specifically type IV collagen, rather than on inducing type I collagen. Lastly, our study highlights a potentially expanded role for MMP2 in photoaging, extending findings from mice that MMP2 and type IV collagen degradation correlate with wrinkle formation.^16,17 Interestingly, the fact that MMP2, but not CRABP2, correlates with wrinkle improvement suggests that MMP2 may be suppressed independently of RAR activation. Future studies can explore the role that MMP2 may play in photoaging and assess its potential as a therapeutic target.

Finally, we found that the initial severity of fine wrinkles in patients strongly predicted their subsequent improvements in response to retinoid therapy (Figure 4D). Although this relationship has not been explicitly described in literature, it has often been anecdotally appreciated among clinicians. What was particularly surprising, however, was that retinoid therapy was not helpful—or was even counterproductive—in patients below a baseline threshold for fine wrinkles and MMP2 expression (Figure 4D-E), according to our linear regression models. This clinical relationship, if true, would certainly influence the calculus of when to prescribe retinoids. Larger trials should be performed to validate this finding and identify other potential predictors of poor clinical outcome.

As noted above, the present study has several limitations. First, the sample numbers for this pilot study were relatively limited, and nonparametric methods for statistical analysis were subsequently used. Second, all of the analyzed patients were White women despite our best efforts to diversify enrollment. In light of our inclusion criteria for moderate to severe photodamaged skin, this result likely reflects the relative protection against photoaging afforded by darker skin types.^22,23 To address this issue, we are actively pursuing studies that specifically examine photoaging in skin of color. Lastly, our study did not control for vehicle differences between the 2 treatment groups. While this may limit conclusions drawn about relative clinical efficacy and toxic effects, the biomarker analysis does not appear to be significantly affected, as intragroup correlations of MMP2 were comparable (Figure 4). Nevertheless, future studies evaluating different retinoid products should attempt to control for vehicle differences wherever possible.

In this randomized clinical trial, there was no significant difference in efficacy between this particular formulation of tretinoin precursors and tretinoin 0.02%. While TTP induced CRABP2 mRNA, confirming RAR activation, no significant changes were detected in procollagen I, MMP1, MMP3, or MMP9 mRNA. Instead, we found that MMP2 expression was significantly decreased in TTP-treated samples and closely correlated with clinical improvements in fine wrinkles. These results highlight the potential role that MMP2 may play in mediating retinoid efficacy in photoaging. Larger studies with more diverse patients should be pursued to validate these findings in a broader population.

Accepted for Publication: March 30, 2022.

Published Online: June 8, 2022. doi:10.1001/jamadermatol.2022.1891

Corresponding Author: Anna L. Chien, MD, Johns Hopkins School of Medicine, Department of Dermatology, 601 N Caroline St, Ste 8060C, Baltimore, MD 21287 (achien3@jhmi.edu).

Author Contributions: Drs Chien and Kim had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Drs Chien and Kim contributed equally to this article.

Concept and design: Chien, Kim, Cheng, Zang, Wallis, Loss, Kang.

Acquisition, analysis, or interpretation of data: Chien, Kim, Cheng, Shin, Leung, Nelson, Suh, Rainer, Okoye, Kang.

Drafting of the manuscript: Chien, Kim, Wallis, Kang.

Critical revision of the manuscript for important intellectual content: Chien, Kim, Cheng, Shin, Leung, Nelson, Zang, Suh, Rainer, Okoye, Loss, Kang.

Statistical analysis: Chien, Kim, Kang.

Obtained funding: Chien, Kim, Kang.

Administrative, technical, or material support: Chien, Kim, Cheng, Leung, Nelson, Zang, Suh, Okoye, Kang.

Supervision: Chien, Loss, Kang.

Conflict of Interest Disclosures: Dr Chien reported grants from SkinMedica during the conduct of the study; and grants from Amorepacific, Cearna, Inc, and Boots-Walgreens outside the submitted work. Dr Kim was a Paul and Daisy Soros Fellow and was supported in part by a grant from the National Cancer Institute of the National Institutes of Health (F30CA236466) and an MSTP training grant from the National Institutes of Health (T32GM007205) during the conduct of the study. Dr Okoye reported grants from Pfizer, and personal fees from Janssen, UCB, Novartis, and Unilever outside the submitted work. Dr Loss reported grants from SkinMedica during the conduct of the study. Dr Kang reported grants from SkinMedica during the conduct of the study; other (advisory board) from Allergan, Galderma, CeraVe, and Unilever outside the submitted work; in addition, Dr Kang had a patent for epidermal growth factor receptor inhibition for retinoid adverse effect prevention with royalties paid from Access Business Group, a patent for novel skin coating for UV protection pending, a patent for use of composition for treating rosacea issued, and a patent for composition and methods for use against acne inflammation issued. No other disclosures were reported.

Funding/Support: This study was funded by SkinMedica.

Role of the Funder/Sponsor: SkinMedica had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Data Sharing Statement: See Supplement 2.

Additional Contributions: We also thank Nathan Archer, PhD, Luis Garza, MD, PhD, and their laboratories for their assistance interpreting data and providing necessary resources. We thank Akiko Iwasaki, PhD, Barbara Kazmierczak, MD, PhD, Fred Gorelick, MD, Peter Aronson, MD, Reiko Fitzsimonds, PhD, Alex Mauzerall, BA, and Cheryl DeFilippo, JD, for their guidance and support. We thank Dustin Dikeman, MA, Yulia Wang, PhD, Ruizhi Wang, MD, MPhil, and Gloria Seho-Ahiable, MS, for their administrative assistance. These individuals were not compensated for this work.