A Randomized Double-blind Placebo-controlled Trial of Oral Aspirin for Protection of Melanocytic Nevi Against UV-induced DNA Damage

The incidence of cutaneous melanoma has been rapidly rising over the past few decades (1) due to many factors including increased exposure to ultraviolet (UV) radiation (2). A randomized controlled trial in Australia demonstrated a 50% reduction in melanoma in sunscreen users (3) and use of sunscreen currently represents the most effective means of primary prevention. However, given that sunscreen is often inadequately applied (4), there would be great value in a systemic adjunct to sunscreen, particularly for high-risk subjects. Several systemic agents have been considered for melanoma chemoprevention, but none have been validated in prospective phase III trials in humans (5) and long-term administration of any drug may be associated with unexpected toxicities.

Although the relationship between UV exposure and melanoma development has long been established (6), the underlying mechanisms through which UV causes melanocyte transformation and tumor development have not been clearly defined. Acute UV exposure induces inflammation, reactive oxygen species, cytokine activation, and DNA damage which have been associated with melanoma development in mouse models (7–9). We suggested over a decade ago (10) that use of a systemic agent limited to periods immediately preceding or during sun exposure that targets UV-induced pathways could counter the cumulative deleterious effects of UV exposure while avoiding the pitfalls of chronic chemoprevention.

Aspirin (ASA) is a safe drug with both anti-inflammatory and antioxidant properties (11) that could potentially be suitable for this purpose. Although there are conflicting epidemiologic studies on the association of ASA use and melanoma risk in humans (12), the potential protective effects of ASA have not been evaluated prospectively in high-risk patients in the context of acute UV exposure. We recently reported that ASA can suppress acute UV-induced inflammation and DNA damage in mouse skin (13). In prior open-label studies in human subjects, we detected ASA metabolites and reduced levels of prostaglandin-E₂ (PGE₂) in plasma and melanocytic nevi following conventional daily ASA dosing (14). Here, we conducted a randomized placebo-controlled trial to determine if ASA could suppress solar-simulated UV (SSUV)-induced inflammation in nevi and protect against SSUV-induced DNA damage.

This study was approved by the Institutional Review Board (IRB #94424) of the University of Utah and conducted in accordance with recognized ethical guidelines (e.g., Declaration of Helsinki). All participants signed an IRB-approved consent form. They were recruited by D.G. from his pigmented lesion clinic at the Huntsman Cancer Institute, in which those with history of numerous or atypical nevi, and/or personal or family history of melanoma are regularly monitored. Subjects that were under age 18, critically ill or mentally handicapped, prisoners, pregnant or breast-feeding, non-English speaking, having history of severe asthma or allergic reaction to ASA, history of bleeding disorder or peptic ulcer disease, intense UV exposure (e.g., tanning bed use) in the preceding month, or those having taken ASA, any non-steroidal anti-inflammatory drug (NSAID), or blood thinner in the preceding two weeks were excluded (Fig. 1A). Females of child-bearing potential had a confirmed negative urine pregnancy test. Subjects were not charged for removal or histological examination of their nevi, and each subject was compensated $400 following their participation. The clinical trial was registered at ClinicalTrials.gov (identifier: NCT04066725) and was conducted in accordance with CONSORT guidelines.

The minimal erythemal dose (MED) needed to produce skin erythema at 24 hours was determined for each subject both prior to and at the end of the trial using a 16S-300–003 solar simulator calibrated with a PMA2100 radiometer (SolarLight, Glenside, PA) as described previously (15). The emission spectrum of the device encompasses wavelengths from 295 to 400 nm, and is weighted to mimic the spectrum of sunlight.

There were four study visits for each subject. At the first visit (day 1), after obtaining informed consent, MED testing was initiated and the first blood sample was obtained. At the second visit (day 2), the MED was determined and unlabeled capsules were provided. At the third visit (approximately day 28), repeat MED testing was initiated and one nevus was SSUV-treated. At the fourth visit (day after 3^rd visit), repeat MED was determined, a second blood sample was obtained, and both the UV-treated and an unirradiated nevus were removed.

Placebo- and ASA-containing capsules were prepared by the Huntsman Cancer Institute investigational pharmacy. Avicel PH105 microcrystalline cellulose NF powder (Letco, NDC 6299120072) was used to fill “00” capsule shells (DRK GRN, NDC 6299141215) for the placebo. Single ASA tablets containing either 81 mg (Rugby, NDC 00536123441) or 325 mg (Bayer, NDC 00284211010) were added along with enough Avicel to fill the remaining capsules. The tops of the capsules were locked into place and added to plastic bottles labeled by subject number according to a randomization scheme provided by the statistician (K.M.B.). The bottles were stored at room temperature. The investigators and subjects were blinded as to the contents of each numbered bottle. Subjects were asked to take a capsule each evening at bedtime, and provided a log sheet to document date and time of each dose. They were contacted each week to confirm adherence and inquire about potential side effects. The last study visit occurred approximately 7–14 hours after the last dose.

Blood was collected from each subject at the start and end of the trial in EDTA-containing vacutainers, and plasma was isolated and stored at −80°C as described previously (14).

Two nevi (>5 mm in diameter) were selected on each subject. These were lesions not clinically suspicious for melanoma by visual inspection and dermoscopy and/or confirmed to be stable by comparison to the subject's baseline photographs used for monitoring during regular clinic visits. At the end of the trial, one nevus was SSUV-irradiated at a dose equivalent to 1 (pink nevi) or 2 (brown nevi) MED (based on initial MED determination for each subject). Approximately 24 hours later, both the unirradiated and SSUV-treated nevus were removed as described previously (16). A representative 1 mm slice from each nevus was placed in 10% formalin for paraffin embedding and sectioning. The remaining nevus fragments were stored at −80°C. A hematoxylin and eosin (H&E)-stained section was later reviewed by a dermatopathologist (S.R.F.) to confirm the lesion was a nevus and not melanoma. The presence of dysplasia was also noted. Additional unstained sections were prepared for immunohistochemistry.

Metabolites of ASA were detected and quantitated in plasma samples by liquid chromatography-mass spectrometry as previously described (14).

PGE₂ was quantitated in plasma and nevus specimens by ELISA using a PGE₂ assay kit as described previously (14).

Sunburn cells were quantitated in H&E-stained sections by light microscopy as previously described (17). Inflammatory cell infiltrates and expression of markers of DNA damage was determined by immunohistochemistry on nevus sections using mouse anti-myeloperoxidase (clone MPO-7, DAKO), rabbit anti-CD3 (Abcam ab5690), mouse anti-CD4 (clone 4B12, Leica Biosystems), mouse anti-CD8 (clone 4B11, Leica), mouse anti-CD163 (clone 10D6, Leica), mouse anti-CD20 (clone L26, Leica), mouse anti-8-oxoguanine (8-OG, Genox Corp.), and mouse anti-cyclobutane pyrimidine dimer (CPD, Cosmo Bio) antibodies as described previously (13).

A biostatistician (K.M.B.) constructed the subject randomization scheme, and performed all statistical analyses using “R” software (Vienna, Austria). Analyses of MED, PGE₂, and sunburn cells was performed on the log scale. Because some sunburn cell counts were equal to zero, 0.5 was added to all sunburn cell counts before taking the logarithm. The results were converted back to the original scale for plotting and interpretation. Paired t-tests were used for differences in PGE₂ within a single treatment group, and Welch's two-sample t tests were used for differences in PGE₂ between treatment groups. Additionally, a combined analysis adjusted for nevus color using linear regression on the log-transformed values with color as an adjustment variable was employed. Multiple regression was used to analyze differences in sunburn cells between treatment groups, adjusting for nevus color (and SSUV dosage). Analysis of covariance with treatment group as fixed effect and pre-treatment value as a continuous adjustment variable was used to analyze differences in post treatment MED and serum PGE₂. P-values from Wald tests are reported for the analysis of covariance and regression models. T tests were used to analyze 8-OG, and the results are presented on the original scale. Analysis was also done on the log scale and with an analysis of covariance model for the difference between the SSUV-irradiated and unirradiated nevi with similar results. For the CPD data, we used non-parametric Wilcoxon tests for pairwise comparisons because the zero values seen in many of the unirradiated nevi violates normality assumptions required for a t-test. For paired analyses of CPD data, zero values were replaced by 0.5 (half of limit of detection). P values ≤ 0.05 were considered statistically significant.

A total of 95 eligible subjects were randomized to receive placebo (n = 32), 81 mg ASA (ASA-81, n = 32), or 325 mg ASA (ASA-325, n = 31) daily for one month (Fig. 1A). There were no significant differences in age, sex, personal of family history of melanoma, body surface area, or hair and eye color among the cohorts (Table 1). Four subjects assigned to the placebo group withdrew due to unrelated events or exposure to ASA or NSAIDs in violation of protocol (Fig. 2). The remaining 91 subjects completed the study. The rate of reported side effects was 3.6%, 6.3%, and 9.7% in the placebo, ASA-81 and ASA-325 cohorts, respectively. One subject developed transient vision loss and withdrew from the study. The other side effects reported were transient auditory symptoms, sleep disturbance, stomach upset, heartburn and bruising, and their incidence is noted in Fig. 2). No significant levels of ASA metabolites were detected in plasma samples from subjects in the placebo cohort, or in pre-trial samples from subjects in either ASA cohort (Fig. 1B and C), confirming subject compliance with protocol. Post-trial plasma samples from the ASA cohorts revealed dose-dependent levels of salicylate (Fig. 1B) and salicylurate (Fig. 1C). Gentesate and salicylacyl glucoronide were not detected.

There was considerable variation in MED values among the subjects, and between the pre-trial and post-trial measurements for each subject (Fig. 3A). The MED increased by approximately 15%–17% in all the subjects, but paired analyses of pre-trial and post-trial values did not reveal significant changes in MED within any cohort or between either ASA cohort and placebo (P = 0.7 for all comparisons).

Plasma PGE₂ levels also varied considerably among all the subjects, but paired analyses comparing post- and pre-trial samples revealed significant reductions of 35–40% for plasma PGE₂ in both ASA cohorts but not in the placebo cohort (Fig. 3B). There was not a significant difference in percent change between the two ASA cohorts (P = 0.69). For nevi, while PGE₂ levels were increased by 1.3%, 8.8%, and 1.6% between paired unirradiated and UV-treated nevi in the placebo, ASA-81 and ASA-325 cohorts, respectively (Fig. 3C), none of these UV-induced increases were significant (P = 0.3–0.9). Similarly, no differences between paired unirradiated and UV-treated nevi were seen among the placebo, ASA-81 and ASA-325 groups when the analysis was adjusted for nevus color with adjustment variables brown and pink (P = 0.19, 0.97 and 0.86, respectively). On the other hand, PGE₂ was significantly reduced among both the unirradiated nevi and UV-treated nevi in both of the ASA cohorts compared to the placebo cohort (Fig. 3D). Reductions in PGE₂ were greater among unirradiated nevi (P = 0.07) and UV-treated nevi (P = 0.04) in the ASA-325 cohort compared to the ASA-81 cohort. Reductions in PGE₂ levels in both SSUV-treated (ASA-81 and ASA-325 vs. placebo, P < 0.0001) and unirradiated (ASA-81 vs. placebo, P = 0.001; ASA-325 vs. placebo, P < 0.0001) nevi remained significant when adjusted for nevus color. Thus, while PGE₂ levels in nevi were not significantly affected by UV treatment, PGE₂ was suppressed in both plasma and all the nevi in subjects receiving ASA.

Sunburn cells undergoing SSUV-induced apoptosis (18) were generally not seen in unirradiated nevi and consistently observed in SSUV-treated nevus specimens. However, there were no significant differences between SSUV-treated nevi in comparing the ASA-81 (P = 0.8) or ASA-325 (P = 0.55) cohorts with the placebo cohort (Fig. 4A). Since brown nevi were treated with twice the MED-equivalent SSUV dose than pink nevi, we performed a stratified analysis for this covariate and again found no significant differences in brown or pink UV-treated nevi between the ASA-81 (P = 0.42 and 0.19, respectively) or ASA-325 (P = 0.98 and 0.37, respectively) cohorts with the placebo cohort.

In our prior studies in mouse skin, acute UV exposure resulted in a mixed inflammatory cell infiltrate composed of macrophages and lymphocytes (13). Most of the SSUV-treated nevi also demonstrated robust inflammatory cell infiltration of the dermis. We selected 6 different SSUV-treated nevi from the placebo group, and performed immunostains for myeloperoxidase and CD-163 to assess macrophage/monocyte and CD3, CD4, CD8, and CD20 to assess lymphocytes in the SSUV-induced inflammatory infiltrates. We found that the predominant inflammatory cell type was CD3⁺ lymphocytes (Supplementary Fig. S1), with a CD4:CD8 ratio of approximately 2:1 as in normal skin (19). Few or no cells expressed the B-cell marker CD20. Finally, there was very minimal staining for myeloperoxidase or CD-163, which was similar to that seen in unirradiated nevi. Thus, we focused our analysis of SSUV-induced inflammation to all nevus sections stained for CD3. As shown in Fig. 4B, both unirradiated and SSUV-treated nevi from each cohort expressed variable expression of CD3 cells and the average value was higher in SSUV-treated compared to unirradiated nevi within each cohort. SSUV-induced CD3 expression was 2–5 fold higher in SSUV-treated compared to unirradiated nevi in all cohorts, however, there was not significantly reduced UV-induced CD3 expression in either ASA cohort compared to placebo (Fig. 4B). In addition, no significant differences were observed in analyses restricted to brown or pink SSUV-treated nevi between the ASA-81 (P = 0.24 and 0.34, respectively) or ASA-325 (P = 0.24 and 0.81, respectively) cohorts and the placebo cohort.

Next, we examined expression of 8-OG and CPD in nevi as a measure of SSUV-induced DNA damage. Both unirradiated and SSUV-treated nevi from each cohort demonstrated variable expression of 8-OG and the average percent positive-staining cells was higher in SSUV-treated compared to unirradiated nevi within each cohort. Notably, there were not significant differences in 8-OG between unirradiated or SSUV-treated nevi in either the ASA-81 (P = 0.48 and 0.10, respectively) or ASA-325 (P = 0.29 and 0.17, respectively) cohort compared to placebo (Fig. 4C). UV-induced 8-OG was approximately 1.5-fold higher in SSUV-treated compared to unirradiated nevi in all cohorts, however paired analyses comparing unirradiated and SSUV-treated nevi did not demonstrate significant differences in SSUV-induced 8-OG in either the ASA-81 (P = 0.07) or ASA-325 (P = 0.37) cohort compared to placebo. For CPD, expression in both unirradiated and SSUV-treated nevi from each cohort was variable and the average percent positive-staining cells was higher in SSUV-treated compared to unirradiated nevi within each cohort. However, there were not significant differences in CPD between unirradiated or SSUV-treated nevi in either the ASA-81 (P = 0.46 and 0.25, respectively) or ASA-325 (P = 0.45 and 0.47, respectively) cohort compared to placebo (Fig. 4D). SSUV-induced CPD was approximately 10–15 fold higher in SSUV-treated compared to unirradiated nevi in all cohorts, however paired analyses comparing unirradiated and SSUV-treated nevi did not demonstrate significant differences in SSUV-induced CPD in either the ASA-81 (P = 0.25) or ASA-325 (P = 0.39) cohort compared to placebo. Thus, ASA administration did not appear to reduce SSUV-induced apoptosis, inflammation, or DNA damage in nevi under the conditions examined.

Finally, we reviewed all the SSUV-treated nevi for dysplasia and found the following proportions with dysplasia in the placebo 11/29 (38%), ASA-81 11/32 (34%), and ASA-325 11/31 (35%) groups. There was no significant difference between the groups (P = 0.96, Fisher's Exact test).

We previously demonstrated that orally-delivered ASA could reduce UV-induced sunburn cell formation, inflammation and DNA damage in mouse skin (13), and showed that a 1-week course of daily ASA could suppress PGE₂ in the plasma and nevi of human subjects (14). Despite achieving similar reductions in plasma and nevus PGE₂ however, we did not observe significant protection against these SSUV-induced sequelae in human nevi of subjects following ASA administration under the conditions examined. It is important to consider, however, a number of differences between the prior mouse study and this randomized controlled trial in human subjects.

First, there were differences in ASA dosing and the time intervals between ASA delivery, UV treatment, and harvest of tissue for analysis in the two model systems. In the prior mouse study, adult mice received 0.4 mg ASA daily by gavage (∼20 mg/kg) approximately 1–2 hours prior to UV exposure and 3–4 hours prior to skin preservation (13). In the present study, subjects in the highest dose group (ASA-325 cohort) took 325 mg ASA daily (∼3–6 mg/kg, dose not adjusted for weight) approximately 7–14 hours prior to SSUV treatment and removal of the nevi. It is possible that subjects in the present study were under-dosed (3–6 vs. 20 mg/kg) and too much time had expired between their last ASA dose and SSUV treatment (7–14 vs. 1–2 hours) and removal of nevi (7–14 vs. 3–4 hours). ASA is rapidly metabolized and not detectable in plasma (20), but examination of its immediate and most stable metabolite salicylate revealed mean levels of approximately 8000 (±2000) ng/mL in the ASA-325 cohort which was comparable to that measured in our prior open-label human study (14) and in adult mice 4 hours after ASA gavage in an earlier study (21). We have previously shown that plasma salicylate levels peak in humans about 4–8 hours after a single 325 mg ASA dose (14). The longer time intervals in the present trial between last ASA dose, SSUV treatment, and nevus harvest compared to the prior mouse studies were necessitated by when it was convenient for most subjects to return for clinic visits. While we could have provided subjects higher daily doses of ASA to achieve greater salicylate levels over a longer period of time, given the roughly 10% incidence of side effects reported in the ASA-325 cohort it is likely that higher doses would not have been tolerable in many subjects.

Despite these limitations, ASA dosing in this trial was associated with 30–50% reductions in both plasma and nevus PGE₂ levels in subjects randomized to ASA. This level of PGE₂ suppression was comparable to that seen in plasma and nevi in the prior open-label human study (14) and in plasma and skin of adult mice in the prior mouse study (13). The mechanism(s) by which ASA protects against UV-induced DNA damage may be multi-factorial and has not been clearly defined. We previously showed that ASA-mediated inhibition of melanoma cell motility and melanin production in vitro was mediated by inhibition of PGE₂ synthesis as addition of PGE₂ to the culture medium was sufficient to reverse these effects of ASA (21). While it is possible that the level of PGE₂ suppression achieved here was insufficient for protection of nevi, it is also conceivable that ASA-mediated protection of melanocytes and nevi involves PGE₂-independent mechanisms which is a subject that merits further investigation.

Additional considerations are the differences between mouse skin and human nevi, the sources of UV employed, and the time point for tissue examination following UV exposure. The epidermis is considerably thicker in human compared to mouse skin and nevi often extend deeply into the dermis (22). UVB-emitting bulbs were used in the prior mouse study, while in the present study we employed SSUV consisting of both UVB and UVA. UVA consists of lower energy but longer wavelengths that may cause less DNA damage but penetrate deeper into the dermis than UVB (23). In our prior mouse studies showing ASA-mediated reduction in UV-induced inflammation and DNA damage (13), the protective effect of ASA was seen at multiple time points (1, 6, 24, and 48 hours) following UV exposure. Since the current trial design involved a second measurement of the MED, which was read 24 hours after SSUV exposure, it was most practical to have the MED initiated on the same day as the SSUV treatment of nevi so both could be evaluated on the same next day (24 hours after SSUV). Thus the spectrum and time course for UV damage and the ability of ASA to protect against it may vary in the context of these different UV sources and tissues.

A prior chemoprevention metabolomics study of ASA in healthy subjects reported that a dosing regimen of 325 mg daily for two months was associated with approximately 20% reduction in plasma levels of the oncometabolite 2-hydroxyglutarate (24). We similarly performed metabolomics in our prior open-label study but did not recapitulate this finding in subjects given 325 mg daily for 1 week (14). We considered that although dosing for one week was sufficient for PGE₂ suppression, longer treatment regimens may be required for a wider spectrum of ASA effects. Thus, we conducted the present trial for a course of one month in balancing these considerations with the efforts associated with extended dosing regimens. It was not feasible to test multiple variables such as higher doses of ASA and duration and timing of drug administration relative to SSUV (before vs. after) exposure in a randomized controlled setting.

ASA remains a safe drug with demonstrated chemopreventive activity in subjects at increased risk for colon cancer (25), although its daily use is not recommended by the U.S. Preventive Services Task Force (26) in healthy subjects over age 60 given the small but increased risk of bleeding (27) and lack of overall survival benefit (28). Our initially proposed paradigm for melanoma chemoprevention, in which an anti-oxidant could be taken in anticipation of UV exposure to prevent UV-induced pro-carcinogenic oxidative damage (10), may be suitable for ASA but further studies are required to determine if alternate dosing regimens or evaluation of nevi at different time points following SSUV treatment could demonstrate protection against UV-induced inflammation and DNA damage.

No disclosures were reported.

N. Okwundu: Data curation, investigation, writing–review and editing. H. Rahman: Data curation, investigation, writing–review and editing. T. Liu: Data curation, writing–review and editing. S.R. Florell: Data curation, writing–review and editing. K.M. Boucher: Formal analysis, methodology, writing–review and editing. D. Grossman: Conceptualization, data curation, formal analysis, supervision, investigation, methodology, writing–original draft, writing–review and editing.

D. Grossman was supported by The Department of Dermatology at the University of Utah and the Huntsman Cancer Foundation.

The Department of Dermatology at the University of Utah and the Huntsman Cancer Foundation (D. Grossman)

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.