Aspirin and nonsteroidal anti-inflammatory drugs can prevent cutaneous squamous cell carcinoma: a systematic review and meta-analysis.

Accepted Article Preview: Published ahead of advance online publication Aspirin and Non-steroidal Anti-inflammatory Drugs Can Prevent Cutaneous Squamous Cell Carcinoma: A Systematic Review and Meta-Analysis Chiho Muranushi, Catherine M Olsen, Nirmala Pandeya, Ade`le C Green

Cite this article as: Chiho Muranushi, Catherine M Olsen, Nirmala Pandeya, Ade`le C Green, Aspirin and Non-steroidal Anti-inflammatory Drugs Can Prevent Cutaneous Squamous Cell Carcinoma: A Systematic Review and Meta-Analysis, Journal of Investigative Dermatology accepted article preview 18 December 2014; doi: 10.1038/jid.2014.531. This is a PDF file of an unedited peer-reviewed manuscript that has been accepted for publication. NPG are providing this early version of the manuscript as a service to our customers. The manuscript will undergo copyediting, typesetting and a proof review before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers apply.

Received 2 July 2014; accepted 15 September 2014; Accepted article preview online 18 December 2014

© 2014 The Society for Investigative Dermatology

Aspirin and Non-steroidal Anti-inflammatory Drugs Can Prevent Cutaneous Squamous Cell Carcinoma: A Systematic Review and Meta-analysis

Chiho Muranushi1, Catherine M Olsen2, Nirmala Pandeya1,2, Adèle C Green2,3

Author Affiliations: 1 School of Population Health, University of Queensland, Brisbane Queensland Australia 2 QIMR Berghofer Medical Research Institute, Brisbane Queensland Australia 3 Institute of Inflammation and Repair, University of Manchester, Manchester Academic Health Sciences Centre, Manchester, UK

Corresponding Author: Professor Adèle C. Green Cancer and Population Studies Group, QIMR Berghofer Medical Research Institute Locked Bag 2000 Royal Brisbane Hospital, QLD 4029, Australia Phone: +61 7 3362 0234 Fax: +61 7 3845 3503 Email: [email protected]

Short title: NSAIDS and cutaneous SCC: a meta-analysis

Abbreviations: CI, confidence interval; COX, cyclooxygenase; NSAID, non-steroidal antiinflammatory drug; OR, odds ratio; RCT, randomizes controlled trial; SCC, squaous cell carcinoma; UV, ultraviolet 1


ABSTRACT Non-steroidal anti-inflammatory drugs (NSAIDS) have received increasing attention as potential chemopreventive agents of skin cancer but evidence is inconsistent. To investigate whether use of aspirin and other NSAIDS reduce the risk of squamous cell carcinoma (SCC), we conducted a systematic review based on published epidemiologic studies, and calculated summary estimates for aspirin, non-aspirin NSAIDS and any NSAIDS use. Summary estimates from 9 studies (5 case-control, 3 cohort, and 1 intervention) indicated significantly reduced risks of SCC among users of non-aspirin NSAIDS (Odds ratio (OR) 0.85, 95% confidence interval (CI) 0.78-0.94) and among users of any NSAIDS (OR 0.82, 95%CI 0.71-0.94) compared with non-users with the effect seen particularly in those with previous actinic skin tumours. A reduced risk was also observed among aspirin users though with borderline statistical significance (OR 0.88 95%CI 0.75-1.03). There was significant heterogeneity between studies regarding SCC risk estimates for aspirin use and any NSAIDS use. These findings suggest that NSAIDS collectively have the potential to prevent the development of cutaneous SCC.

2


INTRODUCTION Cutaneous squamous cell carcinoma (SCC) is one of the most frequently occurring malignancies worldwide and in white populations SCC presents a significant public health burden(Lucas et al., 2008). Aspirin and other non-steroidal anti-inflammatory drugs (NSAIDS) have received much attention as potential chemopreventive agents for various cancers such as colon, breast, lung and prostate cancers (Harris et al., 2005), but evidence in relation to skin cancer is inconsistent.

It is plausible that NSAIDS protect against SCC because, chronic exposure to ultraviolet (UV) radiation, the main cause of SCC (Armstrong and Kricker, 2001), damages skin cells resulting in the release of multiple factors including cyclooxygenase (COX)-2 enzyme and its product, prostaglandin E2, that play key roles in inflammation (Fecker et al., 2007; Rundhaug et al., 2011). These factors also influence skin carcinogenesis (Aggarwal et al., 2009; Maverakis et al., 2010). Over-expression of COX-2 is consistently found in SCC of the skin (Athar et al., 2001; Buckman et al., 1998; Kuzbicki et al., 2011). Non-steroidal anti-inflammatory drugs (NSAIDS) inhibit COX-2 and suppress production of prostaglandins (Rundhaug et al., 2011) and thus may have chemopreventive effects on cutaneous SCC. A topical NSAID, diclofenac, has been shown to be an effective treatment for actinic keratoses (Martin and Stockfleth, 2012; Rivers and McLean, 1997), but consistent evidence of an association between oral NSAIDS use 3


including aspirin and SCC of the skin is lacking. A randomised controlled trial (RCT) reported a 60% reduced risk of SCC with the use of a COX-2 inhibitor, celecoxib, whilst a recent case-control study reported an increased risk of SCC with the use of aspirin or other NSAIDS (Asgari et al., 2010). A past meta-analysis that included studies published up to September 2012 reported no association between use of any NSAIDS (aspirin and other NSAIDS were not distinguished) and SCC (Odds Ratio (OR) 0.88 95% confidence interval (CI) 0.75-1.03) (Zhang et al., 2014). However there have been a number of recent relevant published epidemiological studies that this analysis did not include. To definitively answer the question whether use of aspirin and of non-aspirin NSAIDS can reduce the risk of developing cutaneous SCC, we systematically reviewed all relevant published studies and synthesized the evidence available on the topic.

RESULTS Study Selection and Study Characteristics The collective search terms identified 286 papers (Figure 1). After 49 duplicates were removed, 237 titles and abstracts were reviewed and 11 full articles were retrieved for review (Amari et al., 2011; Asgari et al., 2010; Butler et al., 2005; Clouser et al., 2009; Cook et al., 2005; Elmets et al., 2010; Grau et al., 2006; Jeter et al., 2012; Johannesdottir et al., 2012; Nunes et al., 2011; Torti et al., 2011). Two reports identified 4


did not contain sufficient information to meet the inclusion criteria (Amari et al., 2011; Cook et al., 2005) since one reported an effect estimate for keratinocyte cancers collectively (Amari et al., 2011), and the second reported estimates for total cancer and some major cancers excluding SCC (Cook et al., 2005), leaving 9 studies providing detailed data for inclusion in the meta-analysis. All 9 reported a risk estimate of the association between aspirin or non-aspirin NSAIDS and histologically confirmed SCC (Tables 1 and 2), and all were published in English. One study was a randomized controlled trial (RCT) (Elmets et al., 2010), 3 were cohort (Clouser et al., 2009; Jeter et al., 2012; Nunes et al., 2011) and the remaining 5 were case-control studies (Asgari et al., 2010; Butler et al., 2005; Grau et al., 2006; Johannesdottir et al., 2012; Torti et al., 2011). In all studies, the case definition was newly diagnosed SCC and potential confounding factors were accounted for either by matching or by adjustment. Quality Assessment Six of the 9 studies included in the meta-analysis were considered to be of high quality (Asgari et al., 2010; Butler et al., 2005; Clouser et al., 2009; Elmets et al., 2010; Jeter et al., 2012; Torti et al., 2011); the remaining 3 studies of low to moderate quality (Grau et al., 2006; Johannesdottir et al., 2012; Nunes et al., 2011)(Supplementary Table 1).

5


Use of Aspirin Six studies provided estimates for the association between aspirin use and SCC: 3 cohort studies (Clouser et al., 2009; Jeter et al., 2012; Nunes et al., 2011) and 3 casecontrol studies (Asgari et al., 2010; Johannesdottir et al., 2012; Torti et al., 2011). Two cohort studies reported a significantly reduced risk of SCC with aspirin use (Johannesdottir et al., 2012; Nunes et al., 2011).

Adjusted estimates for ever-use of aspirin ranged from 0.70 to 0.98 in the cohort studies and from 0.75 to 1.38 in the case-control studies (Table 2). The overall summary estimate for ever-use of aspirin compared with non-use was OR 0.88 (95% confidence interval (CI) 0.75-1.03) with significant heterogeneity (p=0.017) (Figure 2) (Table 3). The pooled effect estimate did not differ according to study design, or method of exposure assessment though it tended to be lower in studies conducted among mostly male study populations or people at high risk of skin cancer than among the general population, or lower quality studies (Table 3). There was no evidence of publication bias (p values for Begg 1.000 and Egger 0.854) (Supplementary Figure 1 and Supplementary Table 2).

Use of Non-Aspirin NSAIDS Seven studies, 1 RCT (Elmets et al., 2010), 3 cohort (Clouser et al., 2009; Jeter et al., 2012; Nunes et al., 2011) and 3 case-control (Asgari et al., 2010; Johannesdottir et al., 6


2012; Torti et al., 2011) studies, presented estimates for the association between nonaspirin NSAIDS use and SCC. For exposure to non-aspirin NSAIDS, 2 studies used nonselective NSAIDS (Johannesdottir et al., 2012; Nunes et al., 2011), 1 used propionic acid NSAIDS (Torti et al., 2011), and celecoxib was used in the RCT (Elmets et al., 2010). The adjusted OR for SCC with non-aspirin NSAIDS use was 0.42 in the RCT, ranged from 0.79 to 0.92 in the cohort studies, and from 0.84 to 1.19 in the case-control studies (Table 2). The pooled effect estimate indicated a significant risk reduction (0.85, 95%CI 0.78-0.94) with no evidence of heterogeneity (p=0.628) (Figure 3) (Table 3). The pooled estimate did not differ by study design or exposure assessment method (Table 3). There was no evidence of publication bias (p values for Begg 0.548 and Egger 0.518) (Supplementary Figure 2 and Supplementary Table 2). Use of Aspirin or Non-Aspirin NSAIDS Of 9 studies that presented adjusted estimates of the association between use of either aspirin or non-aspirin NSAIDS and SCC, 3 (the RCT, 1 cohort and 1 case-control study) reported significant reductions in risk of SCC (Elmets et al., 2010; Johannesdottir et al., 2012; Nunes et al., 2011) (Table 2). One study assessed exposure to aspirin only and adjusted analyses for use of non-aspirin NSAIDS (Jeter et al., 2012). The summary estimate of all studies was 0.83 (95%CI 0.71-0.96) with a significant heterogeneity (pvalue 0.003) (Figure 4) (Table 3) but no evidence of publication bias (p values for Begg 7


0.118 and Egger 0.250) (Supplementary Figure 3 and Supplementary Table 2). Studies in those at higher risk of SCC or where exposure assessment was based on pharmacy databases, tended to show stronger inverse associations, though higher quality studies led toward the null association (Table 3).

Dose-response analyses All 9 studies evaluated a possible dose-response (including timing) relationship between either aspirin or non-aspirin NSAIDS and SCC. Only one study, examining aspirin use and risk of SCC reported a statistically significant (p=0.05) dose-response relationship (Jeter et al., 2012).

DISCUSSION We found that use of non-aspirin NSAIDS or any NSAIDS significantly reduced the risk of developing SCC by 15% and 18% respectively. A reduced risk of similar magnitude was also observed among aspirin users though with borderline statistical significance Due to a lack of uniformity of dose units used in different studies, there were insufficient data to meta-analyse the dose-dependent (frequency and duration) relationship between NSAIDS intake and SCC. Though not consistent with a previous meta-analysis of the association between NSAIDS use and cutaneous SCC showing a non-significant 14% 8


reduced risk (Zhang et al., 2014), our results accord with the findings of meta-analyses for other cancers (e.g. of the colon, breast, lung and prostate) (Harris et al., 2005) A topical NSAID (diclofenac) is already approved for use as a treatment for actinic keratoses in the US, Europe and Australia (McGillis and Fein, 2004), and these results raise the possibility that oral aspirin or non-aspirin NSAIDS could be used as chemopreventive agents for cutaneous SCC also. The hypothesized mechanism of action is mitigation of the actions of inflammatory cytokines such as COX-2 and its product prostaglandin E2 in promoting skin carcinogenesis. Overexpression of COX-2 has been observed in actinic keratoses and SCC lesions (Kuzbicki et al., 2011; Muller-Decker et al., 1999), which suggests COX-2 is involved in the multiple stages of skin tumorigenesis and NSAIDS may thus inhibit this inflammatory pathway (Guadagni et al., 2007; Subongkot et al., 2003). NSAIDS appeared to reduce SCC risk to a greater degree in people with a high prevalence of actinic keratoses or a history of keratinocyte cancers than in the general population. This suggests that COX-2 or prostaglandins may also help decrease the multiplicity of new primary skin cancers in susceptible people. Strengths and limitations This meta-analysis was based on a comprehensive literature review and included examination of heterogeneity by various study characteristics and well as an evaluation 9


of study quality. Compared with the previous meta-analysis (Zhang et al., 2014) which examined only the association between any NSAID use and SCC, our review included 2 further studies (Butler et al., 2005; Nunes et al., 2011) and our more comprehensive combined estimates included data from 4 additional studies for any NSAID use, 2 additional studies for aspirin use and 4 additional studies for non-aspirin NSAID use. Seven of the 9 studies included in our review adjusted for reported sun exposure either directly or indirectly (by adjusting for past history of skin cancer) which is important since sun exposure is the primary cause of cutaneous SCC (IARC Working Group on the Evaluation of Carcinogenic Risks to Humans, 2012). The definition of aspirin and nonaspirin NSAIDS use varied across studies however, with participants categorized as temporary or infrequent ‘users’ in some studies being regarded as ‘non-users’ in other studies. As a result, prevalence of use varied widely across studies, ranging from 23% to 77% for ever-use of aspirin or non-aspirin NSAIDS. In addition, referent groups were not consistent across studies, with some defining non-users as those who took neither aspirin nor non-aspirin NSAIDS, while others included aspirin-users in the referent group while estimating the effect of non-aspirin NSAIDS and vice versa without adjusting for the other exposure. Overall however, there was no major difference in risks of SCC among these user-subgroups both within studies or in aggregate. Heterogeneity in the prevalence of NSAIDS use across studies may have reflected the different proportions of 10


participants with health conditions associated with NSAIDS use (e.g. arthritis and coronary heart disease), which in turn could have affected study outcomes through associations with immune system function or sun exposure. We were unable to fully assess the relationship between different doses of aspirin or NSAIDS and SCC. In the studies of aspirin use for example, high and low doses were not analysed separately in 4 of the 6 relevant studies despite the different inherent mechanisms involved. Low-dose aspirin has an antiplatelet effect by reducing production of thromboxane A2 by COX-1 inhibition, whilst high-dose aspirin has an anti-inflammatory effect by reducing the production of prostaglandin E2 by COX-2 inactivation (Vane et al., 1998). Additionally, all non-aspirin NSAIDS were regarded as being the same in our analyses even though different non-aspirin NSAIDS have different kinetics and dynamics (Moore, 2002), and therefore may have quite different effects on risk. Finally, accuracy of exposure assessment in the included studies could be questioned: self-report is known to be inaccurate for non-repetitive NSAIDS use (West et al., 1995) while prescription databases lack information on use of common over-the-counter NSAIDS (Sørensen et al., 2001).

In conclusion, synthesis of existing published data supports a significant inverse association between oral NSAIDS use and incidence of cutaneous SCC, particularly among people with a high prevalence of actinic keratoses or a history of keratinocyte cancers. 11


Further research is required that not only accounts for participants’ sun exposure but that also includes accurate assessment of NSAIDS dosages and information about the reasons for NSAIDS use in order to evaluate the association of NSAIDS (aspirin in particular) and SCC in greater details.

MATERIALS AND METHODS This systematic review was conducted and reported according to the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta- Analyses) statement as a guideline (http://www.prisma-statement.org/). There was no previously published protocol.

Search strategy We searched the PubMed, Web of Science and EMBASE databases, using the search terms “nsaids", "aspirin", "cyclooxygenase", "nonmelanoma", “non-melanoma”, "squamous cell carcinoma", "skin", "cutaneous", "case-control", "cohort", "trial" and “incidence”. A single reviewer identified all potentially relevant studies published in any language to February 2014. Titles and abstracts of articles were reviewed to exclude laboratory-based or therapeutic studies, reviews and duplicate publications; full-texts of the remaining articles were obtained. The reference lists of reviews and retrieved articles 12


and “Related citations” in PubMed and “Times cited” in Web of Science for relevant articles were checked. All reasons for excluding studies were recorded. Study criteria were discussed with the other investigators and disagreements were resolved by consensus. Eligibility criteria Only case-control, cohort or intervention studies that examined oral intake of aspirin or non-aspirin NSAIDS in relation to risk of cutaneous SCC were considered. To be included, the studies had to report relative risk (RR) or odds ratio (OR) with 95% confidence interval (CI) or provide sufficient data to permit these calculations. If multiple publications evaluated the same or highly overlapped study populations, only the one with the most relevant and comprehensive data (e.g. most recently published or largest sample size) was included. Data extraction and quality assessment Extracted data included first author and year of publication, study location, design and duration, source of study population and sex distribution of participants, type of NSAID used (aspirin and/or non-aspirin NSAIDS or a specific NSAID) and exposure assessment (self-report or prescription-based pharmacy database), method of diagnosis (pathology, self-report or registry), total sample size, numbers of exposed cases and noncases, adjusted effect estimates with lower and upper 95% CIs, and adjusted factors in 13


the analyses. Studies were classified by sampling frame (general population, participants with history of skin cancer (defined as “high risk”), participants with higher prevalence (>10) of actinic keratoses (defined as “very high risk”) ), exposure assessment (prescription database, self-report), diagnosis method (histology, cancer registry, other (self-report or not mentioned)), results of quality assessment (low or high) and types of anti-inflammatory drugs (aspirin only, NSAIDS excluding aspirin, any NSAIDS).

We evaluated the quality of the studies by using a scoring system that was designed with reference to the following guidelines: Meta-analysis Of Observational Studies in Epidemiology (MOOSE) (Stroup et al., 2000), Quality Assessment Tool for Systematic reviews of Observational studies (QATSO) (Wong et al., 2008), and Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) (von Elm et al., 2007). Studies were assessed on three aspects, namely study design, method of assessment of exposure to aspirin or non-aspirin NSAIDS, and whether important confounding effects were taken into account during analysis. Two points were allocated for study design, with 2 points given for cohort studies or RCTs, 1 point for population-based casecontrol studies, and no points for clinic-based case-control studies. Studies that assessed exposure based on pill counts or personal recall were allocated one point, those based on pharmacy databases scored 0 points. Studies adjusted for age and sex 14


they received one point, if they were additionally adjusted for a measure of pigmentation or skin type they received 2 points, and if they also adjusted for a measure of sun exposure they received 3 points. A total score of 4–6 was considered high quality, 1–3, low-moderate quality.

Statistical Analysis To obtain the pooled estimate we used a random effects meta-analysis model to account for heterogeneity between the studies (DerSimonian and Laird, 1986). Statistical heterogeneity among studies was explored using a Cochrane Q test for heterogeneity (Cochran, 1954). We also calculated I2 value (ranging from 0% indicating no observed heterogeneity to 100%) to examine the degree of heterogeneity due to between-study variability (Higgins et al., 2003). To identify publication bias towards studies showing positive rather than null results, a funnel plot (Begg and Mazumdar, 1994) was constructed and the Egger regression asymmetry test (Egger et al., 1997) applied. Sensitivity analyses were conducted by removing individual studies one at a time from the pooled analysis and examining their influence on the pooled estimate. Analyses were performed by drug types (aspirin, non-aspirin NSAIDS and any NSAIDS). For studies where relative risks or odds ratios were reported for various doses but not for ever-use, the estimates for ever-use were calculated by combining the dose effect using RREst9 15


(Kinoshita et al., 2002). When effect estimates for non-selective and selective NSAIDS were reported separately, the estimates of non-selective NSAIDS were used. When studies separately reported results for aspirin and non-aspirin NSAIDS for the same study population we included the results with the larger sample size in analyses of use of aspirin or non-aspirin NSAIDS to avoid double-counting people who used both aspirin and NSAIDS. Subgroup Analysis Analyses were performed to assess the consistency of associations between aspirin and non-aspirin NSAIDS and SCC within certain pre-specified subgroups: study design (cohort vs. case-control), study population (general vs. high and very high risked), exposure assessment method (prescription database vs. self-report) diagnosis method (pathological vs. others) and quality of the individual studies (high vs. low). All statistical analyses were performed with the software STATA version 10.0 (Stata corporation, College Station, TX, USA). All p-values were two-tailed.

Conflict of interest The authors state no conflict of interest.

16


Acknowledgement CMO is supported by Program Grant 552429 from the National Health and Medical Research Council of Australia. NP is supported by an Early Career Fellowship from the National Health and Medical Research Council of Australia.

17


REFERENCES Aggarwal BB, Vijayalekshmi RV, Sung B (2009) Targeting inflammatory pathways for prevention and therapy of cancer: short-term friend, long-term foe. Clin Cancer Res 15:425-30. Amari W, Zeringue AL, McDonald JR, et al. (2011) Risk of non-melanoma skin cancer in a national cohort of veterans with rheumatoid arthritis. Rheumatology (Oxford) 50:1431-9. Armstrong BK, Kricker A (2001) The epidemiology of UV induced skin cancer. J Photochem Photobiol B 63:8-18. Asgari MM, Chren MM, Warton EM, et al. (2010) Association between nonsteroidal antiinflammatory drug use and cutaneous squamous cell carcinoma. Archives of Dermatology 146:388-95. Athar M, An KP, Morel KD, et al. (2001) Ultraviolet B(UVB)-induced cox-2 expression in murine skin: an immunohistochemical study. Biochem Biophys Res Commun 280:1042-7. Begg CB, Mazumdar M (1994) Operating characteristics of a rank correlation test for publication bias. Biometrics 50:1088-101. Buckman SY, Gresham A, Hale P, et al. (1998) COX-2 expression is induced by UVB exposure in human skin: implications for the development of skin cancer. Carcinogenesis 19:723-9. Butler GJ, Neale R, Green AC, et al. (2005) Nonsteroidal anti-inflammatory drugs and the risk of actinic keratoses and squamous cell cancers of the skin. J Am Acad Dermatol 53:966-72. Clouser MC, Roe DJ, Foote JA, et al. (2009) Effect of non-steroidal anti-inflammatory drugs on non-melanoma skin cancer incidence in the SKICAP-AK trial. Pharmacoepidemiology and Drug Safety 18:276-83. Cochran WG (1954) The combination of estimates from different experiments. Biometrics 10:101-29. Cook NR, Lee IM, Gaziano JM, et al. (2005) Low-dose aspirin in the primary prevention of 18


cancer: the Women's Health Study: a randomized controlled trial. JAMA 294:47-55. DerSimonian R, Laird N (1986) Meta-analysis in clinical trials. Control Clin Trials 7:17788. Egger M, Davey Smith G, Schneider M, et al. (1997) Bias in meta-analysis detected by a simple, graphical test. BMJ 315:629-34. Elmets CA, Viner JL, Pentland AP, et al. (2010) Chemoprevention of nonmelanoma skin cancer with celecoxib: a randomized, double-blind, placebo-controlled trial. J Natl Cancer Inst 102:1835-44. Fecker LF, Stockfleth E, Nindl I, et al. (2007) The role of apoptosis in therapy and prophylaxis of epithelial tumours by nonsteroidal anti-inflammatory drugs (NSAIDs). Br J Dermatol 156 Suppl 3:25-33. Grau MV, Baron JA, Langholz B, et al. (2006) Effect of NSAIDs on the recurrence of nonmelanoma skin cancer. Int J Cancer 119:682-6. Guadagni F, Ferroni P, Palmirotta R, et al. (2007) Non-steroidal anti-inflammatory drugs in cancer prevention and therapy. Anticancer Res 27:3147-62. Harris RE, Beebe-Donk J, Doss H, et al. (2005) Aspirin, ibuprofen, and other nonsteroidal anti-inflammatory drugs in cancer prevention: a critical review of non-selective COX-2 blockade (review). Oncol Rep 13:559-83. Higgins JP, Thompson SG, Deeks JJ, et al. (2003) Measuring inconsistency in metaanalyses. BMJ 327:557-60. IARC Working Group on the Evaluation of Carcinogenic Risks to Humans (2012) a review of human carcinogens. Part D: Radiation. Lyon. Jeter JM, Han J, Martinez ME, et al. (2012) Non-steroidal anti-inflammatory drugs, acetaminophen, and risk of skin cancer in the Nurses' Health Study. Cancer Causes Control 23:1451-61. Johannesdottir SA, Chang ET, Mehnert F, et al. (2012) Nonsteroidal anti-inflammatory drugs and the risk of skin cancer: A population-based case-control study. Cancer. 19


Kinoshita M, Shiraishi H, Muranushi C, et al. (2002) Determination of molecular mass of acidic polysaccharides by capillary electrophoresis. Biomed Chromatogr 16:141-5. Kuzbicki L, Lange D, Stanek-Widera A, et al. (2011) Different expression of cyclooxygenase-2 (COX-2) in selected nonmelanocytic human cutaneous lesions. Folia Histochem Cytobiol 49:381-8. Lucas RM, McMichael AJ, Armstrong BK, et al. (2008) Estimating the global disease burden due to ultraviolet radiation exposure. Int J Epidemiol 37:654-67. Martin GM, Stockfleth E (2012) Diclofenac sodium 3% gel for the management of actinic keratosis: 10+ years of cumulative evidence of efficacy and safety. J Drugs Dermatol 11:600-8. Maverakis E, Miyamura Y, Bowen MP, et al. (2010) Light, including ultraviolet. J Autoimmun 34:J247-57. McGillis ST, Fein H (2004) Topical treatment strategies for non-melanoma skin cancer and precursor lesions. Semin Cutan Med Surg 23:174-83. Moore DE (2002) Drug-induced cutaneous photosensitivity: incidence, mechanism, prevention and management. Drug Saf 25:345-72. Muller-Decker K, Reinerth G, Krieg P, et al. (1999) Prostaglandin-H-synthase isozyme expression in normal and neoplastic human skin. Int J Cancer 82:648-56. Nunes AP, Lapane KL, Weinstock MA (2011) Association between non-steroidal antiinflammatory drugs and keratinocyte carcinomas of the skin among participants in the Veterans Affairs Topical Tretinoin Chemoprevention Trial. Pharmacoepidemiol Drug Saf 20:922-9. Rivers JK, McLean DI (1997) An open study to assess the efficacy and safety of topical 3% diclofenac in a 2.5% hyaluronic acid gel for the treatment of actinic keratoses. Arch Dermatol 133:1239-42. Rundhaug JE, Simper MS, Surh I, et al. (2011) The role of the EP receptors for prostaglandin E2 in skin and skin cancer. Cancer Metastasis Rev 30:465-80. 20


Sørensen HT, Johnsen SP, Nørgård B (2001) Methodological issues in using prescription and other databases in pharmacoepidemiology. Norsk epidemiologi 11. Stroup DF, Berlin JA, Morton SC, et al. (2000) Meta-analysis of observational studies in epidemiology: a proposal for reporting. Meta-analysis Of Observational Studies in Epidemiology (MOOSE) group. Jama 283:2008-12. Subongkot S, Frame D, Leslie W, et al. (2003) Selective cyclooxygenase-2 inhibition: a target in cancer prevention and treatment. Pharmacotherapy 23:9-28. Torti DC, Christensen BC, Storm CA, et al. (2011) Analgesic and nonsteroidal antiinflammatory use in relation to nonmelanoma skin cancer: a population-based casecontrol study. J Am Acad Dermatol 65:304-12. Vane JR, Bakhle YS, Botting RM (1998) Cyclooxygenases 1 and 2. Annu Rev Pharmacol Toxicol 38:97-120. von Elm E, Altman DG, Egger M, et al. (2007) The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: guidelines for reporting observational studies. Lancet 370:1453-7. West SL, Savitz DA, Koch G, et al. (1995) Recall accuracy for prescription medications: self-report compared with database information. Am J Epidemiol 142:1103-12. Wong WC, Cheung CS, Hart GJ (2008) Development of a quality assessment tool for systematic reviews of observational studies (QATSO) of HIV prevalence in men having sex with men and associated risk behaviours. Emerging themes in epidemiology 5:23. Zhang B, Liang X, Ye L, et al. (2014) No chemopreventive effect of nonsteroidal antiinflammatory drugs on nonmelanoma skin cancer: evidence from meta-analysis. PLoS ONE 9:e96887.

21


List of Figures, Tables and Appendices 

Figure 1. Flow diagram of study selection for meta-analysis of the effect of aspirin or non-aspirin NSAIDS on squamous cell carcinoma (SCC).



Figure 2 Forest plot of the association between aspirin use and SCC (denoted by first author and publication year)



Figure 3 Forest plot of the association between non-aspirin NSAIDS use and SCC (denoted by first author and publication year)



Figure 4 Forest plot of the association between any aspirin or non-aspirin NSAIDS use and SCC (denoted by first author and publication year)



Table 1. Characteristics of the 9 studies included in the meta-analysis of aspirin or non-aspirin NSAID use and risk of cutaneous squamous cell carcinoma.



Table 2. Summary of published results on aspirin or non-aspirin NSAID use and risk of cutaneous squamous cell carcinoma.



Table 3. Meta-analysis results using a random effects model: association of aspirin or non-aspirin NSAIDS use and cutaneous SCC.



Supplementary Figure 1. Funnel plot the relative risk of developing SCC by aspirin.



Supplementary Figure 2. Funnel plot the relative risk of developing SCC by nonaspirin NSAIDS.



Supplementary Figure 3. Funnel plot the relative risk of developing SCC by any 22


aspirin or non-aspirin NSAIDS. 

Supplementary Table 1 Quality assessment of studies in the meta-analysis of aspirin or non-aspirin NSAIDS use and risk of squamous cell carcinoma.



Supplementary Table 2. Results of tests for publication bias using the Begg rank correlation method and the Egger weighted regression method for the analyses of aspirin or non-aspirin NSAIDS use and risk of squamous cell carcinoma.

23


Identification

Records identified through database searching

Additional records identified

Pub Med: 84

through other sources

Web of Sci: 72

(n = 0)

EMBASE: 130

Records after duplicates removed

Eligibility

Screening

(n = 237)

Records screened

Records excluded

(n = 237)

(n = 226)

Full-text articles assessed

Full-text articles excluded,

for eligibility

with reasons

(n = 11)

(n = 2)

Studies included in qualitative synthesis

Included

(n = 9)

Studies included in quantitative synthesis (meta-analysis) (n = 9)

Figure 1. Flow diagram of study selection for meta-analysis of the effect of aspirin or non-aspirin NSAIDS on squamous cell carcinoma (SCC) (Search date: 28th February 2014).

24


Effect

%

authoor&year

Estimate (95% CI) Weight

Jeter (2012)

0.98 (0.86, 1.13)

24.64

Nunes (2011)

0.70 (0.55, 0.88)

18.00

Clouser (2009)

0.71 (0.41, 1.22)

6.54

Johannesdottir (2012)

0.86 (0.76, 0.98)

25.28

Torti (2010)

0.75 (0.55, 1.02)

13.89

Asgari (2010)

1.38 (0.96, 1.97)

11.65

Overall (I-squared = 63.7%, p = 0.017)

0.88 (0.75, 1.02)

100.00

NOTE: Weights are from random effects analysis .413

1

2.42

Figure 2 Forest plot of the association between aspirin use and SCC (denoted by first author and publication year)

Effect

%

authoor&year


Elmets (2010)

0.42 (0.19, 0.93)

1.39

Jeter (2012)

0.92 (0.75, 1.12)

21.76

Nunes (2011)

0.82 (0.64, 1.04)

14.84

Clouser (2009)

0.80 (0.45, 1.42)

2.61


0.85 (0.75, 0.97)

52.88

Torti (2010)

1.19 (0.51, 2.79)

1.21

Asgari (2010)

0.84 (0.56, 1.26)

5.32


0.85 (0.78, 0.94)

100.00


1

5.26

Figure 3 Forest plot of the association between non-aspirin NSAIDS use and SCC (denoted by first author and publication year)

25


Effect

%

authoor&year


Elmets (2010)

0.42 (0.19, 0.92)

2.75

Jeter (2012)

0.98 (0.86, 1.13)

17.61

Nunes (2011)

0.64 (0.52, 0.77)

14.95

Clouser (2009)

0.70 (0.46, 1.06)

7.36


0.85 (0.76, 0.94)

18.85

Torti (2010)

0.78 (0.59, 1.03)

11.53

Asgari (2010)

1.32 (0.92, 1.89)

8.84

Bulter (2005)

0.65 (0.37, 1.14)

4.80

Grau (2006)

0.85 (0.67, 1.07)

13.31


0.82 (0.71, 0.94)

100.00


1

5.26

Figure 4 Forest plot of the association between any aspirin or non-aspirin NSAIDS use and SCC (denoted by first author and publication year)

26


Table 1. Characteristics of the 9 studies included in the meta-analysis of aspirin or non-aspirin NSAID use and risk of cutaneous squamous cell carcinoma First author (Published year)

Study location

Study design

Study duration

Cohort description / source of cases/controls

Sex (% male)

Diagnosis method

Exposure assessment

Exposure definition

Elmets (2010)

US

RCT

2001-2006

10-40 actinic keratoses and a previous histological diagnosis of at least 1 actinic keratosis and/or keratinocyte carcinoma (very high risk for SCC)

82%

Biopsy

Pill counts at each visit

more than 80% of pills (calocoxib 200mg b.i.d.) during 9 months period

Jeter (2012)

US

Cohort

1980-2008

Nurses' Health Study Female registered nurses aged 30-55

0%

Pathology record

Questionnaire

Past or current use in most weeks

Nunes (2011)

US

Cohort

1998-2004

Veterans Affairs Topical Tretinoin Chemoprevention Trial (at least 2 keratinocyte carcinomas on the face or ears within the 5 years prior to enrolment) (high risk for SCC)

98%

Biopsy

Pharmacy database

At least 1 prescription

Clouser (2009)

US

Cohort

1984-1988

SKICAP-AK Trial (a history of at least 10 actinic keratoses) (very high risk for SCC)

68%

Biopsy

Questionnaire

More than once per week during the follow-up period


Denmark

Population-based case-control

1991-2008

Danish Cancer Registry Danish Civil registration System

45%

Database based on morpholog y and histology

Prescription database

Total of >2 prescription during the entire study period

27


Torti (2011)

US


1997-2000

Case: dermatologist and pathology laboratories Control: residents list in New Hampshire

56%

Dermatolo gist and pathology laboratori es

Interview

At least 4 times a week for at least 1 month

Asgari (2010)

US


1994-2005

Kaiser Permanente Northern California members ages 43 to 85

62%

Pathology

Questionnaire (& Pharmacy database)

At least once a week for at least 1 year

Bulter (2005)

Australia


1992-1996

Nambour Skin Cancer Prevention Trial Randomly selected residents aged 20-69

44%

95% Histology

Interview

2 or more times per week for at least 1 year

Grau (2006)

US

Clinic-based casecontrol

1983-1989

Skin Cancer Prevention Study (at least 1 histologically confirmed BCC or SCC after Jan 1 1980) (high risk for SCC)

40%

Biopsy

Questionnaire

At least 1 positive answer in 3 questionnaires (current use)

28


Table 2. Summary of published results on aspirin or non-aspirin NSAID use and risk of cutaneous squamous cell carcinoma. First author (Published year) RCT Elmets (2010)

Cases

240

Type of drugs

NSAIDS**

RR (95%CI)

Adjusted factors

0.42 (0.19-0.93)

Age, sex, Fitzpatrick skin type, actinic keratosis history at screening, skin cancer history, and log-transformed patient time on study

ASA: ASA: 92125 ASA 1337 NSAIDS: NSAIDS NSAIDS: 76181 1123

0.98 (0.86-1.13) 0.92 (0.75-1.12)

Age, questionnaire cycle, reaction of skin to sun exposure, ability to tan, number of severe sunburns, number of moles on left arm, family history of melanoma, UV-B availability at state of residence, menopausal status and use of postmenopausal hormones, height, BMI, physical activity, intake of vitamin C from foods, intake of vitamin D from foods and supplements ASA is also adjusted for smoking status NSAIDs are also adjusted for use of aspirin and acetaminophen

Nunes (2011)

309

1051

ASA NSAIDS*** N+A

0.70 (0.55-0.88) Age, number prior BCC, number prior SCC, Charlson 0.82 (0.64-1.04) Comorbidity Index 0.64 (0.52-0.77)

Clouser (2009)

146

1402

ASA NSAIDS N+A

0.711 (0.413-1.22) 0.797 (0.466-1.42) 0.697 (0.460-1.06)

Cohort studies Jeter (2012)

NA*

Cohort size/Controls

Treatment, age and gender

29


Case-control studies Johannesdottir 1921 (2012)

19163

ASA NSAIDS*** N+A

Torti (2011)

535

462

ASA 0.75 (0.55-1.02) NSAIDS**** 1.19 (0.51-2.79) N+A 0.78 (0.59-1.03)

Age, sex, number of cigarettes smoked per day, skin type, lifelong number of painful sunburns and lifelong cumulative number of hours of sun exposure

Asgari (2010)

415

415

ASA NSAIDS N+A

1.38 (0.96-1.97) 0.84 (0.56-1.26) 1.32 (0.92-1.89)

Eye colour, natural hair colour, skin type, education. history of sunburns, history of high-risk exposures such as UV light, burn scar, non-healing ulcers, radiation treatment, arsenic exposure, exposures to industrial chemicals, history of smoking, history of freckling, outdoor sun exposure, occupational sun exposure, tanning bed use, and family history of skin cancer

Bulter (2005)

86

187

N+A

0.65 (0.37-1.14)

Skin colour, tanning ability, occupation, clinical solar elastosis, hip or knee pain, and sunscreen intervention arm of trial

Grau (2006)

132

1019

N+A

0.85 (0.67-1.07)

Age, sex, centre, risk set time, number of skin cancers prior to study entry, skin type and total number of questionnaires completed

0.86 (0.76-0.98) Charlson Comorbidity Index score, use of systemic 0.85 (0.75-0.97) glucocorticoids, cytostatic or immunosuppressive 0.85 (0.76-0.94) medication, and drugs with pigmenting adverse effects

ASA=acetylsalicylic acid (aspirin); COX=cyclooxygenase; NSAIDs=non-steroidal anti-inflammatory drugs; N+A=any use of aspirin and/or non-aspirin NSAIDs combined * mean number of tumours per patient is presented instead of number of incidence, **celecxib ***non-selective NSAIDS ****propionic acid NSAIDS

30


Table 3. Meta-analysis results using a random effects model: association of aspirin or non-aspirin NSAIDS use and cutaneous SCC Studie s

Aspirin RR (95% CI)

I2 (%)

P het

Studie s

Non-aspirin NSAIDS RR (95% CI)

I2

P het

Studie s

(% )

Any aspirin/non-aspirin NSAIDSuse RR (95% CI) I2 P het (%)

All studies

6

0.88 (0.75-1.03)

64%

0.017

7

0.85 (0.78-0.94)

0%

0.628

9

0.82 (0.71-0.94)

65%

0.003

Study type Cohort Case-control

3 3

0.82 (0.62-1.07) 0.94 (0.71-1.24)

70% 72%

0.037 0.028

3 3

0.87 (0.75-1.01) 0.86 (0.76-0.97)

0% 0%

0.736 0.742

3 5

0.77 (0.56-1.07) 0.87 (0.75-1.01)

85% 42 %

0.001 0.140

Population studied General High a) & Very high b)

4 2

0.94 (0.79-1.11) 0.70 (0.57-0.87)

65% 0%

0.037 0.959

4 3

0.87 (0.79-0.97) 0.75 (0.56-1.00)

0% 20 %

0.808 0.286

5 4

0.91 (0.78 -1.05) 0.70 (0.57-0.86)

57% 41%

0.055 0.168

Exposure assessment Pharmacy database Self-report

2 4

0.80 (0.65-0.97) 0.95 (0.74-1.20)

56% 61%

0.131 0.054

2 4

0.84 (0.75-0.95) 0.90 (0.76-1.07)

0% 0%

0.798 0.864

2 6

0.75 (0.57-0.99) 0.89 (0.76-1.05)

84% 48%

0.013 0.086

Quality assessment High Low

4 2

0.95 (0.74-1.20) 0.80 (0.65-0.97)

0.054 0.131

5 2

0.87 (0.73-1.04) 0.84 (0.75-0.95)

4% 0%

0.384 0.798

6 3

0.84 (0.67-1.06) 0.78 (0.65-0.93)

62% 69%

0.023 0.040

61% 56%

a)All participants a history of skin cancer; b)All participants a high prevalence (>10) of actinic keratoses); na-not applicable

31






Nonsteroidal antiinflammatory drugs--differences and similarities.

Gastrointestinal damage associated with nonsteroidal antiinflammatory drugs.

Differential impairment of aspirin-dependent platelet cyclooxygenase acetylation by nonsteroidal antiinflammatory drugs.

Collagenous colitis in setting of nonsteroidal antiinflammatory drugs and antibiotics.

The use of nonsteroidal antiinflammatory drugs to prevent adherence of Staphylococcus epidermidis to medical polymers.

Informed consent and the prescription of nonsteroidal antiinflammatory drugs.

Systematic review: interactions between aspirin, and other nonsteroidal anti-inflammatory drugs, and polymorphisms in relation to colorectal cancer.

Gastrointestinal damage associated with the use of nonsteroidal antiinflammatory drugs.

RF-Oral Nonsteroidal Anti-inflammatory Drug Use May Reduce the Risk of Cutaneous Squamous Cell Carcinoma.

Risk Factors for Cutaneous Squamous Cell Carcinoma Recurrence, Metastasis, and Disease-Specific Death: A Systematic Review and Meta-analysis.

Preventing upper gastrointestinal bleeding in patients receiving nonsteroidal antiinflammatory drugs.

Oral capecitabine to prevent recurrent cutaneous squamous cell carcinoma in a lung transplant recipient.

CAN ASPIRIN PREVENT COLON CANCER?

Cutaneous spindle cell squamous cell carcinoma.

Diagnostic markers for hyperemesis gravidarum: a systematic review and metaanalysis.

Cutaneous squamous cell carcinoma with osteoclast-like giant cells: a very rare variant of cutaneous squamous cell carcinoma.

Methotrexate and nonsteroidal antiinflammatory drug interactions.

Classification of cutaneous squamous cell carcinoma.

Antiinflammatory drugs and gastrointestinal bleeding: a comparison of aspirin and ibuprofen.

Biology of cutaneous squamous cell carcinoma.

Ketorolac: a parenteral nonsteroidal antiinflammatory drug.

Dietary patterns and oesophageal squamous cell carcinoma: a systematic review and meta-analysis.

Multiple cutaneous squamous cell carcinoma in cutaneous sarcoidosis.

Bioavailability of powdered inclusion compounds of nonsteroidal antiinflammatory drugs with beta-cyclodextrin in rabbits and dogs.