Risk Factors for Actinic Keratoses: A Systematic Review and Meta-Analysis.

Objective: To integrate evidence and assess the risk factors associated with actinic keratosis (AK).
Methods: Unrestricted searches were conducted on five electronic databases, with an end-date parameter of September 2021. We summarized the study characteristics and pooled the results from individual studies by using a random-effects model. The risk of bias was estimated using the Cochrane Risk of Bias Tool, and the quality of evidence was estimated according to the Newcastle-Ottawa Scale.
Results: Sixteen studies were included in final analysis, and we assessed the AK risk among a variety of risk factors. Overall, the male sex (odds ratio (OR): 2.51; 95% confidence interval (CI): 1.94-3.25; P < 0.01), age >45 years (OR = 7.65, 95% CI: 2.95-19.86; P < 0.01), light Fitzpatrick skin phototype (OR = 2.32, 95% CI: 1.74-3.10; P < 0.01), light hair color (OR = 2.17, 95% CI: 1.40-3.36; P < 0.01), light eye color (OR = 1.67, 95% CI: 1.03-2.70; P = 0.04), freckles on face/arms (OR = 1.88, 95% CI: 1.37-2.58; P < 0.01), suffered positive history of other types of non-melanoma skin cancer (OR = 4.46, 95% CI: 2.71-7.33; P < 0.01), sunburns in childhood (OR = 2.33, 95% CI: 1.47-3.70; P < 0.01) and adulthood (OR = 1.50, 95% CI: 1.12-2.00; P < 0.01), severe sunburn (OR = 1.94, 95% CI: 1.62-2.31; P < 0.01), and chronic occupational and/or recreational sun exposure (OR = 3.22, 95% CI: 2.16-4.81; P < 0.01) increased the risk of AK. Moreover, sunscreen use (OR = 0.51, 95% CI: 0.34-0.77; P < 0.01) and history of atopy reduced the risk of AK. Sensitivity analysis yielded consistent results. The included studies showed a high risk of bias.
Conclusion: We confirm several well-known AK risk factors and their quantitative data, and summarized the uncommon risk factors and protective factors. Our results may inform on the design and implementation of AK screening and educational programs. Copyright:

Introduction

Actinic keratosis (AK) is a common cutaneous neoplasia resulting from abnormal proliferation of atypical epidermal keratinocytes.[1] Lesions appear as small slow-growing papules or patches that are dry, erythematous, usually scaly, and sometimes pigmented, preferentially affecting mainly in chronically sun-exposed skin areas.[23] The surrounding skin usually shows evidence of chronic photodamage, such as telangiectasia, cutaneous dyschromia, solar elastosis, or skin atrophy.[4] The natural history of AK may evolve following one of three paths: spontaneous regression, stable existence, or malignant progression. Natural remission benefits from being away from risk factors. However, regressed lesions easily reappear over time.[5]

AK is one of the most frequently seen signs in the outpatient setting and was diagnosed at >10% of dermatology visits. In the UK, 15%–23% of individuals have actinic keratosis lesions, and this value reaches 37.5% among whites 50 years of age or older.[567] Moreover, AK is considered an intraepithelial keratinocyte neoplasm and a precursor of invasive squamous cell carcinoma (SCC).[8] The progression rate of a single AK to SCC was variably estimated at 0%–0.075% per lesion per year. A higher progression rate of 0.53% per lesion per year was estimated if the patient already had SCC elsewhere.[8] Although there have been several clear causes, for example, male gender, advanced age, and a high level of sun exposure, the reported prevalence remains at a high level. For other common risk factors, such as educational level, freckles/naevi on face/arms, and dwelling in a tropical country, the results were inconsistent.

Additionally, prevalence is highly variable across different countries, depending on study setting, UV radiation level, and/or patient characteristics.[910] Especially, there is great variability of major risk factors in the different populations, but a complete overview is missing. Moreover, there is no quantitative data on the harmfulness of male gender, advanced age, and a high level of sun exposure as clear causes. With the increasing incidence of AK, an improved understanding of this issue may have important public health significance given the possibility that prevention of sunburn and some underlying diseases might reduce the incidence of AK.

Methods

This review was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA)[11] and the Cochrane Handbook For Systematic Reviews.[12]

Search strategy

PubMed, Embase, Web of Knowledge, Medline, and Cochrane Central Register databases were systematically searched in September 2021 with no restriction. Relevant studies that assessed the risk of AK under any risk factors were identified. We used and combined the following search terms: (actinic keratosis OR solar keratosis OR senile keratosis) AND (risk factors OR prevalence OR predisposing factors). Reference lists of the retrieved articles and recent reviews were also reviewed to reduce omissions.

Eligibility criteria

An initial screening of titles or abstracts was performed at the initial stage, and a second screening was based on a full-text review. Eligibility was restricted to studies with human participants only, observational research design, at least 10 patients, and high-quality studies selected after quality assessment. Studies had to report odds ratios (ORs) of AK under different specific risk factors with corresponding 95% confidence intervals (CIs) (or data to calculate them). Any patients with a clinical and/or histopathological diagnosis of AK were included, and any population, age, body location, or comparative treatment was allowed. Duration of follow-up was not a criterion. All reported risk factors have been concerned. Moreover, all languages and abstract publications from scientific meetings were included to reduce the chances of publication bias. We excluded review papers, non-peer-reviewed local or government reports, conference abstracts, and presentations in the present review. If there were many studies based on the same research database, only the one that reported the most detailed data was included. Additionally, we assessed potential studies to ensure that there was no duplication of case series.

Data extraction and quality assessment

Two authors (YSL, XSZ) independently initial screening of titles or abstracts and full-text reviews, extracted data, and assessed study quality. Disagreements were resolved through discussions between investigators or with a third independent author (JHW) until a consensus was reached. The following study characteristics were recorded: first author's name, publication year, country/areas, design type, population, center, AK location, participant's gender, participant's age, types of risk factors, and quality score. We used the Newcastle–Ottawa Scale (NOS) to assess the quality of the studies included in the present review.[13] When the study wins seven or more stars, it is considered of higher methodologic quality.

Statistical analysis

Statistical analyses were conducted using MS Excel, Stata 13.0 (College Station, TX: StataCorp, LLP), and Review Manager version 5.0 (The Nordic Cochrane Centre, The Cochrane Collaboration). We used a random-effects model as clinical and methodological heterogeneity between the studies was likely. Heterogeneity tests were performed using the Cochran Q test (P < 0.10 indicates statistically significant heterogeneity)[14] and I2 statistic. Statistical tests were declared significant for a two-sided P value not exceeding 0.05, except where otherwise specified.

Results

Study selection and characteristics

The initial search identified 1035 potential eligible publications from databases. Finally, 16 eligible studies[6101516171819202122232425262728] were identified [Figure 1]. The characteristics of included studies are summarized in Table 1. Among all included studies, 12 studies (75%) were conducted in Europe, 3 studies were conducted in Asia, and 1 study in Oceania. All included studies were designed as an observational study: 7 used case-control study design, 6 used cross-sectional study design, and 3 used cohort study design. Most of the studies (62.5%) were multicenter and performed based on hospital-based participants. Eleven studies (68.75%) were considered of higher methodologic quality. Most AK participants were older and tended to favor males. The mean age of participants in these studies ranged from 57.6 to 79.4 years, with the fraction of male participants ranging from 47% to 100%. Six studies reported the AK locations, basically concentrated in the face/neck, scalp, and extremities.

Figure 1

Flow chart of study identification and selection

Table 1

Characteristics of included studies

Study	Country/Areas	Design type	Population	Center	AK Location	Case			Control			NOS
						AK, n	Age (years)	Male, %	Control, n	Age (years)	Male, %
Flohil SC et al. 2013[6]	The Netherlands	cohort study	population-based	Single-	face, scalp	773	NR	454 (58.7)	1288	NR	473 (36.7)	7
Schaefer I et al. 2014[10]	Germany	cohort study	population-based	Multi-	NR	867	57.6±8.3	684 (78.9)	41317	43.4±10.9	22095 (53.5)	7
Sechi A et al. 2021[15]	Italy	case-control	hospital-based	Single-	NR	90	79.4±7.2	66 (73.3)	90	73.3±8.6	35 (38.9)	7
Warszawik-Hendzel O et al. 2020[16]	Poland	case-control	hospital-based	Single-	NR	200	71 (46-93)	94 (47.0)	200	69 (45-90)	98 (49.0)	7
Vimercati L et al. 2020[17]	Italy	case-control	hospital-based	Multi-	face, scalp, hands, other body areas	217	NR	217 (100.0)	704	NR	704 (100.0)	8
Cerman AA et al. 2018[18]	Turkey	cross-sectional study	hospital-based	Single-	NR	31	70.90±8.97	6 (19.4)	29	71.90±7.18	8 (27.6)	7
Schafer I et al. 2017[19]	Germany	cross-sectional study	population-based	Multi-	NR	NR	NR	NR	NR	NR	NR	7
Fargnoli MC et al. 2017[20]	Italy	cross-sectional study	hospital-based	Multi-	face/neck, scalp, limbs, ear, trunk, limbs, lips.	2181	NR	1374 (63.0)	2423	NR	1067 (44.0)	7
Trakatelli M et al. 2016[21]	European	case-control	hospital-based	Multi-	NR	676	NR	NR	NR	NR	NR	7
Traianou A et al. 2012[22]	European	case-control	hospital-based	Multi-	NR	343	72.6 (37-95)	217 (63.3)	686	61.97 (31-91)	432 (63.0)	8
Hensen P et al. 2009[23]	Germany	case-control	hospital-based	Multi-	face/head, trunk/neck, limb.	331	71.3±8.9	223 (67.8)	383	58.8±11.4	145 (38.0)	7
Campanelli A et al. 2002[24]	Italy	case-control	hospital-based	Single-	face, limbs, chest.	11	NR	NR	50	NR	NR	6
Memon AA et al. 2000[25]	England	cross-sectional study	hospital-based	Multi-	NR	108	NR	82 (75.9)	860	NR	449 (52.2)	5
Araki K et al. 1999[26]	Japan	cross-sectional study	population-based	Multi-	face, neck, forearms, hands.	31	NR	17 (54.8)	4194	NR	1444 (34.4)	5
Frost CA et al. 1998[27]	Australia	cross-sectional study	population-based	Single-	NR	83	NR	NR	110	NR	NR	5
Suzuki T et al. 1997[28]	Japan	cohort study	population-based	Multi-	NR	NR	NR	NR	NR	NR	NR	6

Note: AK, actinic keratoses; NR, not reported

Participant sociodemographic, behavioral characteristics and risk of AK

Risk estimates between participants' sociodemographic, behavioral characteristics, and AK are summarized in Table 2. A total of 10 studies reported the risk of AK among different genders, and the male sex was significantly associated with the risk of AK (OR = 2.51, 95% CI: 1.94–3.25; P < 0.01) [Figure 2a]. However, substantial heterogeneity was observed (I2 = 91%; P < 0.01). A significantly increased risk of AK was found among the participants aged >80 years (OR = 4.33, 95% CI: 1.45–12.92), aged >70 years (OR = 6.26, 95% CI: 4.86–8.07), aged >60 years (OR = 5.44, 95% CI: 2.92–10.14), and aged >45 years (OR = 7.65, 95% CI: 2.95–19.86) compared with the reference group. No obvious association was found in education level and the risk of AK (OR = 0.68, 95% CI: 0.37–1.27, P = 0.23). That is, the educational level could not be considered to have a relationship with the risk of AK. Risk of AK also exists in patients with abnormal BMI (OR = 1.03, 95% CI: 0.79–1.34), alcohol consumption (OR = 1.30, 95% CI: 0.90–1.89), and smoking status (OR = 1.08, 95% CI: 0.86–1.36).

Table 2

The risk of AK under different specific risk factors

Risk factors	Number of studies	OR (95%CI), Test for overall effect	Risk factors	Number of studies	OR (95%CI), Test for overall effect
Participant characteristics			Sun exposure
Male	10	2.51 (1.94-3.25) χ2=104.78, P<0.01, I2=91%	Sunburns in childhood	6	2.33 (1.47-3.70) χ2=77.15, P<0.01, I2=94%
Educational levela	5	0.68 (0.37-1.27) χ2=109.50, P=0.23, I2=96%	Sunburns in adulthood	3	1.50 (1.12-2.00) χ2=1.15, P<0.01, I2=0%
Age >80 years old	3	4.33 (1.45-12.92) χ2=27.70, P<0.01, I2=93%	Severe sunburn	3	1.94 (1.62-2.31) χ2=1.72, P<0.01, I2=0%
Age >70 years old	4	6.26 (4.86-8.07) χ2=11.45, P<0.01, I2=74%	Chronic occupational and/or recreational sun exposure	6	3.22 (2.16-4.81) χ2=79.59, P<0.01, I2=94%
Age >60 years old	5	5.44 (2.92-10.14) χ2=41.44, P<0.01, I2=90%	Dwelling in tropical country	3	1.06 (0.69-1.62) χ2=9.60, P=0.79, I2=79%
Age >45 years old	6	7.65 (2.95-19.86) χ2=178.84, P<0.01, I2=97%	Sunscreen use	5	0.51 (0.34-0.77) χ2=41.27, P<0.01, I2=90%
Abnormal BMIb	2	1.03 (0.79-1.34) χ2=17.34, P=0.84, I2=94%	Drug use
Alcohol consumption	2	1.30 (0.90-1.89) χ2=3.10, P=0.17, I2=68%	Antiplatelet drugs	1	3.35 (1.75-6.39)
Smoking status	3	1.08 (0.86-1.36) χ2=7.19, P=0.51, I2=72%	Calcium channel blockers	2	1.35 (0.46-3.96) χ2=3.51, P=0.58, I2=72%
Pigmentation status			Statins	1	0.95 (0.51-1.77)
Skin phototype I+II (vs. III + IV)	8	2.32 (1.74-3.10) χ2=58.94, P<0.01, I2=88%	Angiotensin-converting enzyme inhibitors	1	2.28 (1.2-4.3)
Hair colorc	5	2.17 (1.40-3.36) χ2=36.69, P<0.01, I2=89%	Angiotensin receptor AT1 blockers	2	4.99 (1.18-21.17) χ2=1.99, P=0.03, I2=50%
Eye colord	4	1.67 (1.03-2.70) χ2=56.39, P=0.04, I2=95%	Antibiotics (Doxycycline)	1	0.98 (0.92-1.04)
Other skin disease			Antibiotics (Ciprofloxacin)	1	0.95 (0.36-2.54)
Freckles on face/arms	5	1.88 (1.37-2.58) χ2=58.13, P<0.01, I2=93%	Thiazides	2	2.14 (1.14-4.03) χ2=0.36, P=0.02, I2=0%
Positive history of other types of NMSC	5	4.46 (2.71-7.33) χ2=19.99, P<0.01, I2=80%	Photosensitizing thiazide diuretics	1	3.18 (1.93-5.26)
Vitiligo	2	5.14 (0.54-48.90) χ2=16.93 P=0.15, I2=94%	Potentially photosensitizing cardiac drugs	1	4.56 (2.92-7.13)
Warts	1	1.11 (0.82-1.50)	NSAID	1	0.63 (0.51-0.83)
Psoriasis	1	0.79 (0.50-1.24)	Carbamazepine	1	0.72 (0.12-4.23)
Baldness	1	7.29 (5.96-8.91)	Quinine	1	0.56 (0.16-1.99)
Family history of skin malignancies	1	1.82 (1.03-3.07)	Loop diuretics	1	1.16 (0.54-2.50)
Atopy ever	1	0.75 (0.68-0.83)	Novel oral anticoagulants	1	0.39 (0.07-2.05)
Vitamin usage			Warfarin	1	0.86 (0.30-2.49)
Vitamin D deficiency	1	0.39 (0.13-1.17)	ACE inhibitors	1	1.00 (0.54-1.87)
Retinoids	2	0.70 (0.31-1.58) χ2=2.01 P=0.39, I2=50%	Beta-blockers	1	1.31 (0.69-2.49)

Note: AK, actinic keratoses; NSAID, nonsteroidal anti-inflammatory drugs; NMSC: non-melanoma skin cancer; a, Junior high school/high school/ academic degree vs. None/primary school; b, BMI <18.5 and BMI ≥25.0; c, other colors in comparison to ‘black’; d, other colors in comparison to ‘blue’

Figure 2

Forest plot of the risk of AK associated with male sex (a), freckles on face/arms (b), and positive history of other skin cancer (c)

Other skin disease and risk of AK

Other skin diseases and AK are summarized in Table 2. A total of 5 studies reported the freckles and the risk of AK, and participants who have freckles on the face/arms had a higher risk of AK (OR = 1.88, 95% CI: 1.37–2.58; P < 0.01) [Figure 2b]. Moreover, participants with a positive history of other types of non-melanoma skin cancer were significantly associated with risk of AK (OR = 4.46, 95% CI: 2.71–7.33; P < 0.01) [Figure 2c]. Only a few studies reported other skin diseases. Among them, baldness and family history of skin malignancies increased the risk of AK, history of atopy reduced the risk of AK, and no obvious association was found in vitiligo, warts, and psoriasis.

Pigmentation status and risk of AK

Eight studies listing Fitzpatrick skin phototype [Table 2]. Compared with skin phototype III + IV, participants who have skin phototype I + II (light Fitzpatrick skin phototype) was significantly associated with risk of AK (OR = 2.32, 95% CI: 1.74–3.10; P < 0.01), and substantial heterogeneity was observed (I2 = 88%; P < 0.01). Furthermore, a significantly increased risk of AK was found among participants who have darker pigmentation in comparison to black hair (OR = 2.17, 95% CI: 1.40–3.36; P < 0.01) and blue eyes (OR = 1.67, 95% CI: 1.03–2.70; P = 0.04). Risk estimates between pigmentation status and AK are summarized in Figure 3.

Figure 3

Forest plot of the risk of AK associated with light skin phototype (a), light-colored hair (b), and light-colored eyes (c)

Sun exposure and risk of AK

Eleven studies reported the sun exposure status. Among them, participants who suffered sunburns in childhood (OR = 2.33, 95% CI: 1.47–3.70; P < 0.01), sunburns in adulthood (OR = 1.50, 95% CI: 1.12–2.00; P < 0.01), severe sunburn (OR = 1.94, 95% CI: 1.62–2.31; P < 0.01), and chronic occupational and/or recreational sun exposure (OR = 3.22, 95% CI: 2.16–4.81; P < 0.01) increased the risk of AK [Figure 4]. However, sunscreen use was significantly associated with the low risk of AK (OR = 0.51, 95% CI: 0.34–0.77; P < 0.01). Three studies reported the risk of AK among participants who had dwelling in tropical country, and no obvious association was found (OR = 1.06, 95% CI: 0.69–1.62; P = 0.79) [Table 2].

Figure 4

Forest plot of the risk of AK associated with sunburns in childhood (a), chronic occupational and/or recreational sun exposure (b), and sunscreen use (c)

Drug use and risk of AK

Many studies list no information of participants' drug use and risk of AK. Among the reported drugs, antiplatelet drugs, angiotensin-converting enzyme inhibitors, thiazides, angiotensin receptor AT1 blockers, photosensitizing thiazide diuretics, and potentially photosensitizing cardiac drugs increased the risk of AK, whereas nonsteroidal anti-inflammatory drugs reduced the risk of AK. No connection was found between calcium channel blockers, statins, antibiotics (doxycycline), antibiotics (ciprofloxacin), carbamazepine, quinine, loop diuretics, novel oral anticoagulants, warfarin, ACE inhibitors, beta-blockers, vitamin D deficiency, retinoids, and the risk of AK [Table 2].

Sensitivity analysis and publication bias

For risk estimates between gender and AK, age and AK, skin phototype and AK, sunburns status and AK, etc., the risk estimates of AK did not materially change after exclusion of any single study. Furthermore, both the funnel plot [Figure 5] and Egger's regression test (P < 0.01) indicated the evidence of publication bias.

Figure 5

Funnel plots of the AK risk associated with male sex (a), sunburns in childhood (b), freckles on face/arms (c), and positive history of other skin cancer (d)

Discussion

Numerous studies have shown that AK has a tendency to become cancerous. The risk factors, pathogenesis, and treatment of AK have attracted increasing attention with the increasing number of cancer patients.[829] Findings from the present meta-analysis indicated the known risk factors associated with AK such as the male sex, age, light pigmentation status, history of skin disease, sun exposure-related factors, and the use of some phototoxic drugs. Independent protective factors were sunscreen use and history of atopy. The present study is the first meta-analysis to summarize the risk factors of AK. With the accumulating evidence and enlarged sample size, we have enhanced statistical power to provide more precise and reliable risk estimates.

As expected, high cumulative UV exposure, including outdoor occupation and outdoor hobbies, with a positive history of sunburns during childhood, etc., were found to influence positively the development of AK.[22] In fact, in people continuously exposed to UV radiation, AK can be multiple, and they can develop into hyperkeratotic lesions clustered in larger fields, increasing the risk of progression into invasive SCC.[30] UVB radiation can create mutations by forming thymidine dimers in the DNA and RNA of keratinocytes, leading to the occurrence of AK. Further, alteration of p53 and its pathways can result in unregulated proliferation of dysplastic keratinocytes, leading to AK formation.[3132] Multiple studies highlight the effectiveness of sunscreen use, which has a dose-dependent effect on decreasing the development of new AKs, as well as increasing the remission rate of existing lesions.[1620] Therefore, it is not surprising that older people, men, and bald people are more likely to develop AK because they have higher cumulative UV exposure, and relatively speaking, they do not pay much attention to sun protection.[2333]

Individuals with Fitzpatrick type I or II skin characteristics, such as fair skin, freckles, light-colored eyes (blue or green), and light-colored hair (blonde or red hair), are more likely to develop AKs as lighter pigmentations are more sensitive to damage from chronic sun exposure.[34] The pooled OR indicated that lower educational level, abnormal BMI, alcohol consumption, and history of smoking were weakly associated with AK. These sociodemographic and behavioral characteristics are likely to slightly change sun-seeking behavior but not meet the long-term sun exposure standard. Furthermore, the association between the AK presence and potentially photosensitizing thiazide diuretics and cardiac drugs was proved.[2235] A plausible explanation might include the known photosensitizing potential and increased UV susceptibility for thiazide diuretics and some cardiac drugs.[34] The present study demonstrated that personal history of any skin malignancies is an independent factor determining the development of AK. Therefrom, there is a possibility that gene expression information might be modified in the person who suffered a personal history of skin malignancies, which result in personal physique sensitivity change to sunlight.[23] No similar studies were reported and it still needs to be further explored.

Preventative efforts are potentially cost-effective, may contribute in reducing new cases of AK, reducing concern about malignant lesions, and may avoid treatment side-effects. Therefore, it seems imperative to educate on the dangers of sun exposure, the use of photoprotection methods (e.g., use of covering clothes, hats, sunglasses, and sunblock and cessation of exposure in the mid-afternoon), as well as the signs before AK onset, more specifically and necessarily for the fair-skinned and elderly population. Various strategies, including workplace policies, provision of shade, and personal protective equipment, also showed positive protection results.[2136] Secondary prevention includes health surveillance performed by occupational physicians and dermatologists.[3738]

This study has some limitations. First, there are some significant differences in characteristics between the whole population and the high-risk population. However, the retrospective and hospital-based design did not allow to assess AK risk factors in the general population. Moreover, three-quarters of the included studies were from Europe; this may restrain the popularization and application of present findings. Second, when assessing many other risk factors listed in Table 2, we only relied on a small number of studies. Related published studies were not plentiful enough or did not qualify for meta-analysis, preventing us from drawing conclusions. Third, the heterogeneity in our study was substantial. However, the heterogeneity was expected and inevitable considering the differences in the ethnic background, study population, and methodology. Moreover, discrepancies between medical institutions may affect the diagnosis and introduce potential biases. Fourth, more than half of the included studies are case-control studies, which may lead to recall bias. Lastly, potential publication bias could influence the findings.

In conclusion, we confirmed several well-known AK risk factors and their quantitative data, and summarized the uncommon risk factors and protective factors for the disease. The summarized risk pattern established here for the probability of developing an AK may serve skin cancer prevention programs as a basis for the improvement of clinical strategies and behavioral purposes.

Financial support and sponsorship

Nil.

Conflicts of interest

There are no conflicts of interest.