The clinical course of actinic keratosis correlates with underlying molecular mechanisms.

Summary

Background:

Actinic keratoses (AKs) are common premalignant skin lesions triggered by excessive UV exposure. The majority of AKs regress or persist, but some progress to squamous cell carcinomas. The biomarkers associated with their persistence, progression and regression have not been characterized.

Objective:

We performed skin biopsies in patients with extensive actinic damage to identify biomarkers that correlate with clinical progression and regression of AKs.

Methods:

This was an observational study of a cohort of patients with extensive actinic damage. AKs were mapped on a clear plastic template in 26 subjects at months 3, 6, 9 and 11. Biopsies were taken from randomly selected, pre-determined AKs and were evaluated for p53, E-cadherin, Snail, Slug and Twist.

Results:

p53 exhibited greater expression in clinically apparent AKs (2.89±1.45) than regressed AKs (0.75±0.96); p<0.01. There also was significantly less membrane E-cadherin, the lack of which is a marker of epithelial-mesenchymal transition (EMT), in clinically apparent AKs (1.89±1.81) than sun-exposed (SE) skin (3.07±1.75); p<0.005. E-cadherin transcription repressors, Snail, Slug and Twist, were increased in AKs compared to SE skin.

Limitations:

Histological biomarkers were not a primary endpoint. In addition, subjects were allowed to apply sunscreens.

Conclusion:

At the molecular level, loss of E-cadherin and an increase in p53 are linked to the dynamic interplay between persistence, progression and regression of AKs.

Introduction

Actinic keratoses (AK), also known as solar keratoses, are common dysplastic epidermal lesions that result from chronic and excessive UV exposure. Clinically, they range from small, flesh colored, erythematous papules to large plaques with yellow or brown scales and crusts that appear on sun exposed areas of the face, scalp, and extremities^1,2. In individuals with multiple AKs, the cumulative lifetime risk of developing an invasive squamous cell skin cancer is as high as 6–10%³. Histologically, AKs are composed of atypical keratinocytes characterized by alternating areas of parakeratosis and orthokeratosis, a loss of polarity, pleomorphic or hyperchromatic nuclei and an inflammatory infiltrate comprised of lymphocytes or plasma cells.

AKs are more frequent in individuals over the age of 45 due to escalating photodamage, from cumulative sun exposure^4,5. Other risk factors include lower Fitzpatrick sun reactive skin types, residence in lower latitudes, male gender, baldness, a prior history of skin cancer and an immunocompromised state^4,6–8. Management of actinically damaged skin represents a considerable burden to the healthcare systems of many countries worldwide⁵. Based on U.S. health claims data, estimates are that over 10 million individuals in the US are evaluated and treated annually for actinically damaged skin, at a cost of $1.68 billion⁵.

The p53 tumor suppressor gene controls cell proliferation, regulates apoptosis, and facilitates DNA repair by interrupting cell cycle progression. p53 mutations are the most common genetic alteration associated with the development of various types of cancers including squamous cell carcinoma (SCCs)^9,10. Chromosomal p53 mutations are present in 90% of SCCs and up to 100% of AKs⁹.

E-cadherin is an intercellular adhesion molecule that plays a significant role in maintaining epithelial architectural integrity. E-cadherin repression permits non-motile epithelial cells to transform into motile mesenchymal-like cells through a process known as epithelial-mesenchymal transition (EMT)^11,12. The resulting cellular phenotype is characterized by loss of intercellular bonds and an increase in migrational ability leading to tumor invasion¹³. E-cadherin expression is a highly regulated process, and the E-cadherin transcription repressors Snail, Slug and Twist play an important regulatory role in EMT¹⁴.

The purpose of this study was to investigate the natural history of AKs in subjects with extensive actinic damage and to determine whether biomarkers might predict which AKs are likely to progress to keratinocyte cancers (KC) (also known as non-melanoma skin cancers [NMSC]. KC and NMSC are used interchangeably in this manuscript) and which are likely to regress back to non-AK-bearing, sun-exposed skin.

Materials and Methods

AK Mapping

This was an observational study of a cohort of patients with extensive actinic damage which used data and biopsy specimens collected from 26 participants in the celecoxib non-melanoma skin cancer chemoprevention trial who were randomized to receive placebo¹⁵. Only samples from patients on placebo were included in the study. Approval for the study was obtained prior to its initiation from the UAB Institutional Review Board. Individuals at high risk for non-melanoma skin cancer were eligible to participate if they were over 18 years of age, with an extensive number of AKs, and Fitzpatrick skin types I, II, or III. To be qualified as high risk, subjects were required to have 10–40 actinic keratoses on the upper extremities, neck, face and scalp along with prior histologic evidence of at least one AK or non-melanoma skin cancer¹⁵. Participants were instructed to abstain from topical prescription and over the counter medications; however, moisturizers, emollients and sunscreens were allowed. Subjects were evaluated for AKs at 0, 3, 6, 9 and 11 months. A study investigator marked the location of each actinic keratosis on a clear plastic template that was placed over the upper extremities, neck, face, and scalp. Lesions with characteristic scaling, keratotic patches, erythema, and sandpaper-like scale were identified as actinic keratoses¹⁵. Study investigators used different plastic templates to mark AKs at each visit and were blinded to the template results from previous visits. At the end of the study, AK templates from months 0, 3, 6, 9 and 11 were compared to assess the number of new, persistent and regressed AKs.

Skin Biopsies and Histological Grading

At baseline, two AKs on the upper extremities were randomly selected for biopsy, one of which was removed at baseline and the other at month 9. If, at month 9, an AK selected for biopsy was no longer present clinically, the site at which it had been located (i.e. regressed AK) was still biopsied. AK biopsies were evaluated based on degree of proliferation and severity of atypia.

Each component in this grading scheme was given a score of 1 to 3 (Table 1). Nodular basal cell carcinoma (BCC) and well-differentiated SCC samples were obtained from the UAB tissue bank and were not from the subjects that had been followed for 11 months.

Table 1.

Histologic Grading Scheme

Atypia	Proliferation	Score
<1/3 epidermal involvement	No downward budding	1
1/3 – 2/3 epidermal involvement	Focal downward budding	2
2/3-full epidermis involvement	Extensive downward budding with irregular contours	3

Biomarker Expression

Biomarker expression was graded based on staining intensity and the continuity of expression. For p53, Snail, Slug and Twist: 0, no staining; 1, minimal; 2, mild; 3, moderate; 4, strong; 5, very strong. For E-cadherin: 0, no staining; 1–2 minimal; 3–4, mild; 5–6, moderate; 7–8, strong; 9–10, very strong.

Immunofluorescence Staining

Paraffin tissue sections after deparaffinization and rehydration were processed further for immunofluorescence staining. Nonspecific antibody binding was blocked with 2% bovine serum albumin (BSA). After overnight incubation with primary antibodies E-cadherin and p53 (Santa Cruz Biotechnology, Dallas, TX), the tissue sections were incubated with fluorescence – coupled secondary antibody and visualized using fluorescence microscopy.

Immunohistochemistry Staining

The paraffin-embedded tissue sections were deparaffinized and rehydrated. The antigen unmasking was done with 10 mM citrate buffer (pH 6.0) and the nonspecific antigen binding was blocked with incubation with 2% BSA. The sections were incubated overnight at 4°C with primary antibodies Snail, Slug and Twist (Santa Cruz Biotechnology, Dallas, TX). This was followed with an incubation with peroxidase coupled secondary antibody and finally, visualization with 3,3’-diaminobenzidine (DAB) substrate.

Statistical Analysis

The primary objective of this analysis was to assess if there are biomarkers which can correlate with progression to skin cancer. The secondary objective was to determine the natural history of actinic keratoses. An independent student’s t-test was conducted to compare biomarker expression grading scores among the various tissue sample groups.

Results

Twenty-six Caucasian subjects with extensive actinic damage on the upper extremities, neck, face and scalp were enrolled in the study. Table 2 summarizes the patient characteristics. A total of 610 AKs were present at baseline with a mean±SD of 23.5±7.0 AKs/person. The total number of lesions remained constant over the 11-month observation period (Table 3). At 11 months, there was a mean increase of only 1.7±11.8 AKs/subject from baseline, which was not statistically significant. This is because over the 11-months, a nearly equivalent number of new AKs appeared as regressed. Overall, 43% clinically regressed without recurrence; 87 (14.3%), 52 (8.5%), 42 (6.9%) and 82 (13.4%) regressed at 3, 6, 9, and 11 months, respectively. The mean time to regression was 6.2 months. On the other hand, 32.6% AKs present at baseline regressed and recurred at one of the follow-up visits; 24% never regressed; 8% regressed twice; and 1% regressed twice and recurred twice.

Table 2.

Patient Characteristics

Age Range (years)	Number of Subjects
30–39	1
40–49	2
50–59	7
60–69	8
70–79	8
Fitzpatrick Skin Type
I	4
II	16
III	6
Sex Distribution
Males	19
Females	7

Table 3.

Number of actinic keratoses (AKs)present on each subject

Time Point	Mean no. of AKs per subject ± SD	Total no. of AKs
Baseline	23.5 ± 7.9	610
Month 3	25.8 ± 13.1	671
Month 6	23.1 ± 11.6	600
Month 9	27.8 ± 16.9	723
Month 11	25.0 ± 16.6	650

AK Histological Features.

There were 18 pairs of biopsies evaluated at baseline and at nine months, for a total of 36 biopsies. Of the 36 AKs clinically present and biopsied, 5 were not confirmed as AKs based on histological analysis (lichen simplex chronicus or solar lentigines). Five of 18 nine month biopsies had regressed and did not show histological evidence of AK. There was not a statistically significant change in the degree of proliferation or atypia of the AKs over the 9-month period of time (Fig 1).

Figure 1.

Histologic score in the atypia and proliferation severity of randomly selected randomly actinic keratoses (AKs) at baseline and biopsied at 9 months. p=0.52, proliferation at baseline vs month 9; p=0.44, atypia at baseline vs month 9; p=0.37, atypia and proliferation (both) at baseline vs month 9.

Biomarker Analysis

Because changes in mutant p53 and E-cadherin are associated with SCC development in animal models.^9,11,16, we sought to determine whether there was a change in p53 and/or E-cadherin expression when sun protected skin, chronic sun damaged skin, and AKs were compared. AKs and sun-exposed skin both expressed p53. Significantly reduced p53 was observed in regressed AKs and non-sun exposed skin samples compared to clinically apparent AKs and sun exposed skin (Fig 2). BCC and SCC specimens were not collected from the patients enrolled in the study. Therefore, representative samples were taken from the UAB tissue bank of cutaneous diseases. Those tissues showed enhanced expression of p53 relative to non-sun-exposed skin.

Figure 2.

a. Average nuclear expression score of p53 in skin biopsy samples. Non sun-exposed skin (NSE), sun exposed skin (SE), clinically present actinic keratoses (AK), regressed AK, squamous cell carcinoma (SCC), basal cell carcinoma (BCC) and combined SCC and BCC i.e. non melanoma skin cancers (NMSC). b) Immunoflourescence of p53 with DAPI (4′,6-diamidino-2-phenylindole). The arrows depict nuclear expression of p53.

Since E-cadherin is associated with progression of skin cancers in animal models¹¹, sun protected skin, chronic sun damaged skin, AKs and KCs were analyzed for E-cadherin expression. There was a progressive decline from non-sun exposed skin to sun exposed skin to AKs (Fig 3). E-cadherin expression in BCC and SCC was similar to that of AKs, suggesting at this point it may be more of a marker of AK persistence rather than progression to KCs. In sites in which AKs had regressed, E-cadherin expression was less, though not significantly, than AKs that had not regressed (Fig 3).

Figure 3.

a. Average E-cadherin expression in skin biopsy samples. Non sun-exposed skin (NSE), sun-exposed skin (SE), clinically present actinic keratoses (AK), regressed AK, squamous cell carcinoma (SCC), basal cell carcinoma (BCC) and combined SCC and BCC i.e. NMSCs. b. Immunoflourescence of E-cadherin with DAPI (4′,6-diamidino-2-phenylindole). Note the cell membrane staining that outlines the cells is greater in the NSE and SE than in the AK, regressed AK, BCC and SCC.

E-cadherin is carefully regulated by the transcription repressors, Snail, Slug and Twist¹⁶. We investigated whether these proteins might be altered in actinically damaged skin. Sun-exposed skin, AKs, BCC and SCC expressed significantly greater levels of Snail, Slug and Twist compared to non-sun exposed skin (Figs 4). In addition, expression levels of these proteins were significantly reduced in regressed AK samples compared to SE skin, AKs and BCC (Figs 4).

Figure 4.

a. Average group score of Snail, Slug, and Twist expression in non sun exposed skin (NSE), sun exposed skin (SE), clinically present actinic keratoses (AK), regressed AK, squamous cell carcinoma (SCC), basal cell carcinoma (BCC) and combined SCC and BCC i.e. non melanoma skin cancers (NMSC). b. Histochemical staining for Snail, Slug and Twist.

Discussion

Actinic keratoses are common pre-neoplastic lesions, which are precursors and risk factors for cutaneous squamous cell carcinomas.^{1,6,7,17–20} Their prevention and treatment reduces the significant likelihood of developing one or more KCs¹. We found that the appearance and regression of AKs is a dynamic process. The same AKs that were present at one time were absent at another, but had the capacity to reappear later. Over 40% of AKs identified at baseline regressed over the 11-month observation period, 33% recurred, 11% regressed twice, and a small number reappeared twice. As a result, the aggregate number of AKs over 11 months remained relatively constant. Moreover, there was no change in the histological severity of the AKs biopsied at 9 months, suggesting that additional insults are required for them to progress on to become squamous cell carcinomas. In this regard, treatment with the cyclooxygenase-2 inhibitor celecoxib has been shown to diminish the incidence of BCCs and SCCs, but has no effect on AKs¹⁵.

The percentage of lesions that regressed in our analysis is similar to that reported in some²¹, but not all studies^22–26. This may be due to differences in the extent of actinic damage, the experimental design, regional AK prevalence, and/or factors such as patient age, gender, sunscreen use and/or amount of UV exposure.

There are multiple options for the treatment of AKs, including, but not limited to, local (liquid nitrogen cryotherapy, electrodessication and curettage) and field (topical 5-fluorouracil, imiquimod, diclofenac, ingenol mebutate, and photodynamic therapy) therapy¹. The concept that AKs can regress and recur has important clinical implications for AK treatment in individuals with extensive actinic damage. Since only a minority of actinic keratoses are present at any one time, either field treatment or repeated aggressive local treatment is necessary for adequate management of these lesions. Based on the observation that AKs can persist, progress or regress, another implication is that it should be possible to intervene to promote regression and/or retard progression of actinic keratoses.

A main goal of this study was to identify biomarkers that are associated with progression and regression of AKs, in order to develop pharmacologic agents that might be employed to inhibit the development of keratinocyte carcinomas (KC). Our analysis found significantly increased p53 expression in AK, BCC, SCC and SE samples compared to non-sun exposed and regressed AKs. p53 alterations have been associated with the early stages of UV-induced carcinogenesis both in animal models and in human studies^9,27. Mutant p53 exhibits a longer half-life than its wild type counterpart and, as a result, is likely to accumulate in the cell nucleus^28–30. We observed that there was a progressive increase in nuclear p53 staining as the skin progressed from actinically damaged skin to AK to KCs. p53 decreased in regressed AKs. These findings suggest that p53 may be a good biomarker of AK progression. While it is uncertain whether AKs can progress to BCCs, it is well-established that people with large numbers of AKs are at increased risk of developing BCCs³. This association is most likely due to the fact that extensive sun exposure causes both.

Down-regulation of E-cadherin expression, which leads to a reduction in intercellular adhesion, has previously been linked to AK and KC progression^31,32. In contrast to the increased expression of p53 with more advanced pre-malignancies and malignancies, there was a progressive decrease in E- cadherin from non-sun exposed skin to sun-exposed skin to AKs. Although significantly less than non-sun-exposed skin, it is important to note that there was no further decline in E-cadherin from AKs to NMSCs, implying that loss of E-cadherin may be required for clinical progression of normal skin to sun damaged skin to AKs, but not for further progression to NMSCs.

Snail, slug and twist contribute to the down-regulation of E-cadherin and consequently have been connected to invasion and metastasis of cancer cells^11,33. Various studies have suggested that such transcriptional repressors are involved in KCs as well^34–36, but none have examined the implications of Snail, Slug and Twist in UV exposed, AK and regressed AK samples. There were higher levels of these proteins in UV exposed skin and in AKs than in control non-UV exposed skin. The corresponding reduction of E-cadherin in these samples suggests that increased activity of these transcriptional regulators in AKs and KCs contribute to the loss of E-cadherin and ultimately to the progression of sun-exposed skin to UV-induced cutaneous tumors.

One limitation of our study was that biomarkers were not a primary endpoint of the study. In addition, for ethical reasons, all patients were allowed to use sunscreen which may itself alter the natural history of AKs.

An estimated 5 million new cases of NMSC will be treated in the United States this year³⁷. While the mortality rate for most NMSCs is not great, they can be locally destructive and their treatment represents a tremendous economic impact on the US healthcare system and healthcare systems around the world⁵. Current methods for their prevention (AK treatment, limiting sun and tanning bed exposure, sunscreens and other photoprotective measures) have not been successful in stemming the increasing incidence of these types of malignancy. Thus, there has been great interest in identifying novel agents for chemoprevention. By examining their ability to alter changes in the expression of p53 and E-cadherin, findings from this study can be employed to determine the potential efficacy of existing and novel measures for the prevention of NMSCs. Moreover, p53, E-cadherin or one of the E-cadherin repressors could be used as a biomarker to identify subsets of AKs that are at high risk of progressing to SCCs. They also confirm that p53 and E-cadherin are relevant targets for the prevention of skin cancer. In this regard, preclinical studies from our laboratory have shown that CP-31398, a pharmacologic agent that produces conformational changes in mutant p53 enhancing its biological activity, confers resistance to UV-induced cancer development when applied topically³⁸. Additional therapeutic strategies that enhance E-cadherin expression or lower mutant p53 levels may be effective in reversing the rising incidence of KC.