1 INTRODUCTION

One in five Americans is affected by skin cancer, making it the most common cancer in the United States. Excluding non-melanoma skin cancer (NMSC), melanoma is the fifth most common cancer, and it is projected that over 100 000 new cases of melanoma will be diagnosed in the United States by the end of 2020. Although melanoma accounts for only 1% of all skin cancers, it causes a majority of deaths and it is expected that almost 7000 people will die from melanoma by the end of the year.[1] Each year, over 5.4 million cases of NMSC are treated in the United States in over 3.3 million people.[2] Despite continuous efforts to promote public awareness about sunburn and skin cancer risk, sunburn remains highly prevalent among American adult populations (Figure 1). Surveys from the Centers for Disease Control (CDC) show that sunburn prevalence remains high, with 50.1% of all American adults and 65.6% of whites aged 18-29 years reporting at least one sunburn every year.[3]

Sunburn prevalence among American adults (adapted from American Cancer Statistics 2013, by American Cancer Society)

Sunburn usually diminishes several days after exposure; however, repeated sunburn causes cumulative genetic and epigenetic damage in skin cells. Although sunburn is a well-established risk factor in skin carcinogenesis, there are often decades-long delays between sunburn events and visible skin tumor lesions (Figure 2). Sunburn-induced molecular changes can persist for years to decades in sun-exposed pre-malignant skin, which can lead to malignant transformation over time. While conventional skin cancer screening methods including dermoscopy are useful in tumor detection, they often fail to detect tumors at early stages due to their inability to detect such cancer-causing molecular alterations prior to visible tumor formation.[4-6] Currently, measurement of skin sun damage relies on the use of minimal erythema dose (MED), which describes the amount of ultraviolent radiation (UVR) that produces visible skin redness within 24 hours after an exposure. The time to reach MED is dependent on the amount of UVR. Yet, MED is not an ideal marker; significant UV-induced molecular damage may occur following sub-MED UV exposure.[7, 8]

UV-induced skin tumor development often spans several decades between initial sunburn events and the formation of visible pre-cancer lesions that will progress to malignancies

Sunburn is largely preventable, and prevention is the most cost-effective strategy to reduce skin cancer incidence and treatment costs.[9] An important factor contributing to the ongoing epidemics of sunburn and skin cancer is related to a low public compliance with UV protection guidelines.[10] This low compliance is partly attributable to the lack of quantifiable risk information for both education and motivation of patients at risk. With the rising rates of skin cancers and treatment costs, there is an urgent need for more effective approaches for prevention and early detection to reduce healthcare spending, morbidity and mortality. The average annual cost for skin cancer treatment increased 125% from $3.6 billion between 2002 and 2006 to $8.1 billion between 2007 and 2011.[11] In contrast, the average annual cost of treatment of other cancers increased by only 25% from $63.7 billion to $79.7 billion during the same time period.

Biomarkers are used in a wide range of cancers in order to provide information about disease development, progression and prognosis. Over the years, there has been significant interest in biomarker development to improve disease prevention and early detection. Molecular signatures have the potential to identify a disease early and risk-stratify individuals based on susceptibility. Given the latency between sunburn and development of skin cancers as well as the challenges of early detection, a biomarker-based test has been of interest in risk assessment to improve skin cancer prevention and decrease time to diagnosis. As the costs of treatment far exceed those of photoprotective strategies, there is significant interest in both primary prevention and screening for high-risk populations to decrease incidence and improve early detection of skin cancers. Prognostic biomarkers will allow these high-risk populations to be identified for the purposes of targeted screening and prevention. Here, we will discuss emerging melanoma and NMSCs biomarkers that can help with risk stratification of the population and targeted primary and secondary prevention for early detection and treatment.

2 RISK FACTORS AND EMERGING BIOMARKERS FOR MELANOMA AND NMSC

UVR exposure leading to sunburn is a well-established risk factor for development of both melanoma and NMSC.[12-15] Biomarkers are molecules whose detection or evaluation provides information about a disease beyond the standard clinical parameters that are gathered by the clinician.[16] While there are various FDA-approved multi-gene panel tests for risk prediction and diagnosis of various cancers,[17] there is no FDA-approved biomarker test for risk stratification. Several studies in the past have attempted to identify UV-responsive genes.[18-23] No consensus UV biomarker panel is available today due to large variations among previous studies and lack of cross-validation of candidate biomarker genes. There are several classes of biomarkers that we will discuss here including markers of susceptibility, exposure, prognosis, progression and metastasis.

2.1 UV radiation as a risk factor

UVA and UVB radiation are mutagenic via induction of dimerization and breaks in DNA structure, with these so-called UV signature mutations commonly identified in melanoma skin cancer. 76% and 84% of primary and metastatic melanomas, respectively, have such signature mutations, and further, mutational burden (sometimes used for classification) correlates with the degree of sun exposure.[24, 25] These UV signature mutations are also seen in NMSC. Actinic keratoses (AK), squamous cell carcinomas (SCC) and basal cell carcinomas (BCC) have all been linked to mutations in TP53, a known tumor suppressor gene, and over 70% of these mutations are attributable to UVR.[26-28]

It has been shown that these somatic mutations are also found in normal sun-exposed skin without evidence of malignancy. One study using deep targeted sequencing of biopsies of sun-exposed eyelid epidermis found that on average, each cell had over 10,000 somatic mutations, most of which had a UV signature mutation.[29] Positively selected mutations were found in 18% to 32% of normal skin cells. Thus, aged sun-exposed skin can contain a significant proportion of cancer-causing mutations while maintaining the normal function of the epidermis, supporting the multi-stage model of carcinogenesis.[30, 31] These studies raise concern regarding the reliability of using mutation-based biomarkers in skin cancer risk assessment.

2.2 Markers of susceptibility

Markers of susceptibility for development of NMSC and melanoma include skin type and presence of heritable mutations. The Fitzpatrick skin phototype classification system is the most commonly used method to assess skin cancer risk. It characterizes skin pigment on a scale of I through VI, from light to dark and includes an individual's self-reported ability to tan or burn, with skin type I having a tendency to burn easily and tan poorly.[32] Evidence shows that the Fitzpatrick skin phototype classification system is a stronger predictor of skin cancer risk than pigmentary phenotypes including hair, eyes and skin color.[33] A limitation of this scale is that it may not be accurate in patients with darker skin tones.[34, 35] The number of common and atypical nevi has also been found to be an independent risk factor for the development of melanoma.[36] A meta-analysis has shown that the presence of over 100 common nevi compared with less than 15 is associated with a relative risk of 6.85 of developing melanoma, and presence of 5 atypical nevi compared with none was similarly associated with a relative risk of 6.36.[36]

Multiple heritable mutations are also associated with risk of NMSC and melanoma. For example, patients with xeroderma pigmentosum have mutations in nucleotide excision repair genes and the risk of developing skin cancer is increased over 1000-fold.[37] Patients with basal cell nevus syndrome have heritable mutations in the tumor suppressor gene PTCH.[38] While family history is an important risk factor for melanoma, these cases constitute only 1%-2% of all cutaneous melanoma.[39] In particular, the cyclin-dependent kinase CD4 gene and cyclin-dependent kinase inhibitor gene CDKN2A confer elevated risk in 20 to 40% of high-risk families.[39]

2.3 Measures of exposure to UV radiation

Several biomarkers measure exposure to UV radiation including the MED, gene expression changes and microRNA levels. MED is the current indicator of skin sun damage. As an indicator, MED is both insensitive and inadequate because significant UV-induced molecular damage may occur after sub-MED UV exposure.[7, 8] Other measures have been used as biomarkers of UVR exposures, including the number of benign nevi present in childhood, which is a known risk factor for the development of melanoma.[36] Additionally, the cyclobutane pyrimidine dimer and pyrimidine (6-4) pyrimidone photoproducts formed as a result of UVR can also be used to measure acute UV damage.

It has been shown that acute UVR exposures can result in significant transcriptomic instability, altering thousands of genes (Table 1).[8, 18] UVR upregulates expression of genes involved in cellular stress and inflammation including protein tyrosine phosphatase receptor type E, thrombospondin-1, inducible costimulatory ligand, galectins, Src-like adaptor protein, IL-10 and CCR7.[19] RNA sequencing examining UVB induced gene expression and found that 2,186 genes were significantly dysregulated in human skin 48 hours after exposure to UVB.[18] These included numerous chemokines and cytokines including interleukin 6 and 24, CCL3, CCL20, CXCL1, CXCL2, CXCL3, CXCL5, COX2 and members of the keratin gene family.[18]

Table 1. Representative UV biomarker genes identified by recent genomic profiling studies Study design UV-induced changes in gene expression qRT-PCR was used to validate selected UVB responsive genes among keratinocytes. Time point was 6 days after exposure to a total of 60 mJ/cm2 UVB radiation[8]

CCNB1, PRSS23, SERPINH1, PLK1 (repressed)

CDKN1A, S100A7, RNASE7, CNFN (activated)

RNA-seq used to perform genome-wide transcriptional profiling after UVB irradiation[18] 2186 genes significantly dysregulated in humans. Most significantly up-regulated genes encoding: IL6, IL24, CCL3, CCL20, CXCL1, CXCL2, CXCL3, CXCL5, COX-2, keratin family RT-PCR and gene set enrichment analysis (GSEA)[19] 40 genes induced > 2.0 fold by UVA/UVB (top 5: CDF15, IL1B, FDXR, PLK2, IL1A) and 24 genes suppressed > 2.0 fold by UVA/UVB (top 5: TFEC, PPFIBP2, LUC7L2, IFIT1, WDR67) UVR-mediated alterations in over 47,000 transcripts using Affymetrix Human Genome U133 Plus 2.0 microarrays[20] 84 genes suppressed > 2.0 fold by UVR (48 of known identity). 99 genes induced > 2.0 fold by UVR (57 known). Top genes elevated: cell cycle regulator CDKN1A, WNT pathway regulator DKK1, receptor tyrosine kinase EPHA2, growth factor GDF15, ferrodoxin reductase (FDXR), p53-inducible protein TP53I3, transcription factor ATF3, DNA repair enzyme DDB2, beta-adrenergic receptor ADBR2. WWOX (interacts with p53 and p73) was consistently suppressed) Chromatin immunoprecipitation (ChIP) studies were performed to characterize UV-induced epigenetic changes after UVR[40] UVR led to genome-wide loss of H3K27ac with regional gains in H3K27ac levels. Association between decrease in H3K27ac levels and reduced gene expression 72 hours after exposure AdductSeq and FreqSeq were used to identify hyperhotspots across the genome[41] UV exposure led to hyperhotspots acquiring cyclobutane pyrimidine dimers up to 170-fold more frequently in melanocytes compared to non UV-exposed cells

Examination of UVR-mediated alterations in epidermal melanocytes found that out of 47,000 transcripts, 84 genes (48 known identity) were suppressed over two-fold by UVR and 99 genes (57 known identity) were induced over twofold by UV radiation.[20] Several genes that target the TP53 pathway were identified including cell cycle regulator CDKN1A, Wnt pathway regulator DKK1, receptor tyrosine kinase EPHA2, growth factor GDF15, ferrodoxin reductase (FDXR), p53-inducible protein TP5313, transcription factor ATF3, DNA repair enzyme DDB2 and beta-adrenergic receptor ADBR2.[20] Epigenetic changes have also been associated with UVR. One study performed histone 3 lysine 27 acetylation (H3K27ac) chromatin immunoprecipitation (ChIP) studies to characterize UV-induced epigenetic changes after UVR.[40] It was found that UVR caused a genome-wide loss of H3K27ac with regional gains in H3K27ac levels. A significant association was found between the decrease in H3K27ac levels and reduced gene expression 72 hours after UV exposure, but not 4 hours after exposure.[40]

A recent study examined the genome of human fibroblasts and melanocytes for genome regions with heightened UV sensitivity.[41] Researchers identified 2000 “hyperhotspots” in the human genome that are up to 170 times more sensitive to UVR compared with the average genome. These hyperhotspots were areas with formation of cyclopyrimidine dimers, the chief UV photoproduct, and they occurred most often in melanocytes. The hyperhotspots were spread across the genome and were most frequent near genes, in particular those that regulate cell proliferation. These hyperhotspots are being investigated as a biomarker for predicting skin cancer development as the major risk of development of skin cancer is prior UV exposure. Given that CPDs accumulate in these hyperhotspots, the regions could be used as objective measures of UV exposure in small skin samples.

2.4 Biomarker of risk stratification of pre-cancerous lesions

Solar UVR is a major factor in the development of both SCC and their precursor lesions, AKs. Approximately 65% of SCCs arise from AK precursor lesions, whereas the estimated rate of progression from AK to SCC is <5%.[42] There is a growing interest in the development of reliable and sensitive tests to identify high-risk AKs and SCCs. One study examined expression of p53, E-cadherin, Snail, Slug and Twist in AKs to identify biomarkers that correlate with clinical progression and regression of AKs. It was found that p53 expression was significantly higher in clinically apparent AKs compared with regressed AKs. There was also significantly less membrane E-cadherin in clinically apparent AKs. Decreased E-cadherin levels are a marker of epithelial to mesenchymal transition. Snail, Slug and Twist are transcriptional repressors of E-cadherin, and these were increased in AKs compared with normal sun-exposed skin.[43] Studies are ongoing to determine genes that can differentiate high-risk AKs from more indolent lesions that will not progress into SCCs, which will facilitate the identification of benign AKs to avoid unnecessary treatment of the 95% of AKs that will not develop into malignant cancerous lesions.

2.5 Markers of disease progression

Biomarkers have also been applied to assess risk of disease progression and metastasis. Phosphorylated signal transducers and activator of transcription (pSTAT1, pSTAT3) are implicated in the development and progression of melanoma. The percentage of pSTAT3 positive melanocytes is associated with the degree of nevi atypia.[44] pSTAT1 and pSTAT3 have opposing functions biologically, and the pSTAT1/pSTAT3 ratio has been evaluated as a potential prognostic indicator with higher ratios in tumor tissue predicting improved overall survival of patients.[44] Additionally, treatment with IFNα increases the ratio in a dose-dependent manner.[44] Molecular markers have been identified that may characterize different stages of melanoma development. The melanoma inhibitory activity (MIA) protein is selectively expressed in melanoma cells but not melanocytes and is involved in tumor development and progression.[45] Serum levels of MIA have been used to distinguish metastatic melanoma from melanoma without lymph node involvement and control groups of patients with dysplastic nevi or BCC but no melanoma.[46] MIA is involved in progression and metastasis of melanoma by interacting with fibronectin and integrin, inhibiting cell matrix contacts and allowing for migration of melanoma cells to other tissues.[45] MIA is also involved in the development of melanoma by affecting the expression of transcriptional regulators including MITF and PAX3 that are involved in melanoma development.[47]

Recently, there are emerging interests in the role of microRNA (miRNA) in melanoma development, progression and metastasis. miRNAs are small (22 nucleotide) single-stranded non-coding RNAs that negatively regulate the expression of over 60% of the human genome. Circulating miRNAs have the potential to be used as a biomarker for the early diagnosis of melanoma.[48] First identified in the peripheral circulation in 2008, they are transported in microparticles or complexed with RNA-binding proteins or lipoproteins which protect them from degradation by RNAse.[49] Several studies have described the use of miRNA in distinguishing patients with melanoma from healthy controls.[48] One panel of 16 up- or down-regulated circulating miRNAs was able to differentiate between these populations with 95% specificity and 98.9% sensitivity.[50]

miRNA levels may also be an indicator of propensity for melanoma metastasis. One study compared miRNA levels among primary non-metastatic melanomas, primary metastatic melanomas, and metastases and found significant differences in expression levels of miR-145, miR-203-3p and miR-205-5p. miR145-5p and miR203-3p were significantly decreased in metastases compared with primary non-metastatic tumors. Additionally, a correlation was found between lower expression of these miRNA and pathological characteristics indicative of tumor aggressiveness such as Breslow thickness > 1 mm, high Clark level, ulceration and mitotic rate over 1/mm2.[51]

3 ROLE OF PRIMARY PREVENTION IN DECREASING THE INCIDENCE OF SKIN CANCER

As UVR is a risk factor for cutaneous malignancy, decreased exposure can prevent development of the genetic and epigenetic changes described. The importance of prevention has been highlighted by studies implicating UVR in nearly 70% and 90% of NMSCs and melanomas, respectively.[52-54] This is a complementary strategy to use in conjunction with UV biomarkers for both primary prevention and cost containment.

Public health campaigns have been implemented globally and are being initiated in the United States as well. These are described in Table 2. Established campaigns, such as SunSmart® in Australia, have shown success with decreased rates of melanoma.[55] It has been estimated that SunSmart® alone has prevented 50 000 cancers and 1400 deaths, saving over $92 million in the process.[56] Public health campaigns have been launched in the United States, but the efficacy of these programmes may not be apparent for decades.[57]

Table 2. Recent studies evaluating effectiveness of primary and secondary prevention measures Primary prevention Campaign Description Findings SunSmart® and Slip! Slop! Slap! Seek! Slide! media campaign[74] Initiated in Australia in 1988. Targeted prevention and detection programme

Successful in terms of melanoma-specific mortality and costs saved[56, 68, 75]

Since implementation of this campaign, the incidence of invasive melanoma in Australia in people under age of 55 is decreasing. The incidence has stabilized in the older population[76]

Ray and the Sunbeatables: A Sun Safety Curriculum®[57]

Initiated at MD Anderson in the United States in 2015

Education programme focusing on sun safety targeted to younger students (preschool to first grade). Now being expanded to include other age groups

Outcomes of this programme are not yet evident given its recent implementation Secondary prevention Skin Cancer Screening Campaign (SCREEN)

Initiated in Schleswig-Holstein, Germany in 2003

Based on full body skin examinations offered from 7/2003 to 6/2004 to individuals over age of 20

There was a significant decrease in melanoma mortality rates (nearing 50%) from the onset of SCREEN. Melanoma mortality rates returned to prior levels after the discontinuation of this programme.[69, 70] University of Pittsburgh screening study

Implementation of a screening programme in adults over the age of 35.

Primary care providers were trained in full body skin examinations, and thickness of melanoma was compared between screened and unscreened groups

Patients receiving annual screening had earlier detection of melanoma and decreased lesion thickness relative to unscreened patients[71]

The primary intervention in these campaigns is the promotion of sun-safe behaviours. These include the use of broad-spectrum sunscreens of sun protection factor (SPF) ≥30 and protective clothing. Regular application of sunscreen has been shown to have sustained, long-lasting effects on the incidence of primary melanomas (HR 0.50, 95% CI 0.24-1.02, P = .05) up to 10 years,[58] and childhood use similarly reduces risk in adulthood.[59]

While primary prevention has been shown efficacious, currently no guidelines exist in the United States for the use of sunscreen to prevent cutaneous malignancy. This lack of governmental support may stem from the mixed results of early studies. Many reasons have been posited for these mixed results, however, including increased sun exposure due to perceived protection,[60] inadequate sunscreen application,[61] limited follow-up periods[62] and latency from prior exposure.[62]

The human papillomavirus vaccine (HPV) has also been shown to prevent and treat keratinocyte carcinomas.[63, 64] One study looked at the prophylactic effect of the HPV vaccine on keratinocyte carcinoma development in two patients with a history of multiple SCCs and BCCs. One of the patients had a mean of 12 new SCCs per year and 2.25 new BCCs per year before vaccination and 4.44 SCCs and 0 new BCCs per year after vaccination (62.5% reduction in SCCs and 100% reduction in BCCs). The second patient developed a mean of 5.5 new SCCs per year and 0.92 new BCCs per year before vaccination and 1.84 SCCs and 0 BCCs per year after vaccination (66.5% reduction and 100% reduction in SCCs and BCCs, respectively). Thus, the HPV vaccine could be a promising prophylactic treatment for patients at high risk of developing keratinocyte carcinomas. Yet, currently the studies are primarily case series and case reports and there is a lack of evidence regarding the efficacy of this vaccine for primary prevention at the population level. Widespread adoption of the HPV vaccine for prevention of cervical cancer may constitute a natural experiment within this arena, providing valuable evidence for the efficacy in the prevention of cutaneous malignancy.

4 BIOMARKERS TO PROVIDE TARGETED SCREENING FOR HIGH-RISK PATIENTS

UV biomarkers can help to identify high-risk patients for secondary prevention in the form of targeted screening. This is important given that there are no recommendations in the United States currently for skin cancer screening. Both the 2016 United States Preventative Services Task Force (USPSTF) and a recent Cochrane review found insufficient evidence to recommend routine skin cancer screening.[65] Risk stratification can help to prevent screening of low-risk populations which may contribute to increased treatment costs with limited mortality benefit through the overdiagnosis of melanoma and skin cancer, as seen with other cancer types such as breast and prostate cancers.[66] Specialists recommend screening populations considered high risk based on the Melanoma Prevention Working Group's response to the USPSTF conclusions (Table 3).[67] Screening of a select group of patients may promote early diagnosis of melanoma, improving quality of life and reducing treatment costs for patients.

Table 3. Specialists’ suggestion for guidelines of screening a high-risk population. Adapted from Advances in Prevention and Surveillance of Cutaneous Malignancies, 2019 with permission from the authors[] Adults aged 35-37 years with one or more of the following risk factors should be screened annually with a total body skin examination Personal History

Melanoma, actinic keratosis or keratinocyte carcinoma

CDKN2A (or other high-penetrance gene) mutation carrier

Immunocompromised

Family History

Melanoma in one or more family members

Family history suggestive of a hereditary predisposition to melanoma

Physical Features

Light skin (Fitzpatrick I-III)

Blonde or red hair

>40 total nevi

Two or more atypical nevi

Many freckles

Severely sun-damaged skin

UVR Exposure

History of blistering or peeling sunburn

History of indoor tanning

The benefits of a nationwide screening programme have already been observed in Europe through population-based studies.[68] For example, the skin cancer screening campaign (SCREEN), implemented in Schleswig-Holstein, Germany in 2003, led to a decrease in melanoma mortality approaching 50%.[69, 70] Following the conclusion of the programme in 2008, the mortality rates returned to the baseline levels. Further supporting these findings, a study at the University of Pittsburgh found that annual full body skin examinations in patients over the age of 35 resulted in earlier identification of melanomas, and that identified specimens were 50% thinner than in unscreened patients.[71]

Currently, screening of high-risk patients is recommended in Australia, New Zealand, the Netherlands and the UK.[67] An Australian study found that such high-risk populations may have a 4-year melanoma risk as high as 18.2%.[72] The population included in this study fulfilled at least one of four criteria: (a) personal history of at least 1 invasive melanoma and dysplastic naevus syndrome, (b) personal history of at least 1 invasive melanoma and a family history of at least 3 first degree or second-degree relatives with MM, (c) personal history of at least 2 primary invasive melanomas with at least 1 occurring in the 10 years prior to recruitment, or (d) confirmed CDKN2A or CDK4 gene mutation. Identifying and monitoring high-risk patients improve outcomes through early detection and are cost-effective.[73] Biomarker-based tests can help to identify this high-risk population and effectively target screening.

5 CONCLUDING REMARKS

The rates of NMSC and melanoma in the United States continue to rise, which is partly attributable to the low compliance with sun protection guidelines in the general population. The non-compliance may be explained by the current difficulty to provide objective risk information to educate and motivate individuals at risk to avoid sunburn. Individuals with a significant sunburn history are often unaware of their risk of developing skin cancer because it can take years or decades for cancer-prone cells to form clinically visible cancer lesions following sunburns. There is an emerging role for biomarkers in risk stratification and early prevention to reduce skin cancer incidence and associated skin cancer care costs. Tumor-associated molecular biomarkers have the potential to detect cancer risk early to enable timely screening, early prevention and intervention. Compared to conventional methods, biomarker-based molecular tests can allow for early detection of cancer-causing molecular changes in sun-exposed skin lesions. Given the long lag between sunburn and visible cancerous lesions in the skin, biomarker-based diagnostic tests focusing on early risk assessment will greatly improve skin cancer prevention. Following biomarker-based risk assessment, primary prevention through education and secondary interventions can be targeted to the high-risk groups. In addition to prevention, biomarkers are also needed to risk-stratify patients and prognosticate specific lesions for appropriate treatment. Recent advances and breakthroughs in next-generation sequencing technology offer an excellent opportunity for biomarker identification and development. Successful validation of biomarker-based tests will likely transform how we assess skin sun damage and skin cancer risk to augment skin cancer prevention and treatment.

ACKNOWLEDGMENTS

Funding support from the NIH/NIAMS grant K01AR064315; the Prevent Cancer Foundation research award; the Columbia University Herbert Irving Comprehensive Cancer Center (P30 CA013696); the Columbia University Skin Disease Research Center (P30 AR44535).

CONFLICT OF INTEREST

The authors have no conflict of interest.

AUTHOR CONTRIBUTIONS

MHT wrote the initial manuscript. FS read and edited the manuscript. LJG read and edited the manuscript. LL involved in conception, writing of initial manuscript and editing. Each author has given final approval for the manuscript to be published.