Actinic keratoses are common pre-cancerous skin lesions that can transform into cutaneous squamous cell carcinoma and whose presence may also be associated with increased risk for melanoma.

Objectives: This project aims to examine the absolute risks of melanoma in patients with actinic keratoses, which are unknown, and to create a melanoma risk prediction model for patients with actinic keratoses, which has not previously been done.

Rationale: Actinic keratoses arise in the setting of chronic UV exposure and affect tens of millions of people in the United States each year. The majority of actinic keratosis clinical care and research focuses on the individual actinic keratosis and its risk of malignant transformation to cutaneous squamous cell carcinoma. However, there is some evidence that the presence of actinic keratoses may be associated with increased risk of melanoma for the patient as a whole. Though they are not currently used this way, actinic keratoses may be an important clinical biomarker of melanoma risk. Unfortunately, there are no current guidelines or recommendations for clinicians to follow for melanoma surveillance or early detection in patients with actinic keratoses. The overarching goal is of this project to provide evidence to guide clinical care and form the foundation for future recommendations on melanoma surveillance and early detection in patients with actinic keratoses, a large and high-risk group.

Methods: In Aim 1, we will determine the absolute risks and timing of melanoma in patients with actinic keratoses using a retrospective cohort design in a dataset of 5 million Medicare beneficiaries. We will use Kaplan Meier and parametric accelerated failure time models to calculate unadjusted and adjusted absolute risk estimates, respectively. Aim 2, we will create and validate a risk prediction model for melanoma in patients with actinic keratoses using a Medicare claims dataset, a commercial claims dataset (Optum, 69 million patients), and UK Biobank (500,000 patients). We will create the risk prediction model in Medicare using routinely collected data available in claims and validate it in Optum. Then we will test the performance of our model using only routinely collected data against that of a model that also includes phenotypic, UV exposure, and genetic information in UK Biobank.

Expected Results: The completion of these aims will build the evidence-base for needed recommendations for tens of millions of patients with actinic keratoses each year in the US who are at increased risk for melanoma. The methodology, results, and research skills developed as part of this application will be utilized in future studies in skin cancer risk prediction. Completion of the proposed research and career development plan will serve as a platform upon which Dr. Wehner can successfully transition to scientific independence in patient-oriented skin cancer research.