# Clinical characteristics and local recurrence risk in patients with multiple actinic keratoses: a retrospective clinical data analysis

## TL;DR

This study identifies factors that increase the risk of recurrence in patients with multiple actinic keratoses and develops a model to predict individual risk.

## Contribution

A novel Cox regression model and nomogram were developed and validated to predict local recurrence risk in patients with multiple actinic keratoses.

## Key findings

- Higher lesion count and incomplete treatment response were identified as independent risk factors for recurrence.
- The predictive model showed moderate discrimination and good calibration for recurrence risk.
- A nomogram was created to enable individualized risk estimation for clinical use.

## Abstract

To evaluate the risk of local recurrence in patients with multiple actinic keratoses (AKs), to analyze the influence of host-, lesion-, and treatment-related factors, and to develop and validate a Cox regression model for predicting recurrence risk.

This retrospective study enrolled 148 patients with multiple AKs during January 2019-February 2022. Baseline characteristics, treatment modalities, and follow-up outcomes were collected. The Kaplan-Meier method was used to estimate cumulative recurrence rates. Cox regression analyses were performed to identify independent risk factors. The interaction effect between lesion count and treatment response was further assessed. A multivariate Cox predictive model was constructed, and its performance was validated using the 24-month calibration curve, Harrell’s C-index, and Brier score. A nomogram was developed for individualized risk prediction.

The cumulative recurrence rates at 12, 24, and 36 months were 22.3%, 35.6%, and 44.7%, respectively. Multivariate analysis identified higher lesion count (11-20: HR = 2.39; > 20: HR = 2.96) and incomplete treatment response (HR = 2.43) as independent risk factors. Immunosuppression and regular sunscreen use were not significant. Although visual analysis suggested elevated risk with more lesions and incomplete response, their interaction term was not statistically significant. The model demonstrated moderate discrimination (C-index = 0.630) and good calibration (Brier score = 0.176). The nomogram enabled individualized risk estimation.

Lesion burden and incomplete treatment response significantly predict recurrence in multiple AKs patients. The developed Cox model and nomogram offer a clinically useful tool for identifying high-risk individuals and optimizing management strategies.