# Nomogram model for predicting incomplete immune reconstitution in people living with HIV based on clinical characteristics

## TL;DR

This study creates a prediction model to identify people with HIV who may not fully recover their immune system despite treatment, helping doctors provide better care.

## Contribution

A novel nomogram model is developed to predict incomplete immune reconstitution in HIV patients based on clinical characteristics.

## Key findings

- Baseline CD4+ T cell count, age at ART initiation, and AST level were confirmed as independent predictors of immune non-response.
- The nomogram achieved AUC values of 0.896 in training, 0.903 in internal validation, and 0.766 in external validation.
- The model shows strong discrimination, calibration, and clinical utility for identifying immune non-responders.

## Abstract

Despite the success of antiretroviral therapy (ART) for human immunodeficiency virus (HIV) in China, immune non-response (INR) remains critical for the long-term quality of life of people living with HIV (PLWH). Although the consensus on diagnosis and management of immunological non-responders in HIV infection (Version 2023) was published in China to standardize diagnostic criteria, prediction models for INR based on these criteria remain scarce. This study aims to develop and validate a nomogram model for early identification of INR risk based on the diagnostic criteria in this consensus, so as to facilitate clinical intervention.

In this retrospective study, the primary cohort included 615 PLWH who initiated ART and completed over 4 years of follow-up at Zhongnan Hospital of Wuhan University (January 2016 to May 2025). They were randomly split into a training set (n=433) and an internal validation set (n=182) in a 7:3 ratio. An external validation set comprised 213 PLWH from Xishui County People’s Hospital (January 2012 to August 2025). Least absolute shrinkage and selection operator (LASSO) and multivariable logistic regression were used to identify independent predictors of INR, and a nomogram was constructed. The receiver operating characteristic (ROC) curves, calibration curves, and decision curve analysis (DCA) were employed to evaluate the discrimination, calibration, and clinical utility of the model in the training set and validation sets, respectively.

LASSO regression identified seven candidate variables: age at ART initiation, baseline CD4+ T cell count, body mass index (BMI), white blood cell count (WBC), hemoglobin (Hb), aspartate aminotransferase (AST), and World Health Organization (WHO) clinical stage. Subsequent multivariable logistic regression confirmed baseline CD4+ T cell count (p < 0.001), age at ART initiation (p = 0.001), and AST level (p = 0.014) as independent INR predictors. The resulting nomogram demonstrated area under the curve (AUC) values of 0.896, 0.903, and 0.766 in the training, internal validation, and external validation sets, respectively. The calibration curve and DCA further indicated satisfactory consistency and clinical net benefit.

The developed nomogram effectively predicts INR risk in PLWH initiating ART, providing clinicians a practical tool for individualized management to improve patient prognosis.