A nomogram for predicting carotid plaque in elderly patients using the ApoB/ApoA1 ratio: a retrospective cohort study

Technical Summary: A Nomogram for Predicting Carotid Plaque in Elderly Patients Using the ApoB/ApoA1 Ratio

Problem Statement
Carotid plaque (CP) is a critical modifiable precursor to cerebrovascular events, with prevalence in the elderly population (≥60 years) ranging from 49% to 90% and increasing with age. While ultrasound is the diagnostic gold standard, its utility as a routine screening tool in primary care is limited by equipment availability, operator expertise, and time constraints. Existing risk scores, such as the Framingham system, are designed for coronary events and may not accurately capture carotid plaque risk, particularly in the elderly where traditional lipid parameters show diminished predictive value. There is a distinct lack of simple, individualized risk stratification tools that rely solely on routine clinical chemistry data to help general practitioners identify high-risk individuals within this high-prevalence population.

Methodology
This study employed a retrospective cohort design involving 452 elderly patients (age ≥60) who visited the general practice clinic of Xuzhou Central Hospital between June 2023 and March 2025.

• Data Collection: Data included demographics, medical history, and routine laboratory tests (fasting glucose, HbA1c, lipid profiles including ApoA1 and ApoB). Carotid plaque was defined via ultrasound as focal intima-media thickening (IMT ≥1.5 mm or localized increase ≥0.5 mm).
• Study Design: The cohort was randomly split into a training set (n=317) and a validation set (n=135) at a 7:3 ratio.
• Variable Selection: Least Absolute Shrinkage and Selection Operator (LASSO) regression with 10-fold cross-validation was used to screen predictors. Variables retaining non-zero coefficients were subsequently analyzed using multivariable logistic regression.
• Model Development: A nomogram was constructed based on the independent predictors identified.
• Evaluation: Model performance was assessed using:
  • Discrimination: Area Under the Curve (AUC) from ROC analysis.
  • Calibration: Calibration plots and the Hosmer–Lemeshow test.
  • Clinical Utility: Decision Curve Analysis (DCA) to evaluate net benefit across threshold probabilities compared to universal treatment or no treatment strategies.
  • Internal Validation: Bootstrap resampling (1,000 iterations) was used to adjust for over-optimism.

Key Contributions
The study developed and internally validated a nomogram specifically for the elderly general practice population that integrates four accessible variables: Age, Sex, HbA1c, and the ApoB/ApoA1 ratio.

• Novel Predictor Integration: It is the first study to construct a multivariable model specifically incorporating the ApoB/ApoA1 ratio to estimate carotid plaque risk in the elderly general-practice setting.
• Re-stratification Capability: Unlike models designed to detect the mere presence of plaque in low-prevalence groups, this tool is designed to re-stratify risk within a high-prevalence cohort (76.33% plaque presence), helping clinicians identify which patients are at the greatest risk for future events.
• Practicality: The model relies exclusively on routine blood tests and demographic data, avoiding the need for immediate ultrasound or specialized imaging for initial risk assessment.

Results

• Predictors: Multivariable analysis confirmed four independent predictors:
  • ApoB/ApoA1 Ratio: The strongest predictor (OR = 5.541, 95% CI 1.108–27.706, P=0.037).
  • HbA1c: (OR = 1.405, 95% CI 1.053–1.874, P=0.021).
  • Age: (OR = 1.075, 95% CI 1.032–1.119, P<0.001).
  • Sex: Female sex was protective (OR = 0.515, 95% CI 0.290–0.917, P=0.024).
• Performance:
  • Discrimination: The model achieved an AUC of 0.704 (95% CI 0.636–0.772) in the training set and 0.782 (95% CI 0.696–0.869) in the validation set.
  • Calibration: The Hosmer–Lemeshow test indicated adequate calibration (Training: P=0.458; Validation: P=0.790).
  • Clinical Utility: DCA demonstrated a net clinical benefit over "treat all" or "treat none" strategies across threshold probabilities of 0.41–0.915 (training) and 0.48–0.99 (validation).

Significance and Claims
The authors claim that this nomogram provides a practical decision-support tool for general practitioners to formulate individualized screening, monitoring, and intervention strategies (e.g., initiating or intensifying statin therapy) for elderly patients.

• Superiority of ApoB/ApoA1: The study highlights that the ApoB/ApoA1 ratio outperformed traditional lipid markers (like Total Cholesterol or LDL-C) and other univariate non-significant factors in the multivariable model, reflecting a more comprehensive balance of pro- and anti-atherogenic lipoprotein metabolism.
• Contextual Relevance: The tool addresses the specific challenge of risk stratification in a high-prevalence population where the clinical question shifts from "does the patient have plaque?" to "which of these patients with plaque is at highest risk?"
• Limitations: The authors modestly acknowledge that the study is single-center and retrospective, requiring external validation in multi-center, diverse populations. They also note that the sample size, while sufficient for internal validation, resulted in wide confidence intervals for the ApoB/ApoA1 ratio, and that the model currently lacks external validation and long-term prospective data on plaque progression or stroke outcomes.