Related Experiment Videos
Risk factors for peripheral atherosclerosis. Retrospective evaluation by stepwise discriminant analysis
Insights
Apolipoproteins and lipid parameters significantly differentiate peripheral atherosclerosis. Mathematical analysis of these lab values improves diagnostic accuracy, reducing classification errors for better patient outcomes.
Area of Science:
- Cardiovascular Medicine
- Biochemistry
- Medical Diagnostics
Background:
- Peripheral atherosclerosis is a complex condition influenced by various risk factors.
- Accurate diagnostic markers are crucial for early detection and management.
- Lipids, lipoproteins, and apolipoproteins are key metabolic indicators.
Purpose of the Study:
- To identify optimal discriminators for peripheral atherosclerosis.
- To compare the efficacy of apolipoproteins versus lipids in classification.
- To develop a more accurate diagnostic model using discriminant analysis.
Main Methods:
- Retrospective study of 49 male patients and 39 male controls (40-60 years).
- Analysis of lipids, lipoproteins, and apolipoproteins, alongside common risk factors.
- Application of linear regression discriminant analysis with variable selection.
Main Results:
- Apolipoprotein A-II/A-B and A-I/B ratios, apolipoprotein B, total cholesterol, and LDL-cholesterol showed significant differences.
- Apolipoproteins were superior discriminators compared to lipids alone.
- A 14-variable model achieved an error rate below 5%, significantly improving classification accuracy.
Conclusions:
- Apolipoproteins are strong indicators for peripheral atherosclerosis.
- Integrating lipid and apolipoprotein data through mathematical models enhances diagnostic precision.
- Advanced discriminant analysis offers a more sensitive and specific classification of peripheral vascular atherosclerosis.
Abstract:
To evaluate the optimal discriminators for peripheral atherosclerosis, we studied retrospectively 49 male patients and 39 male controls between 40 and 60 years of age. In addition to hypertension, cigarette smoking, diabetes mellitus, and hyperuricemia, we determined the most common lipids, lipoproteins, and apolipoproteins. Highly significant differences of median values between patients and controls in decreasing order of magnitude were recorded for apo A-II/apo B, apo A-I/apo B, apo B, total cholesterol, and LDL-cholesterol. A retrospective classification of patients and controls under optimal conditions with one variable (apo A-I/apo B) yielded an error rate of 25%. We found that apolipoproteins were better discriminators for peripheral atherosclerosis than than were lipids or lipoprotein lipids. The application of a linear regression discriminant analysis including 29 variables greatly decreased the rate of error and increased the sensitivity and specificity of the classification. From 229 possible models, we used an economic selection strategy to sort out those which either gave the best segregation or were considered the most practicable. The optimal model with 14 variables gave an error rate of less than 5% for the group studied. Suboptimal models yielded error rates between 13% and 18%. We conclude that a mathematical treatment of laboratory data which includes lipid parameters in addition to apolipoprotein values can improve the classification of peripheral vascular atherosclerosis.