Growth Hormone Assay-Adjusted Standardization Reveals Distinct Clinical Phenotypes in Acromegaly.

Area of Science:

• Endocrinology
• Medical research
• Machine learning applications in healthcare

Background:

• Acromegaly is characterized by excess growth hormone (GH) and insulin-like growth factor 1 (IGF-1).
• Standardization of GH assays is crucial for accurate patient assessment.
• Unsupervised machine learning can identify novel patient subgroups.

Purpose of the Study:

• To define distinct clinical phenotypes in acromegaly.
• Utilize growth hormone (GH) assay standardization and unsupervised machine learning for phenotype identification.
• Correlate identified phenotypes with clinical and biochemical parameters.

Main Methods:

• Multicenter cross-sectional analysis of 416 acromegaly patients.
• Stratification of patients based on standardized GH levels (GHxULN) using binary and four-tier classifications.
• Unsupervised cluster analysis incorporating age, GHxULN, IGF-1xULN, tumor diameter, and T2 signal intensity.

Main Results:

• 36% of patients had GH levels within the normal range (GH-B <1.0×ULN).
• Significant gradients in age, tumor size, IGF-1, and symptom duration were observed across GH categories.
• Cluster analysis identified 3 distinct phenotypes: older with lower GH/IGF-1, intermediate with moderate activity, and younger with aggressive tumors and elevated GH/IGF-1.
• Higher GH categories correlated with increased odds of arthropathy.

Conclusions:

• Standardized GH levels reveal clinically meaningful acromegaly phenotypes.
• These phenotypes correlate with age, tumor characteristics, and disease severity, including arthropathy.
• GHxULN complements IGF-1 in assessing secretory activity, supporting individualized clinical decision-making.