Conference Coverage

Sex-Specific Models Improve ATTR-CM Detection, Particularly in Women Missed by Conventional Screening

Ashton L. Stahl

Key Highlights

  • The study included 28,758 women and 34,468 men with heart failure, including 96 women and 403 men with ATTR-CM.
  • Sex-specific screening models outperformed the Mayo Simple Score, especially among women.
  • The XGBoost model achieved a discrimination score of 96.8 out of 100, compared with 80.3 for the Mayo Simple Score.
  • External validation in a separate heart failure cohort is planned.

Sex-specific screening models improved discrimination of transthyretin amyloidosis with cardiomyopathy (ATTR-CM) in a retrospective heart failure cohort, with particularly strong performance among women. The authors presented the findings at the 2026 American Heart Association Scientific Sessions.

The investigators examined whether routinely collected clinical information at the time of heart failure diagnosis could help identify women missed by conventional ATTR-CM screening. They noted that women may be underdiagnosed because of sex-specific differences in cardiac structure and clinical features.

The study included patients with and without ATTR-CM within an integrated academic healthcare system. We identified patients through a diagnostic algorithm and clinical adjudication. The investigators developed sex-specific XGBoost and L1-penalized logistic regression models using characteristics available when heart failure was diagnosed.

The researchers divided the data into training, validation, and held-out testing sets containing 60%, 20%, and 20% of patients, respectively. The investigators optimized model hyperparameters using the validation set and estimated standard deviations through 1000 bootstraps. The Mayo Simple Score was applied to the test set at its recommended threshold of 6.

Study Findings

The cohort comprised 34,468 men, with a mean age of 72.5 years, and 28,758 women, with a mean age of 70.9 years. The investigators identified ATTR-CM in 403 men and 96 women. When they tested new screening tools designed separately for men and women, the tools performed better than the traditional Mayo Simple Score at distinguishing patients with ATTR-CM from those without the condition.

The improvement was especially significant among women, with the new model achieving a discrimination score of 96.8 out of 100, compared with 80.3 for the Mayo Simple Score. The new models were also better at identifying patients who actually had ATTR-CM while reducing incorrect positive results, particularly among women.

Among 8 women with ATTR-CM missed by the Mayo Simple Score, the sex-specific model identified 5 (62.5%) at a threshold achieving ≥ 80% sensitivity and 7 (87.5%) at ≥ 90% sensitivity. Among 12 missed men, the corresponding numbers were 2 (16.7%) and 7 (58.3%). The leading model features for women were age, polyneuropathy, prealbumin, troponin, and NT-proBNP.

Clinical Implications

According to the investigators, sex-specific models using routinely available clinical characteristics may offer a practical approach to improving ATTR-CM detection at heart failure diagnosis. They proposed the models to improve sensitivity and discrimination in both women and men.

Expert Commentary

“This approach may serve as a practical screening tool to improve the diagnosis of ATTR-CM,” the researchers concluded.

Reference

Jeong H, Prasad P, O’Neill-Dee M, et al. Bridging the diagnostic gap in transthyretin cardiac amyloidosis (ATTR-CM): enhancing ATTR-CM detection at the time of heart failure diagnosis in women through sex-specific ATTR models (SS-ATTR-CMs). Abstract MDP2079. Presented at: American Heart Association 2026; November 9, 2026.