Photo Credit: GreenApple78
The following is a summary of “Machine-learning based subgroups of AL amyloidosis and cumulative incidence of mortality and end stage kidney disease,” published in the September 2024 issue of Hematology by Anand et al.
Immunoglobulin light chain (AL) amyloidosis is a complex disease with diverse treatments and outcomes, and current staging systems primarily focus on limited biomarkers related to cardiac, renal function, and plasma cell dyscrasia.
Researchers conducted a retrospective study improving prognostic accuracy for all-cause mortality and ESKD in patients with AL amyloidosis using unsupervised machine learning and comprehensive clinical and laboratory data.
They analyzed 2,067 patients diagnosed with biopsy-proven AL amyloidosis from the Boston University Amyloidosis Center. The dataset used in the study included 31 clinical symptoms and 28 baseline laboratory values.
The results showed the clustering algorithm identified 3 distinct subgroups within the patient population of AL amyloidosis, low-risk, intermediate-risk, and high-risk. The subgroups exhibited distinct clinical characteristics and differed significantly in the median OS times, estimated at 6.1, 3.7, and 1.2 years, respectively. Notable differences in the 10-year adjusted cumulative incidences of all-cause mortality among the subgroups: 66.8% (95% CI 63.4–70.1) for the low-risk group, 75.4% (95% CI 72.1–78.6) for the intermediate-risk group, and 90.6% (95% CI 87.4–93.3) for the high-risk group. Similarly, the 10-year adjusted cumulative incidences of ESKD were 20.4% (95% CI 6.1–24.5) for the low-risk group, 37.6% (95% CI 31.8–43.8) for the intermediate-risk group, and 6.7% (95% CI 2.8–11.3) for the high-risk group. Finally, a classifier was developed and trained for external validation. The classifier demonstrated high cross-validation accuracy at 85% (95% CI 83–86) utilizing a subset of readily available clinical parameters.
They identified a significant initial step towards integrating precision medicine principles into AL amyloidosis risk stratification, promising improved prediction of all-cause mortality and ESKD.
Source: onlinelibrary.wiley.com/doi/abs/10.1002/ajh.27472
