Model for Predicting Day-100 Stem Cell Transplant-Related Mortality in AL Amyloidosis

Autologous stem cell transplantation (ASCT) in AL amyloidosis effectively produces a high overall hematological response rate, complete response rate, and durable remissions. However, it is a high-dose chemotherapy-based therapy with high morbidity which can be offered only to a subset of AL amyloidosis patients who are considered fit for this therapy. Transplant-related mortality (TRM) in AL amyloidosis has decreased over the past three decades due to improvement in patient selection. New therapeutic options question the utilization of ASCT. A model to predict TRM can improve the decision-making process, patient selection and further reduce TRM.

AL amyloidosis patients (n=1718) from 9 centers globally, transplanted in the years 2003-2020 were included. Pre-ASCT variables of interest were assessed for association with Day-100 all-cause mortality, including age at ASCT (≥ 60 years), sex, light chain isotype, ECOG performance status (ECOG PS; 0, 1 and 2/3), the difference between involved and uninvolved light chains (dFLC; ≥180 mg/L), N-terminal prohormone of B natriuretic peptide (NT-proBNP; ≥1800 pg/mL) or B-type natriuretic peptide (BNP; ≥400 pg/mL), left ventricle ejection fraction (LVEF, <50%), interventricular septum (>16 mm), corrected diffusing capacity for carbon monoxide (cDLCO; ≤65% of predicted), forced expiratory volume in one second (FEV₁; ≤65% of predicted), serum albumin (<2.5 g/dL), estimated glomerular filtration rate (eGFR; <30 mL/min/m²), alkaline phosphatase (>X2 upper limit of normal), and systolic blood pressure (<100 mmHg). Troponins could not be included in our modeling given the multitude of assays used across institutions and periods that could not be reconciled. A random forest (RF) classifier with 10-fold cross-validation was used to assist in variable selection. The final model variables were selected based on statistical performance and clinical relevance. The model was fitted using logistic regression.

The median age at ASCT was 58, and 40% of patients were women. The most frequently involved organs were the kidneys (69%) and the heart (52%). Mayo 2004 stages I and II were collectively present in 83% of patients. Day-100 TRM occurred in 75 patients (4.4%) with the 3 main causes being shock (n=27), cardiac arrhythmia (n=26), and organ failure (n=22). Age at ASCT, sex, and light chain isotype were not predictors of Day-100 TRM. Across the other pre-ASCT variables, adverse features were present from 6.9% (LVEF <50%) to 25.8% (dFLC ≥180 mg/L) of patients. Ten factors were associated with Day-100 TRM on univariate analysis. The odds ratio of these variables ranged from 1.6 to 10.2. RF classifier using all these variables identified a model with an area under the curve (AUC) of 0.72±0.12. Using the importance hierarchy function to refine the model selection, a 4-variable model [NT-proBNP/BNP ≥1800/400 pg/mL, serum albumin <2.5 g/dL, ECOG PS 1 or 2/3, and systolic blood pressure <100 mmHg] was built with an AUC of 0.70±0.12. Based on logistic regression coefficients, ECOG PS 2/3 was assigned two points, while other adverse predictors were assigned 1-point each (model score range 0-5). We then applied the model scoring to our study population. Score distribution was as follows: 0 (n=431, 25.1% of patients), 1 (n=688,40.1%), 2 (n=413, 24.0%), and 3 and above (n=186, 10.8%). Day-100 TRM rates were 0.46%, 3.2%, 5.8%, and 14.5% for 0, 1, 2, and ≥3 points, respectively. The respective rates for the years 2003-2011 were 1.1%, 3.3%, 6.4%, and 19.2%, while for the years 2012-2020 0%, 3.1%, 5.2%, and 8.4%.

We developed a simple clinical tool to predict the Day-100 TRM risk in AL patients undergoing ASCT. This tool will allow more informed decision making weighing the individual benefits and risks of ASCT compared to other therapies.