As a treatment for relapsed or refractory multiple myeloma (MM), carfilzomib has been associated with a significant risk of cardiovascular adverse events (CVAE). The goals of our study were to evaluate the metabolomic profile of MM patients to identify those at high risk prior to carfilzomib treatment and to explore the mechanisms of carfilzomib‐CVAE to inform potential strategies to protect patients from this cardiotoxicity. Global metabolomic profiling was performed on the baseline and post‐baseline plasma samples of 60 MM patients treated with carfilzomib‐based therapy, including 31 who experienced CVAE, in a prospective cohort study. Baseline metabolites and post‐baseline/baseline metabolite ratios that differ between the CVAE and no‐CVAE patients were identified using unadjusted and adjusted methods. A baseline metabolomic risk score was created to stratify patients. We observed a lower abundance of tauroursodeoxycholic acid (T‐UDCA) in CVAE patients at baseline (odds ratio [OR] = 0.47, 95% confidence interval [CI] = 0.21–0.94, p = 0.044) compared with the no‐CVAE patients. A metabolite risk score was able to stratify patients into three risk groups. The area under the receiver‐operating curve of the model with clinical predictors and metabolite risk score was 0.93. Glycochenodeoxycholic acid (OR = 0.56, 95% CI = 0.31–0.87, p = 0.023) was significantly lower in post‐baseline/baseline ratios of CVAE patients compared with no‐CVAE patients. Following metabolomic analysis, we created a baseline metabolite risk score that can stratify MM patients into different risk groups. The result also provided intriguing clues about the mechanism of carfilzomib‐CVAE and potential cardioprotective strategies.
1 February 2026 in Clinical Pharmacology & Therapeutics
Carfilzomib is highly effective in the treatment of multiple myeloma, but it has been associated with cardiovascular adverse events that impact patient outcomes. Our prior global metabolomic analyses indicated an association between hydrophilic bile acids and carfilzomib‐cardiotoxicity risk, although a causal relationship remained to be determined. Here, our objective was to validate the previously identified bile acids in an independent cohort and investigate whether these hydrophilic bile acids play a causal role in carfilzomib‐cardiotoxicity using Mendelian randomization. Using targeted metabolomics, we validated the association between glycoursodeoxycholic acid and carfilzomib‐cardiotoxicity in an independent cohort ( n = 61). We then performed two‐sample Mendelian randomization analyses, using metabolome‐wide association study results to provide the genetic instruments for bile acid levels as exposure and genome‐wide association study summary statistics from the UK Biobank ( n = 484,598) as the outcome data for cardiovascular adverse events. Causal inference was assessed with the inverse‐variance weighted method, followed by multiple sensitivity analyses. Higher glycoursodeoxycholic acid concentration (odds ratio = 0.34, P = 0.032) was associated with lower cardiotoxicity risk after adjusting for hypertension and high levels of brain natriuretic peptides. Mendelian randomization analysis demonstrated a robust causal effect of glycoursodeoxycholic acid on cardiotoxicity risk (β = −0.00065, P = 6.2 × 10 −5 ). Gene enrichment analysis indicated pathways involving potassium channel regulation and thromboxane signaling to be implicated. This integrative metabolomic and genetic investigation supports a potential protective role of glycoursodeoxycholic acid in cardiovascular vulnerability and motivates larger, carfilzomib‐specific studies to evaluate its utility for risk stratification.