Comparison of five tacrolimus measurement methods underscore the need for a standardized method to improve clinical outcomes

Abstract Tacrolimus is a widely used drug to prevent post-transplantation organ rejection by suppressing host immune response. While its efficacy is well established, tacrolimus is also known for narrow therapeutic index and debilitating adverse effects, rendering accurate measurement for therapeutic drug monitoring essential, to minimize drug toxicity and graft loss, and maximize organ rejection prevention. Tacrolimus can be quantitated by immunoassay (IA) and liquid chromatography-tandem mass spectrometry (LC-MS/MS). However, previous pairwise studies showed significant inter-assay variabilities between different IA methods, and between IA vs LC-MS/MS, which may be associated with the fact that IA measures parent drug and its metabolites, while LC-MS/MS measures only parent drug. No clear conclusion has been drawn about the optimal method. Transplant patients may also have their tacrolimus levels measured at different laboratories by different methods, causing inter-laboratory and intra-individual variabilities which may affect tacrolimus dose adjustment and monitoring and lead to adverse clinical outcomes. Here we performed a multi-platform comparison of tacrolimus quantification by two IA and 3 LC-MS/MS assays. A total of 216 specimens were analyzed for tacrolimus levels using chemiluminescence immunoassay (ChLIA, ARCHITECT, Abbott Diagnostic) and electrochemiluminescence immunoassay (ECLIA, Elecsys, Roche Diagnostic), and three LC-MS/MS assays (assigned names LC-MS/MS A (Agilent), Q (Waters), and W (Agilent)) in five different laboratories at four institutions. In one laboratory, the same analyses were performed on two different individual instruments (LC-MS/MS W1 and W2). Comparative results using LC-MS/MS A as reference were analyzed using Deming regression analysis, Pearson correlation, Bland Altman plots and bias calculation using GraphPad Prism version 10.1.1. While the correlations between each method with the LC-MS/MS A were good (r value: lowest 0.948 (LC-MS/MS Q), highest 0.988 (LC-MS/MS W2)), the ChLIA and ECLIA showed positive absolute bias of 2.1 and 1.2 ng/mL, respectively, and proportional bias of +26% and +17%, respectively. In comparison, the absolute bias of LC-MS/MS Q, W1 and W2 were -0.7, -0.02, and -0.1 ng/mL, respectively, with proportional bias of -11%, -0.2%, and -0.7%, respectively. The LC-MS/MS platforms showed tight agreement with one another, while IA methods were noticeably different, both from each other and from the LC-MS/MS, with a consistent positive bias, more pronounced with the ChLIA than the ECLIA. Here, our data-driven approach demonstrated that LC-MS/MS is superior to IA, and among IAs, ECLIA showed higher accuracy compared to ChLIA. However, inter-assay differences remain, emphasizing a need for tacrolimus assay standardization as well as prospective clinical trials to identify the most optimal method of tacrolimus measurement and improve patient outcomes.