A multi-compartment physiologically based pharmacokinetic (PBPK) model coupled with Fourier-ARIMA (auto regressive integrated moving average) for estimation and prediction of polychlorinated biphenyls (PCB) concentration in tuna

Sadia Noon Oishe; Muhammad Zeeshan Jamil; Yoshiki Nishi

doi:10.1016/j.aquatox.2025.107461

A multi-compartment physiologically based pharmacokinetic (PBPK) model coupled with Fourier-ARIMA (auto regressive integrated moving average) for estimation and prediction of polychlorinated biphenyls (PCB) concentration in tuna

Aquat Toxicol. 2025 Jun 18:286:107461. doi: 10.1016/j.aquatox.2025.107461. Online ahead of print.

Authors

Sadia Noon Oishe¹, Muhammad Zeeshan Jamil¹, Yoshiki Nishi²

Affiliations

¹ Department of Mechanical Engineering, Materials Science, and Ocean Engineering, Graduate School of Engineering Science, Yokohama National University, 79-5 Tokiwadai, Hodogaya, Yokohama, Kanagawa 2408501, Japan.
² Division of Systems Research, Faculty of Engineering, Yokohama National University, 79-5 Tokiwadai, Hodogaya, Yokohama, Kanagawa 2408501, Japan. Electronic address: nishi-yoshiki-rg@ynu.ac.jp.

PMID: 40577972
DOI: 10.1016/j.aquatox.2025.107461

Abstract

Polychlorinated biphenyls (PCBs) are persistent environmental contaminants that accumulate in aquatic organisms, posing health risks to humans through the consumption of fish and other aquatic animals. As a top predator, tuna is particularly susceptible to PCB bioaccumulation due to its position in the food chain, highlighting the importance of accurate contamination prediction. This study introduces a novel hybrid methodology that integrates a six-compartment PBPK-inspired model with a Fourier-ARIMA statistical framework to estimate and forecast PCB concentrations in tuna tissues, including blood, muscle, gill, intestine, kidney, and liver, within one year. The PBPK-inspired model simulates physiological bioaccumulation mechanisms, while the Fourier-ARIMA approach improves prediction accuracy by capturing both long-term periodic trends (Fourier) and short-term variations (ARIMA). Simulated concentrations were compared against published experimental data, and a cumulative distribution function (CDF)-based risk assessment was conducted, benchmarked against regulatory thresholds established by the U.S. EPA (47 and 210 ng/g) and WHO (20 ng/g). The model estimated the highest PCB concentrations in the liver (750 ng/g) and kidney (270 ng/g), with muscle tissue showing lower levels (17 ng/g). Probabilistic analysis revealed the liver to be the primary site of PCB accumulation, with approximately 85 % of modeled concentrations exceeding the EPA high-risk benchmark (210 ng/g). In contrast, muscle tissue, the main edible portion, exhibited uniformly low levels, remaining below the WHO threshold of 20 ng/g throughout, suggesting limited dietary exposure risk. This hybrid modeling approach supports dynamic, tissue-level trend prediction and probabilistic risk evaluation, making it a valuable tool for assessing and managing PCB contamination in marine ecosystems.

Keywords: CDF; Fourier-ARIMA; PBPK-inspired model; PCBs; Periodic modeling; Risk assessment.