Analytical review of nanoplastic bioaccumulation data and a unified toxicokinetic model: from teleosts to human brain

Abstract

Nanoplastics (NPs) are increasingly detected in human blood and organs at concentrations reaching hundreds to thousands of parts per million, yet no quantitative framework has linked short-term experimental uptake kinetics to long-term, organ-specific accumulation. Here we analytically review the most reliable uptake and depuration datasets available in teleost fish using a sequential two-compartment toxicokinetic model that distinguishes systemic circulation from tissue-level retention. While anomalous, non-Markovian transport is expected at microscopic scales, we show -- through an explicit theoretic analysis on minimal information -- that such formulations are not identifiable with existing data. Allowing unresolved early-time dynamics to be absorbed into effective, non-zero initial conditions yields an emergent Markovian description that is maximally informative and consistent across species, organs, particle sizes, and exposure levels. When expressed in normalized variables, uptake dynamics collapse onto a universal trajectory governed by a single dimensionless parameter, the systemic excretion capacity, which is generically small under experimental conditions. The resulting scale-free framework reveals systematic power-law dependencies of enrichment and retention times on ambient concentration, particle size, and body mass. Exploiting this structure, we examine the consistency of extrapolations to humans and show that reported organ burdens -- particularly in the brain -- are quantitatively compatible with inefficient systemic clearance and strong lipid-driven partitioning. At steady state, human tissue concentrations follow a robust approximate cubic scaling with lipid fraction, identifying lipid content as the dominant and mechanistically interpretable determinant of chronic nanoplastic accumulation.

Figures (7)

• Figure 1: Schematic representation of the sequential two-level toxicokinetic architecture (S2CT model): the systemic compartment (circulatory system, organ-agnostic) linking the external concentration $C_w$ to the homogenized systemic concentration $C_S$ via the three entry gates (respiratory, oral and skin) and rapid circulatory mixing. Organ-level tissue compartment (organs) in which MNPs diffuse (bottom right panel) from the vascular lumen across the endothelial glycocalyx (eGC), endothelium, and basement membrane into the parenchymal domain of radius $R_1$ (e.g. $R_1 \approx 15~\mu$m) surrounding a capillary of radius $R_0$ (e.g. $R_0 \approx 3~\mu$m). Short-term kinetics are dominated by the eGC, while long-term accumulation reflects organ-specific physicochemical composition.
• Figure 2: (a) Raw data (up to five organs) from the six studies selected habumugisha2023habumugisha2025Zhang2025ZHENG2024Ding2018Choi2023 on NP bioaccumulation process in teleosts (400 measurements). Color codes for each organ are indicated. The opacity of colors indicate: (1) Circles, opacity proportional to ambient concentrations $C_w={5,10,\text{ and }15}$ ppm (study of 2023 habumugisha2023, Danio rerio); (2) Triangles, opacity proportional to particle size (study of 2025 habumugisha2025Danio rerio), ambient concentration 1 ppm; (3) Squares: the selected data from Zhang2025 ( Danio rerio, intestine and gill), ambient concentration 0.45 ppm, $\sim$525 nm PS NPs, with and without protein coronas (ovoalbumin and lysozyme); (4) Pentagons: data (four organs) from Ding2018 ( Oreochromis niloticus), opacity proportional to logarithm of water concentrations of NPs $C_w={0.001, 0.01, \text{ and }0.1}$ ppm. (5) Hexagons: data (four organs) from ZHENG2024 ( Oryzia melastigma) excluding eye and skin for consistency with the rest of data, PS NPs of 120 nm (not their 2 $\mu$m data), and two water environments (fresh and synthetic seawater). (6) Tilted triangles: data (four organs) from Choi2023 ( Zacco platybus), PS NPs of 50 nm, two water concentrations (0.01 and 0.1 ppm). The logarithmic abscissa emphasize the initial short-term dependencies. (b) Universal best fitting of the six uptake data series considered. Black and blue dashed lines are function $y_{Tu}$ given by (\ref{['eq:canonical_T']}) for $\eta=0$ (asymptotic form) and $0.33$, respectively. (c) Universal best fitting of the four depuration data series available from the considered studies on NP bioaccumulation processes habumugisha2023habumugisha2025ZHENG2024Zhang2025. Black dashed line is the identity.
• Figure 3: Power laws found for coefficients $k_p$ and characteristic times $t_c$, as functions of the variables $\varphi=C_w^\phi d^\zeta W^\omega$ with $\{\phi,\zeta,\omega\}$ powers according to Table \ref{['tab:uptake']}. Numerical prefactors are consistent with units of length in microns, mass in grams, and time in days. For the human cases, two representative values of average intake concentrations $C_w$ and for four average values of the particle size $\langle d \rangle=\{20,50,100,200\}$ nm (see Suppl. Info.) are considered.
• Figure 4: The organ-specific concentrations of micro- and nanoplastics, $C_T$ (wet weight basis), as a function of the lipid fraction $f_{\mathrm{lipid}}$. Data compiled from Nihart et al. nihart2025, Liu et al. Liu2024, and Hu et al. Hu2024. Given the relatively small variation of $C_T$ values in the long time scale, compatible with the increment in the concentration of intakes $C_w$ along the years, one can assume $C_T \simeq C_{T\infty}$, the steady state values. The dashed line shows the best-fit power law $C_{T\infty} \propto f_{\mathrm{lipid}}^{3}$, indicating a strong cubic dependence of bioaccumulation on tissue lipid content. Error bars denote the reported experimental uncertainty in both coordinates.
• Figure S1: The normalized kernels $y_T(x)$ and $y_S(x)$ (solid and dashed lines, respectively) for representative small values of $\eta$. (a) Uptake functions $y_{(S,T)u}$. The asymptotic solution (\ref{['eq:asympto_u']}) in the limit $\eta\to 0$ is shown as a black dashed line. For $\eta=1/3$, the asymptotic values of $y_{Tu}$ and $y_{Su}$ coincide. (b) & (c) Depuration functions $y_{(S,T)d}$ for $\xi_d=2$ (b) and $\xi_d=-2$ (c). In this case, $y_{Sd}$ and $y_{Td}$ coincide identically for all $x>0$ when $\eta=1$.