. 2020 Oct;143:105978. doi: 10.1016/j.envint.2020.105978

Table 5.

Assumptions, limitations and uncertainties of the ICF and Httk tools.

	IndusChemFate (ICF)	High-Throughput Toxicokinetics (Httk)
Problem formulation	How can PBK models be used to interpret HBM data in the context of assessing the risk of environmental pollutants?
Assumptions	V_max and K_m were searched in the literature and scaled up to same unit to input into the model. Absorption rate calculated using a QSAR model. Predictions of input parameters based on QSAR. Intake is assumed to come from the oral route, although it is known that it can also come from air, dust or food. Urinary excretion is driven by the lipophilicity. The QSPR calculating solubility assumes that human blood consists to 0.7% of lipids. 8% of arterial blood is turned into primary urine. No tubular resorption of very soluble substances with a log(K_ow) < −1.5. Physiological parameters do not change by sex or over time.	A simple PBK Model will represent the complexity of the human body. All Httk input parameters are correct, except the BW and the p value related to the intrinsic clearance of TCS which were changed (p = 0.06). Intake is assumed to come from the oral route, although it is known that it can also come from air, dust or food. Perfusion-limited kinetics. R_blood2plasma is constant throughout the body. Clearance is assumed to be relative to the amount unbound in whole blood instead of plasma, but converted to use with plasma concentration. Negligible blood volume fractions in all tissues to justify dividing by the tissue volume without a blood volume fraction and partition coefficient dependency in PBK tissue concentration equations. Wetmore data assumed f_ub = 0.005 for chemicals with f_ub below the limit of detection. Linear, non-saturated metabolism.
Input parameters	Sourced from different databases and the public literature which used a wide variety of techniques; Uncertainty in in vitro data (e.g. nominal as opposed to effective concentration is recorded); In vitro to in vivo extrapolation uncertainty (e.g. uncertainty related to the scaling factor used for V_max values); Only metabolism in liver was considered; Lack of detailed consideration of specific protein binding, interaction with intestinal flora, intestinal transport, and excretion by faeces. Prediction for individuals only; body weight can be amended; Lack of certain input parameter values (e.g. V_max and K_m).	Fewer variety of sources and higher consistency of methodology; Uncertainty in in vitro data (e.g. nominal as opposed to effective concentration is recorded); In vitro to in vivo extrapolation uncertainty; High degree of QSAR-generated parameters; Allows only for metabolism in liver; Prediction for populations which are based on U.S. NHANES data; degree of variability to European/Scandinavian population unclear.
Model structure	High number of compartments; Steady state not reached for most compounds; Mass balance can easily be checked but shows errors, especially in children's populations;	Moderate number of compartments; Simulation of metabolite kinetics unavailable; Conversion from chemical amount in renal tubule to urine concentration necessary;
Model output	Uncertainties underlying EDIs and TDIs; Uncertainties in urine concentration measurements, in particular related to the use of a one spot measurement of non-persistent compounds.	Uncertainties underlying EDIs and TDIs; Uncertainties in urine concentration measurements; Size of adult female population above 4.000 individuals gave error message.