Interpretation of HBM data

Human biomonitoring data can rarely be interpreted in isolation.   Apart from well-described examples, such as lead (Pb) in gasoline or methylmercury (MeHg) in seafood,  biomarkers cannot usually be linked directly to a single source or exposure pathway, so do not necessarily reveal how exposures have occurred.  Equally, while biomonitoring can provide invaluable information on current exposures and health effects, monitoring on its own does not provide a robust prediction of how exposures and effects may change under a potential future policy (or other) scenario.  For these and other reasons, biomonitoring invariably needs to be linked to other forms of monitoring and modelling when used in health impact assessments.

One of the most important opportunities for such an approach is in the analysis of the links between exposure, dose and health effect.  In recent years, physiologically based pharmacokinetic (PBPK) modelling has shown to be well-suited to calculate tissue doses of chemicals and their metabolites over a wide range of exposure conditions.  Because these PBPK models are based on the human physiology and anatomy and summarise the behaviour of chemicals in the body, they are often considered to be more realistic than empirical models.  PBPK modeling nevertheless faces a challenge in detecting and quantifying variation in source-to-dose relationships against the noise contributed by the wide range of other factors (e.g. variable exposures, activities, physiology, and pharmacokinetics) that operate within large (and often poorly characterised) human populations.  Linking HBM data with PBPK models can therefore help to address the limitations of both technologies.  

Linking PBPK modelling and HBM has led to a number of new conceptsd and tools for deriving health-based guidance values to support risk assessment and management.  Amongst these, three are of particular note: biomonitoring equivalents, reverse dosimetry and repeated measurements.

1. Biomonitoring equivalents

Biomonitoring Equivalents (BEs) are defined as the concentration of a chemical (or metabolite) in a biological medium consistent with defined exposure guidance values or toxicity criteria (including reference doses and reference concentrations, minimal risk levels or tolerable daily intakes).  These exposure guidance values are estimates of the daily exposure to a chemical that are believed to be without appreciable health risks, and can be used by regulatory agencies as guidelines for making risk management decisions. BEs thus use available pharmacokinetic data and forward dosimetry to calculate levels of biomarkers associated with these exposure guidance values (Figure 1).

Figure 1.  Determination of biomonitoring equivalents

Ideally, calculation of BEs is done using human pharmacokinetic data to relate external dose to biomarker concentrations (pathway 1 in Figure 1).  However, when only animal-based pharmacokinetic information is available (pathway 2), this may be applied, taking into account the appropriate uncertainty or safety factors. 

It needs to be stressed that BE values are screening values.  Comparison of measured biomonitoring levels to BE values can provide an initial, indicative evaluation of the need for follow up for risk assessment, and whether there is a need for additional studies on exposure pathways, potential health effects, other aspects affecting exposure or risk, or other risk management activities. 

The limitations of BE values need to be reognised.  They are not diagnostic criteria or 'bright lines' between safe and unsafe levels, so do not directly identify at-risk groups.  Nor can they be used to evaluate the likelihood of an adverse health effect in an individual or a population, so they are not a direct means of health impact assessment.  Exposure guidance values are set at levels that are designed to be health-protective for daily exposure for a full lifetime of exposure, while, depending on the chemical, biomonitoring data may be informative only about recent exposure levels.  An exceedance of the BE value in a single sample of blood or urine may or may not reflect continuing elevated exposure and does not imply that adverse health effects are likely to occur, but can serve as an indicator of relative priority for further risk assessment follow-up. 

2. Reverse dosimetry

Reverse dosimetry provides a means of source identification for the interpretation of biomarker values.  Establishing the relationship between biomarker data and environmental sources involves the reconstruction of past external exposure.  Ths has been termed 'exposure reconstruction' or 'reverse dosimetry'.  For this purpose, the PBPK model is reversed, and human biomonitoring data are back-transformed into equivalent exposure concentrations,  by using pharmacokinetic data in combination with information regarding the nature of the potential exposures to infer the exposures that are likely to have led to the measured biomonitoring results.

The fundamental problem underlying reverse dosimetry is to relate a measured internal dose, or tissue concentration to an unmeasured external exposure or dose, and beyond that a range of putative sources.  A major challenge in this approach is the need to take into account the inherent variability of the population from which the HBM data arises. As a consequence, the relationship between exposure and dose is such that an inverse association does not exist, is unstable, or is not unique.  Sources are also usually only partically characterised.  To deal with these uncertainties, analysis is usually done by combining a Bayesian analysis (e.g. using through Markov Chain Monte Carlo (MCMC) simulations) with a population model. 

3. Repeated measurements

Typically, large-scale human biomonitoring surveys gather biomarker measurements from single time points. These data are then used to make inferences about longer periods of toxicant intake, assuming that biomarker values are representative of steady-state conditions. However, steady-state conditions require stable biokinetics, a constant rate of exposure, and a dynamic equilibrium among different body tissues.

Figure 2 gives an indication of how a single sampling time may not be representative of steady state biomarker concentrations. At the specific sampling time (t = 10), the biomarker value can be the result of different exposure scenarios. The dashed line implies one high peak exposure episode, the full line a continuous fluctuation around a steady-state situation, and the dotted line a completely steady-state situation. Obviously, information on the biomarker pharmacokinetics is issue has a significant effect on both source identification and potential associations with health effects. 

Figure 2.  The potential effect of non-steady state conditions on the representivity of biomonitoring data

Assuming that biomarker values are representative for a steady-state concentration in the measured matrix is often not justified, and may require additional investigation.  By repeated sampling of individuals with particularly high and particularly low biomarker values (e.g. > P₉₅ and < P5), more insight in the pharmacokinetic behaviour of the biomarker can be obtained.  Collecting additional, contextual data via a  detailed questionnaire aimed at recording specific exposure scenarios during the time period between samplings can also provide more insight into possible sources.  The time-lag between sampling obviously is dependent on the half-life of chemicals, and the matrix that is used for biomarker determination.

Replication could also be achieved by measuring the same chemical in different matrices, or by using multiple biomarkers to describe the presence of a chemical in various matrices.  In these ways, different pharmacokinetic properties of the biomarker are reflected by multiple measurements.  For example, measuring cadmium in both blood and urine could give an indication of the long-term steady state exposure (urine) but also of the more recent, dynamic exposure (blood), thereby enhancing the ability to identify (and discriminate between) putative sources.  The availability of multiple biomarkers then leads to a trade-off in the design of studies: is it more efficient to sample one biomarker over multiple time points or more biomarkers at a single time point?  From a practical point of view, the latter may be preferred, as it is generally much easier and cost efficient to collect multiple biomarkers at a single sampling event in general population studies.  

More information can be found in the attached document 'Interpretation of HBM data', which can be downloaded below.