Development o f physiologically based pharm acokinetic models fo r
the bioaccum ulation o f persistent organic po llutan ts in marine
mammals
Weijs Liesbeth
Systemic Physiological and Ecotoxicological Research (SPHERE), D epartm ent o f Biology, U niversity o f
Antw erp, G roenenborgerlaan 171, 2020 Antw erpen, Belgium
E-mail: liesbeth.weiis@ uantwerpen.be
Factors, such as a greater demand fo r products and more sophisticated industrial techniques, have
caused a grow ing num ber o f chemicals in the environm ent. Due to a lack o f e ffic ie n t m etabolic
breakdown or e lim in a tion processes in organism s and in the environm ent, these chemicals can be
passed on in the aquatic and te rre stria l food webs leading to higher levels in the predator compared
to its prey. Marine mammals are apex predators in marine ecosystems and as such, th e y accum ulate
considerable am ounts o f chemicals th ro u g h th e ir diet. Evidence in the literature has shown th a t
these am ounts can have a m ajor negative im pact on th e ir im m une, endocrine and reproductive
system or even on th e ir survival in general. Marine mammals do not seem to have the appropriate
enzyme systems to be able to deal w ith chemicals. They experience the negative effects o f p ollution
them selves and at the same tim e, pass the chemicals on to th e ir o ffsp ring . As a consequence, levels
o f contam inants in these animals decrease o nly slow ly even th ough the industrial production o f
several chemicals has been restricted or banned in the past.
In order to prevent repetitions o f past situations, it is im p o rta n t to gain knowledge about the
absorption, d is trib u tio n and e lim in a tion o f known or ‘o ld ’ chemicals in marine mammals.
Understanding the kinetics and effects o f old chem icals can be useful to assess the im pact o f new
chemicals w ith com parable properties com pared to the old ones before these new chemicals are
being m anufactured. From th is p oint o f view, studying known, old chem icals is u ndoubtedly useful
fo r the risk assessment o f new com pounds.
In th is thesis, the kinetics o f known or tra d itio n a l chemicals such as PCBs and PBDEs, was
investigated in h arbour porpoises and long-finned p ilo t whales by using physiologically based
pharm acokinetic (PBPK) models. These com puterized m odels com bine physiological inform a tion o f
the organism o f interest and chem ical properties o f the chemical o f interest to reflect the kinetics o f
th a t com pound in the body o f the organism . Sim ilar to exposure experim ents in which all factors
are controlled to m inim ize the large degree o f va riability, m odels fo r w ild populations are also more
reliable if the datasets used fo r evaluation o f the m odels are som ewhat ‘ ro b u s t’ or ‘ u n ifo rm ’. For
practical reasons, trends are easier to visualize and param eters are easier to estim ate if the num ber
o f interfering , external factors are reduced to a m inim um . For theoretical reasons, scattered data
can lead to param eter estim ates th a t are not reflecting the intrinsic, physiological capabilities o f the
species.
Consequently, it was im p o rta n t to investigate firs t which factors were influencing the levels o f
pollutants in the blood since blood is the o nly circulation m edium in the models. Blood samples o f
harbour seals and porpoises o f d iffe re n t health condition, o rig in (captivity versus w ild), gender and
age were analysed. Results showed th a t o nly emaciated animals had deviating concentrations and
profiles o f PCBs and PBDEs in th e ir blood com pared to animals th a t were not emaciated. O f course,
starvation can occur in w ild populations, but is d e fin ite ly not com m on fo r all w ild animals. The
conclusion here was thus th a t datasets o f blood could be used in the bioaccum ulation models.
However, blood is never sampled at the same tim e as tissues in marine mammals. In m ulticom partm ental models as the ones developed in th is w ork, data o f more tissues was preferred to
evaluate the m odel predictions fo r as much com partm ents as possible sim ultaneously.
Because the Black Sea harbour porpoise dataset was both restricted in tim e (animals were from
1997-1998) and space (animals from the Black Sea do not leave the Black Sea area), these results
were preferably used to evaluate the very firs t prelim inary harbour porpoise m odel predictions.
These m odels were developed to explain the bioaccum ulation o f several PCB congeners (PCB 153,
PCB 180, PCB 101, PCB 149, PCB 118, PCB 99, PCB 170) and PBDE congeners (PBDE 47, PBDE 99,
PBDE 100 and PBDE 153) in male harbour porpoises. Model outp uts showed th a t levels o f all PCBs
and PBDEs reached high levels at the end o f lactation period (e.g. firs t year o f life) a fte r which the
grow th d ilu tio n effect and a change in d iet caused a decline in concentrations follow ed by an
increase in concentrations fo r the rest o f the lives o f the Black Sea harbour porpoises. The models
-
1 77
-
were then applied to assess tem poral trends by using the dataset o f harbour porpoises fro m the
North Sea. D uring th is m odelling exercise, levels o f PCBs and PBDEs were found to decrease from
1990 u ntil 2008, although not at the same rate fo r all PCB and PBDE congeners. For some PCB
congeners, the PBPK models were also used to test the m etabolic bio tran sfo rm atio n capacity fo r PCB
11 8, PCB 149 and PCB 101. Results suggested a fa irly weak m etabolic breakdown o f PCB 118 and
an enhanced capacity fo r m etabolic breakdown o f PCB 101 w ith higher age. In contrast, results were
inconclusive about the m etabolic capacities fo r PCB 149. So far, all a ttem pts to estim ate parameters
were perform ed m anually and the se nsitivity o f the param eters on the m odel o u tp u t was tested by a
‘one-at-a-tim e’ o r local se nsitivity analysis. Flowever, th is type o f sensitivity analysis ignores
potential correlations between the parameters. Flence, more sta tistically sound param eter
estim ation m ethods and global sensitivity tests th a t take into account potential interactions
between the param eters were needed in order to im prove the robustness o f the models.
A pplying new m ethods fo r param eter estim ation and sensitivity analyses was, thus, the next step.
So, in the m ost recent PBPK m odel fo r bioaccum ulation o f pesticides (p ,p ’-DDT, p ,p ’-DDE, p ,p ’-DDD)
in h arbour porpoises, param eters were estim ated using Bayes’ theorem executed w ith M arkov chain
Monte Carlo (MCMC) sim ulations. In addition, the influence o f changes in param eter values on the
m odel o u tp u t was tested using global sensitivity analyses. Com pared to all previous PBPK models,
th is m odel fo r bioaccum ulation o f pesticides differed not o nly in the statistical techniques, but also
in its com plexity. Whereas all previous models showed the kinetics o f a single com pound, the
pesticide model showed the kinetics o f p ,p ’-DDT and its tw o m etabolites p ,p ’-DDE and p ,p ’-DDD at
the same tim e ensuring a high co nn ectivity between the kinetics o f these three com pounds. Similar
to the previous harbour porpoise models, the structural model was firs t evaluated using a dataset o f
harbour porpoises from the Black Sea a fte r which the param eter range estim ates were fu rth e r
o ptim ize d using the dataset o f harbour porpoises fro m the North Sea.
The same techniques (Bayesian PBPK m odelling and MCMC sim ulations) were also used fo r a PBPK
model fo r the lifetim e bioaccum ulation o f PCB 1 53 in long-finned p ilo t whales. For th is species, tw o
datasets were available from tw o mass stranding events. Long-finned p ilo t whales have tig h t fa m ily
group bonds so whenever an individual ends up on the beach, all o th e r m em bers o f the group
follow . For the animals, these mass strandings are traum atic, but th ey are a great o p p o rtu n ity fo r
m on itoring and m odelling as a dataset cannot possibly be more ‘ hom ogeneous’. In contrast to the
dataset o f the Black Sea harbour porpoises, o nly blubber samples were available fo r the long-finned
p ilo t whales. The PBPK m odels fo r the bioaccum ulation o f PCB 153 in p ilo t whales are therefore
sm aller than the models in harbour porpoises. Nevertheless, param eters were estim ated w ith the
m ost suitable technique fo r th is type o f models, m aking the p ilo t whale m odel already a useful
fram e w o rk fo r evaluating sim ilar o r more elaborate datasets in the future.
This w ork provides new ideas and innovative approaches to study b io m o n itorin g data. The
bioaccum ulation m odels developed here can already be used as a fram e w o rk to compare to new
datasets, but can also be fu rth e r o ptim ize d and expanded in the fu tu re . These results are,
therefore, not o nly a useful addition to e xistin g knowledge, but provide also new perspectives to
assess p ollutio n and its effects in marine mammals. Such an integrated approach is required to set
up guidelines fo r conservation o f a species. As a result, the m odels developed in th is w ork are
undoubtedly useful to ols fo r risk assessment purposes.
-
178
-