The present study demonstrates, for the first time, procoagulant effects at the circulatory and tissue level in a ferret influenza model, largely proportional to the severity of influenza virus infection. These findings are in line with earlier epidemiological, clinical, animal and in vitro data [6, 8, 13–15, 20, 22–24]. Ferrets have been shown to be an adequate model to study the coagulation cascade [25–27] with PT and APTT normal values varying from 11.6-12.7 and 18.9-22.3 seconds respectively. This is comparable to our 104 pre-inoculation ferret samples (PT 11.7 (+/- 0.1) and APTT 19.8 (+/- 2.2)) .
Like in humans, highly pathogenic avian influenza virus infection causes severe disease in ferrets, which may include bleeding complications and multi-organ failure [28, 29]. In our experiments, HPAI-H5N1 virus inoculated ferrets showed severe disease, which in some cases resulted in spontaneous death. Analysis of the coagulation cascade in these animals confirmed the severity of infection with prolongation of global coagulation assays and signs of activated endothelium. PT and APTT values increased already from 2 dpi onward with individual ferrets showing an increase up to 20 seconds. This observation is suggestive for consumptive coagulopathy which is strengthened by the high levels of fibrin deposition in the lung capillaries. Consumptive coagulopathy could be the result of extreme activation of coagulation, for instance due to increased tissue factor production as is seen in other (severe) viral diseases as Ebola hemorrhagic fever . The exact role for consumptive coagulopathy in highly pathogenic H5N1 infection warrants further research, but hypothetically the excess of coagulation activity could lead to microthrombosis in the pulmonary alveoli leading to respiratory distress or even multi organ failure . The procoagulant changes were seen both at the tissue level and in the circulation, suggested by the TAT increase. The statistically significant increase in D-dimer levels confirms this procoagulant state. However, D-dimer levels were lower in HPAI-H5N1 virus inoculated ferrets compared to ferrets infected with H3N2 virus and especially compared to the ferrets infected with pH1N1 virus. A possible explanation for this phenomenon could be the inhibition of fibrinolysis by high levels of plasminogen-activator type 1 activity (PAI-1) during H5N1 virus infection. Unfortunately we could not test PAI-1 activity in ferret plasma with the currently available human PAI-1 activity assays. Since plasminogen is proven to play an important role in influenza pathogenesis further exploring the biology, activation and inhibition of plasminogen in influenza infection would be of great interest .
The second virus we used in our experiments was pH1N1. Although less severe compared to HPAI-H5N1 virus infected ferrets, pH1N1 virus infection caused severe pneumonia with lung damage in ferrets. While ferrets infected with pH1N1 virus showed remarkably high levels of D-dimer, tissue fibrin deposition was not as prominent as seen in HPAI-H5N1 virus infected ferrets. Activated coagulation in other organs than the respiratory tract or a systemic activation of coagulation could explain this phenomenon. These severe procoagulant changes in the circulation could be the result of a specific immune activation during pH1N1 virus infection. A possible explanation can be found in the work of Monsalvo et al. who showed an excessive amount of pathogenic immune complexes, which are known to have systemic procoagulant effects, in fatal pH1N1 cases [31, 32]. Furthermore, TAT levels significantly increased in the first 4 days after infection and at 4 dpi there was a remarkable prolongation of PT and APTT values up to 4 seconds. The very ‘sudden’ increase of clotting times at 4 dpi is suggestive for a consumptive coagulopathy, possibly similar to what was seen in DIC due to HPAI-H5N1 virus infection and bacterial sepsis . Clotting times had normalized at 7dpi, however, indicating that in contrast to bacterial sepsis, the consumptive coagulopathy is transient and less severe. The ‘sudden’ onset of clotting time prolongation may be of interest to evaluate specific coagulation factor changes during influenza infection.
To evaluate the influence of a more ‘moderate’ influenza virus infection, seasonal H3N2 virus was also included in the experiments. Although this influenza virus in general causes ‘moderate’ disease in humans and ferrets, it did cause significant procoagulant changes in the model with hemostatic alteration comparable to those of pH1N1 virus infected ferrets. However, TAT levels did not increase suggesting a more moderate procoagulant state compared to H1N1- and H5N1 virus infected animals.
Since the ageing human population is prone to both an increase in cardiovascular disease and to complications during and after infection with seasonal and avian influenza viruses [34, 35], further exploration of the interplay between influenza and hemostasis would be of great interest. Most of the associations found in Table 2 show positive correlations between coagulation parameters and markers of inflammation (body weight decrease and relative lung weight increase). This comes as no surprise since the bidirectional cross-talk between coagulation and inflammation has been studied very well, whereby inflammation in general evokes a procoagulant response [36–38]. The specific disturbances in the tightly regulated balance between clotting, anti-coagulation and inflammation could be a target for novel intervention strategies in influenza. Following our observational study, an intervention model could further evaluate the role of coagulation in influenza virus pathogenesis and the potential processes for targeted intervention, for example by targeting protease receptor type-2 (PAR-2) activation in influenza pathogenesis. PAR-2 is an important receptor in both inflammation and coagulation, and recently described to have a major role in the damage seen after the inflammatory response during influenza virus infection [39, 40]. While statins may also be interesting candidates for future studies. Statins may counteract specific inflammatory responses such as seen after acute coronary syndrome, and thereby may decrease mortality when given to influenza patients. Studying the influence of statin treatment on the procoagulant changes during influenza virus infection and the role these changes have in the postulated increased risk of myocardial infarction would be of great interest [41–43].
Collectively the data generated by our study will pave the way for further exploration of novel treatment and intervention strategies for influenza and its complications. Furthermore, based on the correlation between the viral infection - and coagulation parameters in this experiment, coagulation tests could serve as valuable biomarkers predicting disease severity. The ferret model likely offers the best opportunity to explore these options in a preclinical setting optionally also linked to host genetics since ferrets represent an outbred population.