Factors driving the circulation and possible expansion of Crimean–Congo haemorrhagic fever virus in the western Palearctic

Authors

  • A. Estrada-Peña,

    1. Department of Parasitology, Faculty of Veterinary Medicine, University of Zaragoza, Zaragoza, Spain
    Search for more papers by this author
  • F. Ruiz-Fons,

    1. Instituto de Investigación en Recursos Cinegéticos IREC (CSIC-UCLM-JCCM), Ciudad Real, Spain
    Search for more papers by this author
  • P. Acevedo,

    1. Centre de Recerca en Sanitat Animal (CReSA), UAB-IRTA, Campus de la Universitat Autònoma de Barcelona, Bellaterra, Spain
    Search for more papers by this author
  • C. Gortazar,

    1. Instituto de Investigación en Recursos Cinegéticos IREC (CSIC-UCLM-JCCM), Ciudad Real, Spain
    Search for more papers by this author
  • J. de la Fuente

    Corresponding author
    1. Instituto de Investigación en Recursos Cinegéticos IREC (CSIC-UCLM-JCCM), Ciudad Real, Spain
    2. Department of Veterinary Pathobiology, Center for Veterinary Health Sciences, Oklahoma State University, Stillwater, OK,, USA
    • Department of Parasitology, Faculty of Veterinary Medicine, University of Zaragoza, Zaragoza, Spain
    Search for more papers by this author

Correspondence

José de la Fuente, Instituto de Investigación en Recursos Cinegéticos, Ronda de Toledo s/n, 13071 Ciudad Real, Spain. E-mail: jose_delafuente@yahoo.com

Abstract

Aims

To produce a spatial risk map regarding spread of Crimean-Congo haemorrhagic fever virus (CCHFV) in the western Palearctic by linking a process-driven model of the main tick vector, Hyalomma marginatum, to a Next Generation Matrix.

Methods and Results

Process-driven model was composed of deterministic equations that simulate developmental and mortality rates of different tick stages by using temperature and atmospheric water vapour data. The model used climate data at 10-day intervals at a spatial resolution of 10 min over western Palearctic. The model estimated the basic reproduction number, R0, for CCHFV transmission by H. marginatum ticks and evaluated how changes in temperature and biological parameters may alter the geographical range of CCHFV. In particular, variation in the rate of transovarial transmission of CCHFV in the tick produced the greatest change in CCHFV circulation in the tick population. Parameters affecting the rates of tick bite, non-systemic transmission and efficiency of tick-to-tick transmission had little effect on R0. Temperature changes that affect tick development, survival and activity rates increased the suitable area for CCHFV transmission at higher latitudes in the western Palearctic.

Conclusions

Non-systemic transmission had little impact on virus transmission under all scenarios. In the area studied, increase of temperature has no impact on the routes of transmission of CCHFV. However, climate conditions favouring tick survival, which increase infected adult tick populations, together with large numbers of hosts for adults were predicted as the most likely scenario for the spread of the virus in the studied area.

Significance and Impact of the Study

The proposed framework is able to capture the dynamics and the relative contribution of the different routes (hosts, ticks) in the transmission and spread of an important pathogen affecting human health. The high contribution of the transovarial transmission route makes the process highly dependent upon suitable hosts for adult ticks, like large domestic and wild ungulates. Climate seems to have a very reduced effect on such spread.

Introduction

Crimean–Congo haemorrhagic fever (CCHF) is a serious human disease mainly transmitted by ticks of the genus Hyalomma. Interest in the disease increased after the epidemic in Turkey and new viral records reported in areas near Turkey such as the Balkans and Russia (Ergonul and Whitehouse 2007). Studies have focused on outlining the probable routes for virus introduction into Western Europe from the original foci of the disease in Eastern Europe and Turkey (EFSA 2010; Gale et al. 2010; Estrada-Peña et al. 2012a). The finding of CCHF virus (CCHFV) in Western Europe (Estrada-Peña et al. 2012b) demonstrated that the virus is not restricted to Eastern Europe. A viral strain circulating in south-western Mediterranean is a reason of concern about the spread of the virus into northern latitudes (Gale et al. 2011).

The tick vector has three active stages. The immature (larvae and nymphs) commonly feed on the same hosts, which are many species of small mammals and birds. Large ungulates serve as hosts for the adults. Adult female ticks contribute to the infection by transovarial transmission (vertical transmission) of the virus to the eggs (Shepherd et al. 1989). Feeding on infected reservoir hosts or through the nonsystemic transmission of the virus might also infect ticks. Nonsystemic transmission may occur when uninfected ticks feed on the same vertebrate host near to an infected tick, which passes the virus to the uninfected tick with the saliva without host systemic infection (Gordon et al. 1993). Studies on other tick-transmitted viruses such as tick-borne encephalitis virus (TBEV) demonstrated that subtle variations in field seasonal patterns of the tick vector profoundly affect the dynamics of active foci (Randolph et al. 2000). While a large social component has been identified behind the upsurge of tick-borne encephalitis (Šumilo et al. 2008), concerns exist about the spread of the virus of TBEV into areas that recently became suitable for the tick vector because of climate trends.

The spread of pathogens is typically characterized by the basic reproduction number, R0. This quantity is fundamental to the theory and management of infectious diseases. R0 is defined as the expected number of secondary cases produced by a single infected animal in a wholly susceptible population when a homogeneous and well-mixed host population exists. In the model here, R0 represents the number of infected Hyalomma ticks arising from introduction of a single infected tick into a population of susceptible ticks. If R0 is ≥1, the virus can persist and circulate in the tick population, otherwise it does not establish. To work with pathogens that infect multiple hosts, the next-generation matrix (NGM) was introduced (Diekmann et al. 1990). The number of epidemiologically distinct infected and infectious host types determines the dimension of NGM, whose entries are host-type specific R0 (Diekmann et al. 1990). A generalist NGM was developed to illustrate the basic features of transmission rates for some tick-borne pathogens (Matser et al. 2009).

Tick species distribution modelling alone is not valid to infer circulation of ticks-transmitted pathogens when the tick-borne pathogens depend on subtle changes in development, mortality and activity rates of the tick vectors. A dynamic study is necessary for adequate awareness of the endemic potential of CCHFV in Europe. It should explicitly consider tick seasonality, development and mortality rates at discrete time intervals and link these tick processes with the parameters governing the circulation of the virus. This approach could evaluate the effects of climate on tick seasonal processes and thus its effects on virus transmission.

In this study, we developed a dynamic model for the transmission of CCHFV in western Palearctic. The original NGM approach was improved by incorporating a process-driven model for the tick vector, Hyalomma marginatum that allowed estimating tick performance at short time intervals. We specifically explored how changes in the parameters of the model and variations in temperature in the western Palearctic modify the endemic potential of CCHFV in the region through their impact on virus transmission routes and/or tick developmental and survival rates.

Materials and methods

Modelling approach

We used previously developed methods for the analysis of tick-borne pathogens, based on the NGM (Hartemink et al. 2008; Matser et al. 2009; Supporting Information). Instead of using an idealized tick phenology (Davis and Bent 2011), we used a process-driven model of the tick life cycle previously developed and validated (Estrada-Peña et al. 2011). Therefore, basic processes governing the epidemiology of CCHFV arise from tick life cycle processes. The model was applied over the known distribution area of the main tick vector in the western Palearctic, thus accounting for the variability in virus circulation in both time and space. The impact produced on R0 by changes in climate factors and biological parameters of the model were incorporated to produce a dynamic picture of the epidemiological features of CCHFV. This model was not intended for high-resolution analysis, because of the existence of uncertainties at low spatial scales. Instead, it was developed to understand the patterns of virus transmission over large areas.

The Hyalomma marginatum life cycle model

The process-driven model for the main tick vector of CCHFV was composed of discrete, deterministic equations that convert temperature and atmospheric water vapour data into mortality and development rates of different tick stages. The model used climate data at 10-day intervals, at a spatial resolution of 10 min over the western Palearctic, covering from 18°W, 68°N (top left corner) to 49°E, 27°N (bottom right corner). Climate values were obtained from the New LocClim software (Grieser et al. 2006), which includes daily estimates of temperature and water vapour obtained between 1961 and 1990. Daily climate values provided by New LocClim were converted into 10-day intervals by averaging the original data. Model development, parameterization and empirical validation were previously published (Estrada-Peña et al. 2011, see Supporting Information).

NGM for the simulation of R0 for CCHFV

Our contribution to the dynamics of R0 in a tick-CCHFV system is a framework to evaluate the tick phenology with the life's cycle model of the tick as driven by climate features and its impact on viral circulation routes (transovarial, systemic, nonsystemic). We otherwise adhered to published contributions (Hartemink et al. 2008) to build the NGM. Our approach calculated the basic ‘types’ of infections for tick-borne infections with a 5 × 5 matrix structure. The systemic transmission from ticks to hosts, the nonsystemic (cofeeding) transmission among ticks feeding simultaneously on a host (Labuda et al. 1993) and transovarial transmission were used in the model. Briefly, the NGM contained the numbers of individuals that are infected by one infected individual of each type-at-birth during the time it is infectious. Types-at-birth refer to ticks or hosts that become infected (‘birth’ of an infected, sensu Hartemink et al. 2008). Types-at-birth were labelled as (i) ticks infected as an egg, (ii) ticks infected by nonsystemic transmission, (iii) hosts infected by ticks feeding on them and (iv) ticks infected by competent hosts. The model was parameterized with transmission efficiency values taken from the literature for (i) female ticks to eggs (R), (ii) ticks to hosts (P), (iii) hosts to ticks (Q), and (iv) ticks to ticks. Phenology values for one complete generation of the tick then modulated the NGM values for the basic ‘types’ of infection. R0 was computed with the package ‘popbio’ working on the R development framework (R Development Core Team 2011).

Sensitivity of R0 to changes in model parameters and temperature variations

Other than the phenology of the tick and its abundance, the transmission of the virus is further regulated by a series of biological parameters summarized by Matser et al. (2009), which have been considered and evaluated in laboratory studies. These parameters are H (fraction of tick blood meals on infective hosts), i (duration of host infection), R (efficiency of transovarial transmission), θ (efficiency of tick-to-tick transestadial transmission) and P + Q (transmission from hosts to ticks and vice versa). These parameters drive the biological aspects of the vector and the pathogen. Values of pathogen transmission rates were obtained from laboratory studies. Fixed values were used for the biotic parameters of the model, and the transmission rates were further varied over plausible ranges based on literature estimates (Tables S1 and S2). We specifically tested the effects of temperature on tick vector developmental and survival rates and thus indirectly on R0. This is an oversimplified approach if the effects of climate on the expected distribution of the tick are to be evaluated. However, we wanted to assess the claims on the effects of temperature alone on the tick life cycle and thus on the epidemiological potential of CCHF. Further on this, it is ignored how water vapour contents, relevant for computing of tick's processes, would change as a consequence of temperature scenarios. The use of both temperature and water vapour would generate increasing levels of complexity and uncertainty. We therefore focused only on the effects of temperature, and we used the upper and lower temperature limits (5 and 95% percentiles, respectively) recorded over the target geographical region (Grieser et al. 2006). After each run, the sensitivity and elasticity were computed to assess the contribution of each transmission route of the virus under each set of conditions (Matser et al. 2009).

Results

Potential CCHFV circulation areas in western Palearctic

A process-driven model for the life cycle of the tick vector, H. marginatum, showed the range of potential areas for CCHFV circulation in western Palearctic (Fig. 1a). The highest R0 values were observed over large areas of the Mediterranean basin, with areas of lower but ≥1 R0 values in the Balkans, northern Italy, south-western France and even parts of the United Kingdom. Changes in temperatures had a clear influence on the variations of R0. Low temperatures decreased R0 values by 96% when compared to data obtained with average temperatures (Fig. 1b). Using scenarios of high temperatures increased R0 by 116% (Fig. 1c). The size of the territory where R0 ≥ 1 decreased by 87% or increased by 114% with series of low and high temperatures, respectively. As expected, the temperature had an effect on the suitable habitat for the tick vector. The range of R0 ≥ 1 values is higher at northern latitudes when the series of higher temperatures were used. Changes in R0 by temperature increase (decrease) resulted in an increase (decrease) in tick abundance because of lower (higher) mortality rates and faster (slower) developmental periods (P < 0·04 for mortality; P < 0·02 for development). However, contributions of transovarial and nonsystemic transmission routes to viral circulation did not change significantly (P = 0·42) under the scenarios of temperature. These estimated values of high or low temperatures had an impact on the viral transmission rates affecting the life processes of the tick vector.

Figure 1.

Basic Reproduction Rate (R0) for Crimean–Congo haemorrhagic fever virus (CCHFV) in western Palearctic. (a) R0 geographical range. (b, c) R0 changes in simulations with low (b) and high temperatures (c). (d, e) Contribution of the nonsystemic and transovarial transmission routes according to R0 (d) and the latitudinal gradient (e). (image_n/jam12039-gra-0001.png) Non-systemic transmission; (image_n/jam12039-gra-0002.png) transovarial transmission. Areas in white are those where R0 < 1. Low- and high-temperature values are the low and high 5% of the historical climate series in the area.

Viral circulation was mainly supported by transovarial transmission and transmission from and to new competent reservoir hosts. The nonsystemic transmission route had a small contribution to virus dynamics. As R0 values increased, the contribution of transovarial transmission also increased (Fig. 1d). The contribution of transovarial transmission tended to be higher towards southern latitudes, without a clear pattern across a longitudinal gradient (Fig. 1e).

Analysis of model biological parameters

The effect of variations in the critical biological parameters of the model, H (fraction of tick blood meals on infective hosts), i (duration of host infection), R (efficiency of transovarial transmission), θ (efficiency of tick-to-tick transestadial transmission) and P + Q (transmission from hosts to ticks and vice versa) was evaluated. Increments in all these parameters increased R0, and their decrements produced the consequent decrease in R0. However, these parameters made different contributions to virus circulation, the sensitivity of the system and the relative changes in R0 across the geographical range of the study. A decrease in H highlights the relative importance of both transovarial and nonsystemic transmission routes that became the dominant routes in virus circulation. In turn, values of R0 were slightly lower under this scenario. Combined transovarial and nonsystemic transmissions accounted for near 100% of total contributions to virus transmission at low values of H. Increases in H resulted in a highest contribution of the nonsystemic transmission route alone to virus circulation (Fig. 2a).

Figure 2.

Changes in R0 geographical distribution and in the relative contribution of different virus transmission routes according to variations in biological parameters of the model. Chart and map at left and right display the model runs with low and high values of the corresponding biological parameters, respectively. R0 values are displayed in a gradient from dark (R0 = 1) to light (R0 = 35). (a) Fraction of tick blood meals on hosts competent for systemic transmission, H = 0·1 (left) and H = 0·9 (right). (b) Duration of systemic infection in days, i = 1 (left) and i = 20 (right). (c) Efficiency of virus transmission from infected hosts to ticks (P) and from infected ticks to hosts (Q), P or Q = 0·1 (left) and P or Q = 0·9 (right). (d) Efficiency of virus transovarial transmission, R = 0·1 (left) and R = 0·9 (right). (e) Efficiency of tick-to-tick transmission, θ = 0·1 (left) and θ = 0·2 (right). (image_n/jam12039-gra-0001.png) Non-systemic; (image_n/jam12039-gra-0002.png) transovarial.

The increase in i had the effect of decreasing the contribution of both transovarial and nonsystemic transmission routes because of the increase in the contribution of the systemic transmission route. In contrast, the decrease in i produced an increase in both transovarial and nonsystemic transmission routes. Variations in i produced only slight variations in the range of R0 ≥ 1 (P = 0·09, Fig. 2b). Variations in PQ affected R0 in a way similar to i. Therefore, the increase in R decreased the contribution of both transovarial and nonsystemic transmission routes because of the increase in the contribution of the systemic transmission route. A decrease in R produced an increase in both transovarial and nonsystemic transmission routes. The simulated changes in R accounted only slight variations in the range of R0 ≥ 1 (P = 0·18, Fig. 2c). The changes in R alone produced the highest effect on the R0 values (P < 0·001) with a twofold increase in some areas (Fig. 2d). This increase may switch the R0 from <1 to >1 in some areas. A large decrease in θ had a high impact by decreasing R0 values over the entire geographic range of the study (P < 0·01). Using a low efficiency of virus transovarial transmission at high θ increased the impact of the nonsystemic transmission route, which accounted for up to 85% of the total contribution to virus circulation. With increasing θ, the nonsystemic transmission route made the most significant contribution to the increase in the R0 values (Fig. 2e). The nonsystemic transmission route was dominant with low efficiencies of transovarial transmission, accounting for over 90% of the viral circulation in some cases. However, the transovarial transmission route supported around 50% of viral circulation at high R values.

In summary, the system was very sensitive to small changes in transovarial transmission, stressing its contribution to the virus circulation. The system had a low sensitivity to changes in θ, with moderate sensitivity to temperature changes (Fig. 3) but showed a high sensitivity to changes in H only in the highest range of R0 values, supporting the importance of systemic transmission only under some epidemiological conditions.

Figure 3.

Sensitivity of the model to changes in the basic parameters governing the biotic features of Crimean–Congo haemorrhagic fever virus (CCHFV) transmission (H, I, P, Q, R, θ) and temperature (T) with respect to R0 values. (image_n/jam12039-gra-0003.png) H; (image_n/jam12039-gra-0004.png) I; (image_n/jam12039-gra-0005.png) PQ; (image_n/jam12039-gra-0006.png) T; (image_n/jam12039-gra-0007.png) R; (image_n/jam12039-gra-0008.png) θ.

Discussion

The finding of CCHFV in south-western Europe (Estrada-Peña et al. 2012b) suggested that the presence of the virus together with the wide distribution of H. marginatum, its main vector, in the Mediterranean basin, represent a risk for virus circulation in the region (Gale et al. 2011). Our study linked a process-driven tick life cycle model with projection matrices and elasticity analysis as an improvement towards the characterization of the potential endemic area for CCHFV in western Palearctic. Concerns exist about the existence of suitable climate and host abundance conditions to cause an emerging CCHF epidemic in this region (Gale et al. 2011). We ignore basic epidemiological parameters for the transmission of CCHFV and how changes in transmission rates among ticks and competent reservoir hosts affect virus circulation and geographic range. The model showed how changes in climate and virus transmission rates affect CCHFV distribution in the region and how virus transmission routes are affected by these changes.

Two main results were obtained from our simulations, (i) transovarial transmission is essential for CCHFV circulation and (ii) temperature changes affect tick vector seasonal development and mortality rates, thus promoting both higher R0 values and a larger geographical range for viral circulation. However, the effects of temperature alone would not change the relative importance of transovarial transmission on CCHFV circulation. The effects of climate on the epidemiology of CCHFV have been discussed elsewhere (Gray et al. 2009). It is often claimed that warmer temperatures would promote epidemics in the context of the ongoing debate about the impact of climate on tick-borne pathogens (Ergonul and Whitehouse 2007; Maltezou and Papa 2010). However, it was previously shown that climate did not change significantly in the areas of Turkey where the recent and still ongoing CCHF epidemics occur (Estrada-Peña et al. 2011). Our model showed that a rise in temperature might lead to the expansion of the area with endemic potential for CCHFV. Higher temperatures had a direct impact on faster tick development rates and thus a decrease in density-independent mortality, which in turn affected parasite loads on hosts. A secondary impact of higher temperatures is that oviposition would take place well before autumn, improving the survival of developing eggs (Emelianova 2006) and likely promoting the geographical spread of the tick. However, changes in tick seasonality because of climate alone did not promote large changes in R0 or in the contribution of transovarial transmission as the main route for virus circulation. Temperature values used here were the upper and lower 5% statistical limits of the historical series of weather in the region, within the normal temperature variability in the region and thus considered as the most realistic scenario for the analysis.

CCHFV produces persistent infections that last almost for the entire tick life with epidemiological significance for perpetuation of the virus in nature (Kondratenko 1976; Shepherd et al. 1989). CCHFV transovarial transmission has been demonstrated under both natural and experimental conditions, and the epidemiological significance of these findings is well recognized (Kondratenko et al. 1970). The decrease in the efficiency of transovarial transmission had the single largest effect on the decrease in mean R0 values over the studied region. The model was very sensitive to small changes in transovarial transmission rates, further supporting the hypothesis that this transmission route has the main contribution to the maintenance of active virus foci. All data suggest that transovarial transmission together with high systemic transmission between hosts and ticks are the most effective ways for virus circulation in nature. Therefore, it is proposed that higher CCHFV prevalence rates in ticks are expected in sites where adult ticks concentrate on infected ungulates, and where particular climate conditions favour the survival of the tick's eggs that became infected by transovarial transmission. The strong contribution of transovarial transmission on CCHFV circulation has been overlooked in other reports, because it was considered to be dependent only on those ticks feeding on humans, which are accidental hosts (Matser et al. 2009). A tick-pathogen system cannot exclusively rely on the outcome from an accidental host. While clinically affected, humans have no role on virus transovarial transmission in the field because the contribution of adult ticks feeding on ungulates is much larger.

As expected, every change indirectly promoting higher transovarial transmission rates increased R0 values. However, the model showed that nonsystemic transmission plays a minor role in the circulation of CCHFV. Our model did not explicitly consider host density as affecting the dynamics of the system, and we modelled parasitic loads by climate-driven tick survival rates and a negative binomial distribution. With this approach, ticks have a fixed mortality rate, which may affect more or less individuals according to the activity rates driven by climate. This is assumed to model tick infestations when the seasonal patterns of tick abundance are estimated and mortality rates of the many hosts supporting populations of immature ticks are unknown (Poulin 1993). We considered that the use of climate-driven tick activity rates and fixed density-derived tick mortality rates, while not totally free of criticisms, is a realistic way to characterize the impact of the other parameters of the model.

We acknowledged that local processes, which depend on climate and host abundance, might be relevant only under particular conditions. These processes must be incorporated into models of spread after full characterization of tick-dependent processes. Linking a tick life cycle model with recent developments in NGM helped to assess these epidemiological features in a large territory. These findings open an interesting perspective for spatial models that incorporate host rates and seasonality, therefore allowing the use of tick and host phenology. The simulations conducted here showed that warmer temperatures have a positive effect on CCHFV spread into northern latitudes in the western Palearctic by affecting tick vector survival rates. However, the high impact of virus transovarial transmission did not change under realistic scenarios, therefore suggesting that other factors such as hosts for immature ticks have a low impact on the local maintenance and spread of infective foci. The abundance of domestic and wild ungulates that are hosts for adult ticks together with an increasingly suitable habitat for tick survival may be the key factors for CCHFV spread.

In conclusion, a scenario of increased temperature is compatible with the spread of CCHFV in the western Palearctic, because expansion of the habitat suitable for tick vectors and thus the area available for virus circulation. This scenario would affect the whole region and will increase virus circulation in sites where high tick populations may already exist. However, a warmer climate will not affect the relative contribution of the main virus transmission routes. The relative importance of transovarial route of transmission from ticks to ticks may be greatly exacerbated by uncontrolled movements and significant increase in ungulate hosts for adult ticks.

Acknowledgements

This research was supported by EU FP7, ANTIGONE project number 278976. Francisco Ruiz-Fons was funded by a ‘Juan de la Cierva’ contract from the Spanish Ministry for the Economy and Competitiveness. P.A. holds a Beatriu de Pinós fellowship funded by ‘Comissionat per a Universitats i Recerca del Departament d'Innovació, Universitats i Empresa' of the Generalitat de Catalunya, and the COFUND Programme – Marie Curie Actions under 7th Marc Programme of the European Community.

Ancillary