INTRODUCTION
Cryptosporidium is one of the most serious causes of waterborne diarrhoea in humans, with neonates and immunosuppressed individuals particularly at risk. Of the 16 Cryptosporidium species recognized today, eight have been reported from human cases [Reference Thompson1–Reference Leoni4]. However, only three are considered important human pathogens: Cryptosporidium hominis, C. parvum and C. meleagridis. C. hominis is largely restricted to humans, while C. parvum is an important zoonotic agent infecting most, if not all, mammals including humans. It is also a major pathogen of ruminant livestock with peak incidence rates occurring during calving and lambing [Reference Lowery5, Reference McLauchlin6]. The third species, C. meleagridis is primarily an avian pathogen. Although common in parts of Latin America [Reference Xiao7, Reference Cama8], it appears to be rare in Northern Europe [Reference McLauchlin6].
Cryptosporidium is transmitted by highly resistant, long-lived oocysts that are passed fully sporulated by the infected host. People become infected through direct contact with an infected individual or by ingesting contaminated food or water. The zoonotic species C. parvum may also be transmitted via animal-to-person contact.
In Ireland, the combination of high annual rainfall, high livestock stocking densities and the use of unfiltered surface water as drinking water render the public water supply vulnerable to contamination. Since January 2004, when cryptosporidiosis became notifiable in the Republic of Ireland (i.e. cases have to be reported to the medical officer of health), the national communicable disease surveillance agency (the Health Protection Surveillance Centre, HPSC) has reported a total of 431 cases in 2004, 568 in 2005 and 367 in 2006 [Reference Garvey and McKeown9]. These figures correspond to crude incidence rates of 10·2, 13·4 and 8·7/100 000 population respectively [Reference Xiao7]. About 605 cases were notified in 2007 [Reference Cloak10–Reference Cloak13]. As faecal samples are only very rarely sent for laboratory diagnosis, these figures are believed to be a gross underestimate (exact data for Ireland are unavailable but estimates for the United Kingdom state that only about 4·6% of all cases of gastrointestinal disease in the community are sent for laboratory testing [14]). The first large-scale outbreak in Ireland occurred in the spring of 2007 in Galway on the West Coast of Ireland. In addition to about 242 confirmed cases, thousands more were affected by the boil water notice and the economic burden of buying bottled water for a prolonged period of time.
Although the parasite poses a significant public health problem in Ireland, its epidemiology on the island is poorly understood. The species has only been identified in a very small proportion of human cases, and no information whatsoever is available on subtypes that occur in the human population. In this paper we describe the prevalence of C. parvum and C. hominis in 199 human cryptosporidiosis cases collected in Ireland between 2000 and 2007. In addition, we discuss the distribution of C. parvum and C. hominis subtypes in the context of previous reports of subtypes identified in humans and neonatal calves on the island of Ireland [Reference Thompson1, Reference Glaberman15].
MATERIALS AND METHODS
Sample collection
Human faecal samples (n=199, collected between 2000 and 2007) that had been diagnosed Cryptosporidium-positive by the Microbiology staff in 10 hospitals (Cavan General Hospital; Cork University Hospital; University College Hospital, Galway; Midlands Regional Hospital, Westmeath; Midwestern Regional Hospital, Limerick; Our Lady's Hospital for Sick Children, Dublin; Portiuncula Hospital, Galway; St James' Hospital, Dublin; Sligo General Hospital; Waterford Regional Hospital) were sent to the Parasitology laboratories at the University College Dublin School of Agriculture, Food Safety and Veterinary Medicine, for further investigation. These represented roughly 5·5% (2005), 16% (2006) and 10% (2007) of the total incidence reported in these years (no surveillance data were collected in 2000) [Reference Garvey and McKeown9].
Epidemiological data
Where possible the following patient information was collected in conjunction with the diagnostic sample: age, county of residence and date of collection.
Molecular analysis
DNA was extracted according to the methods described by Boom et al. [Reference Boom16] as modified by McLauchlin et al. [Reference McLauchlin17]. This technique involves breaking up oocysts in a mini bead-beater followed by DNA extraction in guanidine thiocyanate buffer. Prior to PCR amplification all DNA extracts were further purified by PVP (polyvinylpyrrolidone; Sigma-Aldrich Ireland Ltd, Dublin, Ireland) precipitation [Reference Lawson18].
Species identification was carried out by nested PCR amplification of the 18S rRNA gene fragment according to Xiao et al. [Reference Xiao19]. To differentiate C. parvum and C. hominis from any other Cryptosporidium spp. that may infect humans, 2 μl of the amplified product were digested with 2 U SspI and restriction buffer in a total volume of 5 μl at 37°C for 1 h. To distinguish C. parvum from C. hominis the same amount of amplified product was digested with 2 U VspI under the same conditions. Amplified and digested products were fractionated on 2% agarose gels and visualized by ethidium bromide staining. The PCR–RFLP results based on the 18S rRNA gene fragment were confirmed by a nested PCR for the amplification of the Cryptosporidium oocyst wall protein gene fragment (COWP) according to the protocol published by Spano et al. [Reference Spano20] and modified by Pedraza-Diaz et al. [Reference Pedraza-Diaz21]. For restriction fragment analysis, 2 μl amplified product were digested with 2 U RsaI in the appropriate restriction buffer at 37°C for 4 h. Sequence analysis of the 60 kDa glycoprotein encoding gene fragment (gp60) was used to subtype 104 randomly selected isolates [Reference Peng22, Reference Alves23].
Positive (purified C. parvum DNA) and negative controls (master mix without a DNA template) were included in each batch of PCR amplification reactions. The resulting PCR products were purified using the QIAquick PCR purification kit (Qiagen, Crawley, UK) and sequenced in both directions (GATC Biotech AG, Konstanz, Germany). The sequences were compared with published sequences using NCBI Blast and aligned with the ClustalW sequence alignment programme. Within each Gp60 allele family (i.e. Ib, IIa and IId), subtypes were identified using the nomenclature proposed by Sulaiman et al. [Reference Sulaiman24]. In short, the subtypes are coded according to the number of trinucleotide repeats (TCA and TCG) in the microsatellite region, A14–A21 indicating the number of TCA repeats and G1–G4 indicating the number of TCG repeats. R1 and R2 are used to indicate the number of ACATCA repeats immediately after the trinucleotide repeat sequences. Gp60 fragment sequences for which there were no identical matches in GenBank were deposited under accession numbers EU272171 to EU272175.
Statistical analysis
The relative numbers of C. parum and C. hominis isolated from male and female patients were compared using χ2 analysis.
RESULTS
In total, 50, 31, 58 and 60 samples were examined in 2000, 2005, 2006 and 2007, respectively. C. parvum accounted for 94% (in 2000), 39% (in 2005), 81% (in 2006) and 68% (in 2007) of all samples that were successfully genotyped. All the remaining samples were identified as C. hominis. In a small percentage of samples (between 2% and 20% of the annual total examined) PCR amplification was unsuccessful. This was either because they contained PCR inhibitors or because they had been wrongly identified as being Cryptosporidium positive.
All samples examined from 2000 had been collected in the west of Ireland. Just two of 50 isolates from that year were C. hominis (4%). The number of samples containing C. parvum and C. hominis obtained from different regions around the country between 2005 and 2007 are shown in Table 1. In 2005, the largest proportion of samples that were typed originated from the mid-west (n=11, 44%) and south-east (n=7, 28%). In contrast, national incidence data from 2005 showed that most cases occurred in the west, the south and the south-east of the country. Just over half of all isolates collected in the mid-west and all of the samples from the east and south-east were identified as C. hominis. Sixty-six percent of all samples examined in 2006 were collected in the south-east (n=35). According to HPSC data this region had the third highest incidence of Cryptosporidium that year [Reference Garvey and McKeown9]. Between 14% and 17% of isolates from the south-east and the midlands were C. hominis. In 2007, unprecedented incidence rates caused by the outbreak in Galway, were reported from the west of the country. As in the previous years, incidences were also high in the mid-west, the south-east and the south of the country. Again most samples were received from the south-east (n=29, 51%). All of the samples collected in the east (n=3) and the midlands (n=5), 75% of the samples from north-west (n=3), 15% of samples from the west (n=2) and 10% of the isolates sent from the south-east (n=3) were identified as C. hominis.
HPSC, Health Protection Surveillance Centre.
† All cases originating from the outbreak in Carlow, 2005.
‡ Some of these cases originated from an outbreak in Portlaw, Co. Waterford, 2006.
§ Unfortunately no samples from the outbreak in Galway were submitted to us, however, a number of C. hominis cases that occurred in the neighbouring County Sligo coincided with the outbreak.
The seasonal distribution of the total numbers of C. parvum and C. hominis identified in the samples received between 2005 and 2007 (no faecal collection dates were available for 2000) is shown in Figure 1 together with the overall cryptosporidiosis incidence reported by the HPSC for each year. A very pronounced spring peak in the number of C. parvum coincided with peaks in the total annual incidences of cryptosporidiosis. A second, much smaller peak appears to occur in late autumn. While small numbers of C. parvum cases were identified throughout the rest of the year, none was detected in September and October in either of the 3 years. Small numbers of C. hominis cases were detected throughout the year. There appears to be a slight accumulation of this species during the spring months but numbers were too low to identify definite trends.
Information on the age of the patient was only available in 50% (n=99) of all samples. Sixty-five percent of all stool samples examined had been collected from children aged <6 years, 81% of all isolates originated from children aged <16 years. A plot of the the age distribution of C. hominis and C. parvum (Fig. 2) indicated a trend towards higher proportions of C. hominis in older age groups.
A total of 104 randomly selected faecal samples were subtyped on the basis of the gp60 locus (25 C. hominis and 79 C. parvum isolates). All C. hominis isolates belonged to the same gp60 subtype which was homologous to a previously described DNA fragment logged under the accession number AY167596. According to the nomenclature described by Sulaiman et al. [Reference Sulaiman24] they were identified as gp60 subtype IbA10G2. Of the C. parvum isolates, 78 belonged to the allele family IIa and one to the allele family IId. The IIa alleles fell into 11 subtypes of which IIaA18G3R1 was by far the most prevalent (Fig. 3). All IIa isolates had only one ACATCA repeat following the trinucleotide repeat region (designated R1). The IId allele was identified as subtype IIdA26G1 (identical to logged sequence AY738185) [Reference Sulaiman24].
DISCUSSION
Among the clinical isolates characterized during this study, C. parvum was predominant in all years except in 2005. In that year samples from an outbreak caused by C. hominis in County Carlow in the south-east [26] were over-represented in our sample selection. Overall C. parvum was identified in 80%, and C. hominis in 20% of all cases. They were the only two species identified. This predominance of C. parvum was also observed in other European countries such as France [Reference Guyot27], Switzerland [Reference Fretz28], Portugal [Reference Alves23], and Ireland's closest neighbour, the United Kingdom [Reference Leoni4, Reference McLauchlin6, Reference Pedraza-Diaz21] including Northern Ireland [Reference Lowery5]. The notable exception to this was Spain where human cases with C. hominis outnumbered those with C. parvum [Reference Llorente29]. Moreover, C. hominis was found to be the predominant species in studies carried out in the Americas, Africa, Australia, and Asia [Reference Xiao30]. Most if not all, larger community-scale outbreaks that have occurred on the island of Ireland over the last number of years have been waterborne. Of these, four were attributed to C. hominis [two outbreaks in the greater Belfast area in 2000/2001 with 117 and 230 confirmed cases [Reference Glaberman15]; the incidence in Carlow mentioned above (26 cases) and the large-scale outbreak in Galway in 2007 (242 cases)], and three to C. parvum (in Belfast in 2000 with 129 confirmed cases [Reference Glaberman15]; Westmeath, 2002 (26 cases) [Reference Bonner31]; Waterford, 2006 (8 cases) [Reference O'Flynn and Roch32], respectively). While the number of community-size outbreaks caused by the two species was similar, the number of people affected by outbreaks due to C. hominis was almost four times higher than the number of cases resulting from C. parvum outbreaks. Our results indicated that among sporadic cases the incidence of the two species was reversed with C. parvum being up to four times more common than C. hominis. This has also been observed in the United Kingdom [Reference McLauchlin6, Reference Pedraza-Diaz21]. Waterborne outbreaks due to C. parvum tend to coincide with lambing or calving [Reference Lowery5, Reference McLauchlin6]. In contrast, outbreaks due to C. hominis are reported to occur throughout the year. Interestingly, however, all C. hominis outbreaks in Ireland in the recent past occurred in the spring. Moreover, outbreaks due to either species occur in both rural and urban areas. As McLauchlin et al. [Reference McLauchlin6] pointed out, the source of water and the proportion of surface water used in public water supply is much more important for determining the predominant route of contamination than the community that is served by the water supply. Considering the island's mild and wet climate ideal for the survival and distribution of oocysts combined with the fact that most drinking water catchments are intensively used for agriculture, it is to be expected that contamination of reservoir water bodies with C. parvum is quite common. That waterborne outbreaks due to C. parvum are relatively rare and that fewer people are affected by them when they do occur may be due to a level of background immunity that already exists in a predominantly rural population.
The overall incidence data of cryptosporidiosis in Ireland released by the HPSC [Reference Garvey and McKeown9–Reference Cloak13, Reference Garvey and McKeown25] shows an uneven distribution across the country, with fewer cases in the east and north-east and an increase in numbers towards the south-east, the midlands and the west coast. This distribution is reflected in a decline in population density from east to west and an increase in the importance of agriculture. While C. hominis was more commonly identified in samples from the east of the country, the numbers were unfortunately too small to draw conclusions. However, our results indicate that in the west of the country, C. parvum is by far the most common cause of sporadic cases.
It is generally agreed that the spring peak in human cryptosporidiosis is due to a sharp increase in environmental pollution with C. parvum oocysts during lambing and calving [Reference Lowery5, Reference McLauchlin6]. This was borne out by the large numbers of C. parvum cases identified during the spring months in this study. In the United Kingdom, the national Cryptosporidium incidence has a bimodal distribution, with the autumn peak thought to be in part due to a second calving event later in the year, although both species are detected during this time [Reference McLauchlin6]. Interestingly in Ireland, the autumn peak is absent [Reference Garvey and McKeown9], although there appeared to be a slight increase in C. parvum cases in late autumn. It may be that autumn calving is less practised than in the United Kingdom or that autumn calves are housed earlier limiting animal-to-human contact. Sporadic C. hominis cases occurred throughout the year. It is generally thought that they are more common in patients with a history of foreign travel [Reference McLauchlin6, Reference Pedraza-Diaz21]. Probably as a result of this they tend to be more prevalent among older patients as was observed in the present study. It may also be the case that adults are more likely to seek medical attention when infected with C. hominis because of its greater pathogenicity [Reference Hunter33].
Ever since it has become obvious that great biological and genetic heterogeneity exists within some Cryptosporidium spp., particularly C. parvum, isolates have been typed to subtype level at numerous loci. By this approach it is hoped to identify the most important transmission routes in an area and aid the sourcing of future outbreaks. In the present study we typed just over half of the human isolates to subspecies level at the gp60 locus. This gene codes for a sporozoite surface glycoprotein, is highly polymorphic and contains a microsatellite region. Worldwide the most common C. parvum gp60 alleles identified in humans are IIa and IIc (formerly known as Ic) [Reference Sulaiman24]. IIa was also the most common human C. parvum allele identified in our study. As it is by far the most predominant allele in cattle [Reference Thompson1, Reference Alves23, Reference Peng34], its predominance in human cryptosporidiosis stresses the importance of zoonotic transmission in Ireland. The most prevalent IIa subtype in our study, IIaA18G3R1, was also the most frequently identified human subtype in a study carried out in Northern Ireland in 2000/2001 [Reference Glaberman15]. Glaberman et al. [Reference Glaberman15] typed C. parvum isolates from an outbreak in the greater Belfast area and from several sporadic cases originating from the west coast of Ireland. The IIaA18G3R1 subtype was also the most prevalent C. parvum subtype identified in neonatal calves in Northern Ireland (55·6%) [Reference Thompson1] (Fig. 3). Of the other IIa subtypes identified in our study, IIaA15G2R1, IIaA17G2R1 and IIaA19G3R1 were also detected in a recent study of human cryptosporidiosis samples collected in Australia [Reference O'Brien35]. Subtype IIaA15G2R1 has also been reported from human cases in such widely dispersed places as Portugal [Reference Alves23], Kuwait [Reference Sulaiman24], Canada [Reference Trotz-Williams36] and the United States [Reference Peng34]. Moreover, this subtype and IIaA17G2R1 were prevalent among neonatal calves in Northern Ireland [Reference Thompson1]. On the other hand, another common calf IIa allele (IIaA19G4R1) that was absent in our study was confined to a particular area in Northern Ireland. The authors suggested that all calves infected with this subtype may have had a common source of infection [Reference Thompson1]. The only other C. parvum allele we detected in one human sample was IId. This allele was as common as IIa among children in Kuwait [Reference Sulaiman24], but apart from that has only been found in a small number of patients and cattle in Portugal [Reference Alves23] and a single HIV+ human isolate from Switzerland [Reference O'Brien35]. The anthroponotic allele IIc, which is the most prevalent C. parvum allele in the Americas and Africa [Reference Xiao30] and others described elsewhere (e.g. IIb, IIf) were not detected in Ireland.
All C. hominis isolates belonged to the same allele Ib, and within this allele to the same subtype, IbA10G2. Interestingly two community-based outbreaks in Northern Ireland in 2000–2001 and several sporadic cases from the north-west of England were ascribed to the same Ib subtype [Reference Glaberman15]. Reports from all over the world indicate that Ib has a wide geographic distribution and has been the cause of several water- and foodborne outbreaks worldwide [Reference Glaberman15]. The dominance of a single C. hominis gp60 allele in the Republic and Northern Ireland contrasts with the large variety of C. hominis subtypes (belonging to alleles Ia, Ic, Id, Ie, If) identified in Portugal, India, Canada and Australia [Reference Alves23, Reference O'Brien35–Reference Muthusamy37]. This homogeneity may be the result of Ireland's geographic isolation. It is possible that IbA10G2 is the only endogenous C. hominis subtype or that it was introduced at some time in the past and has since become established on the island. If so it is probably only a matter of time until other exotic C. hominis subtypes are introduced by returning holiday-makers.
To conclude, it appears that in terms of the overall numbers of people affected, zoonotic transmission of C. parvum is more important in sporadic cryptosporidiosis while C. hominis is most prevalent in outbreak situations. Further studies of subtypes that occur in humans and livestock are necessary to better identify the C. parvum subtypes that are most important to human health and to clarify the role of animals as the source of disease outbreaks.
ACKNOWLEDGEMENTS
We thank the staff at the various hospital microbiology departments for making the faecal samples available to us, particularly Martin Cormican, Catherine Byrne, Michael Carton, Paddy Corrigan, Mike Mitchell and Dennis Barren. Thanks are also due to Patricia Garvey from the Health Protection Surveillance Centre for a critical review of this manuscript. We also acknowledge the Food Safety Promotion Board, Ireland and the School of Agriculture, Food Science and Veterinary Medicine, University College Dublin, Ireland for their financial support.
DECLARATION OF INTEREST
None.