Distribution and Clinical Manifestations of Cryptosporidium Species and Subtypes in HIV/AIDS Patients in Ethiopia

Background Cryptosporidiosis is an important cause for chronic diarrhea and death in HIV/AIDS patients. Among common Cryptosporidium species in humans, C. parvum is responsible for most zoonotic infections in industrialized nations. Nevertheless, the clinical significance of C. parvum and role of zoonotic transmission in cryptosporidiosis epidemiology in developing countries remain unclear. Methodology/Principal Findings In this cross-sectional study, 520 HIV/AIDS patients were examined for Cryptosporidium presence in stool samples using genotyping and subtyping techniques. Altogether, 140 (26.9%) patients were positive for Cryptosporidium spp. by PCR-RFLP analysis of the small subunit rRNA gene, belonging to C. parvum (92 patients), C. hominis (25 patients), C. viatorum (10 patients), C. felis (5 patients), C. meleagridis (3 patients), C. canis (2 patients), C. xiaoi (2 patients), and mixture of C. parvum and C. hominis (1 patient). Sequence analyses of the 60 kDa glycoprotein gene revealed a high genetic diversity within the 82 C. parvum and 19 C. hominis specimens subtyped, including C. parvum zoonotic subtype families IIa (71) and IId (5) and anthroponotic subtype families IIc (2), IIb (1), IIe (1) and If-like (2), and C. hominis subtype families Id (13), Ie (5), and Ib (1). Overall, Cryptosporidium infection was associated with the occurrence of diarrhea and vomiting. Diarrhea was attributable mostly to C. parvum subtype family IIa and C. hominis, whereas vomiting was largely attributable to C. hominis and rare Cryptosporidium species. Calf contact was identified as a significant risk factor for infection with Cryptosporidium spp., especially C. parvum subtype family IIa. Conclusions/Significance Results of the study indicate that C. parvum is a major cause of cryptosporidiosis in HIV-positive patients and zoonotic transmission is important in cryptosporidiosis epidemiology in Ethiopia. In addition, they confirm that different Cryptosporidium species and subtypes are linked to different clinical manifestations.


Introduction
Cryptosporidium is an important protozoan parasite affecting HIV/AIDS patients, causing diarrhea, wasting syndrome, and reduced life quality [1].Since specific therapy or vaccine for the control of this parasite is not yet available, preventing infections depends on avoiding exposure to the parasite and maintaining immune competence.In industrialized nations, access to highly active antiretroviral therapy (HAART) has significantly reduced the morbidity and mortality by cryptosporidiosis [2][3][4].Nonetheless, cryptosporidiosis is still a major threat to AIDS patients who do not have access to HAART, especially in developing countries [5][6][7][8].In industrialized nations, transmission of cryptosporidiosis via contaminated drinking and recreational water and contact with infected farm animals remains a major public health problem in both HIV-positive and immunocompetent persons [9][10][11].
The use of molecular epidemiologic tools has provided new insights into the diversity of Cryptosporidium species infecting humans and animals [12].So far, 26 Cryptosporidium species have been described [13][14][15][16][17].Most human cases are caused by C. hominis and C. parvum [12].The latter also infects some other mammals, notably calves and lambs, and is responsible for most zoonotic infections in humans.Several other Cryptosporidium species are seen in humans at lower frequency, including C. meleagridis, C. felis, C. canis, C. ubiquitum, and C. cuniculus [12].More recently, a new species, C. viatorum, has been described in 10 travelers returning to Great Britain from the Indian subcontinent [17].This species appears to be a human-specific pathogen and has since been found in 2 Swedish travelers to Africa and Latin America [18].In children and HIV-positive persons, differences in clinical manifestations have been observed among different Cryptosporidium species, especially between C. hominis and C. parvum, with the former more virulent than the latter [19][20][21].In addition, infections with C. parvum were associated with chronic diarrhea and vomiting in HIV-positive persons more frequently than was C. hominis [21].
Sequence characterization of the 60-kDa glycoprotein (gp60) gene has been commonly used in subtyping C. hominis and C. parvum [12].Differences have been observed in host specificity among C. parvum subtype families and in virulence among C. hominis subtype families.Thus, C. parvum subtype family IIa is commonly found in calves, IId is mostly found in lambs and goat kids, whereas IIc is mostly found in humans.Within C. hominis, studies in Peru have shown that subtype family Ib was more virulent than other subtype families in children, whereas subtype family Id was more virulent than subtype families Ia and Ie in HIV-positive patients [20,21].
Cryptosporidiosis is endemic in Ethiopia; occurrence rates of 7.6% to 43.6% were reported in HIV/AIDS patients [22][23][24][25][26].Some potential risk factors for cryptosporidiosis occurrence included contamination of drinking water, contact with calves, living in overcrowded households with many family members, and poor personal hygiene.Thus far, only one study has genetically characterized Cryptosporidium spp.from Ethiopia.In the study, 39 of the 41 specimens genotyped had C. parvum, one had C. hominis, and one had both species.All 12 C. parvum specimens subtyped by sequence analysis of the gp60 gene belonged to the subtype family IIa [22].
In the present study, we examined the occurrence of Cryptosporidium infection in HIV/AIDS patients in Ethiopia and characterized Cryptosporidium spp. at the species, subtype family, and subtype levels.We also examined the association between clinical manifestations and infections with specific Cryptosporidium species and subtype families.Data generated from the study have clearly shown a dominance of C. parvum in the study population, importance of zoonotic transmission in the epidemiology of cryptosporidiosis in Ethiopia, and differences in clinical manifestations among Cryptosporidium species and subtypes.

Ethical statement
The research protocol was approved by the Ethical Clearance Committee of the Addis Ababa University.All study participants had given written informed consent before enrollment into the study.When the study participant was a child, written consent was obtained from his or her parent or guardian.Researchers at the Centers for Disease Control and Prevention (CDC) had no contact with patients and no access to personal identifiers.Laboratory work on the study specimens was covered under CDC IRB protocol No. 990115: ''Use of residual human specimens for the determination of frequency of genotypes or sub-types of pathogenic parasites''.

Study population and enrollment
This study was cross-sectional in nature.It was conducted between September 2009 and December 2011 in Addis Ababa, Ethiopia.A total of 520 HIV/AIDS patients were recruited from in-patients (hospitalized) and outpatients attending the Tikur Anbessa Hospital, Addis Ababa University, and patients referred to the study by attending physicians.The criteria for inclusion in the study were documented HIV infection, the ability to provide informed consent by the patient or the guardian, and willingness to provide one stool specimen.A structured questionnaire was used to collect CD4+ cell counts, demographic data, clinical symptoms, HAART history, antibiotics usage, and animal exposure history.Each participant was asked to provide a single fresh stool specimen.

DNA extraction
Stool specimens were stored in 2.5% potassium dichromate at 4uC and shipped to the CDC laboratory in Atlanta for screening and molecular characterization of Cryptosporidium spp.by PCR.After washing the stool specimens twice with distilled water, genomic DNA was extracted from 0.5 ml of fecal materials using a FastDNA SPIN Kit for Soil (MP Biomedicals, Solon.OH) and eluted in 100 ml of reagent-grade water following the manufacturer-recommended procedures.DNA was stored at 280uC until analyzed by PCR.

Cryptosporidium species detection, genotyping and subtyping
Cryptosporidium oocysts present in the specimens were detected by nested PCR amplification of an approximate 830 bp fragment of the small subunit (SSU) rRNA gene as described previously [27].Cryptosporidium species were determined by restriction fragment length polymorphism (RFLP) analysis of the secondary PCR products using endonucleases SspI and VspI [27].PCR products and restriction fragments were subjected to electrophoresis in 1.5% and 2% agarose gels, respectively, and visualized after staining with GelRed (Biotium Inc., Hayward, CA).All secondary PCR products from species other than C. parvum were sequenced to confirm the identification.Specimens that contained C. parvum or C. hominis were further subtyped by DNA sequencing of the nested PCR product of the gp60 gene [28].Each specimen was analyzed at least twice by PCR at each locus using 2 ml of the DNA extraction per PCR.As a positive control, C. baileyi DNA was used in SSU rRNA PCR and C. hominis DNA was used in

Author Summary
The disease burden of Cryptosporidium parvum and role of zoonotic transmission in cryptosporidiosis epidemiology are poorly understood in developing countries.In this study, we examined the distribution and clinical manifestations of Cryptosporidium species and subtypes in HIV/ AIDS patients in Addis Ababa, Ethiopia.Using molecular diagnostic tools, we detected Cryptosporidium infection in 26.9% of 520 HIV/AIDS patients studied.We have shown a very high diversity of Cryptosporidium species and subtypes in these patients, but unlike in other developing countries, C. parvum is overwhelmingly the dominant species in the study community, responsible for ,65% Cryptosporidium infections.The common occurrence of C. parvum zoonotic subtype family IIa, combined with calf contact as a significant risk factor, suggest that zoonotic transmission is important in cryptosporidiosis epidemiology in HIV/AIDS patients in Ethiopia.We have also shown that different Cryptosporidium species and subtypes are linked to different clinical manifestations.Improved hygiene and avoidance of calf contact should be advocated to reduce cryptosporidiosis transmission in HIV/AIDS patients in the study setting.gp60 PCR.A negative control using DNase-free water was also included in each PCR run.

Sequence analysis
PCR products were sequenced using the forward and reverse primers of the secondary PCR.An intermediary sequencing primer gp60-R3 [59-GAG ATA TAT CTT GTT GCG-39] was also used in the sequencing of gp60 PCR products.DNA sequencing was done using the ABI BigDye Terminator v. 3.1 Cycle Sequencing Kit (Applied Biosystems, Foster City, CA, USA) and an ABI3130 Genetic Analyzer (Applied Biosystems).Sequence accuracy was confirmed by sequencing of two PCR products from each positive specimen.Nucleotide sequences obtained were aligned with reference sequences using the ClustalX 1.81 package (http://www.clustal.org/)to identify Cryptosporidium species and C. parvum subtypes.Subtypes of C. parvum and C. hominis were named based on the established nomenclature system [29].Unique sequences generated in this study were deposited in GenBank under accession numbers AB830575 to AB830590.

Statistical analysis
Data from the study were analyzed using the SPSS 20.0 for Windows software (IBM Corp, Armonk, NY) at three levels of parasite categorization: presence of Cryptosporidium, species of Cryptosporidium, and subtype families of C. parvum and C. hominis.Data from persons infected with low-frequency species were pooled based on their genetic similarities.Univariate and multivariate logistic regression modeling were used to analyze the association between Cryptosporidium infection and clinical symptoms or risk-factors while adjusting for potential confounders when the sample size was sufficient.For comparisons at the species or subtype family level, separate models and different subsets of the main dataset were run to examine the effect of each species or subtype family, with Cryptosporidium-negative as the referent.The Hosmer-Lemeshow test was used to assess the goodness-of-fit of each multivariate logistic regression model.The strength of the associations was estimated by odds ratios (OR) and 95% confidence intervals (CI).The association was considered statistically significant when the P value obtained was smaller than 0.05.

Study patients and Cryptosporidium occurrence
Among the 520 HIV/AIDS patients who participated in this study, 276 (53.1%) were males and 244 (46.9%) were females.The median age of the study participants was 41 years (range: 7 months to 86 years), and the mean CD4+ cell count was 278 cells/mL.Almost one-third (32.9%) of the study patients were bedridden and hospitalized.Of the 520 stool specimens examined in this study, 140 (26.9%)were positive for Cryptosporidium by the SSU rRNAbased PCR technique (Table 1).There were no significant differences in prevalence of Cryptosporidium infection among age groups and between the male and female gender (Table 1).No significant association between hospitalization and Cryptosporidium infection was found; percentages of Cryptosporidium infection in hospitalized and non-hospitalized patients were 35.7% and 37.4% (P = 0.83).In addition, no significant difference was observed in the occurrence of vomiting, diarrhea, age, HAART, antibiotic use, animal contact and CD4 between inpatients and outpatients.There was also no association between infections with Cryptosporidium spp. or any specific species and hospitalization rates (data not shown).Cryptosporidium species and C. hominis and C. parvum subtypes Species determination by RFLP was successful for 128/140 Cryptosporidium-positive specimens (Table 2).Cryptosporidium parvum (n = 92) and C. hominis (n = 25) were the species most frequently detected, followed by C. felis (n = 5), C. meleagridis (n = 3), C. canis (n = 2), and C. parvum and C. hominis co-infection (n = 1).All SSU rRNA PCR products from non-C.parvum specimens were sequenced, confirming the RFLP results.In addition, 12 specimens demonstrated a RFLP pattern that was similar to C. parvum, but with a slightly smaller upper SspI band.DNA sequences from two of the specimens were identical to a reference sequence of C. xiaoi (GenBank accession no.JQ413443) and those from 10 specimens were identical to a reference sequence (GenBank accession no.HM485434) of C. viatorum, a recently described species in humans [17].
Subtype family data were obtained from 82 (89.1%) of the 92 participants with C. parvum and showed the presence of subtype families IIa, IIb, IIc, IId, IIe, and a new subtype family genetically related to If in 71, 1, 2, 5, 1, and 2 persons, respectively.In contrast, 19 (76%) of the 25 participants with C. hominis were successfully subtyped and showed the presence of subtype families

Cryptosporidium species/subtypes and clinical manifestations
Cryptosporidium infection was significantly associated with diarrhea in univariate analysis (P,0.001), especially in patients with C. parvum and C. hominis (P,0.001 and P = 0.012 respectively).Vomiting was also more often seen in Cryptosporidium-positive than Cryptosporidium-negative patients (63.6% versus 51.8%; P = 0.018).Patients with C. hominis or rare species (C.meleagridis/C.felis/C.canis and C. xiaoi) were more likely to have vomiting compared with Cryptosporidium-negative patients (P = 0.057 and P = 0.027, respectively; Table 3).
Diarrhea was significantly associated with infections with C. parvum subtype family IIa compared with Cryptosporidium-negative patients (78.9%;P = 0.003).However, there were no significant differences in diarrhea occurrence between patients infected with C. hominis subtype families and Cryptosporidium-negative patients.Patients infected with different C. parvum and C. hominis subtype families also did not show significant differences in the occurrence of vomiting compared to Cryptosporidium-negative patients (P.0.05;Table 3).
Multivariable modeling was attempted for diarrhea and vomiting occurrence in order to adjust for age, gender, type of patients and clinical parameters (HAART and CD4+); however, it did not reveal major differences compared with the crude OR (Table 3).

Cryptosporidium species/subtypes and risk factors
After controlling for other potential risk factors, history of any contact with animals was associated with overall Cryptosporidium infections (OR = 1.6;P = 0.04), and with C. parvum (OR = 2.5; P = 0.002) and its subtype family IIa (OR = 2.1; P = 0.02) in particular.This association was mostly due to contact with calves (OR = 1.6 and P = 0.02 for overall Cryptosporidium infection; OR = 1.8 and P = 0.01 for C. parvum; and OR = 2.0 and P = 0.01 for C. parvum subtype family IIa; Table 4).However, there was no significant association between Cryptosporidium infection and age, gender, HAART history, CD4+ cell counts, antibiotics use, or type of patients (P = 0.56, 0.13, 0.59, 0.90, 0.84, and 0.83, respectively).No significant association was found between Cryptosporidium infection and CD4+ cell counts or HAART history at the species and subtype levels (Tables S1 and S2).

Discussion
The present findings showed that (1) Ethiopian HIV/AIDS patients were infected with a diverse population of Cryptosporidium species, including the unusual species C. viatorum and C. xiaoi; (2) C. parvum was the most frequently detected species; and (3) Cryptosporidium species or subtype families were associated with different clinical manifestations.The dominance of C. parvum in this study is in agreement with the previous observation in a small study in Ethiopia [22].It is, however, in sharp contrast with studies of human cryptosporidiosis in other developing countries where C. hominis dominates [12,30,31].In Europe, the two species are Multivariable logistic regressions were performed to adjust the potential confounders including age, gender, HAART, CD4, and type of patients when sample size was sufficient.na: sample size was too small for multivariate logistic regression analysis.

##
Hosmer and Lemeshow test was applied to test the goodness of fit of multivariate logistics regression models.
*95% CI: 95% confidence intervals.**For each Cryptosporidium species or subtype family, patients with the species or subtype family were taken as ''positive'', patients who were not infected at all were taken as ''negative'' (referent), while patients infected with other species or subtype families were not included in this specific model.doi:10.1371/journal.pntd.0002831.t003almost evenly distributed, with C. parvum being more prevalent in some reports [32] and C. hominis in others [33].Data from this study also indicate that the recently established species C. viatorum is more widely distributed than believed, and humans can be infected occasionally with the sheep and goat parasite C. xiaoi.As expected, the prevalence of cryptosporidiosis in HIV+ patients (26.9%) in this study is substantially higher than in the largely healthy persons (8.7%) in a previous study conducted in the same area [22].Differences in geographical distribution of C. parvum and C. hominis are generally considered a reflection of differences in infection sources and transmission routes [12].C. hominis is transmitted almost exclusively among humans, whereas C. parvum, especially its IIa subtype family, is more likely transmitted zoonotically.In industrialized nations, C. parvum infections have often been linked to contact with farm animals, and C. hominis infections to contact with children with diarrhea [34].Both species have been associated with drinking-water outbreaks [33].The predominance of C. hominis in most developing countries suggests that anthroponotic transmission is more important than zoonotic transmission in cryptosporidiosis epidemiology in developing countries in general [12].In contrast, the dominance of C. parvum IIa subtype family in Ethiopia in HIV/AIDS patients suggests that unlike in other developing countries, Cryptosporidium infection in Ethiopia is mostly transmitted zoonotically.Previously, it was shown that that small numbers of C. parvum infections seen in humans in developing countries were mostly caused by the anthroponotic subtype family IIc [12].
Results of the risk factor analysis support the role of zoonotic transmission in cryptosporidiosis epidemiology in HIV/AIDS patients in Ethiopia.Even though the present study was performed in an urban area, more than 50% of patients had contact with animals, as households in the study area usually have farm animals living inside the residence.Thus, in this study, animal contact, especially with calves, was a significant risk factor for Cryptosporidium, especially C. parvum and its IIa subtype family.The distribution of C. parvum subtypes in this study reinforces the likely occurrence of zoonotic transmission, as the majority of C. parvum infections were caused by IIa subtypes (71/82 specimens subtyped), especially its IIaA15G2R1 subtype (60/82 specimens subtyped), which is a dominant C. parvum subtype in calves around the world [12].In contrast, the anthroponotic C. parvum subtype family IIc was seen in only 2/82 of C. parvum infections.
In addition to IIa, C. parvum subtype family IId was also identified in five Ethiopian HIV/AIDS patients.Subtype family IId of C. parvum is generally considered a sheep and goat parasite [35], but has been found at high frequency in calves in China, Egypt, and Sweden [36][37][38].In this study, only one patient infected with IId had contact with sheep in this study.Genotyping and subtyping studies of domestic animals from the study area and additional case-control-studies are needed to support the conclusion on the importance of zoonotic transmission in cryptosporidiosis epidemiology in Ethiopia.
In agreement with previous observations elsewhere [19][20][21], data from the study suggest that different Cryptosporidium species and subtypes are linked to different manifestations of cryptosporidiosis.As expected, cryptosporidiosis in our study was significantly associated with the occurrence of diarrhea.This association, however, was largely attributable to C. parvum and C. hominis; other species, including the newly described C. viatorum, were less pathogenic than these two species.Likewise, the significant association between Cryptosporidium infection and the occurrence of vomiting was also largely attributable to C. hominis and some less frequent species (C.

#
Multivariable logistic regressions were performed to adjust the potential confounders including age, gender, HAART, CD4, and type of patients when sample size was sufficient.na: sample size was too small for multivariate logistic regression analysis.

##
Hosmer and Lemeshow test was applied to test the goodness of fit of multivariate logistics regression models.

###
For each Cryptosporidium species or subtype family, patients with the species or subtype family were taken as ''positive'', patients who were not infected at all were taken as ''negative'' (referent), while patients infected with other species or subtype families were not included in this specific model.doi:10.1371/journal.pntd.0002831.t004 parvum and C. viatorum were largely not associated with vomiting.Within C. parvum, the IIa subtype family also appeared to be more associated with the occurrence of diarrhea than other C. parvum subtype families.We did not observe any effect of CD4+ cell counts and HAART on the occurrence of cryptosporidiosis in HIV/AIDS patients in this study.This was probably largely the result of severe overall immunodeficiency in the study population, as reflected by the low mean CD4+ cell counts (278 cells/mL) and very high hospitalization rate (32.9%) and occurrence of cryptosporidiosis (26.9%).
In conclusion, Ethiopian HIV/AIDS patients with low CD4+ cell counts had an extremely high occurrence of Cryptosporidium infection, even when they were on HAART.Although the majority of cryptosporidiosis cases were caused by C. parvum, there was a high diversity of Cryptosporidium species, with a significant number of cases caused by the newly recognized C. viatorum.These Cryptosporidium spp.and C. parvum subtypes were linked to different clinical manifestations.Therefore, improved hygiene and avoidance of calf contact among this population should be advocated to reduce the occurrence of Cryptosporidium infections, especially those caused by C. parvum IIa subtypes of calves.

Table 1 .
Occurrence of Cryptosporidium spp.(n = 140 cases) in 520 HIV/AIDS patients in Ethiopia by age and gender*.

Table 2 .
Distribution of Cryptosporidium species and C. parvum and C. hominis subtypes in HIV/AIDS patients in Ethiopia (n = 140).

Table 3 .
Association with diarrhea and vomiting by Cryptosporidium species and of C. parvum and C. hominis subtype in HIV/AIDS patients in Ethiopia.Id, and Ie in 1, 13, and 5 persons, respectively.Within C. parvum, 8 subtypes were from the subtype family IIa.This was followed by subtype families IId with 4, If-like with 2 subtypes, and IIb, IIc, and IIe with 1 subtype each.Thus, participants in the study were infected with 14 subtypes of C. parvum.

Table 4 .
Association between animal contact and infection with Cryptosporidium species or C. parvum and C. hominis subtypes in HIV/AIDS patients in Ethiopia*.