Increased Risk for Entamoeba histolytica Infection and Invasive Amebiasis in HIV Seropositive Men Who Have Sex with Men in Taiwan

Background Incidence of Entamoeba histolytica infection and clinical manifestations and treatment response of invasive amebiasis (IA) in HIV-infected patients have rarely been investigated before. Methodology/Principal Findings At the National Taiwan University Hospital, medical records of HIV-infected patients who received a diagnosis of IA between 1994 and 2005 were reviewed. The incidence of amebiasis was investigated in serial blood and stool samples from 670 and 264 HIV-infected patients, respectively, using serological and specific amebic antigen assays. DNA extracted from stool samples containing E. histolytica were analyzed by PCR, sequenced, and compared. Sixty-four (5.8%) of 1,109 HIV-infected patients had 67 episodes of IA, and 89.1% of them were men having sex with men (MSM). The CD4 count at diagnosis of IA was significantly higher than that of the whole cohort (215 cells/µL vs. 96 cells/µL). Forty episodes (59.7%) were liver abscesses, 52 (77.6%) colitis, and 25 (37.3%) both liver abscesses and colitis. Fever resolved after 3.5 days of metronidazole therapy (range, 1–11 days). None of the patients died. The incidence of E. histolytica infection in MSM was higher than that in other risk groups assessed by serological assays (1.99 per 100 person-years [PY] vs. 0 per 100 PY; p<0.0001) and amebic antigen assays (3.16 per 100 PY vs. 0.68 per 100 PY; p = 0.12). In multiple logistic regression analysis, only MSM was significantly associated with acquisition of E. histolytica infection (adjusted odds ratio, 14.809; p = 0.01). Clustering of E. histolytica isolates by sequencing analyses from geographically-unrelated patients suggested person-to-person transmission. Conclusions/Significance HIV-infected MSM were at significantly higher risk of amebiasis than patients from other risk groups. Despite immunosuppression, amebic liver abscesses and colitis responded favorably to treatment.


Introduction
Invasive amebiasis (IA) is the second most common cause of mortality due to parasite infections worldwide, accounting for 40,000 to 100,000 deaths annually. High risk populations for developing IA include infants, pregnant women, and patients who are taking immunosuppressives [1,2]. Interestingly, IA has not been considered to occur at a higher frequency in HIV-infected patients [3,4]. In industrialized countries, the rare occurrence of IA in HIV-infected patients or persons at risk for HIV infection is probably attributed to the rare intestinal carriage of E. histolytica [4][5][6][7][8][9]. This is in contrast with the relatively frequent carriage of the non-pathogenic E. dispar among men who have sex with men (MSM) who attend sexually transmitted diseases clinics [10][11][12][13]. In a retrospective review of medical records of more than 34,000 HIV-infected patients in the US [9], 111 (0.3%) patients were diagnosed as having E. histoytica or E. dispar infection, and only 2 had extra-intestinal amebiasis. Amebiasis was significantly more prevalent among MSM and patients from E. histolytica endemic areas. However, the interpretation of the results of this study is limited by the retrospective study design and failure to differentiate between E. histolytica and E. dispar [14].
In developing countries, studies comparing the prevalence of amebiasis in HIV-infected and HIV-uninfected persons yielded inconsistent results [15][16][17][18][19][20][21][22]. The interpretation of these studies, however, is difficult because a majority of the diagnosis of amebiasis was based solely on microscopic examination of stool samples, which is an insensitive test that fails to distinguish E. histolytica from E. dispar [14]. In a cross-sectional study using stool antigen detection and polymerase chain reaction (PCR) from Mexico, where amebiasis is endemic, investigators found that HIV-infected patients appeared to have a higher rate, though not statistically significant, of E. histolytica infection than their sexual partners or close contacts [23]. However, those patients colonized with E. histolytica did not develop invasive diseases over the 12month follow-up period.
Over the past few years, we and many investigators in Japan, Taiwan, and Korea have found that IA is increasingly diagnosed among HIV-infected MSM [24][25][26][27][28][29][30][31]. Of the estimated 500 to 600 reported cases of amebiasis annually in Japan, 80% of them occurred in MSM [32] and a substantial proportion of patients with IA were also co-infected with HIV and syphilis [24,30]. In Taiwan, an estimated 5-6% of HIV-infected patients developed IA, and in many IA was the presenting disease of HIV infection [31]. Serologic surveys in the US, Italy, Japan, and Taiwan also demonstrated that MSM, regardless of HIV status, were at an increased risk of exposure to E. histolytica [13,26,31,[33][34][35]. Recent detection of locally acquired amebiasis among MSM who had no recent travel to endemic areas for E. histolytica has raised concerns in Sydney, Australia [36]. Oral-anal sexual contact has been found to be significantly associated with acquisition of E. histolytica infection [37]. Although IA has been considered an increasingly important parasitic infection in HIV-infected patients in three East Asian countries, the incidence of amebiasis and the clinical spectrum and the response of IA to standard metronidazole therapy have not been well studied.
In this study, we conducted a longitudinal follow-up study to assess the incidence of E. histolytica infection among persons with HIV infection at a referral medical center for HIV care in Taiwan. We also described the clinical spectrum and treatment outcome of IA.

Retrospective review of invasive amebiasis cases
Medical records of 1109 consecutive, non-hemophiliac HIVinfected patients aged 15 years or greater were reviewed to identify cases of IA at the National Taiwan University Hospital from June 1994 to December 2005 with the use of a standardized case record form. Of the 1109 patients, 781 (70.4%) were MSM. During the study period, a standardized protocol was followed to investigate HIV-infected patients who presented with gastrointestinal symptoms [28,31]. Those investigations included at least two stool specimens for bacterial cultures and microscopy of concentrated wet mount preparations and modified acid-fast staining; indirect hemagglutination (IHA) assay to detect anti-E. histolytica antibodies (Cellognostics, Boehhringer Diagnostics GmbH, Marburg, Germany); endoscopy and biopsy for histopathologic examinations in patients whose stool examinations were non-diagnostic; abdominal sonography followed by computed tomography for patients with abnormal liver function tests, and space-occupying lesions of the liver. Specific Entamoeba antigen assays using commercial test kits (ENTAMOEBA TEST, TechLab, Branchburg, NJ) followed by polymerase chain reactions (PCR) using specific primers for E. histolytica was introduced after 1 January, 2001 [31].
Definite IA was diagnosed when erythrophagocytic trophozoites and/or positive PCR to E. histolytica were identified in clinical specimens from patients with symptoms compatible with IA, such as colitis and liver abscesses [28,38]. Probable IA was confirmed when a patient with IA symptoms responded to metronidazole monotherapy and the aspirates or blood specimens showed high IHA titers, but microbiological cultures for bacteria, fungi, or histopathological examination of aspirates and biopsy specimens did not reveal any other pathogen. Results of the IHA assay were considered positive if the titer was 128 or greater. AIDS was defined according to the 1993 revised classification system for HIV infection and expanded surveillance case definition for AIDS among adolescents and adults [39]. Highly active anti-retroviral therapy (HAART) was defined as anti-retroviral therapy containing two nucleoside reverse transcriptase inhibitors and a (boosted) protease inhibitor(s) or a non-nucleoside reverse transcriptase inhibitor.

Sero-incidence of E. histolytica infection
Serum samples from HIV-infected patients at baseline were tested for IHA and patients who remained in clinic follow-up from January 2001 to December 2005 were re-tested to determine the sero-incidence of E. histolytica infection. The interval between the two blood samples was at least one year. In patients with samples that tested positive at the last clinic follow-up or the end date of the study (31 December, 2006), serially stored serum samples were retrospectively tested to determine the seroconversion date. Seroconversion was defined as changes from sero-negative at baseline to IHA titers of 128 or greater at subsequent IHA assay; or increases of IHA titers by four-fold or greater. The seroconversion date was defined as the mid-point between the dates when the last sero-negative sample and the first sero-positive sample were collected.

Incidence of intestinal infection with E. histolytica
Sequential stool samples from HIV-infected persons were tested for the presence of stool Entamoeba antigen between 1 January, 2001 and 31 December, 2005. Those patients who were negative for Entamoeba antigen were asked to provide stool samples for follow-up testing using the same method in order to assess the incidence of new acquisition of E. histolytica.The interval between the two stool samples were at least 6 months. The date of new infection was estimated as the mid-point between the date when the last antigen-negative sample and the first antigen-positive sample were collected.
Stool specimens tested positive for E. histolytica/E. dispar antigen were further confirmed by PCR. The primer sets for a multiplex nested PCR were based upon the variable regions between 16Slike rDNAs of E. histolytica (GenBank X56991) and E. dispar (GenBank Z49256) [31]. The procedures to isolate total DNA

Author Summary
Entamoeba histolytica, morphologically identical to but genetically different from E. dispar and E. moshkovskii, is the causative agent of amebiasis. Recently there have been reports of increased risk for amebiasis among men who have sex with men (MSM) due to oral-anal sexual contact in several developed countries. In this longitudinal followup study, the incidence of amebiasis was determined among HIV-infected patients using serological and specific amebic antigen assays. DNA extracted from stool samples containing E. histolytica were analyzed by PCR, sequenced, and compared. Clinical manifestations and treatment response of invasive amebiasis in HIV-infected patients were reviewed. The results demonstrated that HIV-infected MSM were at significantly higher risk of amebiasis than patients from other risk groups. Clustering of E. histolytica isolates by sequencing analyses from geographically unrelated patients suggested person-to-person transmission. Despite immunosuppression, amebic liver abscesses and colitis responded favorably to metronidazole therapy. It is important to investigate in areas of high incidence of both amebiasis and HIV (sub-Saharan Africa) how generalizable these findings are.
from the stool samples and the PCR conditions were described previously [31]. Individual E. histolytica isolates were genotyped by PCR amplification and sequencing of the previously described polymorphic loci, locus 1-2, using one set of primers (R1: CTGGTTAGTATCTTCGCCTGT and R2: CTTACACCCC-CATTAACAAT) previously described [40,41]. PCR was carried out in a 50 ml reaction mixture containing 0.1 mg of DNA, a 1.5 mM concentration of each primer, 2.5 mM MgCl 2 , a 100 mM concentration of each deoxynucleoside triphosphate, and 1.5 U of AmpliTaq Gold DNA Polymerase (Applied Biosystems) with the Taq activation at 95uC for 15 min and 30 cycles of denaturation at 94uC for 30 s, annealing at 45uC for 30 s, and extension at 72uC for 1 min, and then final extension at 72uC for 10 min [40]. The PCR products were fractionated by electrophoresis in 3% NuSieve 3:1 agarose (Cambrex, East Rutherford, USA), stained by ethidium bromide and visualized under UV illumination. After purification using the QIAquick PCR purification kit (QIAGEN), locus 1-2 PCR products were sequenced twice in the forward and reverse directions. The sequences from representative genotypes chosen to infer the phylogenetic trees of locus 1-2 were were manually edited and aligned by using BioNumerics V. 4.01 software (Applied Maths, Kortrijk, Belgium). The study protocols were approved by the Institutional Review Board of NTUH and patients gave written informed consent.

Statistical analysis
All statistical analyses were performed using SAS statistical software (Version 8.1, SAS Institute Inc., Cary, NC, U.S.A.). Categorical variables were compared using x 2 or Fisher's exact test and non-categorical variables were compared using Wilcoxon's rank-sum test. The incidence rate of E. histolytica infection or seroconversion was calculated as number of episode per 100 personyears (PY) of observation. Exact 95% confidence intervals (95% CI) for incidence rates were calculated on the basis of the Poisson distribution. The follow-up duration was from the date with the first stool or blood sample that was negative for E. histolytcia antigen or IHA to the date with the sample that was positive, date of death, or on 31 December, 2006, whichever occurred first. Multiple logistic regression analysis was performed between patients who were diagnosed with newly acquired E. histolytica infection by serologies or antigen assays and those who remained uninfected in order to identify the risk factors associated with E. histolytica infection. All tests were two-tailed. A p value ,0.05 was considered significant.

Sero-incidence of E. histolytica infection
Of 991 patients (89.4%) with available IHA assay results at baseline, 66 patients (6.7%) had IHA titers of 128 or greater. Between January 2001 and December 2005, 670 patients, including 433 (63.6%) MSM, who had more than one blood sample available for follow-up IHA assay, were enrolled in the sero-incidence study (Figure 3). There were no significant differences in demographics and clinical characteristics between the 670 patients who had follow-up IHA assays and the 321 who did not (data not shown). There were no significant differences in median CD4 count and plasma HIV RNA load (PVL) between MSM and patients from other risk groups when the first IHA assays were performed. At baseline, a significantly higher proportion of MSM (7.2%) had IHA titers of 128 or greater than patients from other risk groups (p = 0.006) ( Table 2).
The median interval between the two blood samples was 1054 days (interquartile range [IQR], 606-1857 days) (

Longitudinal survey of intestinal colonization with E. histolytica
Four hundred and sixty-nine (37.6%) patients, including 303 (64.6%) MSM, provided a total of 732 stool samples (range, 2-6 samples; median, 3 samples) for antigen testing between 2001 and 2005. At baseline, 45 (9.6%) patients, including 36 MSM (80%), had stool samples that tested positive for E. histolytica/E. dispar antigen. Two hundred and sixty-four patients, including 165 (62.5%) MSM, who had no intestinal infection with E. histolytica/E. dispar at baseline by specific stool antigen assays submitted more than one stool sample for repeat antigen assays (Table 3). At the  After an observation period of 399 PY, 19 (7.2%) patients were found to be infected with E. histolytica/E. dispar; 9 (47.4%) of the 19 isolates were E. histolytica by PCR. The median interval between the negative and positive antigen tests of the 19  The results from sequencing extracted DNA from the E. histolytica isolates in this study and other isolates of E. histolytica from our previous prevalence study [31] are shown in Figure 4. Case clustering among isolates from MSM was noted (Locus1-2 allelic genotypes B and D), suggesting a common source or person-toperson transmission. However, geographical unrelatedness among those patients with intestinal E. histolytca infection suggests that person-to-person transmission of E. histolytica might have occurred among MSM.
In univariate analysis, patients acquiring amebiasis were predominantly MSM and had significantly higher CD4 counts than those who remained uninfected (315 vs. 157 cells/L; Amebiasis and HIV Infection www.plosntds.org p,0.001) (data not shown). In multiple logistic regression analysis, we found that MSM was the only risk factor that was associated with new acquisition of E. histolytica infection by serologies or antigen assays followed by PCR, with an adjusted odds ratio of 14.809 (95% CI, 1.824, 120.237; p = 0.01) when compared with heterosexuals or patients with other risk behaviors (data not shown). The adjusted odds ratio for new acquisition of E. histolytica infection for every 50-cells/L CD4 increase or 12log 10 copies/ml plasma HIV RNA load decrease was 1.066 (95% CI, 0.975, 1.167; p = 0.16) and 1.180 (95% CI, 0.813, 1.713; p = 0.38), respectively.

Discussion
This is the first longitudinal follow-up study to investigate the incidence of E histolytica infection in HIV-infected patients by examining the incidence rate of intestinal E. histolytica infection and seroconversion of anti-E. histolytica antibodies. We found that HIVinfected MSM were at significantly higher risk for acquisition of E. histolytica infection [31,34]. Despite immunosuppression from HIV infection and the complicated disease course of IA, clinical responses to metronidazole therapy were favorable in terms of rapid defervescence and a low attributable mortality rate.
Exposure to E. histolytica, but not E. dispar, may induce anti-E. histolytica antibody response and development of anti-E. histolytica antibodies may represent either recent or remote exposure to E. histolyica [14,42], although not every person infected with E. histolytica develops an antibody response. Our analysis also showed that acquisition of E. histolytica infection was significantly associated with seroconversion for anti-E. histolytica antibodies despite immunosup-pression from HIV infection. Therefore, such a test may be used as a complimentary tool to understand the epidemiology of E. histolytica among high-risk populations. By using serological surveys, Japanese investigators have found that the seroprevalence of E. histolytica infection in MSM was as high as 13.4-20.4% compared with 1.0% in heterosexuals and 0.8% in prostitutes [24][25][26]33]. Similarly, we found that the seroprevalence of E. histolytica infection among HIVinfected patients remained significantly higher compared with HIVuninfected persons with gastrointestinal symptoms who had their sera tested for anti-E. histolytica antibodies [34].
In this study, we further explored the sero-incidence of E. histolytica infection in HIV-infected patients. The results also indicate that MSM are at increased risk of exposure to E. histolytica infection. The findings of higher seroprevalence and seroincidence of MSM, regardless of HIV status, are caused by higher prevalence and incidence of intestinal infection with E. histolytica. In our previous study, we found that the prevalence of E. histolytica/E. dispar by stool antigen tests was 12.1%, compared with 1.4% healthy controls; and at least 25% of the isolates from HIVinfected persons were confirmed as E. histolytica by PCR [31]. Although the majority of persons infected with E. histolytica are asymptomatic [1,2], more than 80% of the 600 or greater annually reported cases of amebiasis in Japan occurred in MSM [32]. These findings may reflect a decrease in E. histolytica infection in developed countries where improvement of public hygiene and sanitation has reduced the risk of acquisition of E. histolytica through contaminated water or food.
Sharing the identical transmission route with E. histolytica, the transmission of E. dispar among MSM in developed countries is  The clinical manifestations of IA in our HIV-infected patients with significant immunosuppression were similar to those previously described in HIV seronegative patients. Amebic colitis and liver abscesses were the two most common presentations [1,2,29]. The severity of the diseases was reflected by the high proportion of liver abscesses (59.7%) and complications (11.9%) in our cases. Despite low CD4 counts upon diagnosis with IA, the responses to metronidazole therapy with or without combination with antibiotics were favorable, as shown by rapid defervescence within 2 days of therapy initiation and no death attributable to IA.
There are several limitations of our study. First, the risk for exposure to E. histolytica is low in the general population in Taiwan, as reflected by the low seroprevalence (0.12%) of E. histolytica infection among 2500 healthy controls in a recent survey in northern and southern Taiwan [34]. Therefore, generalizations about our findings to areas of higher endemicity of E. histolytica and HIV infection should be cautious. Second, most patients at the late stage of HIV infection who develop HIV-related complications are referred to this hospital. However, those patients with IA had significantly higher CD4 counts than the patients without IA, suggesting that E. histolyica infection may not be associated with immunosuppression in HIV-infected patients. Rather, it is the risky behavior that increases risk of E. histolyica infection and subsequent development of invasive diseases. Third, our study was limited by the small sample size in assessment of the incidence of E. histolytica infection by stool antigen assays during follow-up. Although the incidence of E. histolytica infection is higher in MSM than in heterosexuals and others, the difference does not reach statistical significance. The shedding of E. histolytica may be intermittent, which may reduce the sensitivity of antigen assays if only one stool sample is tested. However, combinations with IHA assays for E. histolytica infection in our study may compensate for this deficiency by increasing the detection sensitivity. In this study, we chose a high titer of 128 as the cut-off value which decreases the possibility of cross-reactions, and seroconversion was significantly associated with newly acquired E. hisitolytica infection. Last, our genotyping methods [40,41] may not be as sensitive enough for detection of genetic differences between the isolates as the new genotyping system that uses 6 tRNA-linked short tandem repeats by Ali and colleagues [43].
In conclusion, HIV-infected MSM in Taiwan are at a higher risk of acquisition of E. histolytica infection and IA than other HIVinfected patients. It should also be investigated whether this is the case in other countries. Certainly physicians, treating MSM with or without HIV infection, should be aware of this potential complication, that until recently, in industrialized countries was seen nearly only in travelers returning from E. histolytica endemic regions.