Study sites

HIV positive (HIV+) individuals were recruited at the HIV Care and Treatment Centre (CTC) of HLH in Haydom, Mbulu district, Manyara region, northern Tanzania. HLH is a 429 bed-hospital serving around 15 400 inpatients and 70 150 outpatients per year [35]. As of December 2012, 1 898 HIV patients had sought care at the CTC clinic. All HIV/AIDS-related investigations and treatment are free of charge to the patient according to the Tanzanian National Policy on HIV/AIDS care and treatment. The villages of residence of the HIV+ patients were used to sample match HIV– individuals. These villages were located in the Mbulu, Hanang and Babati districts of the Manyara region and Iramba district of the Singida region (Fig. 1). All study districts were characterized by poor infrastructure and living conditions, lack of sanitation, and farming as the main source of income. In 2008, the Manyara region reported that 19.6% of rural households reared pigs with the largest number of pigs found in the Mbulu district (51 198 pigs, 53.1% of the pig population in the Manyara region), followed by the Babati district (19 587 pigs, 20.3%) and the Hanang district (15 310 pigs, 5.9%) [36]. The Iramba district reported 42 073 pigs, which corresponds to 86% of the total pig population in the Singida region [37].

According to the Tanzania HIV/AIDS and Malaria Indicator Survey (THMIS) 2011-12 [38], the overall HIV prevalence in Tanzania was estimated to be 5.1% among adults aged 15–49 years (6% among women, 4% among men) during our study period. In the same report, prevalence proportions of 1.5% for the Manyara region and 3.3% for the Singida region were estimated. It is believed that HIV emerged in these regions relatively late (mid 1990s), but has steadily increased ever since and already reached the country’s average in some districts like Babati [39].

Study design and study population

A cross-sectional study was conducted among 170 HIV+ and 170 age (+/− 10 years), gender and village of residence-matched HIV– individuals. The rationale for matching was to limit variation in exposure to T. solium eggs among HIV+ and HIV– participants. Participants were recruited from January 2011 to February 2013. Every day, HIV+ patients attending CTC for regular clinical check-ups and/or HAART were invited to participate in the study. Age, gender and village of residence of each consenting HIV+ patient were entered into an Excel sheet on a daily basis. For the age range of the individual matching partner a range of +/− 10% (with minimum and maximum target age) was calculated. Following this list, a field team including one nurse from CTC and one research assistant went each day to the appropriate villages to identify matching HIV– participants. With the help of village headmen/headwomen and local helpers, potential candidates were visited at their homes and asked to voluntarily participate in this study. The overall recruitment was stopped when a total of 433 participants, resulting in 170 matching pairs, had been identified (Fig. 2). Initial HIV testing of individuals enrolled in CTC was performed using two rapid Ab-tests according to the Tanzanian national algorithm for HIV testing: Alere Determine HIV-1/2 (Abbott laboratories, Abbott Park, IL, USA) and Uni-Gold™ Recombigen® HIV-1/2 (Trinity Biotech, USA). In the case of a positive reading in the Alere Determine HIV-1/2 test the serum sample was subjected to the second rapid Ab-test: Uni-Gold™ Recombigen® HIV-1/2 test. Concordant positive results were interpreted as positive for HIV-Ab. Discordant results were interpreted as inconclusive and the sample was sent to the regional hospital for confirmatory test using two ELISA tests: Enzygnost anti-HIV 1 + 2 Plus ELISA (Behring, Marburg, Germany) and Well-coenzyme HIV recombinant ELISA (Murex, Dartford, England). In this study, HIV– individuals were defined as participants who tested negative using the Alere Determine HIV-1/2 test and HIV+ individuals as participants tested positive following the algorithm described above. Clinical staging of HIV/AIDS was performed according to the revised World Health Organization (WHO) guidelines and HAART treatment status of all HIV+ was noted [40]. Most HIV+ participants (164 or 96%) were receiving HAART and none were under epilepsy or anti-inflammatory treatment at the time of recruitment. For ethical reasons, only individuals aged at least 9 years old were included in the study. Pregnant women were excluded from the study.

Consenting individuals were invited to answer a questionnaire, to undertake a clinical examination and to provide a blood sample for serological testing for taeniosis and CC. Neuroimaging was offered to all HIV+ participants and to only those HIV– participants positive to TSOL serology for ethical reasons.

Blood sample collection and laboratory testing

After counselling by a specially trained nurse, 10 ml of whole blood was collected by venipuncture from each participant. HIV+ and HIV– individuals were informed about the HIV test results immediately thereafter. CD4⁺ counts were determined for all HIV+ patients using a fluorescent activated cell sorter machine (FACScount™, BD Biosciences, US) at the central laboratory of HLH. Blood samples from HIV– individuals were transported in cool boxes to the laboratory and immediately placed into refrigerators until required for further processing. Blood from HIV+ individuals was refrigerated in the HLH central laboratory immediately after drawing. After 1 to 6 h, serum was separated and stored at −20 °C until serological testing for TSOL was performed. On average, two vials with 2 ml of serum were obtained from each individual. Four to twelve months later, samples were shipped to the CDC in Atlanta, USA, for further analyses and one safety back up sample remained at the Parasitology Laboratory of MUHAS to avoid total loss in case of any damage to the sample during the long shipment. Two tests were used to detect the presence of Ab to CC (LLGP-EITB and rT24H-immunoblot), one test was used to detect the presence of antigen (Ag) to CC (Ag-ELISA) and one test was used to detect the presence of Ab to taeniosis (rES33-immunoblot). LLGP-EITB is an enzyme-linked immunoelectrotransfer blot that detects CC-specific Ab to any of seven glycoproteins [41, 42]. The rT24H-test is an immunoblot method that detects CC-specific Ab to a T. solium recombinant Ag [43]. The Ag-ELISA that detects Taenia-Ag in serum is a monoclonal Ab (B158/B60 Ab) based capture ELISA and was performed following the modified protocol described by Dorny et al. [44]. The rES33-test is an immunoblot method that detects adult T. solium specific Ab using a recombinant protein derived from the excretory-secretory proteins of the adult tapeworms [45, 46]. Ab to both of the recombinant peptides, rT24H and rES33, were assessed in the same test. All test results were interpreted independently and blinded by two scientists with many years of experience in reading these T. solium in-house tests. For quality assurance each positive test was repeated at least once as well as 15% randomly selected negative samples.

Definitions

T. solium positive individuals (TSOL+) were defined as those with a positive result in at least one of the four serological tests described above while those negative in all tests were considered T. solium negative (TSOL–). Positive CC cases were defined as those individuals positive to the LLGP-EITB, rT24H-immunoblot or Ag-ELISA test. Taeniosis cases were defined as individuals positive to the rES33-immunoblot test. NCC cases were defined following the revised criteria of Del Brutto [10]. Viable cysts (active NCC) were defined as cystic lesions (with or without visible scolex) or lesions with ring enhancement. Calcified cysts (inactive NCC) were defined as hyperdense lesions with no sign of ring enhancement [47].

Neuroimaging

All HIV+ individuals underwent CT examination at baseline in a Toshiba Auklet Single Slice Spiral CT machine at HLH. HIV–/TSOL+ participants received their CT examination at the Aga Khan Health Centre in Arusha since the CT machine at HLH did not work at the time of the follow-up examinations. All follow-up CT scans of HIV+/TSOL+ were also performed in Arusha for the same reason. This health center used a Siemens Somatom Emotion CT machine. Both machines provided 5 mm slices and took 64 slices per picture. Intravenous contrast medium was used in all patients. CT scan images were evaluated by the local radiologists and sent to the Department of Neurology at the Technical University Munich (TUM) for a second blinded evaluation by a neurologist specialized in NCC diagnoses.

Research questionnaire

Each participant was asked to answer a socio-demographic and a medical history interview questionnaire. Interviews with HIV+ individuals were performed by a clinical officer at the CTC clinic and with HIV– individuals by a trained nurse in their villages of residence. The study nurses delivered the questionnaires at the CTC clinic. The questionnaire was adapted from a previous validated questionnaire used by the Cysticercosis Working Group in Eastern and Southern Africa (CWGESA) and addressed socio-demography, pork consumption habits as well as hygiene and sanitary practices [48]. Questions with an emphasis on self-reported past and present neurological disorders, such as acute and chronic headaches, peripheral nervous system (PNS) and CNS signs and symptoms including epileptic seizures as well as psychiatric disorders were included; so were questions on family history of neurological disease. HIV+ individuals were also asked about their past HIV history including opportunistic infections, HAART duration as well as compliance, amongst others. The questionnaire was developed in English, translated into Swahili and back-translated to English by two independent people. A standard neurological examination was performed on all HIV+ individuals by a medical doctor specializing in neurology.

Follow-up of seropositive individuals

Ten to fourteen months elapsed between sampling and receipt of the serological tests results due to logistical reasons and challenges in tracing patients, therefore HIV+/TSOL+ participants were invited for a second CT scan and blood collection (10 ml from each individual) so that appropriate treatment could be provided. This second clinical examination and follow-up CT scan were performed shortly before treatment and blood samples were taken at HLH. Participants identified HIV–/TSOL+ were invited for their first CT scan and neurological examination, and second blood drawing before treatment. Taeniosis positive participants were treated with niclosamide and symptomatic NCC cases with albendazole and dexamethasone. Present CD4⁺ counts of HIV+ individuals were obtained and all serum samples were again shipped to the CDC in Atlanta for testing as described above. In addition, before treatment, TSOL+ participants were asked to provide three stool samples for copro-Ag-ELISA testing at the CDC, following a protocol described by Guezala et al. [49]. Due to financial limitations in our study copro-Ag testing could only be performed during follow-up.

Statistical analysis

Analysis of associations between potential risk factors including HIV (independent variables) and TSOL (dependent variable) was performed by bivariate approximative tests (χ ²-tests) and exact tests (Fisher’s tests) using Epi Info, version 3.3.2 (CDC, Atlanta, GA). Significant differences were defined as p-values below 0.05. McNemar tests were used for the paired analysis associations between the HIV status and TSOL, as well as for determining all the neurological outcomes. Matched prevalence proportion ratios (mPPR) and their 95%CI were obtained using the csmatch command in Stata 14.0. Statistical significance of associations between HIV and other factors among all matched and unmatched recruited participants and among the HIV positive group were evaluated using the chi-square test for categorical variables and the Kruskal Wallis tests for continuous variables. Matched analyses were performed using Stata 14.0.

Demographic characteristics of HIV+ and HIV– individuals of the matched study population

Initially, 217 HIV+ individuals at CTC and 216 HIV– individuals were recruited to participate in the study. The initial aim of this recruitment was to identify a minimum of 150 matched pairs of HIV+ and HIV– individuals. After the initial recruitment, 40 HIV+ and three HIV– individuals were excluded due to a loss of follow-up or because the individual refused to participate further in the study. Full data sets comprised a completed questionnaire, serological results as well as at least one cranial CT examination and, for HIV+ participants, a neurological examination. After applying the matching criteria, seven additional HIV+ individuals and 43 HIV– individuals were excluded from the group of 177 HIV+ and 213 HIV– individuals, resulting in 170 matching HIV+/HIV– pairs available for analysis (Fig. 2).

HIV related characteristics

At the time of recruitment 161 of 170 HIV+ individuals were on regular HAART therapy, three had discontinued treatment during recent years at least once, and six had not yet started HAART. The most commonly used HAART drugs were the first-line combinations lamivudine/zidovudine or lamivudine/stavudine/nevirapine (76.5%). 46 (27.1%) HIV+ patients reported opportunistic infections in the past year, mostly pulmonary tuberculosis and other respiratory diseases as well as herpes zoster and other infectious skin diseases. The distribution of HIV/AIDS stages at the time of recruitment was: stage I: 71.8%, stage II: 9.4%, stage III: 16.5%, and stage IV: 2.4%.

Association between HIV and relevant past medical/neurological history, current neurological deficits, and TSOL risk factors among the matched pairs

Among the 164 pairs with this information available, HIV+ participants reported consuming pork significantly less often than HIV– participants (mPPR: 0.85; 95%CI: 0.75–0.96), but most of those consuming pork reported to eat pork less than once a month and only properly cooked. Only 1.2% of all study participants reported ever seeing cysts in pork. Of all matched study participants, 97.9% reported having access to latrines at home, and 98.2% of those reported always using the latrine. Only 9.4% of the participants reported washing hands before food consumption; the HIV+ participants tended to wash their hands less often than the matched HIV– participants (mPPR: 0.55; 95%CI: 0.26–1.15) (Tables 2 and 3).

There were six pairs where the HIV+ individual reported living in a household with a member with a past history of tapeworm infection, while the HIV– matched participant did not; in only two pairs did the HIV– participant reported such history and the HIV+ did not (mPPR: 2.33; 95%CI: 0.70–7.82). Fewer HIV+ participants tended to report anthelminthic treatment in the past than the matched HIV– participants (mPPR: 0.57; 95%CI: 0.30–1.05), but the difference was not significant (Tables 2 and 3).

Distribution of TSOL tests results among the matched pairs and in the initially recruited population

There was no association between HIV and TSOL in this population with four discordant pairs each of HIV+/TSOL+ with HIV–/TSOL– and HIV+/TSOL– with HIV–/TSOL+ (mPPR: 1.0; 95%CI: 0.25–4.00). This corresponds to an overall prevalence of 2.4% (95%CI: 0.6–5.9) of TSOL among both HIV+ and HIV– participants. Comparing detailed test results, no significant differences were observed between matched HIV+ and HIV– individuals regarding the sero-prevalence of taeniosis-Ab (mPPR: 0.5; 95%CI: 0.05–5.51), CC-Ab (mPPR: 1.0; 95%CI: 0.25–4.00), and CC-Ag (mPPR undefined because there was only one discordant pair with HIV+ positive to the Ag-ELISA and HIV– negative to Ag-ELISA) were detected (Fig. 2, Tables 2 and 3).

Of the 404 participants initially recruited with sera available, which include the 340 matched participants, the prevalence of CC-Ab was 2.5% (95%CI: 1.2–4.5). For more clinical, serological and risk factor data of those 10 TSOL+ individuals refer to Additional file 2: Table S1. The prevalence of Ab to taeniosis was 1.2% (95%CI: 0.4–2.9). All five of 404 rES33 positive individuals also had Ab to CC-Ag or positive Ag-results. Only three (0.7%; 95%CI: 0.4–2.9) participants were positive for the presence of CC-Ag. The prevalence of TSOL was slightly higher among those presenting with headaches or seizures (3.7%; 95%CI: 0.8–10.4) than among those without such symptoms (2.2%, 95%CI: 0.9–4.4), but this difference was not statistically significant (Tables 2 and 3).

The district of origin and serological results were available for 399 out of 433 initially recruited individuals. The prevalence proportion of TSOL was considerably higher among participants living in the Mbulu district (5.2%, 95%CI: 1.9–10.4) than in other districts (0.8%; 95%CI: 0.1–2.8) (Fig. 1), a difference which was statistically significant (p = 0.008) (Table 1).

Association between CD4⁺ counts, HIV stages, HAART and TSOL

Radiological and serological results of TSOL+ individuals among the matched pairs and initially recruited population

Details on the four HIV+ patients (P1-P4) and the four HIV– patients (P5-P8) who were also TSOL+ in addition to two HIV– participants (P9-P10) who could not be matched but were TSOL+ (initially recruited study population; Fig. 2) are provided in Table 4. More clinical details on these ten patients are shown in Additional file 2: Table S1.

Overall, all four HIV+/TSOL+ and three of four HIV–/TSOL+ participants of the matched study population agreed to have at least one CT scan at the time of their first and second clinical examination. The two unmatched HIV–/TSOL+ individuals (P9, P10) received a CT scan at their first clinical examination due to the severe clinical condition they were in (epilepsy, encephalopathy). None of the 166 HIV+/TSOL– participants presented with brain lesions compatible with NCC, but three (P1, P3 and P4) of the four HIV+/TSOL+ had definitive NCC. The fourth participant did not show any lesions in the brain (P2). This would suggest an overall specificity of 99.4% (95%CI: 96.7–100) of the CC tests (Ag-ELISA, LLGP-EITB, rT24H-immunoblot) to detect NCC among the HIV+ patients in this population. There were too few cases to estimate the sensitivity with any precision. Overall, there was perfect agreement between the EITB and rT24H for the detection of Ab to CC (Tables 3 and 4).

Among the three matched HIV–/TSOL+ participants with CT scan results, two were found to have definitive NCC (P5, P7) and one other had probable NCC (P6). The latter presented with an obstructive hydrocephalus and a mass in the 4^th ventricle. Among all HIV– participants positive to TSOL and receiving neuroimaging, four out of five had lesions of NCC, suggesting that, although there were only small numbers, HAART may not impact on the proportion of NCC among those positive for CC in people with HIV/AIDS. The proportion of NCC was therefore 1.8% among the matched HIV+ participants, but cannot be estimated in the matched HIV– participants since only three had a CT scan.

Among the seven NCC cases, five presented with calcified lesions without perilesional edema and were free of any viable cysts. Among the two unmatched HIV–/NCC participants, one was found to have multiple viable cysts in the parenchyma, some of them with perilesional edema (P10), and the other had several calcified cysts (P9). The number of detected lesions per NCC case in this study ranged from one to ten, except in P10, where more than 20 active lesions could be identified. No lesions outside the parenchyma were detected (Table 4).

Description of symptoms among patients with NCC

Four (1 HIV+, 3 HIV–) of eight participants with definitive or probable NCC presented with neurological symptoms. Two of the symptomatic participants presented with headaches (1 HIV+ with acute and 1 HIV– with chronic headaches) and two HIV– individuals with epilepsy. One HIV– participant (P10), a boy aged 14 years, was unconscious at the time of admission to HLH with signs of brainstem involvement and generalized increased muscle tone leading to flexion contractures of all four limbs, most likely in the context of increased intracranial pressure. P10 was the only participant with positive serological results for all tests (CC-Ab, CC-Ag and taeniosis-Ab). The CD4⁺ counts of the only symptomatic (presenting with acute headaches) HIV+ individual (P4) was 170 cells/μl, the lowest level of the four HIV+ individuals who were positive to TSOL (Table 4).

Follow-up of TSOL+ individuals

Of the ten participants found positive to TSOL at baseline, three were lost to follow-up, including one unmatched HIV– participant who died for unclear reasons (strong headaches and abdominal pain reported before death). Five (4 HIV+ and 1 HIV–) of the seven followed participants provided sera and six (4 HIV+ and 2 HIV–) provided a stool sample. Seven (4 HIV+ and 3 HIV–) participants had a stool examination at the time of follow up, but all of them had a negative copro-Ag-ELISA test result. The only HIV+ participant positive to rES33 at baseline was still positive at follow-up. Among the five participants positive to CC at baseline, one HIV+ participant became sero-negative to the tests detecting Ab. Interestingly, this participant (P3) had an increase of 92 cells/μl in CD4⁺ (from 276 to 368 cells/μl; increase of 33% from baseline), in contrast to the three other HIV+ participants for whom the CD4⁺ counts decreased slightly or remained similar. Only three participants accepted to undergo a second CT examination. Two of them did not show any changes, whereas P10, who had shown several viable cysts on the first CT scan, revealed mostly calcifications as well as a few viable cysts. This young (14-year-old) patient was hospitalized and, after serological and radiological confirmation, received anti-inflammatory (dexamethasone) and anthelminthic (albendazole, niclosamide) treatments repeatedly during the whole study period. There were no major improvements in the symptoms after treatment and the patient was transferred to a home for children with special needs after the study. Overall, there was no change in the clinical presentation of all the TSOL+ individuals at follow-up. None of the asymptomatic NCC cases reported new clinical signs/symptoms (Table 4, Additional file 2: Table S1).