Cerebral Toxoplasmosis Mimicking Subacute Meningitis in HIV-Infected Patients; a Cohort Study from Indonesia

Background HIV-associated subacute meningitis is mostly caused by tuberculosis or cryptococcosis, but often no etiology can be established. In the absence of CT or MRI of the brain, toxoplasmosis is generally not considered as part of the differential diagnosis. Methodology/Principal Findings We performed cerebrospinal fluid real time PCR and serological testing for Toxoplasma gondii in archived samples from a well-characterized cohort of 64 HIV-infected patients presenting with subacute meningitis in a referral hospital in Indonesia. Neuroradiology was only available for 6 patients. At time of presentation, patients mostly had newly diagnosed and advanced HIV infection (median CD4 count 22 cells/mL), with only 17.2% taking ART, and 9.4% PJP-prophylaxis. CSF PCR for T. Gondii was positive in 21 patients (32.8%). Circulating toxoplasma IgG was present in 77.2% of patients tested, including all in whom the PCR of CSF was positive for T. Gondii. Clinically, in the absence of neuroradiology, toxoplasmosis was difficult to distinguish from tuberculosis or cryptococcal meningitis, although CSF abnormalities were less pronounced. Mortality among patients with a positive CSF T. Gondii PCR was 81%, 2.16-fold higher (95% CI 1.04–4.47) compared to those with a negative PCR. Conclusions/Significance Toxoplasmosis should be considered in HIV-infected patients with clinically suspected subacute meningitis in settings where neuroradiology is not available.


Introduction
In settings of Africa and Asia, the most common cause of subacute meningitis in patients with advanced HIV infection is either tuberculous or cryptococcal infection [1,2]. However, in many patients, the etiology of subacute meningitis cannot be established [1,3]. In line with a large retrospective cohort of adult meningitis patients in South Africa, where 52.8% had no definite diagnosis despite extensive microbiological testing [1], we could not identify the causative pathogen in 48.9% of HIV-infected meningitis patients in an Indonesian setting [4].
Toxoplasmosis is a common and serious central nervous system (CNS) infection in patients with advanced HIV infection [5][6][7][8], although its incidence has decreased with introduction of antiretroviral treatment (ART) [6,9]. Cerebral toxoplasmosis mostly presents as cerebral mass lesions with headache, confusion, fever, lethargy, seizures, cranial nerve palsies, psychomotor changes, hemiparesis and/or ataxia [10]. Some of these symptoms may also mimic meningitis, but cerebral toxoplasmosis is generally not considered as a differential diagnosis of subacute meningitis in HIV-infected patients. This is especially the case in low-resource settings where no CT or MRI can be performed. We have therefore examined if toxoplasmosis can be diagnosed in HIVinfected patients presenting with subacute meningitis of unknown origin in Indonesia, using cerebrospinal fluid (CSF) PCR for T. gondii.

Ethics statement
Anonymized CSF and blood samples were used from an already-existing hospital collection, from a cohort of patients collected as part of a project 'Optimization of diagnosis of meningitis', approved by the Ethical Committee of Hasan Sadikin Hospital/Medical Faculty of Universitas Padjadjaran, Bandung, Indonesia (No. 85/FKUP-RSHS/KEPK/Kep/EC/2006). As this study was done using already existing sample collection, no separate consent was asked for this study. HIV testing is done routinely with oral informed consent for all patients with suspected meningitis in Hasan Sadikin hospital, after 24% were found HIVpositive in a previous cohort study of 185 patients in the same hospital [4]. Consent is obtained from closest relatives (husband/ wife or parents) for those patients who are unstable or unconscious at time of presentation. With approval from the ethical committee HIV testing was done anonymously afterwards for those who had died before consent could be obtained.

Setting and patients
We included adult patients presenting with suspected meningitis at Hasan Sadikin Hospital, the top referral hospital for West Java Province, Indonesia, between December 2006 and October 2010. Clinical data including outcome was recorded in individual case report form. Definite TB meningitis was diagnosed if CSF culture or real time PCR were positive for M. tuberculosis, cryptococcal meningitis if either CSF India Ink examination, culture or cryptococcal antigen testing were positive, and toxoplasmosis if CSF T. gondii PCR was positive. HIV testing is done routinely for patients presenting at this hospital, but cerebral CT-scanning is rarely done in this setting and is not covered by the government health insurance for the poor.

Laboratory examinations
CSF cell count and differentiation, protein and glucose were measured. CSF microscopy was done for cryptococci, acid-fast bacilli and bacterial pathogens. CSF was cultured for Mycobacterium tuberculosis (solid Ogawa and liquid MB-BacT, Biomerieux), bacterial pathogens (blood agar, chocolate agar, and brain-heart infusion) and fungi (Sabouraud). Cryptococcal antigen (CALAS, Meridian Diagnostics) testing was done on CSF samples following the manufacturer's instructions. Five to 7 ml CSF samples were used for molecular testing. After centrifugation of CSF samples at 30006g for 10 minutes, DNA was extracted from 200 ml of CSF sediment by using QIAmp DNA mini kit (Qiagen, USA). CSF M. tuberculosis real time PCR was done using IS6110, a repeated insertion sequence specific for M. tuberculosis, as a target [11].
Measurement of CD4-cell count for HIV-patients only became available during the time of the study and was measured only for those who survived for more than 4 days. Real time PCR for T. gondii, using the multicopy B1 gene of the T. gondii as the target as described elsewhere [12], was performed to archived CSF samples at Radboud University Nijmegen Medical Centre. CSF specimens from 22 HIV-negative meningitis patients (16 with definite TB meningitis, 2 with bacterial meningitis, and 4 with no definite diagnosis), and nine patients with non-infectious CNS diseases, all recruited at Hasan Sadikin Hospital, were used as controls for T. gondii PCR. These samples were collected during the study period over a similar time scale compared to the case CSF samples. Toxoplasma immunoglobuline G (toxoplasma IgG) were measured by electro chemiluminescent assay (ECLIA, Elecsys, Roche) in archived serum samples of patients included in the study.

Data analysis and statistics
Characteristics of patients with definite tuberculosis, cryptococcosis and toxoplasmosis were compared using Chi-square test for proportions and Mann-Whitney U test for continuous variables. Progression to death using 2-month mortality data was examined by Kaplan-Meier estimates.

Results
During the period, 401 patients presented with clinical meningitis, 76 were diagnosed with HIV infection, and 64 had archived CSF samples and were included in this study. Patients included in the study presented after a median 7 days, with meningismus (86.0%), headache (80.8%), lowered consciousness (33.3%), fever (28.8%), hemi-or tetraparesis (28.6%), cranial nerve palsies (12.5%), and seizures (10.9%). HIV was newly diagnosed in 53 patients (82.8%). All 11 patients previously diagnosed with HIV were taking ART, and 6 were using co-trimoxazole as Neck stiffness, headache and fever, the classical signs of meningitis, were equally common in patients diagnosed with toxoplasmosis, cryptococcosis and tuberculosis, as were most other signs and symptoms, except hemiparesis ( Table 1). None of the patients with toxoplasmosis had received ART or co-trimoxazole prophylaxis prior to admission with meningitis. CT scans were available for 6 patients, including 4 with a positive T. gondii PCR. Three showed signs of hydrocephalus, one a hypodense lesion that showed no enhancement using contrast, and two were normal. No mass lesions typical for cerebral toxoplasmosis were seen.
CSF cell count and protein were normal or mildly elevated in patients with toxoplasmosis, and hypoglycorrhachia was less common compared with tuberculosis or cryptococcosis

Author Summary
If HIV-infected patients present with seizures, focal neurological symptoms or confusion, a CT-scan or MRI of the brain is normally made. If mass lesions are found (and the CD4 cell count is sufficiently low), cerebral toxoplasmosis is suspected, and often treated empirically. However, some of the symptoms of cerebral toxoplasmosis may mimic those of subacute meningitis. Therefore, in settings where no cerebral imaging can be performed, HIVassociated cerebral toxoplasmosis may be under-diagnosed. We retrospectively looked for toxoplasmosis in a cohort of HIV-infected patients presenting with subacute meningitis in an Indonesian hospital, where neuroradiology was not available for most patients. Patients mostly came with newly diagnosed and advanced HIV infection and few were on HIV-treatment or PJP-prophylaxis. Molecular testing of cerebrospinal fluid (CSF) was positive for Toxoplasma gondii in 32% of patients, serology was positive in 78%. Clinically, in the absence of neuroradiology, toxoplasmosis was difficult to distinguish from tuberculosis or cryptococcal meningitis. A positive CSF T. gondii PCR was associated with a two-fold increased mortality. We conclude that toxoplasmosis should be considered in HIV-infected patients with clinically suspected subacute meningitis in settings where neuroradiology is not available.
( Table 1). CD4 counts, missing in two-thirds of patients due to early death or the unavailability of CD4 cell testing during the initial phase of the cohort study, were low in all but one patient. Table 2 lists the CSF findings of individual patients, Figure 1 is a graphic representation of the CSF cell count, protein and glucose ratio, showing the overlap in CSF findings   Patients with confirmed cryptococcosis received amphotericine B, followed by fluconazole; all others received empiric tuberculosis treatment combined with adjunctive corticosteroids [13]. No toxoplasmosis treatment was given, as T. gondii PCR was performed retrospectively and was not available at time of presentation. Eight patients were lost to follow up and were not included in Kaplan Meier analysis. Mortality among those with positive CSF T. gondii PCR was 2.16-fold (95% CI 1.04-4.47) higher compared to those who had a negative PCR result; median survival was 7 days for toxoplasmosis, 7 days for tuberculosis meningitis, 110 days for cryptococcosis, and 32 days for patients with an unknown cause of meningitis ( Figure 2).

Discussion
In our cohort of HIV-infected patients presenting with clinical signs and symptoms of CNS infection, CSF T. gondii PCR was positive in 32.8% of patients, sometimes in conjunction with tuberculosis. In the absence of CT or MRI of the brain, toxoplasmosis could not be distinguished from tuberculosis or cryptococcosis. Mortality in this cohort of newly diagnosed and advanced HIV infection was extremely high and associated with a positive T. gondii PCR.
Cerebral toxoplasmosis typically causes space occupying lesion(s), leading to subacute or acutely developing confusion, with or without focal neurological deficits [14]. In the absence of CT or MRI of the brain, common findings like headache, fever, hemiparesis and decreased level of consciousness [10] may mimic those of meningitis [4,[15][16][17]. In previous series of cerebral toxoplasmosis [18,19], meningeal signs have been reported in 3 to 16% of the patients, although in many reports neck stiffness is not mentioned [14]. Although rare, cases of spinal cord toxoplasmosis have also been reported [20]. No typical mass lesions were found in 6 patients with an available CT scan. This is not surprising, as this study depended on the availability of CSF samples, that would not have been obtained if typical mass lesions had been found.
We used T. gondii PCR for diagnosis of cerebral toxoplasmosis. In previous studies CSF T. gondii PCR had a sensitivity of 50-60% to confirm cerebral toxoplasmosis in HIV-infected patients [21,22]. The sensitivity is possibly higher among patients with meningoencephalitis compared to those with space-occupying lesions only, but this has not been examined. Specificity of T. gondii  PCR is high, between 97 and 100% [22][23][24]. The positivity rate of 32.8% in our study might therefore be an underestimate, especially in the category of patients in whom no other pathogen was isolated despite extensive microbiological testing. In our cohort, toxoplasmosis could not be distinguished clinically from tuberculosis and cryptococcosis. From our previous series [4], CSF samples were available for the current study for 36/47 HIV-infected patients. Ten out of 17 patients who were diagnosed with 'probable TB meningitis' and 'unknown' in the previous study were found to have a positive T. gondii PCR (and no bacteriological confirmation of tuberculosis) in the current study. Diagnosis of cerebral toxoplasmosis is usually based on clinical findings and CT or MRI of the brain. However, if cerebral imaging is lacking, toxoplasmosis may not be considered. Positive toxoplasma serology, which has a high sensitivity but very poor specificity, is helpful to exclude but not to confirm cerebral toxoplasmosis, although some reports suggest that high toxoplasma IgG titers are only found in patients with symptomatic toxoplasmosis [25]. Indeed, in our study, patients with a positive T. gondii PCR had a higher IgG titers compared to those who had a negative PCR. An autopsy study from India provides further support for the notion that cerebral toxoplasmosis is not always considered; among 233 HIV patients, toxoplasmosis accounted for 6.8% of deaths, but in none of these cases toxoplasmosis had been suspected clinically [26].
The incidence of cerebral toxoplasmosis varies between countries [14] and is related to the seroprevalence of toxoplasmosis in the general population [19,25]. In the United States, toxoplasma seroprevalence varies from 3% to 30%, whereas in France 73%-90% of the population is infected [10]. Reported seroprevalence rates were varied from 13-31% in the general population, and 45-68% in HIV patients in studies from several developing countries [8,27,28]. In our study, 78% of patients had detectable toxoplasma IgG, but this does not reflect the seroprevalence in the general population or among unselected HIV-infected patients, as only meningitis patients were examined.
Mortality in this cohort of patients was very high, higher compared to reported rates in other series [8,9,29]. One explanation is that patients mostly presented with advanced and untreated HIV infection. In addition, no toxoplasmosis treatment was provided, as PCR was done retrospectively on archived samples. In our previous study, HIV infection was associated with a 2.5-fold increased mortality among patients presenting with meningitis [4]. Data from the current study suggests that this may is at least in part attributable to a high prevalence of (unrecognized and untreated) toxoplasmosis. Future studies should examine the benefit of timely diagnosis and/or empiric treatment of toxoplasmosis for patients in settings like ours. Empiric treatment for subacute meningitis in HIV-infected patients should probably also include tuberculosis, which is difficult to exclude as culture is slow and microscopy and commercial PCR assays have insufficient sensitivity [30].
Our study suffers from several limitations. Most importantly, no CT or MRI of the brain could be performed. In addition, clinical data, CD4 cell counts and other laboratory parameters were missing in a number of patients. Despite these limitations the data strongly suggest that toxoplasmosis should be included in the differential diagnosis of HIV-infected with clinically suspected subacute meningitis, and that molecular testing or empiric treatment for toxoplasmosis should be considered in these patients, especially if no CT or MRI can be performed. Obviously, timely diagnosis and treatment of HIV will help prevent this severe opportunistic infection.

Supporting Information
Checklist S1 STROBE checklist for the manuscript: ''Cerebral toxoplasmosis mimicking subacute meningitis in HIV-infected patients; a cohort study from Indonesia''. (DOC)