Cryptococcal antigen prevalence in HIV-infected Tanzanians: a cross-sectional study and evaluation of a point-of-care lateral flow assay
Cryptococcal antigen (CRAG) screening at antiretroviral therapy (ART) initiation and pre-emptive antifungal treatment for those testing positive could prevent many cases of cryptococcal meningitis (CM). To investigate whether CRAG screening would be feasible in Tanzania, we conducted a cross-sectional study measuring CRAG prevalence in ART clinic patients and comparing the novel lateral flow assay (LFA) with the cryptococcal latex agglutination (LA) test.
Consecutive HIV-infected outpatients with CD4 counts <200 cells/µL, who were ART naive or had been on ART for <6 months, were screened for CRAG using the LA and LFA kits. For further assay validation, HIV-infected inpatients with suspected cryptococcal disease were also tested using the LA and LFA kits.
Cryptococcal antigen was detected in seven of 218 ART clinic attendees (3%). Six patients (5%) with CD4 cell counts ≤100 cells/µL (n = 124) were CRAG-positive. Agreement between the LA and LFA test in the 218 outpatients was 100%. Another 101 inpatients were tested for CRAG, of whom 56 (55%) were CRAG-positive on both the LA and LFA tests. One patient was positive using the LFA test but negative on the LA test. The overall agreement between the two assays was 99.7%, kappa coefficient 0.99 (standard error 0.06, P < 0.001).
Five percentage of ART clinic patients with CD4 cell counts ≤100 cells/µL in northern Tanzania had asymptomatic cryptococcal antigenaemia, suggesting that CRAG screening would be worthwhile in the Tanzanian ART programme. The LFA is a reliable, cheap and practical alternative to LA for detection of CRAG.
Le dépistage de l'antigène cryptococcique (CRAG) à l'initiation du traitement antirétroviral (ART) et du traitement préventif antifongique pour ceux dont le test se révèle positif pourrait prévenir de nombreux cas de méningite cryptococcique. Afin de déterminer si le dépistage du CRAG est réalisable en Tanzanie, nous avons mené une étude transversale mesurant la prévalence du CRAG chez les patients des cliniques ART et en comparant le nouveau test de l’écoulement latéral (LFA) à celui de l'agglutination du latex (LA) pour la détection de la cryptococcose.
Des patients ambulatoires consécutifs infectés par le VIH avec des taux de CD4 < 200 cellules/μl, naïfs pour l’ART ou qui étaient sous ART depuis moins de 6 mois, ont été dépistés pour CRAG à l'aide des kits LFA et LA. Pour de plus amples validation des tests, les patients infectés par le VIH et hospitalisés avec une suspicion de cryptococcose ont également été testés en utilisant les kits LFA et LA.
CRAG a été détecté chez 7/218 (3%) visiteurs de la clinique ART. Six patients (5%) avec une numération des CD4 ≤ 100 cellules/μl (n = 124) étaient positifs pour CRAG. La concordance entre les tests LFA et LA chez les 218 patients ambulatoires était de 100%. Un autre ensemble de 101 patients hospitalisés a été testé pour CRAG, dont 56 (55%) se sont révélés positifs à la fois pour LFA et LA. Un patient a été positif pour le test LFA mais négatif pour le test LA. La concordance globale entre les deux tests était de 99,7%, coefficient kappa: 0,99 (erreur standard: 0,06, P < 0,001).
5% des patients de la clinique de ART dont la numération des CD4 ≤ 100 cellules/μl dans le nord de la Tanzanie avaient une antigénémie cryptococcique asymptomatique, ce qui suggère que le dépistage du CRAG serait utile dans le programme ART tanzanien. Le test LFA est une alternative au test LA, fiable, bon marché et pratique pour la détection du CRAG.
La detección del antígeno criptocócico (AgCr) al comienzo de la terapia antirretroviral (TAR) y el tratamiento antifúngico preventivo para aquellos que dan positivo, podría prevenir en muchos casos la meningitis criptocócica (MC). Con el fin de investigar si el despistaje del AgCr sería factible en Tanzania, hemos realizado un estudio croseccional para medir la prevalencia del AgCr en pacientes de una clínica de TAR y comparado la nueva prueba de flujo lateral (PFL) con la prueba de aglutinación en latex (PAL) para criptococo.
Se realizó el despistaje del AgCr en pacientes consecutivos de consultas externas infectados con VIH y conteos de células CD4 < 200 células/ml, que no habían recibido TAR o llevaban menos de 6 meses en tratamiento. Se utilizaron kits para las pruebas de aglutinación en latex y de flujo lateral. Para una validación extra de las pruebas, también se probaron los kits de PAL y PFL con pacientes ingresados, infectados con VIH, en los que había sospecha de infección por criptococo.
Se detectó AgCr en 7 de 218 pacientes de la clínica de TAR (3%). Seis pacientes (5%) con conteos de células CD4 ≤ 100 células/ml (n = 124) eran positivos para AgCr. La concordancia entre la PAL y la PFL en los 218 pacientes de consultas externas era del 100%. A otros 101 pacientes ingresados se les realizaron las pruebas para AgCr, y de estos 56 (55%) dieron positivo, tanto mediante PAL como PFL. Un paciente dio positivo utilizando la PFL pero negativo con la PAL. La concordancia entre los dos ensayos era del 99.7%, coeficiente de kappa de 0.99 (error estándar 0.06, P < 0.001).
Un 5% de los pacientes de la clínica de TAR con conteos de células CD4 ≤ 100 células/ml en el norte de Tanzania tenía una antigenemia criptocócica asintomática, lo cual sugiere que el despistaje de AgCr sería válido dentro del programa TAR en Tanzania. EL PFL es una alternativa fiable, barata y práctica a la PAL para la detección de AgCr.
Cryptococcal meningitis (CM) is estimated to cause over 500 000 deaths annually in sub-Saharan Africa (Park et al. 2009) and is a leading cause of mortality in HIV-infected patients initiating antiretroviral therapy (ART) in Africa (Jarvis & Harrison 2007; Lawn et al. 2008; Walker et al. 2012). Recent data suggest that most CM cases that develop after ART initiation could be prevented by screening patients for subclinical cryptococcal infection at ART programme entry, using cryptococcal antigen (CRAG) detection assays, and pre-emptively treating those who test positive with high-dose fluconazole (Jarvis et al. 2009, 2010, 2012; Meya et al. 2010). Highly sensitive and specific CRAG detection assays, usually latex agglutination (LA) kits, have been available for many years, but such assays require laboratory infrastructure and capacity often lacking in low-resource settings.
The development of a new simple to use point-of-care lateral flow dipstick assay (LFA) has the potential to markedly facilitate both CRAG screening of patients initiating ART and earlier diagnosis of patients presenting with symptomatic cryptococcal disease in much of Africa (Jarvis et al. 2011; Lindsley et al. 2011).
The prevalence of cryptococcal antigenaemia in patients entering ART programmes in Tanzania is unknown. This is important to quantify as the cost-effectiveness of CRAG screening for patients initiating ART depends upon the CRAG prevalence in the patient population screened (Meya et al. 2010; Jarvis et al. 2013). We performed a cross-sectional study to determine the prevalence of cryptococcal antigenaemia in patients initiating or who had recently initiated ART at a clinic in northern Tanzania. In parallel, we evaluated the performance of the novel cryptococcal LFA compared with the conventional LA test, both in the ART clinic attendees and in symptomatic inpatients with suspected cryptococcal disease.
Patients were recruited and samples collected between 1 September 2011 and 29 February 2012 at Kilimanjaro Christian Medical Centre (KCMC), Moshi, the tertiary referral hospital serving northern Tanzania. The study was approved by the Kilimanjaro Christian Medical University College Research Ethics Committee. All study participants provided written informed consent. For unconscious patients in the hospital, primary relatives and/or spouses provided written informed consent. Consecutive HIV-infected outpatients aged 13 years and older with CD4 cell counts <200 cells/μl (FACS Count, Becton Dickinson, Mountain View, CA, USA) who were either ART naive or had been on ART for <6 months were recruited at the care and treatment clinic (CTC). For further validation of the lateral flow assay, HIV-infected inpatients aged 13 years and older on the medical wards at KCMC with suspected CM presenting during the study period were also recruited. Patients taking fluconazole or treated for CM in the 6 months prior to study enrolment were excluded. Demographic, clinical and laboratory data, including a full symptom questionnaire, were recorded at study enrolment, and 3 ml of whole blood was collected from each participant. Serum samples were obtained by centrifugation and tested for CRAG using the latex agglutination (LA) and lateral flow assay (LFA) kits (Immuno-Mycologics Inc, Norman, OK, USA) as per manufacturer's instructions. Positive results were reported to the responsible clinicians for further investigation and treatment at their discretion. Data were analysed using Stata, version 12.0 (StataCorp, College Station, Texas, USA). CRAG prevalence with 95% confidence intervals (95% CIs) was calculated in the outpatient cohort. Baseline variables were compared across CRAG-positive and CRAG-negative groups with the Mann–Whitney U test, the Chi-squared test or Fisher's exact test as appropriate. Associations between ART status, CD4 cell count and CRAG positivity were examined using cross tabulations and calculation of odds ratios, and a simple logistic regression model with likelihood ratio testing. Agreement between LA and LFA CRAG tests was assessed in both the outpatient cohort, and the whole study population (inpatients and outpatients) using Cohen's kappa coefficient. Statistical significance was defined as P ≤ 0.05.
The study included 218 outpatients. 43% (n = 94) were male, the median age was 39 (IQR 31–48) years, 44% (n = 96) were on ART, and the median CD4 cell count was 96 (IQR 69–115) cells/µL. Seven patients (3%, 95%CI 1–7%) were CRAG-positive, all on both the LA and LFA test. The median LA CRAG titre was 1:128 (IQR 1:32–1:512). None of the patients had symptoms suggestive of CM on direct questioning (headache, neck stiffness, photophobia or vomiting). CRAG-positive patients had significantly lower median CD4 cell counts than CRAG-negative patients: 19 cells/µL (IQR 9–58 cells/µL) vs. 97 cells/µL (IQR 72–116 cells/µL), P < 0.001 (Table 1). Patients with a CD4 cell >50 cells/µL had markedly lower odds of being CRAG positive than those with CD4 cell counts ≤50 cells/µL after adjusting for ART status (adjusted odds ratio 0.03, 95% confidence interval 0.01–0.08). ART-naive patients had higher odds of being CRAG-positive than patients on ART in unadjusted analysis; however, this association did not remain significant after adjustment for CD4 cell count (Table 2). When analysis was restricted to just those patients with a CD4 cell count ≤ 100 cells/µL (n = 124), six patients (5%, 95% CI 2–10%) were CRAG-positive (Table 1). Agreement between the LA and LFA test in the 218 asymptomatic patients was 100%.
Table 1. Baseline demographic, clinical and laboratory data
|Outpatients|| n = 218 ||211 (97%)||7 (3%)|| |
|Sex (% male)||43% (94)||44% (92)||29% (2)||0.7|
|Age (years)||39 (31–48)||39 (31–48)||38 (30–55)||0.6|
|ART (% on ART)||44% (96)||45% (95)||14% (1)||0.1|
|CD4 count (cells/μl)||96 (69–115)||97 (72–116)||19 (9–58)||<0.001|
|Outpatients CD4 ≤100 cells/μl||n = 124||118 (95%)||6 (5%)|| |
|Sex (% male)||44% (54)||44% (52)||33% (2)||0.5|
|Age (years)||38 (30–48)||37 (30–47)||43 (30–55)||0.4|
|ART (% on ART)||39% (48)||40% (47)||17% (1)||0.4|
|CD4 count (cells/μl)||74 (54–89)||75 (59–90)||17 (9–36)||<0.001|
|Inpatients||n = 101||44 (44%)||57 (56%)|| |
|Sex (% male)||53% (54)||57% (25)||51% (29)||0.6|
|Age (years)||38 (30–48)||40 (32–51)||38 (30–47)||0.5|
|ART (% on ART)||27% (27)||36% (16)||19% (11)||0.06|
|CD4 count (cells/μl)||33 (18–78)||82 (48–104)||25 (12–32)||<0.001|
|Headache||67 (66%)||20 (45%)||47 (82%)||<0.001|
|Neck stiffness||18 (18%)||0 (0%)||18 (32%)||<0.001|
|Photophobia||1 (1%)||0 (0%)||1 (2%)||1.0|
|Coma||13 (13%)||1 (2%)||12 (21%)||0.006|
|Vomiting||22 (22%)||7 (16%)||15 (26%)||0.2|
|Fever||53 (52%)||15 (34%)||38 (67%)||0.001|
|Cough||63 (62%)||28 (64%)||35 (61%)||0.8|
Table 2. Association between cryptococcal antigen detection (by LFA), ART status and CD4 cell count in ART outpatient clinic attendees (n = 218)
|ART status||ART (n = 96)||1% (1)||1||<0.001||1||0.2|
|No ART (n = 122)||5% (6)||3.3 (1.7–6.7)||1.7 (0.7–3.9)|
|CD4 cell count||0–50 cells/μl (n = 29)||17% (5)||1||<0.001||1||<0.001|
|51–100 cells/μl (n = 95)||1% (1)||0.02 (0.01–0.09)||0.03 (0.01–0.08)|
|>100 cells/μl (n = 94)||1% (1)||0.02 (0.00–0.09)||0.02 (0.01–0.07)|
Table 3. Agreement between cryptococcal lateral flow assay (LFA) and latex agglutination (LA) tests in 319 Tanzanian patients
|Latex agglutination (LA)|
An additional 101 inpatients with clinically suspected cryptococcal disease were tested for CRAG using both the LA and LFA kits (Table 1), of whom 56 (55%) were CRAG-positive on both the LA and LFA tests. One patient was positive using the LFA test but negative on the LA test. This patient had a CD4 cell count of 129 cells/µL, was on ART, and was symptomatic with cough and fever, but no headache, neck stiffness or photophobia. Thus, in the total cohort of 319 in- and outpatients, 64 patients were LFA positive, 63 of whom were also LA positive. No patients were LA positive but LFA negative (Table 2). The overall agreement between the two assays was 99.7%, giving a kappa coefficient of 0.99 (standard error 0.06, P < 0.001).
Cryptococcal antigenaemia was detectable in 5% of patients with CD4 cell counts ≤100 cells/µL attending an ART clinic in northern Tanzania. These are the first CRAG prevalence data from ART clinic attendees in Tanzania and are consistent with other East African data showing a high prevalence of subclinical cryptococcal infection in asymptomatic HIV-infected individuals with advanced immune suppression(Lara-Peredo et al. 2000; Tassie et al. 2003; Liechty et al. 2007; Meya et al. 2010; Meyer et al. 2013). Liechty et al. (2007) found a 5.8% prevalence of cryptococcal antigenaemia in a cohort of 377 patients with CD4 cell counts ≤100 cells/µL and no prior history of CM at ART initiation in rural Uganda. A second study in Kampala, Uganda, detected a higher CRAG prevalence of 8.8% in a similar cohort of 295 patients (Meya et al. 2010), and a recent Kenyan study reported a CRAG prevalence of 11% in a cohort of 514 ART-naive patients with CD4 cell counts ≤100 cells/µL (Meyer et al. 2013). These findings are also in keeping with data from Southern African ART programmes, where CRAG prevalence in patients screened prior to treatment initiation ranges from four to seven per cent (Jarvis et al. 2009; NHLS 2013). The reason for the slightly lower CRAG prevalence found in our study than the recent studies from Kampala (Meya et al. 2010) and western Kenya (Meyer et al. 2013) may relate to our inclusion of patients on ART. This was carried out for operational reasons to maximise sample size within the recruitment time frame. The duration of ART permitted prior to patient recruitment was limited to 6 months, and the available data suggest that cryptococcal antigenaemia persists for over 1 year in most patients following treatment for CM (Jarvis et al. 2011). A small number of patients with asymptomatic low-level antigenaemia may have cleared their infection within the first 6 months of ART (Jarvis et al. 2009; Kwan et al. 2012), leading us to underestimate the true CRAG prevalence. More importantly, as CM usually develops very early after ART initiation (Jarvis et al. 2010) and has a high case-fatality rate, a number of these patients may have been missed as they would either be dead, or excluded from the study due to their previous CM, again leading to an underestimation of the CRAG prevalence in the section of the cohort on ART.
Although the cross-sectional method of our study means that we have no information about what subsequently happened to patients with subclinical cryptococcal infection, evidence from several independent studies demonstrates that asymptomatic cryptococcal antigenaemia is a strong risk factor for CM and mortality in patients initiating ART (Liechty et al. 2007; Jarvis et al. 2009; Meya et al. 2010). The CRAG prevalence of 5% found in this cohort of ART clinic attendees with CD4 cell counts ≤100 cells/µL suggests that CRAG screening and pre-emptive treatment for CRAG-positive patients would be a worthwhile and almost certainly cost-effective (Meya et al. 2010; Jarvis et al. 2013) intervention in the Tanzanian ART programme.
Our results also highlight the good agreement between the conventional LA CRAG detection assays and the dipstick format LFA, further adding to the accumulating data demonstrating close agreement between the two tests (Jarvis et al. 2011; Lindsley et al. 2011; Binnicker et al. 2012; McMullan et al. 2012; Hansen et al. 2013). The only discordance between the LA and LFA tests in our study was in a single patient with pulmonary symptoms who was positive using the LFA but had a negative LA. Rather than a ‘false positive’ LFA result, this may have been a ‘false negative’ LA result in a patient with low-level cryptococcal antigenaemia, reflecting the ability of the LFA to detect lower levels of antigen than the conventional LA assays (Gates-Hollingsworth & Kozel 2013).
These data both emphasise the high burden that HIV-associated cryptococcal disease places on in-patient hospital services in Tanzania (Kisenge et al. 2007; Wajanga et al. 2011), with 57 patients diagnosed with cryptococcal infection during a six-month period, and provide further justification for targeted CRAG testing of HIV-infected inpatients and improved resources for the treatment of CM in Tanzania (Wajanga et al. 2011).
In conclusion, 5% of patients with CD4 cell counts ≤ 100 cells/µL attending an ART clinic in northern Tanzania had detectable asymptomatic cryptococcal infection, suggesting that CRAG screening and targeted pre-emptive antifungal treatment in this patient group may reduce morbidity and mortality due to CM. The LFA is a reliable, cheap and practical alternative to LA for detection of CRAG in regions such as Tanzania.
Immuno-Mycologics Inc (IMMY) donated the latex agglutination and lateral flow assay kits. The ministry of health and social welfare is acknowledged for sponsoring the postgraduate training. We thank the nurses at the CTC who helped with data collection and the patients involved in the study. Funding was obtained from the Ministry of Health and Social Welfare, government of the United Republic of Tanzania. The sponsors were not involved in study design, data analysis or preparation of the manuscript.