Cryptococcal Antigenemia in HIV therapy-experienced Ugandans with Virologic Failure

Mpoza, Edward; Rajasingham, Radha; Tugume, Lillian; Rhein, Joshua; Nabaggala, Maria Sarah; Ssewanyana, Isaac; Nyegenye, Wilson; Kushemererwa, Grace Esther; Mulema, Vivienne; Kalamya, Julius; +5 more... Kiyaga, Charles; Kabanda, Joseph; Ssali, Mina; Boulware, David R; Meya, David B; (2019) Cryptococcal Antigenemia in HIV therapy-experienced Ugandans with Virologic Failure. Clinical infectious diseases. ISSN 1058-4838 DOI: https://doi.org/10.1093/cid/ciz1069 Downloaded from: http://researchonline.lshtm.ac.uk/id/eprint/4655311/ DOI: https://doi.org/10.1093/cid/ciz1069


INTRODUCTION
Cryptococcal meningitis is responsible for 15% of AIDS-related deaths and is the leading cause of adult meningitis in Sub-Saharan Africa (1)(2). Annual global deaths from cryptococcosis are estimated at 181,100, with 75% of these deaths in Sub-Saharan Africa (1). Cryptococcal antigen (CrAg) is an independent predictor of mortality and can be detected in blood by latex agglutination a median of three weeks before developing initial symptoms of clinical cryptococcosis (3)(4). CrAg screening is cost effective, has survival benefit, and is recommended by the World Health Organization (WHO) in persons with CD4 count <100 cells/mcL, and can be considered below 200 cells/µL (5)(6)(7)(8)(9). Among ART-naïve outpatient populations in low and middle income countries, the CrAg prevalence in 2014 averages 6% with some regional variations (1). Among Ugandan ART-naive outpatients in 2004-2006 with CD4 <100 cells/mcL the CrAg prevalence is 8.8% (10).
There is a substantial proportion of ART-experienced patients with virologic failure presenting with fulminant cryptococcosis (11). During 2013-2017 in Uganda, nearly half of the patients presenting with cryptococcal meningitis were ART-experienced (11). These ARTexperienced persons with virologic failure are at risk for cryptococcosis, yet they are not included within WHO CrAg screening guidelines' high risk category (5,12). As monitoring of those receiving ART is scaled up from CD4 to virologic monitoring in resource limited settings, CD4-based algorithms for CrAg screening may not be readily applicable to ART-experienced populations. Baseline CD4 counts may be performed at HIV diagnosis but not routinely while monitoring HIV treatment. The Uganda HIV guidelines 2018 recommend CrAg screening in ART naïve or treatment failure patients with CD4 cell counts less than 100cells/mcL (13). In the absence of CD4 monitoring, it is unknown whether to screen persons with virologic failure.
Given the substantial proportion of ART-experienced patients presenting with cryptococcosis, new viral load-based CrAg screening strategies may be useful. Uganda has also adopted the 'test and treat' strategy which will enroll more people on ART irrespective of their CD4 counts. In addition we are now encouraging more virologic monitoring compared to CD4 monitoring. This creates a challenge for CrAg screening which are based on CD4 counts. We conducted this study to determine the CrAg prevalence and clinical outcomes of CrAg-positive persons among ART-experienced people with virologic failure. We evaluated the potential threshold of virologic failure where CrAg screening should be considered, and cost of CrAg screening at this threshold.

Study Design, Setting and Participants
This cross-sectional study was conducted at the Uganda National Health Laboratory Services (UNHLS), which performs centralized viral load monitoring in Uganda (~95% of all HIV viral load testing). The UNHLS processed a total of 843,020 viral load tests for adults in 2017 of which 10% (84302/843020)had unsuppressed viremia of >1,000 copies/mL (14). Dried blood spot and plasma samples from health facilities are delivered to the UNHLS via a hub transport system for HIV viral load testing (15). We retrospectively evaluated stored plasma samples of 1,186 ART-experienced HIV-infected adults (≥18 years) with suspected virologic failure with HIV viral loads above >1,000 copies/mL. The sample size was estimated using Buderer's formula (using an absolute precision of 0.1, a confidence Interval of 95%, an estimated prevalence of 8.9%(from a CrAg study in Nigeria which had largely ART experienced people; 5 (16)) and assumed a sensitivity of 50% giving us 1079 subjects to which we added a 10% margin for possible missing samples making a total of 1186 subjects. We tested left over plasma samples after viral load testing from specimens collected during September 2017-January 2018. The samples from September 2017-January 2018 were the most recent easily accessible and retrievable samples. We didn't have stored samples of plasma for patients who had dried blood spots. CrAg LFA testing on dried blood spots has lower sensitivity compared to plasma (unpublished data). CrAg testing was performed from March to July 2018 followed by retrospective chart review to determine 6-month outcomes for the CrAg-positive patients.

Study procedures
Using the UNHLS database, we identified all patients with virologic failure and left over

Ethic Reviews
We obtained ethical review and approval from the Makerere University School of Medicine research and ethics committee, as well as Uganda National Council of Science and Technology, and the University of Minnesota. A waiver for informed consent was obtained as CrAg testing was performed on leftover plasma samples, and the study posed no more than minimal risk to participants.

Statistical methods
Data analysis was primarily descriptive with variables summarized by mean with 95% confidence interval (CI), median with inter -quartile range (IQR), and number (percentage). We tested continuous variables via nonparametric Wilcoxon Rank-Sum test and categorical data via Chi-square. We additionally used logistic regression to assess whether CrAg-positivity was associated with quantitative HIV viral load thresholds.

Costing
Once CrAg prevalence was determined, we used the inverse to calculate the number needed to test to detect one CrAg+ person. The cost of CrAg screening using the lateral flow assay in Uganda is $3.50 (17). We multiplied the cost of screening by the number needed to test to identify the cost to detect one CrAg positive person. Another two patients developed cryptococcal meningitis during these 6 months were treated and survived. Thus, meningitis-free survival at 6-months was 61% (14/23; 95%CI, 35% to 76%).
We also did an analysis including those who were lost to follow up. We used 6 month outcome findings from a CrAg+ cohort in Uganda which found 14% mortality rate, 77% meningitis free survival and 9% progression to meningitis(10). If we applied this to the 10 without survival data, we would have 8 who survived meningitis free,1 who died and 1 who developed cryptococcal meningitis. This would then make a 63% (22/35 )meningitis free survival rate. If we were to consider that all the 10 without survival data died then the meningitis free survival would be 40% (14/35).
We determined the likely outcome of those lost to follow up (including those without outcome data) by comparing their sex, median ages and viral loads with those who survived and died. We 9 used the Fischer's exact test for the sex comparison across the three groups and found no significant difference (p=0.42). The Kruskal Wallis test was used to compare the medians for age and viral load across the three groups and the test was not significant for both age (p = 0.42) and viral load (p = 0.10). Therefore, the likely outcome of the individuals lost to follow up may be similar to those who were accounted.

Cost of CrAg screening
We identified 4.2% CrAg prevalence among persons with a HIV viral load >5000 copies/mL. Thus, the number needed to test to detect one CrAg positive person is 25. At a cost of $3.50 for the CrAg lateral flow assay (17), the cost to detect one CrAg positive person among those with virologic failure is $87.50.

Discussion
The CrAg prevalence was 3.0% in ART-experienced HIV patients with virologic failure and 4.2% among those with >5,000 HIV RNA copies/mL. To our knowledge, this is the first study to evaluate CrAg prevalence in ART-experienced patients with virologic failure, irrespective of CD4 count. This CrAg prevalence rate was similar to that in ART-experienced populations in studies in South Africa 2.8% , Brazil 3.1% (18)(19), and in Ethiopia 4.1% (20) but lower than that in studies in Nigeria (8.9%) and Ethiopia (8.4%) though all these studies included participants responding to ART or didn't report viral loads (12,16). CrAg positivity was shown to significantly increase with higher viral loads indicating a higher risk for cryptococcosis in people with more fulminant virologic failure. We found an all-cause 6-month meningitis-free survival rate of 61% among CrAg-positive persons, of whom at least 6 survivors received fluconazole therapy. Absent this fluconazole therapy, meningitis-free survival was ~40%, although the 95% confidence interval was wide.
We calculated that it would cost $87.  (11,21). WHO guidelines recommend CrAg screening for persons with CD4<100 cells/L prior to initiating ART (9, 22), however, for the ART experienced persons no guidelines exist. Given that virologic monitoring is emphasized more than CD4 monitoring in this population, using a viral load threshold to identify who needs CrAg screening becomes more pertinent. The guidelines acknowledge that CrAg screening in ART-experienced persons needs to be further evaluated. Based on our findings, in the absence of CD4 monitoring, we would recommend CrAg screening among persons with virologic failure >5,000 copies/mL as this threshold where we found a CrAg prevalence of 4.2%.
The study had limited access to patient medical information, thus missed some clinically relevant data, such as CD4 cell count, which could have been useful in further characterizing the population of interest. This could also have introduced information bias in characterizing patients as meningitis free. The potential seasonal distribution of cryptococcal infections may have influenced prevalence results. While we only tested plasma specimens, whether dried blood spots would have a different CrAg prevalence is unclear as the CrAg LFA sensitivity is lower in dried blood spots than in plasma due to the extraction process (unpublished data). Viral load testing is being scaled up replacing CD4 monitoring yet absolute CD4 counts provide the risk stratification to target evaluation for opportunistic infections, including CrAg screening. The cost of CD4 testing (~$6) is more than the cost of CrAg testing (<$4), thus from a program implementation perspective, using CD4 testing to then narrow the pool to select for CrAg testing is not efficient, unless a CD4 is already being performed (17,23). We should consider expanding the threshold for CrAg testing by including those with viral load >5,000 copies/mL which would decrease the 13

Disclaimer:
The views expressed in this publication are those of the author(s) and not necessarily those of AAS, NEPAD Agency, Wellcome Trust or the UK government. None of the authors has a potential conflict of interest