Cryptococcus meningitis is caused by Cryptococcus neoformans, an encapsulated yeast-like organism ubiquitous in the environment.^1,2 Although exposed to C. neoformans pathogens, most people with a normal immune system do not acquire this infection. However, those with a lowered immune response are susceptible to cryptococcal infection, which has a particular tropism for the central nervous system, frequently causing fatal cryptococcal meningitis (CCM).³ This is a common opportunistic infection in Southeast Asia and sub-Saharan Africa,¹ and is an Acquired Immune Deficiency Syndrome (AIDS)-defining illness in patients with late-stage HIV infection. It remains a major cause of HIV-related mortality worldwide with the largest burden of the disease being in sub-Saharan Africa, where health care access is often limited.^4,5,6

South Africa (SA) has the largest HIV epidemic in the world, with approximately 7.06 million South Africans living with the virus in 2017 (12.6% of the total South African population).⁷ Since 2004, the antiretroviral therapy (ART) programme in SA has grown considerably, and by 2016, it was estimated that 3 929 000 (56%) people living with HIV (PLWHIV) were on treatment.^8,9 In 2015, SA adopted the United Nations AIDS 90–90–90 ambitious but achievable treatment targets to help end the AIDS epidemic by 2020. In this programme, it is expected that 90% of the South African population will be aware of their HIV status; 90% of those who test positive will be on ART, and 90% of those who are on ARTs will be virally supressed by 2020.

Despite these ambitious goals and the increased coverage of ART in SA, the incidence of HIV-associated CCM remains high with a case-fatality rate ranging between 30% and 50%.^2,4,10,11 There is a high susceptibility to cryptococcal infection and the subsequent development of CCM among patients with a CD4 count ≤ 100 cells/mm³ as well as a high associated mortality.^12,13 Therefore, in 2011, the World Health Organization (WHO) recommended that prior to initiating ART in populations with a high prevalence of cryptococcal (Cr) antigenaemia, all ART-naïve adults with a CD4 count ≤ 100 cells/mm³ should be routinely screened. People with evidence of cryptococcal disease should then be assessed for evidence of CCM and treated when positively diagnosed.^14,15

In 2014, the South African Department of Health adopted the WHO recommendations,¹⁴ and a screen-and-treat intervention was added to the South African HIV guidelines¹⁶ which require that all HIV-positive patients with a CD4 ≤ 100 cells/mm³ be screened for Cr antigenaemia. All those who are Cr antigen positive (CrAg +ve) should be assessed for CCM, and those with symptoms suggestive of CCM should have a lumbar puncture (LP) and be treated if the diagnosis is confirmed. The purpose of this proactive screen-and-treat intervention is to reduce CCM-related deaths.^10,16

The aim of this study was to assess the effectiveness of the current screen-and-treat intervention for HIV-positive patients with a CD4 count ≤ 100 cells/mm³ who presented for care either at a large regional hospital in Durban or at a clinic that referred patients to the hospital, during 2015 and 2016.

Prince Mshiyeni Memorial Hospital (PMMH) is a 1200-bed regional hospital situated on the outskirts of Umlazi Township, Durban, and serves a population of approximately 2 million people with an estimated HIV prevalence of 16.9%.¹⁷ There are 17 clinics in the hospital ‘catchment area’ that refer the patients to the hospital. All laboratory investigations (bloods, cerebrospinal fluid [CSF], etc.) from these clinics and PMMH hospital are sent to the National Health Laboratory Service (NHLS) laboratory based at the hospital for analysis. In accordance with the National HIV guidelines, all HIV-positive patients presenting to any of the 17 clinics or to PMMH must have a baseline CD4 count, and those specimens where the CD4 count is ≤ 100 cells/mm³ are reflexively screened for the presence of CrAg using the same blood specimen. Those patients who are CrAg +ve are assessed by the clinicians for symptoms suggestive of CCM (headache, photophobia and/or confusion), and if symptomatic, are offered an LP to rule it out as part of the screen-and-treat intervention.

This study was divided into two parts, the first being a review of the data from the NHLS laboratory at PMMH of all those with a CD4 count ≤ 100 cells/mm³. This was followed by the second part, which was a review of the hospital records of all patients referred for an LP for suspected CCM and all those admitted to the hospital with confirmed CCM.

Ethical approval for the study was granted by Biomedical Research Ethic Committee (BREC REF No: BE402/16) of the University of KwaZulu-Natal (UKZN). Permission to conduct the study was granted by Prince Mshiyeni Memorial Hospital (PMMH) management and National Health Laboratory Service (NHLS) management.

In the NHLS component of the study, 3702 patients presented for care at PMMH and the 17 clinics between June 2015 and May 2016 with a CD4 ≤ 100 cells/mm³. Although progress is being made towards achieving the 90–90–90 targets, many patients continue to present late in the progression of the disease. The findings of a recent study showed that 85.5% of HIV-positive South Africans were aware of their status,⁹ and as ARTs have been widely available since 2004, it is a concern that so many patients continue to present only when their HIV is at an advanced stage. These findings suggest that despite extensive publicity to encourage people to test early and regularly for HIV and to know their status, many only become aware of their HIV status when they develop symptoms, by which time their CD4 counts are likely to be very low. Since the early 1990s, considerable resources have gone into HIV campaigns encouraging safe sexual behaviour and the importance of knowing one’s status. However, it would appear that a more nuanced understanding of the factors influencing when people choose to test for HIV and wish to start treatment is needed if all members of the population are to be encouraged to test early to know their HIV status and access ART.

In this study, only 5.1% (190 out of 3702) of the patients with a CD4 count of ≤ 100 cell/mm³ had Cr antigenaemia. These findings are higher than the 2.1% with Cr antigenaemia reported from a study in Soweto,¹⁸ consistent with studies in SA and UK,^19,20,21 lower than the 13% reported from a 2009 study in Cape Town,² and much lower than the 19% found in Uganda²² and the 17.7% reported in Cambodia.²³ With C. neoformans being ubiquitous in the environment, it is not clear why the percentage of patients with Cr antigenaemia varies widely between patients with a low CD4 count, and may have to do with the incidence of cryptococcal disease in the population being studied.¹⁸

Jarvis et al. reported that a negative CrAg was a 100% negative predictor for the development of CCM in patients with a low CD4 count.² This is helpful for clinicians, as it means that among the 3512 patients who were CrAg –ve in this study, there was no need to provide primary prophylaxis against CCM.^2,11,15 However, based on a cohort of patients tested in 2011, Govender et al. reported that a small number of patients (7 out of 1430; 0.5%) developed CCM despite their initial CrAg being negative,¹⁸ as did Pongsa who reported that 0.8% (1 out of 119) of patients developed CCM despite having an initial CrAg that was negative.²⁴ Govender concluded that these patients with very low CD4 counts developed CCM because of the delay in initiation of ARTs. The current South African guidelines recommend that all patients who are CrAg –ve do not need primary prophylaxis with fluconazole, but need to be fast-tracked onto ARTs as the subsequent improvement in the CD4 count is the best defence against CCM.^15,24

Cryptococcal antigen immunoassays have been shown to be effective in identifying patients at risk of developing subclinical cryptococcal infection before ART initiation.^2,25 Despite this, there is currently no consensus about how to manage patients who are CrAg +ve. The current South African HIV guidelines recommend that only those patients with symptoms suggestive of CCM should have an LP to confirm the diagnosis, and that those who are asymptomatic should receive high-dose fluconazole to prevent its development.^15,24 Whilst a 2005 Cochrane review reported a reduction in the incidence of cryptococcal disease, there was no reduction in mortality associated with CCM from such approach.²⁶

In this study, laboratory records show that only 2.6% (5 out of 190) of patients who were CrAg +ve had an LP at PMMH as part of the screening programme. In addition, no patient identified for an LP in this manner was diagnosed with CCM on LP, which suggests that clinicians were not effective in identifying patients with possible early onset of CCM. These results are very different from the findings from Gauteng province, where 41% (99 out of 244) of patients who were CrAg +ve were symptomatic, of whom 56 (57%) had an LP and 59% (33 out of 56) were diagnosed with CCM.²⁷ Furthermore, the study in Gauteng reported that 26% (8 out of 31) of the asymptomatic CrAg +ve patients had CCM on LP.²⁷ Further studies are needed to understand the discrepancy between these findings, which may be because of the large patient load in Umlazi, the ad-hoc manner in which patients who are CrAg +ve are assessed for symptoms of CCM, or the challenges of referring patients from the clinics to the hospital. Additional research is needed to determine how to effectively screen patients with low CD4 counts for symptoms of CCM.

Alternative approaches to managing patients who are CrAg +ve include providing CrAg titres to help clinicians decide which patients require an LP,^2,24 or an LP for all patients who are CrAg +ve to exclude CCM, regardless of whether or not they are symptomatic.²⁵ In a study in Thailand, where all patients who were CrAg +ve had an LP, 25% (3 out of 12) of patients had active CCM,²⁴ whilst a retrospective review of patients who developed CCM in Cape Town showed that 28% (13 out of 46) who were CrAg +ve developed CCM.² A follow-up study in Gauteng in 2012–2014, among HIV patients with a CD4 count ≤ 100 cells/mm³ who were CrAg +ve, showed that 25% had subclinical CCM.²⁷

Based on the review of hospital records, 20% (38 out of 190) of patients who had been identified as CrAg +ve presented to the hospital with symptoms of CCM had their diagnosis confirmed LP. This finding is consistent with the study in Thailand,²⁴ but much lower than the 66% – 71% reported in earlier studies among patients with stage IV HIV disease.^28,29 These 38 patients may represent those patients who were lost to follow-up, which is not an uncommon occurrence,¹⁸ or be a missed opportunity for early identification of CCM and initiation of treatment prior to the onset of obvious symptoms of CCM. In this study, the clinicians appeared to be unable, in the screening process, to identify those patients who went on to develop CCM. This suggests that patients with low CD4 counts who are CrAg +ve and who develop CCM either have no or very subtle symptoms that are not recognised as possible precursors to the development of CCM. However, there may be patients who go on to develop CCM once they have started on ARTs, as was the experience in the study in Cape Town.^2,25 More research is needed to understand the early symptoms that might be predictive of CCM.

Cryptococcal meningitis continues to have a significant mortality, despite the introduction of high-dose fluconazole into the South African treatment guidelines. Based on the data available at PMMH, the overall mortality in this study was 30%, which is consistent with other studies.^6,11,18 It is important to note that the mortality was significantly higher in the cohort with a CD4 count ≤ 100 cell/mm³ (31.6%) and in the unknown CD4 cohort (30.8%) than in those with a CD4 count > 100 cell/mm³ (20%), supporting the call to identify patients who are HIV-positive and to initiate ART before they develop profound immunological failure. This was not an unexpected finding, as many studies have linked CCM-related mortality to a lower CD4 count.^2,19,20,30

In the light of the high mortality associated with CCM, the challenges associated with early identification of subclinical CCM, the cost effectiveness of the screen-and-treat intervention (estimated to be cost-effective at a CrAg threshold of 0.6%),^18,20,30 the high pickup rate of CCM in asymptomatic CrAg +ve patients,^24,27 as well as the resources available in SA, serious consideration should be given to recommending LPs for all patients with a low CD4 count who are CrAg +ve.

Although this study is limited by the small numbers, with 20% mortality in those patients whose CD4 count ranged between 101 and 150 cells/mm³, consideration should include increasing the CD4 count level to 150 cells/mm³ for patients who are screened for CrAg. In 2015, Govender et al. reported a 97% yield of incident antigenaemia when the CD4 threshold was increased from 100 cells/mm³ to 150 cells/mm,^3,18 suggesting that this increase in the CD4 cut-off threshold may help to timely detect CCM in this group of patients. Further study is needed in this area as well as assessing the effect of giving patients who are CrAg +ve prophylactic fluconazole, in keeping with the current guidelines, to determine what impact this would have on the development of CCM.

Because of the design of this study, it was not possible to follow up with patients who had Cryptococcus antigenaemia. Only a few patients were admitted for an LP following the screen-and-treat intervention, and the results need to be treated with caution. It is possible that some patients were referred but never presented to PMMH for LP and that some symptomatic patients were seen and managed at other health institutions, although the referral system is designed so that clinics in the Umlazi area refer patients to PMMH.

The current ad-hoc system at busy public sector clinics does not appear to be effective identifying and/or referring patients with subclinical symptoms of CCM nor in reducing its associated mortality. Although screening for Cr antigenaemia is considered to be cost-effective, the benefits are currently not being realised. This may be because of the challenges associated with identifying patients with latent or early CCM as well as the challenges within the health care system that include large numbers of patients, overloaded staff and the failure of patients to return for follow-up. Although this is a small descriptive study limited to one location, it does provide data based on day-to-day practice with caution being advised when interpreting the data. Based on the findings of this research, we recommend that a system be considered to ensure that all patients with Cr antigenaemia are offered an LP to exclude CCM, and that the cut-off point for assessing Cr antigenaemia be increased to a CD4 count of 150 cells/mm³.

It is also very important that all patients with a low CD4 count be fast-tracked onto ARTs so as to reduce the risk of developing CCM. Further research is needed to assess the impact of these recommendations, and the effectiveness of other interventions aimed at reducing the mortality from CCM. This includes the use of oral fluconazole in patients who have Cr antigenaemia but not CCM, and high-dose fluconazole in combination with Amphotericin B for the management of CCM.