Infection with the saprophytic yeast Cryptococcus neoformans is a well-recognized complication of immunosuppression. During the 1990s, as a consequence of underlying HIV infection, cryptococcal meningitis has become the leading reported cause of adult meningitis in sub-Saharan Africa [1–7]. In addition, it is a leading cause of bloodstream infection in HIV-infected adults in most [8,9] but not all parts  of the developing world. Mortality from cryptococcal disease is inevitable in the absence of appropriate therapy.
Despite the frequent reports and seriousness of C. neoformans as a pathogen in regions of high HIV prevalence, there is no reporting of community-based incidence data. Consequently, the relative importance of this infection as a cause of morbidity and mortality is uncertain, and the role of prevention and treatment cannot be defined or prioritized. In the absence of antiretroviral therapy, care of HIV-infected adults in the developing world is focused on infection prophylaxis and treatment of concomitant infections. At present, specific recommendations exist for the use of isoniazid and co-trimoxazole. Primary chemoprophylaxis may be a suitable approach for cryptococcal disease prevention [11,12], particularly with the high cost of treatment and subsequent secondary prophylaxis. In addition, further evaluation of a previously reported protein conjugate cryptococcal polysaccharide vaccine may be warranted [13,14]. The potential benefit of these interventions for management may only be estimated with appropriate natural history data.
A cohort of over 1000 HIV-1-infected Ugandan adults was established in October 1995, initially to investigate the value of 23-valent pneumococcal vaccine for the prevention of invasive pneumococcal disease . Participants’ health was comprehensively surveyed throughout this study and we describe the natural history of cryptococcal disease in this population and discuss the implications for disease prophylaxis.
The pneumococcal trial methodology has been described elsewhere . In brief, the study was conducted at two community-based HIV care clinics (AIDS Support Organization of Uganda, TASO; and the Ministry of Health clinic, Uganda Virus Research Institute) in Entebbe, Uganda. HIV-1-infected adults (age 15 years or older) in World Health Organization (WHO)  clinical stages 1, 2 or 3 were invited to take part in a double-blind, randomized, placebo-controlled trial of 23-valent pneumococcal polysaccharide vaccine.
Enrolment followed signed informed consent sought in the local languages. Following a standardized clinical interview for clinical staging, a blood sample was drawn. Baseline CD4 T cell counts and haematological indices were measured. Follow-up clinical and haematological assessments were performed 1 month after enrolment and then 6 monthly. On each occasion, serum and plasma samples were archived. Participants had open access to the study clinics and were encouraged to attend whenever ill. Each episode of illness was investigated and managed according to set protocols recorded in the study procedure manual, and in accordance with standard clinical practice.
In the event of a patient failing to attend a routine clinic appointment, a study field-worker would visit the participant at home and encourage clinic attendance or arrange a visit from a trial physician. If he/she refused further participation in the cohort, had left the study area or could not be traced by virtue of an inaccurate address, he/she was deemed a defaulter. If the patient was dead, a verbal autopsy was performed. This included systematic questioning of a relative or carer about date of death and nature and duration of symptoms at the time of death.
If the terminal illness had not been managed by the study clinicians, or investigations had been incomplete, retrospective testing of a preterminal archived serum sample was undertaken to look for the presence of cryptococcal antigen (CRAG).
Therapy for cryptococcal disease was not routinely available in either community clinic. A charitable donation of itraconazole (50 mg capsules) was made during the course of the pneumococcal vaccine trial. A small number of individuals used this agent for therapy and secondary prophylaxis.
Cryptococcal infection was defined by the isolation of C. neoformans or a positive CRAG test at a titre of 1 in 8 or greater from blood or cerebrospinal fluid (CSF). A ‘definite’ diagnosis of cryptococcal disease was made if an individual was febrile with confirmed infection, in the absence of any other demonstrable explanatory pathogenic process. A ‘probable’ diagnosis of cryptococcal disease was made when a preterminal serum sample was positive for CRAG in the absence of confirmatory clinical information. Serum samples were screened if available within 3 months of the date of death.
The HIV status of all participants was confirmed following enrolment using two standardized enzyme immunoassay  (Recombigen HIV-1/2, Cambridge Biotech Corp., Cambridge, Massachusetts, USA, and Welcozyme HIV-1/2, Welcome Diagnostics, Hertford, UK). CD4 T cell counts were performed using a FACScount system (Becton Dickinson, San Jose, California, USA). This system was unable to discriminate between technical failure and a true zero CD4 T cell count. Standard microbiological techniques were used for processing and investigating clinical specimens. Blood cultures used two bottles: one a brain heart infusion broth, the other a nutrient broth. CSF was cultured on sheep blood agar and chocolate blood agar and cultured at 35°C in 5% carbon dioxide. The laboratories participated in the UK external quality assessment scheme NEQAS.
The identity of all C. neoformans isolates were confirmed by a positive urease test, ability to grow at 37°C, brown colony formation on Bird Seed agar and the presence of CRAG. Serotyping was performed using the Cryptocheck agglutination test (Tatron Laboratories, Tokyo, Japan).
Cryptococcal antigen testing was performed on CSF or serum using a Cryptococcus latex test (Latex-Crypto test, Immuno-Mycologics, Norman, Oklahoma USA) in accordance with the manufacturer's guidelines.
Study data were collected by manual recording of information in a standardized format. Data entry was performed in duplicate in databases created in Foxpro(Microsoft, Redmond, Washington State, USA). Analyses were carried out using STATA statistical software version 5.0 (Stata Corp., College station, Texas, USA).
An individual's follow-up commenced on the day of enrolment and was censored at date of death, default or 1 January 1999 (study end date), whichever was the earliest. To allow for progressive changes in the clinical and CD4 T cell staging, person-years of observation (pyo) were calculated in 6-monthly blocks; the period under observation being ascribed to a particular CD4 T cell group or clinical stage on the basis of the measurement at the beginning of this period. The CD4 T cell count from the preceding steady-state visit was used when describing the association of clinical events and the status of HIV infection. CD4 T cell counts at the time of illness were not used as they may have been lowered by the intercurrent event .
A Cox proportional hazard approach  was used to investigate the strength of association, with multiple adjustments for potential confounders identified by a preliminary univariate analysis. All tests of statistical significance were two sided.
Duration of cryptococcal antigenaemia was assessed by testing sequential archived serum samples. Antigen seroconversion was taken as the date of the first positive CRAG test.
To ensure cryptococcal antigenaemia was a specific marker of cryptococcal disease and not a consequence of advanced HIV-associated immunosuppression, enrolment serum samples from asymptomatic individuals with advanced HIV disease (CD4 T cell count < 50 × 106 cells/l) were subjected to CRAG testing. To further ensure these individuals represented a suitable cryptococcal disease-free comparison group, they must have survived at least 6 months from enrolment.
Ethics and trial conduct
The Liverpool School of Tropical Medicine Research Ethics Committee and the AIDS Research Subcommittee of the National Council of Science and Technology approved the pneumococcal vaccine trial. An independent data monitoring and ethics committee (DMEC) was established. In addition a Trial Steering Committee (TSC) supervised the trial progress and conduct, according to Medical Research Council good clinical practice guidelines .
Up to the beginning of January 1999, 1372 HIV-1-infected adults had been followed for a total of 1902 pyo, with a median follow-up of 1.2 years [interquartile range (IQR), 0.6–3.3]. Median age in the cohort was 30 (IQR, 26–36); 70% of participants were female. Nearly one-third of the cohort had died, 444 deaths (32%; rate 233/1000 pyo), and a further 130 participants (9.5%; rate 68/1000 pyo) defaulted. Deaths were strongly associated with low CD4 T cell count and advanced WHO clinical stage (Tables 1–3).
Cryptococcal disease was diagnosed in 77 participants (69 definite and eight probable), an overall crude incidence rate of 40.4/1000 pyo (36.2/1000 definite cases). Rates of disease were strongly associated with low CD4 T cell count and advancing WHO clinical stage (Tables 1 and 2). The median CD4 T cell count at diagnosis of cryptococcal disease in 76 of 77 patients for whom counts were available was 16 × 106 cells/l (range, 1–365); in 66 (86%) the CD4 T cell count was < 100 × 106 cells/l. The total lymphocyte count was < 1000 × 106 cells/l in 18 (23%), and < 2000 × 106 cells/l in 54 of the 77 patients. Cryptococcal infection was associated with 17% of all deaths in the cohort. Median survival from date of diagnosis was 26 days (range, 0–138) in the 69 definite cases. Median survival times were greater in 11 individuals who used itraconazole therapy: 38 days compared with 13 in those who received no therapy.
Demographic and clinical parameters were investigated for their association with cryptococcal disease. By univariate analysis, CD4 T cell count, WHO clinical stage, total lymphocyte count, male sex and herpes zoster infection were shown to be associated with cryptococcal disease. In a multivariate regression model, these variables, with the exception of total lymphocyte count, remained independent and strong predictors of cryptococcal disease (Table 3).
Clinical presentation was often non-specific (Table 4). Headache was present in about half of all patients but classical signs of meningitis were present in only 13 (18%). C. neoformans was the principal cause of meningitis in the cohort. Meningitis, confirmed by CSF examination, was diagnosed in 21 individuals over the period of the study; the meningitis was cryptococcal in aetiology in 17 (81%). The four other patients had pneumococcal (one), tuberculous (one) and undiagnosed lymphocytic meningitis (two). When symptom frequency in the 69 patients with definite cryptococcal disease was compared with the 1075 investigated non-cryptococcal illness episodes, the presence of neurological symptoms and signs were specific markers of cryptococcal disease but were insensitive (Table 4). The positive and negative predictive values in this cohort in respect for cryptococcal disease was 76% and 95%, respectively, for meningism and 47% and 95%, respectively, for focal neurological deficit.
Confirmation of disease was based on blood culture in 19 patients, CSF examination in nine, blood and CSF examination in eight and CRAG testing in 33. The eight patients with probable cryptococcal disease were identified from screening stored preterminal serum samples available for 219 individuals out of 375 deaths that were not definitely known to be a result of cryptococcal disease. Serum CRAG testing was positive in all culture-proven patients. Blood cultures were performed on 51 of the 69 patients and were positive in 27 (53% sensitivity). White cell counts in CSF examinations were < 4 × 106 cells/l in 14 specimens and 10, 47 and 415 × 106 cells/l in the remaining three. All 36 cryptococcal isolates were confirmed as C. neoformans var grubii (previously var neoformans serotype A).
Out of the total of 77 patients, 42 had stored serial serum samples and a seroconversion sample. Median minimum duration of CRAG positivity was 41 days (range, 1–297) in 34 individuals who had not received systemic antifungal therapy and 52 days (range, 22–241) in eight individuals who received itraconazole therapy. Median CRAG titre at death in these 42 individuals was 1 in 1024 (range 1 in 16 to 1 in 65 536). CRAG titre showed a significant inverse relationship with sampling time before death (Fig. 1). CRAG positivity preceded clinical symptoms by a median of 22 days (range, 5–234 for the 37 definite cases), four individuals (11%) being positive for more than 100 days. Seven of the individuals who developed cryptococcal disease were retrospectively identified as CRAG positive at the time of study enrolment. No CRAG antigenaemia was found in 58 individuals with advanced HIV disease without features of cryptococcal disease.
Cryptococcal disease is a common and important HIV-related problem across most of sub-Saharan Africa [2,5–7,21] and is usually diagnosed in patients with meningitis. Our findings show that cryptococcal disease is one of the leading contributors to death in HIV-infected adults in Uganda and commonly presents without the specific features of meningitis. The scale of the problem would suggest further evaluation of preventive strategies to be appropriate.
Previous reports from Africa have focused on cryptococcal disease presenting as meningitis and emphasized the frequency of neurological abnormalities . The findings from this community-based cohort suggest a broader and perhaps less-specific presentation of cryptococcal disease. Only half of the patients complained of headache and in only one-fifth was there clear physical signs of meningism. Whether individuals with non-neurological presentations also had central nervous system involvement is uncertain as systematic assessment of CSF was not undertaken when blood culture or CRAG testing was positive. Respiratory complaints were frequent and pulmonary cryptococcosis was almost certainly present in some of the patients as a component of a disseminated fungaemic infection. Chest radiographic abnormalities were present in 47% of films but these were performed on only 32 patients. Furthermore an abnormal chest radiograph bore no relationship to the presence of respiratory symptoms and signs and, as with the principal symptoms and signs, it would appear to have poor diagnostic discriminatory power. Although early presentation may explain in part the low frequency of clinical signs, a significant burden of cryptococcal disease may be going unrecognized if only neurological illness is investigated or if specific mycological investigations are not performed on febrile patients with advanced (stage 3/4) HIV/AIDS disease.
Cryptococcal antigen testing proved to be the most sensitive diagnostic test. We have previously shown it to be a specific discriminator of cryptococcal disease from other febrile illnesses . By testing serum samples from individuals with CD4 T cell counts < 50 × 106 cells/l and no apparent cryptococcal disease, we have also gone some way to confirming its specificity as a marker of cryptococcal infection and not a marker of advanced HIV disease.
CRAG positivity preceded the onset of clinically recognizable symptoms by a median of 22 days, although 11% of the patients had preceding cryptococcal antigenaemia for greater than 100 days. The serum CRAG titre was associated with survival time in this group; however, in practical clinical terms, a single titre probably carries little predictive prognostic value and routine titration of serum to record a level is unnecessary. Nevertheless, the overall strong temporal association between CRAG positivity and death supports the reliability of this test, not only as an indicator of systemic cryptococcal infection but also as a predictor of disease and subsequent death. This has been previously suggested by investigators in North America , although in their setting of unrestricted access to antifungal therapy, the treatment of asymptomatic antigenaemia obscured the relationship between antigenaemia and outcome. Our findings would support their view that the treatment of asymptomatic antigenaemia is correct. Whether routine screening for cryptococcal antigenaemia would be of value in an African population with highly restricted access to antifungal therapy is unclear: early diagnosis is only likely to be of benefit if it allows the use of simplified and less-expensive oral antifungal treatments. Further evaluation of this strategy is appropriate.
Rates of cryptococcal disease in this cohort were high, double those seen in North American reports (17–20/1000 pyo) [24,25] that predated the introduction of potent antiretroviral therapy. This excess risk persisted, with the exclusion of the seven patients who were asymptomatic but CRAG positive at the time of their enrolment. There is no reason to suspect geographic differences in HIV-associated immunosuppression in sub-Saharan Africa. Therefore, the rates in this cohort are likely to reflect the situation in much of the rest of Africa. However, cryptococcal disease is absent or infrequently reported in some African settings , suggesting either failure of diagnosis or an important exposure/environmental component of risk.
Advanced HIV-related immunosuppression proved to be the strongest independent risk factor for disease. Male sex was also shown to be a risk, similar to reports from the United States . This finding may represent incomplete adjustment for stage of disease in the multivariate analysis, as in general men enrolled in the cohort with more advanced HIV disease. However, male sex has been associated with increased risk of C. neoformans var gatii infection in non-HIV-related populations in Australasia , supporting a possible sex-linked (behavioural, biological or environmental) risk factor. The association with past herpes zoster infection was also unexpected. However, work in a separate cohort of HIV-infected adults in Uganda showed the cumulative incidence of herpes zoster infection had a linear relationship with duration of HIV infection and was not associated with more rapid progression or death . Therefore, herpes zoster is a marker of duration of HIV infection and immunosupression; once again, the association may represent incomplete adjustment for stage of disease in the multivariate model.
C. neoformans var grubii was responsible for all events and this is a consistent feature of HIV-associated cryptococcal disease . The reasons for the preponderance of this serotype are unclear. Some authors have suggested the possibility of a single global clone, with specific pathogenicity for individuals with HIV-related immunosuppression [26,29]. It is believed to be widespread in the environment, unlike other serotypes that have a limited environmental habitat, and its frequency of isolation may reflect this. A better understanding of the environmental residence of cryptococci and its relationship to disease is needed to help to plan appropriate avoidance and prevention strategies.
Survival following the development of clinical disease was poor. Effective management of established cryptococcal disease was not available to us and this continues to be the situation for most HIV-infected individuals in sub-Saharan Africa. Amongst the 69 definite cases, cryptococcal infection was believed to be the primary pathology leading to death. We did not identify alternative explanations antemortem and the lack of postmortem examinations and the advanced state of HIV-associated immunosupression does not allow the cause of death to be definitively ascribed. Irrespective of the contribution to death, cryptococcal meningitis is a distressing and painful illness that is compounded by rudimentary palliative care and difficulties in prescribing opiate analgesia in much of Africa. Reappraisal and further evaluation of prevention strategies are appropriate, given the frequency of disease, difficulties of providing therapy and poor outcome.
Intermittent fluconazole therapy was effective at preventing cryptococcal infections in studies in the United States . However, it shows little impact on survival  in a setting where cryptococcal disease is a small contributor to overall HIV mortality. The principal concerns over Fluconazole use in an African population are cost and generation of widespread azole resistance in other fungi. Appropriate targeting of at-risk groups would help to limit both. Use of a CD4 T cell count < 200 × 106 cells/l would be an appropriate marker of risk, but is unworkable in practice because of the lack of both funds and appropriate laboratory facilities. Clinical stage 3 and 4, while a less-sensitive indicator of risk than CD4 T cell counts, is simple and inexpensive for use in resource-limited health-care settings. Total lymphocyte counts were of limited value used in isolation and added little discriminatory power when combined with clinical stage (data not shown). A vaccine-based prevention strategy would, in principle, be the most relevant approach. In the absence of access to potent antiretroviral therapy in the foreseeable future for the majority of African populations, possible vaccine strategies and a previously described vaccine construct should be evaluated further [13,14].
Coinfection with C. neoformans is a frequent and serious problem in HIV-infected Africans. Under-recognition of the infection will occur if diagnostic tests are confined to patients with meningitis. Therapy of established disease is likely to remain difficult even with greater access to oral azole antifungal drugs. Strategies to use these agents as prophylaxis need assessment in the developing world; in the longer term, vaccine development should be deemed a priority in the control and prevention of cryptococcal disease.
We thank the staff and members of the TASO Entebbe office for their enthusiasm and support and Dr S. D. K. Sempala, director of the Uganda Virus Research Institute, for his encouragement.
1. Maher D, Mwandumba H. Cryptococcal meningitis in Lilongwe and Blantyre, Malawi. J Infect 1994, 28: 59–64.
2. Bergemann A, Karstaedt AS. The spectrum of meningitis in a population with high prevalence of HIV disease. Q J Med 1996, 89: 499–504.
3. Moosa MY, Coovadia YM. Cryptococcal meningitis in Durban, South Africa: a comparison of clinical features, laboratory findings, and outcome for human immunodeficiency virus (HIV)-positive and HIV-negative patients. Clin Infect Dis 1997, 24: 131–134.
4. Heyderman RS, Gangaidzo IT, Hakim JG. et al
. Cryptococcal meningitis in human immunodeficiency virus-infected patients in Harare, Zimbabwe. Clin Infect Dis 1998, 26: 284–289.
5. Hakim JG, Gangaidzo IT, Heyderman RS. et al
. Impact of HIV infection on meningitis in Harare, Zimbabwe: a prospective study of 406 predominantly adult patients. AIDS 2000, 14: 1401–1407.
6. Bogaerts J, Rouvroy D, Taelman H. et al
. AIDS-associated cryptococcal meningitis in Rwanda (1983–1992): epidemiologic and diagnostic features. J Infect 1999, 39: 32–37.
7. Gordon SB, Walsh AL, Chaponda M. et al
. Bacterial meningitis in Malawian adults: pneumococcal disease is common, severe, and seasonal. Clin Infect Dis 2000, 31: 53–57.
8. Archibald LK, den Dulk MO, Pallangyo KJ, Reller LB. Fatal Mycobacterium tuberculosis
bloodstream infections in febrile hospitalized adults in Dar es Salaam, Tanzania. Clin Infect Dis 1998, 26: 290–296.
9. Archibald LK, McDonald LC, Rheanpumikankit S. et al
. Fever and human immunodeficiency virus infection as sentinels for emerging mycobacterial and fungal bloodstream infections in hospitalized patients ≥ 15 years old, Bangkok. J Infect Dis 1999, 180: 87–92.
10. Arthur G, Nduba VN, Kariuki SM, Kimari J, Bhatt S, Gilks CF. Trends in bloodstream infections among human immunodeficiency virus-infected adults admitted to a hospital in Nairobi, Kenya, during the last decade. Clin Infect Dis 2001, 33: 248–256.
11. Powderly WG, Finkelstein D, Feinberg J. et al
. A randomized trial comparing fluconazole with clotrimazole troches for the prevention of fungal infections in patients with advanced human immunodeficiency virus infection.NIAID AIDS Clinical Trials Group.
N Engl J Med 1995, 332: 700–705.
12. Singh N, Barnish MJ, Berman S. et al
. Low-dose fluconazole as primary prophylaxis for cryptococcal infection in AIDS patients with CD4 cell counts of ≤ 100/mm3
: demonstration of efficacy in a positive, multicenter trial. Clin Infect Dis 1996, 23: 1282–1286.
13. Casadevall A, Mukherjee J, Devi SJ. et al
. Antibodies elicited by a Cryptococcus neoformans–
tetanus toxoid conjugate vaccine have the same specificity as those elicited in infection. J Infect Dis 1992, 165: 1086–1093.
14. Devi SJ. Preclinical efficacy of a glucuronoxylomannan–tetanus toxoid conjugate vaccine of Cryptococcus neoformans
in a murine model. Vaccine 1996, 14: 841–844.
15. French N, Nakiyingi J, Carpenter LM. et al
. 23-Valent pneumococcal polysaccharide vaccine in HIV-1-infected Ugandan adults: a double-blind, randomised and placebo controlled trial. Lancet 2000, 355: 2106–2111.
16. The WHO International Collaborating Group for the Study of the WHO Staging System. Proposed 'World Health Organization staging system for HIV infection and disease': preliminary testing by an international collaborative cross-sectional study. AIDS 1993, 7: 711–718.
17. Nunn AJ, Biryahwaho B, Downing RG. et al
. Algorithms for detecting antibodies to HIV-1: results from a rural Ugandan cohort. AIDS 1993, 7: 1057–1061.
18. Gilks CF, Ojoo SA, Ojoo JC. et al
. Invasive pneumococcal disease in a cohort of predominantly HIV-1 infected female sex-workers in Nairobi, Kenya. Lancet 1996, 347: 718–723.
19. Cox DR. Regression models and life tables. J Roy Stat Soc 1972, 34: 187–220.
20. Medical Research Council. MRC Guidelines for Good Clinical Practice in Clinical Trials
. London: HMSO; 1998.
21. Okongo M, Morgan D, Mayanja B, Ross A, Whitworth J. Causes of death in a rural, population-based human immunodeficiency virus type 1 (HIV-1) natural history cohort in Uganda. Int J Epidemiol 1998, 27: 698–702.
22. Lara-Peredo O, Cuevas LE, French N, Bailey JW, Smith DH. Cryptococcal infection in an HIV-positive Ugandan population. J Infect 2000, 41: 195.195.
23. Feldmesser M, Harris C, Reichberg S, Khan S, Casadevall A. Serum cryptococcal antigen in patients with AIDS. Clin Infect Dis 1996, 23: 827–830.
24. Sorvillo F, Beall G, Turner PA. et al
. Incidence and factors associated with extrapulmonary cryptococcosis among persons with HIV infection in Los Angeles County. AIDS 1997, 11: 673–679.
25. Oursler KA, Moore RD, Chaisson RE. Risk factors for cryptococcal meningitis in HIV-infected patients. AIDS Res Hum Retroviruses 1999, 15: 625–631.
26. Chen S, Sorrell T, Nimmo G. et al
. Epidemiology and host- and variety-dependent characteristics of infection due toCryptococcus neoformansin Australia and New Zealand.Australasian Cryptococcal Study Group.
Clin Infect Dis 2000, 31: 499–508.
27. Morgan D, Mahe C, Malamba S. et al
. Herpes zoster and HIV-1 infection in a rural Ugandan cohort. AIDS 2001, 15: 223–229.
28. Casadevall A, Perfect JR. Cryptococcus neoformans. Washington, DC: American Society of Microbiologists; 2001.
29. Franzot SP, Hamdan JS, Currie BP, Casadevall A. Molecular epidemiology of Cryptococcus neoformans
in Brazil and the United States: evidence for both local genetic differences and a global clonal population structure. J Clin Microbiol. 1997, 35: 2243–2251.