Predictors of outcome in Cryptococcus neoformans var. gattii meningitis

Q J Med 1996; 89:423-428 Predictors of outcome in Cryptococcus neoformans var. gattii meningitis R.A. SEATON 1 ' 2, S. NARAQI 2, J.P. WEMBRI 2 and D.A. WARRELL 1 From f/ie 1 Centre for Tropical Medicine, Nuffield Department of Medicine, University of Oxford, UK, and 2 The Department of Clinical Sciences, Medical Faculty, University of Papua New Guinea, Papua New Guinea Received 14 December 1995 and in revised form 12 February 1996 Summary In Papua New Guinea, Cryptococcus neoformans var. gattii meningitis has a high fatality rate even in immunocompetent patients. Our retrospective study attempted to identify marker of poor prognosis. Of 88 immunocompetent patients, 30 (34.1%) died, usually soon after admission, and mortality was higher in men (p = 0.025) and older patients (p= 0.039). Death was associated with altered consciousness ( p < 0.001), a history of convulsions prior to treatment (p = 0.002) and a maximum systolic blood pressure of >150 mmhg (p = 0.017). These data suggest that death results from raised intracranial pressure and subsequent tentorial herniation. However, CSF opening pressure measured on admission was raised in 29/36 (81 %) patients and did not predict outcome. In survivors, relapse was uncommon and was not predicted by discharge serum cryptococcal antigen titres, which were frequently raised on completion of therapy in asymptomatic patients. Mortality may be reduced if efforts are made to lower intracranial pressure in those patients who present with markers of poor prognosis. Introduction Cryptococcus neoformans is an encapsulated fungus which causes a life-threatening meningo-encephalitis. Two varieties of this organism are recognized, var. neoformans and var. gattii. The variety neoformans, which has been well studied, occurs worldwide 1 and usually causes disease in immunosuppressed individuals, 2 " 4 particularly patients with HIV infection. 5 In contrast, var. gattii infection occurs predominantly in immunocompetent individuals 6 " 8 living in tropical or sub-tropical regions. 1 In the Central province of Papua New Guinea (PNG), cryptococcal meningitis is the commonest type of chronic meningitis with an estimated annual incidence of 42.8 cases/million population. The majority of patients have no underlying immunosuppressive state, and are infected with the variety gattii. In PNG, the in-hospital case fatality rate in immunocompetent patients has been reported as 22.4%, with an overall case fatality rate of 40.8%. Patients are frequently disabled with blindness, which occurs in 31% of survivors. 9 These data contrast with other published studies in immunocompetent patients which have reported a much lower fatality rate with few or no visual complications. 6 ' 7 To improve our understanding of the pathogenesis of this infection and so potentially to improve patient management, this study investigated those demographic, clinical and laboratory variables which might predict a poor outcome in immunocompetent patients. Our investigation into those factors associated with visual loss have been reported elsewhere. 10 ' 11 Methods The study was based in the Port Moresby General Hospital (PMGH), a 600-bed university teaching Address correspondence to Dr R.A. Seaton, Infection and Immunodeficiency Unit, King's Cross Hospital, Clepington Road, Dundee DD3 8EA Oxford University Press 1996

424 R.A. Seaton etal. hospital in the nation's capital. Retrospective data were collected from the casenotes of patients admitted to the Port Moresby General hospital between 1979 and December 1992 and nationwide prospective data were collected from January 1993 to September 1995. using x 2 statistics or Fisher's exact test (two-tailed). Relative risks (RR) and 95% CIs for death were calculated. Continuous data were compared using the Wilcoxon two-sample test or student t-test as appropriate. Calculated p values of <0.05 were considered significant. Diagnosis Diagnosis of cryptococcal meningitis was based on a history of chronic headache and vomiting with the finding of budding yeasts on India-ink-stained cerebrospinal fluid (CSF), cryptococcal antigen titres in the CSF or serum >1 :32, or a positive culture of CSF on Sabouraud's media. The variety type was determined by subculturing colonies on canavinine glycine-bromothymol blue (CGBB) agar; var. gattii gives the medium a characteristic cobalt blue colour. 12 Biotyped cases were confirmed serologically by the Australian mycology reference laboratory in Adelaide. Inclusion and exclusion criteria Patients whose cultures were confirmed as var neoformans, and those who had underlying immunosuppression including HIV infection were excluded. Only patients whose outcomes were known were included. Patients who absconded from hospital during their treatment course were excluded. Data collection Clinical and laboratory data were recorded along with CSF findings in a proportion of patients who underwent lumbar puncture. CSF pressure was categorized as 'raised' if the opening pressure exceeded 200 mm of CSF. Cryptococcal antigen concentration was measured in both CSF and serum by a latex agglutination method (Crypto-LA, Carter Wallace). The majority of sera were heat- and proteaseinactivated prior to testing. Anti-cryptococcal therapy All patients received standard anticryptococcal therapy with parenteral amphotericin B (0.3-1 mg/ kg/day) and oral flucytosine (150 mg/kg/day in divided doses) for more than 4 weeks, except those who died shortly after admission. Data storage and statistical analysis All data were recorded on standardized proformas and stored and analysed on the EPI-info 5 statistical software package. The proportion of patients who died was calculated for a series of clinical and laboratory variables and results were compared by Results A total of 88 patients met the stated inclusion criteria. Of the 49 patients previously reported between 1979 and November 1990, 9 37 had sufficient information in the case notes to merit inclusion. From November 1990 to January 1993, a further 12 patients were identified from case records. Prospective studies began in January 1993, continuing until September 1995, and a total of 60 patients were identified nationwide, of whom 39 met the inclusion criteria. The reasons for exclusion were lack of clinical data (13), absconding early in the treatment course (2), HIV infection diagnosis (4) and C. neoformans var. neoformans infection (2). Thirty-nine patients had CSF cultured, and all were positive for C. neoformans. Of the twentyseven CSF isolates biotyped on CGBB agar, all were confirmed as var. gattii. Consecutive CSF isolates biotyped during the prospective study period revealed 18/19 (95%) isolates in immunocompetent patients to be of the variety gattii. A total of 30 (34.1%) patients died during their hospital admission. Variables examined for an association with a poor outcome are given in Tables 1 and 2. The fatality rate was significantly higher in male patients (p = 0.025) and those who died were older on admission (p = 0.039). The time that symptoms of meningitis had been present prior to admission was similar in the two groups and did not predict outcome (p = 0.956). However, those patients who presented with a history of convulsions were significantly more likely to die (p = 0.002). Of the clinical signs on admission, only an altered level of consciousness (ranging from confusion to obtundation and coma) predicted an outcome of death (p<0.001). A maximum blood pressure reading of >150mmHg at any time during admission was also significantly associated with death, but other signs such as papilloedema, abducens nerve palsy and CSF opening pressure of greater than 200 mm/csf were not. No associations were found between death and other CSF findings, including cryptococcal antigen titres. However, patients who died tended to have higher serum cryptococcal antigen titres. Patients who died did not survive long after admission. The median time from admission to death was 8 days (range 1-110 days), and the mean dose

Predicting outcome in gattii meningitis 425 Table 1 Clinical and demographic variables on admission examined for an association with an outcome of death Died (n = 30) Survived (n = 58) RR (95% Cl) P Male sex Urban residence Age (years) Symptoms (weeks) Altered consciousness Convulsions Papilloedema Maximum systolic BP >150mmHg Cranial nerve palsy on admission Abducens palsy Other neurological signs 25/30 (83.3%) 12/25 (48%) 30(15-55) 5 (1-17) 27/30 (90%) 13/27 (48.1%) 19/28 (67.9%) 8/23 (34.8%) 7/27 (25.9%) 2/27 (7.4%) 5/30 (16.7%) 33/58 (56.9%) 34/57 (59.6%) 24 (7-46) 4 (1-20) 15/58 (25.9%) 8/57 (14%) 43/53 (81.1%) 4/40 (10%) 18/57 (31.6%) 12/57 (21.1%) 4/58 (6.9%) 2.59 (1.1-6.07) 0.72 (0.38-1.39) 9.86 (3.23-30.12) 2.79 (1.57-4.93) 0.65 (0.35-1.18) 2.44 (1.36-4.4) 0.83 (0.4-1.7) 0.4 (0.1-1.5) 1.76 (0.9-3.42) 0.025 0.46 0.039 0.956 < 0.001 0.002 0.287 0.017 0.784 0.208 0.264 For continuous data, median values are shown with the range given in brackets. For categorical data, proportions are followed by percentages in brackets. RR (95% Cl), relative risk (95% confidence interval). Table 2 Investigation results on admission examined for an association with an outcome of death Died (n = 30) Survived (n = 58) RR (95% Cl) CSF opening pressure ^200 mm of CSF CSF India ink stain positive Serum cryptococcal antigen titre CSF cryptococcal antigen titre Total CSF white cell count per mm 3 CSF protein (g/l) CSF glucose (mmol/l) 7/10(70%) 22/26(84.6%) 0.56(0.19-1.64) 17/19(89.5%) 33/35(94.3%) 0.68(0.24-1.95) 1 :4096 (-1 :262144) 1 :2048 (1 :32-1 :65536) 1 :2048 (-16384) 31.5 (0-714) 0.83 (0.5-6.3) 2.5 (0.5-8.3) 1 :2048 (0-32768) 91 (0-1434) 0.84 (0.2-3.45) 1.9 (0.1-5.0) 0.370 0.607 0.092 0.779 0.139 0.984 0.146 For continuous data, median values are shown with the range given in brackets. For categorical data, proportions are followed by percentages in brackets. RR (95% Cl), relative risk (95% confidence interval). of amphotericin B received was 0.169 g (SD 0.462). In contrast, survivors were treated for a median time of 10.5 weeks (4-52 weeks) and received a mean dose of 1.6 g (SD 0.92) amphotericin B. The mean follow-up for survivors was 15 months (range 2-132 months). Modes of death In all but two of the patients in whom the mode of death was documented, death occurred following coma and cardiorespiratory arrest. This was associated with raised systolic blood pressure in eight (Table 1). Death did not appear to be precipitated by lumbar puncture, which was performed in 64 patients. In one patient, death occurred following a cardiorespiratory arrest late in the treatment course, and followed an infusion of amphotericin B. In a further patient, treatment was complicated by jaundice and acute renal failure. Only one autopsy was performed, and this demonstrated widespread meningo-encephalitis with small cerebral abscesses. Sequelae of infection in survivors Of the 58 patients who survived, six later relapsed and one died on the second admission. Three patients required long-term treatment with anticonvulsants following cure of meningitis, one who had convulsions prior to treatment and two who developed convulsions during therapy. Visual loss and hearing loss were common sequelae. Visual loss was documented in 30/57 (52.6%) of survivors and 21/57 (36.8%) were blind. A full description of the patterns of visual loss and the value of corticosteroid treatment in preventing visual deterioration is given else-

426 R.A. Seaton etal. where. 10 ' 11 Pure tone audiometry was performed in 35 survivors, of whom 22 (62.9%) were documented to have some degree of sensorineural hearing loss. Two patients who survived were blind with severe bilateral sensorineural deafness. Cryptococcal antigen titres and CSF findings in survivors In twelve patients followed prospectively (eight of whom were confirmed to have infection with var. gattii), serum cryptococcal antigen titres were measured before and on completion of anticryptococcal therapy as well as at follow up (Table 3). By completion of therapy, all patients were asymptomatic having received curative doses of amphotericin B. In 11/12 patients, serum cryptococcal antigen titres fell during treatment although in 10/12 titres were >1 :32 on discharge. Despite these high discharge titres patients remained asymptomatic and in 9/12 further falls in titres occurred despite receiving no further anticryptococcal therapy. Of 15 patients in whom India ink examination was carried out before and after treatment, nine (60%) had persistence of encapsulated C. neoformans although cultures were negative, and none relapsed. Discussion This study has attempted to identify those features of Cryptococcus neoformans var. gattii meningitis on admission which will predict a poor outcome. Although it is possible that not all our patients were infected with var. gattii, 95% of CSF isolates in the prospectively studied immunocompetent patient were confirmed as this variety. We have no reason to suspect that prior to 1993 the proportion of var. gattii isolates in immunocompetent patients was any different. The likelihood of inclusion of var. neoformans has been further reduced by excluding all patients who had evidence of underlying immunosuppressive disorders. We have confirmed earlier reports that this disease has a high mortality (34.1%) as well as a high rate of serious sequelae in survivors. In Australia, deaths due to var. gattii meningitis are less common, occurring in 0/20 patients in one series 7 and 4/26 (15.4%) in a second series. 6 In non-immunosuppressed patients in Brazil, however, where var. gattii accounts for 58.3% of infections, the mortality rate of 38.6% was comparable to that in our series. 8 In our investigation for poor prognostic signs we identified a number of associations. A significantly higher proportion of men died, and those patients who died tended to be older than the survivors. Altered consciousness and a history of convulsions were strongly predictive of death, as was a maximum systolic blood pressure >150mmHg. Altered consciousness has previously been identified as an important adverse prognostic sign in AIDS-related cryptococcal meningitis 13 as well as in predominantly healthy patients with cryptococcal meningitis in Singapore. 14 Altered consciousness, convulsions and raised systolic blood pressure may reflect raised intracranial pressure and subsequent tentorial herniation. However, CSF opening pressure, measured by lumbar puncture on admission, was frequently raised in both those who died and those who survived, and was therefore not a good predictor of outcome. It has recently been shown that CSF opening pressure is not significantly associated with an outcome of visual loss in PNG. 10 In PNG, neuroradiological scans are not available, but elsewhere cerebral CT scan abnormalities have Table 3 Serum cryptococcal antigen titres following cessation of therapy Patient Cryptococcal antigen titres in sera Pre-treatment Immediate post-treatment Follow-up Months from discharge Total dose of amphotericin B (g) 1 2 3 4 5 6 7 8 9 10 11 12 1 :8192 1:1024 1:8192 1:8192 1:1024 1:512 1:32 1:4 1 :64 1 :16 1 :64 1 :64 1 :16 1 :4 1:8 0 1:64 0 0 22 8 3 8 11 5 12 12 9 13 16 6 0.746 3.237 2.500 2.340 2.425 1.930 0.900 1.200 1.250 2.585 1.860 1.000

Predicting outcome in gattii meningitis 427 been commonly observed in immunocompetent patients with cryptococcal meningitis. Mass lesions, hydrocephalus, oedema and other cerebral abnormalities have been documented in 57-69% of var. gattii cases in Australia 6 ' 7 and in Brazil 55% of immunocompetent patients had cerebral CT abnormalities. 8 In Australia, neurosurgical procedures are frequently done, and this probably accounts for the lower mortality there observed. 6 ' 7 CSF findings on admission were not significantly different between the two outcome groups, although those who died tended to have higher serum cryptococcal antigen titres and lower white-cell counts in the CSF. Diamond and Bennett identified high CSF opening pressure, low glucose concentrations, low CSF leucocyte counts, as well as high cryptococcal antigen titres in the CSF and serum, as associated with a poor outcome. 15 The more marked differences in CSF findings in this American series perhaps reflect the high proportion of immunosuppressed patients who fail to mount a good immune response to infection. These differences may also reflect the variety type, with Americans infected predominantly with the variety neoformans. In AIDS-related cryptococcal meningitis, high CSF cryptococcal antigen titres were also predictive of death. 13 Survivors were followed up for a median time of 15 months, and relapse occurred infrequently, in 6/58 (10%) patients. Despite relapse, five of these patients were successfully treated. In Australia, 20% of patients with gattii relapsed. 6 In HIV-infected patients with cryptococcal meningitis, relapse occurs in up to 50% of patients who complete therapy and are culture-negative, 16 and maintenance therapy is usually required. Relapse in HIV infected patients has been shown to be a true recrudescence 17 due to subclinical foci of infection. We did not find discharge cryptococcal antigen titres to be a useful predictor of relapse. In the twelve patients in whom serial serum cryptococcal antigen titres were measured, high discharge titres continued to decline in the majority despite no further treatment being given. This probably reflects dead, non-viable antigen agglutinating with the test reagent, or a lower load of antigen which can be managed by these immunocompetent patients. Likewise, examination of CSF on completion of treatment frequently showed persistence of encapsulated organisms with negative culture results. We did not document a significant rise in titre of antigen in any patient which might be more indicative of relapse as has been observed in HIV-antibody-positive patients. 18 These data help to better understand the pathophysiology of C. neoformans var gattii meningitis. It is suggested that the high mortality observed in this series results from the untreated effects of raised intracranial pressure. If survival is to be improved, we suggest more aggressive measures be undertaken to lower intracranial pressure in patients who present with an altered state of consciousness, convulsions or raised systolic blood pressure. Prospective studies of outcome in patients with poor prognostic signs treated with aggressive measures to lower intracranial pressure should help to confirm or refute these preliminary observations. Acknowledgements Andrew Seaton was funded by the Colt Foundation, UK. We are grateful to Professor Isi Kevau and Drs Barnabus Mavo, Adolf Saweri, James Amini, Ben Miam, Jo Amban, David Lalloo and Ian Laurenson, who gave permission to include their patients in this series. Mrs Trixie Wain-Marjen and Mr Paul Kila performed the cryptococcal antigen tests, and Dr David Ellis of the Australian mycology reference laboratory, Adelaide kindly performed confirmatory testing of the CSF isolates. Thanks also to Professor A. Seaton for critical reading of the manuscript. References 1. Kwon-Chung KJ, Bennett JE. Epidemiologic differences between the two varieties of Cryptococcus neoformans. Am J Epidemiol 1984; 120: 123-30. 2. Collins VP, Gelhorn A, Trimble JR. Coincidence of cryptococcosis and disease of reticuloendothelial and lymphatic systems. Cancer 1951; 4: 883. 3. Schroter GPJ, Temple DR, Husberg BS, Weil R, Starzl TE. Cryptococcosis after renal transplantation: Report of ten cases. Surgery1976; 79(3): 268-77. 4. Kaplan MH, Rosen PP, Armstrong D. Cryptococcosis in a cancer hospital. Cancer 1977; 39: 2265-74. 5. Kovacs JA, Kovas AA, Polis M, Wright C, Gill VJ, Tuazon CU, Gelmann EP, Lane HC, Longfield R, Overturf G, Macher AM, Fauci AS, Parrillo JE, Bennett JE, Masur H. Cryptococcosis in the Acquired Immunodeficiency Syndrome. Ann Intern Med. 1985; 103: 533-8. 6. Mitchell DH, Sorrell TC, Allworth AM, Heath CH, McGregor AR, Papanaoum, Richards MJ, Gottlieb T. Cryptococcal disease of the CNS in Immunocompetent Hosts: Influence of Cryptococcal variety on Clinical Manifestations and Outcome. Clin Infect Dis 1995; 20: 611-16. 7. Speed B, Dunt D. Clinical and host differences between infections with the two varieties of Cryptococcus neoformans. Clin Infect Dis 1995; 21: 28-34. 8. Rozenbaum R, GoncalvesJR. Clinical epidemiological study of 171 cases of cryptococcosis. Clin Infect Dis 1994; 18: 369-80. 9. Lalloo D, Fisher D, Naraqi S, Laurenson I, Temu P, Sinha A, Saweri A, Mavo B. Cryptococcal meningitis (C. neoformans var. gattii) leading to blindness in previously healthy Melanesian adults in Papua New Guinea. QJ Med 1994; 87: 343-9. 10. Seaton RA, Verma N, Naraqi S, Wembri J. Ocular

428 R.A. Seaton et al. manifestations of Cryptococcus neoformans var. gattii meningitis in immunocompetent patients. 1. Predictors of visual loss (submitted). 11. Seaton RA, Verma N, Naraqi S, Wembri J. Ocular manifestations of cryptococcal meningitis in immunocompetent patients. 2. Effect of corticosteroids (submitted). 12. Kwon-Chung KJ, Polacheck I, Bennet JE. Improved diagnostic medium for separation of Cryptococcus neoformans var. neoformans (serotypes A and D) and Cryptococcus neoformans var. gattii (serotypes B and C). J Clin Microbiol 1982; 15: 535-7. 13. Saag MS, Powderly WC, Cloud GA, Robinson P, Crieco MH, Sharkey PK, Thompson SE, Sugar AM, Tuazon CU, Fisher JF, Hyslop N, Jacobson JM, Hafner R, Graybill JR, Dismukes WE, and the NIAD mycoses study group and the AIDS clinical trials group. Comparison of amphotericin B with fluconazole in the treatment of acute AIDs associated cryptococcal meningitis. N Engl J Med 1992; 326: 83-9. 14. Tjia TL, Yeow YK, Tan CB. Cryptococcal meningitis. J Neurol Neurosurg Psychiatr 1985; 48: 853-8. 15. Diamond RD, Bennett JE. Prognostic factors in cryptococcal meningitis. Ann Intern Med 1974; 80: 176-81. 16. Zuger A, Louie E, Holzman RS, Simberkoff MS, Rahal JJ. Cryptococcal disease in patients with the Acquired Immunodeficiency syndrome. Ann Intern Med 1986; 104: 234-40. 17. Spitzer ED, Spitzer SG, Freundlich LF, Casadevall A. Persistence of initial infection in recurrent Cryptococcus neoformans meningitis. Z.anceM993; 341: 595-6. 18. Powderly WG, Cloud GA, Dismukes WE, Saag MS. Measurement of cryptococcal antigen in serum and cerebrospinal fluid: Value in the management of AIDSassociated cryptococcal meningitis. Clin Infect Dis 1994; 18:789-92.