Longitudinal Clinical Findings and Outcome among Cryptococcus gattii Patients in British Columbia

Abstract

Cryptococcus gattii (Cg) infection emerged in British Columbia in 1999. A longitudinal, clinical description of patients has not been reported. Medical records were reviewed for Cg patients identified through surveillance (1999-2007). Risk factors for Cg mortality were explored using multivariate Cox regression; longitudinal patterns in serum cryptococcal antigen (SCrAg) titers and the probability of chest cryptococcomas over time were estimated using cubic B-splines in mixed-effects regression models. Among 152 patients, 111 (73.0%) were culture confirmed. Isolated lung infection was present in 105 (69.1%) patients; 47 (30.9%) had central nervous system infection, with or without lung involvement. Malignancy was the provisional diagnosis in 64 (42.1%) patients. Underlying diseases were present in 91 (59.9%) patients; 23 (15.1%) were immunocompromised, and 23 (15.1%) had asymptomatic disease. There were only 2 (1.8%) culture positive relapses, both within 12 months of follow-up. The estimated median time to resolution of lung cryptococcomas and decline in SCrAg titer to <1:8 was 2.8 and 2.9 years, respectively. Cg-related and all-cause mortality among culture-confirmed cases at 12 months' follow-up was 23.3% and 27.2%, respectively. Cg-related mortality was associated with age >50 years (hazard ratio [HR], 15.6; 95% confidence interval [CI], 1.9-130.5) and immunocompromise (HR, 5.8; CI, 1.5-21.6). All Cg-related mortality occurred among culture-positive cases within 1 year of diagnosis. Cryptococcomas and serum antigenemia were slow to resolve. However, late onset of failed therapy or relapse was uncommon, suggesting that delayed resolution of these findings does not require prolongation of treatment beyond that recommended by guidelines.