Risk factors for C. gattii infection were evaluated in 2 ways. First, a case–control study was conducted on a subset of case-patients (1999–2001) to examine which medical and environmental exposures increased the odds of infection. Second, risk behaviors and underlying medical conditions of all case-patients (1999–2007) were compared with those of the general BC population by using existing data from population-based surveys and studies.

This controlled study identifies demographic, medical, and behavioral risk factors for C. gattii acquisition. Notably, ≈40% of C. gattii–infected patients were immunocompromised. This represents a departure from the prevalent view that C. gattii infects otherwise healthy persons. However, the proportion of immunocompromised patients remains lower than estimates of immunocompromise for patients infected by C. neoformans, many of whom have severe immunologic disorders (26).

In contrast to previously reported findings of an association between cryptococcal species infection and male sex, even when adjusting for HIV status (14), C. gattii patients in British Columbia were not statistically more likely to be male. It has been hypothesized that sex differences may be the result of differential environmental exposure during occupational or leisure activities. In British Columbia, sex differences are unlikely, given the widespread detection of the fungus in the environment, including urban and semiurban settings. Previous studies that have demonstrated sex differences were not species specific and likely included a disproportionate number of persons with C. neoformans infection; thus, gender may not influence acquisition of C. gattii specifically.

C. gattii patients in British Columbia were more likely than the provincial population to be >50 years of age. The relative absence of pediatric case-patients (4 patients in 9 years) confirms reports in other jurisdictions (16). This finding may have resulted from differential exposure, as suggested by serosurveys in Australia (16). In a New York study, however, most children had acquired antibodies to C. neoformans by the age of 5 years (27). Alternately, advancing age may be a risk factor simply because older persons are more likely to have medical risk factors for C. gattii infection or have had their immune function decline with age. In the case–control analysis, no significant difference in age was detected between case-patients and controls; therefore, this relationship could not be studied. The lack of age association between case-patients and controls may have been because of bias toward the inclusion of older controls because they were selected on the basis of a recent chest radiograph showing no abnormalities—a procedure more likely to be routinely performed in older patients.

In the case–control study, C. gattii–infected case-patients were significantly more likely than controls to have received systemic corticosteroid therapy. Increased susceptibility to cryptococcal infection after corticosteroid therapy has previously been described (28). Mouse experiments by Levine et al. in the 1950s demonstrated that mice, which usually were able to localize an induced subcutaneous cryptococcocal infection, could not do so after an intraperitoneal injection of 2.5 mg of cortisone acetate (10). Among US patients with culture-confirmed cryptococcal meningitis, persons who had received >20 mg of prednisone or who had Hodgkin lymphoma or other lymphoreticular malignancies were at significantly higher risk of dying than those without these conditions (11). In another case series of Cryptococcus-infected patients, disseminated disease developed in 59% of those treated with corticosteroids (12). More recently, corticosteroid use has been identified as a risk factor for disseminated disease in HIV-negative patients with pulmonary cryptococosis (29). Because oral steroids are sometimes used to treat cryptococcal infections, these findings present a medical challenge.

In the case–control study, a history of pneumonia or other lung conditions (including emphysema, chronic bronchitis, COPD, sarcoidosis) was associated with cryptococcosis. Pneumonia alone may not be an independently associated risk factor because it is associated with many lung conditions. Certain respiratory conditions, including bronchial obstruction, predispose the patient to recurrent bacterial pneumonia (30,31), and bronchopulmonary disorders may predispose a person to pulmonary cryptococcosis (32) because airway damage leads to a decreased barrier to infection. C. gattii–infected case-patients were not more likely than population controls to have experienced COPD, when that factor was examined alone. Notably, asthma was not identified as a risk factor by either the case–control or population-level analyses, although this condition, along with COPD, has been suggested as a risk factor (31).

Invasive cancers of all types were significantly more common in C. gattii–infected patients than in the general population. This finding held true for subgroups of patients with lung cancer and leukemia/lymphoma. Lymphoproliferative malignancies have traditionally been considered risk factors for cryptococcal infection and disseminated disease (14), and this appears to remain true for C. gattii–infected populations specifically. Although invasive cancers likely lead to an increased risk for infection, C. gattii infection may have been an incidental finding in patients with lung cancer, with the detection of the former facilitated through imaging and invasive procedures used for diagnosing the latter. In the case–control study, MORs also suggested that odds of infection were elevated for those with a history of invasive cancers; however, this difference was not significant.

During 1999–2007, six cases of culture-confirmed C. gattii infections were documented in HIV-positive residents of British Columbia. Although such cases are recognized (33,34), HIV infection has generally been associated with C. neoformans infection. This study highlights that HIV infection, although rare in C. gattii patients, is significantly more likely to be found in this patient group than in the general population. In fact, the strength of this relationship may have been underestimated: since HIV-positive persons with cryptococcal infection of undetermined species were excluded from the case definition, some C. gattii infections among HIV-positive patients may have been missed. Although the specific reason is unknown, HIV status may be associated with C. gattii infection because of differences in strain pathogenicity and host immune response.

Three outdoor activities—chopping wood, pruning, and cleaning up branches—were protective against infection. The reason for this is not clear because these activities would potentially have exposed the person to C. gattii in trees. Ongoing low-level exposure possibly results in immunity, preventing symptomatic infection if the person is exposed to a higher dose. More likely, these more vigorous activities are undertaken by healthy persons and reflect a general state of wellness rather than a protective effect of these specific behaviors. In contrast, outdoor building repairs represented a risk factor for infection. This may have been due to the disturbance of rotting wood colonized with C. gattii. Environmental experiments in British Columbia have demonstrated substantially increased airborne concentrations of C. gattii when colonized trees are felled or chipped (35).

Population comparisons suggest that significantly more C. gattii–infected case-patients are current smokers. Because smoking compromises the protective barriers in airways, it has been linked to increased risk for many lung infections (36). A previous study of AIDS patients with C. neoformans infection, which identified smoking as a risk factor for disseminated disease, suggested that fungus may be inhaled along with smoke particulates and deposited in small airways (17). Air-sampling studies of C. gattii in British Columbia have found that the airborne propagule is sufficiently small to be deposited in the upper lung (i.e., from 3.3 μm to >7 μm (37).

This study has several limitations. First, to ensure that asymptomatic C. gattii–infected patients were not included as controls, we required controls to have had chest radiographs showing no abnormalities. A selection bias may have been introduced, which could have reduced the likelihood of showing significant differences among variables, such as smoking, that might be associated with having a radiograph taken, leading to more conservative estimates or an inability to detect true differences in exposures between case-patients and controls. Similarly, the small sample size of the case–control study may have limited its power to detect significant exposures. Therefore, when possible, population comparisons were also performed to validate case–control findings.

In addition, risk factors for both analyses were collected by self-report, which may have been inaccurate. Also, the case–control analysis examined risk factor information in the 3 months before disease onset. Subsequent research has shown that the incubation period for this disease is typically longer (38). This long incubation period may have affected the study’s ability to accurately detect behavioral and environmental risk factors; medical risk factors were unaffected because these were reported as “ever diagnosed with.” Despite a case definition specifically designed to exclude case-patients infected with C. neoformans, some misclassification may have occurred, given that only 50% of infections in case-patients included in the case–control study and 57% of infections in case-patients included in population comparisons could be confirmed by laboratory typing.

Although most cases of C. gattii infection do occur in otherwise healthy persons, our findings suggest that infection is associated with immunosuppressive states induced by oral corticosteroid use and invasive cancers as well as with weakened pulmonary function resulting from previous lung infections and smoking (Table 3). Despite HIV infection being rare among C. gattii–infected case-patients, it also occurred more frequently in infected persons than in the general population. Recent evidence indicates that C. gattii is now spreading in the Pacific Northwest of the United States (19,39). Physician awareness of risk factors should assist with diagnosis of this serious but treatable infection in areas where the disease is emerging.