We interviewed persons from whom C. gattii serotype B was cultured through December 31, 2005, and who did not report contact with Vancouver Island or other known disease-endemic areas. We conducted telephone interviews by using a standard questionnaire to assess demographic information, travel history, risk factors for infection, underlying medical conditions, and clinical symptoms. Risk factors and travel exposures were assessed for the 1-year period before the onset of illness (or before diagnosis, in asymptomatic cases). Health authorities in neighboring provinces (Canada) and states (USA), where the disease is not reportable, were provided with case investigation forms and encouraged to investigate cryptococcal disease in immunocompetent persons.

Reports of animal cases were informally collected through veterinary networks in BC. Cases from the United States were reported by state veterinary epidemiologists. Infection in the animals was diagnosed histologically or identified as C. gattii serotype B by culture. None of the animals had traveled to Vancouver Island or other disease-endemic areas.

From October 2001 through December 2005, environmental sampling was undertaken in the BC mainland, the BC Gulf Islands, and the US Pacific Northwest. We sampled 22 map grids as defined by the 1:50,000-scale National Topographic System of Canada (NTS) and US Geological Survey (USGS) mapping system. Geographic data were assembled as described elsewhere (Kidd et al., unpub. data). Purposive sampling was conducted at selected sites and areas surrounding the homes of persons with C. gattii infection, those who reported travel to Vancouver Island and those who did not. Sampled environments included front and back yards, walking trails, public parks, and recreational areas. Trees, small woody debris, soil, air, and water were sampled as described elsewhere (Kidd et al., unpub. data).

Sample positivity was scored binarily. C. gattii concentration was expressed as CFU/gram, CFU/m³, and CFU/100 mL in soil, air, and water, respectively. The concentrations of multiple samples were described by the geometric mean and geometric standard deviation. When more than 1 sample was taken from a single sampling point (e.g., the same tree), only the first sample was included.

We initially cultured the samples on Staib media (13). Resulting dark brown colonies were grown on canavanine-glycine-bromothymol blue (CGB) agar (14) to differentiate C. gattii from C. neoformans and then serotyped (Crypto-check, Iatron Laboratories, Tokyo, Japan).

Molecular types were identified by a previously described PCR-based restriction fragment length polymorphism (RFLP) method (15), which was adapted for further discrimination of variation within the VGII molecular type (9). The URA5 gene was amplified as previously described (15) and then completely digested at 37°C in a 20-μL reaction containing 1× NEB2 buffer, 1× bovine serum albumin, and 4 U each of Hha I, Dde I, and BsrG I (New England Biolabs, Inc., Ipswich, MA, USA). RFLP products were subjected to electrophoresis and visualized on a 3% agarose gel prestained with ethidium bromide. Control strains were used for each possible C. gattii RFLP pattern: WM179 (VGI), NIH444 (VGIIa), RB28 (VGIIb), WM161 (VGIII), and WM779 (VGIV). C. neoformans strains WM148 (VNI), WM626 (VNII), WM628 (VNIII), and WM629 (VNIV) were also included.

Multilocus sequence typing was performed for selected isolates by using methods previously described (8) with the use of 2 additional loci, PLB1 and IGS (16). We isolated total DNA from histopathology specimens (n = 3) by using the DNeasy Tissue kit (QIAGEN Inc., Mississauga, Ontario, Canada). Cryptococcal-specific PCR-RFLP was conducted as described above. The internal transcribed spacer region (ITS1-5.8S-ITS2) was amplified and sequenced for identification of C. gattii–specific polymorphisms (17).

Five persons with culture-confirmed C. gattii, 3 in BC and 2 in Oregon, did not report exposure to Vancouver Island or other cryptococcal disease–endemic areas (Figure 2). Case-patients 1 through 4 received a diagnosis or reported symptom onset from September through December 2004. Case-patient 5, who had a fatal infection, received a diagnosis in December 2005.

Case-patient 1 was a 47-year-old man living in BC who was hospitalized with cough, chills, night sweats, nausea, loss of appetite, muscle pain, headache, and neck stiffness. Both lung and brain cryptococcomas were identified. He had chronic hepatitis C infection and a history of drug addiction. At the time of infection, he smoked 20–40 cigarettes/day. His residence, a farmhouse undergoing significant renovations, was located in NTS grid 092G/05 on the coast north of Vancouver. Environmental exposures included yard and landscaping work at this property.

Case-patient 2 was a 48-year-old woman living in BC who experienced shortness of breath, fever, chills, headache, night sweats, loss of appetite, nausea, and muscle pain. A lung mass was identified by computed tomography. She had no known underlying health conditions. She resided on the BC lower mainland within NTS grid 092G/02; her last visit to Vancouver Island was 4 years before the onset of her illness. In the year before onset, considerable deforestation had occurred near her residence to clear land for housing developments. During this period, she also traveled ≈1 day/week to garden centers and nurseries within NTS 092G/02 to obtain shrubs, trees, and new topsoil for yard landscaping she carried out at her residence.

Case-patient 3 was a 73-year-old woman living in BC who had chronic renal failure requiring dialysis and a history of lung disease and breast cancer. She was asymptomatic; a cryptococcal lung nodule was identified radiographically after she had hip surgery in December 2004. No nodule was apparent on imaging conducted <2 months earlier, which suggests recent acquisition. The patient resided in NTS grid 092G/07. She last visited Vancouver Island 14 years before her diagnosis. She had reduced mobility and consequently little outdoor exposure.

Case-patient 4 was a 59-year-old man living in Oregon who began to experience cough, shortness of breath, fever, chills, weight loss, nausea, and muscle pain in December 2004. He had undergone a kidney transplant in September 2003 and reported scarring of lung tissue due to his occupation. His place of residence was located within USGS grid 44123-A1B4. He had not traveled outside Oregon in the year before symptom onset.

Case-patient 5 was an 87-year-old man living in Oregon (USGS grid 45122-C5D8). He was hospitalized in December 2005 with meningitis, accompanied by fever, weight loss, and loss of appetite. His medical history included chronic lymphocytic leukemia, and he had taken oral steroids in the year before diagnosis. He had traveled to parts of Oregon, Washington, and Colorado during his exposure period.

In BC, a retrospective review of companion animal cases identified 8 culture-confirmed serotype B cases, which occurred in a ferret, a llama, and 6 cats. Specimens were collected from December 2003 through December 31, 2005; animal residences were located throughout the BC lower mainland (Figure 2).

In Washington, 3 cats with cryptococcal disease residing in USGS 48122-G1H4, close to the BC-USA border, were reported from February through June 2005 (Figure 2). All cases were diagnosed by histopathologic examination, and no cultures were obtained.

From October 2001 through December 2005, 3% of 2,033 off-island environmental samples were positive for C. gattii (Table 1). Swab samples included 45% of samples from trees and other structures (n = 925), 38% from soil (n = 781), 15% from air (n = 304), and 1% from water (n = 23).

Five positive air samples were recovered from 2 focal areas of the lower mainland at 2 different times (Figure 3). The first 2 were collected from air in NTS grid 092G/02 on the same day in October 2002. No other positive samples were detected in 2003 despite further sampling at this site and others. In July 2004, a third positive air sample was collected from grid 092G/02 and 2 more from grid 092G/01. Despite extensive sampling in these areas and other parts of the BC lower mainland (n = 1,140), C. gattii was not isolated from colonized sources such as trees or soil.

Among environmental samples taken outside Vancouver Island, C. gattii was most often recovered from the Gulf Islands (Table 1). NTS grid 092B/14 had the highest proportion of positive samples (52/220; 24%). Colonized sources were identified from a single Gulf Island within this grid in March, May, and June of 2004. In 2005, positive soil and tree samples were obtained in February, August, September, and October from 3 of the Gulf Islands (NTS grids 092B/14 and 092F/01) and Washington (USGS grid 48122-G1H4). From among 50 samples collected a month apart in Washington within 10 km of the BC border (USGS grid 48122-G1H4), a single soil sample and a swab of a fence post were positive. An additional 27 samples from other sites along the US side of the border (USGS grids 48122-G1H4 and 48122-G5H8) were negative, as were all 197 samples collected from several areas of Oregon (USGS grids 44123-A1B4 and 45122-C5D8). The geometric mean concentration of detected C. gattii among soil, water, and air samples taken from various sites outside Vancouver Island is summarized in Table 2.

Table 3 and Figure 4 summarize the molecular subtyping results for human and animal isolates from persons and animals with no recent exposure to Vancouver Island, as well as environmental isolates obtained from off-island locations.

Of the 5 human cases, 3 were attributed to the VGIIa molecular type, 1 to the VGIIb molecular type, and 1 to the VGI molecular type. However, although all 3 molecular types have been identified among clinical and environmental isolates from Vancouver Island, multilocus sequence typing (MLST) results indicated that both the VGIIa and VGIIb strains from Oregon cases were genetically distinct from previously characterized Vancouver Island isolates (9,16). The Oregon case 4 VGIIa isolate differed from Vancouver Island VGIIa at 1 locus, while Oregon case 5 VGIIb differed from Vancouver Island isolates at 4–5 loci, where it was more similar to Vancouver Island VGIIa than VGIIb (Table 4).

Cryptococcal DNA isolated from the formalin-fixed, paraffin-embedded tissue of 3 cats in Washington belonged to the VGIIa molecular type. MLST profiles could not be determined in these cases because of the relatively poor quality and yield of DNA from the fixed tissue.

Most off-island environmental isolates that were typed belonged to the VGIIa molecular type. These included 4 of the 5 isolates from lower mainland air samples (the fifth could not be separated from contaminants) and 90% of 20 typed isolates from the NTS grid with the highest proportion of positive off-island environmental samples (092B/14). All tested environmental VGIIa isolates from BC and Washington possessed identical MLST profiles to those of representative isolates from Vancouver Island (Table 3).