Epidemic of Invasive Pneumococcal Disease, Western Canada, 2005–2009

A single clone of Streptococcus pneumoniae serotype 5 caused this epidemic.

B efore the advent of antimicrobial drugs, outbreaks of invasive pneumococcal disease were numerous. Since then, however, outbreaks have been less frequently reported and have involved fewer persons, usually those confi ned to closed settings such as hospitals or military barracks (1,2). Even more rare have been large outbreaks or epidemics of invasive pneumococcal disease; if and when they do occur, they tend to be caused by a limited number of pneumococcal serotypes (2)(3)(4).
The serotype of a Streptococcus pneumoniae bacterium is designated according to the organism's polysaccharide capsule, its major virulence factor. Worldwide, 91 polysaccharide capsular serotypes have been identifi ed (5,6). A small subset of serotypes is responsible for most large outbreaks; these serotypes typically include, but are not restricted to, serotypes 1, 4, 5, 9V, 12F, and 23F (2).
Before 2005, large outbreaks of pneumococcal disease, including invasive pneumococcal disease caused by serotype 5, were rare in Canada. In 2002, an outbreak caused by S. pneumoniae in northern Quebec, Canada, was reported, and blood culture identifi ed 10 cases as being caused by a serotype 1 strain (7). We report a large epidemic of invasive pneumococcal disease caused by S. pneumoniae serotype 5 in Canada that occurred during [2005][2006][2007][2008][2009]. The study received approval from the institutional research review committees of the health regions and the University of Alberta ethics review board.

Materials and Methods
In Canada, invasive pneumococcal disease is nationally notifi able. For this study, cases of invasive pneumococcal disease were defi ned according to the national case defi nition: isolation of S. pneumoniae from a normally sterile site, such as blood, cerebrospinal fl uid, pleural fl uid, biopsy

Epidemic of Invasive Pneumococcal
Disease, Western Canada, [2005][2006][2007][2008][2009] tissue, joint aspirate, pericardial fl uid, or peritoneal fl uid (8). In the provinces affected by the 2005-2009 epidemic, clinical diagnostic microbiology laboratories were required by provincial health authorities to submit isolates from patients with invasive pneumococcal infections to their respective provincial laboratories, which would then send them to the National Centre for Streptococcus, in Edmonton, Alberta, for capsular serotyping and antimicrobial drug resistance epidemiologic profi ling. For this study, 1 isolate per case was counted. Multiple isolates of the same serotype collected from the same patient within a 30-day period were considered to account for 1 case. Regardless of serotype, isolates collected from the same patient >30 days after the fi rst isolate were counted as separate cases.

Clinical Data Collection
To elucidate features of disease caused by S. pneumoniae serotype 5, we reviewed all cases of invasive pneumococcal disease in the northern Alberta area reported from 2005 through 2009. During the study period, Alberta was subdivided into 9 health regions. For cases originating in health regions 4 through 9 (located in northern Alberta), an extensive medical chart review was conducted. The total population for these health regions in 2008 was 1,888,881 (www.health. alberta.ca/documents/Population-Projections-2006.pdf). The clinical data collected were patient age, sex, aboriginal status (i.e., First Nations heritage), homelessness, substance abuse, type of invasive pneumococcal disease, outcome, and concurrent conditions (Table 1).

Identifi cation and Serotyping
As part of its serotyping program, the National Centre for Streptococcus pneumococcal surveillance confi rmed isolates as S. pneumoniae according to morphologic appearance and optochin susceptibility (9). All pneumococcal isolates that exhibited a positive quellung reaction when commercial type-specifi c antiserum (Statens Serum Institute, Copenhagen, Denmark) was used were assigned a serotype (10). Strains that were susceptible to optochin but for which no serotype was assigned were further tested by using the AccuProbe Streptococcus pneumoniae Culture Identifi cation Test (Gen-Probe, San Diego, CA, USA) to confi rm species identifi cation.

Antimicrobial Drug Susceptibility Testing
Drug susceptibility was determined by using the reference broth microdilution method described by the Clinical and Laboratory Standards Institute (11). The following antimicrobial drugs were tested: penicillin, cefotaxime, ceftriaxone, chloramphenicol, erythromycin, clindamycin, tetracycline, trimethoprim/sulfamethoxazole, levofl oxacin, and vancomycin. All antimicrobial agents were purchased from Sigma-Aldrich Canada Ltd, Oakville, Ontario, Canada. Interpretation of MICs was based on Clinical and Laboratory Standards Institute performance standards that were current at the time of testing (M100-S15 through M100-S17) (12,13).

Pulsed-fi eld Gel Electrophoresis and Multilocus Sequence Typing
S. pneumoniae chromosomal DNA was prepared as described (14). Chromosomal DNA was restricted with 20 U of SmaI (New England Biolabs, Beverly, MA, USA), and pulsed-fi eld gel electrophoresis (PFGE) was performed by using a CHEF DR-III apparatus (Bio-Rad Laboratories [Canada] Limited, Mississauga, ON, Canada) for 23 h. The parameters used were as follows: initial pulse 3.5 s, fi nal pulse 23.5 s, voltage 6 V/cm, and temperature 13°C. Salmonella Braenderup U9812 was used as a molecular size marker. The macrorestriction pattern was analyzed by using Bionumerics version 5 (Saint-Martens-Latem, Belgium).

Statistical Analyses
Data were analyzed by using SAS software version 9.1 (SAS Institute, Inc., Cary, NC, USA). Possible factors associated with serotype 5 among patients with pneumococcal disease were assessed. We examined the association of each demographic, substance abuse, and concurrent condition variable with outcome variable serotype 5 (yes or other serotype). For continuous variables, we used the t test or Mann Whitney U test as appropriate. For categorical variables, we used the χ 2 or Fisher exact test. For variables that were signifi cant (p<0.20) on univariable analyses, we used the following multivariable logistic regression model: In model 1, the signifi cance of variables was assessed by using the Wald statistic. The variables that were signifi cant at p<0.05 were retained in the model. All others were removed from the model unless they were possible confounders. In the fi nal model, we tested βs, the effect of each variable on log odds of serotype 5 after adjustment for other associated variables. To calculate rates, we used the populations that were current for each province in January 1, 2009 (16). In western Canada during 2000-2009, the numbers of serotype 5 and other serotype isolates identifi ed increased ( Figure 2). The increased number of isolates submitted for typing after the onset of the epidemic indicates greater interest on the part of public health offi cials in western Canada in identifying circulating serotypes from patients with invasive pneumococcal disease in their provinces.

Results
The epidemic primarily affected young adults (median age 41 years) (Figure 3). Only a small subset of cases occurred among patients <5 years of age and even fewer in those >65 years of age. Most patients were male (637 male, 395 female, and 16 unknown).

Specimen Source
The sources of specimens for the serotype 5 isolates from across Canada were as follows: 988 isolates from blood, 33 from lung/pleural fl uid, 9 from cerebrospinal fl uid, 7 from synovial fl uid, 7 from chest/hip/leg fl uid, 3 from pericardial fl uid, and 1 from peritoneal fl uid. For the univariable and multivariable analyses, isolates from patients with serotype 5 and nonserotype 5 invasive S. pneumoniae were collected from northern Alberta only (1,112 cases).

Patient Characteristics
According to univariable analysis, serotype 5 was more prevalent than other serotypes among patients who were male (66.2% vs. 57.0%), of First Nations heritage (21.8% vs. 10.0%), or homeless (16.1% vs. 4.7%) ( Table 1). Among the substance-abuse categories, associations with tobacco use, alcoholism, and illicit drug use were considered signifi cant (p<0.001 for each; Table 1). With respect to concurrent conditions, cases of invasive pneumococcal disease caused by serotype 5 were signifi cantly associated with cancer within 5 years before onset of invasive pneumococcal disease, cardiovascular disease, hematologic abnormalities, diabetes mellitus, cirrhosis, chronic renal failure, musculoskeletal impairment, and hepatitis C ( Table  1). For patients with bacteremia and pneumonia, invasive pneumococcal disease caused by S. pneumoniae serotype 5 occurred signifi cantly more often with pneumonia than did that caused by other serotypes (94.7% vs. 74.6%; Table  1). In addition, meningitis was more common for patients in the non-serotype 5 group than in the serotype 5 group (8.0% vs. 0.7%, respectively; p<0.001; Table 1). Death was less associated with infection caused by serotype 5 than by other serotypes (3.2% vs. 14.1%, respectively; Table 1).

S. pneumoniae Serotype 5 Characteristics
Antimicrobial drug susceptibility testing of 1,009 isolates indicated that all S. pneumoniae serotype 5 isolates tested were susceptible to cefotaxime, ceftriaxone,   To determine whether this clone had been found in the United States, we compared it with 6 serotype 5 isolates from the US Centers for Disease Control and Prevention. Three isolates were from a small cluster of cases in San Francisco, California, in 2002 and 3 were from sporadic cases in the United States in 2006 (B. Beall, pers. comm.). Of these 6 isolates, the RFLP pattern for 5 isolates was identical to that of the epidemic clone and 1 isolate had a single band difference, suggesting that the serotype 5 clone in western Canada had been circulating in the United States in 2002 and 2006. This clone might have been imported into Canada from the United States; however, it might also have been imported from elsewhere in the world because sequence type (ST) 289 is the major circulating serotype 5 clone.

Discussion
Large epidemics of pneumococcal disease might go unrecognized unless surveillance programs are in place to document fl uctuations in serotype prevalence, as reported here. The year-to-year variability of invasive pneumococcal disease caused by S. pneumoniae serotype 5 seen in some countries might actually refl ect serotype 5 outbreaks similar to what we have described (18). For example, in 2000 in Mali, Africa, 50% of the isolates recovered from children with invasive pneumococcal disease were serotype 5, yet 2 years later; this percentage had dropped to a small portion of the total cases (19,20). This serotype 5 variability has also been reported in Chile and Israel (21,22). In Israel during 1989-1998, serotype 5 was the second most common serotype (serotype 1 was the most common) that caused invasive pneumococcal disease (12%-13% of cases among children <15 years of age) (21).
Although in other countries the number of S. pneumoniae serotype 5 cases might vary from year to year, in Canada no variability for serotype 5 was evident until the 2005-2009 epidemic. Few serotype 5 isolates had been documented since 1991, when the National Centre for Streptococcus fi rst began performing pneumococcal serotyping to support national surveillance in Canada, until 2005. This serotype 5 strain has been demonstrated elsewhere in the world, not just Canada. Data from the MLST database and published reports indicate that the Colombia 5 ST289 clone has been reported in countries in Europe, Latin America, and Africa and in the United States (http://spneumoniae.mlst.net) (23)(24)(25)(26). In addition, the rate of resistance to trimethoprim/sulfamethoxazole by the Colombia 5 ST289 clone has been reported as 80.8% (27) and 58.2% (17) of the Colombia 5 ST289 strains from Latin American countries.
The S. pneumoniae serotype 5 epidemic mostly affected middle-aged men (median 41 years of age). Other risk factors were homelessness and First Nations heritage, although these factors accounted for a small percentage of the population. Because invasive pneumococcal disease reportedly affects homeless populations, the fi nding that homelessness was a major demographic factor associated with this epidemic is not surprising (28)(29)(30). A recent study from Toronto, Ontario, Canada, found that incidence of invasive pneumococcal disease was greater in the homeless population than in the general population (30) and that the variables associated with the serotype 5 epidemic (tobacco use, alcohol abuse, illicit drug use) were associated with invasive pneumococcal disease. Serotype 5 pneumococci were not identifi ed in this study.
In December 2006, investigators found S. pneumoniae serotype 5 affecting persons of First Nations heritage living near the city of Calgary, Alberta, and persons living in innercity Calgary; Edmonton; and Vancouver, British Columbia (31)(32)(33). These reports indicated that the variables associated with invasive pneumococcal disease caused by this serotype were homelessness, use of illicit drugs, First Nations heritage, alcoholism, and hepatitis B or C, thereby corroborating our fi ndings for those cases in northern Alberta (31,32). Recovery of this serotype in locations other than inner cities in western Canada (including northern Saskatchewan) suggests its spread beyond the larger metropolitan areas of western Canada (34).
A public health response to the epidemic occurred throughout western Canada. Regional health authorities conducted vaccination programs focused primarily on homeless populations in large metropolitan areas. They used the 23-valent pneumococcal polysaccharide vaccine, which contains serotype 5. As a result of these largescale pneumococcal vaccination campaigns, the National Advisory Committee on Immunization issued an advisory statement recommending use of the 23-valent pneumococcal vaccine for homeless persons and injection drug users (35). Examples of public health measures used to address the outbreak in 2 health regions are contained in reports from British Columbia focusing on S. pneumoniae serotype 5 outbreaks in the Vancouver downtown eastside and in the city of Kelowna (35,36). In Vancouver, investigators found that the serotype 5 strain accounted for 78% of cases of invasive pneumococcal disease. The major risk factors reported were use of crack cocaine and residence in Vancouver's downtown eastside, an impoverished part of that city where most of the illicit-drug users and homeless persons live (36). As a result, Vancouver Coastal Health authorities targeted rooming houses, shelters, food banks, and other community locations (32). In Kelowna, public health nurses and health care providers focused a pneumococcal vaccination program on persons who were homeless and/or addicted to illicit drugs or alcohol; at the time of their report, they had vaccinated ≈1,000 at-risk persons (37). A strength of our study is the ability of the centralized laboratory to capture and document shifts in the epidemiology of pneumococci in Canada. Regionalization of serotyping of pneumococci has the potential to miss changes in serotypes that can occur rapidly.
A weakness of our study is the lack of clinical data for all cases of invasive pneumococcal disease caused by S. pneumoniae serotype 5 that occurred during this epidemic. Logistically, gathering all of these data was not possible; however, the clinical data from northern Alberta do indicate some of the clinical variables involved and the concurrent conditions associated with serotype 5 cases. Another limitation might be that the variables for persons with invasive pneumococcal disease caused by S. pneumoniae serotype 5 (patient demographics, substanceabuse associations, concurrent conditions, type of pneumococcal disease, and outcomes) were compared with those for persons with other pneumococcal disease rather than with a healthy (nondiseased) control group. However, we thought it useful to try and determine among those with invasive pneumococcal disease whether differences existed among disease caused by serotype 5 and other serotypes.
We do not know why the epidemic was focused in western Canada and why large numbers of cases did not spread to eastern Canada or the United States. Clearly, we do not understand all the dynamics associated with large invasive pneumococcal disease epidemics.
In conclusion, we document a rare large-scale outbreak of invasive pneumococcal disease in western Canada caused by a single clone of S. pneumoniae. The clone possessed a serotype 5 polysaccharide capsule and ST289, indicating that the clone is derived from the international Pneumococcal Molecular Epidemiology Network clone Colombia 5 -19 originally described in Colombia (18). RFLP comparing a collection of S. pneumoniae serotype 5 isolates from the United States with the epidemic clone from western Canada showed that all isolates were identical, suggesting that this strain has been circulating within the United States. However, without direct evidence, we do not know from what part of the world this clone was originally imported into Canada.