The results of our genetic analyses suggest that the most recent outbreak of cholera in Kenya (beginning in January 2009) represented several distinct genetic lineages of V. cholerae that emerged simultaneously, perhaps facilitated by environmental and behavioral factors around the country. Thus, these data suggest that these outbreaks likely resulted from endemic foci rather than from recent introduction and spread by travelers.

In contrast, previous outbreaks of cholera in Kenya have been attributed to the spread of V. cholerae by travelers. Outbreaks in the 1990s and in 2005 were attributed to a clone carried by travelers to Nairobi; from there, it spread to other locations in the country (4). Likewise, the well-publicized 2010 cholera outbreak in Haiti also appears to have resulted from the introduction of V. cholerae by travelers (16). In addition, there have been repeated introductions of V. cholerae into previously disease-free locations by travelers from Southeast Asia and other areas where cholera is endemic (17). On a smaller scale, V. cholerae is thought to persist around the African Great Lakes Region because, although it becomes extinct in areas, it is still present in other localities and can spread when weather conditions are favorable (7).

Genetic uniformity is expected when V. cholerae is spread from a single source. This expectation conflicts with the observation that isolates with Inaba and Ogawa serotypes occurred during the outbreak. However, the expectation is consistent with the genetic evidence that all isolates had the El Tor biotype and identical (by Tenover’s criteria) (18) PFGE patterns (J.O. Oundo, et al. unpub. data). The nearly identical (single band) differences in the PFGE patterns are consistent with whole genome sequencing results showing that the toxigenic El Tor lineage of O1 V. cholerae has few large (several kilobytes) insertions or deletions (19) and few nucleotide changes (17). Large insertions or deletions or nucleotide changes in restrictions sites are the mutations that produce altered PFGE patterns; their absence in the El Tor lineage is consistent with the minimal number of PFGE patterns.

All V. cholerae strains that we studied from the 2009–2010 outbreak in Kenya had the classical ctxB allele, which is consistent with a clonal origin. The earlier population of V. cholerae is presumed to have contained only the El Tor allele because the classical ctxB allele was not detected before 1989 in isolates with the El Tor background (20). Thus, the presence of the classical allele in all isolates from the 2009–2010 outbreak most likely represents a rapid shift in allele frequency. Such a shift occurred in O1 V. cholerae in Kolkata, India, when allele frequency changed from 100% El Tor in 1989 to 100% classical ctxB in 1995 (20). It is of critical clinical importance that, the classical ctxB allele, when found in the background of an El Tor strain, has been associated with a more severe form of cholera than that caused by a strain with the El Tor ctxB allele (21,22).

Our MLVA genotyping results for 170 V. cholerae isolates from the 2009–2010 outbreak in Kenya showed extensive genetic diversity: we found 5 clonal complexes and 19 singleton genotypes. In contrast, studies using PFGE found genetic uniformity among V. cholerae isolates from 1994 to 2010 (5,8) (J.O. Oundo, et al. unpub. data). These differences may be explained by the finding that MLVA is superior to PFGE for discriminating between isolates of V. cholerae O1 (11–13).

An alternative explanation for the substantial variation observed among V. cholerae isolates from Kenya is that their tandem repeat loci may result from a rapid accumulation of mutations. However, 3 findings have documented the relative stability of those loci. First, an analysis of the variation occurring during a month-long serial passage revealed that the 3 large chromosomal loci (the primary determinants of membership in the various clonal complexes) were largely stable (13). Second, the clonal complexes remained distinct during 3 consecutive years in Bangladesh (13), longer than the 17 months of the 2009–2010 outbreak in Kenya. Third, all isolates from the cholera outbreak that began in October 2010 in Haiti belonged to a single clonal complex (16). These observations and the co-occurrence of distinct clonal complexes are consistent with the idea that there were multiple simultaneous outbreaks of cholera in Kenya.

Diverse MLVA genotyping results are found in countries where V. cholerae is endemic. In India, 6 clonal complexes were detected in V. cholerae O1 isolates (23), and in Bangladesh, 7 clonal complexes were detected (12). Consistent with these findings, we detected 5 clonal complexes in isolates from Kenya. The genetic diversity is also consistent with data from single-nucleotide polymorphism analyses that showed the several distinct waves of immigration of V. cholerae into Kenya (17), if the descendants of these immigrants settled and survived. The presence of multiple distinct lineages across the country supports the notion that there have been multiple introductions of V. cholerae and that over time these strains spread countrywide. The multiplicity of clonal complexes is in stark contrast to the apparent uniformity of the ctxB allele. An earlier study demonstrated that the ctxB alleles, as part of a mobile genetic element, were found in distinct locations in different MLVA clonal complexes (14).

The timing of the observed diversity is consistent with multiple separate outbreaks and not with the spread of disease by travelers. In February 2009, isolates from highland region, including Nairobi, belonged to clonal complex 1 and might have spread from the lake region. However, isolates from Moyale in the arid and semi-arid region belonged to clonal complex 5, and the time between the initial observations in the southern part of the country and the far north was too short for the isolates to have evolved into a new clonal complex. In addition, the far north is a region where travel is difficult and people from south seldom travel. Thus, these 2 outbreaks appear to be temporally, geographically, and genetically separate. As stated, the mutations necessary for 1 clonal complex to evolve into another are not expected to occur in a few months. Thus, during the 2009–2010 outbreak, the occurrence of isolates from a second complex in the highland region (clonal complex 3) and the arid and semi-arid region (clonal complex 2) cannot be explained by evolutionary changes. In addition, no source was found from which travelers could have introduced these V. cholerae strains. A similar argument can be made for clonal complex 4 isolates and for each of the singletons scattered across the country. That is, the time to their appearance was too short to represent an evolutionary change, and there was no source from which a traveler spread the genetically distinct isolates. Thus, our analyses led to the conclusion that V. cholerae is endemic in Kenya. The extensive genetic variation among the isolates exposes a limitation in our sampling scheme. Our sampling should have included many more isolates from each region and each time period. The observed genetic differences permit distinguishing between regions, but the small number of samples limits what can be inferred about what is happening within a region.

Kenya has experienced multiple major outbreaks of cholera, and different regions of the country have reported different attack rates. During the 2009–2010 outbreak, all of the isolates collected contained the classical ctxB allele. Despite this genetic uniformity, our MLVA results showed that the isolates had extensive genetic variation within and between geographic locations. The genetic relatedness studies we performed showed that 5 clonal complexes and 106 different genotypes were part of the outbreak; thus, V. cholerae had several genetic lineages. Our data are consistent with V. cholerae isolates being endemic throughout Kenya.