We examined 400 isolates identified as NAG vibrios that were collected during 2002–2009 from 11,904 stool specimens from patients with diarrhea admitted to the Infectious Diseases and Beliaghata General Hospital, Kolkata, India. Specimens were screened for common enteric pathogens, according to standard protocols (7). Oxidase, string test, and arginine dihydrolase–positive strains that did not agglutinate with V. cholerae O1 polyvalent or O139 monovalent antiserum were further confirmed as V. fluvialis by using a multiplex PCR targeting the toxR gene of V. fluvialis and the ompW gene of V. cholerae (8,9). Isolates were also subjected to PCRs targeting different virulence-associated genes encoding the repeat in toxin (rtxA, rtxC), heat-stable enterotoxin (stn), type 3 secretion system (vcsC2, vcsV2, vcsN2 and vspD), cholera toxin (ctxA), toxin co-regulated pilus (tcpA), thermostable direct-hemolysin (tdh), TDH-related hemolysin (trh), V. fluvialis hemolysin (VFH), and metalloproteases, according to published methods (10–12).

Expression of VFH was determined in vitro by using erythrocytes from rabbit and sheep. Cytotoxin assay was performed with HeLa and Chinese hamster ovary cell lines by using sterile culture filters of the V. fluvialis strains that were isolated as a sole pathogen. Antimicrobial drug susceptibility testing was performed by using the disk diffusion method with commercially available disks (Becton Dickinson, Sparks Glencoe, MD, USA), according to Clinical and Laboratory Standards Institute criteria (13). Because these guidelines do not include interpretive criteria for V. fluvialis, breakpoints for Enterobacteriaceae were adopted. Escherichia coli ATCC 25922 was used as a quality control strain.

Pulsed-field gel electrophoresis was performed according to the PulseNet standardized protocol for V. cholerae (14). Gel Compare II software (Applied Maths NV, Sint-Martens-Latem, Belgium) was used for electrophoresis pattern comparison that runs on Dice similarity index and unweighted pairgroup with arithmetic mean method.

Among the 400 isolates presumptively identified NAG vibrios, multiplex PCR confirmed 131 and 269 strains (each strain representing a case) as V. fluvialis and V. cholerae, respectively. The overall prevalence rate of V. fluvialis among 11,904 hospitalized patients with diarrhea was 1.1%. Abrupt appearance of V. fluvialis was identified in 2002, although the surveillance of diarrheal infection was initiated at the Infectious Diseases and Beliaghata General Hospital in 1996 (www.niced.org.in/annual_reports.htm). The isolation rate of V. fluvialis gradually increased from 0.7% in 2002 to 2.2% in 2009 (Table 1). Of the 131 strains of V. fluvialis, 43 (33%) were identified as the sole pathogen; the remaining 88 (67%) were isolated as a co-pathogen with either V. cholerae, V. parahaemolyticus, E. coli, Shigella spp., parasites, or enteric viruses (data not shown). Among the mixed infections, V. fluvialis with V. cholerae was isolated most often (17%), followed by V. fluvialis and V. parahaemolyticus (8%). The presence of Vibrio spp. as mixed pathogens indicates that these patients likely acquired the infection from contaminated water or food. We analyzed the date of admission and place from where the patients resided and found no evidence for clusters of infection or small outbreaks caused by V. fluvialis.

V. fluvialis infection was much more often detected in adults (73%) than in children <5 years of age (27%). Clinical symptoms of sole infection caused by V. fluvialis were similar to that of cholera: watery diarrhea (86%), severe dehydration status (28%), and abdominal pain (12%) (Table 2). Several previous investigations have identified cholera-like diarrheal outbreaks caused by V. fluvialis (4,5).

All the V. fluvialis strains were negative for the virulence genes commonly reported in V. cholerae and V. parahaemolyticus, but >90% were positive for genes encoding VFH and metalloproteases. More than 80% of the strains expressed hemolysin against rabbit and sheep red blood cells. Hemolysin is a widely distributed virulence factor in most pathogenic Vibrio spp. Metalloprotease produced by V. fluvialis is related to hemagglutination proteases of V. vulnificus, which enhances permeability and hemorrhagic activities (12). These factors may increase the virulence of V. fluvialis and contribute to diarrhea.

When the culture filtrates were tested, cytotoxic effect was readily noticed in the Chinese hamster ovary and HeLa cell lines, i.e., cytoplasmic vacuolation, cell rounding, and destruction of the monolayer. In most strains isolated as a sole pathogen, the cytotoxic endpoint titer was 2–256 (Technical Appendix Table 1). The cell vacuolation phenomenon has been reported as a virulence factor in several enteric pathogens (Technical Appendix References).

In this study, V. fluvialis strains were highly resistant to ampicillin (92%), streptomycin (85%), furazolidone (85%), and sulfamethoxazole/trimethoprim (70%) (Technical Appendix Table 2). About half the number of strains were resistant to ciprofloxacin and 45% to nalidixic acid; the lower resistance rate for nalidixic acid compared with fluoroquinolones is unexpected and warrants further investigation to confirm the additional mechanisms. In a previous study, we found that some V. fluvialis strains carried the plasmid-mediated quinolone resistance gene allele qnrA1 and a gene encoding the aminoglycoside acetyltransferase (aac(6')-Ib-cr), which reduces ciprofloxacin activity (15). Fluroquinolone resistance and intermediate susceptibility to erythromycin (92%) are the unique features of the V. fluvialis isolated in this study; this trend was not recorded in other Vibrio spp., e.g., V. cholerae and V. parahaemolyticus.

Although the V. fluvialis strains exhibited distinct NotI restriction profiles in the denrogroam analysis, at least 4 major clades were identified (Figure). Clades A and B, with strains isolated during 2002–2007, exhibited less antimicrobial drug resistance than did clade C and D strains identified during 2008–2009; multidrug-resistant strains, especially those resistant to fluroquinolones, were identified in higher numbers in clades C and D (Figure).

Our results demonstrate an emerging trend of prevalence of V. fluvialis among patients with acute diarrhea patients in Kolkata. The expression of cytotoxic activity and hemolysin may contribute to understanding the pathogenicity of V. fluvialis. Further epidemiologic studies are necessary to elucidate the public health importance of V. fluvialis–mediated diarrhea.

Results of assays of clinical Vibrio fluvialis strains to determine ability to lyse rabbit erythrocytes and cytotoxic effect on Chinese hamster ovary and HeLa cells and antimicrobial drug resistance of V. fluvialis.