Cholera is endemic in Laos and O1 and non-O1 V. cholerae (including serogroups 11, 14, 16, 21, 41, 43, 68 and 169), but not 0139, have been recorded [12, Phouthavane et al. unpublished]. Non cholera-toxin producing, non-O1, non-O139 V. cholerae have been described from blood cultures and ascitic fluid, although they are exceedingly rare isolates. Consumption of undercooked shellfish, exposure of open wounds to salt and fresh water, drinking of well water, ascites, cirrhosis, renal failure and haematological malignancies have all been associated with non-O1, non-O139 disease 345. The mortality of recorded non-O1 bacteraemia is high at 24–62% 67891011. Concurrent infection with organism(s) that may damage the bowel mucosa may predispose patients to V. cholerae non-O1 bacteraemia 12. Of 28 patients with non-01 V. cholerae bacteremia recorded in Taiwan over 5 years, 95% had ascites and 75% had cirrhosis, perhaps because increased intestinal permeability in cirrhotic patients predisposes them to invasive disease 13. Unlike O1 biotype El Tor, but like O139 V. cholerae, O21 possess a capsule, which is thought to facilitate survival in the blood stream 12. V. cholerae 01 bacteremia has also been described, but is more rare than non-01 14.

In January 2006, a 20 year old female rice farmer, from Nayxaythong District, on the outskirts of Vientiane City, was admitted to Mahosot Hospital, Vientiane, Lao PDR (Laos) with a 7 day history of abdominal pain, fever, rigors, watery diarrhoea, yellow eyes, leg oedema, anuria and malaena. She had collected many apple snails 'hoy pak kwang' (Pomacea canaliculata) and mussels 'hoy kee' (Ligumia species) from the mud on the edge of a nearby lake and washed them before boiling them for 30 minutes. She ate partially cooked snails as she boiled them and became unwell 7 days later. The fully cooked snails were eaten, with chilli sauce, by another villager who remained well. On admission she was had a fever (39.5°C) with jaundice, abdominal tenderness, splenomegaly, tachycardia, shortness of breath, lower leg oedema and ascites. She had no skin lesions and had no evidence for prior cirrhosis, antacid or steroid use, gastric surgery or haemoglobinopathy. Her haematocrit was 35%, with a peripheral white cell count of 8.9 10⁹/L (66% neutrophils). Serum creatinine, alkaline phosphatase and glucose were within the reference ranges, with raised total serum bilirubin 41 μmol/L (reference <14.5 μmol/L), AST 75 IU/L (<37 IU/L) and ALT 170 IU/L (<40 IU/L). Abdominal ultrasound demonstrated ascites and splenomegaly with normal liver and kidneys. Stool was not cultured. She was treated empirically with intravenous ampicillin, gentamicin and diuretics. Analysis of tapped ascitic fluid revealed many white cells, 90% of which were neutrophils.

Oxidase positive, curved, non-motile Gram negative rods were grown from 2 of 2 blood cultures, after 2 days incubation and the ascitic fluid, after 1 day of incubation. API 20E (bioMeriuex, France) identified Vibrio cholerae with 99.9% confidence. The organism grew in 0% sodium chloride but not in 6% or 10% sodium chloride and did not agglutinate when tested against O1 and O139 antisera, but did against O21 antisera (Department of Bacteriology, National Institute of Infectious Disease in Tokyo, Japan). By disc diffusion testing (Kerby-Bauer, according to NCCLS guidelines 2003) the organisms were susceptible to ampicillin, ciprofloxacin, ceftriaxone and tetracycline. She was taken home moribund, at her relatives' request, after 2 days in hospital and died soon afterwards.

In order to investigate whether we could grow 021 V. cholerae from the probable source of infection, in May 2006 specimens of P. canaliculata and Ligumi species were collected from the same lake as above and cultured in TCBS according to the methods of Ottaviani et al. 15. No V. cholerae were grown but Aeromonas, Enterobacter, Proteus and Escherichia species were identified.

The PCR techniques of Nandi et al. 16 were used to confirm that the organisms contained the outer membrane protein W (ompW) and regulatory protein (toxR) of Vibrio cholerae and to determine whether they contained cholera toxin (ctxA) and toxin-coregulated pilus (TCP) virulence genes (Table 1 &2). The bacteria from blood culture and ascitic fluid were re-plated on Modified Drigalski Agar (BTB agar, Eiken Kagaku Co Ltd, Tokyo, Japan) to obtain pure growth and DNA extraction was performed (Wizard, A1120, Promega, WI, USA). Biotype Classic O1 Inaba strain 569B and Vibrio cholerae O21, recovered from a patient with diarrhoea in Vientiane Capital, were used as positive control and reference strains, respectively and compared with human isolates of serogroup O21 109-68 from India (from diarrhoeal stool), and environmental isolate 418-03 from the USA (seawater, Chesapeake Bay, Virginia) (both from National Institute of Infectious Diseases (NIID) Tokyo). Primers of Outer membrane protein W (ompW) were prepared according to 16 and outer membrane protein U (ompU), outer membrane protein K (ompK), toxin-coregulated pilus (TCP), cholerae toxin A (ctxA) and Regulatory protein (toxR) primers were designed (Table 2) and DNA amplification performed (i-cycler, BioRad) with analysis using 1.8% agarose gel electrophoresis. V. cholerae O21 cultures from both blood and ascitic fluid from our patient and 109-68 and 418-03 had ompW and toxR gene regions, confirming them to be V. cholerae 16, but had no ctxA, ompU, omp K, TCP gene regions, which were present in the Classic 569B strain (Figure 1, Table 1). Scanning electron microscopy (SEM; S-5200, Hitachi, Japan) of the patient's isolate, serogroup O21 109-68 and 418-03 (Figures 2, 3, 4) demonstrated that, although 109-68 and 418-03 did possess flagellae, the organisms isolated from the patient did not. The flagellin regulation gene flaA was demonstrated by PCR in the patient's isolates and 569B with specific oligonucleotide primers (Table 3), which we designed based on the work of Klose and Mekalanos 17 who demonstrated that immotility could be caused by mutations in the flaA gene alone.

Fatal bacteremic non-O1, non-O139 V. cholerae is reported from Laos. The organisms were unusual in being non-motile. They did however possess the Vibrio flagellin flaA gene. Further research on the determinants of the non-motility is required. Given the importance of fishing and water in Lao society and the high prevalence of cirrhosis secondary to hepatitis B and C infection 24, V. cholerae non-01 bacteraemia may be a more important disease than this case report suggests.

The authors declare that they have no competing interests.

RP was in charge of diagnosis and edited the ms, MN performed PCR assays and helped write ms, EA performed diagnosis work up and SEM, and edited ms, VD, MV and OL performed diagnosis work up and edited ms, CEM assisted with diagnostic workup, reviewed molecular analysis and edited ms, SN performed diagnosis workup and helped write ms and PNN assisted with diagnostic workup, wrote the first draft and edited ms. All authors read and approved the final manuscript.