The patients were enrolled in a febrile illness study in 4 rural hospitals in Chiang Rai and Khon Kaen Provinces, in northern and northeastern Thailand, respectively, during February 2002–March 2003. One of the patients (no. 45-00250) was enrolled as an afebrile patient. The 4 patients had some common clinical symptoms, such as headache, myalgias, dizziness, and fatigue. In addition, 3 of the patients had elevated levels of liver enzymes. All patients reported trapping or killing rats or seeing rats inside or around their houses during the 2 weeks before onset of symptoms, and all patients owned a dog and/or cat (Table).

To determine if the patients were infected with Bartonella species, we tested blood clots from each patient for the presence of the bacterium. For testing, we used a liquid growth medium (Bartonella α-Proteobacteria growth medium [BAPGM]) as a pre-enrichment step (11), improved molecular assays for Bartonella detection (12), and established molecular methods. The blood clots were inoculated into freshly prepared BAPGM. To avoid potential contamination, we did not supplement the medium with animal blood. Blank BAPGM controls were included for each inoculation. The inoculants and the blank controls were incubated aerobically at 35°C in 5% CO₂ for 7 days.

DNA was extracted from this pre-enriched medium by using the QIAamp DNA Mini Kit (QIAGEN, Valencia, CA, USA). DNA was analyzed following published protocols (12–14). In brief, real-time PCR, nested PCR, and conventional PCR were used to target Bartonella-specific regions in the transfer-messenger RNA (ssrA) gene, the citrate synthase (gltA) gene, and the 16S–23S rRNA internal transcribed spacer, respectively. Positive controls (B. doshiae) and negative controls were included within each PCR run to evaluate the presence of appropriately sized amplicons and contamination, respectively. Amplicons were recovered from PCRs by gel purification (QIAGEN) and sequenced in both directions.

Sequences for the 3 targets were similar to those for the type strain of B. vinsonii subsp. arupensis. The gltA sequences (338 bp) from all 4 patient samples were identical to those for previously reported variants from a febrile patient in Thailand (GenBank accession no. GQ200857) and from stray dogs in Thailand (GenBank accession no. FJ946836). These gltA sequences were 0.8% divergent from the type strain of B. vinsonii subsp. arupensis (GenBank accession no. AF214557).

The ssrA sequences (251 bp) revealed 2 similar variants. One variant was identical to the type strain of B. vinsonii subsp. arupensis (GenBank accession no. JN029783) and was identified in samples from 3 of the patients (45-00250, 45-01217, and 45-01239). The other variant (GenBank accession no. JN394654), from patient 45-00025, was 2.8% divergent from the type strain of B. vinsonii subsp. arupensis.

In addition, two 16S–23S rRNA internal transcribed spacer variants were identified. One variant (GenBank accession no. JN402327), identified in samples from 3 patients (45-00025, 45-01217, and 45-01239), was 0.4% divergent from the type strain of B. vinsonii subsp. arupensis (GenBank accession no. AF312504). The other variant (GenBank accession no. JN402328), from patient 45-00250, was 0.9% divergent from the type strain of B. vinsonii subsp. arupensis (GenBank accession no. AF312504).

In Thailand, at least 2 Bartonella species, B. henselae and B. tamiae, have been reported in association with human diseases (2,15). In addition, Bartonella DNA was detected recently in acutely ill patients from Thailand (7). gltA sequences revealed a broad range of Bartonella species in these patients, including B. elizabethae, B. tribocorum, B. rattimassiliensis, and B. vinsonii subsp. arupensis. This finding suggests that numerous species of Bartonella may be associated with acute illness in Thailand.

In the current study, sequence data from 3 genetic targets provide additional evidence to confirm infection with B. vinsonii subsp. arupensis in patients in Thailand. Although the potential source(s) and mechanism(s) of transmission of this bacterium to humans remain unclear, the previous finding of B. vinsonii subsp. arupensis among stray dogs in Thailand (9) and the fact that 3 of the 4 patients in this report owned dogs could suggest dogs might be a source of the bacterium. Commonality of rat exposure may suggest that rats can also be potential reservoir hosts of B. vinsonii subsp. arupensis. Further investigation is needed in this regard.

It is known that diseases of humans and animals can translocate across the globe. For example, bubonic plague, caused by Yersinia pestis, was spread by infected rats on ships traveling from Asia to other continents. Bartonella elizabethae and other rat-associated Bartonella species that originated in the Old World have been similarly translocated to the New World and other regions (1,8). Reports of infections caused by B. vinsonii subsp. arupensis in humans in France, Russia, and Nepal, suggest that the role of this bacterium as a human pathogen may be more geographically widespread than previously believed.

Given the postulated New World origin of this pathogen, its ecology might serve as a model system to examine a possible translocation of Bartonella species across broad geographic regions and even between hosts. In addition, our finding of B. vinsonii subsp. arupensis in patients in Thailand suggests that this pathogen may be responsible for some undiagnosed cases of febrile illness and other types of illnesses in Thailand and possibly other countries in Asia.