HIV, FUO, and Bartonella

Abstract & Commentary

Synopsis: Bartonella is a cause of unexplained fever in HIV-infected patients.

Source: Koehler JE, et al. Prevalence of Bartonella infection among human immunodeficiency virus-infected patients with fever. Clin Infect Dis. 2003;37:559-566.

Koehler and colleagues at University of Califonia in San Francisco and the CDC examined the prevalence of Bartonella infection in patients with persistent or recurrent fever of at least 2 weeks duration. Patients who had received tetracycline, a macrolide, or a rifamycin in the previous 2 weeks were excluded, as were HIV-infected patients with a history of Mycobacterium avium complex (MAC) infection. In addition, HIV-infected patients must have had a negative MAC blood culture within the previous 4 weeks.

Of the 382 patients studied, 95% of whom were HIV infected, 68 (18%) had evidence of Bartonella infection as determined by indirect fluorescent antibody testing, culture, or PCR. Nineteen patients (5%) had serological titers > 1:64—levels previously associated with active Bartonella infection. Twelve patients had organism identification by either culture or PCR of blood and/or tissue. When the 12 patients were carefully examined, cutaneous bacillary angiomatosis was detected in 6. In a nested, case-control study, only bacillary angiomatosis and elevated alkaline phosphatase level were significantly associated with Bartonella infection.

Comment by Stan Deresinski, MD

Eighteen percent of persistently or intermittently febrile HIV-infected patients with previous negative MAC blood cultures had evidence of infection with either B henselae or B quintana, although only a minority of these had direct evidence (culture or PCR) of the presence of the organism. At the same time, only 14% of the cohort proved to have disseminated MAC infection, making it less prevalent than bartonellosis, but this likely was the result of selection bias since patients with positive MAC cultures in the 4 weeks before study entry were excluded. Two percent of the patients proved to have disseminated histoplasmosis.

One can certainly question the relevance of the diagnosis of active Bartonella infection based on a single serological test. Koehler et al cite studies indicating that the background prevalence of Bartonella antibodies in an afebrile population is 4-7%. It is possible that higher titers might be expected in HIV-infected patients as a consequence of the polyclonal increase in gamma globulin levels commonly seen in this population.

Despite such caveats, this study clearly demonstrates that Bartonella infection is a cause of "fever of unknown origin" in AIDS-infected patients with advanced immunodeficiency—the median cell count of the Bartonella patients in this study was only 35/mm3. The problem is in diagnosis. While antibody testing for Bartonella is readily available through a number of commercial laboratories, PCR is less readily available and the performance of these tests may not be as good as those used here. Blood culture can be obtained with a little cooperation from your laboratory. In this study, 10 mL of blood was cultured by lysis-centrifugation with inoculation for 3 weeks. Nonetheless, the diagnosis of this infection is important since it is treatable.

Dr. Derenski is Clinical Professor of Medicine, Stanford; Associate Chief of Infectious Diseases, Santa Clara Valley Medical Center.