We retrospectively reviewed clinical and laboratory data in cases of Desulfovibrio bacteremia at Kameda Medical Center, a tertiary-care hospital in Kamogawa, Japan, during January 2020–June 2025; we included episodes in which Desulfovibrio spp. were isolated from blood cultures. We considered all positive blood culture bottles collected during the same clinical episode as 1 case. We counted a new episode only when it was clearly associated with new symptoms or signs, a new anatomic focus, or resolution of a previous episode. We assessed outcomes during hospitalization, which we defined as the period from initiation of antimicrobial therapy through hospital discharge. The ethics committee of Kameda Medical Center approved the study (approval no. 25-061) and waived the requirement for informed consent because of the retrospective study design and use of deidentified data.

We processed blood cultures with the BACTEC FX system (Becton, Dickinson and Company, https://www.bd.com) and incubated them at 35°C for <7 days in accordance with our routine protocol. We defined the time to positivity as the interval from the start of incubation to the first instrument-flagged positive bottle. We performed routine identification by MALDI-TOF mass spectrometry and a desulfoviridin assay. We performed species identification with a MALDI Biotyper using the MBT Compass Library version 13 (Bruker Daltonics GmbH, https://www.bruker.com). We considered scores >2.0 as species-level identifications and scores <2.0 as uncertain. We confirmed species by 16S rRNA gene and whole-genome sequencing (Appendix Tables 1, 2). For whole-genome sequencing, we generated paired-end reads on an Illumina MiSeq instrument (Illumina, https://www.illumina.com), assembled reads de novo, evaluated assembly quality, and assigned species by comparing average nucleotide identity with type strains and reference genomes. We identified antimicrobial drug resistance determinants from draft assemblies and determined antimicrobial drug susceptibility by microdilution in Brucella broth on dry plates (Eiken Chemical, https://www.eiken.co.jp) incubated anaerobically at 35°C–37°C for 48–96 hours. (Appendix).

We identified 8 episodes of Desulfovibrio bacteremia among 4,431 patients with positive blood cultures (0.2% [95% CI 0.1%–0.4%]). All patients were >65 years of age (median 81 years; interquartile range 77–86 years); the presumed source in 7 episodes was abdominal infection (Table 1). The primary initial symptoms were fever (4/8) and abdominal symptoms (5/8) (Appendix Table 3). Median time to positivity was 4.1 days (range 2.9–5.5 days), and 3 episodes became positive on incubation day 6 (Table 1). We assessed outcomes at discharge; 1 patient died. Case 7 was considered of uncertain clinical significance because symptoms had resolved without antimicrobial therapy by the time of culture notification and the patient declined further evaluation; he was later confirmed to be alive when he returned for care for an unrelated illness ≈1 year later.

Figure 1

Figure 1. Gram-stained smear from a positive anaerobic blood-culture bottle in a study of Desulfovibrio bacteremia at a tertiary-care hospital in Japan, 2020–2025. A) D. desulfuricans spiral form; B) ...

Figure 2

Figure 2. Phylogenetic tree of 8 Desulfovibrio isolates from study of Desulfovibriobacteremia at a tertiary-care hospital in Japan, 2020–2025. Tree is based on 16S rRNA gene sequences and...

We identified 2 isolates each of D. desulfuricans, D. fairfieldensis, D. falkowii, and D. legallii (Table 2). Gram stains from anaerobic blood culture bottles showed curved gram-negative rods (Figure 1). The desulfoviridin assay was positive for all isolates. Seven isolates passed genome quality thresholds and were assigned to species by average nucleotide identity and digital DNA–DNA hybridization (Figure 2; Appendix Tables 4–8). The remaining isolate was identified as D. legallii by 16S rRNA gene sequencing because its genome assembly did not meet completeness criteria. Previous reports emphasized D. desulfuricans and D. fairfieldensis as predominant causes of bacteremia (2), whereas our series also included D. falkowii and D. legallii. Bacteremia caused by D. falkowii or D. legallii has been reported infrequently (5,8). MALDI-TOF mass spectrometry did not identify D. falkowii or D. legallii (Table 2); that finding was consistent with a previous report of D. legallii bacteremia in which MALDI-TOF mass spectrometry failed to identify the species (5). Sequence data are available in DDBJ/GenBank under BioProject PRJDB35884 (Appendix Table 9).

Antimicrobial susceptibility testing by broth microdilution showed low MICs for ampicillin/sulbactam and metronidazole (Table 2; Appendix Table 10). Piperacillin/tazobactam MICs were 32 to >64 μg/mL for all isolates (Table 2), consistent with previous observations (6,9). We did not perform Etest susceptibility testing (bioMérieux, https://www.biomerieux.com), so could not assess agreement with broth microdilution MICs. Five isolates had ceftriaxone MICs ≥32 μg/mL and carried β-lactamase genes, including bla_DES-1-like, bla_MUN-1, or bla_CfiA-like (Table 2), suggesting that β-lactamase activity contributes to elevated ceftriaxone MICs in some isolates. DES-1 has been described in D. desulfuricans (7). MUN-1 is an Ambler class A extended-spectrum β-lactamase (10). CfiA-family class B metallo-β-lactamases have been described in the Bacteroides fragilis group (11). D. falkowii isolate KML2505 carried bla_MUN-1 with 100% identity and 100% coverage to reference isolate WP_206340447.1. D. desulfuricans isolates KML2506 and KML2508 each carried DES-family class A β-lactamases (bla_DES-1-like) with 81%–82% identity and 100% coverage to the closest reference WP_063860095.1 isolate. In addition, D. fairfieldensis isolates KML2502 and KML2504 harbored subclass B1 metallo-β-lactamase homologs (bla_CfiA-like) with 47% identity and 94% coverage to the closest reference, WP_005808062.1.

In this case series, Desulfovibrio bacteremia was associated with multiple species, including D. desulfuricans, D. fairfieldensis, D. falkowii, and D. legallii, suggesting broader species diversity than previously appreciated. Antimicrobial drug susceptibility testing showed low MICs for metronidazole and ampicillin/sulbactam, whereas MICs for piperacillin/tazobactam were high in all isolates. Routine MALDI-TOF mass spectrometry did not identify D. falkowii or D. legallii bacteria. Curved gram-negative rods in anaerobic blood culture bottles and a positive desulfoviridin assay may prompt suspicion for Desulfovibrio infection, which can guide empiric therapy while confirmatory identification is pending.