TEXT

Top Abstract Text References

The Centers for Disease Control group dysgonic fermenter 3 (DF-3) comprises unnamed gram-negative rods phenotypically resembling Capnocytophaga species (6). Comparative 16S rRNA sequence analysis, however, has revealed that DF-3 is not a close relative of the capnocytophagas but constitutes a separate genus clustering together with Bacteroides forsythus and Bacteroides distasonis (6). The possible causative role of DF-3 in diarrhea was first noted 10 years ago in a patient with common variable hypogammaglobulinemia (8). Since then, 21 DF-3 stool isolates collected from immunosuppressed patients have been reported, but there was an association with diarrhea in only 50% of the cases (4).

Only one case of DF-3-associated bacteremia, which occurred in a patient with acute lymphocytic leukemia, has been reported (1). Here, we describe the isolation of DF-3 from blood and stool samples collected from a patient with acute myelocytic leukemia.

A 45-year-old male, admitted for acute myelocytic leukemia, was started on cytostatic therapy (thioguanine, cytarabin, and daunorubicin) and prophylactic oral ciprofloxacin (500 mg twice daily) plus intravenous amphotericin B (25 mg three times weekly). Twelve days later he developed aplasia, with a polymorphonuclear leukocyte count of <1,000 cells per µl, and became febrile. The antibiotic regimen was changed to imipenem (500 mg three times daily) and teicoplanin (400 mg daily) when both of two blood cultures (BacT/Alert; Organon Teknika, Turnhout, Belgium) grew Streptococcus agalactiae. However, fever persisted and a further blood culture taken 5 days after the first blood cultures became positive showed growth signals 3 days later in both the aerobic and anaerobic bottles. Twenty-four-hour subcultures on blood agar revealed greyish colonies (diameter, 1 to 2 mm) of small, gram-negative coccobacilli that did not grow on MacConkey agar. Fever resolved and leukocytes reappeared (>1,000 polymorphonuclear leukocytes per µl) a few days later. The isolate was sent to the Department of Medical Microbiology in Zurich, Switzerland (DMMZ), where it was identified as DF-3. A stool specimen of normal consistency was collected 2 months later, when the patient was readmitted for the first cytostatic consolidation therapy.

The stool was streaked onto a selective medium containing Columbia agar base No. 2 (Difco Laboratories, Detroit, Mich.) with 5% sheep blood, 15 mg of cefoperazone/liter, 7.5 mg of vancomycin/liter, and 2 mg of amphotericin B/liter (2) and incubated at 37°C in 5% CO₂ for 2 days. Identification was based on reactions for catalase, oxidase, indole, urease, nitrate reduction, esculin hydrolysis, and triple sugar iron (TSI) agar and carbohydrate fermentation (1%) in CTA Medium (Becton Dickinson Microbiology Systems, Cockeysville, Md.) (9). Fatty acids derived from glucose fermentation were determined by gas-liquid chromatography (5). Cell wall fatty acid analysis was performed by using the Microbial Identification System (MIS) (Microbial ID, Inc., Newark, Del.) (5). MICs were determined by using the E test (AB Biodisk, Solna, Sweden) on Mueller-Hinton agar (Becton Dickinson) with 5% sheep blood. Three reference DF-3 strains (72-0679, 83-0449, and 83-0498; kindly supplied by K. Bernard, Laboratory Centre for Disease Control, Ottawa, Canada), a strain (IMM 3/95) from the collection of the DMMZ, and the blood and stool isolates of the patient were included in typing experiments. Whole-cell DNA was analyzed by pulsed-field gel electrophoresis following digestion with the restriction enzymes SmaI, XbaI, PstI, I-LeuI, and NotI and the combination of PstI and BamHI (New England Biolabs Inc., Beverly, Mass.) according to standard procedures as described (7). The method of ribotyping has been described previously (3). Briefly, chromosomal DNA was digested with two restriction enzymes (PvuII and EcoRI) and separated by agarose gel electrophoresis. After Southern blotting, hybridization was done with biotin-labelled plasmid pKK3535 containing an rRNA operon of Escherichia coli. Hybrids were then visualized by using the BluGene Kit (Gibco-Bethesda Research Laboratories, Gaithersburg, Md.).

Stool (collected 10 weeks after isolation of DF-3 from the patient's blood) initially revealed heavy growth of pinpoint-size translucent colonies, which later became larger and took on a greyish appearance. Both isolates developed a fruity odor and showed negative reactions for catalase, oxidase, indole, urease, and nitrate reduction, positive reactions for esculin hydrolysis, acidification of butt and slant of TSI, and acid production from glucose, sucrose, maltose, and xylose but not from mannitol. This identified the organism as DF-3. Metabolic fatty acids produced by the blood isolate were propionic, lactic, and succinic acids, and the major cellular fatty acids were a-C_15:0 (26%), i-3-OH-C_16:0 (21%), i-C_14:0 (13%), 3-OH-C_16:0 (6%), C_16:0 (5%), i-C_15:0 (4%), C_15:0 (3%), C_16:17c (3%), and C_18:19c (3%). In line with previous results (1, 2, 4), the blood isolate was susceptible to amoxicillin-clavulanate, clindamycin, erythromycin, piperacillin-tazobactam, rifampin, tetracycline, and trimethoprim-sulfamethoxazole (Table 1). Pulsed-field gel electrophoresis did not reveal visible bands in any run. However, ribotyping of all six DF-3 strains by using either PvuII or EcoRI revealed five clearly different patterns except that the blood and stool isolates of the patient were indistinguishable with both enzymes (Fig. 1).

View larger version (60K): [in this window] [in a new window]   FIG. 1.   Ribotyping patterns after cleavage with EcoRI (a) and PvuII (b). Lanes: 1 and 2, stool and blood isolates, respectively, of the patient; 3 to 6, epidemiologically unrelated DF-3 isolates.

Twenty-two of 28 reported isolates of DF-3 were isolated from stool. In six other cases (4) the gastrointestinal tract was the putative source of infection. The first attempt at isolation of DF-3 from stool on a selective agar for Campylobacter spp. at 35°C was successful because DF-3 is resistant to cefoperazone and vancomycin (8). Our patient did not have diarrhea at that time or during the bacteremia with fever. The detection of DF-3 in the stool 10 weeks after the episode of bacteremia can be explained either by persistence of DF-3 or by reinfection of this particularly susceptible immunocompromised individual. Since the ribotypes of both isolates were indistinguishable and because all epidemiologically independent strains exhibited different patterns, any reinfection must have taken place from the same source. The lack of isolation of DF-3 outside the human body and its predominant isolation from stool specimens strongly suggest that the gastrointestinal tract is its natural habitat, and therefore, persistence of DF-3 rather than reinfection is the most plausible explanation. Broad-range antibiotic regimens in immunocompromised patients could select for growth of DF-3 because the organism is resistant to a wide range of antibiotics, including aminoglycosides, glycopeptides, cephalosporins, and carbapenems. Our isolates were resistant to the antibiotics given during the bacteremic episode, and fever disappeared only at the time of reappearance of blood leukocytes, i.e., with the restitution of immune function. We therefore believe that DF-3 is an opportunist with rather low pathogenicity in healthy hosts.