We developed a method to identify mycobacteria by sequencing hypervariable regions of the polymerase chain reaction-amplified 16S ribosomal RNA gene. This method is nearly specific for mycobacteria and uses positive culture from liquid or solid medium without the needfor lengthy subculture. It shortens identification time to 3 days, which is much faster than the conventional biochemical method (mean, 8 weeks). It applies to all mycobacteria (approximately 100 species), unlike current nucleic acid hybridization methods, which probe only 4 species. The identifications are the same or are species specific for the well-characterized mycobacteria (59/68 [87%]) or more accurate for recently proposed species (9/68 [13%]). The method requires a single sequencing reaction, which is efficient and cost-effective. Therefore, this method is clinically and academically useful.

We used a polyphasic approach (sequencing analysis of the 16S ribosomal RNA gene and phenotypic analyses) to characterize 36 strains of Roseomonas species isolated from blood. Five strains, represented by strain MDA5176 (M.D. Anderson Cancer Center), were identified as Roseomonas gilardii. One strain belonged to Roseomonas genomospecies 4. The 22 strains represented by strain MDA5527 showed significant differences genotypically and phenotypically with R gilardii and other Roseomonas species and represented a new Roseomonas species; Roseomonas mucosa sp nov was proposed to denote its prominent mucoid, almost runny colonies. Eight strains, represented by strain MDA5605, had minor differences with R gilardii and displayed obvious pink to red colonies; Roseomonas gilardii subsp rosea subsp nov was proposed. For subspecies differentiation, R gilardii was proposed to be R gilardii subsp gilardii subsp nov. Unique patterns of biochemical reactions were established for these Roseomonas species, which may assist routine identification of these organisms. All 36 strains and 2 American Type Culture Collection strains were susceptible to amikacin and ciprofloxacin but resistant to cefepime and ceftazidime. They also were frequently susceptible to imipenem and ticarcillin-clavulanate but far less susceptible to ceftriaxone, trimethoprim-sulfamethoxazole, and ampicillin. R mucosa strains were most resistant, whereas R gilardii subsp gilardii strains were most susceptible.

Five cases of catheter-related bacteremia caused by Gordonia terrae are reported. All patients who also had the primary diagnosis of cancer experienced nonneutropenic fever as a result of G. terrae infection. All patients were treated successfully with antibiotics, with the requirement of catheter removal for 2 patients who had systemic infections.

We report the design and evaluation of a quantitative real-time polymerase chain reaction (PCR) assay to diagnose invasive mold infection (IMI) by detecting mold DNA in the serum. This assay detected 200 fg to 20 ng (5-log range) mold DNA and permitted a cutoff of 110 fg (3 genomes). Human or candidal DNA was not amplified. Specificity also was demonstrated by negative results in all 35 patients (76 serum samples) with unlikely IMI at the cutoff. For patients with possible, probable, and documented IMI diagnosed by a combination of clinical, microbiologic, and histologic criteria, this real-time PCR showed positivity in 40% (12/30), 68% (19/28), and 85% (11/13) cases, respectively, in testing of multiple serum samples. The overall serum positivity rate for these patients was 15.1% (73/483). Quantitative analysis of the positive serum samples estimated the bodily circulating mold burden to be 1.6 x 10(5) genomes (5.3 ng) by geometric mean with 4.2 x 10(7) genomes (1,400 ng) the highest. These results suggest that for the diagnosis of IMI, this real-time PCR may be a promising alternative to other invasive methods. Further evaluation is underway.