The Brief Case: Cough in an Immunocompromised Patient

DISCUSSION

This case highlights the fact that distinguishing between Haemophilus and Brucella is important not only clinically but also for maintaining the safety of laboratory personnel working with highly infectious organisms.

Haemophilus, a genus of Gram-negative coccobacilli important in both human and zoonotic disease, is transmitted via respiratory droplets or direct contact. The species appears as short (1.5-μm) coccobacilli, sometimes in pairs or chains, on a Gram stain (1). H. influenzae is particularly important as a cause of exacerbation of chronic obstructive pulmonary disease and CAP. Strains of H. influenzae may produce one of six distinct capsular polysaccharides (which provide the basis for serotype designation) or may be nonencapsulated. Nonencapsulated H. influenzae strains are referred to as nontypeable. An enriched medium, with factors X (hemin) and V (NAD/NAD⁺) added, is required for the isolation of H. influenzae. The bacterium typically does not grow on sheep blood agar except around colonies of Staphylococcus aureus (known as a “satellite phenomenon”). This phenomenon occurs because both X and V factors are present in erythrocytes, and hemolysis by S. aureus releases both, thus allowing for growth on the blood agar plate (1). H. influenzae can also be identified through automated systems, including matrix-assisted laser desorption ionization–time of flight (MALDI-TOF) systems.

In comparison, Brucella, also a small Gram-negative coccobacillus, is easily transmitted by aerosol and percutaneous inoculation, requiring a low infective dose (2). Most automated blood culture instruments can detect Brucella in <5 days, although detection is often delayed, typically 3 to 4 days, in contrast to <36 h for most other pathogens (2). While rapid methods to identify suspected agents of bioterror (such as MALDI-TOF) have been described, not all laboratories have this capability, and not all have validated the instrument for highly infectious agents (3).

Thus, adherence to the steps outlined in the American Society for Microbiology Sentinel Level Clinical Laboratory Guidelines for identification of a suspect agent of bioterrorism is particularly important in order to distinguish between the two species (2). First, small (0.4- by 0.8-μm), poorly staining Gram-negative coccobacilli are seen on a Gram stain. Following the identification of suspicious colonies, plates should be taped shut, and all further work should be performed in a biosafety cabinet. Brucella will typically grow on blood but, like Haemophilus, grows poorly and after prolonged incubation. Both grow equally well on chocolate agar; however, Brucella appears as pearl-white, nonhemolytic colonies, while Haemophilus appears as “morning dew drops.” Neither species grows well on MacConkey agar, and both are oxidase and catalase positive. Finally, unlike most Haemophilus species, Brucella is rapidly urease positive, often within 15 min to 24 h. At that point, the presumed identification is Brucella, and further identification with automated systems should be avoided due to the high concentration of organisms as well as a high risk for aerosolization (2, 3).

Key components in the initial misidentification of this bacterium led the microbiology team to suspect Brucella rather than Haemophilus. First, the satellite test was initially negative, prompting the exclusion of Haemophilus as a potential organism. However, the satellite test has a number of problems that can result in false-negative results. Dealler et al. (4) estimated that approximately 10% of satellite tests are interpreted falsely or have to be repeated. One explanation is that Haemophilus generally has poor growth on the blood agar medium, despite the addition of S. aureus. In addition to this, the plate itself can be inadvertently inoculated with growth factors or contaminated with other bacteria. Finally, the test is not frequently performed and thus may be misinterpreted by clinical microbiology labs. Another component in the initial misidentification of the bacterium was the rapidly positive urease test. The urease test is used to determine if an organism can split urea using the enzyme urease. The test utilizes Christensen’s urea agar, a medium that contains urea (2). If urease is present, ammonia is released, forming ammonium hydroxide, which alkalinizes the medium area of inoculum and changes the color from yellow-orange to pink (Fig. 2). Very few organisms can degrade urea in <4 h (considered to be a rapidly positive test). These organisms include Proteus spp., Helicobacter spp., and, importantly, Brucella spp. However, urease test positivity is variable in Haemophilus species, with a positive reaction occurring more classically with species such as Haemophilus aegyptius and Haemophilus haemolyticus. It is particularly important, however, that urease positivity in H. influenzae is in fact quite variable, and if a positive reaction should occur, it is typically less rapid. It is not known what percentage of H. influenzae strains have a positive urease reaction. Our case demonstrated a rapidly positive urease test, within 15 min, which would lead one to consider Brucella spp. rather than Haemophilus spp.

• Open in new tab
• Download powerpoint

FIG 2

(A) Growth of colonies of H. influenzae on chocolate agar after 24 h. (B) Satellite growth of Haemophilus around a streak of Staphylococcus aureus on sheep blood agar after 24 h. (C) Positive urease slant test for Haemophilus influenzae.

Finally, it is important to discuss the antimicrobial therapy that this patient received. The patient was initially treated with azithromycin and later completed inpatient therapy at an outside institution; thus, the duration and type of antibiotics she received there remain unknown. Susceptibility testing on the isolated H. influenzae strain revealed that it was negative for β-lactamase, and no further susceptibility testing was done on the isolate. Interestingly, the initial treatment course with azithromycin is recommended as standard-of-care therapy for patients without comorbidities or risk factors for resistant pathogens (5). However, given the patient’s active malignancy and immunosuppressive therapy (and the presence of bacteremia), single-agent oral therapy would not be appropriate, and combination therapy, including a β-lactam (such as a third-generation cephalosporin) would be advised. With regard to macrolide resistance, H. influenzae isolates worldwide show low-level intrinsic resistance that acts in combination with mutational and acquired resistance. Resistance typically occurs in the presence of an intrinsic or acquired efflux pump, or through mutations in 23S rRNA and other ribosomal proteins (6). The clinical significance of macrolide resistance to H. influenzae is debated, since MICs are generally relatively low for azithromycin (from 0.1 to 4 μg/ml), and antibiotic failure is more common when the isolate demonstrates higher MICs (6). However, antimicrobial susceptibility testing for Haemophilus is not typically done, since the in vitro testing demonstrates considerable inter- and intralaboratory variation and does not typically reflect the patient’s outcome (6).

We report a case of Haemophilus infection that was rapidly urease positive, with an initially negative satellite test—both characteristics that would typically be concerning for Brucella identification. Furthermore, the patient failed treatment with an antimicrobial generally expected to treat Haemophilus. Nevertheless, appropriate handling of the specimens via institutional protocols was of the utmost importance. Clinical microbiologists must have an understanding of the technical principles that are necessary for the recognition and differentiation of a potential biological threat from other species.