Syndrome of Rochalimaea henselae Adenitis Suggesting Cat Scratch Disease

Case Reports

Patient 1

A 68-year-old white man with a history of mild hypertension was seen in the emergency department; he complained of 72 hours of fever to 39.2 °C and an enlarging mass at his left elbow. In the emergency department his temperature was 39.4 °C and he had a small eschar in the interphalangeal space of his left hand between the third and fourth digits that he reported had been present for 2 months. His environmental exposures included small lacerations from rose bushes he tended and contact with a pet cat. His leukocyte count was 17.3 × 10⁹/L with a predominance of neutrophils and band forms. The patient was given a 10-day course of cefadroxil. He became afebrile during therapy but because the left epitrochlear mass was enlarging and becoming painful, he again sought medical attention.

At this time he had a small 1 × 0.5-cm nonhealing ulcer without eschar on the dorsum of the 3 to 4 interphalangeal space of the left hand and a 4 × 4-cm, tender, nonfluctuant left epitrochlear lymph node. No other adenopathy was present, and the rest of the examination was unremarkable. Laboratory test results, including complete blood count, chemistry profile with liver functions studies, urinalysis, and erythrocyte sedimentation rate, were normal. Result of a test for HIV-1 was negative. A chest roentgenogram was normal. Serologic test results for tularemia, brucellosis, and syphilis were negative. A purified protein-derivative skin test result was negative with positive controls. An excisional biopsy of both the interdigital lesion and the epitrochlear node was done. Histopathologic examination revealed similar processes in both sites characterized by necrotizing granulomas in a background of chronic inflammation, with multinucleated giant cells and a mixed perivascular infiltrate (Figure 1). No organisms were seen with tissue acid fast or Gram stain, Warthin-Starry stain, or Gomori-methenamine-silver stain.

View larger version:

• In this window
• In a new window

Figure 1. Granulomatous inflammation in the skin lesion. (Original magnification, × 50.) Rare multinucleated giant cells in the skin lesion. (Original magnification, × 470.) Perivascular inflammation in the involved lymph node. (Original magnification, × 230.) All samples were stained with hematoxylin and eosin. Photomicrographs of the papular hand lesion and lymph node of Patient 1.Top left.Top right.Bottom.

A sample of the lymph node was cultured, and an organism was isolated that was identified as R. henselae. The patient was treated with a 28-day course of doxycycline and ciprofloxacin. The lesion healed after 10 days, and he has remained symptom free for 9 months.

Patient 2

A previously healthy Hispanic 27-year-old man developed a tender left axillary lymph node while traveling through Hawaii, Thailand, Guam, and Bangladesh as a flight crew member. Three days later, he developed fevers to 38.9 °C, light-headedness, fatigue, and nausea with one episode of vomiting, chills, diaphoresis, and headache. The systemic symptoms resolved after 3 days, but the node continued to enlarge and became more painful. When we examined him, he was afebrile and complained only of the painful node that limited use of his left arm. The patient owned and handled cats, dogs, and goats and had received tick bites, although not on the affected limb.

He had a 4 × 5-cm nonfluctuant, tender left axillary lymph node with minimal overlying erythema, a 1 × 1-cm left epitrochlear lymph node that was nontender, and a 1 × 1-cm eschar on the palmar surface of his left hand. The examination was otherwise unremarkable. Values were normal for a complete blood count, serum chemistry tests, urinalysis, and erythrocyte sedimentation rate. Serologic test results for tularemia, brucellosis, plague, syphilis, and scrub typhus were negative, as was a test for HIV-1. The purified protein derivative skin test was negative with positive controls. An excisional biopsy of the axillary lymph node was done, revealing hard, inflamed, adherent, matted nodes at surgery. The patient was given a 14-day course of cephradine followed by a 28-day course of doxycycline after surgery. Histopathologic examination of the lymph node sample showed findings similar to those of the skin and lymph node biopsies of Patient 1.

Cultures of the biopsy material grew rare, small colonies of R. henselae after a prolonged incubation. Concurrent cultures of the blood remained negative after 30 days. The patient recovered and has remained symptom free for 12 months.

Methods

Microbial Isolation

Biopsy material was placed on saline-soaked gauze and carried directly to the microbiology laboratory, where it was ground in a tissue grinder and plated onto chocolate and CDC anaerobic blood agar (sheep blood with added hemin and L-cysteine; BDMS, Cockeysville, Maryland), Jem Bec plates (BDMS) and placed into liquid phase media, including cooked meat broth (BDMS) and eugonic broth (modified, Remel, Lexena, Kansas). Fungal and mycobacterial cultures were also done. These materials were then incubated at 35 °C in 5% CO₂, with plates kept upright for 24 hours and then inverted. Plates were incubated for up to 6 weeks under these conditions to optimize the isolation of fastidious organisms. For determination of X and V factor dependence, cultures were plated onto brain-heart infusion agar with X, V, and XV strips (BDMS).

Broth Microdilution Antimicrobial Susceptibility Tests

The two R. henselae isolates described above and a previously described blood isolate [7] were tested for susceptibility to a wide array of antimicrobial agents by a broth microdilution susceptibility method analogous to that used to test Haemophilus influenzae. Specifically, microdilution trays were prepared with twofold concentration increments of the various antimicrobial agents incorporated into Haemophilus test medium [21, 22] dispensed into 100-µL/well aliquots. The bacterial inoculum was prepared from growth of the test strains that had been isolated from a single colony and subcultured on enriched chocolate agar and incubated for 5 to 7 days at 35 °C in 5% CO₂. The final inoculum density used for all susceptibility tests was approximately 5 × 10⁵ colony-forming units per mL derived by suspending growth in 0.9% NaCl to a turbidity of 0.5 McFarland standard. Plate counts verified the appropriateness of the final inoculum. Microdilution trays were incubated for 7 days at 35 °C in ambient air before minimum inhibitory concentrations were interpreted in the usual manner.

Whole-Cell Fatty Acid Analysis

Analysis of whole-cell fatty acids was done with a Hewlett Packard 5890 gas chromatography system and software (Microbial Identification System, Version 3.0, Microbial ID, Inc., Newark, Delaware). Cultures of R. henselae, R. quintana, and A. felis (strain B.V., Armed Forces Institute of Pathology, Washington, DC) were grown for 7 days on CDC blood agar at 35 °C in increased CO₂, and then colonies were harvested with a loop. The cellular material was then saponified, methylated, extracted, base-washed, and analyzed as recommended by the manufacturer.

DNA Analysis

The identification of the bacteria was confirmed by using a combination of the polymerase chain reaction (PCR) amplification of the citrate synthetase gene sequence and subsequent restriction fragment length polymorphism analysis of the resultant amplified DNA product [6, 23]. The PCR-amplified DNA was subjected to restriction-endonuclease digestion with various enzymes (HinfI, HhaI, MseI, and TaqI). The sizes of resulting DNA fragments were compared by polyacrylamide gel electrophoresis with DNA fragments prepared simultaneously from other Rochalimaea strains, including R. quintana (ATCC VR-358, isolate Fuller), R. vinsonii (ATCC VR-152, vole agent), and R. henselae, (Houston-1 prototype isolate).

Results

Colonies were noted in the area of application on the CDC anaerobic blood plates at day 33 and day 13 in the first and second cases, respectively. Growth was slower on chocolate agar and did not occur on the other media used. Afipia felis (strain B.V.) was subcultured onto blood agar under identical conditions to show that Afipia could be grown using the method described here to isolate R. henselae. Rochalimaea colonies were small, nonhemolytic, rough and dry, and yellow to grey in color. Gram stain revealed small, curved, pleomorphic gram-negative rods. The organism was oxidase and catalase negative and X-factor dependent. Growth occurred more quickly on subsequent subcultures. The most rapid growth was attained using human blood agar. Rochalimaea quintana has previously been shown to grow better on human than on sheep or horse blood agar [24].

Analysis of whole-cell fatty acids from the node isolates demonstrated profiles extremely similar to a previously identified R. henselae isolate [7] that was subcultured with similar media and conditions (Table 1). Major fatty acids included 18:1 ω 7c, 18:0, and 16:0. Others, including 18:1 ω 9c, 17:0, and 16:1 ω 7c, were present in small amounts. All the R. henselae isolates had a higher proportion of 18:0 to 16:0 fatty acid than R. quintana, consistent with data obtained using different agar media [25]. Notably absent was the characteristic 11-methyloctadec-12-enoic acid moiety that was seen in the A. felis isolate (estimated chain length, 18:080 ± 0.003) [26, 27].

Antimicrobial susceptibility testing assessed by the broth microdilution method showed favorable minimum inhibitory concentration values for several drugs tested, including ampicillin, second- and third-generation cephalosporins, rifampin, tetracycline, chloramphenicol, trimethoprim-sulfamethoxazole, aminoglycosides, macrolides, and an azalide (Table 2). Minimum inhibitory concentration values were variable for quinolones and were less favorable for the first-generation cephalosporins. This is notable because one patient we describe had increasing adenopathy after treatment with a first-generation cephalosporin, showing a correlation between in vitro resistance and in vivo failure.

Restriction fragment length polymorphism analysis of PCR-amplified citrate synthetase gene DNA from both of the lymph node isolates showed that each resultant restriction-endonuclease-cleaved DNA fragment comigrated with a comparable DNA fragment generated from the Houston-1 type strain of R. henselae (11 sets of comigrating fragments). In contrast, only three restriction-endonuclease-cleaved DNA fragments derived from the lymph node isolates comigrated with PCR-amplified citrate synthetase gene DNA from comparable fragments derived from either R. quintana or R. vinsonii.

Discussion

The diseases currently attributed to R. henselae include bacillary angiomatosis, peliosis hepatis, bacteremia, and aseptic meningitis. The cases we describe add a syndrome of unilateral adenitis with a distal papular lesion to this spectrum. Both patients discussed had tender adenopathy in an extremity; cat exposure; and histopathologic findings compatible with cat scratch disease [15, 28, 29]. Routine cultures of the nodes would ordinarily have shown no growth and would have been discarded at 72 hours. These factors, in association with a slowly developing regional lymphadenopathy and the presence of a skin lesion consistent with a primary inoculation papule, suggest a clinical diagnosis of cat scratch disease. A pattern of tissue histopathologic findings in cat scratch disease has been described that is characterized by granulomatous necrosis with a mixed cellular infiltrate and reactive follicular hyperplasia [18, 28]. Rare, multinucleated giant cells and a perivascular mixed cellular infiltrate have also been found [16, 29, 30]. Pleomorphic bacilli may be found in Warthin-Starry-stained tissue specimens but often are not seen. Although the histopathologic changes described for cat scratch disease are only suggestive of the disease and are not specific [15], they are entirely consistent with the lymph node and skin papule findings observed in the two cases we describe.

A clinical diagnosis of cat scratch disease can be made in both cases related above by meeting three of the four criteria, which include: “1) a history of animal [usually a cat or dog] contact, with the presence of a scratch or primary dermal or eye lesion; 2) aspiration of sterile pus from the node [a presumptive diagnostic test] or culture and laboratory data that exclude other etiologic possibilities; 3) a positive cat scratch disease skin test; and 4) a node biopsy with histopathologic findings consistent with cat scratch disease” [29]. Use of the cat scratch disease skin test antigen is severely limited by both the absence of a commercial source and by the potential for transmission of hepatitis C, HIV, or other pathogens that could contaminate the human-derived antigen [29]. Because the diagnosis of cat scratch disease has been made mainly on clinical presentation and tissue histopathologic examination, rather than on a microbiologic isolate and because the isolation of R. henselae requires that culture plates be held much longer than is customary in clinical laboratories and be incubated in an increased CO₂ environment, it is possible that undiagnosed R. henselae infection may be a cause of reported cat scratch disease.

The bacterium isolated, R. henselae, has been shown to differ from Afipia felis, a recently described prokaryote previously proposed as the etiologic agent of cat scratch disease, both metabolically (the former is oxidase negative; the latter, oxidase positive), as well as by whole-cell fatty acid content [25, 27]. Genomic differences are also clearly present between these two bacterial pathogens. The 16S ribosomal gene sequence analysis [1, 6, 7, 31], quantitative citrate synthetase PCR-restriction fragment length polymorphism analysis [6], and DNA hybridization studies [25] have previously shown that R. henselae is a unique species related to other species of the genus Rochalimaea and that A. felis is only distantly related to the genus Rochalimaea. Of the techniques currently available for unambiguous identification of Rochalimaea isolates, the PCR-restriction fragment length polymorphism method is relatively rapid and simple to perform. Such analyses used in confirming the identification of these two lymph node isolates show them to be genotypically identical to R. henselae by all criteria used for this analysis. Likewise, the Wilford Hall isolates and prototype R. henselae isolate were distinct from other recognized species within the genus Rochalimaea.

Laboratories attempting to isolate R. henselae from tissue should inoculate the sample directly onto a fresh blood or chocolate agar plate in increased CO₂ and hold the plates for as long as 6 weeks. Desiccation of the plates can be prevented by storing them in a gas-permeable, moisture-retaining envelope. Application of a relatively large inoculum of infectious material to culture plates may be important for isolation of the organism. Other patients we have seen who have had similar clinical presentations and serologic evidence of R. henselae infection have not had the organism isolated from aspirates of nonfluctuant lymph nodes. Because both R. henselae and the A. felis strain evaluated grew well in Haemophilus test medium, this liquid medium may be useful for isolation and broth dilution susceptibility studies of both of these potential etiologic agents of cat scratch disease. Blood cultured in automated systems may not register as positive for growth by CO₂-sensing techniques and should be subcultured onto solid media, as described above, or cultured using a lysis-centrifugation system [7]. Slow growth, appearance as a pleomorphic, curved gram-negative rod on gram stain, X-factor dependence, and metabolic tests (negative oxidase, catalase, and urease) help to identify the organism. Membrane fatty acid analysis by gas chromatography, a tool used increasingly in clinical microbiology laboratories, can contribute substantially to identification. The PCR-restriction fragment length polymorphism analysis performed on the isolates described here serves as an independent, definitive confirmatory method.

The natural history of untreated infection and the evolution of the various manifestations of R. henselae disease (adenitis, bacillary angiomatosis, peliosis hepatis, bacteremia, and aseptic meningitis) remain undefined. It is similarly unclear whether therapeutic requirements differ for these separate clinical forms of disease. Therapeutic end points can be confusing because R. henselae appears able to cause persistent bacteremia after resolution of symptoms, as has been described with R. quintana[9, 10] and Bartonella[8]. Antimicrobial therapy of bacteremic R. henselae infection in both immunocompromised and immunocompetent patients has been frequently associated with relapse, particularly when a tetracycline was used [6, 7, 25]. Erythromycin, noted to treat bacillary angiomatosis effectively [32, 33], is also effective in selected cases of peliosis hepatis [3] and is variably effective in R. henselae bacteremia [5, 7, 25]. Both patients with adenitis were treated with a prolonged antibiotic course because of the potential relapsing nature of a Rochalimaea infection and the trend described in cases of bacteremia for cures to be associated with prolonged therapy.

The question of whether R. henselae exists extracellularly, intracellularly, or both while infecting humans affects whether intracellular penetration is a necessary factor for optimal antimicrobial therapy. This issue has also been raised with R. quintana, which demonstrates in vitro antimicrobial susceptibilities that are similar to those of R. henselae[34]. Rochalimaea quintana in tissue culture [35] as well as in the louse vector [10] appears adherent and extracellular to host cells. Rochalimaea henselae can grow in entirely cell-free medium in vitro and might be treated effectively with a penicillin if no in vivo intracellular focus were present. Therapeutic failures with β-lactam type antibiotics for susceptible isolates suggest that some portion of the organisms either survive intracellularly in human infection or are sequestered in a tissue compartment into which β-lactam antibiotics cannot penetrate. Treatment using antimicrobics with good intracellular penetration may be prudent until correlation of in vitro and in vivo susceptibility data are available.

All three patients with culture-proven R. henselae infection diagnosed at Wilford Hall USAF Medical Center during the past 18 months owned and handled cats. These included two patients with unilateral adenitis and one previously described patient with persistent bacteremia [7]. None, however, recalled a scratch associated with the onset or location of their illness. Rochalimaea henselae has recently been isolated from the blood of a Rochalimaea-seropositive cat [36], supporting the association between cats, humans, and R. henselae in cat scratch disease. The first patient we described recalled having received a tick bite immediately before the onset of his symptoms. Amblyomma americanum, the Lone Star tick, is the principal tick that bites humans in the San Antonio, Texas area during the summer. It feeds on a variety of animals and can transmit the agent of Rocky Mountain spotted fever [37]. In view of these associations, it is possible that cats, cat fleas, or ticks, particularly Amblyomma americanum, may play a role in the epidemiology of R. henselae infection.

The views expressed in this article are those of the authors and do not reflect the official policy of the Department of Defense or other departments of the United States Government.