This issue of Annals includes two articles describing different manifestations of infection caused by Rochalimaea species, a slow-growing, fastidious, gram-negative bacillus [1]. Rochalimaea species play a significant role in two syndromes, cat scratch disease (R. henselae) and bacillary angiomatosis (R. henselae and R. quintana), for which, until recently, attempts to isolate and culture the responsible pathogen(s) have been difficult. Application of new technologies that analyze the 16S rDNA present in bacteria has helped solve these diagnostic dilemmas. This is possible because the genes for rRNA contain sequences that have changed only slightly over time, along with moderately conserved sequences and highly variable ones. The DNA from bacteria in tissue can be isolated and the 16S rDNA sequence amplified using the polymerase chain reaction. The similarities of the organisms can then be deduced by comparing their sequences.

Cat scratch disease was recognized as a clinical syndrome in the 1930s, but the first report was not until 1950 [2, 3]. A history of contact with cats is found in 90% of patients and antecedent cat scratch in 60% [4]. Often the initial symptom is the formation of a small erythematous papule or pustule at the site of the scratch that persists for several weeks. Subsequently, lymph nodes draining the site of inoculation become enlarged and tender. Involved nodes undergo sequential changes characterized by reticulum cell hyperplasia, granuloma formation, microabscess development, and suppuration. Low-grade fever and malaise are seen in 30% of patients. Less often patients experience rash, hepatosplenomegaly, lytic bone lesions, granulomatous conjunctivitis, pneumonitis, and central nervous system involvement [5]. This process usually lasts 2 to 4 months and resolves spontaneously. Antibiotics have not been shown to be clearly beneficial [4, 6].

Its clinical and epidemiologic features support an infectious cause. With no method to identify the causative organism, diagnosis has depended on an array of typical signs and symptoms, a history of animal exposure (usually to a cat or dog), a biopsy result from the node that is histopathologically consistent with cat scratch disease, and a positive skin test to an extract from lymph nodes from patients with this disease.

A breakthrough occurred in 1983 when Wear and coworkers [7] identified bacilli in a lymph node from a patient with cat scratch disease. Attempts to culture a pathogen were unsuccessful. However, the Warthin-Starry silver impregnation stain revealed many small rod-shaped bacteria that were Gram's stain negative. Prospective analysis of 39 lymph nodes with histologic changes consistent with the disease demonstrated bacilli in 29 of 34 patients by Warthin-Starry stain.

In 1988, English and coworkers [8] successfully cultivated a gram-negative organism from lymph nodes of 10 of 19 patients with cat scratch disease. Characterization of these isolates demonstrated gram-negative, oxidase-positive, nonfermenting rods subsequently designated as Afipia felis [9]. However, rapidly increasing evidence supports a role for R. henselae in most cases of cat scratch disease. Regnery and coworkers [10] at the Centers for Disease Control and Prevention, using an indirect fluorescent assay for antibody to R. henselae, found significant titers in 88% of persons suspected of having the disease compared with 3% of healthy controls. Of note, only 10 of 41 affected patients had antibody to A. felis. Five cat scratch disease skin test antigen preparations were tested by polymerase chain reaction-based methods for evidence of Rochalimaea and Afipia DNA. All had positive responses for the primers to detect Rochalimaea species and showed no evidence of Afipia [11].

The report by Dolan and colleagues [12] in this issue of Annals provides further support for an etiologic role of R. henselae. Two adult patients with illnesses consistent with cat scratch disease had excisional biopsies of the involved nodes. In both cases, small pleomorphic, gram-negative bacilli were isolated. Whole-cell fatty acid profiles were similar to R. henselae but differed from A. felis. Analysis of polymerase chain reaction-amplified DNA restriction endonuclease, fragment-length polymorphism revealed additional evidence that the isolates were R. henselae. Now the ability to cultivate the organisms and analyze their 16S rDNA and also measure specific antibody from patients with the disease should clarify the relative etiologic roles of Rochalimaea and Afipia. These advances also should further our understanding of its clinical spectrum and provide a basis for determining optimal therapy.

Bacillary angiomatosis, characterized by proliferation of small blood vessels in skin and visceral organs, has been recognized increasingly since the onset of the acquired immunodeficiency syndrome (AIDS) epidemic. In 1983, one of us (RTS), with Stoler and associates [13], reported the case of a 32-year-old man with AIDS in whom multiple firm, nontender subcutaneous nodules developed. Histologic evaluation revealed a proliferation of histiocytic and endothelial cells with a diffuse infiltration by polymorphonuclear leukocytes. Warthin-Starry silver staining showed small bacilli distributed diffusely in the interstitium. Since this initial report, the spectrum of this syndrome has been characterized more fully [14, 15]. Cutaneous involvement can take multiple forms [14]. Superficial lesions of the papillary and upper reticular dermis appear as red, purple, or skin-colored papules that may ulcerate. Deeper subcutaneous nodules are skin colored, mobile, or fixed to underlying structures. Histologic examination shows that cutaneous lesions are composed of circumscribed lobular aggregates of ectatic capillaries lined with protuberant endothelial cells surrounded by a mucinous stroma with a predominantly neutrophilic inflammatory infiltrate. Variable degrees of endothelial cell nuclear atypia and necrosis are seen that may be mistaken for angiosarcoma. In contrast, Kaposi sarcoma, which may be confused macroscopically with bacillary angiomatosis, is characterized by slit-like vessels with spindle-shaped endothelial cells and a lymphocytic inflammatory infiltrate [14]. In bacillary angiomatosis, bacteria often can be demonstrated with Warthin-Starry silver staining. Involvement of visceral organs, particularly the liver and spleen, may occur concomitantly or independently of the cutaneous lesions. These visceral lesions are characterized by multiple, dilated, blood-filled spaces (often called peliosis) separated by normal parenchyma and fibromyxoid stroma containing a mixture of inflammatory cells, dilated capillaries, and clumps of rod-shaped, Warthin-Starry stained bacteria. Although most reported cases have occurred in human immunodeficiency virus-infected patients, bacillary angiomatosis can be seen in apparently immunocompetent persons, as reported by Tappero and coworkers [16] in this issue of Annals and previously [17-19]. Unlike cat scratch disease, the various forms of bacillary angiomatosis respond to treatment with antibiotics such as erythromycin and the tetracyclines.

Attempts to culture the responsible pathogen from tissue and blood of patients with bacillary angiomatosis were unsuccessful until Slater and coworkers [17] isolated a motile, curved, gram-negative bacillus in primary culture of blood after lysis-centrifugation in five patients with fever. Growth required the use of fresh blood or chocolate agar for periods of 2 to 4 weeks [17, 18]. Cellular fatty acids of these isolates were compared with various bacteria and found to be similar yet distinct from Rochalimaea quintana, the agent of trench fever [17]. Simultaneously, Relman and coworkers [20] amplified, using polymerase chain reaction, the 16S ribosomal RNA gene from bacillary angiomatosis lesions and demonstrated sequence similarity to R. quintana. More recently, Koehler and associates [21] cultivated organisms from bacillary angiomatosis cutaneous lesions of four patients. Analysis of the amplified 16S rDNA sequence indicated that the organism was R. quintana in three patients and R. henselae in one.

Rochalimaea species are closely related to Bartonella bacilliformis, which also causes an angioproliferative host response and are more distantly related to Afipia. As with cat scratch disease, the ability to detect Rochalimaea infection by direct culturing of blood and tissue on solid agar, after coculture with vascular endothelium, and by 16S rDNA analysis of infected tissues should help clarify the spectrum of disease caused by these organisms (Table 1). We need to understand why in one host Rochalimaea infection manifests as cat scratch disease and in another as bacillary angiomatosis; why the latter responds to antimicrobial drugs, whereas cat scratch disease rarely does; and when disease is caused by new infection rather than possible reactivation of latent infection. Continued interchange between astute clinicians and molecular biologists is likely to answer these intriguing questions.