INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Bartonella henselae is the causative agent in most cases of cat scratch disease (CSD) a common cause of subacute regional lymphadenopathy in mostly immunocompetent children and adults. Patients are typically scratched or bitten by a cat, and after 3 to 10 days, skin lesions such as pustules or papules develop at the inoculation site. During the next 1 to 3 weeks, regional lymph nodes enlarge, remain stationary for another 2 to 3 weeks, and then resolve spontaneously over an additional period of 2 to 3 weeks (3). These typical clinical manifestations and a history of cat contact should lead to the presumptive diagnosis of CSD. The diagnosis can be confirmed by detection of antibodies to B. henselae in the patient's sera (13, 14, 17), by histopathological examination (10, 12, 20), and by molecular detection of B. henselae DNA from the patient's biopsy (1, 2, 4, 7, 10, 12, 20). Histopathological findings in the lymph nodes depend on the stage of infection. There may be lymphoid hyperplasia, arteriolar proliferation, and reticulum cell hyperplasia early in the course of infection. Granulomas with central areas of necrosis, multinucleated giant cells, and stellate multiple microabscesses may be found in later stages (3, 11). However, histopathological findings are typical but not specific for CSD. Infections caused by other agents, such as lymphogranuloma inguinale caused by Chlamydia trachomatis, atypical mycobacteriosis, yersiniosis, tularemia, brucellosis, certain mycoses, and chronic granulomatous disease of childhood must be considered in the differential diagnosis (11). Detection of B. henselae DNA in tissue samples therefore would be useful to confirm histologically suspected CSD.

Recently, several PCR-based assays have been developed for detection of Bartonella DNA in clinical samples. Large differences were found concerning the sensitivities of these assays, depending on whether fresh or formalin-fixed, paraffin-embedded tissue was investigated.

In a retrospective study, we compared the sensitivities of two PCR assays: one assay was based on the amplification of a 296-bp fragment of the 16S rRNA gene as described by Relman et al. (15), and the second assay amplified parts of the Bartonella htrA gene encoding a 60-kDa heat shock-like protein as described by Anderson et al. (1). Additionally, a genotype-specific PCR for B. henselae (5) was performed with all lymph nodes to differentiate between the two different genotypes of B. henselae involved in CSD.

The study examined lymph nodes from 60 patients with histologically suspected CSD. From 24 of these 60 patients, serum samples taken at the time of surgery were available for serological testing.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Lymph node samples. Paraffin-embedded lymph node biopsies from 60 patients with histopathologically suspected CSD were included in this study. The samples were obtained retrospectively for a period of 7 years, from January 1989 to December 1996, by the Institute of Pathology.

Histopathological investigation. The lymph node specimens were fixed in 10% buffered formalin, embedded in paraffin, cut at 2 to 3 µm, and routinely stained with hematoxylin and eosin. Twelve paraffin-embedded lymph nodes without any histologic evidence of CSD were used as negative controls. Warthin-Starry staining was not performed in our study.

DNA extraction. DNA was extracted from the formalin-fixed, paraffin-embedded lymph node biopsies by using a commercially available kit (Qiagen GmbH, Hilden, Germany) as proposed by the manufacturer. The extracted DNA was used as a template in the PCR assays. Purified DNA from cultured bacterial strains of B. henselae (Houston-1; ATCC 49882) was used as a positive control.

Amplification of Bartonella DNA. The primers p24E (5'CCTCCTTCAGTTAGGCTGG3') and p12B (5' GAGATGGCTTTTGGAGATTA3'), previously described by Relman et al. (15), were used to amplify a 298-bp fragment of the Bartonella 16S rRNA gene by PCR as described elsewhere (16). The reaction mixture consisted of bovine serum albumin (8 ng/µl), deoxynucleoside triphosphates (200 µM), primers (117 nM each), Taq polymerase (Pharmacia Biotech [2 U]), and 2.5 µl of extracted DNA in 50.0 µl of TBE (Tris-borate-EDTA) buffer. The mixture was overlaid with two drops of light mineral oil. PCR amplifications were performed in an automated thermal cycler (Robocycler 40; Stratagene) with initial denaturation (95°C, 5 min), followed by 30 cycles of denaturation (94°C, 1 min), annealing (57°C, 1 min), and extension (72°C, 90 s), with a single final extension at 72°C for 3 min.

Serological testing. Serum samples taken from 24 of the 60 patients with suspected CSD at the time of lymph node biopsy were available. All sera were stored frozen at 70°C. Serological testing for immunoglobulin G (IgG) and IgM antibodies to B. henselae was performed with a commercially available indirect immunofluorescence antibody test (Bios, München, Germany) as described previously (17). Titers of <1:64 were regarded as negative.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

The patients' characteristics (age, gender, and site and diameter of the infected lymph nodes) are shown in Table 1. Nineteen of the extirpated lymph nodes had been localized as cervical, 16 as axillar, and 13 as inguinal, respectively.

Histopathology. Histopathological examination of the 60 extirpated lymph nodes showed epithelioid cells next to necrotic tissue particles, necrotizing granulomatous inflammation, multiple stellate microabscesses with mixed hyperplasia, and perilymphadenitis, compatible with the diagnosis of CSD.

Amplification of Bartonella DNA. In 39 of the 60 lymph nodes (65%), B. henselae DNA could be detected by PCR. With the primer pair p12B-p24E, a positive PCR result was obtained from 36 lymph nodes (60%), whereas the second primer pair, CAT1-CAT2, amplified B. henselae DNA only in 26 (43.3%) of the 60 samples. Concordant positive results were obtained from 23 of the 39 lymph nodes, 13 samples were positive only with primer pair p12B-p24E, and 3 samples were positive only with primer pair CAT1-CAT2. By type-specific PCR, 23 of the 39 PCR-positive lymph nodes (59%) were found to belong to genotype I, and 9 (23%) belonged to genotype II, whereas 7 (18%) lymph nodes were negative in both type-specific PCR assays (Table 1). No specimen negative in the B. henselae PCR reacted with the genotype-specific primers. Only the 26 specimens positive in the PCR with primers CAT1 and CAT2 reacted with the B. henselae-specific oligonucleotide RH1. None of the 39 PCR-positive lymph nodes reacted with probe RQ1 (B. quintana).

Serological testing. All but 1 of the 24 serum samples available showed elevated IgG antibodies to B. henselae. In addition, most of them contained elevated IgM antibodies. Bartonella DNA could not be detected by the three PCR assays of the lymph nodes in three of the patients with high antibody titers (Table 1, samples 19, 22, and 24 from the serology group). In only 1 serum sample (no. 23) were both IgG and IgM titers negative, but in this case, all PCR assays remained negative as well (Table 1). It remains to be clarified in this case if the lymphadenopathy was caused by B. henselae or by another agent. Of the 23 patients with CSD confirmed by both histology and serology, 20 had PCR-positive lymph nodes with a primer reaction pattern comparable to that of the unselected population.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

The clinical features of CSD were described nearly 50 years ago (8), but B. henselae as the etiological agent of this disease was recognized only a few years ago and confirmed by serological and molecular studies (1, 2, 4, 5, 10, 12, 13, 14, 17, 20). Even today, the symptoms of CSD remain often unrecognized, and the diagnosis is based on the histological examination of a surgically removed lymph node or a biopsy. With the Warthin-Starry silver stain, the bacilli can be detected in tissue specimens, but the technique is difficult and the result is not specific for Bartonella. By this method, Scott et al. (20) demonstrated a few pleomorphic bacilli compatible with the CSD agent in only 14% of 42 formalin-fixed lymph node biopsies.

A better approach to a specific diagnosis is provided by DNA amplification methods. In the same study by Scott et al. (20), B. henselae DNA was found in 27 of 42 (64%) histologically defined lymph node biopsies and in 23 of 34 (68%) specimens from patients with CSD diagnosed both clinically and by histology (Table 2). The first primers for molecular identification of the agent of bacillary angiomatosis were described by Relman et al. (15). Using these primers and Southern blotting of the PCR products, Bergmans et al. (4) found B. henselae DNA in a high percentage of their CSD patients. These results were confirmed in our study. In 1994, Anderson et al. (1) described a new primer set for PCR detection of Bartonella DNA in specimens from CSD patients. This primer pair has been used in many studies (2, 10, 12) with good results (Table 2). However, the problem with all evaluations so far has been the lack of a defined "gold standard" for CSD or for the presence of B. henselae. Neither histology nor clinical symptoms or serology alone is satisfactory. Reliability, however, increased in most studies if two criteria were used in combination. Thus, in our study, the sensitivity of PCR (with both PCR assays) increased from 65 to 87% if histology was confirmed by the results of serology. A similar increase was seen by Bergmans et al. (4) (Table 2). Although not specific, the histological diagnosis of CSD appears to be quite accurate in cases confirmed by serology. Only 1 of our 24 patients for whom serum was available had no serological evidence for a B. henselae infection. This suggests a correct histopathological result in 96% of the cases studied.

In the absence of a gold standard for diagnosis of CSD, we compared our PCR results with the histopathological interpretation of the investigated lymph nodes. However, histopathological findings are typical but not specific for CSD. Especially in cases with negative PCR results and lack of serological testing, we have to consider that the histopathological findings might be caused by other agents and that these patients had been suffering from a disease other than CSD. Although in our study two different primer pairs were used, only 65 and 87% of the samples without and with serology, respectively, were positive, and the results were even less satisfying if the percentages for the primers were considered separately. A small number of bacteria, below the detection limit, is a possible cause. However, we assume that the false-negative reactions are more likely due to the various steps of fixation and embedding of the tissues known to damage DNA. The variable results obtained with the three primer pairs (including genotype-specific PCR) suggest random destruction of the DNA, with the smallest target (type-specific) resulting in the highest sensitivity (82%). The assumption is supported by the fact that with untreated or frozen lymph nodes, PCR often showed a higher detection rate (1, 2, 4). However, the sensitivity of the PCR assays increased from 65 to 87% in our study when two diagnostic criteria (histopathology and serology) were combined.

The results of our study and those of others (5, 9, 19) indicate that at least two genotypes of B. henselae are involved in CSD. Bergmans et al. (5) demonstrated that the majority (32 of 41 samples) of the lymph nodes from patients with CSD in The Netherlands contained B. henselae genotype I (78%), 7 of 41 belonged to genotype II (17%), and 2 samples (5%) were found to be negative in both type-specific PCRs. Similarly, in our study, 59% (23 of 39) of the PCR-positive patients were infected with B. henselae genotype I, 23% (9 of 39) were infected with B. henselae genotype II, and 7 (18%) of the lymph nodes were negative in both type-specific PCRs. In contrast, a study in Switzerland of 34 human clinical specimens containing B. henselae DNA had revealed 9 infections with type I but 25 infections with type II (6).

Furthermore, 16 of 17 B. henselae isolates from Southern German cats belonged to genotype II, and only 1 isolate was of genotype I (18). These results suggest that different genotypes of B. henselae are prevalent in different geographic regions (e.g., The Netherlands, Germany, and Switzerland) or that B. henselae genotype I could be more pathogenic to humans than genotype II (genotype of cat isolates versus genotypes in human lymph nodes in Germany).

We conclude that the detection by PCR of B. henselae in tissues of patients with suspected CSD is an useful diagnostic method complementing histopathological and serological analysis. At least two different primer pairs should be used for higher sensitivity, especially in prefixed materials. The different distributions of the two genotypes in cats and humans have yet to be explained sufficiently. In addition, whether the clinical presentation is somehow dependent on the type of the infecting strains remains to be analyzed.