Cat-scratch disease (CSD) is largely due to infection with Bartonella henselae. Microbiologic detection is difficult, and molecular testing is not readily available. A monoclonal antibody (mAB) to B henselae has become commercially available. We evaluated the usefulness of immunohistochemical analysis (IHC) for diagnosing CSD on surgical specimens and compared these results with polymerase chain reaction (PCR) detection and serologic testing for B henselae. We studied 24 formalin-fixed, paraffin-embedded (FFPE) cases of lymphadenitis with histologic and/or clinical suspicion of CSD. Control cases included 14 cases of lymphadenopathy other than CSD. FFPE tissue sections were evaluated with an mAB to B henselae, Steiner silver stain (SSS), and PCR that targeted B henselae and Bartonella quintana. Positive cases were as follows: SSS, 11 (46%); PCR, 9 (38%); and IHC, 6 (25%). Only 2 cases (8%) were positive for all 3 studies. All control cases were negative for IHC and PCR. The diagnostic sensitivity of these 3 tests is low for CSD. SSS seems to be the most sensitive test but is the least specific. PCR is more sensitive than IHC and may, therefore, serve as a helpful second-line test on all IHC– cases.
Polymerase chain reaction
Cat-scratch disease (CSD) is a common cause of chronic lymphadenopathy among children and adolescents, and the major etiologic agent underlying it is the bacterium Bartonella henselae.1,2 The most frequent cause of pediatric infectious lymphadenitis is a bacterial infection, specifically streptococcal and/or staphylococcal infection.3 CSD is also a common cause of lymphadenopathy in childhood, estimated to occur in approximately 22,000 patients annually in the United States.3 Most patients with CSD have subacute lymphadenitis and an associated cutaneous inoculation site due to a cat scratch or bite. The local or regional lymphadenopathy generally develops 1 to 3 weeks after the primary inoculation, which is usually in the head, neck, axilla, or supraclavicular region. Atypical manifestations include mammary, oculoglandular, hepatosplenic, cardiopulmonary, central nervous system, and bone involvement.
Because B henselae is very fastidious and difficult to culture, the diagnosis of CSD has historically relied on clinical history, serologic studies, and/or histologic evaluation.4–7 Serologic assays have poor sensitivity and specificity because some patients never mount a detectable antibody response and other patients may remain serologically positive long after exposure and recovery from the illness. In addition, there is extensive serologic cross-reactivity between B henselae and Bartonella quintana, as well as with other intracellular bacteria.7,8
Lymph node biopsy is often performed to determine the cause of lymphadenopathy. The histopathologic findings in lymphadenopathy associated with CSD are nonspecific, however, and often depend on the stage of the disease. Reactive follicular hyperplasia is present in the early stages of the disease, followed by the development of the more characteristic stellate suppurative granulomas. These granulomas usually demonstrate acellular necrotic centers, with a surrounding zone of histiocytes and an outer zone of lymphocytes and plasma cells. The stellate microabscesses that develop within the granulomas may become confluent at a later stage.
B henselae is a gram-negative bacterium, but this organism does not readily stain with a Gram stain. Traditionally, silver impregnation stains such as Warthin-Starry or Steiner stains have been used to aid in the microscopic identification of B henselae organisms, which appear as pleomorphic bacilli in clumps or as single forms.9–11 The silver stains are cumbersome and expensive to perform. They are also difficult to interpret, given the high background of silver precipitate in necrotic material and within macrophages. Molecular testing using the polymerase chain reaction (PCR), often combined with Southern blotting, is becoming more widely available but has yet to become the standard method of diagnosis in clinical laboratories for the identification of B henselae.
Identification of B henselae by immunohistochemical analysis is an interesting diagnostic alternative. The goals of this study were to evaluate the detection of B henselae by immunohistochemical analysis in formalin-fixed, paraffin-embedded (FFPE) sections of lymph nodes suspected to be involved by B henselae and to compare the immunohistochemical results with available serologic data, Steiner silver stain (SSS) results, and a PCR assay.
Materials and Methods
We retrieved 24 FFPE lymph node biopsy specimens that were suggestive of CSD from the surgical pathology files of Baystate Medical Center, Springfield, MA, for a 10-year period. The study was approved by the institutional review board of Baystate Medical Center. Lymph nodes were included in the study group if CSD was a clinical consideration or if the characteristic histologic features of granulomas with central stellate microabscesses were present in evaluation of H&E-stained slides. Negative external control cases (n = 14) included nonspecific reactive lymphoid hyperplasia (n = 4), Toxoplasma lymphadenitis (n = 2), sarcoidosis (n = 2), Mycobacterium avium intracellulare lymphadenitis (n = 2), Hodgkin lymphoma (n = 1), dermatopathic lymphadenopathy (n = 1), Rosai-Dorfman disease (sinus histiocytosis with massive lymphadenitis) (n = 1), and Staphylococcus aureus lymphadenitis (n = 1). Pathology reports and available medical records were reviewed for pertinent clinical information, including patient age, sex, chief complaint, clinical history, serologic test results, culture results, and antibiotic therapy.
Biopsy tissue was processed routinely in 10% buffered formalin and embedded in paraffin. In addition to H&E stains, Grocott-methenamine silver stains were available in 22 study cases, Ziehl-Nielsen stains in 23 cases, and a Gram stain in 2 cases. All 24 study cases were stained with an SSS.
We evaluated 5-μm-thick sections of FFPE tissue by immunohistochemical analysis with the B henselae monoclonal antibody (clone H2A10, Biocare Medical, Concord, CA). The antibody was optimized by using microwave antigen retrieval with EDTA buffer (pH 9.0). The slides were heated to 97°C for 2 consecutive 10-minute heat cycles, with cooling for 10 minutes between heating cycles. At the end of the second cycle, the slides remained in the EDTA buffer for 20 minutes at room temperature. Immunohistochemical evaluation was performed on an automated DAKO AutoStainer platform (DAKO, Carpinteria, CA) with the primary antibody diluted at 1:100. A 45-minute primary antibody incubation was performed at room temperature. A peroxidase polymer–based detection system was used (Signet Labs, Cambridge, MA) with diaminobenzidine as the colorogenic reagent, followed by hematoxylin counterstain. All slides were reviewed separately by 3 of us (G.C.C., L.P., and C.N.O.).
Primer pairs were designed to target a 153-base-pair fragment of the 16S ribosomal RNA gene present in B henselae and B quintana.12–14 Following deparaffinization and Proteinase K digestion, DNA was extracted from two 25-μm sections of routinely processed, FFPE lesional tissue for all cases. DNA amplification was performed by PCR, and β-actin served as a “housekeeping” gene to ensure intact DNA in the samples. Positive control samples included purified genomic DNA obtained from previously positive patient samples. Reagent blanks and known positive cases of Yersinia enterocolitica and Mycobacterium tuberculosis were used as negative control samples. Size separation of amplicons was performed on 8% native polyacrylamide gel stained with ethidium bromide and visualized with UV light. PCR analysis of all samples was performed in duplicate. All cases were analyzed in the same laboratory, the Infectious Disease Molecular Research Laboratory at the University of Arkansas for Medical Sciences, Little Rock.
Serologic testing was ordered in 9 of the study cases. The method used was indirect immunofluorescence, and the tests were performed at the Centers for Disease Control and Prevention, Atlanta, GA. Serologic testing was considered positive if the IgG titers against B henselae and/or B quintana were equal to or greater than 1:64 or if IgM titers were equal to or greater than 1:20. Culture results were available in 10 of the study cases.
The mean age of the 24 patients in the study group was 17.6 years (range, 1–47 years); 17 were males. Lymph node biopsy specimens were obtained from the axillary region (10 cases), neck (5 cases), inguinal region (5 cases), preauricular region (1 case), and submandibular region (3 cases). Lymph nodes were often matted. One patient had simultaneous inguinal and axillary lymphadenopathy. In 8 of the 24 study cases, recent scratches by cats were reported. Exposure to cats was denied in 4 study cases. Exposure to cats was unknown in the remaining 12 cases.
A history of preoperative antibiotic therapy was obtained in 14 (58%) of the study cases. In the remaining 10 study cases, preoperative antibiotic therapy was not administered or not documented. The antibiotic agents were cephalexin, azithromycin, clindamycin, ampicillinclavulanic acid, and/or trimethoprim-sulfamethoxazole.
Suppurative granulomas and/or multiple stellate microabscesses were identified in 23 of 24 study cases and were morphologically compatible with the diagnosis of CSD Image 1A. One case demonstrated only reactive follicular hyperplasia Image 1B with nonnecrotizing granulomas, but this case was retained in the study group because the patient was serologically positive for B henselae. In 11 study cases (46%), SSS revealed several coccobacillary structures compatible with Bartonella species Image 2A. In 1 of the control cases an SSS was available, but this case was retained in the control group because Mycobacterium avium was identified in the culture. All of the other histochemical stains, including Grocott-methenamine silver stain (22 cases), Ziehl-Nielsen stain (23 cases), and Gram stains (2 cases), were negative in the study cases.
A, Coccobacillary microorganisms compatible with Bartonella spp (Steiner silver stain, ×100). B, Immunohistochemical identification of Bartonella henselae in an area of necrosis in a suppurative granuloma (B henselae monoclonal antibody, ×100).
Microorganisms compatible with B henselae were identified by immunohistochemical analysis in 6 (25%) of 24 study cases Image 2B. The immunoreactive bacteria were identified within the areas of necrosis of the suppurative granulomas. Organisms were not identified immunohistochemically in any of the 14 control cases. Of the 24 study cases, 9 (38%) were positive by PCR for Bartonella species DNA Table 1, and all 14 control cases were negative for Bartonella species by PCR. It is interesting that the identification of Bartonella species was not entirely concordant when immunohistochemical results were compared with PCR results (Table 1). Two cases were positive by immunohistochemical analysis and PCR, whereas 4 cases were positive by immunohistochemical analysis and negative by PCR. Conversely, 7 cases were positive by PCR and negative by immunohistochemical analysis. Of the study cases, 11 (46%) were positive by SSS. The results of all 3 tests are summarized in Table 2.
Serologic evaluation for B henselae was available in 9 study cases, and 7 of these cases were positive. One of the serology-positive cases demonstrated organisms by immunohistochemical analysis, and 4 were positive by PCR for Bartonella species. All microbiologic culture results were negative for Bartonella species.
The traditional approach to the diagnosis of CSD is based on 4 criteria, any 3 of which are sufficient to establish the diagnosis: (1) contact with a cat and presence of a scratch or other evidence of an inoculation site, (2) a positive CSD skin test or serologic test, (3) regional lymphadenopathy without other etiologic evidence for the adenopathy, and (4) the characteristic histopathologic features in tissue biopsy specimens.
Tissue biopsy is one of the most useful ways to diagnose CSD and to render other causes less likely in the differential diagnosis. Clinical suspicion of CSD may be reinforced by the histologic findings in lymph node biopsy material, sometimes in conjunction with serologic studies. However, neither histologic features nor clinical symptoms alone are entirely sufficient for the diagnosis of CSD, and, as discussed, there are numerous problems inherent in the serologic tests. Silver impregnation stains have historically been the preferred histochemical stain for detection of Bartonella species; however, organisms are only variably detectable, and these stains are technically challenging owing to the high background of silver precipitate. Although 11 (46%) of our study cases were positive using an SSS, only 6 of these specimens were concomitantly positive by immunohistochemical analysis and/ or PCR. This finding suggests that several of the cases interpreted as positive by SSS may represent false-positives. The identification of B henselae species is further complicated by the difficulty in culturing this fastidious microorganism and a lack of a defined “gold standard” method for the diagnosis of CSD or for detecting the presence of B henselae and other Bartonella species. Thus, there is a need for improved, clinically useful methods for the diagnosis of these infections.
This study assessed the usefulness of immunohistochemical analysis in the diagnosis of CSD using a monoclonal antibody (clone H2A10, Biocare Medical). This monoclonal antibody has been successfully used in previous studies to specifically demonstrate the presence of B henselae in cases in which CSD was considered on clinical and histopathologic grounds.15,16 According to the manufacturer, this antibody is of the IgG2a subclass, reacts with a 43-kDa epitope present only in B henselae, and shows no cross-reactivity with other Bartonella strains.
Our study revealed that 9 (38%) of the study cases clinicopathologically suspected of being CSD had PCR evidence of Bartonella DNA (Table 1). This finding is similar to the results of a previous study that found that 39% of 289 lymph node specimens from patients with lymphadenopathy had CSD confirmed by at least 1 of 3 PCR assays.17 Of our 24 study cases, 6 (25%) were positive by immunohistochemical analysis. Only 2 of our study cases were simultaneously positive by PCR and immunohistochemical analysis (Table 1). Seven of our study cases were positive by PCR and negative by immunohistochemical analysis Table 3. In 5 of these cases, preoperative antibiotics had been administered (Table 3). The discordance between the immunohistochemical and PCR results could be due to sampling bias (especially if the cases have only a focal presence of low numbers of bacteria) or to the inherent greater sensitivity of PCR compared with immunohistochemical analysis for CSD. In addition, because the immunohistochemical assay targets B henselae only and the PCR assay used in this study targets B henselae and B quintana, it is conceivable that some of the study cases that were positive by PCR but negative by immunohistochemical analysis could be B quintana infections.
All study cases that were negative by PCR and immunohistochemical analysis had histologic features consistent with CSD. These negative results can likely be explained by several factors, including reduced numbers of bacteria due to preoperative antibiotic therapy and low numbers of bacteria in early infection.
In a similar study, investigators tested the usefulness of immunohistochemical analysis using the same H2A10 clone for B henselae in 6 confirmed cases of CSD, 3 cases of bacillary angiomatosis, and 24 reactive (control) lymph nodes.15 All of their CSD and bacillary angiomatosis cases were positive by immunohistochemical analysis, showing easily identifiable bacilli. This is higher than the 25% immunoreactivity for microorganisms we report. They also found that immunostaining revealed more bacteria than did a Warthin-Starry stain. In another study involving 23 FFPE lymph nodes from 16 patients clinically suspected of having CSD, 9 of whom were serologically confirmed to have CSD, 9 (39%) of 23 nodes were positive by immunohistochemical analysis, including 3 from seronegative patients.16 Their control cases (7 lymph nodes from tuberculosis patients) were all negative by immunohistochemical analysis. Most of their specimens (14 cases) were positive for the Bartonella pap31 gene by PCR.16
In 7 of 9 of our study cases for which serologic data were available, the serologic testing for B henselae was positive. Among the serology-positive cases, 4 were positive by PCR, and only 1 case was positive by immunohistochemical analysis (the latter was also PCR positive) (Table 3). In 3 of the study cases, serologic testing for B henselae was positive and immunohistochemical analysis and PCR for CSD were negative. False-negative immunohistochemical and PCR results could be due to a previous infection, with elimination of the bacteria in the lymph nodes while antibodies against B henselae remained present, suggesting that the results of immunohistochemical analysis and PCR may depend on the duration of illness. It is possible that in patients with negative immunohistochemical and PCR results a positive serologic result could reflect a remote past infection.8,17–19 Our data underscore previous assertions that serologic assays should not be a mainstay for the diagnosis of CSD.
This study demonstrates that the diagnostic sensitivity of immunohistochemical analysis is low for the detection of B henselae and that PCR is more sensitive than immunohistochemical analysis. However, given the lower cost, shorter turnaround time, and greater availability of immunohistochemical analysis for CSD, we recommend performing immunohistochemical analysis for B henselae if CSD is suspected clinically, serologically, and/or histologically. We also recommend performing PCR on all cases negative by immunohistochemical analysis in which CSD is suspected owing to the higher sensitivity of PCR in comparison with immunohistochemical analysis and SSS. Because immunohistochemical and PCR testing can be performed on FFPE routine tissue samples, all testing can be performed from the same patient sample and there is no need for rebiopsy to obtain additional fresh tissue.
. Detection of Bartonella henselae DNA by two different PCR assays and determination of the genotypes of strains involved in histologically defined cat scratch disease. J Clin Microbiol. 1999;37:993–997.
. Seroprevalence of antibodies to Bartonella henselae in patients with cat scratch disease and in healthy controls: evaluation and comparison of two commercial serological tests. Clin Diagn Lab Immunol. 1998;5:486–490.
Gabriel C.Caponetti, LironPantanowitz, SharonMarconi, Jennifer M.Havens, Laura W.Lamps, Christopher N.OtisAm J Clin Pathol(2009)131 (2):
250-256DOI: http://dx.doi.org/10.1309/AJCPMNULMO9GPLYUFirst published online: 1 February 2009 (7 pages)