Case Definition

Patients referred to our laboratory from 1999 to 2004 were classified as definite cases of R. africae infection if this rickettsia was isolated from clinical specimens or a positive PCR detection was associated with a positive serologic test result. Of these, we selected those for whom inoculation eschar biopsy specimens had been formalin fixed and paraffin embedded for histopathologic analysis.

Serology, PCR, and Culture

Immunofluorescence assays, the reference diagnostic method, were carried out by using both R. conorii strain Seven (Malish, ATCC VR-613T) and R. africae strain ESF-5 as antigens (6). Titers of 64 for IgG and 32 for IgM in patient serum specimens were considered evidence of recent infection by a Rickettsia sp (12–14). Western blotting and cross-absorption procedures were performed (15–17). The microorganism was isolated in the manner previously reported (6). DNA was extracted from material frozen at –80°C. DNA was extracted from ground eschar biopsy specimens by using the QIAmp Tissue kit (Qiagen GmbH, Hilden, Germany), according to the manufacturer's recommendations. These DNA extracts were used as templates in a PCR assay incorporating the ompA primers 190-70 and 190-701 (18). For negative controls, we used sterile water processed as described and DNA extracted from a heart valve from a patient with degenerative valvulopathy. For positive controls, we used DNA from R. montanensis strain M/5-6. Testing was performed in a blinded manner. All positive PCR products were sequenced in both directions (19). When regular PCR did not detect rickettsiae in skin biopsy specimens, we used a nested PCR incorporating the ompA-amplifying primer sets AF1F–AF1R and then AF2F–AF2R for the nested amplification (1). Every 6 specimens, we incorporated the above-described negative controls. To avoid contamination, we did not include any positive control in the assay. Positive nested PCR products were identified by sequencing with the AF2F–AF2R primer pair.

Histologic Analysis and Immunohistochemical Detection of R. africae

Formalin-fixed, paraffin-embedded skin biopsy specimens of the inoculation eschars were cut to 3 μm in thickness and stained with hematoxylin-eosin-saffron by routine staining methods. Serial sections of each tissue specimen were also obtained for special staining or immunohistochemical investigations. Immunohistochemical analysis was performed on paraffin-embedded skin biopsy sections by use of the monoclonal mouse antibody AF8-F3 produced against R. africae (20). The specificity of this antibody has been tested on a variety of rickettsiae, and cross-reactivity was found only with R. parkeri and R. sibirica (20). The immunohistologic procedure, in which an immunoperoxidase kit was used, has been described elsewhere (21). Briefly, after deparaffinization, each tissue section was incubated with the monoclonal anti–R. africae antibody diluted 1:1000 in phosphate-buffered saline. After the sections were incubated with the primary antibody, immunodetection was performed with biotinylated immunoglobulins, followed by peroxidase-labeled streptavidin (HistoStain plus kit, Zymed, Montrouge, France) with amino-ethyl-carbazole as substrate. The slides were counterstained with Mayer hematoxylin for 10 min. For each case, 3 level tissue sections were systematically evaluated by immunohistochemical analysis, and a negative control was created by using an irrelevant monoclonal mouse antibody. Moreover, to test the specificity of our monoclonal antibody, skin biopsy specimens from patients with Mediterranean spotted fever (15 cases), acute eczematous dermatitis (2 cases), psoriasis (2 cases), and lichen planus (1 case) served as negative controls.

Quantitative Image and Statistical Analyses

To characterize the immune response in inoculation eschars during ATBF and Mediterranean spotted fever, paraffin sections were stained with the polymorphonuclear leukocyte marker CD15 (Immunotech, Marseille, France), the macrophage marker CD68 (Dako, Trappes, France), the T-lymphocyte marker CD3 (Dako), the B-lymphocyte marker CD20 (L26, Dako), and the endothelial cell marker Factor VIII–related antigen (Dako) by using the peroxidase-based method described above. The antibodies anti-CD15 and anti-CD3 were ready to use, whereas the antibodies anti-CD68, anti-CD20, and anti-Factor VIII–related antigen were used at a working dilution of 1:1000.

The evaluation of the proportion of polymorphonuclear leukocytes, macrophages, T lymphocytes, and B lymphocytes, as well the relative proportion of neovascularization in skin tissue specimens, was determined by quantitative image analysis (22). In brief, histologic images were digitized and transferred to a computer system. Using the image analyzer SAMBA 2005 (SAMBA Technologies), which is a specific interactive program that provides visual control of analysis, we analyzed the CD15-positive, CD68-positive, CD3-positive, CD20-positive, and Factor VIII–positive surfaces in tissue sections to determine the percentages of the total surface area covered by neutrophils, macrophages, T and B lymphocytes, and endothelial cells, respectively. For each set of observations, the surfaces of 10 randomly chosen areas were studied at a magnification ×100, and the surface areas of each immunohistologic parameter were measured. The average areas were calculated by comparison with the area of the whole tissue sample. The Mann-Whitney U test was used for the statistical comparisons of values that were obtained for each immunohistologic parameter for skin biopsy specimens from patients with ATBF and patients with Mediterranean spotted fever; p<0.05 was considered significant.

Value of Immunohistochemical Techniques

The sensitivity, specificity, and positive and negative predictive values of immunohistochemical techniques for the diagnosis of ATBF were calculated. The sensitivity of immunohistochemical analysis was compared to that of culture, regular PCR, nested PCR, and serologic testing by using the Fisher exact test.

Histologic and Immunohistochemical Analyses

Figure 1

Figure 1. Fibrinoid necrosis of a vessel in the dermis (arrow) with perivascular inflammatory infiltrates mainly composed of polymorphonuclear leukocytes (hematoxylin-eosin-saffron; original magnification ×250).

Figure 2

Figure 2. Coagulative necrosis (*) of the dermis surrounding necrotic vessels (arrow) (hematoxylin-eosin-saffron; original magnification ×250).

Figure 3

Figure 3. Inoculation eschar from a patient with African tick-bite fever showing numerous dermal inflammatory infiltrates mainly composed of polymorphonuclear leukocytes (immunoperoxidase staining with an anti-CD15 antibody; original magnification ×100).

Figure 4

Figure 4. Quantification of inflammatory changes in inoculation eschars from patients with African tick-bite fever (Rickettsia africae, n = 8) and patients with Mediterranean spotted fever (R. conorii, n = 15). Surface areas...

Eschars of ATBF cases were histologically dominated by vascular injury and perivasculitis, which were detected in all skin biopsy specimens examined. The host response to the vascular injury was manifested as intramural and perivascular infiltration by polymorphonuclear leukocytes, small lymphocytes, and macrophages (Figure 1). Vascular damages were endothelial swelling and vascular fibrinoid necrosis (Figure 2). Mural and occlusive fibrin thrombi were observed in a few blood vessels. Vascular injuries were associated with cutaneous necrosis (Figure 2). Moderate-to-severe cutaneous necrosis was present in 7 of 8 cutaneous biopsy specimens. When cutaneous necrosis was found, it was associated with vascular thrombosis. No hemorrhages were noted. Quantitative immunohistochemical analysis showed that the immune response in skin biopsy specimens from patients with ATBF more predominantly involved polymorphonuclear leukocytes than did specimens from patients with Mediterranean spotted fever (p<0.01, Figure 3 and Figure 4). In contrast, the inflammatory infiltrates in patients with Mediterranean spotted fever were mainly characterized by macrophages (p = 0.04, Figure 4) and T and B lymphocytes. The relative amounts of vessel formation were similar in the 2 diseases.

Figure 5

Figure 5. Immunohistochemical detection of Rickettsia africae in the inoculation eschar of a patient with African tick-bite fever. Note the location of the bacteria in the endothelial and inflammatory cells of a blood...

Six of 8 eschar biopsy specimens contained R. africae detected by immunohistochemical analysis (Table 2). Rickettsial antigen was observed in the endothelium and inflammatory cells organized in and around blood vessels (Figure 5). None of the control skin biopsy samples, including the R. conorii–infected cutaneous specimens, showed immunoreactivity with the monoclonal mouse antibody AF8-F3. Finally, no statistical correlations were found between the age of the cutaneous lesions and the amount and type of inflammatory cell infiltrate as well as the amount of rickettsial antigen found in the lesion.

The sensitivity, specificity, and positive and negative predictive values of immunohistochemistry were 75%, 100%, 100%, and 91%, respectively. The sensitivity of immunohistochemical tests was not significantly different from that of culture (6/8 vs 4/8, p = 0.6), regular PCR (6/8 vs 6/8, p = 1), nested PCR (6/8 vs 8/8, p = 0.5), and serology (6/8 vs 4/8, p = 0.6) (1,12,28).