Atypical Cat-Scratch Disease in Children: Report of Seven Presentations Ranging From Hepatosplenic Disease to Horner Syndrome

1. Introduction

Cat-scratch disease (CSD) is an infectious disease caused by the Gram-negative rod Bartonella henselae (BH). It is usually benign and self-limited. It affects children or young adults in contact with infected cats. Cats, especially under one year old, get infected by cat fleas (Ctenocephalides felis). Humans are contaminated by cat saliva or by scratch. However, the absence of cat in the direct environment of patient and the lack of history of scratch does not exclude the diagnosis of CSD. Transmission by dogs has also been rarely reported (1).

Typically, CSD leads to subacute regional lymphadenitis, possibly associated with mild fever. Symptoms occur between 14 days to several weeks after the inoculation. Scratch lesion is rarely found. Lymphadenitis tends to resolve spontaneously within 4 - 6 weeks, although surgical drainage of necrotic collection is indicated in 12 - 16% of the cases (2). Less specific symptoms such as prolonged fever, hepatosplenic lesions, osteomyelitis, radiculopathy or neuroretinitis have also been associated with CSD (1). In this paper, we reported a series of seven children with atypical presentations of CSD. We reviewed and discussed the incidence of these unusual manifestations, the therapeutic approaches recommended, and the biological tools available to confirm the diagnosis of CSD.

2. Case Presentation

Seven children with a broad spectrum of atypical symptoms were diagnosed with CSD in several pediatric centers in Belgium. The clinical histories of these patients are summarized in Table 1. Four children mentioned the presence of cats at home, but only one patient reported a history of cat scratch. Five children (patients 1 to 5) presented abdominal pain associated with systemic symptoms such as persistent fever and loss of weight. In these patients, the radiological evaluation revealed liver and/or spleen nodular infiltrations. Bone lesions (in thoracic or lumbar vertebrae) were detected in three patients (patients 3 to 5) with back pain. A systemic infection without localized lesion was identified in patient 6 who presented prolonged fever and diffuse muscle pain. Patient 7 presented preauricular and cervical lymphadenopathy associated with ptosis, miosis and enophtalmy, suggestive of Horner syndrome.

3. Discussion

These series illustrated that CSD induces symptoms that surpass the classical subacute lymphadenitis associated or not with slight fever. Those symptoms can occur independently on the immune status of the host. The frequencies of these atypical presentations of CSD are listed in Table 2. In immunodeficient patients, CSD is rather associated with specific manifestations such as bacillary angiomatosis and peliosis hepatis (1). As found in our series, prolonged fever and hepatosplenic CSD are the most common signs of disseminated BH infection. In pediatrics, CSD is considered as the third cause of fever of unknown origin (FUO) after infection by Epstein Barr virus and osteomyelitis, and before urinary tract infection (1, 3). Fever can last for several weeks and may be associated with headaches and diffuse muscle pain. Hepatosplenic lesions, often revealed by abdominal pain, have been described in 10 - 14% of CSD patients (1). They consist of parenchymal micro-abscesses resulting from a hematogenous spread of the pathogen (1, 2). Hepato- and/or splenomegaly have been clinically detected in some, but not all cases. The dissemination of BH may also result in osteomyelitis (4). The most frequent bones involved are vertebrae, pelvis and ribs. Patients present with localized pain possibly along with inflammatory signs in the adjacent soft tissues, although some bone lesions are asymptomatic and are only detected by scintigraphy and/or radiology (4, 5). Magnetic resonance imaging (MRI) allows early detection prior to tomodensitometry or X-ray (5).

The neurological manifestations of CSD are very rare and reported in only 1 - 2 % of CSD cases. These symptoms include encephalopathy (90% of neurological manifestations), neuroretinitis (acute visual loss due to optic nerve edema and macular exudate), radiculitis and transverse myelitis (7). The Parinaud oculoglandular syndrome, characterized by unilateral conjunctivitis with conjunctival granuloma, has also been associated with preauricular and/or submaxillary CSD lymphadenopathy. It is attributed to the inoculation of BH in the conjunctiva (1). Patient 6 presented with back and inferior limbs pain which may correspond to radiculitis. While bone lesions were excluded by radiological workup, no further examination was performed to investigate the presence of neurological lesion in this patient. Patient 7 presented with miosis, ptosis and enophtalmy, suggesting Horner syndrome. It was the first time that this syndrome was associated with CSD. It is usually due to a lesion along the sympathetic pathway in the area of neck, head or eye. Our hypothesis was that enlarged lymphadenitis probably compressed the sympathetic nerve and induced Horner syndrome. This syndrome completely disappeared concomitant with lymphadenitis regression.

Since BH is difficult to culture from human specimens, the diagnosis of CSD mainly resides on serological investigations. Two techniques are available to detect antibodies against BH. The immunofluorescent antibody assay (IFA) is considered more reliable and more sensitive than the enzyme-linked immunosorbent assay (EIA) (6, 8). The sensitivity and specificity of immunoglobulin (Ig) G detection by IFA have been reported respectively around 88% and 97%. IgM detection is slightly less sensitive and specific (respectively around 70 - 90% and 87.5 - 100%) (6, 8). The timing is also important for the interpretation of these analyses. IgM and IgG occur within 1 - 8 weeks after the onset of the disease. IgM lasts for three months while IgG remains increased up to two years after infection. An increase of both IgM and IgG confirms the diagnosis of CSD (6). However, an isolated elevation of IgM or IgG, in particular in the context of atypical presentation, needs to be interpreted with caution. A second analysis repeated after 3 - 6 weeks may be useful in the interpretation. A delayed elevation of IgM level on the second sample concomitant with IgG increase is considered as a proof of CSD. Likewise, an amplified increase of IgG level on the second sample (above 1/512 or at least four times the previous level) confirms the diagnosis of CSD (6). In our series, all the patients but two had an increase in both IgM and IgG levels. The high level of IgG (> 1/1280) in patients 4 and 6 and the presence of lymphadenopathy possibly with a cat scratch on the finger in patient 4 led us to establish the diagnosis of CSD. Polymerase chain reaction (PCR) is also useful in the detection of BH (8). This analysis can be performed on any possibly infected tissue (i.e. lymph, liver, bone, etc.). PCR allows early detection, prior to seroconversion, with a high specificity and a better sensitivity than serological investigations. When combined, these analyses reach a sensibility of more than 92% (2). The major limitation of PCR is the requirement of invasive tissue sampling. On blood, PCR has a lower sensitivity and its use in clinics remains controversial (1, 7). None of our patients was diagnosed using PCR.

As for typical CSD lymphadenitis, the role of antibiotics in the management of uncomplicated fever and/or hepatosplenic lesions is controversial and most authors only recommend symptomatic treatment (1, 2). While drugs such as azithromycin, doxycycline, rifampicin, gentamicin, ciprofloxacin or trimethoprim-sulfamethoxazole have a potential interest, only azithromycin (with a dose of 10 mg/kg on day one, followed by 5 mg/kg from days 2 - 5) has demonstrated its efficacy on lymphadenitis (9). There has been no randomized trial on atypical CSD. In CSD osteomyelitis, prolonged antibiotherapy up to six weeks is recommended, but the drug and the duration of treatment remain debated (5). Azithromycin has good bone diffusion and no significant toxicity in children (5). Trimethoprim–sulfamethoxazole, ciprofloxacin or rifampicin are also characterized by good bone diffusion and could therefore be used against BH. No pediatric recommendation has been offered for neurological manifestation of CSD. In adults, a combination of doxycycline and rifampicin has been proposed (10). In the absence of response, the use of corticosteroids has also been suggested while there is no published data in this regard. We acknowledge that the therapeutic approach in our patients was quite variable. This was due to the fact that patients were diagnosed and treated in several centers. All of them were treated with antibiotics despite the absence of consensus in the approach of uncomplicated forms of CSD. Their clinical outcomes were favorable, although it was not necessarily correlated to the initiation of antibiotherapy.

This report underlines that CSD should be evocated not only in front of loco-regional lymphadenitis, but also in a broad spectrum of presentations in children. In this context, history and clinical examinations looking for cat scratches are often not contributive. In addition, serological workup represents a widely available tool for the diagnosis of CSD. Therefore, we suggest including this testing in the routine evaluation of children with prolonged fever of unknown origin, or unexplained liver, spleen or bone lesions. We also highlight that, when unclear, a second delayed serological assay and/or a PCR on the possibly infected tissue may be useful to precise the diagnosis of CSD. Further investigations are required to better understand the natural progression of BH infection in human, identify the factors leading to self-limited or systemic dissemination, and define the most efficient therapeutic approach of this disease in children.