Abstract
In addition to Legionella pneumophila, about 20 Legionella species have been documented as human pathogens. The majority of infections by non-pneumophila Legionella species occur in immunocompromised and splenectomized patients. Here, we report a case of ‘classical’ lobar pneumonia caused by Legionella longbeachae in a splenectomized patient receiving corticosteroids for chronic immune thrombocytopenia. Tests for Legionella antigen were negative. L. longbeachae was immediately detected in bronchoalveolar fluid by PCR and subsequently confirmed by culture on legionella-selective media. The features of Legionnaires' disease in immunocompromised patients with special emphasis on significance and detection of non-pneumophila species are reviewed.
Introduction
Legionella is consistently reported among the most commonly identified pathogens in community- and hospital-acquired pneumonia (Stout & Yu, 1997). About 20 different pathogenic species have been reported, with Legionella pneumophila accounting for more than 80 % of human infections. The most commonly isolated non-pneumophila Legionella spp. are Legionella longbeachae (3.2 %), Legionella bozemanae (2.4 %), Legionella micdadei, Legionella dumoffii and Legionella feeleii (2.2 % combined) (Yu et al., 2002). L. longbeachae is far more common in Australia (30.4 %), where infection has been associated with exposure to potting mixes (Yu et al., 2002; O'Connor et al., 2007).
Legionnaires' disease can be acquired by the inhalation of contaminated aerosols or by microaspiration of contaminated water (Stout & Yu, 1997). Domestic aquatic reservoirs have been described as a source both in nosocomial and community-acquired infections (Pedro-Botet et al., 2002).
The clinical features of pneumonia caused by Legionella spp. are diverse. Early symptoms include fever, malaise, myalgia, anorexia and headache. Cough is only slightly productive. During the course of illness, fever exceeding 40 °C, stupor, and respiratory and even multiorgan failure may develop. Elevation of creatinine kinase and diarrhoea have been reported as disease-specific symptoms for legionellosis in a comparative study (Sopena et al., 1998).
The incidence of community-acquired pneumonia caused by Legionella is probably underestimated as many species and serogroups are not properly detected by commercially available tests (Roig & Rello, 2003). This issue is of importance as patients with community-acquired legionellosis are more likely to require admission to the hospital and the intensive care unit compared to patients with pneumonia of other causes (Sopena et al., 1998).
Legionnaires' disease has been historically referred to as ‘atypical’ pneumonia, based on its clinical presentation and the notion that chest radiographic findings are neither lobar nor consolidating, as seen in the ‘classical’ pyogenic pneumonia (Tan et al., 2000). We report the case of a patient with unilateral lobar cavitating pneumonia in which L. longbeachae was the only aetiological agent. In addition, we summarize the special features of Legionnaires' disease in immunocompromised patients.
Case report
A 69-year-old man presented to the emergency room in October 2006 with complaints of dyspnoea on effort, somnolence, a non-productive cough, fever and night sweats. History was significant for chronic idiopathic thrombocytopenic purpura for which he underwent splenectomy in 2002 and relapsed thereafter. Three weeks prior to admission, he experienced fatigue and a new episode of severe thrombocytopenia, for which he had received dexamethasone at a dose of 40 mg per day for 3 days.
On admission, his temperature was 39.4 °C, heart rate was 117 beats min−1, blood pressure was 115/72 mmHg and respiratory rate was 26 min−1. Physical examination revealed fine crackles and marked dullness of the right middle lobe. Petechiae and multiple ecchymoses were present on the lower extremities, trunk and oral mucosa.
Laboratory testing was remarkable for leukocytosis (13.4 nl−1), thrombocytopenia (3 nl−1) and elevated C-reactive protein (CRP, 392 mg l−1, normal <5 mg l−1). Hepatic, pancreatic and renal parameters as well as urinalysis were normal. Capillary blood gas analysis confirmed moderate hypoxia and respiratory alkalosis due to hyperventilation (pO2 66 mmHg, pCO2 26 mmHg, pH 7.51, HCO3 21.1 mmol l−1).
Chest radiograph and computed tomography revealed a distinct confluent infiltration of the right middle lobe including two small cavitations. Pleural effusions were absent. Bronchial obstruction or external compression could be excluded (Fig. 1⇓).
(a, b) Chest radiograph on admission showing airspace consolidation of the right middle lobe. (c) Chest-CT additionally revealed two small cavitations (∼10×10 mm, black arrows) and excluded bronchial obstruction or external compression. Pleural effusions were absent.
Microbiology
Gram and acid-fast stains of bronchoalveolar lavage (BAL) fluid were negative. A Legionella-genus specific PCR detected L. longbeachae DNA in the BAL fluid within 18 h [nucleic acids extracted from an equivalent of 25 μl BAL fluid (QIAamp DNA Mini kit; Qiagen) were used in a PCR (40 cycles with primers U24 (5′-CGC CTT CGC CAC TGG TGT TGT T-3′) and L65 (5′-AAC GCG TAG GAA TAT GCC TT AGA-3′))]. Identification was enabled by sequence determination of the amplified 16S rDNA fragment (capillary electrophoresis GeneticAnalyser; Applied Biosystems).
Colonies typical for Legionella grew on a Legionella medium (Legionella-BCYE; Becton Dickinson) after several days of incubation, and were identified as L. longbeachae by partial sequencing of the 16S rRNA gene (Bottger, 1989). Legionella urine antigen enzyme immunoassay (Biotest) tested negative. Cultures for other bacteria (including mycobacteria), fungi and respiratory viruses were negative. All blood cultures remained negative. On the basis of these results, we made a definitive diagnosis of L. longbeachae pneumonia.
Management and clinical course
During initial antibiotic therapy with intravenous piperacillin/tazobactam for 24 h, the clinical symptoms worsened markedly. Identification of L. longbeachae in BAL fluid prompted a change of antibiotic therapy to intravenous moxifloxacin. The patient recovered quickly and was switched to oral moxifloxacin for a total of 12 days. Thrombocytopenia markedly improved as the patient recovered.
Discussion
In the case reported here, a clinical presentation with high fever, markedly elevated CRP and the radiographic presentation of cavitating pneumonia in a splenectomized patient was suggestive of classical pneumonia caused by Streptococcus pneumoniae. In contrast, Legionnaires' disease has been historically referred to as ‘atypical’ pneumonia based on its clinical presentation and the long-held belief that chest radiographic findings are neither lobar nor consolidating (Tan et al., 2000). Three-quarters of patients with L. pneumophila pneumonia present with an abnormal chest X-ray. Patchy pneumonic infiltrates, predominantly of the lower lobes, but also circumscribed or even lobar consolidations are observed. Cavitations are rather uncommon (Tan et al., 2000). No study has addressed the radiographic manifestations of legionellosis in the immunocompromised host, but cavitations are probably more common than in immunocompetent patients (Di Stefano et al., 2007; Muder et al., 1987; Tan et al., 2000; Fraser et al., 2004; Dowling et al., 1983; Senecal et al., 1987). In conclusion, Legionella spp., like other causes of ‘atypical’ pneumonia, cannot be confirmed or excluded by distinct radiographic patterns.
Clinical practice guidelines recommend testing for Legionella in selected patients, including seriously ill patients without an alternative diagnosis, older and immunocompromised patients, as well as patients nonresponsive to beta-lactam antibiotics (British Thoracic Society Standards of Care Committee, 2001; Mandell et al., 2007). Legionella spp. do not grow on standard microbiology media, and are usually not detected by blood culture or Gram stain or culture of sputum. Culture isolation from BAL fluid is considered the gold standard of diagnosis, but may not always be helpful for the clinician because it takes 4–10 days. Moreover, currently available selective media are probably suboptimal for the isolation of non-pneumophila Legionella spp. (Muder & Yu, 2002). The availability of tests for Legionella antigen in the urine resulted in a decreasing use of cultures and serological studies (Benin et al., 2002). The major drawback is that urinary antigen tests are virtually limited to L. pneumophila (Roig & Rello, 2003). Due to the shortcomings of culture and urinary antigen tests, PCR appears to be a promising tool for the simultaneous, rapid and reliable detection of many different Legionella species. Most of the commercially available assays target species-specific regions within the 16S and 5S rRNA genes and in the macrophage inhibitor potentiator (mip) gene. PCR techniques using ribosomal genes as target have the potential to provide a rapid diagnosis of several Legionella spp. with the use of readily obtainable respiratory tract specimens (Bencini et al., 2007; Diederen et al., 2006).
Numerous case reports suggest a higher rate of non-pneumophila Legionella spp. among immunocompromised and splenectomized patients than in immunocompetent hosts (Jaeger et al., 1988; Jernigan et al., 1994; Korman et al., 1998; McClelland et al., 2004; Schwebke et al., 1990; Wilkinson et al., 1987; Fang et al., 1990; Garcia et al., 2004; Lang et al., 1990; Radaelli et al., 1991; Singh et al., 2002). L. longbeachae was first described as a cause of pneumonia in two Californian patients (McKinney et al., 1981). Since then, only nine cases have been reported outside Australia. Interestingly, three of those patients were splenectomized (Gorelik et al., 2004; Lang et al., 1990), and three others received immunosuppressive drugs because of systemic lupus erythematodes (Garcia et al., 2004; McClelland et al., 2004) or heart transplantation (Korman et al., 1998).
The primary host defence mechanism against Legionella is cell-mediated immunity, similar to other intracellular pathogens. Depression of cell-mediated immunity by glucocorticoids and immunosuppressive drugs may predispose the host to Legionnaires' disease (Schlossberg & Bonoan, 1998). Transplant recipients carry the highest risk (Singh et al., 2002). Corticosteroids and disorders associated with immunosuppression, e.g. cancer, are independent risk factors (Poupard et al., 2007). Hairy cell leukaemia also increases the risk of Legionella infection because of monocyte deficiency and dysfunction (Fang et al., 1990; Radaelli et al., 1991). In contrast, the incidence of Legionnaires' disease is not higher in neutropenia, acute leukaemia and HIV infection (Schlossberg & Bonoan, 1998).
Loss of the spleen due to splenectomy, spleen irradiation, infarction or sickle cell disease increases the risk of infection with S. pneumoniae, Streptococcus meningitidis and Haemophilus influenzae. An increased risk of legionellosis has not been recognized. However, there is a striking coincidence of splenectomy and infections with non-pneumophila Legionella spp., as reported here and in previous reports by Gorelik et al. (2004) and Lang et al. (1990). The spleen, by delivery of antigen, regulation of lymphocyte traffic, opsonization, phagocytosis and as a site of lymphocyte priming, plays an important role during any bacterial infection. Accumulating reports of infection with non-pneumophila spp. in splenectomized patients suggest that spleen-related factors might be especially important in the defence against these species.
Different Legionella species must be considered as causal agents of pneumonia in immunocompromised hosts, even if radiographic appearance suggests pneumococcal pneumonia. Antigen assays and serology can fail to detect non-pneumophila Legionella species. PCR testing of BAL fluid should be considered in high-risk patients.