Isolation of Legionella oakridgensis from Two Patients with Pleural Effusion Living in the Same Geographical Area

Abstract

Two cases of Legionnaire's disease caused by Legionella oakridgensis were diagnosed at the university hospital in Nantes, France. The two patients' isolates were identified by means of phenotyping and genotyping methods. Epidemiological investigations concluded that the first case was hospital acquired while the second case was considered community acquired.

CASE REPORTS

Case 1.

A 69-year-old man with a 1-year history of rheumatoid arthritis was admitted to Hotel Dieu hospital on 20 September 1993 with idiopathic thrombopenic purpura requiring splenectomy. He was discharged 9 days after surgery and returned home to a town located 50 km from Nantes. Fifteen days later he was readmitted to the hospital for fever, dyspnea, and chest pain. Bilateral ronchi and crackles were heard on auscultation of the lungs; the chest X-ray film showed left pleural effusion, and computed tomography revealed an encysted pleural pocket. Bronchial brushing performed on 26 October with culture on buffered charcoal yeast extract (BCYE) agar yielded a gram-negative Legionella-like bacterium. Cultures of pleural fluid were sterile. Serologic tests for Legionella pneumophila were negative. He recovered after receiving intravenous (i.v.) cefotaxim and oral ciprofloxacin for 6 days, followed by oral roxithromycin for 2 weeks.

Case 2.

A 79-year-old woman living in Nantes who had a 2-year history of rheumatoid arthritis and Parkinson's disease was admitted to Hotel Dieu hospital on 21 November 1993 for severe pneumonia of the left lung with chest pain. Chest radiography revealed bilateral lower lobe pneumonia with interstitial opacities and left pleural effusion. Bronchial brush culture performed on 22 November yielded gram-negative bacilli on BCYE agar after 9 days of incubation. The patient was first treated with oral ciprofloxacin and oral amoxicillin-clavulanate for 10 days and then with oral roxithromycin for 5 days. She recovered from the pneumonia and was discharged.

Microbiological investigations.

On primary culture, the two isolates grew on BCYE agar supplemented with 0.1% α-ketoglutarate (BCYEα) and BCYEα agar supplemented with glycine, vancomycin, and colistin, but not on blood agar or on BCYEα agar without l-cysteine. Each isolate yielded positive catalase and oxidase tests, negative glucose fermentation, and no urease or nitrate reduction. Neither isolate was autofluorescent under UV light (365 nm). These biochemical characteristics are shared by Legionella hackeliae serogroup 1, Legionella jordanis, Legionella londiniensis, Legionella moravica, Legionella sainthelensi serogroup 1, Legionella santicrucis, Legionella spiritensis, and Legionella oakridgensis (although L. oakridgensis reference strain OR-10/ATCC 33761 is described as oxidase negative [7]). The two isolates were tested by direct immunofluorescent assay (DFA) with unadsorbed antisera as previously described (1) and reacted strongly with antisera against L. oakridgensis, L. sainthelensi serogroup 1, and L. hackeliae serogroup 1.

Random amplified polymorphic DNA (RAPD) typing was performed on the two isolates by using primer SK2 (5′-CGGCGGCGGCGG-3′) (6), and infrequent-restriction-site PCR (IRS-PCR) was carried out using primers FGPS1490-72 (5′-TGCGGCTGGATCCCCTCCTT-3′) and FGPL132′-38 (5′-CCGGGTTTCCCCATTCGG-3′) (8). RAPD profiles were consistently reproducible in several independent determinations, and the two profiles obtained were more than 91% related to that of the L. oakridgensis type strain (Fig. 1) and to those of six L. oakridgensis isolates cultured from four French thermal spas and two hospitals (one in Lille, France, and one in Portugal), all of which had previously been characterized by RAPD (5) (not shown). These profiles were less than 80% related to those of all other Legionella species (Fig. 1). Similarly, the 16S-23S intergenic spacer region PCR profiles of the clinical isolates were identical to that of the L. oakridgensis type strain OR-10/ATCC 33761 (data not shown). The two clinical isolates were therefore identified as L. oakridgensis (deposited in the American Type Culture Collection as strains ATCC 700515 and ATCC 700516), and the diagnosis of L. oakridgensis pneumonia with pleural effusion was established in both patients.

FIG. 1.

RAPD profiles of L. oakridgensis strains and other Legionella reference strains. Lanes M, molecular weight marker Ready-load 100 bp (Gibco BRL); lane 1, L. oakridgensis ATCC 33761; lane 2, strain 930101868/ATCC 700515 (patient 1, Nantes); lane 3, strain 930101937/ATCC 700516 (patient 2, Nantes); lane 4, strain 88020285; lane 5, strain 88020400; lane 6, strain 88020413; lane 7, strain 88020417 (lanes 4 to 7 are environmental isolates from Nantes hospital); lane 8, L. oakridgensis strain 91020991 (thermal spa Bourbon-Lancy); lane 9, L. oakridgensis strain ATCC 33761; lane 10, L. hackeliae ATCC 35250; lane 11, L. jordanis ATCC 33263; lane 12, L. londiniensis ATCC 49505; lane 13, L. moravica ATCC 43877; lane 14, L. sainthelensi ATCC 35248; lane 15, L. santicrucis ATCC 35301; lane 16, L. spiritensis ATCC 35249.

As patient 1 was considered to have hospital-acquired legionellosis, the source was investigated by analyzing samples of the hospital's hot water distribution system. Cultures yielded Legionella isolates which were identified as L. oakridgensis by using the same molecular methods as above (phenotypically, the isolates were oxidase positive, like the patients' isolates) (Fig. 1, lanes 4 to 7).

The clinical and environmental isolates were compared by means of IRS-PCR with PstI and XbaI restriction enzymes. The IRS-PCR profiles of our clinical and environmental isolates were compared to those of the six L. oakridgensis control environmental isolates mentioned above. The reproducibility of the method was 100% when each isolate was tested three times. Nine different IRS-PCR patterns were obtained, each composed of three to nine bands ranging from 200 to 1,000 bp, confirming the discriminatory power of the method (Fig. 2). The two patients' isolates and two of the four Nantes hospital isolates each had a unique IRS-PCR profile (Fig. 2, lanes 1, 3, 6, 9), and the third and fourth Nantes hospital isolates each had a distinct profile (Fig. 2, lanes 2 and 4).

FIG. 2.

IRS-PCR profiles of L. oakridgensis reference strains and isolates. Lanes M, molecular weight marker VI (Boehringer Mannheim); lane 1, strain 89020417; lane 2, strain 88020285; lane 3, 89020400; lane 4, strain 89020413 (lanes 1 to 4 contain isolates from the water distribution system of the Nantes hospital); lane 5, strain 92203624 (from the water distribution system of a Lille hospital); lane 6, strain 930101868/ATCC 700515 (patient 1, Nantes); lane 7, L. oakridgensis strain OR-10/ATCC 33761; lane 8, strain 9020498 (thermal spa, Aix-les Bains); lane 9, strain 930101937/ATCC 700516 (patient 2, Nantes); lane 10, strain 90020551 (from the water distribution system of a Portuguese hospital).

L. oakridgensis was originally isolated from water in the environment (7, 13) and was subsequently found to be responsible for several human cases of pleurisy (4, 10, 11). However, these human infections were diagnosed by DFA on clinical specimens and were not confirmed by culture of the organism (4, 10, 11). Our two cases of L. oakridgensis legionellosis were diagnosed on the basis of strain isolation. Isolates were recovered from bronchial brush specimens of patients with pleural effusion and/or pneumonia. Both patients had a previous history of immunological disorders treated with steroids (idiopathic thrombocytopenic purpura and rheumatoid arthritis, respectively), possibly rendering them more susceptible to infection. Pleural effusion was also present in the three previously reported cases of human L. oakridgensis infection (4, 10, 11), suggesting that this species might produce specific virulence factors (adhesins and/or extracellular products) favoring tropism for pleural tissue.

Biochemical tests poorly distinguish L. oakridgensis from other Legionella species (2, 12), and cross-reactions occur with L. sainthelensi in DFA (3). Definite L. oakridgensis identification is more easily achieved with molecular techniques, such as RAPD and ISR-PCR (5, 8), and we have successfully applied these methods in the present report. If pulsed-field gel electrophoresis and IRS-PCR have proven useful for epidemiological comparison of L. pneumophila isolates (9), we show here that IRS-PCR can be applied to L. oakridgensis for microbiological investigation of clinical cases. Case 1 was considered hospital acquired, with probable infection directly from the hospital water: L. oakridgensis isolates from the patient and from hospital sources had the same IRS-PCR profiles. Case 2 was considered community acquired, as the patient had not been in contact with the hospital for several months before onset, even though the isolate had an IRS-PCR profile identical to that of the first patient's isolate. As patient 2 lived in Nantes, it is conceivable that the town water distribution system contained the same L. oakridgensis strain, but cultures were not done at the time. Hence, patients from the same geographical area can be infected by an unusual Legionella strain that has only rarely been associated with human infections.