Clinical utility of a Legionella pneumophila urinary antigen test in a large university teaching hospital

Abstract

OBJECTIVE:

To determine the clinical utility of diagnosing Legionella pneumonia by urinary antigen testing (LPUAT) in a low prevalence centre.

DESIGN:

The results of LPUATs were abstracted and analyzed from the authors' laboratory information system. Medical records were reviewed in detail for all positive tests and a random sample of 50 negative tests.

SETTING:

The Queen Elizabeth II Health Sciences Centre, a large university hospital complex.

POPULATION STUDIED:

Patients who were admitted from the emergency room with pneumonia or who had developed pneumonia in hospital and who had a LPUAT performed between April 1998 and October 2000.

MAIN RESULTS:

One thousand one hundred fifty-four tests were performed on 1007 patients. Seven patients had nine positive LPUATs. Three of these patients had confirmed Legionella pneumophila pneumonia. Three others had probable or possible L pneumophila pneumonia. There was one probable false positive. Six of the seven patients were already on empirical anti-L pneumophila therapy. Of the 50 negative tests reviewed in detail, 31 patients were on one of the antibiotics of choice for L pneumophila at the time the test was ordered; in 21 (68%) of these patients the negative result did not lead to a change in therapy.

CONCLUSIONS:

The cost to diagnose each case of Legionella pneumonia by LPUAT was approximately $5,770 and most patients were already on appropriate antibiotics. In patients with negative tests, antibiotics were often not changed in response to the test result. Rigorous screening of patients is required to increase pretest probability for LPUAT to be justified.

Legionella pneumophila is responsible for 1% to 4% of cases of community-acquired pneumonia requiring hospitalization (1). Although the clinical presentation and radiographic appearance may be suggestive of Legionella pneumonia, clinicians are not able to differentiate it from other bacterial causes of pneumonia on clinical grounds (1). In theory, early diagnosis of a suspected Legionella infection would permit appropriate antibiotic therapy in a timely fashion. In practice, guidelines for the treatment of community-acquired pneumonia recommend the empirical use of macrolides and fluoroquinolones, which are the drugs of choice for L pneumophila infections (2-4).

Several methods exist for diagnosing Legionella infection. Sputum culture is favoured by many because it has the ability to detect all species and subgroups of Legionella and has a specificity of 100% (1). High specificity is especially advantageous for low prevalence diseases. However, the sensitivity is approximately 70% (1), and positive results typically take three to five days. Serological diagnosis is limited due to the time delay for seroconversion (5). Direct fluorescent antibody (DFA) testing has the ability to provide results in a time frame able to influence clinical management and has a specificity of close to 100% (5). However, DFA is technically demanding and insensitive. As with sputum culture, DFA has limited usefulness when patients cannot produce sputum.

L pneumophila by urinary antigen testing (LPUAT) is a rapid tool for early diagnosis of Legionella infection (6-14). An enzyme immunoassay (EIA) for detecting L pneumophila serogroup 1, which accounts for between 50% and 70% of cases of Legionella pneumonia (15,16), has been shown to be both sensitive and specific. In addition, some studies have shown that the test is capable of cross-reacting with several L pneumophila serogroups and even other Legionella species (6,9). Specificity is typically more than 99% (7-9).

Since the authors had been using a urinary antigen assay since 1998, they wished to determine how many additional cases of L pneumophila pneumonia had been diagnosed and whether negative test results influenced a change in the antimicrobial agent used or the duration of the treatment.

METHODS

The results of the Binax Legionella Urinary Antigen EIA tests (Binax Incorporated, USA), L pneumophila cultures and antibody studies, which were performed at the Queen Elizabeth II Health Sciences Centre between April 1998 and October 2000, were abstracted using the authors' laboratory information system. To determine the extent to which patient care was influenced by the results of the EIAs and whether negative test results led to appropriate changes in management, the charts of all patients who tested positive were reviewed, as well as a randomly chosen sample of 50 cases in which the patients tested negative. The selection was done by choosing the cases of every 21st patient on an alphabetically ordered list of those who had the test during the study period.

The Binax Legionella Urinary Antigen EIA 96 well test (Binax Incorporated, USA) was performed in accordance with the manufacturer's instructions. In brief, patient urine was added to microtitre wells coated with polyclonal rabbit antibody specific for L pneumophila serogroup 1 antigen. Captured antigen was detected with anti-Legionella horseradish peroxidase labelled conjugate. Unbound antigen and conjugate were decanted and the wells were washed. A colour developer was added and absorbance was read on a microplate reader. Samples with absorbencies greater than three times the negative controls were considered positive.

Respiratory secretions submitted for Legionella species cultures were planted onto buffer charcoal yeast extract agar and buffer charcoal yeast extract agar with vancomycin and polymixin B. Plates were incubated at 37°C in carbon dioxide and examined daily for seven days. Organisms with colony morphology, Gram stain characteristics and that did not grow on sheep blood agar were further identified by DFA staining using L pneumophila group 1 to 6 antisera.

The Legionella indirect fluorescence antibody testing was performed by the Ministry of Health laboratory in Toronto, Ontario. A fourfold rise or a rise in titre to greater than 1:256 from the acute specimen in an acute-convalescent pair was considered evidence of recent infection.

Definitions

Risk factors for L pneumophila pneumonia:

Transplant recipient; taking systemic steroids or cytotoxic chemotherapy; diagnosed with chronic obstructive pulmonary disease (COPD), end stage renal disease, cancer, diabetes mellitus or AIDS; and active cigarette smoker.

Proven case of L pneumophila pneumonia:

Clinical picture compatible with pneumonia including appropriate radiographic changes and a positive culture for L pneumophila or a fourfold rise in convalescent anti-L pneumophila antibody.

Probable case of L pneumophila pneumonia:

Clinical picture compatible with pneumonia including appropriate radiographic changes, presence of risk factors other than smoking for L pneumophila pneumonia, and no laboratory evidence of other infectious etiologies.

Possible case of L pneumophila pneumonia:

Clinical picture compatible with pneumonia including appropriate radiographic changes and no laboratory evidence of other infectious etiologies.

RESULTS

One thousand one hundred fifty-four LPUATs were performed on 1007 patients during the study period. Seven patients had nine positive tests. Their breakdown into proven, probable or possible cases of L pneumophila pneumonia and whether they were already on anti-Legionella therapy at the time of the test result is shown in Figure 1. Of the seven patients, six were already on anti-Legionella therapy before the test result was available.

Figure 1.

Patients with positive urinary antigen tests

Information relating to the LPUAT's influence on therapy in the 50 cases with negative test results is shown in Figure 2. Of the 31 cases in which the patients were taking anti-Legionella medication, 21 (68%) had no change in therapy with the negative test result. It is of note that 28% of those with negative LPUATs had no evidence of pneumonia on chest radiograph when subsequently read by a radiologist. Thirty-six per cent of patients with negative LPUATs had no recognized risk factors for L pneumophila pneumonia. Data on patient demographics, presence of recognized risk factors for L pneumophila pneumonia and final chest radiographic reports is listed in Table 1.

Figure 2.

Patients with negative urinary antigen tests. CXR Chest x-ray

Table 1.

Patient demographics, presence or absence of recognized risk factors and final chest radiographic reports

	Positive LPUATs	Negative LPUATs
Number	7	50*
Median age (range)	64 (40-77)	70 (33-90)
Men 5/7 (71%)	17/47	(36%)
Suspected cases of community-acquired pneumonia†	5/7 (71%)	38/50 (76%)
Presence of risk factors for Legionella pneumonia	4/7 (57%)	32/50 (64%)
CXR suggestive of pneumonia	6/7 (86%)	36/50 (72%)

^*Forty-seven patients, 50 pneumonia episodes

^†The remaining cases were hospital acquired. LPUAT Legionella pneumophila urinary antigen test

Details of the cases with positive LPUATs

Proven positive cases

Case 1:

A 53-year-old male smoker with COPD presented with symptoms and signs of community-acquired pneumonia. Bloodwork showed a slightly increased white blood cell (WBC) count and a sodium level of 125 mmol/L. A chest radiograph revealed bronchopneumonia. He was started on cefuroxime and erythromycin therapy. An LPUAT and Legionella sputum culture were performed on the second day. The culture grew L pneumophila on day 4 and the LPUAT was positive. Cefuroxime was discontinued on day 5. A repeat LPUAT was also positive. Serology was not performed.

Case 2:

A 40-year-old male smoker presented with symptoms and signs of severe community-acquired pneumonia. Bloodwork showed a normal WBC count but a left shift and a sodium level of 130 mmol/L. A chest radiograph revealed multilobar pneumonia. He was initially treated with levofloxacin, which was later changed to ciprofloxacin to cover a second organism. An LPUAT ordered on day 1 was positive. L pneumophila sputum culture and a repeat LPUAT ordered on days 7 and 8, respectively, were both negative, but acute and convalescent antibody titres revealed a fourfold rise.

Case 3:

A 71-year-old male smoker with COPD and ischemic heart disease developed right-sided airspace disease evident on a chest radiograph the day after an emergent coronary artery bypass graft. He was started on ceftriaxone therapy for nosocomial pneumonia. A chest radiograph on postoperative day 3 showed progression of the pneumonia and the patient required reintubation. A positive LPUAT on the sixth postoperative day resulted in a discontinuation of the ceftriaxone. Treatment with erythromycin, ciprofloxacin and rifampin was started. Sputum samples collected on the sixth and seventh post-operative day to confirm the LPUAT result grew L pneumophila. Serology was not performed.

Probable positive cases

Case 4:

A 64-year-old male heart transplant recipient with COPD presented with symptoms and signs of pneumonia. He had been discharged from hospital three days before, after a two-day admission for flu-like symptoms. Bloodwork showed an increased WBC count (30x10⁹/L) and a chest radiograph revealed multifocal airspace disease. Treatment with levofloxacin was started. An LPUAT was positive, but three sputum samples for Legionella culture collected on days 4 and 5 were negative. Serology was not performed. Sputum culture for respiratory viruses was negative. He improved and was discharged after 17 days. Four weeks later he was readmitted with right upper lobe pneumonia and was started on levofloxacin. An LPUAT was positive, but Streptococcus pneumoniae susceptible to levofloxacin was isolated from a blood culture. The levofloxacin was continued. The patient improved and was discharged after five days of hospitalization.

Possible positive cases

Case 5:

49-year-old male cigar smoker with no risk factors for Legionella pneumonia was started on oral cefuroxime therapy for a right lower lobe community-acquired pneumonia two days before admission. He was admitted with worsening symptoms. Bloodwork showed an elevated WBC count (16x10⁹/L) and a sodium level of 130 mmol/L. A chest radiograph was not repeated. He was started on intravenous erythromycin and cefuroxime. An LPUAT ordered on admission was positive. The cefuroxime was discontinued and the dose of erythromycin was increased. He recovered quickly and was discharged after four days. No other tests for Legionella were ordered.

Case 6:

A 67-year-old female smoker with no other risk factors for Legionella pneumonia developed a left upper lobe segmental nosocomial pneumonia during a hospital stay for psychiatric illness. She was started on levofloxacin, which was then changed to azithromycin on the following day. An LPUAT was positive. She remained on azithromycin and recovered. No other tests for Legionella were ordered.

Probable false positive case

Case 7:

A 77-year-old woman with type 2 diabetes mellitus was admitted to a community hospital and treated for congestive heart failure and a possible community-acquired pneumonia with cefuroxime and azithromycin. She developed acute renal failure and was transferred to the Queen Elizabeth II Health Sciences Center on day 5. A chest radiograph performed on arrival revealed congestive cardiac failure, but no evidence of pneumonia. The patient was afebrile and her serum WBC count was normal. Antibiotics were discontinued. An LPUAT was ordered on day 7. The LPUAT was positive but it was felt that this was a false positive. The patient improved without antibiotic treatment. A repeat LPUAT performed three weeks later was negative. Serology was performed 24 days after her admission to the community hospital and was also negative. Cultures for Legionella were not performed.

DISCUSSION

Although certain risk factors, clinical features and laboratory abnormalities may suggest a diagnosis of L pneumophila pneumonia, clinical differentiation from common bacterial pneumonias is usually not possible (17,18). The diagnosis of L pneumophila pneumonia is often not considered and, in one large study, it was suggested that only 3% of sporadic cases of Legionnaires' disease are correctly diagnosed (19). In some centres, L pneumophila is second only to S pneumoniae as a cause of community-acquired pneumonia requiring admission to a critical care unit (17,20). A prompt diagnosis by the detection of L pneumophila antigen in urine would, therefore, seem highly desirable in view of its sensitivity and specificity by immunoassay. Legionella antigen testing can be rapidly performed on urine, is usually detectable at the time of presentation and is highly specific. On the other hand, these same characteristics may lead to its overuse in patients with a low pretest probability of L pneumophila infection. It could be argued that, because current guidelines for the empirical treatment of community-acquired pneumonia suggest the use of macrolides or respiratory fluoroquinolones, a specific diagnosis is less important than it once was. Although macrolides and fluoroquinolones are likely to be continued when no diagnosis is forthcoming, the dose and duration may be altered or a second agent added if Legionnaires' disease is diagnosed.

The cost to diagnose each case of Legionnaires' disease using the LPUAT was approximately $5,770 (materials and supplies only). Six of the seven patients with a positive LPUAT were already on, or, in one case, had already been, treated with antibiotics active against L pneumophila when the tests were ordered. The patient who was not taking an appropriate antibiotic did have a change in antibiotic because of the test result. For both cases in which the patient had been started on dual therapy (cephalosporin and macrolide), the cephalosporin was discontinued when the diagnosis of L pneumophila pneumonia was made, but in one of these cases, a positive Legionella culture was available. It is of note that 28% of patients with negative LPUATs did not have evidence of pneumonia on chest radiograph when the test was ordered. In these cases the diagnosis of pneumonia was usually made by housestaff before the chest radiograph was seen or before their interpretation of the chest radiograph was confirmed by the radiologist. In 62% (31 of 50) of the cases, the patient was started on empirical therapy for L pneumophila and in 68% (21 of 31) of these cases the treatment was not influenced by the negative LPUAT. During the period studied 448 specimens from 343 patients were submitted for Legionella culture. Only three specimens, from two patients, were positive. Both of these patients had positive LPUATs. No other Legionella species were recovered during the study.

In summary, the authors feel that the LPUAT may have some clinical utility when it is used to confirm a diagnosis of L pneumophila pneumonia. However, the pretest probability of L pneumophila pneumonia should be reasonably high and results should alter patient care or have an impact on infection control measures. Criteria to identify patients at high risk for L pneumophila pneumonia should be developed in the context of local epidemiology. There were too few positives to recommend what those criteria might be, however, the literature would suggest advanced age, cigarette smoking, chronic lung disease, recent surgery and immunosuppression (14,16). Unfortunately, most patients admitted to hospital with pneumonia will have at least one risk factor. Perhaps only the most severely ill patients, and those with suspect single source pneumonia, should be tested.