Importance of Q fever in community acquired pneumonia

Abstract

Coxiella burnetii appears to be endemic in animals in the Mauricie region of Quebec, and causes some human cases of Q fever annually. Unlike in other rural areas, patients in this study experienced few respiratory symptoms. To determine whether C burnetii pneumonia is underdiagnosed, adults admitted to hospital for community acquired pneumonia were included in a one-year serological study. Significant immunofluorescent antibody (IFA) titres in four of 118 patients with pneumonia (fewer than 4%) were studied. Clinical presentation, standard laboratory tests and epidemiological data did not allow identification of these cases; however, Q fever increased during the warm months. There were no detectable complement fixing (CF) antibodies in these four cases. C burnetii causes few cases of pneumonia in Mauricie. IFA seems to be a more sensitive test than CF.

Q fever was recently identified as an endemic zoonosis in Mauricie, a partially rural area in the central part of Quebec, located near the Eastern Townships. The first Canadian cases of Q fever were reported in this area at the end of the 1950s. A regional study conducted in 1992 and 1993 showed an animal seroprevalence for Coxiella burnetii of 28% (82 of 297) in cats and 12% (12 of 104) in cattle (unpublished data). Nine human cases were diagnosed during an 18-month period, and a 20% human seroprevalence in a local slaughterhouse was documented (2). Clinical cases were primarily middle-aged men who reported an abrupt onset of a high fever, severe headache, fatigue and myalgia. None of the patients had clinical or radiological evidence of pneumonia; this finding was remarkable because Q fever is known to cause an atypical pneumonia (3–13). For example, in 10 rural Nova Scotia hospitals, C burnetii was responsible for 21.8% of community acquired pneumonia in 1983 (3). Q fever may be underdiagnosed because clinical symptoms, x-rays and standard laboratory tests are often nonspecific (13). Complement fixation (CF) serology can miss 20% to 46% of cases (3,14). Q fever is important because the infection does not respond to beta-lactams or erythromycin, and there is a risk of reactivation in the future. To understand the role of Q fever in adults hospitalized for community acquired pneumonia in Mauricie, a one-year serological study was conducted.

PATIENTS AND METHODS

The Centre Hospitalier St-Joseph is a 300-bed, adult hospital in Trois-Rivières, a city with a population of 50,000, located in a partly rural area north-east of Montreal (regional population of 450,000). It is a reference centre for care in respirology and infectious diseases with four respirologists and three infectious disease specialists on staff. Between November 1992 and November 1993, the hospital admission list was reviewed daily to identify patients admitted with suspected pneumonia. Inclusion criteria were three or more clinical symptoms including sputum, chest pain, fever, dyspnea, changes in consciousness, abnormal lung auscultation and leukocytosis, and chest x-ray compatible with pneumonia.

Of 184 patients with an initial diagnosis of pneumonia, 66 were excluded because of unconfirmed diagnosis (39), rapid discharge (six), refusal or inability to consent (six), rapid death (five) or nosocomial complications (12 patients transferred from other centres).

The remaining 118 patients were enrolled in the study after providing written consent. They answered a brief epidemiological questionnaire concerning contact with animals during the previous month, occupation and home location. One serological sample was drawn at enrollment and a second one after a minimum three-week interval. Analysis was done at Laboratory Centre for Disease Control in Ottawa. Two assay methods were used for the first 41 patients; subsequently, CF was used only on samples that tested positive in the immunofluorescent antibody (IFA) test.

IFA test:

Coxiella burnetii phase II antigen prepared from the Nine Mile strain was purchased from Centers for Disease Control and Prevention (CDC, Atlanta, Georgia), and antigen-coated wells were prepared as recommended by CDC. Sera to be tested were diluted in phosphate buffered saline (PBS) containing 3% normal yolk sac. Sera were tested at screening dilutions of 1:32 because fluorescence is less specific under this dilution for most Rickettsiae. Positive sera were retested at initial dilutions of 1:16, using phase I antigen. Test sera were added to antigen-coated wells and incubated in a moist chamber at 37°C for 35 mins. Slides were washed twice with PBS and then air dried. A fluorescent isothiocyanin-labelled goat antihuman conjugate purchased from Sigma Chemical Company (Missouri) diluted in PBS containing a 1:400 dilution of Evan’s Blue was added to each well, and the slides were incubated in a moist chamber at 37°C for 35 mins. The slides were washed twice with PBS, air dried, mounted using phosphate-buffered glycerol and then examined by ultraviolet microscopy at a magnification of 400.

CF test:

CF tests were conducted by a modification of the microtitre method described by Sever (15). Two units of anti-sheep hemolysin, two units of complement and four units of antigen were used in the test. The sheep cell concentration was reduced to 0.4% for greater sensitivity. Initial dilutions of sera tested were 1:8. Q fever phase II antigen used was purchased from Instituto Immunlogico Italiano (ISMUNIT), Rome, Italy.

The criteria for a positive serological diagnosis was a fourfold rise in antibody titre or, if only a convalescent serum was available a titre of 256 or greater by IFA, or 64 or greater by CF. Phase II:phase I ratio is generally greater than one in acute cases and one or less in chronic cases. Immunoglobulin (Ig) M was not consistently tested, but was completed in two ambiguous cases.

RESULTS

Of the 118 patients enrolled, nine did not complete the study: five died before the second serum sample could be taken, and four were noncompliant. Of the 109 remaining patients, 104 were negative for C burnetii serology. Four patients had positive serological results (Table 1); a fifth patient had a nondiagnostic reaction.

TABLE 1.

Epidemiological data and serological results for patients with Coxiella burnetii pneumonia

Case	Sex (age years)	Period of hospitalization	Animal contacts	Tested after onset of symptoms (days)	Serological data
					IFA		CF
					Phase II	Phase I
1	Male (76)	May	Cats, cows	6	<16	NT	AC
				27	32		<4
2	Female (67)	July to December	Rabbits	6	<32	NT	NT
				26	<32		NT
				56	32		<4
				88	64		<4
				271	32		<4
3	Male (79)	August	Cats, dogs	11	<16	NT	<4
				33	64		<4
				255	<32		<4
4	Male (80)	October	Cats, dogs	5	256	32	<4
				48	256	64	AC
				166	128	32	<4

AC Anticomplementary reaction; CF Complement fixation test; IFA Immunofluorescent antibody test; NT Not tested

The first patient, a cat owner, often walked near dairy farms. He was admitted with bacteremic pneumococcal pneumonia and initially responded to treatment with penicillin. Six days later, fever and cough recurred, and he did not respond to 10 days of cefadroxyl treatment. Finally, the fever subsided with ciprofloxacin therapy, and the infiltrate regressed very slowly. Some anticomplementary reaction was initially present in this patient’s serum.

The second patient was in contact with her grandchildren’s young rabbits; she was hospitalized four times over a five-month period for pneumonia. Pseudomonas aeruginosa was identified in her sputum cultures during her first two hospitalizations. She received different antibiotics, some of which were quinolones. Bronchiectasis was not present and the cause of her repeated pneumonias was not apparent.

The third patient had an abrupt onset of fever, headache and liver tenderness, followed by severe pneumonia six days later. There was no clinical response to ceftriaxone and erythromycin. Aspiration pneumonia was then suspected. Subsequently he received clindamycin and recovered a few days later.

Five days after the onset of symptoms, the fourth patient had a high phase II titre by IFA, a high phase II:I ratio of 8 and later an anticomplementary reaction. However, IgM was not detectable. No febrile episodes had been reported within the past three years, and the patient remained in good health in the following year.

The fifth patient, who had occasional contact with a dog, was admitted with bacteremic pneumococcal pneumonia which responded to penicillin. He had an IFA titre of less than 32 on the first serum, and a low titre (32) in the second sample drawn five weeks later. CF was negative and IgM antibodies were negative by IFA. This was considered to be the consequence of a previous infection.

The four cases were similar to the 105 controls in sex (male 75% versus 51%), age (75 years versus 66), employment rate (0% versus 14%), animal contacts (100% versus 70%) and home location. The four cases occurred between May and October (four of 53 pneumonia during these months), but none occurred between November and April.

DISCUSSION

Ninety-two per cent of enrolled patients completed the study. There was an average of 27 days between the taking of the two serological samples. C burnetii does not appear to be an important cause of community acquired pneumonia in this area. The rate (less than 4%) is much lower than the 21.8% observed in rural Nova Scotia 11 years ago (3).

Our patients were relatively old, often retired and likely exposed to a smaller inoculum than subjects with occupational acquisition. The results did not provide a population-based estimate of the incidence of C burnetii pneumonia, but documented the limited importance of this microorganism in the illnesses of adults sufficiently ill to require hospitalization. Other cases that were clinically mild likely occurred in this area during the study period, but remained undetected.

The literature suggests that the clinical manifestations of Q fever vary with age (3,5,16–18), strain of organism (13,19,21), presence of immunosuppression (20–23), density of inoculum (4,5) and route of infection (12,24). Our previously reported human cases (2) were similar to those reported in Ontario (25), with a febrile syndrome and abnormal liver tests, but rarely manifesting respiratory symptoms.

From these observations, neither clinical manifestations nor standard laboratory tests were useful in identifying cases. Similar observations were reported in Nova Scotia where C burnetii pneumonia was correctly suspected in only 33% of patients, even when patients presented more typically with fever, chills, severe headache, anorexia and a nonproductive cough (13). The literature suggests that approximately 45% of patients do not present with respiratory symptoms (4,11). The clinical presentations of our patients varied, perhaps due to well known complications that occur in elderly patients, and included secondary infection (12%), congestive heart failure (11%), gastrointestinal hemorrhage (3%), respiratory failure (3%) and renal failure (2%) (27). Severity of infection is generally moderate but about 25% of admitted patients have rapidly progressive pneumonia (13) and death may occur (12).

Epidemiological observations collected here were not useful, except perhaps the observation that greater incidence of disease was seen during the warm months. Tracking of animal contact was unhelpful; in rural areas animal contacts are frequent. In Spain, seasonal variation was related to the reproductive cycle of domestic animals, with most infections occuring in young men (10,11). Laboratory observations were also unhelpful. Half of the patients had mild elevation of transaminases. In addition sedimentation rate was often elevated, but white blood cell counts remained normal (11,13). Radiographical findings also vary, although some authors have reported round pulmonary opacities (4), subsegmental and segmental pleural-based alveolar opacities (13) and slow clearing infiltrates (26), suggesting the presence of C burnetti.

Due to the lack of specific clinical or laboratory findings, underdiagnosis of C burnetti pneumonia is likely.

Serological cross-reactions that would explain a false C burnetii antibody increase during bacterial or rickettsial infections are not described in the literature (14,27,28,33). Although CF testing is believed to be quite specific, it is generally conceded to be less sensitive and more labour intensive than IFA (3,14,20). Anticomplementary activity of serum is frequently present in Q fever, which poses a diagnostic problem (4,14,29). Antibodies are detected earlier by IFA and seroconversion may be missed, (3) so specific IgM serology could theoretically aid the diagnosis of acute infection. However, IgM antibodies are often detected more than six months after an acute infection (30,31), as well as during the chronic infection phase (32,33). The sensitivity of IgM antibodies is also poor (14), and the test is not currently available to clinicians. The fourth case, who did not remember any febrile episode in the three years before admission to hospital and stayed in good health the following year, was considered an acute case, but a recrudescence could not be excluded.

This study does not suggest that C burnetii is a frequent cause of pneumonia in our geographic area. Therefore, routine serological testing for this agent is not recommended. Further studies are required to determine the exact role of C burnetii in the pathogenesis of pneumonia in the elderly. In the near future it is hoped that there will be more information about the incidence and clinical manifestations of Q fever in Canada.