Introduction

Q fever is a zoonosis, caused by the Gram-negative coccobacillus Coxiella burnetii (C. burnetii). There is a large animal reservoir, with goats, sheep and cattle being the most common source of human infections [13]. After primary infection, an estimated 1–5 % of patients progress to chronic Q fever, which can become manifest years after initial infection. Endocarditis and infections of aneurysms or vascular prostheses are the most common manifestations [4, 5]. Pre-existent cardiac valvular disease, aortic aneurysm, vascular grafts, immunocompromised state, and pregnancy are reported risk factors for the development of chronic Q fever [68]. Since 2007, the southern part of the Netherlands suffered from a major Q fever outbreak, with more than 4,000 reported symptomatic cases [9, 10]. As the majority of patients have mild or asymptomatic acute infection, the actual incidence is probably much higher. Due to hygienic measures, vaccination of goats and culling of pregnant goats, the acute Q fever epidemic in the Netherlands has subsided [9]. Nowadays, the brunt of the epidemic in the Netherlands is chronic infections [11]. The diagnosis of chronic Q fever is rather challenging and relies on a combination of clinical symptoms, risk factors, serology, polymerase chain reaction (PCR) and diagnostic imaging. The Dutch consensus group for the diagnosis of chronic Q fever categorized patients in three levels of certainty based on these factors [12]. According to this guideline, approximately 300 chronic Q fever cases have been registered in the Netherlands. Unfortunately, the sensitivity of C. burnetii detection in blood, by PCR or culture, is low. Therefore microbiological evidence still relies on serology. In the Dutch guideline, an IgG phase I titer of 1:1024, measured with an indirect fluorescent antibody test (IFAT), is considered as a threshold above which chronic Q fever is considered [12]. Recently, it was shown that there is a strong association between the level of the IFAT IgG phase I titer and PCR-confirmed (proven) chronic Q fever, and it was recommended that IFAT should be used as a screening tool for chronic Q fever [13]. However, immune fluorescence techniques are not used in all laboratories, which may use enzyme-linked immunosorbent assays (ELISA) and complement fixation tests (CFT) instead.

If left untreated, chronic Q fever leads to morbidity and mortality rates up to 60 % [5]. Long-term antibiotic treatment, preferably consisting of hydroxychloroquine and doxycycline, and sometimes aggressive surgery are required in patients with established chronic Q fever [5, 14]. While the guideline can be used for treatment guidance, the duration of treatment is not defined and is done empirically. Currently, based on available literature, treatment is continued until the IgG phase I titer as measured by IFAT has declined fourfold, but with a minimum of 18 months. However, when applying this, we observed examples of treated individuals with sustained clinical response but no detectable effect on IgG phase I titers. Reportedly, compared to patients recovered from acute Q fever, patients with chronic Q fever more often have IgA phase I antibodies, but IgA antibody testing has been abandoned since the introduction of an IFAT IgG cutoff in 1994 [15, 16]. Here, we compared kinetics of antibody responses, measured with IFAT, CFT and ELISA in early- and follow-up serum samples from patients with chronic Q fever, in order to find improved serological correlates for patient management.

Materials and methods

Case definition

Patients were classified as having proven, probable or possible chronic Q fever according to the recent Dutch guideline [12]. Briefly, the proven classification is used for patients with either positive C. burnetii PCR on blood and/or tissue or with an IFAT phase I IgG titer ≥1:1024 in combination with a proven (cardio) vascular infection as determined by diagnostic imaging. Probable infection is considered in C. burnetii PCR negative patients with an IFAT phase I IgG titer ≥1:1024 in combination with known risk factors for chronic Q fever, such as heart valve and vascular prosthesis, but without diagnostic signs of infection by positron emission tomography–computed tomography (PET-CT) or transoesophageal echocardiography. Possible chronic Q fever is used for patients with IFAT phase I IgG titer ≥1:1024 without signs or risk factors for chronic Q fever. According to the literature, successful treatment was defined as patients with a fourfold decline in IFAT IgG phase I titer in 18 months or more [17].

Patients

A total of 49 patients, diagnosed between May 2009 and April 2012 from four different hospitals were included; 32 patients were classified as proven, 12 as probable and five as possible cases. Antibiotic treatment was started in 32 patients after diagnosis (26 proven and six probable cases). Six proven cases died due to Q fever complications before antibiotic treatment was started. Clinical parameters associated with chronic Q fever, such as fever, abdominal pain and weight loss, as well as treatments (e.g. surgical treatment) were collected from the patient records. Based on these characteristics patients were clustered in three groups: asymptomatic cases, symptomatic cases and symptomatic cases with surgical intervention during antibiotic treatment.

Serum samples

For comparative analysis of serological test results at time of first sampling (t = 0), serum samples from all 49 patients were available. Ten patients were lost to follow-up due to death or transfer to another hospital, and 13 patients were diagnosed with chronic Q fever recently and therefore follow-up sera were limited to 1 year or less. Therefore, a remaining 189 serum samples were analyzed from 26 patients (16 proven and ten probable cases) treated for at least 18 or 24 months (3–13 samples per patient) to evaluate the antibody kinetics by use of the different methods. All sera were stored at –80° Celsius before antibody detection using CFT and ELISA was performed. Antibody testing using CFT as well as ELISA were done simultaneously (one batch). IgG phase I antibodies tested by IFAT were tested according to the standard operation performance (SOP) of our laboratory; all sera were tested twice, if a difference of two dilution steps or more was observed, serum was retested again.

IFAT (Focus Diagnostics, Cypress, USA)

IgG phase I and II, and IgM phase I and II were measured according to the manufacturer’s instructions with initial screening of a 1:32 dilution of serum in PBS. Read out was done visually using a fluorescence microscope at 400× magnification. Positive samples were further diluted in this test to establish an endpoint titer. Titers were expressed as the reciprocal of the highest serum dilution showing the diagnostic pattern of fluorescence.

CFT (Virion/Serion, Würzburg, Germany)

Phase I and II antibodies were measured according to the manufacturer’s instructions. After the procedure steps, the ready-to-use hemolytic system (Virion/Serion, Germany) was used before visually measuring the titers. Dilutions with 100 % of hemolysis were defined as positive. A positive result was defined as having an endpoint dilution of ≥1:8. In order to avoid prozone effect all sera were diluted to 1:512.

ELISA (Virion/Serion, Würzburg, Germany)

IgG phase I and IgA phase I were processed on a fully automated 4-plate ELISA processing system (DSX). Dilution protocol was used according to the manufacturer’s instructions, using 1:100 dilution for IgG as well as for IgA phase I. Data were analyzed according to the Virion/Serion protocol, reporting IgG and IgA phase I. IgG and IgA phase I were positive whenever the measured absorbance was more than 10 % above the extinction of the cut-off control. Ambiguous results were added to negative results. Extinctions were expressed in OD values.

Data analysis

Statistical analysis included computations of frequencies and analysis of agreement between ELISA IgG phase I, ELISA IgA phase I, IFAT IgG phase I and CFT phase I. To test significant differences between the three test methods, chi-square test was used or Fisher’s exact test when frequencies has an expectancy of five or less. To investigate the kinetics of IgG phase I antibodies, ratios of the titer or OD values at the end of treatment (18 or 24 months) and at the time of diagnosis (time point 0) were calculated for each method. A ratio above 1 was in accordance with a rise in titer and below one was in accordance with a decline in titer.

Results

At time of diagnosis

At t = 0, IgG phase I antibodies were more frequently detected with ELISA compared to CFT but this difference was not statistically significant (93.9 % [46/49] and 83.7 % [41/49], respectively, p = 0.20) (Fig. 1). This difference was entirely explained by differences in test outcomes for patients with proven Q fever, where IgG phase I antibody ELISA results were in full agreement with the IFAT (100 %; 32/32), whereas five cases were not considered serologically positive by using CFT testing (83.3 %; 27/32) (p = 0.053). Overall, using CFT phase I, eight out of 49 patients (16.3 %) were undetected. These eight patients had been categorized as proven (five), probable (two) and possible (one) chronic Q fever cases by IFAT based testing combined with criteria as defined in the Dutch consensus guideline. At t = 0, IgA phase I antibodies were not detected using ELISA in five patients. There was no difference in detection between proven and probable cases; in both categories one patient was undetected. Otherwise, three (out of five) possible cases tested negative.

Fig. 1
figure 1

Comparison of serological tests by percentage of positive results at time of first diagnostic sampling (t = 0) (ELISA IgA phase I, CFT IgG phase I, ELISA IgG phase I and IFAT IgG phase I)

Follow-up

The kinetics of IgG phase I antibodies (measured with IFAT, CFT and ELISA) of 26 patients are shown in Table 1 and Fig. 2. Patients were clustered in three groups: asymptomatic individuals (10), cases with symptoms (8), and cases with surgical intervention during antibiotic treatment (8). In general, it is visible that all kinetic options are possible: decreasing, stable or even increasing antibody levels. Patients were divided in three groups to examine whether the IFAT IgG phase I course reflects the clinical course of the patients better. Kinetics of the IgG phase I antibodies, based on the last and the first serum, is expressed per patient for IFAT, CFT and ELISA (Fig. 2). In the group of asymptomatic patients, the ratio was ≤ 0.25 (= ≥ fourfold decline) for IFAT in six out of ten patients. This differs from the symptomatic group with or without surgery: in both groups the minority of patients showed a significant decline using IFAT (ratio ≤ 0.25 in 3/8 patients in both groups). The numbers are however too small to draw statistical conclusions. When looking for kinetics of antibodies with the other serological methods, a poor correlation was found with the direction of response between measurements with the different assays. With CFT, a ratio ≤ 0.25 was rarely observed in all three groups, and most of the time (17/23) no significant decline or even an increase was calculated in the ratio. Patients 14, 42 and 12 were excluded, because CFT was negative in the initial sample. With ELISA, calculations differ from CFT and IFAT, because values are not expressed in titers, but in extinctions instead. Therefore, calculation of the ratio was leading to two options: below or above 1. Here again, a poor correlation was observed between the direction of antibody kinetics (increase or decrease) measured with ELISA versus IFAT.

Table 1 Results of anti IgG phase I test using CFT, IFAT (expressed in titers) and ELISA (expressed in OD value) in 26 patients before and after antibiotic treatment for chronic Q fever of at least 18 months
Fig. 2
figure 2

Twenty-six patients had treatment for at least 18 months. Each bar corresponds with one patient. Ratios of IgG phase I, measured with IFAT, CFT and ELISA, between time of diagnosis and end of treatment; a ratio above 1 is in accordance with a rise, and a ratio below 1 is in accordance with a decline, in titer. ↑ = patient with negative CFT value in initial sample

Discussion

In the present study, based on 49 chronic Q fever patients, we showed that qualitative differences exist in the currently available serological methods for establishing chronic Q fever diagnosis. Also, the use of serological testing as prognostic marker in patients at the end of a treatment period could not be established.

In the Netherlands, patients are classified as proven, probable or possible chronic Q fever based on serological and PCR results together with clinical symptoms and diagnostic imaging outcomes [12]. Antibiotic treatment is recommended in all proven Q fever patients. Most of the probable patients are treated with antibiotics as well, although this decision is taken after multidisciplinary consultation. Possible Q fever cases are monitored without antibiotic treatment. The cornerstone of serology in chronic Q fever is IgG phase I measured with IFAT and a level of 1:1024 is agreed as a threshold. In Dutch as well as in other research groups it was demonstrated that high IgG phase I titer and proven chronic Q fever are strongly associated [13, 18]. However, the specificity of the thresholds is low. In different (case) reports the clinical significance of high IgG phase I titers was criticized, especially in asymptomatic patients, monitored after acute Q fever patients or in patients identified during screening programs [19, 20]. Van de Hoek et al. showed that in a follow-up cohort of 686 acute Q fever patients, discrimination between acute and chronic Q fever was difficult as high IFAT phase I antibodies were found in both groups [21]. Although in a much smaller cohort, Min Nan et al. showed the same [20]. Both authors conclude that a high IgG phase I titer in an asymptomatic patient is of limited diagnostic value. In the present study we asked ourselves whether other serological techniques could be used to reduce to this specificity problem.

Different commercial ELISA kits are available for Q fever diagnostics. For diagnosis of acute Q fever, ELISA is proven to be a reliable test [22] and IgG phase II ELISA is routinely used nowadays for this purpose. ELISA is adapted for automation and therefore more suitable for screening purposes than IFAT. Recently, it was demonstrated that the sensitivity of ELISA for high IFAT IgG phase I antibody titers (≥1:1024) is 100 % [23]. Using the same commercial test we found in our study also 100 % sensitivity for the 30 proven patients in detecting IgG phase I. IgG phase I was undetected with ELISA in three patients (two probable and one possible). The numbers are too small to draw conclusions, but one could speculate about the possible added value of a negative ELISA IgG phase I in classifying chronic Q fever patients.

Five proven patients were missed in the initial serum with CFT. Prozone effect was excluded, as all sera were diluted and tested to 1:1024. Lower sensitivity could be explained by the fact that CFT is not IgG specific.

We tested for IgA phase I antibodies in order to find support for treatment decisions in patients without clinical symptoms but with risk factors for chronic Q fever. However, in our patient series, no significant difference in IgA phase I detection was found between proven and probable patients. In the patients with possible chronic Q fever, the majority (3/5) tested negative. This is in accordance with older studies, where IgA phase I antibodies, measured with IFAT or ELISA, were more often detectable in sera from patients with chronic Q fever compared to acute or past infections [15, 24, 25]. The value of IgA antibodies in the diagnostics of chronic Q fever has become less prominent in the last two decades, but it could be interesting to re-explore its importance more extensively using sera from all Dutch chronic Q fever patients.

We compared the use of different serological assays to monitor treatment effect, as has been recommended for IFAT. The results were quite diverse, showing a poor correlation between clinical response and serological response, and poor correlation between different methods in terms of direction of antibody response. The poor correlation between clinical and serological response may be related by the great diversity in the patient group. There are differences in the organs involved, in co-morbidity and in the time point of diagnosis, i.e. some patients are diagnosed with symptomatic disease, others are traced in screening programs. This heterogeneity challenged the interpretation of serological differences between patients. It has been shown that antibodies to C. burnetii (IgG as well as IgM) are detectable years after the initial infection in fully recovered patients [16, 22], and that follow-up should be extended to 18–24 months [26]. Literature on the use of other assays for serological follow-up is extremely limited, and these were too small to draw conclusions [25, 27].

ELISA could be a good alternative, because there is no observer bias due to the automated way of testing and a half-time is reported to be less than IFAT. However, the ELISA used in our study is designed qualitative, instead of quantitative. Earlier case reports mentioned good serological responses in endocarditis patients, including the former study of Peter et al. [25, 28, 29]. On the basis of our results we could not advise the ELSIA IgG phase I assay for follow-up of chronic Q fever patients. However, transforming the assay quantitatively with the use of calibration series could be interesting.

In conclusion, based on the data from this study we can conclude that there is no added value of current alternative tests in the diagnosis of chronic Q fever, compared to IFAT testing. In addition, we were unable to confirm the value of phase I IgG IFAT as a diagnostic marker for clinical response. We conclude that there is a need for improved prognostic and diagnostic markers.