Friday 15 March 2013
Chronic Q fever in The Netherlands
Promotor: Prof. dr. Andy Hoepelman
Defence: 15 March 2013
Q fever is a complex disease, with both acute and chronic manifestations. Detected in 1935, its name was derived from the phrase ‘query’. Nowadays, this name is still relevant: many aspects of this disease remain unsolved and disputed. With new insights from the recent outbreak of Q fever in the Netherlands, we tried to fill some of the gaps in the knowledge about Q fever.
Part 1. Burden and magnitude of Q fever in the Netherlands
From 2007 on there has been a large outbreak of Q fever in the Netherlands, which subsided in 2010 only after interventions from the government in the veterinary sector.
In chapter 2 we have estimated the acute Q fever-associated in-hospital mortality rate. Available information from six hospitals in the Q fever affected areas revealed 749 patients admitted with acute Q fever, of which nine patients (1%) died. The patients who succumbed from, or with, Q fever were of older age (age >65 years), were male, and had serious underlying diseases, which already raised the a-priori risk of death. This mortality rate is comparable to earlier studies in other countries. As there is no obligation for registration of hospitalized patients with acute Q fever and patients who died of acute Q fever, these numbers could even be overestimated. Moreover, in the beginning of the Q fever epidemic many clinicians were not aware of this disease entity and Polymerase Chain Reaction (PCR) as a diagnostic tool for acute Q fever was not introduced yet. This may have caused some acute Q fever cases to have been left undiagnosed. This study illustrates the low mortality rate of acute Q fever.
To assess whether early antibiotic treatment inhibits antibody responses after an acute Q fever episode, we have studied IgG antibody responses in symptomatic patients diagnosed and treated either before, or during the development of serologic response to C. burnetii in chapter 3. Serological profiles of 181 acute Q fever patients of whom 91 were diagnosed early (ED) (defined as negative IgM phase II with enzyme-linked immunosorbent assay (ELISA) and positive PCR) and 90 were diagnosed late (LD) (defined as positive/dubious IgM phase II with ELISA and positive immunofluorescence assay (IFA)), were analysed. Serologic profiles consisting of IgG phase I and II titres, were determined using IFA at three, six, and 12 months. At 12 months, 28.6% of ED patients and 19.5% of LD patients had high IgG antibody titres (defined as IgG phase II titres ≥ 1:1024 and IgG phase I ≥ 1:256), which was not statistically significant. In addition, we also found no significant differences in frequencies of IgG phase I and IgG phase II antibody titres at all follow-up for adequately and inadequately, and early- and late-treated patients. This study indicates that early diagnosis and antibiotic treatment of acute Q fever do not preclude development of the IgG antibody response.
In chapter 4, we have estimated the magnitude of the Q fever outbreak in the Netherlands by extrapolating the prevalence of C. burnetii antibodies in two high-risk groups for development of chronic Q fever in 2010 and 2011. The 785 patients in these populations were all living in the catchment area of the Jeroen Bosch Hospital (JBH) in ’s-Hertogenbosch, which is located in the centre of the epidemic area. Seropositivity for IgG phase II antibodies to C. burnetii was present in 10.7% of patients (95% confidence interval (CI) = 8.5–12.9%). Seroprevalence was not influenced by age, gender or area of residence. By extrapolation of these data, we estimated that 40,600 persons (95% CI = 32,200–48,900) in the JBH catchment area have been infected with C. burnetii. This figure by far exceeds the number of notified symptomatic acute Q fever patients and illustrates the magnitude of the past Dutch Q fever epidemic.
Part 2. Diagnosis and classification of chronic Q fever
Faced with a large Q fever epidemic in the Netherlands, the Dutch Q fever Consensus Group was constituted. First, a guideline for the diagnosis of acute Q fever was formulated (see introduction). However, after 2009 an increasing number of chronic Q fever patients was observed, with a significant amount of morbidity and mortality. Diagnosis of chronic Q fever proved to be complex and uniformity in diagnosis of chronic Q fever was lacking. Therefore, a review of the available literature, which was limited to case reports and mainly retrospective and descriptive analyses of a small number of patients, was performed. A proposal for a new guideline for the diagnosis of chronic Q fever was formulated, which combines clinical, radiological and microbiological factors. This guideline, in which chronic Q fever is categorized as proven, probable or possible, and the literature it is based on is described in chapter 5. Proven chronic Q fever is classified as either positive C. burnetii PCR (or culture) in blood or tissue, in absence of an acute infection, or patients with phase I IgG ≥ 1:800 or phase I IgG ≥ 1024 (depending on in house IFA technique or commercial IFA technique, respectively), in combination with definite endocarditis according to the revised Duke criteria, or evident aneurysm or vascular graft infection on computed tomography (CT), positron emission tomography (PET)-CT, duplex ultrasound or magnetic resonance imaging (MRI). Long-term antibiotic treatment should always be initiated in case of proven chronic Q fever and frequent clinical and serological monitoring, at least three-monthly, is warranted. Surgical intervention should be considered in absence of favourable response to treatment. Probable chronic Q fever is diagnosed in those patients with phase I IgG ≥ 1:1024 who have established risk factors for chronic Q fever, show echocardiographic abnormalities not meeting the revised Duke criteria, probable rare manifestations of Q fever (e.g. hepatitis, osteomyelitis) or signs of systemic inflammation. It is advised to discuss whether to start antibiotic treatment in a multidisciplinary team. Three-monthly clinical and microbiological follow-up, and radiographic imaging (echocardiogram, PET/CT) whenever clinical improvement stagnates or worsens, is advised. Possible chronic Q fever is diagnosed in patients with solely a phase I IgG ≥ 1:1024, without any of the manifestations mentioned in the categories proven and probable, and largely reflects patients with no chronic Q fever at all, but only elevated antibodies against C. burnetii. Antibiotic treatment is not warranted. The follow-up regime is equal to patients with probable chronic Q fever.
Publication of this manuscript led to discussion in the international medical literature. Subsequently, Professor Raoult from France proposed another diagnostic guideline for chronic Q fever. In chapter 6 a comparison is made between this diagnostic guideline and that of the Dutch Q fever Consensus Group. An important difference in the diagnostic criteria proposed by prof. Raoult and the Dutch Q Fever Consensus Group is the diagnostic value attributed to C. burnetii PCR positivity of blood samples, which is not considered proof of chronic infection by prof. Raoult. We showed that with prof. Raoult’s criteria, approximately 30% of proven chronic Q fever cases would be missed, including at least four patients that eventually died due to chronic Q fever-related causes, and almost all probable and possible cases. Although specificity of the Dutch guideline might be lower, sensitivity is importantly higher.
In chapter 7 we have studied the height of phase I IgG antibody titres at the time of positive PCR in blood, at diagnosis, and at peak levels in patients with proven, probable or possible chronic Q fever. Positive C. burnetii PCR on blood was seen in 65% of proven chronic Q fever cases, demonstrating the relative low sensitivity of PCR for the diagnosis of chronic Q fever. In contrast to previous reports, positive PCR was associated with high phase I IgG titres. In patients with proven chronic Q fever, phase I IgG titres were significantly higher compared to patients with probable and possible chronic Q fever. The positive predictive values (PPV) of phase I IgG titres for proven chronic Q fever, when compared to possible chronic Q fever, at titres 1:1024, 1:2048, 1:4096, ≥1:8192, were 62%, 67%, 77%, >86%, respectively. On the other hand, sensitivity was 98%, 95%, 81% and <60%, respectively. These figures indicate that raising the cut-off titre for the diagnosis of chronic Q fever would lead to an unacceptable number of missed chronic Q fever cases. Whereas, with the current cut-off of 1:1024, the number of chronic Q fever cases is overestimated.
Part 3. Chronic Q fever, risk groups, morbidity and mortality
To reduce morbidity and mortality of chronic Q fever it is of importance to identify high-risk groups for the development of this manifestation. Knowledge on risk factors could be used in early case-finding, before complications occur, by screening of high-risk populations. In chapter 8 a case-control study is presented, which aimed to identify and quantify risk factors for development of chronic Q fever after C. burnetii infection. Co-morbidity, cardiovascular risk factors, medication and demographic characteristics from patients with proven, probable or possible chronic Q fever (cases), were compared with patients who had acute Q fever in 2009, but did not develop chronic Q fever (controls). Previous valvular surgery, vascular prosthesis, aneurysms, renal insufficiency, and increasing age were identified as major risk factors for the development of chronic Q fever in this study. Remarkably, an association between (mild) non-surgical heart valve pathology and chronic Q fever was not found.
A previously defined high-risk group for chronic Q fever, namely patients with a history of cardiac valve surgery, was evaluated in chapter 9. Patients with a history of cardiac valve surgery in the Q fever epidemic area were selected for microbiological screening, by means of analysis of phase I and II IgG against C. burnetii. If titres were above a defined cut-off level, C. burnetii PCR was performed. Chronic Q fever was diagnosed according to the consensus guideline of the Dutch Q Fever Consensus Group. We found a seroprevalence of C. burnetii antibodies of 20.4% in this population. Proven or probable chronic Q fever was identified in 7.8% of seropositive patients, which was remarkably lower than incidence figures in previous reports of this high-risk group. The risk for chronic Q fever seemed to be independent of type of valvular surgery, although numbers of patients with chronic Q fever were small.
In response to the rising number of chronic Q fever cases after 2009, and the many questions around this disease, a Dutch National Chronic Q Fever Database was implemented, to address these questions. All Dutch hospitals that treat chronic Q fever patients were actively approached to include patients in this database. In chapter 10 we describe the characteristics and outcome of chronic Q fever patients included until five years after the start of the Q fever epidemic. We specifically focused on mortality of chronic Q fever. In total, 284 chronic Q fever patients were identified, of whom 151 (53.7%) had proven, 64 (22.5%) probable and 69 (24.3%) possible chronic Q fever. The majority of proven and probable chronic Q fever patients had a vascular focus of infection (56.7%), while endocarditis (34.9%) was less prevalent. Only 27.0% recalled an acute Q fever episode. In our population, mortality among proven and probable chronic Q fever patients related to chronic Q fever was 13.0%: 9.3% among endocarditis patients and 18.0% among vascular chronic Q fever patients. Older age and presentation with vascular complications necessitating acute surgical intervention were the major risk factors for chronic Q fever mortality.
Part 4. Important examples of clinical manifestations of Q fever endocarditis
Endocarditis is one of the major manifestations of chronic Q fever. Nevertheless, it is not always recognized. Early treatment can prevent substantial morbidity and mortality. In this part we describe two manifestations of Q fever endocarditis, which may facilitate (early) case finding by clinicians.
Endocarditis due to a concomitant infection with C. burnetii and another causative agent has thus far been incidentally described in seven cases only. In chapter 11 three other patients with endocarditis by C. burnetii concomitant with another bacterial pathogen are described.
Q fever endocarditis can present very subtle, with no or few symptoms. Cardiac valve infection with C. burnetii is difficult to differentiate from degenerative changes of cardiac valves on echocardiogram. In chapter 12 we presented three cases of chronic Q fever, in which diagnosis was delayed until after cardiac valve surgery, warranting better surveillance for C. burnetii infection in case of cardiac valve surgery in Q fever endemic areas.
Incidence of acute and chronic Q fever
During the Q fever epidemic in the Netherlands, over 4000 cases of acute Q fever were registered.1 This is an underestimation of the total number of C. burnetii infections, due to a significant amount of asymptomatic infections. The percentage of asymptomatic human C. burnetii infections is much higher in our studies than the previously reported 50-60%.2;3 In chapter 9, only 4.9% of seropositive patients with a history of heart valve surgery had an actual history of acute Q fever. Seroprevalence studies, described in chapter 4, and among blood donors4, demonstrate a tenfold higher number of C. burnetii infected individuals, than the 4000 cases reported by the Dutch governmental health institutes. Follow-up of a former screening in the general population in which the incidence of Q fever antibodies was 2.4% in 2006,5 in combination with analysis of notified acute Q fever cases in this population, could give more insight in the percentage of asymptomatic Q fever infections and the number of people potentially at risk of chronic Q fever.
In the literature it is stated that 1-5% of patients develop chronic Q fever after acute Q fever,6;7 although definitions of chronic Q fever were based on serologic results only. We have found a similar incidence in our studies for all chronic Q fever cases, including possible chronic Q fever. Proven and probable chronic Q fever cases had experienced an episode of acute Q fever less often, whereas most patients with possible chronic Q fever were identified in follow-up programmes after acute Q fever. In chapter 10 we demonstrated that the risk for development of chronic Q fever after a known acute Q fever episode is approximately 2.5% if all chronic Q fever cases are taken into account, and 1.5% if limited to proven and probable chronic Q fever cases. These incidence numbers were confirmed by van der Hoek et al, who demonstrated 1.6% progression to chronic Q fever in a group of 686 unselected patients diagnosed with acute Q fever in 2007.8 Nevertheless, we have also demonstrated that the majority of patients with chronic Q fever did not recall an episode of acute Q fever reflective of the fact that the percentage of asymptomatic infections is much larger than previously reported. Therefore, the percentage of people who will develop chronic Q fever after primary C. burnetii infection, whether symptomatic or asymptomatic, is presumably much lower than 1%.
Risk factors for chronic Q fever development and follow-up regimes
In the light of a high percentage of asymptomatic individuals with a primary C. burnetii infection, there is a large population at risk for chronic Q fever, who do not benefit from current follow-up programs for patients with acute Q fever. Follow-up programs in the Netherlands consist of serological examinations for at least one year after the acute Q fever episode. There is controversy in the international medical literature, whether all patients with acute Q fever should be screened for heart valve defects by echocardiogram.6;9;10 A Dutch study demonstrated no benefit of a screening echocardiogram for all acute Q fever patients.9 Moreover, in chapter 8 risk factors for chronic Q fever were analysed, in which (mild) cardiac valvulopathy, in contrast to previous cardiac valvular surgery, was not identified as risk factor. Incidence of chronic Q fever in case of C. burnetii infection and cardiac valvulopathy was previously estimated at 40%, and even higher in case of heart valve prosthesis. However, our screening study among patients with a history of heart valve surgery, demonstrated a much lower chronic Q fever incidence of 7.9%. Another recent Dutch screening study in a Q fever epidemic area of high risk groups for chronic Q fever, including both patients with heart valve prosthesis, aortic aneurysms and vascular prosthesis showed that 19% of C. burnetii seropositive patients had chronic Q fever.11 Importantly, both studies detected patients with chronic Q fever before occurrence of complications, with no or minor symptoms, and, in case of suspected endocarditis, no abnormalities on echocardiogram. Previously proposed high risk factors consist, besides pre-existing cardiac valvulopathy, of vascular grafts and aneurysms, immunosuppression and pregnancy.7;12-16 Identified risk factors for chronic Q fever in this thesis (chapter 8) were previous cardiac valvular surgery, vascular prosthesis, aneurysms, renal insufficiency, and increasing age. An association between immunosuppression and chronic Q fever could not be confirmed in this study, although total numbers of patients with immunosuppression were very small. However, in the international medical literature concerning chronic Q fever, clear definition of immunosuppression and statistical empowerment proving an association with chronic Q fever is lacking.13 Numbers were too small to reliably evaluate pregnancy as a risk factor for chronic Q fever. However, during the Dutch Q fever epidemic only one case of pregnancy-related chronic Q fever has been identified (see also chapter 10).17 A novel finding was the association between mild renal insufficiency and chronic Q fever, which may be explained by the fact that renal insufficiency is associated with vascular disease. Increasing age also predisposed for the development of chronic Q fever, which was also illustrated in another Dutch study by van der Hoek et al.8 Explanation probably lies in the increased prevalence of cardiovascular diseases and the decrease of cellular immunity during aging.18;19 Age >60 appeared the best age cut-off above which the risk for chronic Q fever is increased significantly. Confirmation of the findings of this study and elucidation of the risk factors immunosuppression, pregnancy and even non-surgical cardiac valvulopathy is needed from future research.
It could be debated if the current follow-up regime for detection of chronic Q fever needs adjustment. The benefit of screening all patients after acute Q fever is relatively low (1-2% of patients develop chronic Q fever), while mainly patients with possible chronic Q fever are detected. The total costs of the current follow-up regimes during and after the Dutch Q fever epidemic are unknown, but need evaluation. More importantly, a large amount of patients is missed with follow-up regimes, as most chronic Q fever patients, especially proven and probable cases, did not recall an episode of acute Q fever. Results of the studies described in this thesis indicate that early case-finding of chronic Q fever patients in case of a Q fever epidemic may benefit more from screening programs for risk populations. The highest risks for chronic Q fever were observed in patients with a history of valvular surgery, aortic aneurysms and vascular prosthesis. Therefore, we would recommend serological screening for chronic Q fever in epidemic areas in these patient groups. Older patients (age >60) and patients with mild renal disease seem also at risk, probably because of previous unidentified cardiovascular disease. The risks of chronic Q fever for immunocompromised patients and pregnant women are until now unclear. Cardiac non-surgical valvulopathy was previously indicated as an important risk factor for chronic Q fever (risk of up to 40%), but could not be identified as such in our study. For patient groups with age >60 years, mild renal disease, non-surgical cardiac valvulopathy, immunosuppression, or pregnancy, we would recommend thorough follow-up after acute Q fever. In light of the identified risk factors in this thesis, it seems to be more beneficial in case of acute Q fever, to perform a thorough medical examination and history taking for determination of the follow-up plan, instead of follow-up of all acute Q fever cases. Screening for risk factors with abdominal ultrasound, especially when vascular risk factors are present, may be more profitable than performing an echocardiogram to detect (minor) valvulopathy.
Manifestations of chronic Q fever
The most remarkable difference compared to reports from the international literature concerning chronic Q fever is the large number of vascular infections we have identified in the Netherlands. This manifestation is numerically at least of equal importance as endocarditis. In chapter 10 we found significantly more patients with a vascular focus of infection (56.7%) than patients with endocarditis (35.3%). This disparity with international observations might be caused by strain-specific differences and regional differences in hosts genetic background. Nevertheless, chronic Q fever vascular infections are increasingly described in literature from other countries in the last years.13;20-23 Therefore, another explanation might be the increased awareness for this previously relatively unacknowledged manifestation among clinicians and microbiologists, especially in the Netherlands. Of concern is the fact that prognosis of vascular chronic Q fever is far worse than of Q fever endocarditis as was previously described in the international literature but also observed in our studies.6;12 In chapter 10 we described that the majority of patients who probably or definitely died of chronic Q fever-related complications, had a vascular focus of infection (79%). As vascular chronic Q fever often presents with severe life threatening complications like acute aneurysms or prosthetic dysfunction, it is possible that patients with this manifestation were missed, because they succumbed before diagnosis of chronic Q fever could be made.
Diagnosis of chronic Q fever
The diagnostic criteria of chronic Q fever have recently led to debate in the international literature. A positive C. burnetii PCR, in the absence of signs of acute Q fever, proves chronic Q fever, but this method has low sensitivity in blood, which was also demonstrated in this thesis (chapter 7). In contrast to previous reports, we found a positive association between the height of phase I IgG titres and positivity of PCR on blood samples. Sensitivity of C. burnetii PCR on infected tissue is probably much higher, but definite proof is currently lacking mainly because the focus of infection of chronic Q fever is usually situated in difficult accessible locations like heart valves, aneurysms and vascular prosthesis. Until now, diagnosis of chronic Q fever relied mainly on serological examination with IFA as most commonly used tool. An elevated phase I IgG titre of >1:800, based on a French in-house IFA, or >1:1024, based on a commercially available IFA (Focus Diagnostics), has been used for the serological diagnosis of chronic Q fever.8;24 In this thesis, we have demonstrated that high phase I IgG titres have a high positive predictive value (PPV) to differentiate proven chronic Q fever from possible chronic Q fever, especially if titres exceed 1:4096. Nevertheless, sensitivity of high phase I IgG titres is low: 60% in case of phase I IgG titres >1:4096. In contrast, phase I IgG titres between 1:1024 and 1:4096 have high sensitivity (98% to 81%), but a low PPV (62.2% to 76.5%). These results of this study emphasizes that serology is not a diagnostic tool on its own, but should be interpreted in relation with the clinical background. Uniformity in diagnosis is important for uniform treatment and follow-up advice and adequate comparison of results of future research on chronic Q fever. The diagnostic guideline of the Dutch Q fever Consensus Group combines PCR, serology, clinical data and radiological data. In this guideline, suspected chronic Q fever cases are divided in three groups in decreasing likelihood of chronic Q fever: proven, probable and possible. In comparison with the recent guideline proposed by prof. Raoult, we have demonstrated that the guideline of the Dutch Q fever Consensus Group has better sensitivity and also takes more rare manifestations of chronic Q fever into account. Although part of the patients with probable chronic Q fever and most patients with possible chronic Q fever will eventually not have actual chronic Q fever, we do think that these patients should all be analysed for a chronic Q fever focus and should remain under close follow-up. We do not advocate treatment for all these patients.
The best way to solve the dispute about chronic Q fever diagnostic criteria is a more reliable strategy to establish this manifestation of Q fever. With the current diagnostic arsenal, consisting of PCR and serology, differentiation between past and chronic Q fever is insufficient. A gold standard for the diagnosis of chronic Q fever is lacking. Research into better chronic Q fever diagnostic criteria and methods are in progress, but more time is needed to evaluate these new developments.25-29
PERSPECTIVES AND FUTURE RESEARCH
The recent outbreak of Q fever in the Netherlands has brought about new perspectives on this complex disease. We found some important dissimilarities between findings from the international medical literature on the topic of Q fever, and findings obtained from the Dutch Q fever epidemic. There are three main reasons for the observed dissimilarities between international studies and studies presented in this thesis. First, until now, most research was based on work from a single reference laboratory and clinic for C. burnetii infection (Marseille, France). While this centre has performed important and solid work on the topic of Q fever, using data from a reference centre can cause selection bias. Second, strain specific differences, and differences in host genetic background, could lead to distinct clinical presentations and risk profiles.30 Third, former studies often analyzed cases from both endemic and epidemic origin, which were identified over a considerable time span, while the Dutch studies analyzed cases from a single four-year-lasting epidemic.
To determine better follow-up strategies after a Q fever outbreak, there is a need for clarification of the risk for chronic Q fever after C. burnetii infection. Although it is evident that patients with a history of valvular surgery, aortic aneurysms and vascular prosthesis possess a high-risk, this needs to be elucidated for other potential risk-factors like non-surgical cardiac valvulopathy, immunosuppression, (mild) renal insufficiency and pregnancy. Besides, it is important to determine the number of patients with asymptomatic primary infection, as we have proven that they are also at risk for chronic Q fever. Data from the Dutch epidemic indicate that the number of asymptomatic primary infections is far beyond the formerly stated 60%. Screening in the general population and comparing incidence of Q fever antibodies before and after the Dutch Q fever epidemic in combination with analysis of notified acute Q fever cases in this population, could give more insight in the percentage of asymptomatic Q fever infections. Moreover, with these data a new case-control study, as described in chapter 8 could be performed, now also including asymptomatic primary C. burnetii infections.
We have demonstrated that chronic Q fever still has a high mortality, especially in case of vascular infection. Early case-finding, by targeted screening in risk-groups, could improve prognosis. On the other hand, best treatment strategies are not fully elucidated yet. Although not subject of this thesis, in daily practice the antibiotic regime for chronic Q fever, consisting of doxycycline and hydroxychloroquine, has many side-effects and efficacy is unclear. Also, the role of surgical intervention in case of chronic Q fever needs clarification. The Dutch Chronic Q Fever Database could provide more answers about this topic in the future. Nevertheless, as most research on Q fever is descriptive and retrospective from nature, there is also a need for randomized trails, especially for treatment regimes.
This thesis illustrates that Q fever, and in particular chronic Q fever, is a complex disease with diverse manifestations and still many queries. We have specifically focused on long-term sequels of C. burnetii infection. The most important manifestations of chronic Q fever in the Netherlands are infections of aortic aneurysms, vascular prosthesis, and, to a slightly lesser extent, cardiac valves. Major risk factors for development of chronic Q fever are, correspondingly, aortic aneurysms, vascular prosthesis and history of valve surgery. Other risk factors, including increasing age, mild renal disease, non-surgical cardiac valvulopathy, immunosuppression, and pregnancy, should also be taken into account, but need further evaluation. Early case-finding in epidemic areas could diminish morbidity and mortality of chronic Q fever, for which we would like to recommend targeted screening for risk groups. Diagnosis of chronic Q fever is debated, but until further research offers better diagnostic tools, we feel that the diagnostic guideline of the Dutch Q fever Consensus Group is most accurate.