Correspondence |
Molecular detection of Coxiella burnetii in blood and sera during Q fever
Unité des Rickettsies Faculté de Médecine Université de la Méditerranée Marseille France email: didier.raoult{at}medecine.univ-mrs.fr
Sir,
We read with interest the intriguing article by Marmion et al.1 which describes the molecular detection of Coxiella burnetii in two cohorts of patients from Australia and the UK, 5 years and 12 years after initial infection, respectively. Their data suggest that C. burnetii DNA may persist up to 12 years in blood of patients who are eventually cured, or those suffering only from fatigue. Moreover, authors found that PCR targeting single-copy genes such as Com1 and 16S rRNA genes, wase more sensitive than that targeting the 1S1111a repetitive element.2 According to these authors, PCR targeting 1S1111a was negative in the UK cohort, presumably due to the absence of these insertion elements in the strain of C. burnetii that was the causative agent of Q fever in this patient population.
In our opinion, such a notion is conjecture, since C. burnetii was not isolated from any of the patients in the UK cohort. Indeed, we routinely used in our laboratory a real-time quantitative PCR assay with Taqman probe targeting the IS1111a and IS30a repetitive elements for the diagnosis and follow-up of patients with Q fever either in biopsies or in blood and sera,3,4 and we never obtained such results. From March 2004 to April 2005, we have tested blood and sera of 546 patients with strict PCR conditions4 and clear definitions of clinical cases.5,6 Using our primers and probes targeting these two genes, among the 546 patients being tested we have obtained 34 positive PCR in blood and/or sera (6.2%). Among these 34 positive patients, 22 patients had active acute Q fever, six had active Q-fever endocarditis, four had a vascular infection, one had spondylodiscitis, and one was a pregnant woman with Q fever infection, confirmed by culture of C. burnetii from placenta. All these positive samples were obtained at the time of the diagnosis, before antibiotic treatment, and all these patients were followed and none remained positive in blood after 1 month of treatment.
For these reasons, we decided to compare our real-time PCR assay for specificity and sensitivity to the three set of primers and probes published by Marmion et al. We have selected 80 blood samples from our collection, extracted the DNA again, and perform the PCR in the same conditions as those of Marmion.1 Among the 80 blood samples, 12 were from patients with active acute Q fever, 30 from patients who had recovered from acute Q fever, 5 from patients with active Q-fever endocarditis, 29 from patients who had recovered from Q fever endocarditis, and four from pregnant women. The results for the four different real-time PCR assays (Table 1) show that only blood from patients with active Q fever (acute or endocarditis) were positive (9/80 = 11.3%). Moreover, real-time PCR were positive in the same manner (the mean of cycle thresholds (Ct) values ranged from 29.76 to 37.74) when targeting the IS1111 repetitive element or the Com1 gene, whereas PCR targeting the 16S rRNA gene was positive in only three cases (two patients with an acute form and one patient with a vascular infection), with a mean Ct value of 37.58 (36.86–38.61) (p<0.05). Finally, in order to determine the sensitivity of the different primers and probes, we used four freshly isolated strains of C. burnetii and evaluated the cut-off detection for each strain by serial ten-fold dilutions in a same run of PCR. The results (Table 2) show that the two real-time PCR assays targeting the IS1111a repetitive element had a higher sensitivity (1 to 2 log more) than those targeting Com1 and 16S rRNA genes.
|
|
Thus, we believe that the results recently published by Marmion et al are doubtful, and should be interpreted cautiously. In fact, other results given in this manuscript are intriguing, especially the fact that IFA staining and culture of cells inoculated with positive-PCR samples were all negative. This is also discrepant with our current knowledge with the sensitivity of the shell vial culture assay in samples from patients with active Q fever.7–9 According to serological data given in Table 2a and b, all patients in this series except one (patient 2, who had an endocarditis) probably had past evidence of infection, and results obtained using Com1 and 16S rRNA genes are more likely false positives due to a lack of specificity, since the DNA of any bacterial species may cause contamination10,11 or because amplicons generated with the Com1 primers from the positive control had contaminated all the subsequent assays. The authors claimed that the hypothetical isolate responsible of the outbreak in Birmingham does not contain the insertion elements. This is also a speculation that is not in accordance with current knowledge, since the IS1111a gene has been found all over the world in all strains of C. burnetii.12 Moreover, the five valve extracts from patients with Q-fever endocarditis used as positive controls in this study (Table 3b) were positive either using IS1111 or Com1 primers.
In our view, the results of the PCR should be interpreted in conjunction with pertinent clinical data and the results of such conventional microbiological tests as serological tests and bacteriological culture. We believe that, due to its high sensitivity and specificity, the 1S1111a repetitive element is the best target gene for the detection of C. burnetii in patients with active C. burnetii infections. Such studies reporting a high ratio of positive PCR in asymptomatic patients, such as for Whipple disease,13 are rarely confirmed.14,15 We believe that surprising PCR results should be interpreted with much caution.
References
1. Marmion BP, Storm PA, Ayres JG, Semendric L, Mathews L, Winslow W, Turra M, Harris RJ. Long-term persistence of Coxiella burnetii after acute primary Q fever. Q J Med 2005; 98:7–20.
2. Seshadri R, Paulsen IT, Eisen JA, et al. Complete genome sequence of the Q-fever pathogen Coxiella burnetii. Proc Natl Acad Sci USA 2003; 100:5455–60.
3. Fenollar F, Fournier PE, Raoult D. Molecular detection of Coxiella burnetii in the sera of patients with Q fever endocarditis or vascular infection. J Clin Microbiol 2004; 42:4919–24.
4. Brouqui P, Rolain JM, Foucault C, Raoult D. Concomitant infection with Q fever and Plasmodium falciparum malaria in a patient returning from Comores. Am J Trop Med Hyg 2005; in press.
5. Rolain JM, Lecam C, Raoult D. Simplified serological diagnosis of endocarditis due to Coxiella burnetii and Bartonella. Clin Diag Lab Immunol 2003; 10:1147–8.
6. Tissot-Dupont H, Thirion X, Raoult D. Q fever serology: cutoff determination for microimmunofluorescence. Clin Diag Lab Immunol 1994; 1:189–96.
7. Raoult D, Vestris G, Enea M. Isolation of 16 strains of Coxiella burnetii from patients by using a sensitive centrifugation cell culture system and establishment of strains in HEL cells. J Clin Microbiol 1990; 28:2482–4.
8. Spyridaki I, Gikas A, Kofteridis D, Psaroulaki A, Tselentis Y. Q fever in the greek island of Crete : detection, isolation, and molecular identification of eight strains of Coxiella burnetii from clinical samples. J Clin Microbiol 1998; 36:2063–7.
9. Musso D, Raoult D. Coxiella burnetii blood cultures from acute and chronic Q-fever patients. J Clin Microbiol 1995; 33:3129–32.[Abstract]
10. Heininger A, Binder M, Ellinger A, Botzenhart K, Unertl K, Doring G. DNase pretreatment of master mix reagents improves the validity of universal 16S rRNA gene PCR results. J Clin Microbiol 2003; 41:1763–5.
11. Corless CE, Guiver M, Borrow R, Edwards-Jones V, Kaczmarski EB, Fox AJ. Contamination and sensitivity issues with a real-time universal 16S rRNA PCR. J Clin Microbiol 2000; 38:1747–52.
12. Glazunova O, Roux V, Freylikman O, Sekeyova S, Fournous G, Tyczka J, Tokarevich N, Kovacova E, Marrie T, Raoult D. Multispacer Sequence Typing (MST) for Characterizing Coxiella burnetii source and outbreaks. Emerg Infect Dis 2005; in press.
13. Ehrbar HU, Bauerfeind P, Dutly F, Koelz HR, Altwegg M. PCR-positive tests for tropheryma whippelii in patients without Whipple's disease. Lancet 1999; 353:2214.[CrossRef][ISI][Medline]
14. Fenollar F, Fournier PE, Raoult D, Gerolami R, Lepidi H, Poyart C. Quantitative detection of Tropheryma whipplei DNA by real-time PCR. J Clin Microbiol 2002; 40:1119–20.
15. Maiwald M, Von Herbay A, Persing DH, Mitchell PP, Abdelmalek MF, Thorvilson JN, Fredricks DN, Relman DA. Tropheryma whippelii DNA is rare in the intestinal mucosa of patients without other evidence of Whipple disease. Ann Intern Med 2001; 134:115–19.
Response
Sir,Most observers would agree that C burnetii can persist after initial infection. The unresolved questions are where, how often and in what form.
We advanced two core findings. First, that PCR assay of bone marrow aspirates collected up to five years after acute primary Q fever in Australian patients and at 12 years after infection in patients in the Birmingham Q fever outbreak detected C. burnetii DNA in, respectively, 65% and 
88% of samples. With one exception, all had Q fever antibody. Second, that PCR assay of separated PBMC from the Birmingham patients showed significant variation in the frequency of detection of coxiella DNA. Rarely in those who remained asymptomatic, commonly in those still ill at 12 years with a chronic fatigue syndrome (the post Q fever fatigue syndrome: QFS1–4).
New observations and concepts must, as part of the scientific process, expect a sceptical reception. But scepticism without relevant experimental tests is a hollow exercise (see reference 5). Proper, effective scientific due process for validation requires that several centres, independent of the group making the original report, try to repeat the new observations without preconceptions based on tangential past experience. Most importantly, the repetition should start from identical or near identical types of samples collected from comparable patients under their control, and should adhere precisely to the laboratory protocols used by the original workers.
Drs Rolain and Raoult have not followed this paradigm. Their critique is a mix of comments related partly to misapprehensions, to the use of data from patient groups we did not study and to the questioning of conclusions we did not reach. Finally they advance again6 the common, trivial explanation for unexplained or counter-intuitive PCR findings—amplicon contamination. But this time specifically suggesting contamination of the Birmingham patient samples with Com1 amplicons—adopting a formal possibility we had reviewed and dismissed in our paper.
We correct some of their misinterpretations below. We comment also on the limitations of the PCR and culture methods available to detect persistent paucibacterial infection, as exhibited by C. burnetii. Our responses should be read in conjunction with the text of the original paper.
Specific responses
We did not state that C. burnetii might persist up to 12 years in the blood of asymptomatic recovered patients or that of patients suffering only from fatigue. We examined bone-marrow aspirates and PBMC packs, not ‘blood’, (which might mean serum, plasma or whole blood). Our Table 3a on the Birmingham outbreak, shows that of group 3 (’Q fever no sequelae’) only 3/22 were positive in PBMC samples with the Com1 PCR, whereas in the same group, 11/12 bone-marrow aspirates were positive. Group 5 (‘QFS‘ in Table 3a), were not suffering only from ‘fatigue’ but from some or all of a persistent constellation of eight or more symptoms making up the post Q fever fatigue syndrome (QFS).1–4 QFS patients in Group 5 met the criteria laid down for chronic fatigue syndrome by the CDC group.7 In contrast to group 3, group 5 (‘QFS’) were significantly more often positive with Com1 PCR in their PBMC (5/8) as well as in the bone marrow (4/5). So the findings were by no means uniform between the two groups. The significance of these differences was analysed in detail in our Discussion.
Relative sensitivities of the IS1111a and Com1 PCR assays for detection of C. burnetii
We have used the Insertion Sequence PCR since 1997, following the report of Hoover et al.8 and informal discussions at the American Society of Rickettsiology meeting in 1997. However, we use various primer sets and probes devised ‘in house’. From this experience, we agree that in general it is more sensitive than a PCR directed against a target sequence in a single-copy gene such SOD9 or from recent experience targets in Com1.10 That is why we used it in the late 1990s.11 We did not state otherwise in our paper. The equal or superior performance of our IS1111a assay over our Com1PCR assays was clearly illustrated (Table 3b in the original paper). Both primer sets reacted well with the positive control (5 coxiella cells from the Q Vax vaccine, Henzerling strain) and Q-fever endocarditis valve extracts from Australian patients. In contrast, specimens from Australian QFS patients, that presumably contained fewer coxiellas, reacted differently. For example, bone-marrow extracts from the Australian QFS patients showed IS1111a 65% positive vs. Com1–25% positive. The Birmingham specimens were the exception, with Com1 positive and IS1111a negative.
In fact, Rolain and Raoult's Tables 1 and 2 show that both their ‘in house’ IS1111a PCR, and their versions of our IS1111a, Com1 and 16S rRNA gene PCR and primer sets gave essentially the same proportional number of positives and a similar distribution of ct values with a range of their clinical specimens. Note that the latter were from clinical states different to those in our study. Nevertheless the concordance is reassuring, as complete details of our DNA extraction method were not requested by Rolain and Raoult, and there were two errors in the base sequences we gave for our PCR assays as they originally appeared in the QJM (later corrected, Q J Med 2005; 98:237–8. With MGB probes, for TMRA read Non Fluorescent Quencher).
Note also that Rolain and Raoult obtained six positive (50%) out of 12 of the ‘active’ (presumably acute phase) Q fever samples with either their PCR primer sets, or their version of ours (Rolain and Raoult, Table 1). This rate is substantially lower than that (
84%) observed in Adelaide in routine examinations with sera or blood samples from patients in the acute phase of illness and before antibodies have appeared (Turra et al.12 and below). This discrepancy calls into question the sensitivity of their PCR assay and by extension, that of the shell vial culture method12 we used.
Precise details about the time from onset of illness at which the ’active’ specimens were collected are not apparent from Rolain and Raoult's letter. We take the paper by Fournier and Raoult14 as a detailed account of their current practice and results.
Comments on sensitivity of PCR assays in relation to culture/isolation detection of C. burnetii
Recent studies from the Infectious Diseases Laboratory in our Institute explored PCR assays for detection of C burnetii in serum taken shortly after onset of acute primary Q fever. The patients came from a small Q fever outbreak in the country north of Adelaide. Turra et al.12 used PCR assays with Q fever IgM antibody–negative serum samples taken from patients during the period up to 11 days after onset. C. burnetii DNA was detected with Com1 primer sets in 58% of samples, and with IS1111a primer sets in 84% of samples. The combined results were 89% positive. Rates fell sharply once antibody appeared. All patients were subsequently confirmed as Q-fever-antibody positive.
These findings are in striking contrast to those of Fournier and Raoult,14 who also used the insertion sequence primers and the Roche Light Cycler. They obtained only 16 positives (24%) out of 66 patients whose sera were taken in the period 1–14 days after onset of illness, before antibiotic treatment and while still seronegative. Sixteen (26%) of the 61 seronegative samples from the patient group were positive.
So what rate of positive isolations and therefore PCR positives might be expected in the bacteraemia during the acute phase of Q fever?
Derrick's report15 on his original group of nine patients detected C. burnetii (aka R. burneti) in all patients by guinea pig inoculation with whole blood or extract of ground clot, inoculated IP. Samples were taken during the period 3–13 days after onset while patients were still febrile. In a later paper, Derrick16 noted that during the primary fever of the acute attack ‘almost every guinea pig inoculated with blood during this phase became infected’. Interestingly, this was not the finding during later febrile relapses or after the 15th day of prolonged fever (when presumably antibody had appeared).
Nearer the present day, Nagaoka et al.17 reported isolation of coxiellas from 13 (72%) of 18 Q fever patients using acute-phase sera inoculated into A/J mice. And To et al.18 isolated C. burnetii in A/J mice inoculated with acute phase serum from all of 23 pneumonia patients ultimately verified by seroconversion to Q-fever-antibody positive. There was 100% agreement18 (Table 3) between their PCR assay results and isolation of the coxiella in mice.
The isolation rates of the coxiella from acute phase blood or serum samples in Derrick's guinea pigs and in the two Japanese studies match the high frequency of positive PCR results obtained by Turra et al.12 in Adelaide.
The reason for the very low rate of positives obtained by Fournier and Raoult14 with acute-phase sera is unclear, as the sensitivity of their PCR assay was claimed to be one DNA copy. We note however that 69/100 patients tested by the French workers came from one outbreak. One intriguing possibility is that the strain in this outbreak was IS1111a-deficient and resembled that in the Birmingham outbreak. Might re-testing the sera with Com1 primer sets be illuminating?
Our failure to isolate C. burnetii from positive bone marrow samples
We have covered this aspect in detail in the Discussion of our paper and refer the reader to it. We would add, however, that in relation to the sensitivity of the of the shell vial method13 we used, the report by Musso and Raoult19 found that of 66 blood cultures taken from 66 acute Q-fever patients before antibiotic treatment, only 11 (16–17%) were positive; again a proportion very difficult to reconcile with the much higher ‘acute phase’ isolation rates obtained by Derrick15,16 using guinea pigs and the Japanese workers17,18 using A/J mice.
Absence of C. burnetii insertion sequence in Birmingham samples
We reported what we found, and made it quite clear in our Discussion that a final resolution of the matter has to await the isolation of the Birmingham outbreak strain. Efforts to do so continue. We merely note that workers using the analogous insertion sequence in M. tuberculosis as a PCR target must have regarded it as an invariable, conserved component of the organism, until samples from a small group of Vietnamese patients20 indicated otherwise.
Amplicon contamination and the Birmingham samples
Considerable effort was put into precautions to avoid and detect amplicon or specimen cross-contamination, and these are described in the paper. However, Rolain and Raoult appear to make two concrete suggestions to support their earlier generalized assertion.6
The first—in fact echoing points already considered in our paper—is that most or all Birmingham bone marrow samples, negative in the insertion sequence PCR assay, were subsequently contaminated with Com1 amplicon. Hypothetically, from that generated from the Q Vax controls, or Australian samples, or from one solitary but genuine positive from among the Birmingham samples.
Second, that our 16S rRNA gene primer sets were non-specific and widely reactive with bacterial DNA in addition to that of C. burnetii.
Our responses to these propositions were and are as follows:
(i) Contamination was not detected in any of a large series of non-template controls in test runs; and in particular, in those processed in parallel with a re-extraction of the original Birmingham bone marrow samples by one worker and PCR assay of the extracts and non-template controls by a second worker.
(ii) The re-extracted DNA from the bone marrows positive with Com1 primer sets were then shown to be positive with the (sequence-unrelated) 16S rRNA gene primer set and probe, thus indicating the presence of a larger DNA fragment than a Com1 amplicon.
(iii) The specificity of the amplicons from the 16S rRNA gene PCR was confirmed by flourometric probe binding, and by sequencing, and demonstration of complete or near complete identity with known C. burnetii 16SrRNA gene sequences, but not with phylogenetically-related bacteria.
(iv) We did not use so called ‘Universal’ primers. Wide species homology at higher annealing temperatures than we used with the C burnetii 16S rRNA gene primer set and probe was not found with DNA from 13 species of bacteria. The collection included such species as: (e.g.) M. tuberculosis, Legionella pneumophilia, Pseudomonas spp, Acinetobacter spp, and in relation to ‘Universal‘ primer sets, E. coli.
In conclusion
Studies of persistent paucibacterial infections and their immunological consequences are important. We hope that other workers, including Drs Rolain and Raoult, will attempt to repeat our observations with bone marrow from Q fever cases taken at intervals soon and long after the primary acute Q fever. This will require PCR assays able to detect coxiella DNA in 85% of ‘acute phase’ sera from acute Q fever and in normal bone marrow seeded with 5 or fewer coxiella cells, using our DNA extraction methods (available on request).
Also, the actual sensitivity of the shell vial method for paucibacterial, as distinct from pluribacterial infections, needs to be checked with suspensions of C. burnetii of high and low virulence and of known direct bacterial count.21 The same suspensions should be titrated in SCID mice,22 chick embryo yolk sac21 and guinea pigs.
Q Fever Research Group Adelaide Australia
Department of Environmental and Occupational Medicine University of Aberdeen Aberdeen UK
References
1. Marmion BP, Shannon M, Maddocks I, Storm PA, Penttila IA. Protracted debility and fatigue after Q fever. Lancet 1996; 347:977–8.[Medline]
2. Ayres JG, Flint N, Smith EG, et al. Post infection fatigue syndrome following acute Q fever: follow up study of patients involved in the 1989 outbreak in the West Midlands. Q J Med 1998; 91:105–23.
3. Wildman MJ, Smith EG, Groves J, et al. Chronic fatigue following infection by Coxiella burnetii (Q fever):ten-year follow-up of the 1989 UK outbreak cohort. Q J Med 2002; 95:527–38.
4. Penttila IA, Harris RJ, Storm P, Haynes D, Worswick DA, Marmion BP. Cytokine dysregulation in the post-Q fever fatigue syndrome. Q J Med 1998; 91:549–60.
5. Marmion BP. Eaton Agent—Science and Scientific Acceptance: A Historical Commentary. Rev Infect Dis 1990; 12:338–53.[ISI][Medline]
6. Raoult D. Q fever: still a mysterious disease. Q J Med 2002; 95:491.
7. Fukuda K, Straus SE, Hickie I, Sharpe MC, Dobbins JC, Komoroff A. The chronic fatigue syndrome: a comprehensive approach to its definition and study. Ann Intern Med 1994; 121:953–9.
8. Hoover TA, Vodkin MH, Williams JC. A Coxiella burnetii repeated DNA element resembling a bacterial insertion. sequence. J Bacteriol 1992; 174:5540–8.
9. Heinzen RA, Frazier ME, Mallavia LP. Coxiella burnetii superoxide dismutase gene, cloning, sequencing and expression in Escherichia coli. Infect Immun 1992; 68:3814–24.
10. Zhang GQ, Nguyen SV, To H, et al. Clinical evaluation of a new PCR assay for detection of Coxiella burnetii in human serum sample. J Clin Microbiol 1998; 36:77–80.
11. Harris RJ, Storm PA, Lloyd A, Arens M, Marmion BP. Long term persistence of Coxiella burnetii in the host after primary Q fever. Epidemiol Infect 2000; 124:543–9.[CrossRef][Medline]
12. Turra M, Chang G, Whybrow D, Higgins G, Qiao M. Diagnosis of Q fever in infection by PCR on sera during a recent outbreak in rural South Australia. 4th International Conference on Rickettsiae and Rickettsial Diseases. Logrono, Spain, June 2005. Abstract P80. Ann NY Acad Sci 2005; in press.
13. Raoult D, Vestris G, Enea M. Isolation of 16 strains of Coxiella burnetii from patients using a sensitive. centrifugation cell culture system and establishment of strains in HEL cells. J Clin Microbiol 1990; 28:2482–4.
14. Fournier PE, Raoult D. Comparison of PCR and serological assays for early diagnosis of acute Q fever. J Clin Microbiol 2003; 41:5094–8.
15. Derrick EH. ‘Q‘ fever, a new fever entity: clinical features, diagnosis, and laboratory. investigation. Med J Aust 1937; ii:281–99.
16. Derrick EH. The course of infection with Coxiella burnetii. Med J Aust 1973; 1:1051–7.[ISI][Medline]
17. Nagaoka H, Akiyama M, Sugieda M, et al. Isolation of Coxiella burnetii from children with influenza-like symptoms Japan. Microbiol Immunol 1996; 40:147–51.[ISI][Medline]
18. To H, Kako NJ, Zhang GQ, Otsuioca H, Owaga M, et al. Q fever pneumonia in children in Japan. J Clin Microbiol 1996; 34:647–51.[Abstract]
19. Musso D, Raoult D. Coxiella burnetii blood cultures from acute and chronic Q fever patients. J Clin Microbiol 1995; 33:3129–32.[Abstract]
20. Yuen LK, Ross BC, Jackson KM, Dwyer B. Characterization of Mycobacterium tuberculosis strains from Vietnamese patients by Southern blot. hybridization. J Clin Microbiol 1993; 31:1615–18.
21. Ormsbee R, Peacock M, Gerloff G, Tallent G, Wire D. Limits of Rickettsial Infectivity. Infect Immun 1978; 19:239–45.
22. Andoh M, Naganawa T, Hotta A, et al. SCID Mouse Model for Lethal Q fever. Infect Immun 2003; 71:4717–32.
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||