Old and new diagnostic approaches for Q fever diagnosis: Correlation among serological (CFT, ELISA) and molecular analyses

Comparative Immunology, Microbiology and Infectious Diseases 35 (2012) 375–379

Contents lists available at SciVerse ScienceDirect

Comparative Immunology, Microbiology and Infectious Diseases journal homepage: www.elsevier.com/locate/cimid

Presented at the 6th International Meeting on Rickettsia and Rickettsial Diseases at Heraklion, Crete, Greece on June 5–7, 2011

Old and new diagnostic approaches for Q fever diagnosis: Correlation among serological (CFT, ELISA) and molecular analyses A. Natale a,∗ , G. Bucci a , K. Capello a , A. Barberio b , A. Tavella c , S. Nardelli a , S. Marangon a , L. Ceglie a a b c

Istituto Zooprofilattico Sperimentale delle Venezie, Legnaro (PD), Italy Istituto Zooprofilattico Sperimentale delle Venezie, Vicenza, Italy Istituto Zooprofilattico Sperimentale delle Venezie, Bolzano, Italy

a r t i c l e

i n f o

Article history: Received 18 July 2011 Received in revised form 1 March 2012 Accepted 2 March 2012 Keywords: Q fever Coxiella burnetii ELISA CFT Real-time PCR

a b s t r a c t The objective of this study was to evaluate the performance of the complement fixation test (CFT) with respect to ELISA for the serological diagnosis of Q fever and to assess the role of serology as a tool for the identification of the shedder status. During 2009–2010, sera from 9635 bovines and 3872 small ruminants (3057 goats and 815 sheep) were collected and analyzed with CFT and ELISA. In addition, 2256 bovine, 139 caprine and 72 ovine samples (individual and bulk tank milk samples, fetuses, vaginal swabs and placentae) were analyzed with a real-time PCR kit. The relative sensitivity (Se) and specificity (Sp) of CFT with respect to ELISA were Se 26.56% and Sp 99.71% for cattle and Se 9.96% and Sp 99.94% for small ruminants. To evaluate the correlation between serum-positive status and shedder status, the ELISA, CFT and real-time PCR results were compared. Due to the sampling method and the data storage system, the analysis of individual associations between the serological and molecular tests was possible only for some of the bovine samples. From a statistical point of view, no agreement was observed between the serological and molecular results obtained for fetus and vaginal swab samples. Slightly better agreement was observed between the serological and molecular results obtained for the individual milk samples and between the serological (at least one positive in the examined group) and molecular results for the bulk tank milk (BTM) samples. The CFT results exhibited a better correlation with the shedder status than did the ELISA results. © 2012 Elsevier Ltd. All rights reserved.

1. Introduction Q fever is a zoonotic disease that is considered emerging or re-emerging in many countries. The clinical presentation of Q fever is non-specific in most animal species. In ruminants, late abortions and stillbirths can occur but reproductive disorders are frequently the only manifestation of the disease in a herd/flock [1]. Q fever has

∗ Corresponding author at: Serology and Virology Laboratory, Istituto Zooprofilattico Sperimentale delle Venezie, Viale dell’Università 10 – 35020 Legnaro (PD), Italy. Tel.: +39 0498084354; fax: +39 0498084351. E-mail address: [email protected] (A. Natale). 0147-9571/$ – see front matter © 2012 Elsevier Ltd. All rights reserved. doi:10.1016/j.cimid.2012.03.002

only recently been included in the Community Summary Reports on Zoonoses, and a standardized definition was suggested for a clinically affected herd/flock [2]. The overall mean of all of the available data indicates a prevalence between 15% and 30% at the animal level for cattle, sheep and goats [3]. The majority of the published studies address seroprevalence rather than shedding prevalence, although the shedding prevalence is needed to estimate the risk of transmission of the infection between ruminants, between herds and from ruminants to humans [3]. Diagnosis, particularly the determination of the shedder status, is a critical and expensive process that is not yet completely standardized. There is no officially prescribed test for the serological diagnosis of Q fever. Among

376

A. Natale et al. / Comparative Immunology, Microbiology and Infectious Diseases 35 (2012) 375–379

the various techniques that can be employed, the three most often used are the indirect immunofluorescence assay (IFA), ELISA and the complement fixation test (CFT). Typically, ELISAs are preferred for practical reasons. Currently, no IFA is commercially available for ruminants. Numerous reports have shown a weak sensitivity of CFT compared with other methods [2,4–6]. At present, ELISA is therefore the recommended choice for seroprevalences studies. ELISA kits based on antigens prepared from a ruminant isolate are recommended over kits based on the Nine Mile reference strain [7], even though there are no published data on the gain in sensitivity obtained using the ruminant antigens [3]. The presence of specific anti-C. burnetii IgG antibodies provides evidence of a recent infection and of past exposure. Serological assays are suitable for the screening of herds or flocks, but interpretation at the individual animal level is not possible. Indeed, a significant proportion of animals that shed C. burnetii bacteria and even some Q fever-aborted animals are found to be seronegative [5,8–10]. It has been proposed that the interpretation of the results at the herd/flock level requires at least six ewes or goats and ten cows (with those aborted as a priority). Both serological responses and bacterial evidence are often necessary to establish the presence of the infection [2]. With regard to the direct detection of Coxiella, the presence of Coxiella burnetii can be demonstrated in various ways depending on the type of sample and the purpose of the investigation [2,11]. The ability to detect and quantify C. burnetii DNA by real-time PCR has dramatically enhanced diagnostic and research approaches. For laboratory diagnosis in the context of serial abortions and stillbirths, samples should be collected from aborted fetuses, placentae and vaginal discharge soon after abortion or parturition. If possible, vaginal swabs from the day of parturition (or taken fewer than 8 days after) should be collected to limit the number of false-negative PCR results. When difficulties in the interpretation of diagnostic results are encountered, an association with a positive serological result at the herd or flock level is useful. Milk from the tank, individual milk or colostrum samples and vaginal and fecal samples can be collected to investigate bacterial shedding [12]. However, the detection of shedders is still complicated because the shedding dynamics are not well known [2,9–11]. Currently, PCR and real-time PCR are considered the most sensitive and rapid means for the identification of shedding animals. The specificity levels of different laboratories for the detection of Coxiella DNA in different spiked matrices (PBS, placenta, milk and aborted fetuses) and of different protocols are comparable [13,14]. Regarding sensitivity, PCR tests directed against the multiple-copy target IS1111 (real-time and conventional) were found to be superior to tests detecting single-copy genes. The limits of molecular diagnosis are due to the different modes of shedding (in birth products, milk, vaginal mucus, feces and urine); the different possible shedding periods, which are specific for each animal species; and the intermittence of shedding [9,10,15]. Therefore, the PCR-based diagnosis of shedder status at the individual level is currently an

expensive process, requiring many tests to be performed on different matrices at different periods of time. The objective of this study was to investigate the performance of CFT with respect to ELISA and to assess the potential role of some specific serological protocols as a tool for the determination of shedder status. 2. Materials and methods 2.1. Sample source The routine diagnostic activity performed in our laboratory for Q fever permitted collection of the data for the present study. Most of the samples were analyzed following three types of requests: - Request to determine the most frequent causes of abortion/infertility in “problem” herds/flocks: Typically, the laboratory receives groups of 10–30 sera (sometimes samples for the whole herd/flock) with a request to run tests for a panel of possible pathogenic causes, such as Brucella, Neospora, BVDV, Chlamydia, Q fever, BHV1 and Leptospira. Sometimes sera are accompanied by vaginal swabs and BTM samples. - Request to determine the causes of abortion based as part of a regional cattle monitoring program that requires that a panel of serological analyses be performed for the aborting cow (Neospora, BVDV, Chlamydia, Q fever, BHV1, Leptospira and Brucella) and that etiological examinations be performed on the aborted fetus/placenta (bacteriological examination and molecular detection of Neospora, BVDV, Chlamydia, Q fever and Leptospira); vaginal swabs can be analyzed if the fetus is not available. - Request for specific serological/molecular investigations of herds with suspected of having Q fever. These investigations are usually performed on a panel of sera from “problem” animals together with vaginal swabs from selected animals, feces and BTM or individual milk samples. The bias in the data collection method indicates that the results of both the serological and the molecular analyses cannot be considered prevalence estimates. 2.2. Sample collection A total of 9635 bovine sera and 3872 small ruminant sera (3057 goats and 815 sheep) collected during 2009 and 2010 as part of the routine diagnostic activity described above were considered for the present study. In addition, 2256 bovine, 139 caprine and 72 ovine samples, including individual and bulk tank milk (BTM) samples, fetuses, vaginal swabs and placentae, were collected for molecular analyses. Most of BTM samples routinely received for diagnostic activity are made up 50–100 individual milk samples; a few samples are derived from smaller (10–49 subjects) or larger (101–250) groups of animals. Each BTM sample had a volume of 50 ml. Samples for serological and molecular analyses were not collected at the same time for all the subjects and, even when both types of tests were performed, not all of the

A. Natale et al. / Comparative Immunology, Microbiology and Infectious Diseases 35 (2012) 375–379

records in the database contained sufficient information to link the serological and molecular results. Therefore, the analysis of the correlation between the serological and molecular result was possible only for a subset of the samples. More precisely, these correlations were evaluated for 3027 out of the 9635 bovine sera; the results for these 3027 serum samples were compared with the results for 375 fetuses, 199 vaginal swabs, 390 individual milk samples and 120 bulk tank milk (BTM) samples. 2.3. Serological assays Sera samples were analyzed using two independent tests: - a manual CFT performed in agreement with the Manual of Diagnostic Tests and Vaccines for Terrestrial Animals [1], using a cut-off titer of 1:10 and a commercial antigen (Siemens Healthcare Diagnostic Products); - an indirect ELISA using a commercial kit (Chekit Q fever, IDEXX Laboratories), following the manufacturer’s instructions; the value-% was calculated as (ODsample-ODneg)/(ODpos-ODneg) × 100, and samples with a value-% lower than 30% were classified as negative, samples with a value-% between 30 and lower than 40 were classified as doubtful, and samples with a value-% equal or higher than 40 were classified as positive. 2.4. Molecular assays Molecular assays were performed using a real-time PCR commercial kit (ADIAVET® COX REALTIME, Adiagène). Pre-treatment and sample processing were performed according to the manufacturer’s instructions. The extraction of Coxiella DNA from different matrices was performed according to the manufacturer’s instructions (QIAamp DNA mini kit, Qiagen), as follows: - From milk: 400 ␮l of individual milk or BTM was transferred to a 1.5 ml microtube after vortexing. - From feces: 1 g of feces was placed in a 10 ml sterile tube, 5 ml of sterile water was added, and the sample was vortexed for 30 s. The samples was the centrifuged at 3000 × g for 2 min, and 150 ␮l of the supernatant was transferred to a 1.5 ml microtube. - From vaginal swabs: 1 ml of sterile water was added to the vaginal swab, the sample was vortexed for 30 s, and 200 ␮l of the supernatant was transferred to a 1.5 ml microtube. - From tissues: 2–3 g of organ tissue was cut from the sample and then disrupted with a tissue lyser. PBS was added to dilute the sample 1:10 (w/v). The sample was homogenized for 2 min at medium speed. The homogenized sample was then transferred into a 10 ml tube, and 200 ␮l of the sample was used for the extraction. The instrument employed for real-time PCR was a Roche LightCycler 2.0 or an Applied Biosystems 7900 HT Fast RealTime System. The diagnostic sensitivity of the real-time PCR analysis of the BTM samples was determined to be sufficient

377

Table 1 Serological results in cattle and small ruminants. Cattle

ELISA CFT

Small ruminants

+

−

%

+

−

%

1630 456

8005 9179

16.92 4.73

241 26

3631 3846

6.22 0.67

to detect 1 shedder cow in a group of 250 milking cows (internal kit validation, data not shown). The individual milk samples that tested positive in the real-time PCR assay were divided into two classes depending on the cycle threshold (Ct) cut-off value of 31. For this method, Ct = 31 represents a concentration of 103 bacteria/ml (ADIAVET® COX REALTIME, validation data sheet, November 2010). Animals whose BTM samples had Ct values ≤31 were classified as high shedders, whereas animals with Ct values >31 were classified as low shedders. 2.5. Statistical analyses The two serological methods were compared with each other, and the relative sensitivity (Se) and specificity (Sp) of CFT were calculated using ELISA as the reference test. From the whole database described above, a statistical comparison between the serological and molecular results was possible only for subjects for whom both serological tests and at least one molecular test were performed. The data for 3027 out of the 9635 bovine sera were compared with the data for 375 fetuses, 199 vaginal swabs, 390 individual milk samples and 120 bulk tank milk (BTM) samples. To evaluate the agreement between the serological results and shedder status, Cohen’s Kappa with a 95% confidence interval was calculated for ELISA and CFT vs real-time PCR. The Landis and Koch [16] guidelines were used for the interpretation of the results. Using the BTM PCR results, the correlations between high and low shedders and the CFT results were assessed using Fisher’s exact test. 3. Results 3.1. Serology The serological results obtained by means of ELISA and CFT are shown in Table 1. These results make it clear that the number of positive results was higher for ELISA than for CFT (16.92% vs 4.74% for cattle; 6.22% vs 0.67% for small ruminants). The percentage of positive results must not be taken as a prevalence estimate because of the bias in the data collection system, with collection being based on clinical suspicion of disease and on abortion/infertility problems. The comparison of the serological results and the relative sensitivity (Se) and specificity (Sp) of CFT, using ELISA as the reference test, are shown in Table 2. CFT has a very high Sp for cattle and for small ruminants (99.71% and 99.94%, respectively) but a low Se (26.56% and 9.96%, respectively). The agreement (K) between ELISA and CFT was 0.17 (poor) for small ruminants and 0.37 (fair) for cattle.

378

A. Natale et al. / Comparative Immunology, Microbiology and Infectious Diseases 35 (2012) 375–379

Table 2 CFT performances respect to ELISA. CFT vs ELISA performances

Sensitivity % (95% CI)

Specificity % (95% CI)

Cattle (N = 9635) Small ruminants (N = 3872)

26.56 (24.43–28.78) 9.96 (6.49–14.50)

99.71 (99.56–99.81) 99.94 (99.80–100)

Table 3 Molecular analyses results (samples collected during routine diagnostic activity). Sample material

PCR+

Placenta/fetus Vaginal swab Individual milk Bulk tank milk (BTM)

Table 5 Correlation among serological (CFT and ELISA) and molecular results.

15 (4.0%) 14 (7.0%) 64 (16.4%) 53 (44.2%)

375 199 390 120

a

The molecular analysis results are shown in Table 3. The table shows the total number of examined samples for each different matrix and the percentage of positive samples. As indicated for Table 1, these percentages must not be considered prevalence estimates. 3.3. Molecular vs serological results The correlation between the serological (CFT and ELISA) and molecular results is shown in Tables 4 and 5. In Table 4, the PCR results (divided by matrices) can be compared with the results of the serological analyses performed with ELISA and CFT for the same subjects at the same time of sampling. For BTM, the correlation with the serological results was considered positive if a positive BTM test corresponded to at least one serum-positive subject in the linked examined group. In Table 5 the same data are shown as a statistical correlation (Cohen’s Kappa and relative confidence interval), revealing that there was no agreement between the Table 4 Correlation among serological (CFT and ELISA) and molecular results. CFT

Serum

Real time PCR Placenta/fetus + − Total Vaginal swab + − Total Individual milk + − Total BTM + − Total

+

−

+

−

7 78

8 277

2 25

13 335

370 7 36

375 7 144

3 10

194 40 100

11 175

0.07 (−0.01;0.16) 0.15 (0.01;0.30) 0.21 (0.12;0.30) 0.39 (0.23;0.55)

0.05 (−0.08;0.17) 0.16 (−0.05;0.38) 0.36 (0.23;0.48) 0.54 (0.38;0.69)

ELISA doubtful samples were not considered.

serological results and the molecular results obtained for fetuses (0.07 for ELISA, 0.05 for CFT) and vaginal swabs (0.15 for ELISA, 0.16 for CFT). Higher levels of agreement were observed between the serological and molecular results for the individual milk samples (0.21 for ELISA, 0.36 for CFT) and between the serological (at least one positive in the examined group) and molecular results for the BTM samples (0.39 for ELISA and 0.54 for CFT). In Table 6, the CFT serological status (positive or negative) is compared with the real-time PCR positive status for individual milk samples, and the molecular results are divided into two categories: Ct ≤ 31 and Ct > 31, corresponding, respectively, to high and low milk shedders. High shedders cows (Ct ≤ 31) are mainly classified as CFT positive (p = 0.024). 4. Discussion The assessment of shedder status is a critical part of Q fever diagnosis because of the different modes of shedding and the discontinuity in shedding. The repeated molecular tests on different types of samples needed to accurately determine shedder status are expensive. The use of serology, a more economical tool, for shedder identification, especially for those subjects with epidemiological relevance (high shedders, chronic shedders), was not successfully attempted previously. ELISA is a sensitive technique that is easy to perform and standardize, but it has some limitations in monitoring activity: the shedder status is highly correlated with ELISA positivity, especially when the samples tested by ELISA present a high S/P value

199 24 220

27 27

384 42 26

CFT

Serum Real time PCR Placenta/fetus Vaginal swab Individual milk BTM

3.2. Molecular analyses

ELISAa

ELISAa

Kappa (95%CI)

Total PCR samples

37 299

Table 6 High and low shedders in individual milk (cows) compared to the serological CFT status.

390 11 41

120

Serology correlated to BTM: see Section 2. a ELISA doubtful samples were not considered.

35 9

18 58 120

CFT − + Total a

Ct ≤ 31

Ct > 31

Total

6 (33.33%) 12 (66.67%) 18 (100%)

30 (66.67%) 15 (33.33%) 45 (100%)

36a (57.14%) 27 (42.86%) 63 (100%)

For one sample the Ct value was not available.

A. Natale et al. / Comparative Immunology, Microbiology and Infectious Diseases 35 (2012) 375–379

[9]; however, the converse, that ELISA positivity is correlated with shedder status, is not true because the ELISA status stays positive for long time after seroconversion, even in the absence of active shedding [2]. Furthermore, serum-negative shedders have been identified [5,8–10,17]; the negative serum tests are currently considered to be the result of a lack of sensitivity of the test, possibly due to the animals being in the early part of the seroconversion phase or to an antigenic difference between the strain used for the test and the strain infecting the animal [7]. The aim of the present study was to investigate the potentialities of ELISA and CFT, alone or in combination, as possible tools for the determination of shedder status. The study results confirm the low sensitivity of CFT [2], especially for small ruminants, but a very high specificity was observed. The fair (cattle) or poor (small ruminants) agreement between CFT and ELISA is due simply to the low sensitivity of CFT (Tables 1 and 2). With regard to the correlation between the serological and molecular results (Tables 4 and 5), no statistically significant agreement was observed between the individual serological status (for both ELISA and CFT) and the results of the molecular analyses when performed on placenta/fetuses/vaginal swabs. A possible explanation for this lack of correlation may be the sampling method that is currently applied: in the case of abortion, following the regional standard operating diagnostic procedure, the collection of both serological (serum from the aborting cow) and molecular samples (placenta, fetal organs or vaginal swab) is performed at the same time. Therefore, it is possible that seroconversion has not yet occurred in the cow at the time of abortion, and that another serological test should performed at a later time, at least for cows exhibiting molecular Coxiella positivity. The agreement between the serological and molecular results was slightly better when individual milk and BTM samples are considered, and the CFT results are better correlated with the shedder status than the ELISA results, suggesting that the presence of one or more CFT-positive animals within a herd is correlated with the active circulation of Coxiella. Serology is not a completely reliable screening test for the detection of shedders within a herd, but the serological results are fairly/moderately correlated with individual and herd shedding in milk. With regard to individual milk shedding (Table 6), high-shedder cows (Ct ≤ 31) are mainly classified as CFT positive (p = 0.024), as previously observed [18]. Further studies are needed to investigate the possible correlation between CFT positivity and chronic shedder status because there have been previous reports of repeated positive PCR results in parallel with persistent CFT-positive status [19].

379

Acknowledgment This study was partially funded by the Italian Ministry of Health (RC IZSVe 04/08). References [1] Rousset E, Sidi-Boumedine K, Thiery R. Q fever. In: OIE terrestrial manual; 2010 [on-line edition, chapter 2.1.12], http://www.oie.int/ fileadmin/Home/eng/Health standards/tahm/2.01.12 Q-FEVER.pdf. [2] Sidi-Boumedine K, Rousset E, Henning K, Ziller M, Niemczuck K, Roest HIJ, Thiéry R. Development of harmonised schemes for the monitoring and reporting of Q fever in animals in the European Union. Question No. EFSA-Q-2009-00511. [3] Guatteo R, Seegers, Taurel AF, Joly A, Beaudeau F. Prevalence of Coxiella burnetii infection in domestic ruminants: a critical review. Vet Microbiol 2011;149:1–16. [4] Kittelberger R, Mars J, Wibberley G, Sting R, Henning K, Horner GW, et al. Comparison of the Q-fever complement fixation test and two commercial enzyme-linked immunosorbent assays for the detection of serum antibodies against Coxiella burnetii (Q-fever) in ruminants: recommendations for use of serological tests on imported animals in New Zealand. N Z Vet J 2009;57:262–8. [5] Rousset E, Durand B, Berri M, Dufour P, Prigent M, Russo P, et al. Comparative diagnostic potential of three serological tests for abortive Q fever in goat herds. Vet Microbiol 2007;124:286–97. [6] Roest HIJ, Buijs RM, Döpfer D, Bölske G, Christoffersen AB, Frangoulidis D, et al. Comparison of serological assays for the detection of antibodies against Coxiella burnetii in serum of ruminants. In: Communication at the 4th annual scientific meeting Med-Vet-Net. 11–14 June 2008. [7] Rodolakis A. Q fever, state of the art, epidemiology, diagnosis and prophylaxis. Small Ruminants Res 2006;62:121–4. [8] Arricau-Bouvery N, Rodolakis A. Is Q fever an emerging or reemerging zoonosis? Vet Res 2005;36(3):327–49. [9] Guatteo R, Beaudeau F, Joly A, Seegers H. Coxiella burnetii shedding by dairy cows. Vet Res 2007;38(6):849–60. [10] Rousset E, Berri M, Durand B, Dufour P, Prigent M, Delcroix T, et al. Coxiella burnetii shedding routes and antibody response after outbreaks of Q fever-induced abortion in dairy goat herds. Appl Environ Microbiol 2009;75:428–33. [11] Samuel JE, Hendrix LR. Laboratory maintenance of Coxiella burnetii. Curr Protoc Microbiol 2009 [chapter 6: Unit 6C.1]. [12] Kim SG, Kim EH, Lafferty CJ, Dubovi E. Coxiella burnetii in bulk tank milk samples, United States. Emerg Infect Dis 2005;11(4):619–21. [13] Duquesne V, Sidi-Boumedine K, Prigent M, TylewskaWierzbanowska S, Chmielewski T, Krogfelt K, et al. A multicenter PCR-ring trial for C. burnetii detection in veterinary clinical samples: an approach to standardisation of methods. In: 4th Annual scientific meeting Med-Vet-Net. 2008. [14] Jones R.Communication at the 14th international symposium for the world association of veterinary laboratory diagnosticians (WAVLD). 17–20 June 2009. [15] Rodolakis A, Berri M, Héchard C, Caudron C, Souriau A, Bodier B, et al. Comparison of Coxiella burnetii shedding in milk of dairy bovine, caprine, and ovine herds. J Dairy Sci 2007;90:5352–60. [16] Landis JR, Koch GG. The measurement of observer agreement for categorical data. Biometrics 1977;33(1):159–74. [17] Guatteo R, Beaudeau F, Berri M, Rodolakis A, Joly A, Seegers H. Shedding routes of Coxiella burnetii in dairy cows: implications for detection and control. Vet Res 2006;37:827–33. [18] Bucci G, Ceglie L, Nardelli S, Gagliazzo L, Barberio A, de Mateo M, et al. Approccio pratico al controllo della Febbre Q nella bovina da latte. Buiatria. J Italian Assoc Buiatrics 2010;5(1):25–9. [19] Natale A, Busani L, Comin A, De Rui SL, Buffon S, Nardelli S, et al. First report of bovine Q-fever in north-eastern Italy: preliminary results. Clin Microbiol Infect 2009;15(2):144–5.