Screening of post-mortem tissue donors for Coxiella burnetii infection after large outbreaks of Q fever in The Netherlands

Donor serum samples

Serum samples were included from all Dutch post-mortem tissue donors between October 2010 and June 2011, from whom at least one tissue was approved at initial assessment. All serum samples were obtained within 24 hours post-mortem, unless there was haemodilution or insufficient quality, in which cases (pre-transfusion) ante-mortem samples were used for testing. Standard donor selection criteria were applied during the study. Regarding Q fever the following donors were excluded: donors with proven acute Q fever; donors with signs of acute Q fever (such as flu-like symptoms, pneumonia without a clear cause or identified pathogen or hepatitis); and donors with a high risk of acute or chronic Q fever, such as donors with occupational hazard (i.e. farmers and veterinarians).

Donor tissue samples

Tissue samples from all donors who tested positive for IgG antibodies against phase 2 of C. burnetii were collected and stored for detection of Coxiella DNA (provided that permission for transplantation-related research had been given).

Detection of Coxiella antibodies

Serum samples from the post-mortem tissue donors were tested for IgG antibodies against phase 2 of C. burnetii using the CE-marked Serion enzyme immunoassay (EIA) test (Serion, Clindia Benelux, Leusden, the Netherlands). The cut-off values for EIA (borderline) positivity were determined according to the manufacturer’s instructions. Borderline reactive samples were considered positive. Confirmation of positive samples was performed using an immunofluorescence assay (IFA) for IgG antibodies against phase 1 and 2 of C. burnetii (Focus Diagnostics, Cypress, CA) following instructions of the manufacturer, using a cutoff dilution of 1:32. An IgG phase 1 antibody titer ≥ 1/1024 was considered suspect for chronic Q fever [11–13]. Both serologic tests have, to our knowledge, never been used for post-mortem blood samples. Cadaveric specimens are of lower quality than regular blood samples, showing more false-positive and false-negative reactions. Validation of the EIA and the IFA was therefore performed prior to the start of the study.

The Serion anti-Coxiella phase 2 IgG EIA was validated by testing 45 randomly selected donated post-mortem samples. One of the 45 samples tested positive; this result was confirmed by IFA. The average EIA signal (measured as the optical density signal to cutoff ratio) was not different between negative post-mortem samples and negative samples from 92 healthy blood donors from the Northwestern part of the country (OD/CO 0.107 ± 0.112 versus 0.104 ± 0.008, p = 0.87). The clinical specificity in post-mortem samples could not be determined, since no samples from post-mortem tissue donors historically proven to be reactive were available.

Effects of the cadaveric nature of the samples on the sensitivity were assessed by spiking samples. A panel of 20 samples from post-mortem tissue donors, showing various degrees of haemolysis, was spiked with 1/8 volume of serum from a healthy blood donor who tested positive for both anti-Coxiella phase 1 and 2 IgG [14]. The average increase in OD caused by the spiking was slightly higher for post-mortem samples than for sera from 18 healthy blood donors (ΔOD = 0.556 ± 0.075 versus 0.486 ± 0.080, respectively; p = 0.009). The signal increase was not significantly different in hemolytic samples compared to normal post-mortem samples, suggesting the EIA test is sufficiently robust for measuring post-mortem samples of relatively low quality.

A small-scale validation of the IFA for measuring post-mortem serum samples was performed by measuring 27 random post-mortem samples. All samples tested negative and no high background fluorescence was observed. Further validation was done by spiking samples suspected to be false-positive in the EIA (see results section). 18 samples with a varying degree of EIA reactivity that were not confirmed by IFA were spiked with 1/8 volume of serum from a healthy blood donor positive for phase 1 and 2 IgG. The fluorescence intensity after spiking did not significantly differ between EIA-negative samples, borderline-positive samples and IEA-positive samples, suggesting that the negative IFA results (of the unspiked samples) were not caused by signal inhibition.

Detection of Coxiella DNA

Donated tissues from donors positive for C. burnetii antibodies were tested for the presence of C. burnetii DNA by PCR. Corneas were frozen at −20°C until analysis as soon as serology results were known. For cornea donors, a wedge of cornea, including scleral rim, was used for DNA extraction and PCR testing. For skin donors, skin samples, frozen in an 85% glycerol buffer, and skin storage solution were tested. For heart valve donors, aortic and pulmonary valves, aortic and pulmonary artery wall and myocardium were tested. If available, samples were used from cryopreserved grafts, stored for transplantation. If no grafts for transplantation were available, sampling was done from formaldehyde-fixed remnants of the heart that are routinely stored for histological examination. These samples were embedded in paraffin and cut into ribbons before DNA extraction. For musculoskeletal tissue donors, bone marrow samples were taken at retrieval and stored at −20°C for PCR testing.

To efficiently extract DNA from tissues, proteinase K digestion was performed prior to DNA extraction. One fourth of the cornea, ±50-100 mm3 of skin tissue, heart valve tissue or bone marrow was digested with 25 μL of proteinase K (20 mg/ml; Roche Diagnostics GmbH, Mannheim, Germany) and 225 μl digestion solution (0,5% SDS, 21 mM Tris–HCl). Paraffin-embedded tissues were cut into sections. Approximately 1–1.5 cm² of sectioned tissue was put in 250 μL proteinase K digestion solution. Digestions were performed in a thermoshaker at 55°C overnight at 1400 rpm. The subsequent day DNA was extracted using a NucliSens EasyMAG extraction system (bioMérieux, Boxtel, Netherlands). Samples were processed according to manufacturer’s instructions and eluted in 60 μL elution buffer. Ten μL of DNA isolate was added to the PCR, which was performed as previously described [15]. An internal control was added to each sample before EasyMAG isolation and a real-time PCR to detect the internal control target was run parallel to the C. burnetii PCR to monitor DNA extraction and PCR inhibition [15]. For cornea samples and paraffin-embedded samples an additional PCR to detect human albumin DNA was performed to ensure sufficient input material in each PCR. Albumin PCR cycle threshold values for cornea PCR’s varied from 19 to 20.3 and for paraffin-embedded tissues from 26.4 to 28.9.

Data collection

Donor characteristics, such as age, gender, cause of death, clinical characteristics, and place of residence were recorded. Three criteria for increased geographical risk of Q fever were applied. First, living in a four-digit postal code area where at least one Q fever patient was reported in the preceding 3 months. Second, living within a five-kilometre radius of an infected farm, where C. burnetii was detected in the bulk tank milk. Third, living in a three-digit postal code area in which the Q fever incidence was higher than 20/100.000 inhabitants in any of the years 2007 to 2010. Approximately 15% of the Dutch population lives in this area, where 86.6% of the Q fever cases were reported. The data on Q fever incidence were obtained from the Dutch National Institute for Public Health and the Environment. The data on bulk tank milk positive farms were obtained from the Dutch Food and Consumer Product Safety Authority. The five-kilometer radius from infected farms to the residence of each donor was determined by measuring the distance between both postal codes.

For all donors who tested positive for C. burnetii antibodies, additional clinical, occupational, geographical and -if available- histological data were gathered, to determine the likelihood of previous infection with C. burnetii, and to establish the presence of signs of chronic Q fever. Information was gathered by reviewing charts, interviewing the general practitioner or next of kin and by reviewing autopsy results and results of histological examination of remnant hearts after heart valve donation, if performed. Factors that were considered risk factors for chronic Q fever are heart valve disease, heart valve or vascular prosthetics, aortic aneurysms and being immunocompromised [16].

Ethics

Consent for donation, including testing for transmittable infectious diseases, was obtained prior to donation from either the donor, as registered in the Dutch donor registry, or from the legal next of kin. Because of the Q fever outbreak in The Netherlands the Dutch Health Advisory board deemed additional testing for C. burnetii necessary. Since consent for testing for transmittable infectious diseases was obtained and all data were analysed anonymized no specific approval by an Ethical Committee was needed for this study according to the Code of Conduct for Health Research, as implemented by the Dutch Federation of Biomedical Scientific Societies, which is followed by the involved organisations.

Data analysis

All data were entered in a database and analysed with the statistical software program SPSS 15.0 for Windows (SPSS Inc., Chicago, IL). For statistical analysis of differences between groups, ANOVA or Chi-squared tests were used, when appropriate. P-values ≤0.05 were considered statistically significant.

Donor characteristics

Between October 2010 and June 2011, 1033 consecutive donors donated at least one approved tissue, and were tested for IgG antibodies against phase 2 of C. burnetii. The mean age of the included donors was 65.6 ± 12.3 years (range 0–85) and 664 (64.3%) of them were male. 950 Donors donated corneas (92.0%), 187 skin (18.1%), 139 heart valves (13.5%), 13 thoracic aortas (1.3%), and 86 musculoskeletal tissues (8.3%).

Thirty-three donors (3.2%) lived in a postal code area, where in the previous three months a Q fever patient had been reported. Eighty donors (7.7%) lived within a five-kilometre radius of a previously contaminated farm, and 147 (14.3%) donors lived in the high incidence Q fever area.

Serological testing

Fifty of the 1033 donors (4.8%) tested positive for IgG antibodies against phase 2 of C. burnetii by EIA. In 31 donors (3.0%) the reactivity was confirmed by IFA. Twenty-four samples were reactive for both phase 1 and 2, while seven samples were only reactive for phase 2 IgG. One sample showed very high background fluorescence with a few discernible bacteria and was scored positive. One donor (0.1%) had IgG phase 1 and 2 titers of both 1:4096 in the IFA, indicative for chronic Q fever [16]. There was a remarkable difference between the two kit lots used for this study. 12/547 (2.2%) of the donors tested (borderline) positive with the first kit lot (SBA.AG), while 38/486 (7.8%) tested positive with the second lot (SKA.CD, p < 0.001). All donors for whom IgG reactivity was not confirmed by IFA had been tested with the second lot. Usually IFA is more sensitive [17]. Thus, it is seems highly likely that the EIA results were false-positive. Spiking experiments confirmed this (see Methods section). Comparing signals from all measured samples showed elevated background signals for lot SKA.CD and increased numbers of borderline and positive samples using this lot (Figure 1).

These data suggest a problem with cross-reactivity with unknown antigens leading to false-positive reactivity. Indeed, re-testing the 19 discordant samples with a new lot that became available after the study showed ten non-reactive, eight borderline and one positive sample in this lot. Based on these results, we consider only donors who tested positive in both EIA and IFA C. burnetii seropositive.

PCR testing

Characteristics of donors who tested positive for C. burnetii phase 2 IgG

The mean age in IFA-confirmed seropositive donors was 67.0 ± 9.9 years (range 46–83 years). In the age group from 40–59 years 2.4% of the donors was seropositive, in the group from 60–79 years 3.2% and in group from 80–85 years 3.9%. 22/31 seropositive donors were male (71.0%). The donated tissues are shown in Table 1. Five donors had risk factors for chronic Q fever, namely vascular prosthesis (N = 1), heart valve disease (N = 1), artificial heart valve (N = 1), aortic dissection/aneurysm (N = 2). None of these donors with risk factors for chronic Q fever had a serological profile indicating chronic disease, and in two donors autopsy was performed, which showed no abnormalities. In four seropositive donors the remnant hearts after heart valve donation were examined, which resulted in identification of one donor with a focal pericarditis. However, PCR testing of samples of the heart of this donor did not reveal any C. burnetii DNA. The donor who had a serological profile suspect for chronic Q fever was a male donor, 68 years old, who died of an acute cardiac arrest due to a myocardial infarction. The donor did not live in a high risk area for Q fever and had no work related risk factors for Q fever, signs of past infection or risk factors for chronic infection. No autopsy was performed. He donated corneas and skin, which tested negative for C. burnetii DNA. No seropositive patients lived in an area where in the last 13 weeks an acute Q fever patient had been reported, but 6 seropositive donors lived within a five-kilometer radius of an infected farm. Eleven donors (35.5%) lived in a high incidence Q fever area.

There was no significant difference in age between groups of seropositive or seronegative donors. There was no significant correlation between gender and seropositivity; the percentage of males was 71.0% in seropositive versus 64.1% in seronegative donors, respectively. There was no increased level of seropositivity in donors living in areas where in the previous 13 weeks acute Q fever patients had been reported or in donors living within a five-kilometer radius from an infected farm. In contrast, the percentage seropositivity was higher in donors living in high incidence Q fever areas as compared to donors living elsewhere (7.5% vs. 2.3%, P = 0.002).