The number of specimens analyzed in this study was relatively small therefore the null finding could be due to chance. Previous meta-analysis showed the occurrence of C. pneumoniae
in atheromatous lesions compared to non-atheromatous vessels of controls was 59% versus 3.1%[8
]. In our study Chlamydia
spp. were not detected in any of the atheromatous carotid specimens which is the same as detection rates in a study of abdominal aortic aneurysms carried out by Lindholt et al
].Weiss et al
only detected C. pneumoniae
in one coronary atheroma [6
]. C. pneumoniae
was only detected in 4/50 (8%) coronary atherectomies from patients in Germany [7
Several measures were taken to ensure that our results were not false negatives. The sensitivity of the nPCR was determined by comparing the endpoints with titrations of C. trachomatis and C. pneumoniae of knownTCID50. The nested approach gave a 103 increase in sensitivity. This was to ensure that the lack of detection of Chlamydia spp. was not due to lack of sensitivity of the nPCR. We looked for the presence of inhibitors in the tissue specimens. An aliquot of each extracted DNA specimen was spiked with a weak dilution of C. trachomatis DNA and amplified by PCR to exclude such inhibitors. We did not demonstrate any inhibition to our assay. In retrospect it would have been useful to spike the tissue sample with whole chlamydia particles prior to extraction to determine whether chlamydia DNA was being lost at the extraction step. Further studies should look at this aspect.
The size of specimens and method of DNA tissue extraction may account for the differences in the rate of detection. The heterogeneity of distribution of the organism within the tissue may elude detection by various methods as noted by Jackson et al [4
]. Jantos et al [7
]who used a different method and Paterson et al [12
]who looked at larger samples of carotid endarterectomy and carotid and coronary arteries from autopsy also failed to detect C. pneumoniae
. In this study the Qiagen tissue DNA kit, which is specific for DNA extraction from tissue samples, was used. The cumulative tissue analyzed in this study argues against the negative findings resulting from inadequate amounts of tissue investigated. In order to ensure that adequate DNA was extracted in this study, the HLADRB gene was amplified in each specimen. Results of chlamydia PCR in each individual specimen were only considered valid when the HLADRB gene was successfully amplified and when inhibition to the PCR was excluded. Care was also taken at all stages to avoid contamination of specimens, especially at the stage of DNA tissue extraction.
The difficulty of obtaining adequate numbers of disease-free and age and sex matched controls has been a problem in previous investigations, especially in an elderly population. Taylor-Robinson suggested the best control comprises disease-free specimens adjacent to abnormal areas in vessels from the same subjects [8
]. One study showed that C. pneumoniae
was not detected by electron microscopy in adjacent tissue without disease [5
]. In our study external controls from 8 age and sex matched post mortem patients were used when we were not able to identify adjacent tissue free of atheroma on macroscopic examination in patient specimens. Although 13/41 (32%) of the controls were found retrospectively on histology to have signs of early atheroma development, Chlamydia
spp. were not detected. There is much evidence from many workers that C. pneumoniae
is present in atheromatous material including studies using electronmicroscopy, antigen detection and even culture.[13
] However sensitive techniques such as PCR should be the gold standard for detection of organisms at low levels in atheromatous material. The wide range in detection rates in the literature is therefore surprising.
Apfalter et al have recently published a multicenter comparison trial of DNA extraction methods and PCR assays for detection of C. pneumoniae
in endarterectomy specimens.[14
] Despite this study an explanation for the wide discrepancies is unclear. In the Apfalter et al
. study, the majority of methods (10/16) found all 15 specimens negative. Hence the majority consensus among the expert laboratories was that all the specimens were negative. Three methods were able to detect the low positive control (0.01 inclusions). Of these 3 most sensitive methods 2/3 found all 15 specimens negative. In contrast, of the 2 laboratories that detected C. pneumoniae
in the most clinical specimens, one does not detect the lowest positive control. This less sensitive laboratory/method (A) was the one that had the most positive results (9/15).
Unfortunately the study design employed by Apfalter et al. was not ideal for assessing specificity as only one negative control was employed. Suspiciously, the two laboratories that detect the most positives share a common extraction methodology (Boehringer High Pure PCR template preparation kit). This is a relatively intricate method with 6 spin steps. The consequent increase in manipulation steps over that associated with other extraction methods makes this method inherently potentially prone to amplicon and carry-over contamination. Working in a routine clinical laboratory, we avoid such potentially non-robust methods because of the potential associated specificity problems. The positive control data (samples C1-4) do not suggest that the Boehringer High Pure PCR template preparation kit confers a higher sensitivity as 2 methods not using this kit had better sensitivity than method A, and a further 7 labs had equivalent sensitivity to method A. The two 'best' laboratories yield 9/15 (lab A) and 5/15 (lab B) positive results. The concordance between these two laboratories is not accessible from the data presented in the paper. We are not told if the 5/15 positives in lab B are included in the 9/15 positives in lab A. In addition the positive PCR products were not sequenced. Thus there was no attempt to show if the positive products represent real positive samples or are related to amplicon or carry-over contamination from laboratory or control samples. PCR contamination is an insidious problem. PCR contamination remains a viable explanation for the constellation of results presented by Apfalter et al. It is possible that the literature in this area is biased by false positive PCR results.
This work in the current study was carried out in a diagnostic laboratory (Regional Virus Laboratory, Belfast). We routinely use nested PCR assays as diagnostic procedures. In this setting we necessarily place much emphasis on specificity, including robust extraction methodologies and minimization of potential for contamination.
A publication bias against negative findings in well-conducted studies is something to be guarded against as it compromises future systematic review. We feel that this may be a particular problem in this subject area. We feel that there is a particular role for electronic journals such as BMC in which space, in contrast to paper journals, is not a limitation. Unlimited publication space will increase the chance that papers with negative findings are published.
spp. were not detected in carotid atheroma in this group of patients despite a high prevalence of C. pneumoniae
antibodies comparable to that previously reported in the Northern Ireland population [15
], an area with a high prevalence of atherosclerotic disease. Weiss et al who only detected the organism in one coronary atheroma by PCR also found that seroprevalence for C. pneumoniae
antibodies did not correlate with the presence of C. pneumoniae
in coronary atheroma in the patients in his study [6
]. This implies that if there is any causal association between C. pneumoniae
and atherogenesis then it must be indirect or that it is a variable that is not essential for atheroma development.