Sequence capture-PCR improves detection of mycobacterial DNA in clinical specimens

The rapid identification of mycobacterial DNA in clinical samples by PCR can be useful in the diagnosis of tuberculous infections, but several large studies have found that the sensitivity of this approach is not better than that of culture, In order to improve the sensitivity of detection of mycobacterial DNA in clinical specimens from patients with paucibacillary forms of tuberculosis, we have developed a procedure permitting the specific capture of mycobacterial DNA in crude samples prior to amplification, thereby concentrating the target sequences and removing irrelevant DNA and other potential inhibitors of the amplification reaction (sequence capture-PCR). By using this approach to capture and amplify two different sequences specific for organisms of the Mycobacterium tuberculosis complex (IS6110 and the direct repeat region), it was possible to detect as little as one genome of mycobacterial DNA in samples containing up to 750 mu g of total DNA, representing a 10- to 100-fold increase in sensitivity compared with that obtained by purifying total DNA prior to amplification, Detection of the IS6110 sequence in pleural fluid samples from patients with tuberculous pleurisy by sequence capture-PCR gave positive results in 13 of 17 cases, including 3 of 3 culture-positive samples and 10 of 14 culture-negative samples, In contrast, when total DNA was purified from these samples by adsorption to a silica matrix prior to amplification, only the three culture-positive samples were positive by PCR, The sensitivity of detection of the direct repeat sequence in these samples by sequence capture-PCR was similar to that oil IS6110 and, in addition, permitted immediate typing of the strains from some patients, We conclude that sequence capture-PCR improves the sensitivity of detection of mycobacterial DNA in paucibacillary samples. This approach should be useful in detecting rare target sequences from organisms implicated in other pathologic processes.