Journal of Infection (1998) 36, 273-277
Detection of Mycobacterium tuberculosis in Serum Samples
Using the Polymerase Chain Reaction
M. van Staden *~, E. van der Ryst 2, E. M. Attwood ~, M. L. Hendricks ~, G. Joubert 3 and
D. J. V. WeicW
~Department of lnternaI Medicine (G73), University of the Orange Free State, P.O. Box 339, Bloemfontein, 9300,
South Africa, 2Department of Virology, University of the Orange Free State, Bloemfontein, South Africa,
and 3Department of Biostatisties, University of the Orange Free State, BIoemfontein, South Africa
Conventional methods for the diagnosis of Mycobacterium tuberculosis infections have serious limitations. To
determine whether amplification of M. tuberculosis DNA in serum by the polymerase chain reaction (PCR) might
be a useful additional diagnostic tool, we tested 329 clinical specimens using primers specific for the IS6110 insertion
sequence of the M. tuberculosis complex. The samples consisted of 30 serum samples from healthy controls, 114
serum samples from patients with diagnoses other than tuberculosis (including immunosuppressive disorders), 59
samples from patients with a clinical picture suggestive of tuberculosis, and 78 serum samples from patients with
proven M. tuberculosis infection. Both serum, and representative samples from anatomical regions suspected of
being infected, were collected from a further 48 patients for comparison with serum PCR. Serum PCR identified 72/
78 (92%; 95% confidence interval CI: 84%-97%) patients with proven tuberculosis, and 49/59 (83%; 95% CI:
71%-92%) patients with suspected tuberculosis. In the group of patients with other diagnoses, 30/114 (26%; 95%
Ch 18%-34%) tested positive, while none of the specimens from the healthy control group were positive (95% Ch
0%-12%). Serum PCR results also compared favourably with other clinical specimens obtained from the same
patient. Serum PCR can, therefore, be a useful additional technique for the early diagnosis of M. tuberculosis
infection, but it does not necessarily indicate active infection.
Introduction
Tuberculosis is a major cause of morbidity and mortality
in developing countries. Moreover, the human im-
munodeficiency virus (HIV) epidemic poses an additional
risk factor for the progression of inactive (or dormant)
tuberculosis to active disease. One of the principal reasons
for the failure of tuberculosis control programmes in
developing countries is the inability to detect infectious
cases early enough, a Conventional methods for the
laboratory diagnosis of tuberculosis have been very use-
ful, but have serious limitations. 2 The Mantoux test is
not very specific, while acid fast staining of sputum
smears is not very sensitive. Culture of Mycobacterium
tuberculosis is a more sensitive technique, but can take
several weeks. The polymerase chain reaction (PCtl) is a
rapid diagnostic technique which has the potential to
overcome some of these limitations such as a lack of
sensitivity and specificity. 3 Several studies on the use of
PCR for the detection and identification of M. tuberculosis
in clinical samples have been published. 4-1°
In this study we aimed to evaluate the use of PCR of
* Address all correspondenceto: M. van Staden.
Accepted for publication 28 July 1997.
serum samples for the diagnosis of tuberculosis, especially
extrapulmonary infection. In these cases it is often difficult
to obtain material from the site of infection for microscopic
analysis, culture or PCR, while serum is easily obtainable
using minimal facilities. A rapid blood-based diagnostic
test for mycobacteria will also be of considerable benefit
in circumstances where tuberculosis presents as an occult
disseminated infection. Furthermore, in patients with the
acquired immunodeficiency syndrome (AIDS) or other
immunosuppressive disorders, giehl-Neelsen (ZN) stain-
ing of sputum smears is less sensitive. I1 13 We report here
the results of PCR on serum samples from 329 patients,
and compare the results with their clinical diagnoses.
Patients and Methods
Clinical specimens
Clotted peripheral blood samples were obtained from: (a)
patients with proven tuberculosis infection either by ZN
staining and/or culture positivity on Lowenstein-Jensen
(LJ) slope, typical chest X-ray with positive Mantoux skin
test, or increased adenosine deaminase (ADA) levels in
body fluids (n= 78); and (b) from patients suspected of
having tuberculosis based on clinical criteria (n=59)
0163-4453/98/030273 +05 $12.00/0 © 1998 The British InfectionSociety