At present, the rapid diagnosis of pulmonary tuberculosis rests with m
icroscopy. However, this technique is insensitive and many cases of pu
lmonary tuberculosis cannot be initially confirmed, Nucleic acid ampli
fication techniques are extremely sensitive, but when they are applied
to tuberculosis diagnosis, they have given variable results. Investig
ators at six centers in Europe compared a standardized PCR system (Amp
licor; Roche) against conventional culture methods. Defined clinical i
nformation was collected. Discrepant samples were retested, and inhibi
tion assays and backup amplification with a separate primer pair were
performed. Mycobacterium tuberculosis complex organisms were recovered
from 654 (9.1%) of 7,194 samples and 293 (7.8%) of 3,738 patients. Fo
ur hundred fifty-two of the M. tuberculosis isolates from 204 patients
were smear positive and culture positive. Among the culture-positive
specimens, PCR had a sensitivity of 91.4% for smear-positive specimens
and 60.9% for smear-negative specimens, with a specificity of 96.1%.
Analysis of 254 PCR-positive, culture-negative specimens with discrepa
nt results revealed that 130 were from patients with recently diagnose
d tuberculosis and 94 represented a presumed laboratory error. Similar
analysis of 118 PCR-negative, culture-positive specimens demonstrated
that 27 discrepancies were due to presumed uneven aliquot distributio
n and 11 were due to presumed laboratory error; PCR inhibitors were de
tected in 8 specimens. Amplicor enables laboratories with little previ
ous experience with nucleic acid amplification to perform PCR. Disease
in more than 60% of the patients with tuberculosis with smear-negativ
e, culture-positive specimens can be diagnosed at the time of admissio
n, and potentially all patients with smear-positive specimens can imme
diately be confirmed as being infected with M. tuberculosis, leading t
o improved clinical management.