Background: Whipple disease is a chronic, multisystem disorder associa
ted with infection with Tropheryma whippelii, an organism that has not
yet been grown on artificial media. In some cases, the diagnosis of W
hipple disease is uncertain if it is based on histology alone. Althoug
h antibiotic regimens of various durations have been used, the disease
recurs in about one third of cases. No test for cure is available. Ob
jective: To develop a test that is more sensitive and specific than hi
stologic examination to diagnose Whipple disease and monitor the effec
ts of antibiotic therapy. Design: Retrospective, laboratory-based eval
uations of stored tissue specimens. Patients: 30 patients with clinica
lly diagnosed, histologically confirmed Whipple disease and 8 patients
in whom Whipple disease was clinically suspected but who did not have
definitive histologic evidence. Measurements: Pretreatment and post-t
reatment biopsy specimens of the small bowel and lymph node were teste
d by polymerase chain reaction for the presence of T. whippeli DNA. Re
sults: Results on PCR were positive in 29 of the 30 specimens from pat
ients with histologically confirmed disease (sensitivity, 96.6%; speci
ficity, 100%) and in 7 of the 8 specimens from patients in whom diseas
e was clinically suspected. Small-bowel biopsy specimens were obtained
after treatment from 17 patients (median duration of follow-up, 119 m
onths); specimens from 12 of these patients had positive results on PC
R. When these cases were correlated with therapeutic outcome, it was f
ound that 7 of the 12 patients had clinical relapse during subsequent
follow-up or had never responded to treatment (positive predictive val
ue, 58% [95% CI, 28% to 85%]). In contrast, none of the 5 patients who
se post-treatment biopsy specimens had negative results on PCR had rel
apse (negative predictive value, 100% [CI, 48% to 100%]; P = 0.044). N
o correlation was found between post-treatment histology and clinical
outcome (P > 0.2). Conclusions: Polymerase chain reaction is highly se
nsitive and specific when used to confirm the diagnosis of Whipple dis
ease, to identify inconclusive and suspicious cases, and to monitor re
sponse to therapy. A negative result on PCR may predict a low likeliho
od of clinical relapse; a positive test result that remains positive d
espite therapy may be associated with a poor clinical outcome. Histopa
thologic evaluation of post-treatment specimens does not predict clini
cal cure or relapse.