PURPOSE: To determine the accuracy of magnetic resonance (MR) imaging
in evaluating the invasive cervical carcinoma response to concurrent c
hemotherapy and radiation therapy. MATERIALS AND METHODS: MR imaging w
as performed before and after concurrent chemotherapy and radiation th
erapy in 18 patients with locally invasive cervical carcinoma. Surgery
followed neoadjuvant therapy in all patients. The presence of a lesio
n, signal intensity, zonal anatomy integrity, vaginal and parametrial
invasion, and lymph node enlargement was determined. Posttreatment MR
and histopathologic findings were correlated. RESULTS: Fourteen patien
ts had histopathologic confirmation of MR findings: Twelve had true-ne
gative and two had true-positive findings. (Two had microscopic neopla
stic foci beyond the spatial resolution of MR images; these foci do no
t change Surgical treatment planning and probably do not influence pro
gnosis. Therefore, these two patients were considered to have complete
response.) Four patients had false-positive findings; the hyperintens
e lesion on posttreatment MR images was due to a tunnel cluster patter
n (focal hyperplasia of the endocervical glands with inflammation) in
three patients and necrosis in one patient, without any evidence of ne
oplastic tissue. Thirty-three of 36 parametrial halves and 67 of 72 va
ginal fornices were correctly interpreted on posttreatment images. Inv
olvement of three parametrial halves and five fornices was overestimat
ed at MR, because edema or inflammation was not distinguishable from t
umor. CONCLUSION: MR imaging is 78% accurate in evaluation of tumor re
sponse; in 22% of patients, however, benign conditions were not distin
guishable from tumor.