PURPOSE: To analyze the value of magnetic resonance (MR) imaging after
radiation therapy for cancer of the cervix. MATERIALS AND METHODS: Ei
ghty-nine MR images were retrospectively studied in 69 patients aged 4
6.3 years +/- 11.5. MR findings of tumor recurrence and irradiation ch
anges were correlated with time after radiation therapy; paracentral r
adiation dose (dose to point A); and in patients with pretreatment ima
ges, primary tumor size and stage. RESULTS: Overall accuracy of MR in
diagnosis of tumor recurrence was 78% (positive predictive value, 65%;
negative predictive value, 97%). In MR examinations less than 6 month
s after the beginning of radiation therapy, accuracy (69%) and specifi
city (46%) were significantly lower than in examinations more than 6 m
onths later (88%, P = .0032; 81%, P = .0166, respectively). Comparison
of pre- and posttreatment MR findings and knowledge of stage or initi
al tumor size did not affect MR results. CONCLUSION: Overall, diagnosi
s was best with unenhanced T2-weighted images, but in patients with ad
nexal or pelvic sidewall recurrence and in patients with treatment com
plications (eg, fistula formation), contrast enhancement did help.