RESUMEN
OBJECTIVES: To assess the accuracy and interobserver variability of T2-weighted (T2W) and contrast-enhanced dynamic (CE-Dyn) magnetic resonance imaging (MRI) in predicting the results of transrectal biopsy in patients with suspected recurrent prostate cancer after external beam radiotherapy. METHODS: A total of 22 patients with increasing prostate-specific antigen levels after external beam radiotherapy for prostate cancer underwent T2W and CE-Dyn MRI of the prostate. The CE-Dyn sequence (acquisition time 30 seconds) was repeated three times after the injection of gadolinium. All patients underwent subsequent transrectal biopsy. Three independent readers interpreted the MRI scans. The MRI and biopsy results were correlated in 10 prostate sectors (the sextants of the peripheral zone, the two transitional zones, and the two seminal vesicles). RESULTS: Biopsy cores were obtained in 147 prostate sectors. Of these, 63 were positive for cancer in 19 patients. On the T2W images, the three readers interpreted as positive for cancer 15, 15, and 13 of the 19 patients showing cancer at biopsy. They interpreted as negative 3, 0, and 1 of the 3 patients showing no cancer at biopsy. On CE-Dyn images, the three readers correctly classified all the patients as positive or negative for cancer. The T2W and CE-Dyn MRI findings were concordant with biopsy results in, respectively, 81 to 95 and 107 to 117 prostate sectors (P <0.001 and P <0.01 for readers 1 and 2 and was nonsignificant for reader 3). The interobserver agreement was better for CE-Dyn images (kappa = 0.63 to 0.70) than for the T2W images (kappa = 0.18 to 0.39). The MRI-calculated tumor volumes and the mean biopsy core invasion rates were significantly correlated on the CE-Dyn images for all readers. They correlated significantly on T2W images only for one reader. CONCLUSIONS: CE-Dyn MRI depicts the intraprostatic distribution of recurrent cancer after external beam radiotherapy more accurately and with less interobserver variability than T2W MRI.
Asunto(s)
Imagen por Resonancia Magnética/métodos , Recurrencia Local de Neoplasia/patología , Próstata/patología , Neoplasias de la Próstata/patología , Anciano , Biopsia/métodos , Humanos , Masculino , Persona de Mediana Edad , Variaciones Dependientes del Observador , Neoplasias de la Próstata/radioterapia , Ultrasonografía IntervencionalRESUMEN
Our objectives were to determine time-enhancement curves of prostate cancer, peripheral zone, and adenoma at gadolinium-enhanced MR imaging, and to determine if a high-spatial/low-temporal dynamic imaging could be accurate in depicting prostate cancer, or if a higher temporal resolution (and a lower spatial resolution) should be favored. Thirty-nine patients with prostate cancer underwent MR imaging before radical prostatectomy by using T1- and T2-weighted axial images and a single-slice dynamic gadolinium-enhanced sequence (40 images; one image per 6 s; injection of 20 ml at 2 ml/s). After analysis of the pathologic specimens, four region-of-interest (ROI) cursors (cancer, peripheral zone, adenoma, and muscle) were retrospectively placed on dynamic images. Time-enhancement curves of the ROIs were obtained. The theoretical accuracy of a 30-s dynamic multislice MR sequence in depicting cancer within peripheral zone and adenoma (ROC curves) was calculated from these curves. On average, prostate cancer enhanced more and earlier than peripheral zone and adenoma, but there were great interindividual variations. For start delays ranging from 12 to 84 s, the areas under the ROC curves ranged from 0.602 to 0.698 for the depiction of cancer within adenoma and from 0.614 to 0.827 for the depiction of cancer within peripheral zone. The best results were obtained with a 36-s start delay. In conclusion, we found a 30-s scanning window which seems to allow a good depiction of cancer within peripheral zone. Because of largely overlapping enhancement patterns, cancer will probably not be depicted within adenoma by dynamic imaging, at least by using low temporal resolution.