Endorectal MRI and MR spectroscopic imaging of prostate cancer: developing selection criteria for MR-guided focal therapy

Abstract

Purpose: To investigate criteria that can identify dominant treatable prostate cancer foci with high certainty at endorectal magnetic resonance imaging (MRI) and MR spectroscopic (MRS) imaging, and thus facilitate selection of patients who are radiological candidates for MR-guided focal therapy.

Materials and methods: We retrospectively identified 88 patients with biopsy-proven prostate cancer who underwent endorectal MRI and MRS imaging prior to radical prostatectomy with creation of histopathological tumor maps. Two independent readers noted the largest tumor foci at MRI, if visible, and the volume of concordant abnormal tissue at MRS imaging, if present. A logistic random intercept model was used to determine the association between clinical and MR findings and correct identification of treatable (over 0.5 cm3) dominant intraprostatic tumor foci.

Results: Readers 1 and 2 identified dominant tumor foci in 50 (57%) and 58 (65%) of 88 patients; 42 (84%) and 48 (83%) of these were dominant treatable lesions at histopathology, respectively. Within the statistical model, the volume of concordant spectroscopic abnormality was the only factor that predicted correct identification of a dominant treatable lesion on T2-weighted images (odds ratio=1.75; 95% confidence interval=1.08 to 2.82; P value=0.02). In particular, all visible lesions on T2-weighted imaging associated with at least 0.54 cm3 of concordant spectroscopic abnormality were correctly identified dominant treatable tumor foci.

Conclusion: Patients with dominant intraprostatic tumor foci seen on T2-weighted MRI and associated with at least 0.54 cm3 of concordant MRS imaging abnormality may be radiological candidates for MR-guided focal therapy.

Keywords: MR imaging; MR spectroscopic imaging; focal therapy; high intensity focused ultrasound; prostate cancer.

Figure 1

Distribution by size of prostate cancer foci (n = 300 foci) detected at histopathological review of radical prostatectomy specimens (n = 88 cases).

Figure 2A

Photomontage showing an axial T2-weighted image with an overlaid grid (left side of montage) and the spectral array corresponding to the overlaid grid (right side of montage) in a 68-year-old man with a PSA of 8.2 and Gleason 7 prostate cancer on transrectal ultrasound guided biopsy. A focus of low T2 signal intensity (large arrow on left) corresponds to a cluster of voxels with elevated choline peaks (small arrows on right) and is consistent with tumor.

Figure 2B

Corresponding section from a histopathological tumor map prepared after step-section review of the radical retropubic prostatectomy specimen from the patient featured in Figure 1A shows a focus of tumor in the right mid-gland that corresponds to the abnormality seen in Figure 1A (a second smaller tumor focus anteriorly is not visible on MR imaging). Handwritten annotations refer to tumor size (cm) and Gleason score. The posterior tumor was associated with extracapsular extension, as indicated by tumor extending beyond the margin of the prostate.

Figure 3

The area under ROC curve was 0.86 (95% CI = 0.77 to 0.92), suggesting a good performance of the model. At a threshold of 0.54cm³ of spectroscopic abnormality, the model has a sensitivity of 59.8% and specificity of 100%, correctly classifying 67% of cases.