Presentation Authors: Edwin Jonathan Aslim, Terence Yu Xi Law*, Kenneth Chen, Lui Shiong Lee, Henry Sun Sien Ho, Weber Kam On Lau, John Shyi Peng Yuen, Christopher Wai Sam Cheng, Nye-Thane Ngo, Yan Mee Law, Kae Jack Tay, Singapore, Singapore
Introduction: Accurate imaging is crucial in focal therapy for prostate cancer. There is limited data for the prediction of cancer size using multiparametric MRI (mp-MRI). We evaluate the size correlation between mp-MRI detected and whole-mount histology confirmed prostate cancer, and the impact of high-grade cancer on the size prediction for focal therapy.
Methods: This IRB approved study (CIRB 2017/2651) enrolled all radical prostatectomies operated between 1 January 2015 to 31 July 2017. The inclusion criteria were pre-operative mp-MRI performed in our institution, either pre-biopsy or >1 year from biopsy, and available whole-mount histology. Those with prior prostate cancer treatments were excluded. Each lesion was outlined on histology, with the high-grade (HG)) components (Gleason 4 and 5) colour-coded differently. Lesions on imaging were matched to histology (entire lesions and HG components), and measured at the same axial plane taking the mean of 2 readings. Spearmanâ€™s rank correlation was used to compare mp-MRI and histology. Specimen shrinkage correction used the maximum transverse prostate diameter as a reference. Bland-Altman plots were used to assess size discrepancies between imaging and histology, and evaluate the adequacy of mpMRI in measuring entire lesions or HG components. Per lesion analysis was performed, with statistical significance defined as p < 0.05.
Results: There were 54 cases, comprising 162 lesions on either mp-MRI and histology combined. The sensitivity of mp-MRI to detect cancer was 81% (101/125), with a false positive rate of 27% (37/138). Mean prostate shrinkage was 9% (range -10 to 28). Taking only true positives, mean lesion area was 105.9mm2 (SE 9.4, 95%CI: 87.2 to 124.7) on mp-MRI, and 103.9mm2 (SE 11.3, 95%CI: 81.5 to 126.4) on histology after shrinkage adjustment. The mean HG lesion area was 50.0mm2 (SE 9.6, 95%CI: 31.0 to 69.1). Comparing imaging with histology, rho=0.72 (95%CI: 0.61 to 0.80, p < 0.0001), and for imaging with HG cancers, rho=0.53 (95%CI: 0.38 to 0.66, p < 0.0001). On Bland-Altman test, the mean area differences was 1.98mm2 (95%CI: -10.8 to 14.8, p=0.76), with a sweet-spot range of 50 to 150mm2(8 to 14mm in diameter), where size discrepancies were less than 100mm2 (11mm in diameter). For cancer lesions within this range, a 5.5mm treatment margin covers 100% of high-grade and 98% of all cancer lesions.
Conclusions: There is good correlation between lesion size on mp-MRI and histology. The sweet-spot may represent the ideal size range for focal ablation, with a safety margin of 5.5mm.