INTRODUCTION: Active surveillance (AS) is the preferred management approach for patients with low and select favorable intermediate risk prostate cancer. Patients on AS, however, face a high risk of cancer progression. Currently, AS protocols typically include periodic prostate multiparametric MRI (mpMRI) and biopsies. There is a need for additional tools to risk stratify patients to minimize the burden of prostate biopsies and better select patients for AS. A pixel by pixel Habitat Risk Score (HRS) classification,¹ an automatic method for identification of suspicious lesions on mpMRI was applied in a prospective AS clinical trial.

MATERIAL AND METHODS: HRS analyzes the ADC and DCE-MRI and assigns a pixel-by-pixel score from 1 to 10 in increasing fashion with tumor aggressiveness. HRS was derived in correlation with histopathological Grade Group (GG) from radical prostatectomy.^1-3

We evaluated HRS in 210 AS patients in “The Miami MAST Trial“ (ClinicalTrials.gov: NCT02242773) enrolled between 2014 and 2020. The study protocol included an mpMRI and confirmatory biopsy within 18 months of the diagnostic biopsy and every subsequent year for 36 more months, unless there was a histopathologic progression, defined as (i) more than 4 cores with any grade cancer, (ii) more than 2 cores with GG2 cancer, (iii) any single core with GG3 or higher cancer, (iv) a GG1 at diagnosis upgraded to GG2. mpMRI acquisition was consistent with the recommendations for PI-RADSv2.⁴ Suspicious-for-cancer regions were outlined in Dynacad (InVivo, Gainsville, FL) by a radiologist. MRI/Ultrasound (MRI-US) fusion biopsies were carried out in UroNav (InVivo, Gainsville, FL). Targeted and standard template biopsies were collected from each patient. We reviewed patients that progressed at the 1^st, 2^nd or 3^rd year of AS. Patients were classified as “True Progressors” if: (i) the lesion(s) that resulted with a biopsy upgrade were targeted at all previous biopsy sessions; and there was evidence that (ii) the lesion(s) grew; and/or (iii) a new lesion appeared; and/or (iv) there was GG upgrade. The remaining patients were classified as “Early Progressors” when the lesion(s) that resulted in a biopsy upgrade were NOT targeted until the last session (“PIRADS miss”) or the lesion(s) were annotated for targeting in Dynacad, but it was concluded with high degree of certainty that the biopsy needle did not hit the targets (“Needle miss”).

HRS was applied and compared to: (i) radiology targets; (ii) biopsy histopathology. The volumes of HRS6-8 were recorded.

RESULTS: 35 patients progressed at 1^st (n = 18); 2^nd (n = 13) or 3^rd year (n=4), resulting in 91 mpMRI datasets. In addition, the diagnostic scans were available for 19 patients. Thus a total of 110 mpMRI studies were analyzed: 46 were acquired on 3T Discovery MR750 (GE), 48-3T Skyra, 13-3T TrioTim and 3-1.5T Symphony (Siemens). HRS was not feasible in 17 cases due to missing sequence in the mpMRI acquisition or suboptimal ADC or DCE imaging.

In Figure 1 the imaging studies from a patient, classified as a “True progressor” are presented. Representative axial slices on T2-weighted MRI, ADC and early enhancing DCE through the lesion at Diagnostic, Confirmatory and 1st Yr imaging/biopsies are shown. The lesion, indicated with yellow arrows, was targeted at all time points, but it resulted in an upgrade (GG4) in the 1^st yr. HRS indicates both the lesion’s growth and increased aggressiveness scores (HRS6 to HRS9). Using similar analysis, we identified 11 (31%) “True progressors” in the dataset. In Figure 2, the median of HRS6 to HRS9 volumes in “True Progressors” are shown at Confirmatory, 1^st, and 2^nd year, indicating the tumor evolution both in volume and aggressiveness.

The remaining patients were classified as “Early Progressors”. In Figure 3, an example of a patient in the “PIRADS miss” group is shown. The anterior lesion was not targeted at Confirmatory biopsy and the 1^st Yr biopsy resulted in GG4 cancer. Of note, HRS clearly identified the lesion as dominant on both exams. Using similar analysis, we identified 14 (40%) “PIRADS miss” in the dataset. Assuming that HRS reduces the time of surveillance for these patients at a minimum of one year, the cumulative reduction in surveillance would be 14 years. And finally, 10 (29%) patients were identified as “Early Progressors”/Needle miss. The overall classification of the patients is shown in Figure 4.

DISCUSSION: HRS as an important tool to facilitate the targeting of the dominant lesion and quantification of tumor habitat progression. The study illustrates the power of HRS as being much more quantitative and objective, as compared to PIRADS. Currently, HRS is integrated in MRI-US fusion biopsy platform for prospective evaluation.

Acknowledgements

The research was supported by the National Cancer Institute of the National Institutes of Health under Award Number P30CA240139, RO1CA189295, R01CA190105, and U01CA239141.