Login to Access Video or Poster Abstract: MP18-06
Sources of Funding: Blue Cross Blue Shield of Michigan

Introduction

Cross-sectional imaging is performed during staging of high-risk prostate cancer to identify lymph node (LN) metastases. We investigated the performance and utilization rates of pre-robot-assisted radical prostatectomy (RARP) CT and MRI to identify LN metastases.

Methods

Using the MUSIC registry, we identified all patients undergoing RARP (3/2012 to 9/2016), grouping them by imaging prior to surgery (CT, MRI, none). Primary outcome was detection of LN metastases at RARP. In general, imaging studies were considered positive if LN>8mm in short axis were identified.

Results

Prior to 6489 RARP, 1783 patients underwent CT (27%), 282 underwent MRI (4.3%), and 4424 had no pretreatment abdominal imaging (68%). Pre-treatment factors were significantly different in these 3 populations (Table 1). For example, D&[prime]Amico high-risk patients represented 57.5%, 25.3%, and 8.3% of the CT, MRI, and no imaging cohorts, respectively (p<0.001). Predominant pattern 4 disease was present in 58%, 40%, and 23% and stage pT3/T4 cancer represented 48%, 36%, and 24% of the CT, MRI, and no imaging groups, respectively (both p<0.001). Among patients with Gleason 8-10 disease at final pathology, 30% were not imaged before RARP. Overall, 225 patients (3.5%) had pathologic LN involvement, including 0% low, 2% intermediate, and 9.4% high-risk patients. Suspicion for LN involvement was identified on 2.8% of CT (n=50) and was associated with higher Gleason score (sGS 9/10: 53%) and pT stage (pT3b/T4: 44%). Interestingly, many more patients with pN+ disease at RARP had a negative CT (n=123, 7.1%) than a positive CT (n=12, 24%), yielding a sensitivity of 8.9%, specificity of 97.7%, NPV of 93% and PPV of 24%.

Conclusions

Overall, 32% of patients (and 75% of those with high-risk cancer) underwent CT or MRI prior to RARP. Suspicion for LN metastases on CT was predictive of higher risk disease, but was a poor predictor of presence of LN metastases (positive predictive value: 24%). These data have implications for patients with and without suspicion of LN metastasis on CT. Patients with suspicious LNs might be managed as having disseminated disease and not be offered definitive local treatment. Conversely, patients with &[prime]negative&[prime] or no imaging may not receive PLND despite metastatic LNs in 2% and 9.4% of those with intermediate- and high-risk cancer at RARP.

Funding

Blue Cross Blue Shield of Michigan