Login to Access Video or Poster Abstract: MP43-15
Sources of Funding: None

Introduction

1) Evaluate the association between detection of perineural invasion (PNI) on a surveillance biopsy and grade reclassification (GR or Gleason score (GS) ≥ 7) during active surveillance (AS); 2) assess whether the presence of PNI alone should exclude men from AS.

Methods

The study cohort was 1559 men enrolled in an AS registry from 1995 to 2016, following diagnosis of favorable risk prostate cancer. The outcome of interest was GR on surveillance biopsy. A logistic model was used to evaluate the relationship between PNI and GR on next surveillance biopsy. The predictive accuracy of PNI for GR was compared with a recently published multivariable GR risk prediction tool that included period of diagnosis, age, cancer volume, risk status, PSA density, and number of previous biopsies without GR. Predictive performance was evaluated using concordance statistics (AUC), calibration and decision curve analysis

Results

Of 1559 men with a median follow-up of 4 years (IQR 2-7 years), 156 (10%) had PNI detected on at least one surveillance biopsy. Three hundred and nineteen men (20%) had GR. PNI was detected in a significantly higher proportion of men with GR than men without GR (19.7% vs. 7.5%, p = <0.0001). PNI was significantly associated with GR, OR = 2.91 (95% CI, 2.06 - 4.10, p = < 0.001; AUC = 0.58) in a univariable analysis. The multivariable prediction tool had an AUC of 0.75 (calibration slope = 0.84) for predicting GR with no significant gain in predictive accuracy by incorporating PNI (AUC of 0.77, calibration slope = 0.86). A decision curve analysis showed a positive net benefit of using multivariable risk prediction tools over PNI alone for future biopsy outcome predictions.

Conclusions

: Shared decisions on AS selection and monitoring should be based on individualized risk assessments from multivariable risk prediction tools rather than presence or absence of a single risk factor.

Funding

None