Characterizing the learning curve of MRI-US fusion prostate biopsies
Introduction and Objective- MRI-US fusion prostate biopsies are becoming a common procedure to diagnose prostate cancer. Fusion biopsies are a complex procedure and thus surgical expertise is vital. There is paucity of information regarding the learning curve for fusion biopsies. We aim to study the amount of experience needed to be both accurate and time-efficient in this procedure
Patients and methods- We prospectively collected data on all MRI-US fusion prostate biopsies performed between 4/2014-8/2017. We used two parameters to define the learning curve. The first was efficiency measured by time from end of anesthesia to end of procedure. The second was accuracy measured by cancer detection rate. We analyzed data in blocks of 10 biopsies utilizing a time-series analysis. The end of the learning curve was defined as the graph reaching a plateau. We preformed separate analysis for transrectal and transperineal biopsies.
Results- We completed 779 fusion biopsies (523 transrectal and 256 transperineal). Patients median age was 64 (IQR 60-72) and median PSA 6.33 (IQR 3.67-10.05). Prostate cancer was diagnosed in 358 (46%).
Efficiency – Procedure time decreased from 45 minutes in the first 10 transrectal fusion biopsies to 15 minutes after 100 biopsies and remained stable ( p<0.0001). Time decreased from 55 minute in the first ten transperineal biopsies to 18 minutes after 120 biopsies (p<0.0001).
Accuracy- In transrectal fusion biopsies detection rate for PIRADS 3 lesions increased from 35% to 50% after 100 biopsies. For PIRADS 4 detection rate increased from 50% to 70 % after 100 biopsies, and for PIRADS 5 detection increased from 80% to 90% after 70 biopsies.
In transperineal fusion biopsies detection rate increased from 40% to 55% after 70 cases` for PIRADS 4 from 50% to 80% after 50 cases and for PIRADS 5 increased from 80 to 95%.
Conclusions- AS far as we know this is the first study to systematically test the learning curve of fusion biopsies. We demonstrate that about 100 cases are needed to become proficient.