[Transition to personalized software mpMR/US fusion prostate biopsy: a stratified retrospective analysis of 322 patients with development of a CHAID model].

INTRODUCTION MRI-targeted biopsy is the standard method for prostate cancer diagnosis; however, it is typically supplemented with perilesional or systematic biopsy to minimize the risk of false-negative results. Nevertheless, several studies have demonstrated the feasibility of performing MRI-targeted biopsy alone in carefully selected patients.

Objective

To identify factors allowing MRI-targeted biopsy to be performed alone without compromising the detection of clinically significant prostate cancer (csPCa).

Materials And Methods

A retrospective analysis was conducted in patients who underwent transperineal software mpMR/US fusion and saturation biopsy. Inclusion criteria were PSA more or equal 2 ng/ml and/or the presence of a suspicious lesion on digital rectal examination (DRE) and/or transrectal ultrasound (TRUS), as well as a PI-RADS v2.1 score more or equal 3. Factors included in the stratified analysis were: type of biopsy, PI-RADS score, suspicious lesion on TRUS or DRE, PSA level, PSA density, and lesion size. To formulate clinical rules, a CHAID decision tree algorithm was developed with the following variables: detection of csPCa on software mpMR/US fusion biopsy, PI-RADS score, PSA density, lesion size, age, and type of biopsy. CsPCa was defined as ISUP grade more or equal 2.

Results

A total of 322 patients were included (median age 63 years, PSA 6.7 ng/ml, prostate volume 48 cm). Primary biopsy was performed in 241 patients (74.8%) and repeat biopsy in 81 patients (25.2%). The detection rate of csPCa did not differ significantly between software mpMR/US fusion and combined biopsy (23.6% vs 30.4%; p=0.051). The additional diagnostic value of saturation biopsy for csPCa was 6.8% (range 0-28.6%). Stratified analysis showed no added value of saturation biopsy in primary biopsy patients with PI-RADS 5 lesions, PI-RADS 3-4 lesions with adverse factors, and in repeat biopsy patients with PI-RADS 5 lesions and adverse factors. The developed CHAID model demonstrated an accuracy of 80.2%, sensitivity of 100%, and specificity of 60.7%. According to the model, saturation biopsy was considered unnecessary in patients with PI-RADS 5 lesions with adverse factors, as well as in patients with PI-RADS less or equal 4 lesions combined with PSA density less or equal 0.20 ng/ml/cm and lesion size less or equal 10-12 mm. Application of the CHAID algorithm could reduce the number of saturation biopsies by 60.5% with a missed csPCa risk of less or equal 3.3%.

Conclusion

The presented results indicate that combined biopsy can be safely replaced by only MRI-targeted approaches in carefully selected patients without a substantial risk of missing csPCa, thereby creating a basis for the implementation of personalized prostate biopsy strategies.