Elsevier

European Journal of Radiology

Volume 95, October 2017, Pages 307-313
European Journal of Radiology

Research article
Sub-differentiating equivocal PI-RADS-3 lesions in multiparametric magnetic resonance imaging of the prostate to improve cancer detection

https://doi.org/10.1016/j.ejrad.2017.08.017Get rights and content

Highlights

  • Objective T2WI and DWI helped improve cancer yield for equivocal (PI-RADS 3) lesions.

  • Biopsy recommendation improved PPV to 32% for GS 7–10 and 61% for any cancer.

  • No-biopsy recommended had equivalent NPV to a negative MRI (PI-RADS 1–2) at 92%.

  • The criteria-based score system could potentially avoid 48% of biopsies in the cohort.

Abstract

Purpose

To evaluate sub-differentiation of PI-RADS-3 prostate lesions using pre-defined T2- and diffusion-weighted (DWI) MRI criteria, to aid the biopsy decision process.

Methods

143 patients with PIRADS-3 index lesions on MRI underwent targeted transperineal-MR/US fusion biopsy. Radiologists with 2 and 7-years experience performed blinded retrospective second-reads using set criteria and assigned biopsy recommendations. Inter-reader agreement, Gleason score (GS), positive (PPV) predictive values (±95% confidence intervals) were calculated and compared by Fisher’s exact test with Bonferroni-Hom correction.

Results

43% (61/143) patients had GS 6–10 and 21% (30/143) GS  3 + 4 cancer. For peripheral zone lesions, significant differences in any cancer detection were found for shape (0.26 ± 0.13 geographical vs. 0.69 ± 0.23 rounded; p = 0.0055) and ADC (mild 0.21 ± 0.12 vs marked 0.81 ± 0.19; p = 0.0001). For transition zone, significantly increased cancer detection was shown for location (anterior 0.63 ± 0.15 vs. mid/posterior 0.31 ± 0.14; p = 0.0048), border (pseudo-capsule 0.32 ± 0.14 vs. ill-defined 0.61 ± 0.15; p = 0.0092), and ADC (mild 0.35 ± 0.12 vs marked restriction 0.68 ± 0.17; p = 0.0057). Biopsy recommendations had 62% inter-reader agreement (89/143). Experienced reader PPVs were significantly higher for any cancer with “biopsy-recommended” 0.61 ± 0.11 vs. “no biopsy” 0.21 ± 0.10 (p = 0.0001), and for GS 7–10 cancers: 0.32 ± 0.10 vs. 0.08 ± 0.07, respectively (p = 0.0003).

Conclusion

Identification of certain objective imaging criteria as well as a subjective biopsy recommendation from an experienced radiologist can help to increase the predictive value of equivocal prostate lesions and inform the decision making process of whether or not to biopsy.

Introduction

Multiparametric prostate MRI (mpMRI) has become established in the diagnostic pathway of men with prostate cancer [1], [2], [3] and is now increasingly used in the pre-biopsy setting to allow selection of men with significant cancer for biopsy, while avoiding biopsy and unnecessary treatment in men without an MRI lesion [4], [5].

The recently updated Prostate Imaging-Reporting and Data System (PI-RADS) guidelines are aimed at standardizing MRI acquisition and interpretation using a 5-point scoring system [6], [7]. However, when MRI is being used to guide the clinical decision making process either in the context of a previous negative biopsy, or in biopsy naïve patients, this 5-point scale has to be translated into a binary decision of whether to biopsy or not. A PI-RADS score of 1–2 is considered a “negative” MRI, and has a >90% negative predictive value (NPV) for significant disease [8], [9], thus biopsy can be reasonably avoided. Conversely, a PI-RADS 4–5 lesion is of high probability and targeted biopsy is warranted. An intermediate PI-RADS 3 lesion, however, straddles this decision making process, and biopsy in this case is under debate [10], [11], [12]. The overall detection of cancer in indeterminate lesions has been shown to vary from 6.5% to 60% for any cancer and 4.1% to 21% for significant cancer [10], [13], [14], [15], [16]. This needs to be considered in the context of a “miss rate” of around 10% for a PIRADS score of 1–2. Importantly, detection rates have been shown to be higher in the peripheral zone [14] and as high as 40% in the context of a second-biopsy population [15], suggesting some PI-RADS 3 lesions deserve biopsy, whereas others could be safely deferred. Informing management of such lesions is particularly relevant given the reported prevalence of indeterminate of 20.5–26.3% using earlier Likert-based systems [10], [16], [17], [18] is predicted to increase with a switch to using the PI-RADS-version 2 reporting system [19].

The aim of this study therefore was to evaluate if equivocal PI-RADS 3 lesions on mpMRI of the prostate can be further differentiated using pre-defined T2- and diffusion-weighted imaging (DWI) criteria, in order to aid in the biopsy decision process.

Section snippets

Study population

This single-institution retrospective study was part of an evaluation of transperineal prostate biopsies with the need for informed consent for data analysis waived by the local ethics committee. From January 2013 to April 2016, 155 consecutive patients with a dominant (index) lesion considered to be equivocal on mpMRI (PI-RADS 3) underwent transperineal prostate biopsies at our tertiary center. 4 patients were excluded due to hip replacements, 8 patients were excluded as their scans were

Results

59% (85/143) of index lesions were called in the transition zone and 41% (58/143) in the peripheral zone. At transperineal biopsy, 43% (61/143) patients had a GS 6–10 prostate cancer in the target area, 21% (30/143) patients had a GS  3 + 4 cancer, and 6% (9/143) a GS  4 + 3 cancer, resulting in an overall positive predictive value of 0.43 for any cancer and 0.21 for significant GS 7–10 cancer. The clinical characteristics are shown in Table 2. The PPV for each objective imaging criterion for either

Discussion

Our study shows that re-evaluation of equivocal MRI lesions by an experienced uroradiologist, using only topographical, T2WI, and DWI and assessing set imaging criteria, improved diagnostic accuracy. Adding a subjective recommendation of whether or not to biopsy a lesion improved the cancer yield to 32% for GS 7–10 cancers and justified the “deserves biopsy” recommendation. Conversely the NPV of 0.92 for “avoid biopsy” lesions is equivalent to the NPV of a negative MRI (PI-RADS 1–2), which is

Conclusions

Identification of certain objective imaging criteria as well as a subjective biopsy recommendation from an experienced radiologist can help to increase the predictive value of equivocal prostate lesions and inform the decision making process of whether or not to biopsy.

Conflicts of interest

None.

Acknowledgements

The authors acknowledge research support from Cancer Research UK, National Institute of Health Research Cambridge Biomedical Research Centre, Cancer Research UK and the Engineering and Physical Sciences Research Council Imaging Centre in Cambridge and Manchester and the Cambridge Experimental Cancer Medicine Centre.

References (34)

  • National Collaborating Centre for Cancer (UK)

    Prostate Cancer: Diagnosis and Treatment

    (2014)
  • P.R. Carroll et al.

    NCCN guidelines insights: prostate cancer early detection, version 2.2016

    J. Natl. Compr. Cancer Netw.

    (2016)
  • H.U. Ahmed et al.

    Is it time to consider a role for MRI before Prostate biopsy

    Nat. Rev. Clin. Oncol.

    (2009)
  • H.U. Ahmed et al.

    The PROMIS study: a paired-cohort, blinded confirmatory study evaluating the accuracy of multi-parametric MRI and TRUS biopsy in men with an elevated PSA

    J. Clin. Oncol.

    (2016)
  • N.K. Yerram et al.

    Low suspicion lesions on multiparametric magnetic resonance imaging predict for the absence of high-risk prostate cancer

    BJU Int.

    (2012)
  • J.S. Wysock et al.

    Predictive value of negative 3T multiparametric prostate MRI on 12 core biopsy results

    BJU Int.

    (2016)
  • N.L. Hansen et al.

    The influence of prostate-specific antigen density on positive and negative predictive values of multiparametric magnetic resonance imaging to detect Gleason score 7-10 prostate cancer in a repeat biopsy setting

    BJU Int.

    (2016)
  • Cited by (40)

    • Natural History of Patients with Prostate MRI Likert 1-3 and Development of RosCaP: a Multivariate Risk Score for Clinically Significant Cancer

      2023, Clinical Genitourinary Cancer
      Citation Excerpt :

      The level of evidence on the long term outcomes and safety of this practice is nevertheless limited,3 and meta-analysis show that 9%-10% patients with a negative mpMRI are at risk of harbouring csCaP.1,2 The optimal decision-making strategy for equivocal lesions (PI-RADS/Likert score 3) is less clear, and csCaP diagnosis rates in patients may range significantly (3.4% - 46.5%).7-9 Indeed, the National Institute for Health and Care Excellence (NICE) highlights that more evidence is required on longer term outcomes and the most suitable surveillance protocol for patients with a Likert score on mpMRI < 3 in whom no initial biopsy is performed.4

    • Temporal changes of PIRADS scoring by radiologists and correlation to radical prostatectomy pathological outcomes

      2022, Prostate International
      Citation Excerpt :

      PIRADS 3 lesions suffer most from the experience of radiologists, possibly as the lesion is to be indeterminate or equivocal for csPCa, whereas PIRADS 4-5 lesions are proven to accurately detect significant and adverse pathology, ranging from 40 to 55% and 80 to 90% for PIRADS 4 and 5 lesions, respectively20,21. As such, the prevalence of PIRADS 3 lesions vary significantly among studies, from one in three (32%) to one in five (22%),22 and experts are more likely to confidently reclassify lesions categorized as PIRADS 3 as 1–2 or 4–523, leading some to caution immediate biopsy in PIRADS 324. In this study, we found that within the same group of radiologists, low risk tumors (GG ≤ 2) were nearly twice more likely to be interpreted as PIRADS 1–2 in cohorts separated by two years.

    • Prostate Imaging Reporting and Data System 3 Category Cases at Multiparametric Magnetic Resonance for Prostate Cancer: A Systematic Review and Meta-analysis

      2020, European Urology Focus
      Citation Excerpt :

      One study also used a Likert scoring system [10]. Multiparametric MRI was performed at 3 T in 18 studies [9,15,16,18–20,23,26,27,29–34,36,38,39], at 1.5 T in four studies [21,25,35,37], and at 3 or 1.5 T in six studies [3,10,17,22,24,28]. Multiparametric MRI sequences were T2W, DWI, and DCE in 22 studies [3,9,10,15–18,20,22,24–36,39]; four studies used only T2W and DWI sequences [19,35,37,38]; and one study also added MR spectroscopy [21] (Table 1).

    View all citing articles on Scopus
    View full text