Expert readers may reduce overcalling of prostate cancer and minimize negative biopsies.

Hansen NL, Koo BC, Gallagher FA, Warren AY, Doble A, Gnanapragasam V, Bratt O, Kastner C, Barrett T. Eur Radiol. 2016 Oct 24. [Epub ahead of print] PubMed PMID: 27778089.

Short summary 

Prostate MRI reports from dedicated uroradiologists at a tertiary center were compared to reports from 28 different radiologists from 7 regional referring centers. Radiologist's agreement and accuracy were compared. 

Conclusion

• Agreement between the two groups of radiologists was poor.
• MRIs were more often called negative in subspecialist reads (41 % vs 20 %).
• Second readings by subspecialist uroradiologists significantly improved NPV and PPV.

Patient cohort

Retrospective assessment of 158 prostate MRIs of men referred to a tertiary center either for repeat biopsy in patients under active surveillance or patients with prior negative biopsy and ongoing suspicion of prostate cancer.

The New and the Good

• A recent study on the influence of reader experience on diagnostic performance.
• Interesting subanalysis of performance among three patient groups (negative vs. equivocal vs. positive).

Limitations

• Referring radiologist's expertise level unknown.
• Lesions deemed positive in initial reading but negative in "expert's reading" were not targeted in biopsy.
• Experts were unblinded to initial report.
• Selection bias due to only including patients who were referred to the tertiary centre.
• The minimum requirements regarding acquisition parameters (i.e. in-plane resolution, etc.) for the MRI study to be included were not explicitly stated. 

Possible consequences for clinical practice

Experience, expertise (and elaborated MRI protocols) are important for prostate MRI. It may be discussed whether some sort of certification (gained through courses or minimal numbers of prostate MRIs read over the course of a year) should become recommended for interpreting MRI of the prostate.

Possible implications for PIRADS v3

Apart from recommendations and minimal requirements regarding technical parameters, a further version of PIRADS may suggest a minimal number of prostate MRI / year recommended to read for a radiologist or a number of prostate MRIs / year to perform for a center in order to safely perform and interpret prostate MRI studies...

Future study ideas derived from this paper

A subanalysis on whether the technical parameters of the MRIs referred for second reading correlated with disagreement between radiologist groups or with diagnostic performance would have been interesting and could have validated the proposed minimal requirements recommended bi PIRADS. 

Personal comment

I agree with the proposition of the authors that training of radiologists should be done and to then provide a "competency certification". It remains questionable who would provide the training and under what conditions.