Accuracy of Multiparametric MRI for Prostate Cancer Detection: A Meta-Analysis

This meta-analysis of the current literature found 7 studies meeting their inclusion criteria for assessment of multiparametric MRI for prostate cancer (defined as using both T2-weighted imaging and functional MR techniques DWI and DCEMRI). All studies had to include comparison with either biopsy or surgical histopathology. Studies were assessed for quality using QUADAS-2.

Overall the meta-analysis of pooled data demonstrated a specificity of 0.88 and sensitivity of 0.74 for prostate cancer detection, with variable NPV ranging from 0.65 to 0.94.

The meta-analysis is methodologically sound, but does have a limited pool of only 7 studies to assess. This provides a good overview of the current literature and provides further evidence of the role of MRI in assessment of patients with suspected or diagnosed prostate cancer.

PICO Analysis:
Population: Patients with suspected prostate cancer.
Intervention: Multiparametric prostate MRI.
Comparison: Surgical or Biopsy histopathology.
Outcome: Sensitivity & Specificity.

Reference: Pubmed Full Text

Diagnostic accuracy and patient acceptance of MRI in children with suspected appendicitis

This prospective study compares 3 strategies for imaging suspected appendicitis. Ultrasound alone, US+conditional MRI and MRI alone. The sensitivity of MRI alone and US + conditional MRI was 100%, ultrasound alone was found to be significantly lower at 76%.
The specificity was found to be comparable among the three investigated strategies; MRI alone 89%, US + conditional MRI 80% and US alone 89%.

The study on a technical basis is quite good, with patients enrolled consecutively and prospectively. Surgical diagnosis or 3mo followup including contact of all patients primary physician was used for clinical followup. While no flowchart is provided in the text documenting the different arms it appears none were lost to followup and all 'positive' cases did proceed to surgery. The study population had a relatively high prevalence of appendicitis at 56%.

Limitations are that the US was performed by radiology residents, and the surgical management is not clear for all cases which makes assessment of the applicability of these results to different practices (such as in Canada or US) somewhat challenging, and may explain the relatively low sensitivity of US alone despite relatively high subjective quality evaluation of the US studies.

PICO Analysis:
Population: pediatric patients 4-18yo (mean 12yo)
Intervention: US alone, MRI alone or US + conditional MRI
Comparison: Surgical dx or 3mo clinical followup
Outcome: Sensitivity and Specificity

Reference: Pubmed Full Text

Prostate Cancer: Can Multiparametric MR Imaging Help Identify Patients Who Are Candidates for Active Surveillance?

Retrospective study of 133 patients who underwent multiparametric MRI before radical prostatectomy.  The MRI results were compared to clinical scoring systems, with pathology (tumour<0.5cc) as the reference standard for assessing eligibility for Active Surveillance. 14/133 would have been eligible for AS based on pathologic results, and the MRI was highly accurate for predicting eligibility for AS.

D'Amico system:  sensitivity 93%, PPV 25%, and accuracy 70%.

Epstein criteria: sensitivity 64%, PPV 45%, and accuracy 88%. 

CAPRA system: sensitivity 93%, PPV 20%, and accuracy 59%.
Multiparametric MR: sensitivity 93%, PPV 57%, and accuracy of 92% (P < .005). 

This is limited by its retrospective nature, but provides further evidence of the utility of multiparametric MRI in the evaluation of prostate cancer.

PICO Analysis:
Population:

• 133 patients (mean age, 59.3 years) 
• Mean PSA 6.73 ng/mL (median, 4.39 ng/mL) 

Intervention: Multiparametric MR imaging at 3.0 T before radical prostatectomy.
Comparison: Conventional clinical assessment scores (the D'Amico, Epstein, and CAPRA scoring systems)
Outcome: Radical prostatectomy pathology specimen (gross tumour <0.5cc) as Active Surveillance eligible.

Reference: Turkbey et al., Radiology, July 2013, 268(1): 144-152.
Links: Pubmed Full Text