DWI is the dominant sequence in multiparametric prostate MRI for evaluating the index lesion in the PZ lesion and the secondary sequence for the TZ.
The background signal suppression progressively increases with increasing b-values on DWI.
The normal PZ is not fully suppressed and continues to show mildly increased signal on DWI because of T2-shine through effects,
resulting in obscure some tumors on DWI of b values ,
higher b-value DWI may improve the contrast to noise ratio between cancerous and normal tissues via removing T2-shine through effects.
However,
they has lower SNR and may develop image distortions that diminish anatomic clarity and visualization of normal landmarks of prostate,
resulting in decreased diagnostic performance value for detecting the prostate cancer [2].
Therefore,
a number of studies have been performed to determine the optimal b value on prostate DWI [2-10].
Several studies have reported as comparing b = 1000 and 2000 s/mm2 at 3T DWI,
DWI of b = 2000 s/mm2 was superior to DWI of b = 1000 s/mm2 for prostate cancer detection by quantitative methods [3,
4,
9].
Wang et al [5] reported DWI using b = 1500 s/mm2 was more effective than DWI using b = 1000 or 2000 s/mm2.
Rosenkrantz et al [2] demonstrated that using a b = 1500 to 2500 s/mm2 was optimal for prostate cancer detection after comparing different computed b values (1000 to 5000 s/mm2).
Until now,
many studies have focused on the value of cancer detection on high b-value DWI.
No studies have assessed the differences for PI-RADs v2 scoring between different b-value on DWI.
In daily practice,
radiologists use the visual assessment and score for the index lesion in the prostate,
which may be somewhat subjective [12].
Therefore,
assessment of the variability of PI-RADs v2 scoring between conventional and higher b-value DWI is very important,
especially for different readers.
Our results demonstrated that DWI1000 at 3T demonstrated excellent agreement with DWI1500 in both experienced and less-experienced readers for PI-RADS v2 scoring in evaluating prostate cancer.
Furthermore,
the sensitivity and accuracy of DWI1000 and DWI1500 in the detection of PZ CSCs were similar for the both readers.
Interestingly,
recent studies [3,
6] have reported that diagnostic performance of ADC maps derived from different b values showed no difference regarding detection of prostate cancer,
even though DWI using b = 1500 or 2000 s/mm2 had higher sensitivity.
These findings suggest that somewhat limitations for detecting prostate cancer on DWI1000 as compared with DWI using b = 1500 or 2000 s/mm2 may be overcome using the ADC maps of b= 1000 s/mm2.
In our study,
inter-reader agreements of DWI between b = 1000 and b = 1500 s/mm2 were all excellent and were greater than those of the previous reports for PI-RADS v2 [13,
14].
Rosenkrantz et al [13] reported a κ value of 0.535–0.619 for b = 1500 s/mm2 and Muller et al [14] reported a κ value of 0.46 for b = 2000 s/mm2 images.
A potential reason was explained by methodological differences.
Rosenkrantz et al [13] assessed inter-reader variability for six experienced radiologists from different academic centers.
Our study included experienced and less-experienced readers from a single center.
Additionally,
potential differences of image quality on DWI due to different MR vendors or different acquisition parameters might affect the results for PI-RADS v2 scoring.
In our study,
the ROC curve analysis demonstrated that the AUC of the experienced reader was similar with that of recent studies [4,
8,
9,
17],
although different study designs and heterogeneous study population have been used.
This study has several limitations.
First,
this was a retrospective study.
A larger,
prospective study is warranted.
Second,
the numbers of CSCs were relatively small resulting in lower specificity for predicting CSCs.
Third,
regarding inter-reader variability of assessing PI-RADS v2,
only two readers from a single center assessed the index lesion.
Further large study including more readers or institutions should be performed.
In conclusion,
DWI1000 at 3T demonstrated excellent agreement with DWI1500 in both readers for PI-RADS version 2 scoring in evaluating prostate cancer.
Therefore,
we believed that conventional DWI using b-value of 1000 s/mm2 may be sufficient for the evaluation of prostate cancer.