This study delves into the significance of this innovative functional MRI technique. When used in conjunction with PI-RADS V2.1, APT-weighted MRI helps to distinguish between csPCa and benign prostate lesions or cisPCa. It means APT-weighted imaging could provide additional value to PI-RADS V2.1. The combination of PI-RADS V2.1 and the APT max value yielded the most effective results in distinguishing csPCa from benign lesions or cisPCa in the whole prostate gland, with an AUC of 0.883. Moreover, the combined approach was successful at distinguishing csPCa from benign lesions or cisPCa in PZ (AUC = 0.885 vs 0.798, P = 0.036), indicating a significant improvement in performance. However, in TZ, the addition of APT-weighted signal values to PI-RADS V2.1 did not lead to a noteworthy improvement in diagnostic accuracy, with the AUC increasing from 0.791 to 0.853–0.865 (P = 0.202).
While PI-RADS V2.1 showed good overall performance in diagnosing csPCa, its specificity for PZ csPCa was low, potentially leading to unnecessary biopsies[23, 24]. In our evaluation, the highest Youden's index was obtained at a PI-RADS V2.1 specificity of 0.706. Currently, APT-weighted MRI has shown promise as a good predictor for detecting PCa in PZ[17, 18]. Our study revealed that combining APTmax with the PI-RADS V2.1 score improved the specificity of diagnosing PZ csPCa and successfully enhanced the diagnostic efficiency. Moreover, APT signal values may help address the limitations of PI-RADS V2.1 and contribute positively to the specificity of identifying different csPCa lesions in PZ.
For detecting csPCa in TZ, APT-weighted signals values were included in the multivariate model. It was found to be beneficial, albeit with a minor additional impact when combined with PI-RADS V2.1, in predicting csPCa in the TZ. According to the PI-RADS V2.1, T2WI is considered the most crucial sequence for identifying and characterizing prostate lesions in TZ, followed by DWI. However, the morphological features of TZ csPCa on T2WI in the context of PI-RADS V2.1 are subjective. Therefore, other quantitative parameters should be further analyzed to improve the diagnostic performance of PI-RADS V2.1 for the detection of TZ csPCa.
Our results indicated that APT max, APT mean, and APT min were independent predictors of csPCa, the APT-weighted signal values in patients with csPCa was generally greater than that in patients with benign prostate lesions or cisPCa. However, these conclusions are not consistent with the findings of Yang et al. [25], who demonstrated that APTmax and APTmean could accurately distinguish between malignant and benign prostate lesions and that the differences in APTmin were not statistically significant. The reason could be that the samples were different. In the present study, it was diagnostic for csPCa but not for all PCa cases.
Guo et al[26] reported that APT-weighted signal value can provide more accurate lesion characterization for distinguishing TZ PCa from BPH, with an AUC of 0.812. According to the findings of Yin et al[27], APT imaging can be used to discriminate PCa from BPH, with an AUC of 0.8. Qin et al[28] reported that APT imaging performed well in PCa risk classification and had reproducible cutoff values in TZ and PZ. In line with our findings, these early results suggest that APT-weighted imaging is a promising valuable imaging marker for detecting PCa.
Theoretically, increased APT-weighted signal values indicate enhanced concentrations of proteins and peptides due to abnormal protein synthesis by rapidly dividing tumor cells and altered cellular metabolism in high-grade malignancies. Malignant prostate lesions exhibit more active metabolism than benign lesions, leading to a denser cell arrangement, reduced intercellular space, and greater secretion of macromolecules and peptides in csPCa, all of which correspond to elevated APT-weighted signal values[29, 30].
The sub-analysis demonstrated moderate diagnostic accuracy, with APTmax values of 0.6 indicating the potential to differentiate csPCa in patients with PI-RADS 3 lesions. Although PI-RADS 3 lesions are commonly observed and exhibit a moderate to high risk of malignancy, their optimal treatment remains under investigation[19, 31]. Stratifying these lesions using the PI-RADS V2.1 algorithm continues to be challenging. Our results indicated that APTmax exhibited high sensitivity and low specificity for PI-RADS 3 lesions, suggesting that APTmax has moderate diagnostic accuracy but a relatively high false-negative rate. This finding implies that functional MRI sequences such as APT imaging may not serve as effective stand-alone predictive markers for distinguishing benign and malignant prostate lesions within PI-RADS V2.1 3 lesions. Our preliminary findings provide initial support for utilizing APT-weighted imaging for stratifying PI-RADS 3 lesions.
Strengths and limitations
To the best of our knowledge, this study is the first to assess the effectiveness of APT-weighted imaging in diagnosing csPCa compared to that of combination model (PI-RADS V2.1 and the 3D APT approach). The 3D APT sequence offers a comprehensive scan of the entire prostate area, a higher signal-to-noise ratio, and less image distortion than does the 2D APT sequence[32, 33]. Notably, we also conducted an initial investigation into the diagnostic accuracy of APT-weighted signal values for detecting csPCa in patients with PI-RADS 3 lesions.
There are a few potential limitations to consider. First, this study was retrospectively conducted at a single center, potentially leading to patient selection bias that may restrict generalizability. Hence, the current findings may require further validation in prospective multicenter studies involving a larger patient cohort. Second, using freehand ROI analysis may introduce artificial errors that could impact accuracy. Additionally, it is not possible to completely eliminate the risk of undetected necrosis or cystic changes, potentially contaminating the results. Utilizing methods such as histograms and iconography may offer more objectivity and enhance accuracy. Third, in some cases, the reference standard was MRI-based targeted biopsy via transrectal ultrasound, which could overlook potential lesions that were negative on MRI but positive on pathology.