Evaluation of the normal-to-diseased apparent diffusion coefficient ratio as an indicator of prostate cancer aggressiveness

The assessment of local aggressiveness of prostate cancer (PCa) is of key importance for appropriate management of this disease. The increase in life expectancy of the general population combined with efficient screening methods will lead to an increase in the number of new PCa cases [1]. These cases will tend to be more localized and at an earlier stage. Over the last few years, new focal methodologies have emerged for PCa treatment, such as high-intensity focused ultrasound, cryotherapy, focal laser ablation, intensity-modulated radiation therapy, radiofrequency ablation, and others. These relatively new therapeutic methods come as alternatives to existing treatment modalities such as radical prostatectomy or radiotherapy [2]. The diagnosis of PCa is based on physical (i.e., digital rectal) examination, prostate specific antigen (PSA) levels, and, when PSA levels are abnormal, blind biopsy. All of these assessments are of low diagnostic accuracy [3‐5]. Multiparametric magnetic resonance imaging (MRI) has shown great potential, with significant improvement in detection, localization, and characterization of PCa [6, 7]. This is of great importance for the selection of precise treatment options, and for minimizing treatment-related morbidity. Patients with tumors rated with a Gleason score of 6 or greater represent an important target population for PCa detection and effective therapy [8, 9].

According to recent guidelines [10], multiparametric MRI of the prostate includes three functional MR sequences: diffusion weighted imaging (DWI) [11], dynamic contrast enhancement-MR [12], and MR spectroscopy [13]. These sequences contribute to the PI-RADS scoring system of prostate lesions [10], with DWI as the main contributor. With a short acquisition time and high contrast resolution between tumor and normal prostatic tissue [14], DWI and apparent diffusion coefficient (ADC) mapping demonstrate an inverse and significant correlation with the Gleason score [7, 14]. Although ADC can estimate tumor aggressiveness, there is still a large overlap between ADCs of different tumors with the same Gleason score [7]. The purpose of this study was to establish a simple method for PCa assessment of tumor aggressiveness using ADC ratios between prostate cancer and healthy prostatic tissue.

We analyzed the images of 43 men who underwent prostate MRI examinations with an endorectal coil in our institution between February 2011 and August 2012. Written informed consent had been obtained from patients prior to the MR examinations, as well as before all interventions, such as prostatic biopsy. The requirement to obtain written, informed consent for the retrospective analysis of the data was waived. To be included in our study, subjects needed to have biopsy-proven PCa from a standard 20-core transperineal saturation biopsy. We excluded five cases because of post-biopsy hemorrhage, five cases because of motion artifacts, and 11 cases because of unavailable or negative biopsy specimens. Thus, 22 cases (median age: 64.5 years; range: 52–75 years) were further selected in this retrospective single-center study. All subjects had undergone MRI with an endorectal coil including T2-weighted images (T2WI) and DWI 4–6 weeks after PCa confirmation by biopsy.

Median PSA serum levels were 15.5 ng/mL (range, 6.8–100 ng/mL) and median prostate volume was 44 mL (range, 31–67 mL). On saturation biopsy, cancer was detected in 157 of the 440 cores (36%). After assigning the cancer-positive cores to the 16-region standardized prostate reporting scheme, 128 out of 352 regions were cancer positive (36%). The median Gleason score was 7 (range: 6–9). Gleason score 6 tumors were detected in 12 regions, Gleason score 7 tumors were detected in 38 regions, Gleason score 8 tumors were detected in 31 regions, and Gleason score 9 tumors were detected in 46 regions. From the positive tumor sectors, 86 (67%) were peripherally located, while 42 (33%) were centrally located.

Our study shows promising results in the assessment of PCa aggressiveness using tumor-to-normal ADC ratios. In this study, both central and peripheral ADC ratios demonstrated a better sensitivity, specificity, and AUC than the interpretation of ADCs alone, with the most promising results obtained for the ratios computed using normal ADCs measured in the periphery of the gland. It is not yet possible to find a cut-off value to delineate malignant tissue from benign because of the overlap of ADCs of PCa and those of normal prostatic tissue; even Gleason 9 tumors showed ADCs up to 1.27 × 10^-3 mm²/s. This finding is well known; Litjens et al. concluded in their study that tumor ADCs should not be considered absolute because of the influence of “background” variation of normal peripheral zone (PZ) tissue composition [16]. This statement is supported by the results of our study, in which we were able to demonstrate better AUC for ADC ratios in comparison with tumor ADCs alone. In our study, the ADCs of the PCa-positive biopsy cores showed statistically significant differences from the biopsy cores of normal PZ and central gland (CG). These results are consistent with results previously published [17‐19]. The mean ADC for positive PCa cores in our group (0.79 ± 0.18 × 10^-3 mm²/s) was slightly higher than the one reported by Woodfield et al. [14] (0.737 ± 0.154 × 10^-3 mm²/s, range, 0.108–1.263 × 10^-3 ) but lower than results published recently by Yamamura et al. [18] (0.96 ± 0.24 × 10^-3 mm²/s), Rinaldi et al. [19] (0.99 ± 0.15 × 10^-3 mm²/s) and Nagayama et al. [20] (1.07 ± 0.35 × 10^-3 mm²/s) (p > 0.5). The lower ADCs for positive cores in our study group were probably because of the high number of Gleason score 8 and 9 biopsy-proven sectors, which exceeded 60% of all PCa positive biopsies. This large number of biopsy-proven, high-grade PCa is likely the consequence of larger tumors in this group of patients and a higher percentage of positive biopsy cores (for example, all 20 biopsy cores were tumor-positive in two patients).

Healthy PZ prostatic tissue is rich in tubular structures, allowing water molecules to move easily. As a consequence, the ADCs of healthy PZ are high. Published ADC mean values for PZ have been 1.65 ± 0.21 × 10^-3 mm²/s and 1.73 ± 0.27 × 10^-3 mm²/s [15, 18]. Our study of the normal PZ shows similar results, with a mean ADC of 1.71 ± 0.32 × 10^-3 mm²/s. Normal CG structure is composed of two different tissues, a glandular component with high signal intensity on T2WI and a stromal component with low signal intensity on T2WI, which can mimic PCa. The ADCs of normal CG should be lower than normal values from the PZ because of benign stromal hyperplasia changes taking place in the central gland. Oto et al. showed that there is a significant difference between ADCs of hyperplasic stromal nodules (1.27 ± 0.21 × 10^-3 mm²/s) and hyperplastic glandular nodules (1.73 ± 0.28 × 10^-3 mm²/s) [21]. Our measurements were done on the entire CG including both macroscopically stromal and glandular areas and the mean ADC was 1.44 ± 0.22 × 10^-3 mm²/s, which is lower than the PZ ADC and close to the results published by Oto et al. [22].

When comparing the Gleason scores, the ADC values in subjects with Gleason 6 and 7 were significantly lower than those with Gleason 8 and 9. As an additional observation, patients with Gleason 8 and 9 were significantly younger (p < 0.0001). We hypothesize that the ADC differences arise from natural variations in prostate physiology in younger patients, owing to less significant fibrotic and atrophic changes of prostate glandular structure that would tend to decrease the ADC measurements [17].

In the last several years, many authors have investigated the correlation between ADC and Gleason score. Woodfield et al. [14] demonstrated a statistically significant difference between ADCs of low-grade tumors (Gleason 6) and intermediate-grade tumors (Gleason 7) and between low-grade tumors and high-grade tumors (Gleason 8 and 9). Verma et al. [7], in a large study of 197 prostate tumors, concluded that ADCs negatively correlated with the Gleason score. Somford et al. [22] showed the capacity of ADC mapping to predict the presence of high-grade tumors in patients with Gleason 6 (3 + 3) findings on biopsy. Although there are promising results with DWI, there are still discrepancies between ADCs of PCa and the final Gleason score in the many papers published. The differences among ADCs can be explained by physiologic factors such as age and tumor size and by technical factors (e.g., acquisition parameters, software, and the use of endorectal coils). The results of our study are very encouraging; we demonstrated that there is a statistically significant difference between ADCs of low- and intermediate-grade tumors compared with high-grade tumors. Our results are consistent with those reported by Woodfield et al. [14].

PCa is a major health problem. The diagnostic methodology remains practically the same, with ultrasound-guided systematic biopsy as the main diagnostic modality. Systematic biopsy has important limitations; it can miss up to 35% of cancers [23], and the Gleason score resulting from systematic biopsy has to be upstaged in up to 50% of the patients after radical prostatectomy [24]. In addition, Nam et al. concluded in their extensive study that complications after TRUS-guided biopsy have increased dramatically in the last 10 years [25]. With, new less-invasive focal therapy techniques and the active surveillance approach (which is based only on biopsy results), it is of great importance to improve PCa detection, characterization, and staging while reducing biopsy-related morbidity as much as possible.

The results of this study show the potential of ADC mapping in estimating the aggressiveness of PCa. By analyzing the ADC ratios when applying DWI sequences in PCa imaging, we can detect high-grade tumors more accurately than when carrying out analysis based on ADC alone. From our point of view, ADC mapping has the potential to guide the biopsy needle into the most aggressive region of a malignant lesion, resulting in fewer and more accurate biopsies, better staging, and more information to select appropriate treatment options.

We have to state several limitations: (1) We used a systematic biopsy as the reference standard, even though it may miss a substantial percentage of prostate cancers. However, it is the method of choice for cancer detection. (2) Our study population was relatively small. Patients with Gleason 8 and 9 scores tend to be younger than patients with Gleason 6 and 7 scores, so the cases could not be age-matched. (3) We do not have data on intraobserver and interobserver variability. (4) Prostate biopsy was performed prior to MRI examination, which might have influenced the ADC measurements. The presence of microscopic hemorrhage undetectable on standard MRI sequences might have altered the ADC values and underestimated the diagnostic value of ADC ratios.

The ADC tumor-to-normal ratio may be more predictive of tumor grade than analysis based on ADCs alone.