Transrectal ultrasound examination of prostate cancer guided by fusion imaging of multiparametric MRI and TRUS: avoiding unnecessary mpMRI-guided targeted biopsy

Abstract

The purpose of this study was to explore transrectal ultrasound (TRUS) findings of prostate cancer (PCa) guided by multiparametric magnetic resonance imaging (mpMRI) and to improve the Prostate Imaging Reporting and Data System (PI-RADS) system for avoiding unnecessary mpMRI-guided targeted biopsy (TB). From January 2018 to October 2019, fusion mpMRI and TRUS-guided biopsies were performed in 162 consecutive patients. The study included 188 suspicious lesions on mpMRI in 156 patients, all of whom underwent mpMRI-TRUS fusion imaging-guided TB and 12-core transperineal systematic biopsy (SB). Univariate analyses were performed to investigate the relationship between TRUS features and PCa. Then, logistic regression analysis with generalized estimating equations was performed to determine the independent predictors of PCa and obtain the fitted probability of PCa. The detection rates of PCa based on TB alone, SB alone, and combined SB and TB were 55.9% (105 of 188), 52.6% (82 of 156), and 62.8% (98 of 156), respectively. The significant predictors of PCa on TRUS were hypoechogenicity (odds ratio [OR]: 9.595, P = 0.002), taller-than-wide shape (OR: 3.539, P = 0.022), asymmetric vascular structures (OR: 3.728, P = 0.031), close proximity to capsule (OR: 3.473, P = 0.040), and irregular margins (OR: 3.843, P = 0.041). We propose subgrouping PI-RADS score 3 into categories 3a, 3b, 3c, and 3d based on different numbers of TRUS predictors, as the creation of PI-RADS 3a (no suspicious ultrasound features) could avoid 16.7% of mpMRI-guided TBs. Risk stratification of PCa with mpMRI-TRUS fusion imaging-directed ultrasound features could avoid unnecessary mpMRI-TBs.

INTRODUCTION

Transrectal ultrasound (TRUS) has been widely used for biopsy guidance in patients with suspected prostate cancer (PCa). In addition, as an imaging modality, TRUS has also dramatically improved the visualization and examination of PCa. Usually, hypoechoic lesions are the typical TRUS findings for the diagnosis of suspected PCa. However, there has been controversy regarding the use of hypoechogenicity as the diagnostic clue for PCa. On the one hand, it has been shown that 40% of PCa lesions are hyperechonic or isoechoic on TRUS;1 on the other hand, approximately 55%–60% of hypoechoic lesions in the posterior prostate were proven to be benign by TRUS-guided biopsy.2 Therefore, the role of TRUS for PCa diagnosis is questionable in current practice. Recently, some new TRUS techniques have been developed and utilized in clinical practice, such as contrast-enhanced ultrasound, sonoelastography, and high-resolution microultrasound combined with multiparametric ultrasound. However, these techniques still cannot meet clinical requirements due to a lack of standardization and large-scale evaluations of interreader variability.3,4 To further improve the diagnosis of PCa using TRUS as an imaging tool, it is necessary to identify suspected PCa lesions from another perspective.

On imaging, the behaviors of PCa lesions appear distinctly different, ranging from low-risk indolent tumors to aggressive high-risk tumors. This discrepancy may be because the pathological features of PCa are multifocal and ill-defined, and the tumor progresses along the capsule of the prostate. In addition, benign prostate hyperplasia (BPH), acute or chronic prostatitis and calcification may also affect the imaging readouts. In addition to TRUS, multiparametric magnetic resonance imaging (mpMRI) has become a promising diagnostic approach for PCa in the clinic since it can depict PCa lesions more clearly than TRUS.

Compared with traditional TRUS-guided prostate systematic biopsy (SB), MRI-guided targeted biopsy (TB) could efficiently increase the detection rates of clinically significant prostate cancer (csPCa, defined as Gleason score ≥3+4 in at least 1 biopsy core5), while decreasing the number of biopsy cores and reducing the unnecessary diagnosis and treatment of clinically insignificant prostate cancer (insignPCa).6,7,8,9 Henning et al.10 suggested that the addition of MRI-guided TB offered benefits for patients with an anterior lesion on MRI and those with multiple lesions. In addition, the latest Guidelines of the European Association of Urology have recommended the usage of the Prostate Imaging Reporting and Data System (PI-RADS)11 for the acquisition and interpretation of mpMRI data and suggested that TB should be performed in cases of PI-RADS scores 3–5.12 However, some studies showed that the interreader reproducibility remained moderate at best using the PI-RADS version 2 system,13,14,15 and the detection rates of PCa with PI-RADS scores 3 and 4 were only 24.8% and 39.1%, respectively.16 In other words, a large number of PI-RADS score 3 and 4 lesions underwent unnecessary biopsies. The number of unnecessary biopsies would further increase if insignPCa lesions are taken into account. Thus, a more practical and reliable prebiopsy PI-RADS, which could further categorize PI-RADS score 3 or 4 lesions and stratify their malignant risk, is required to increase the resource utilization efficiency and reduce the associated procedural discomfort of biopsy.

The advent of MRI-TRUS fusion imaging makes it possible to re-evaluate PCa lesions on TRUS with mpMRI as a reference. This so-called second-look TRUS has the potential to identify additional imaging features that are overlooked with conventional TRUS. We hypothesized that second-look TRUS guided by MRI-TRUS fusion imaging may be helpful in further categorizing PI-RADS score 3 lesions and stratify their malignant risk, which would reduce the number of unnecessary MRI-guided TBs. To confirm the hypotheses, this retrospective study was carried out to investigate the relationship between second-look TRUS features and PCa, and a revised PI-RADS based on the significant second-look TRUS features was proposed with the aim of reducing the number of unnecessary MRI-guided TBs.

PATIENTS AND METHODS

Patients

This retrospective study was approved by the Ethics Committee of the Shanghai Tenth People’s Hospital of Tongji University (Shanghai, China; approval No. SHSY-IEC-4.0) and informed verbal consent was obtained from each patient. From January 2018 to October 2019, 162 patients underwent transperineal mpMRI-TRUS fusion imaging-guided biopsies with positive mpMRI findings (PI-RADS scores 3–5). Six patients were excluded because their MRI and TRUS images showed diffuse PCas and overlapping lesions. The remaining 156 patients all underwent mpMRI-TRUS fusion imaging-guided TB and 12-core transperineal SB. Of the 156 patients, 15 (9.6%) had a previous negative biopsy, and 141 (90.4%) had no previous biopsy. The mean age of the 156 patients was 69.9 (range: 45–87) years. The mean prostate-specific antigen (PSA) level was 15.8 (range: 2.7–86.7) ng ml^–1, and the mean prostate volume was 44.4 (range: 17–161) ml.

mpMRI protocol and interpretation

mpMRI was performed using a 3.0-T Discovery MR system (Verio, Siemens, Erlangen, Germany) with a surface phased array coil. mpMRI pulse sequences (T1-weighted imaging, T2-weighted imaging, dynamic contrast-enhanced imaging, diffusion-weighted imaging, and apparent diffusion coefficient maps) were acquired according to the PI-RADS version 2 guidelines.11 All mpMRI scans were interpreted together by two experienced radiologists (JHL, with 3 years; and BHZ, with >7 years of experience in reading prostate MRIs using PI-RADS). The location and maximum diameter of the lesions with PI-RADS scores 3–5 were recorded before biopsy.

MRI-TRUS fusion imaging and biopsy

TRUS, MRI-TRUS fusion imaging, and all biopsy procedures were performed using a MyLab Twice US scanner (Esaote, Genoa, Italy) equipped with a fusion software package, a 6–9 MHz biplane endorectal transducer (TRT33; Esaote) and a real-time transducer navigation system. First, the MRI-TRUS fusion procedure was performed by an experienced operator with the help of the two radiologists as follows: (a) the mpMRI data were imported in Digital Imaging and Communications in Medicine (DICOM) format into a fusion software-based platform; (b) the patients were placed in the lithotomy position and remained motionless, and the mpMRI series were fused with sagittal real-time TRUS in image registration mode; (c) in navigation mode, the targets on MRI were identified, and the resulting MRI-TRUS fusion images with targets, including grayscale and color Doppler ultrasound images, were recorded in the MRI-TRUS fusion workstation; and (d) the targets for biopsy were identified on MRI (2–4 cores per target). Then, 12-core transperineal SB was performed. Finally, samples were acquired and numbered consecutively before pathological examination.

Second-look ultrasound imaging analyses

Second-look TRUS images were interpreted together by three radiologists who were blinded to the biopsy results (YCC, with 15 years of experience in urogenital ultrasound; and YKS and LHX, both with 5 years of experience in urogenital ultrasound). First, the suspicious areas on TRUS corresponding to MRI-TRUS fusion images were identified and localized. The suspicious areas on TRUS were classified as areas with echogenic changes or/and asymmetric vascular structures that were not classified as non-mass types. The suspicious TRUS areas showed iso/hyperechogenicity or hypoechogenicity compared with the contralateral prostate parenchyma. The margins of the suspicious areas on TRUS were classified as well circumscribed or irregular. Calcifications and asymmetric vascular structures relative to those in the contralateral prostate parenchyma were recorded when present in the suspicious areas on TRUS. The shape was classified as taller-than-wide (its anteroposterior dimension is greater than its transverse dimension) or not. The border was classified as close proximity to the capsule or away from the capsule. If the maximum diameter of the suspicious TRUS areas was greater than that of the suspected lesions on MRI, it was classified as larger than MRI.

Data and statistical analyses

The associations between second-look TRUS features and PCa were evaluated using the Chi-square test or Fisher’s exact probability test to determine the suspicious TRUS features of PCa that showed statistical significance. For visible TRUS lesions, the logistic regression analysis was performed to determine the independent predictors of PCa among suspicious TRUS features. After analysis, a regression equation and fitted probability of PCa were obtained. The linear-by-linear association (LLA) was used to evaluate the linear association between the fitted probability of PCa and the number of independent suspicious TRUS features. The data were analyzed using the SPSS version 25.0 software package (SPSS Inc., Chicago, IL, USA). P < 0.05 was considered to indicate a statistically significant difference.

RESULTS

Baseline characteristics of suspicious lesions on mpMRI

The baseline characteristics of the suspicious lesions on mpMRI are presented in Table 1. Among 188 lesions in 156 patients, the mean lesion size as determined by measuring the maximum diameter on mpMRI was 13.9 (range: 5–33) mm. Lesions were located in the peripheral zone in 63.8% of cases and in the central zone in 36.2%. According to the PI-RADS classification, 47.9%, 38.8%, and 13.3% of lesions were graded as score 3, 4, and 5, respectively. Of the 188 lesions, 167 were visible on second-look TRUS guided by mpMRI-TRUS fusion imaging, while the remaining 21 appeared as non-mass types on second-look TRUS.

Table 1.

Baseline characteristics of 188 magnetic resonance imaging suspicious lesions in 156 patients

Variable	Lesions, n (%)
Location
Central zone	68 (36.2)
Peripheral zone	120 (63.8)
PI-RADS
3	90 (47.9)
4	73 (38.8)
5	25 (13.3)
Number of lesions per patient
1	160 (85.1)
2	25 (13.3)
3	3 (1.6)
TRUS suspicious areas
Non-mass type	21 (11.2)
Present	167 (88.8)

PI-RADS: Prostate Imaging Reporting and Data System; TRUS: transrectal ultrasound

Detection rates of PCa and PI-RADS

Of the 156 patients, 98 had PCa and 92 had csPCa. In addition, of the 188 lesions, 105 were PCa and 100 were csPCa. The detection rates of PCa based on TB alone and SB alone were 98.0% (96 of 98) and 83.7% (82 of 98), and the detection rates of csPCa were 100.0% (92 of 92) and 73.9% (68 of 92), respectively.

In the PI-RADS score 3 group, 20.0% (18 of 90) were diagnosed as PCa and 18.9% (17 of 90) were csPCa. Similarly, 84.9% (62 of 73) were diagnosed as PCa and 79.5% (58 of 73) were diagnosed as csPCa when the PI-RADS score was 4. Finally, the detection rates of PCa and csPCa were 100.0% (25 of 25) when the PI-RADS score was 5, and all 25 lesions of PI-RADS score 5 were visible on TRUS.

Of the 21 non-mass types, the detection rate of PCa was 19.0% (4 of 21). All were csPCa lesions, and all were detected by TB alone. Of the 4 csPCa non-mass type lesions, 2 had PI-RADS scores of 3, and 2 had PI-RADS scores of 4.

Second-look TRUS features and PCa

Univariate analysis showed that the following second-look ultrasound features were significantly associated with PCa: the presence of visible suspicious lesions on TRUS (P < 0.001), visible lesions in the peripheral zone (P = 0.004), hypoechogenicity (P < 0.001), irregular margins (P < 0.001), close proximity to the capsule (P < 0.001), taller-than-wide shape (P < 0.001), asymmetric vascular structures (P < 0.001), and calcifications (P = 0.002), as shown in Table 2 and Figure 1.

Table 2.

Association between prostate cancer and various second-look transrectal ultrasound features

Parameter	PCa lesions (total=105), n	Benign disease (total=83), n	P
TRUS suspicious area			<0.001
Non-mass type	4	17
Present	101	66
Suspicious location			0.004
Central zone	28	40
Peripheral zone	77	43
The maximum diameter			0.314
Larger than MRI	30	25
Not	71	41
Echogenicity			<0.001
Hyper/isoechogenicity	6	34
Hypoechogenicity	95	32
Margin			<0.001
Well circumscribed	6	41
Irregular margins	95	25
Shape			<0.001
Not	34	50
Taller-than-wide	67	16
Border			<0.001
Away from capsule	27	61
Proximity to capsule	74	5
Asymmetric vascular structures			<0.001
Absent	27	57
Present	74	9
Calcifications			0.002
Absent	79	63
Present	22	3

MRI: magnetic resonance imaging; PCa: prostate cancer; TRUS: transrectal ultrasound

Figure 1.

csPCa in a 68-year-old man with PSA level of 5.6 ng ml^‒1 and PI-RADS score 5. (a) TRUS shows features indicative of PCa, including the visible TRUS lesion, peripheral zone, hypoechogenicity, irregular margin, close proximity to the capsule, and taller than wide shape. (b) Color Doppler TRUS shows asymmetric vascular structures (dotted circle) indicative of PCa. (c) Fusion imaging of TRUS and contrast-enhanced MRI shows the PCa lesion (dotted circle). csPCa: clinically significant prostate cancer; PSA: prostate-specific antigen; PI-RADS: Prostate Imaging Reporting and Data System; TRUS: transrectal ultrasound; PCa: prostate cancer; MRI: magnetic resonance imaging.

For the 167 visible lesions identified by TRUS, multivariate analysis with a generalized estimating equation showed that the significant predictors were hypoechogenicity (OR: 9.595, P = 0.002), taller-than-wide shape (OR: 3.539, P = 0.022), asymmetric vascular structures (OR: 3.728, P = 0.031), close proximity to the capsule (OR: 3.473, P = 0.040), and irregular margins (OR: 3.843, P = 0.041; Table 3). Among them, hypoechogenicity represented the strongest significant risk factor (OR: 9.595; Table 3). In addition, the receiver operating characteristic (ROC) curves were plotted to assess the fitted probability of PCa with the logistic regression equation. The area under the ROC curve was 0.937, and the sensitivity and specificity were 94.1% and 91.8%, respectively.

Table 3.

Multivariate analysis in predicting prostate cancer with second-look transrectal ultrasound features

Parameter	B	s.e.	OR	95% CI	P
Suspicious location	0.450	0.520	1.568	0.565–4.348	0.388
Echogenicity	2.261	0.719	9.595	2.345–39.252	0.002*
Margins	1.245	0.607	3.473	1.058–11.401	0.040*
Shape	1.264	0.553	3.539	1.197–10.465	0.022*
Border	1.346	0.660	3.843	1.053–14.021	0.041*
Asymmetric vascular structures	1.316	0.609	3.728	1.130–12.303	0.031*
Calcification	0.793	1.053	2.210	0.280–17.422	0.451

^*P<0.05, statistically significant difference. PCa: prostate cancer; TRUS: transrectal ultrasound; s.e.: standard error; OR: odds ratio; CI: confidence interval; B: regression coefficient

In lesions with none of the 5 significant predictors described above, the fitted probabilities and detection rate of PCa were 1.6%–5.3% and 5.6% (1/18), respectively. The corresponding values were 5.4%–19.8% and 7.4% (2/27) for those with 1 of the 5 significant predictors (Figure 2), 16.8%–47.9% and 44.0% (11/25) for those with 2 of the 5 significant predictors, 42.9%–77.0% and 63.1% (12/19) for those with 3 of the 5 significant predictors, 81.9%–96.4% and 93.8% (30/32) for those with 4 of the 5 significant predictors, and 96.5%–98.9% and 97.8% (45/46) for those with all 5 significant predictors. LLA analysis showed that the fitted probability of PCa increased as the number of suspicious US features increased (P < 0.001).

Figure 2.

BPH in a 75-year-old man with PSA level of 6.1 ng ml^‒1 and PI-RADS score 3. (a) TRUS shows a visible hyperechoic lesion, with features of close proximity to prostate capsule, with well circumscribed margins, and without taller-than-wide shape. (b) Color Doppler TRUS shows the lesion without asymmetric vascular structures (dotted circle). (c) T2WI shows the heterogeneous encapsulated nodule (dotted circle). (d) Markedly hyperintense is shown on high b-value DWI. (e) Focal markedly hypointense is shown on ADC map (dotted circle). (f) DCE MRI shows focal enhancement corresponding to the lesion demonstrating features of BPH on T2WI (dotted circle). BPH: benign prostate hyperplasia; PSA: prostate-specific antigen; PI-RADS: Prostate Imaging Reporting and Data System; TRUS: transrectal ultrasound; T2WI: T2-weighted imaging; DWI: diffusion-weighted imaging; ADC: apparent diffusion coefficient; DCE: dynamic contrast-enhanced; MRI: magnetic resonance imaging.

Stratification of PI-RADS score 3 lesions with TRUS features

The distribution of PCa diagnoses associated with the number of significant TRUS features is listed in Table 4. Based on these findings, we suggest revising the PI-RADS score 3 into categories of 3a (no suspicious US features), 3b (one suspicious US feature), 3c (two or three suspicious US features), and 3d (four or five suspicious US features). Ultimately, the number of detected PCa lesions classified as PI-RADS 3a was 0, which would have avoided mpMRI-guided TB in 16.7% (15 of 90) of men with PI-RADS score 3 lesions.

Table 4.

The distribution of prostate cancer detection associated with the number of significant transrectal ultrasound features

Parameter	PI-RADS score 3		PI-RADS score 4		PI-RADS score 5
	PCa lesions (total=18), n	Benign disease (total=72), n	PCa lesions (total=62), n	Benign disease (total=11), n	PCa lesions (total=25), n	Benign disease (total=0), n
Number of significant TRUS features
0	0	15	1	2	0	0
1	1	23	1	2	0	0
2	2	10	9	4	0	0
3	2	5	5	2	5	0
4	7	1	18	1	5	0
5	4	1	26	0	15	0
Non-mass type	2	17	2	0	0	0

PCa: prostate cancer; TRUS: transrectal ultrasound; PI-RADS: Prostate Imaging Reporting and Data System

DISCUSSION

As an economical and simple imaging examination approach, TRUS was first investigated in a clinical trial in the early 1970s. Since then, TRUS has been mainly used to assess prostate volume as well as prostate anatomical variations (such as enlarged median lobe) and to direct biopsy, which is still the gold standard for PCa diagnosis. Despite the increased application of prostate TRUS, there is no universal reporting and data processing system for radiologists and clinicians to interpret and report prostate images. Moreover, the idea that “conventional TRUS is not reliable at detecting PCa” was brought forth, as TRUS had poor sensitivity and specificity for diagnosing PCa.17 PCa may not always appear as a hypoechoic lesion on TRUS, and the cancer detection rates were fairly similar for hypoechoic and isoechoic lesions (9.3% and 10.4%, respectively).18 In addition, multifocality and echo overlap with BPH or prostatitis may also result in invisible PCa on TRUS.

Because of the shortcomings of TRUS as well as the accuracy of mpMRI in detecting and localizing suspicious PCa lesions, the combination of second-look TRUS with mpMRI guidance was implemented to evaluate the TRUS features predicting PCa. Several TRUS features in our study, such as hypoechogenicity, irregular margins, and taller-than-wide shape, were identified as independent features of PCa and were similar to those widely used in Breast Imaging Reporting and Data System (BI-RADS)19 and Thyroid Imaging Reporting and Data System (TI-RADS).20 Regarding progression along the capsule of the prostate and symmetry of the prostate gland anatomy, the TRUS features of “close proximity to capsule” and “asymmetric vascular structures” were revealed as independent TRUS features of PCa. However, unlike malignant tumors of the breast and thyroid, PCa is almost solid without cystic-solid mixed components. In addition, our results suggested that the calcification was not a risk factor for PCa, which might be because calcification in the prostate may be caused by prostatitis or degenerative effects of prostate tissue.

In our study, the visible lesions on TRUS with no suspicious TRUS features had a fitted probability of 1.6%–5.3% for PCa, indicating that the lesions could be followed up rather than proceeding with TB or even SB. BI-RADS and TI-RADS are widely applied in clinical practice, and the stratification of cancer risk was established by using the number of suspicious US features.19,20 We found that as the number of suspicious TRUS features increased, the fitted probability and number of detected PCa lesions also increased. With the help of reassessments of second-look TRUS features and the number of suspicious features, we proposed a modified prostate TRUS interpretation to provide better communication and reduce confusion between clinicians and patients.

mpMRI examinations of the prostate have been widely utilized for PCa detection, risk stratification, and MRI-guided biopsy. Moreover, combining this modality with PSA density could improve the detection rate of csPCa with MRI-guided biopsy.21 In addition, some new diagnostic techniques, such as the SelectMDX (MDxHealth, Nijmegen, The Netherlands) test and ⁶⁸Ga-prostate-specific membrane antigen (PSMA) positron emission tomography/computed tomography (PET/CT), have been utilized in clinical practice and are as important as mpMRI in the diagnosis of csPCa.22,23,24 According to the literature, the PCa detection rates for PI-RADS score 3, 4, and 5 lesions were 19.4%–24.8%, 39.1%–78.0%, and 82.8%–86.9%, respectively.16,25 Similarly, in our study, the PCa detection rates for PI-RADS score 3, 4, and 5 lesions were 20.0%, 84.9%, and 100.0%, respectively, suggesting that PI-RADS, especially PI-RADS score 3, should be restratified to obtain better clinical diagnosis results. mpMRI-guided TB increases diagnostic accuracy over that of TRUS-guided SB by decreasing the number of biopsy cores, maintaining or improving csPCa detection rates, and reducing the detection of insignPCa.8 Pepe et al.26 found that revising PI-RADS score 3 could reduce the number of scheduled repeated prostate biopsies. Recent high-quality studies have suggested implementing mpMRI-TB for PCa diagnosis in patients with prior negative biopsy.27,28,29 Nonetheless, because adopting this approach for all patients who received mpMRI-TB would not maximize the detection of significant PCa in the most cost-effective manner and could lead to unnecessary biopsies, especially in patients with a PI-RADS score of 3, the necessity of mpMRI-TB in the first biopsy is still debated.

The revised PI-RADS score 3 with second-look TRUS features was further defined into category 3a (no suspicious US features), 3b (one suspicious US feature), 3c (two or three suspicious US features), and 3d (four or five suspicious US features). Such classifications will help to further optimize the fitted probability of PCa detection to avoid TB in patients with PI-RADS score 3 lesions and aid in shared decision-making with patients.

Several limitations exist in our study. First, this was a retrospective study with a small sample size, thus selection bias may have existed. Second, MRI-guided high-resolution microultrasound and multiparametric ultrasound rather than conventional TRUS could reveal more ultrasound features associated with PCa. Third, we did not calculate the risk of PCa in all patients using the TRUS features, so further studies are still needed to validate the practical value of the modified PI-RADS score.

CONCLUSIONS

Risk stratification of PCa with mpMRI-TRUS fusion imaging-directed ultrasound features is promising for clinical practice, and the revised PI-RADS score could avoid unnecessary mpMRI-guided TBs.

AUTHOR CONTRIBUTIONS

GX, JW, and BY performed the research. GX and JHL wrote the paper. YKS, BHZ, JHL, YCC, and LHX analyzed the data. JW, LPS, and HXX designed the research study. All authors read and approved the final manuscript.

COMPETING INTERESTS

All authors declare no competing interests