Contrast-enhanced transrectal ultrasonography: Measurement of prostate cancer tumor size and correlation with radical prostatectomy specimens

Authors

  • Ting Yue Qi,

    1. Department of Ultrasonography, Xinhua Hospital Affiliated to Shanghai Jiaotong University School of Medicine, Shanghai, China
    Search for more papers by this author
  • Ya Qing Chen,

    Corresponding author
    1. Department of Ultrasonography, Xinhua Hospital Affiliated to Shanghai Jiaotong University School of Medicine, Shanghai, China
    • Correspondence: Ya Qing Chen M.D., Department of Ultrasonography, Xinhua Hospital Affiliated to Shanghai Jiaotong University School of Medicine, No. 1665, Kongjiang Road, Shanghai 200092, China. Email: joychen1266@126.com

    Search for more papers by this author
  • Jun Jiang,

    1. Department of Ultrasonography, Xinhua Hospital Affiliated to Shanghai Jiaotong University School of Medicine, Shanghai, China
    Search for more papers by this author
  • Yun Kai Zhu,

    1. Department of Ultrasonography, Xinhua Hospital Affiliated to Shanghai Jiaotong University School of Medicine, Shanghai, China
    Search for more papers by this author
  • Xiao Hong Yao,

    1. Department of Pathology, Xinhua Hospital Affiliated to Shanghai Jiaotong University School of Medicine, Shanghai, China
    Search for more papers by this author
  • Jun Qi

    1. Department of Urology, Xinhua Hospital Affiliated to Shanghai Jiaotong University School of Medicine, Shanghai, China
    Search for more papers by this author

Abstract

Objectives

To determine the accuracy of contrast-enhanced transrectal ultrasonography for tumor size measurements of hypoechoic prostate cancer foci located in the peripheral zone.

Methods

A total of 55 men scheduled for radical prostatectomy, with biopsy-proven cancer in hypoechoic foci located in the peripheral zone, were consecutively enrolled in the present prospective study. Each patient underwent grayscale ultrasound and contrast-enhanced transrectal ultrasonography of the prostate according to a standardized protocol. The maximum tumor diameter on grayscale imaging and contrast-enhanced transrectal ultrasonography was compared with that determined using histopathology.

Results

A mean underestimation was documented to be approximately 3.9 mm and 0.6 mm for grayscale and contrast-enhanced transrectal ultrasonography imaging, respectively. Grayscale and contrast-enhanced transrectal ultrasonography imaging underestimated measurements by 76.67% (46 of 60) and 48.33% (29 of 60), whereas overestimated measurements were 20% (12 of 60) and 26.67% (16 of 60), respectively. A strong correlation was observed between contrast-enhanced transrectal ultrasonography and histopathological measurements (r = 0.91, P < 0.0001). A weak linear correlation was found between grayscale and histopathological measurements (r = 0.59, P < 0.0001). Bland–Altman analysis results were in complete accordance with correlation analysis results. For cases with maximum histopathological tumor diameters ≤10 mm and >10 mm, 40% (6 of 15) and 86.67% (39 of 45) were index tumors, respectively (P < 0.0001).

Conclusions

Contrast-enhanced transrectal ultrasonography is significantly more accurate than conventional grayscale imaging for measuring prostate tumor size, especially for tumors with a diameter >10 mm, and it might have a role in preoperative assessment of prostatic index tumor sizes.

Abbreviations & Acronyms
CETRUS

contrast-enhanced transrectal ultrasonography

CEUS

contrast-enhanced ultrasonography

CPS

cadence contrast pulse sequencing

DRE

digital rectal examination

MVD

microvessel density

PCa

prostate cancer

PSA

prostate-specific antigen

PSAD

prostate specific antigen density

PZ

peripheral zone

TRUS

transrectal ultrasonography

Introduction

Accurate non-invasive measurement of PCa tumor volume soon after diagnosis could improve determination of tumor prognosis and assist with rational selection of treatment protocols.[1] Several previous studies have documented that simple measurements of index tumor dimensions were correlated with tumor volume,[2] and the prognostic value of these measurements has previously been shown.[3-5] Therefore, as an alternative to tumor volume measurements derived from pathological examination of radical prostatectomy specimens, this simple and cost-effective technique was recommended for routine clinical practice.[3]

Large PCa tumors are more easily detected with diagnostic imaging methods than are small tumors. Therefore, prostate lesions identified with diagnostic imaging are generally more likely to be index tumors. TRUS is currently the most often used imaging modality for diagnosing prostate cancer. The most common diagnostic appearance of PCa with grayscale TRUS imaging is as a hypoechoic lesion in the PZ.[6] Although grayscale TRUS can be used to directly measure prostate tumor size, these measurements generally underestimate tumor size.[7]

The MVD of PCa tissue is increased by the proliferation of neovessels.[8] However, these microvessels are below the resolution of conventional TRUS.[9] CETRUS has recently shown promise as a means of diagnosing PCa because of its improved visualization of tumor microvessels.[10-14] As prostate tumors are strongly vascularized and neo-angiogenesis is evident at the vital border, CETRUS might be a more accurate modality than grayscale ultrasound for delineation of prostate tumor boundaries and tumor size assessment.

The present prospective primary study was designed to assess the accuracy of CETRUS for preoperative prostate tumor size measurements in patients with biopsy-proven cancer in hypoechoic foci located in the PZ.

Methods

Patients

Between February 2011 and March 2012, 135 men with hypoechoic lesions in the PZ consecutively underwent CETRUS before prostate biopsies. Patients with biopsy-proven cancer in the PZ hypoechoic foci, without prior treatment, and that were scheduled for radical prostatectomy after the diagnosis of PCa, were consecutively enrolled in the present prospective study (n = 55). The study was approved by our local ethics committee, and written informed consent was obtained from all patients.

Imaging protocols

Each patient was evaluated by a sonographer with 4 years of experience in TRUS and CETRUS using a Sequoia 512 unit (Siemens Medical Solutions, Mountain View, CA, USA) fitted with a transrectal endfire probe (EV8C4-S) operating at a frequency of 7.0 MHz. TRUS evaluations were carried out at baseline and again during intravenous bolus injection of SonoVue (Bracco, Milan, Italy). Scanning was carried out with the lowest possible pressure applied to the prostate in an effort to reduce the degree of prostate deformation as much as possible. Grayscale images with the largest transverse cross-section of each hypoechoic tumor were recorded for analyses.

All patients had CETRUS examinations after grayscale imaging procedures. The scanner was set in cadence CPS mode with a probe frequency of 8.0 MHz. The acoustic power was set at a mechanical index of 0.11 and the dynamic range was fixed at 81 dB. The CPS gain setting was set to automatic optimizing. The transverse plane with the largest cross-section of each hypoechoic tumor was selected and measured with CETRUS. We used a dual-scan mode where real-time CETRUS and grayscale images were simultaneously shown on the monitor to assure good overlap of these two modalities. SonoVue (2.4 mL) was infused intravenously, followed by a 0.9% normal saline flush (5 mL). Each plane was observed continuously for approximately 90 s. Injections were repeated after elimination of contrast agent from the prostate for patients with more than one hypoechoic tumor. All acquired digital videos were recorded on the internal hard disk of the scanner for subsequent analysis.

Image interpretation and analysis

The baseline grayscale images of hypoechoic tumors were analyzed by another sonographer who was blinded to the CETRUS data and pathological results. The maximum diameter of each hypoechoic tumor on grayscale imaging was measured and recorded (Fig. 1a).

Figure 1.

Tumor size measurements obtained from each modality and histopathological examinations in a 58-year-old man with a PSA level of 13.38 ng/mL and a Gleason score of 4 + 4. (a) The tumor margin (white arrows) and the maximum diameter of hypoechoic tumor on grayscale imaging (white line). (b) The tumor margin (white arrows) and the maximum tumor diameter on CETRUS imaging (white line). (c) The tumor margin (yellow dotted line) and the maximum histopathological size of the corresponding tumor on grayscale and CETRUS images (white line).

All CETRUS data were independently reviewed frame-by-frame on the scanner by two other sonographers who were blinded to the grayscale imaging and pathological results. When tumor enhancement was at peak level, the corresponding static CETRUS image was selected for determination of maximum tumor diameter. When their evaluations differed, the images were evaluated by a third experienced investigator with 5 years of extensive CETRUS prostate experience. A consensus was reached for static CETRUS image selection discrepancies, for determination of maximum tumor diameters. For measurements, the tumor edge was defined as the end of the abnormally enhanced focus at the time of the lesion's maximal contrast enhancement. The CETRUS image's maximum tumor diameter was measured and recorded (Fig. 1b).

Radical prostatectomy pathological analysis

After radical prostatectomy, prostatectomy specimens were fixed overnight in 10% formalin and coated with India ink. The entire gland was cut in 3–4 mm sections perpendicular to the urethra from the apex to the base, and sections were then divided into halves or quadrants to fit in conventional tissue cassettes for paraffin embedding. Seminal vesicles were separately amputated. Micro-slices were subsequently placed on glass slides and stained with hematoxylin–eosin. An experienced pathologist reviewed all tissue sections, and cancerous areas were outlined on slides with a pen. Pathological tumor stage, Gleason score, surgical margin status and the presence of seminal vesicle invasion were recorded. The maximum histopathological size of the corresponding tumor on grayscale and CETRUS images was determined by marking both ends of the outlined tumor with a pen and measuring this distance directly from the glass slide (Fig. 1c). The latest study by Wolters et al. showed that the fixation and processing of the radical prostatectomy specimens did not result in significant reduction in tissue volume.[15] Therefore, a shrinkage factor was not considered in the measurement of the prostate tumor size in the present study. To most closely match corresponding transverse grayscale or CETRUS image and pathological step-section slice, some anatomical landmarks within the same region of the prostate, which were used successfully in our previous study,[16] were applied.

Statistical analysis

MedCalc statistical software version 11.4.2 (MedCalc software, Mariakerke, Belgium) was used for all statistical calculations. Statistical tests were all two-sided. P-values were based on two-sided testing at a 5% significance level. Agreement between tumor size measurements obtained from each modality and histopathological examinations were evaluated with Pearson correlation coefficients and Bland–Altman analysis.

Results

The clinical and pathological characteristics of patients are shown in Table 1. Overall, 60 hypoechoic tumor foci were identified in the PZ of 55 patients based on histopathology maps. A total of 50 patients had a single hypoechoic tumor focus and five patients had two hypoechoic tumor foci. Overall, 75% (45 of 60) of hypoechoic tumor foci were index tumors.

Table 1. Clinical and pathological characteristics of patients
VariableMean ± SD (range) or no. (%) cases
No. patients55
DRE18 (32.73)
Age (years)65.53 ± 7.18 (45–75)
PSA (ng/mL)10.90 ± 4.67 (4.2–21.8)
Prostate volume (mL)33.74 ± 11.07 (19.26–52.14)
PSAD (ng/mL per mL)0.31 ± 0.14 (0.16–0.58)
Pathological stage 
T2 
T2a6 (10.91)
T2b2 (3.64)
T2c9 (16.36)
T3 
T3a33 (60)
T3b5 (9.09)
Pathological features 
Organ-confined17 (30.91)
Prostatectomy Gleason ≥750 (90.91)
Extracapsular extension38 (69.09)
Positive surgical margins32 (58.18)
Seminal vesicle invasion5 (9.09)

The mean maximum diameter of PCa tumors was 12.52 mm (range 5.6–25 mm), 15.85 mm (range 8.7–28 mm), and 16.43 mm (range 7–28 mm) measured with grayscale imaging, CETRUS and histopathological examination, respectively. The maximum tumor diameter for CETRUS and histopathology was significantly higher than for grayscale imaging (P < 0.0001 and P < 0.0001, respectively). There was no significant difference in the maximum tumor diameter between CETRUS and histopathology (P > 0.05) (Table 2). The mean absolute difference in tumor size differed between grayscale imaging (3.92 mm) and CETRUS imaging (0.58 mm; P < 0.0001). The mean absolute difference for tumors with maximum histopathology diameters ≤10 mm and >10 mm was 1.89 mm and 1.4 mm for CETRUS imaging, and 0 mm and 5.22 mm for grayscale imaging, respectively (P < 0.01 and P < 0.0001, respectively). Tumor size measurements made on the basis of grayscale and CETRUS imaging generally underestimated tumor size based on histopathology. For grayscale and CETRUS imaging, the underestimated measurements were 76.67% (46 of 60) and 48.33% (29 of 60), and the overestimated measurements were 20% (12 of 60) and 26.67% (16 of 60), respectively.

Table 2. Correlations between maximum tumor diameter differences and pathological tumor sizes
Maximum histopathological tumor diameter (mm)No. (%) lesionsDifference, mean ± SD (mm)P-value*
GrayscaleCETRUS
  1. *P-value by paired sample t-tests.
≤1015 (25)0 ± 2.591.89 ± 1.55<0.01
>1045 (75)–5.22 ± 5.39–1.40 ± 2.62<0.0001
Total60 (100)–3.92 ± 5.33–0.58 ± 2.78<0.0001

There was a clear tumor boundary in 71.67% (43 of 60) of foci using CETRUS imaging, whereas only 28.33% (17 of 60) foci at grayscale imaging had a clear boundary (P < 0.0001) (Table 3). Furthermore, foci had unclear (23.33%; 14 of 60) and clear (23.33%; 14 of 60) boundaries using both ultrasound techniques. The maximum tumor diameter measured with grayscale imaging was significantly smaller than with histopathological examination, regardless of the tumor boundary status (P < 0.01; Table 4). For cases with clear boundaries on CETRUS imaging, no significant difference was found in the maximum tumor diameter between CETRUS and histopathological examination (P > 0.05). However, for cases with unclear boundaries on CETRUS imaging, there was a significant difference in the maximum tumor diameter between CETRUS and histopathological examination (P < 0.05). Additionally, using CETRUS, a clear tumor boundary was found in 33.33% (5 of 15) and 84.44% (38 of 45) of foci in maximum histopathological diameter ≤10 mm and >10 mm groups, respectively. In comparison, an unclear tumor boundary was found in 66.67% (10 of 15) and 15.56% (7 of 45) of foci with maximum histopathological diameter ≤10 mm and >10 mm groups, respectively. The correlations between the Gleason score and tumor boundary status on CETRUS and grayscale imaging are shown in Table 5. Of 60 tumor foci, 15 foci had a Gleason score <7 and the remaining foci had a Gleason score ≥7. Using CETRUS, the tumor boundary status was clear in 33.33% (5 of 15) and 84.44% (38 of 45) of the foci with Gleason scores <7 and ≥7, respectively (P < 0.001). However, on grayscale imaging, the percentage of tumors with clear boundaries was 40% (6 of 15) and 24.44% (11 of 45) in the two groups, respectively (P > 0.05).

Table 3. Comparisons of tumor boundary status for CETRUS and grayscale imaging
 Grayscale imagingTotalP-value*
ClearUnclear
  1. *P-value by McNemar's χ2-test.
CETRUSClear142943<0.0001
Unclear31417
Total 174360 
Table 4. Correlations between tumor boundary status and measurement accuracy using different imaging methods
Tumor boundary conditionnMaximum diameter mean ± SD, range (mm)P-value*nMaximum diameter mean ± SD, range (mm)P-value*
GrayscaleHistopathologyCETRUSHistopathology
  1. *P-value by paired samples t–test.
Clear1713.65 ± 4.57 (6.5–23)18.76 ± 5.91 (9–28)<0.014317.24 ± 4.39 (10.5–24)17.88 ± 6.50 (7–28)>0.05
Unclear4312.06 ± 5.23 (5.6–25)16.03 ± 6.87 (7–28)<0.011715.65 ± 3.05 (8.7–28)13.35 ± 3.50 (8–25)<0.05
Table 5. Correlations between Gleason score and tumor boundary status on CETRUS and grayscale imaging
Gleason scoreTumor boundary statusTotal
GrayscaleP-value*CETRUSP-value*
ClearUnclearClearUnclear
  1. *P-value by McNemar's χ2-test.
<769>0.05510<0.00115
≥7113438745
Total1743 4317 60

A strong positive correlation was observed between CETRUS and histopathological examinations (r = 0.91, P < 0.0001), whereas a weak positive linear correlation was found between grayscale and histopathological examinations (r = 0.59, P < 0.0001; Fig. 2).

Figure 2.

Correlation analyses. (a) Correlation between grayscale and histopathological examinations (r = 0.59 [CI 95%: 0.39–0.73], P < 0.0001). (b) Correlation between CETRUS and histopathological examinations (r = 0.91 [CI 95%: 0.86–0.95], P < 0.0001).

Using histopathology as a gold standard, the mean size, mean difference from pathological sizes and the limits of agreement for each method are presented in Bland–Altman plots (Fig. 3). The mean differences were −3.9 mm (SD 5.3; 95% CI −5.3 mm to −2.5 mm) for grayscale imaging and −0.6 mm (SD 2.8; 95% C, −1.3 mm to 0.1 mm) for CETRUS imaging, respectively. The corresponding results of the limits of agreement (mean ± 1.96 SD) were −14.4 mm to 6.5 mm for grayscale imaging and −6.0 mm to 4.9 mm for CETRUS, respectively. The corresponding maximum absolute value of the difference was 16 mm, and 6 mm for grayscale and CETRUS imaging, respectively. Furthermore, four values (6.67%) for grayscale imaging were beyond the limits of agreement. In contrast, no value was beyond the limits of agreement for CETRUS imaging. Compared with CETRUS (0.6 mm or 3.65%), the average bias between grayscale and histopathological examination was 3.9 mm or 23.74%.

Figure 3.

Bland–Altman plots show agreement between tumor size measurements obtained from histopathological examination, and (a) grayscale imaging and (b) CETRUS imaging. Horizontal (solid) lines for mean differences correspond to differences between size measured at imaging and size measured on pathologic slices, and broken lines represent mean differences at 95% confidence limits (± 2 SD). Observations within 95% limits of agreement likely result from measurement error; observations outside 95% limits of agreement likely result from true differences between both methods.

Discussion

CEUS was recently developed to provide detailed and real-time perfusion imaging. It improved the visualization of tumor vascularity in many organs, including the prostate.[17, 18] Meanwhile, it has been postulated that this technique reliably visualizes neovascularization within and around tumors, and can potentially be used for tumor boundary identification and lesion characterization.[19-21] A recent study by van Esser et al. has documented that CEUS is more accurate than grayscale ultrasound for preoperative size assessment of invasive ductal breast carcinomas.[21] However, there is no literature examining the potential value of CEUS on PCa tumor size determination. The present study used Pearson correlation coefficients and Bland–Altman analysis to evaluate agreement between grayscale, CETRUS and histopathological examination. We also evaluated the accuracy of CETRUS measurements of PCa tumors located in the PZ.

In the present study, the mean maximum PCa tumor diameter varied for grayscale imaging (12.52 mm), CETRUS (15.85 mm) and histopathological examinations (16.43 mm). The correlation between histopathological examination and CETRUS was more robust than for grayscale imaging (r = 0.91, P < 0.0001 vs r = 0.59, P < 0.0001). We used Bland–Altman analysis as an alternative method for assessing agreement between methods, because exclusive use of correlation and regression analysis for comparing two methods of clinical measurement is misleading.[22] Furthermore, Bland et al. also suggested that a plot of the difference against the average of the standard and new measurements is advantageous over plots of the difference against the standard measurement.[23] Therefore, this method was also applied in the present study. As a result, based on the Bland–Altman analysis, the concordance, limits of agreement and average bias for CETRUS were more acceptable than for grayscale imaging in clinical settings, and are therefore preferable for clinical practice. The potential for CETRUS to substitute for histopathological examination was further supported by Bland–Altman analysis.

In the present study, the mean maximum tumor diameters measured with grayscale imaging were significantly smaller than for CETRUS (P < 0.0001) and histopathological examination (P < 0.0001), and there was not a statistical difference between CETRUS and histopathology (P > 0.05). A possible explanation for this difference lies in these two imaging techniques' differing capacities to visualize true tumor boundaries. Because microvascular structure changes associated with PCa tumors occur earlier than morphological changes, grayscale imaging could not delineate the real tumor boundaries of PCa, and it could not visualize tumors beyond hypoechoic areas in most cases. Regardless of tumor boundary status, the maximum tumor diameters measured with grayscale imaging were significantly smaller than with histopathological examination (P < 0.01), with a mean underestimation of 3.9 mm. This suggests that grayscale imaging essentially underestimates prostate tumor size, as has been previously reported.[7] In comparison, contrast agent used with CETRUS improved the detection of low-volume blood flow by increasing signal strength from small vessels, signal-to-noise ratio and delineation of neovascular anatomy. Furthermore, contrast agent vibrations generate higher harmonics than surrounding tissues.[8] Therefore, cancerous regions of the prostate gland can be distinguished from surrounding tissues with CETRUS based on neo-angiogenesis at the tumor's borders. This accounts for CETRUS underestimates of tumor size (0.6 mm), compared with grayscale imaging (P < 0.0001). Additionally, for CETRUS most of the cases with clear tumor boundaries had Gleason scores ≥7. A possible explanation for this finding is the positive correlation between the Gleason score and MVD in PCa.[13] Consequently, CETRUS is a more sensitive and accurate imaging technique than grayscale imaging for identifying PCa's true borders.

However, not all measurements with CETRUS were accurate in the present study. Cases with unclear boundaries based on CETRUS imaging (28.33%; 17 of 60) had greater maximum tumor diameters, compared with histopathological measurements (P > 0.05). This indicates that tumor boundary status is an additional factor that affects the accuracy of CETRUS measurements. Most (58.82%; 10 of 17) of these erroneous measurements were associated with tumors ≤10 mm, and CETRUS overestimated the size of these smaller tumors by an average of 1.9 mm. These inaccuracies could be a result of small tumor's low malignancy and low density of small blood vessels,[24] because these factors could result in the generation of correspondingly low signals. This could compromise the potential for CETRUS to distinguish cancerous and normal surrounding tissues for small-sized tumors. However, just 25% (15 of 60) of cases had maximum histopathological tumor diameters ≤10 mm in the present study population, so further study is warranted to confirm these findings.

Finally, we used 10 mm as a threshold, because 0.5 cc (10 mm in diameter) is an established threshold for prognosis.[25] The present study also produced several interesting results. For cases with maximum histopathological tumor diameters ≤10 mm and >10 mm, 40% (6 of 15) and 86.67% (39 of 45) were index tumors, respectively (P < 0.0001). These findings showed that CETRUS is a potentially useful imaging technique in preoperative assessment of prostatic index tumor sizes.

The present study was not devoid of limitations. One limitation was patient selection; only patients with biopsy-proven cancer in the PZ hypoechoic lesions were included. Further validation of CETRUS would require consideration of all patients. Another limitation was the use of multiple blocks and slides per cross-section instead of whole-mount processing. Thus, foci of tumors on separate slides could potentially cause measurement errors. However, as previously stated, this is less relevant.[26]

In conclusion, CETRUS is more accurate for measuring prostate tumor size than conventional grayscale imaging, especially for tumors with diameters >10 mm, and this technique might be applicable in preoperative assessment of prostatic index tumor sizes.

Acknowledgments

This study was funded by grants from the Innovation Fund for Doctor of Shanghai Jiaotong University School of Medicine (project number: BXJ201123), the Crossover Fund of Engineering with Medicine of Shanghai Jiaotong University (project number: YG2011MS43) and the Science and Technology Commission of Shanghai Municipality Foundation (project number: 10JC1411400 and 10411952000). Relevant staff from the department of urology and pathology of Xinhua Hospital have provided a great deal of help in the gathering of data.

Conflict of interest

None declared.

Ancillary