SEARCH

SEARCH BY CITATION

Keywords:

  • elastography;
  • prostate cancer;
  • staging;
  • ultrasonography;
  • extracapsular extension;
  • index lesion

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. PATIENTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONFLICT OF INTEREST
  8. REFERENCES
  9. Supporting Information

Study Type – Diagnostic (exploratory cohort) Level of Evidence 2b

What’s known on the subject? and What does the study add?

Current studies evaluating real-time elastography in patients prior to radical prostatectomy reported sensitivities between 57% and 100% for detection of prostate cancer.

This is the first prospective study comparing the findings of real-time elastography and conventional gray-scale ultrasound with final pathology. A significant improvement for cancer detection as well as detection of extra capsular is shown by adding the attributes of tissue elasticity to current gray-scale imaging.

OBJECTIVE

• To evaluate whether transrectal real-time elastography (RTE) improves the detection of intraprostatic prostate cancer (PCa) lesions and extracapsular extension (ECE) compared with conventional grey-scale ultrasonography (GSU).

PATIENTS AND METHODS

• In total, 229 patients with biopsy-proven PCa were prospectively screened for cancer-suspicious areas and ECE using GSU and RTE.

• The largest tumour focus detected by RTE was defined as the index lesion.

• The prostate gland was stratified into six sectors on GSU and RTE, which were compared with histopathological whole mount sections after radical prostatectomy.

RESULTS

• Histopathologically, PCa was confirmed in 894 out of 1374 (61.8%) evaluated sectors and ECE was identified in 47 (21%) patients.

• Of these 894 sectors, RTE correctly detected 594 (66.4%) and GSU 215 (24.0%) cancer suspicious lesions.

• Sensitivity was 51% and specificity 72% using RTE compared to 18% and 90% for GSU.

• RTE identified the largest side specific tumour focus in 68% of patients.

• ECE was identified with a sensitivity of 38% and specificity of 96% using RTE compared to 15% and 97% using GSU.

CONCLUSIONS

• Compared with GSU, RTE provides a statistically significant improvement in detection of PCa lesions and ECE.

• RTE enhances GSU, although improvement is still needed to achieve a clinically meaningful sensitivity.


Abbreviations
ECE

extracapsular extension

GSU

grey-scale ultrasonography

IL

index lesion

PCa

prostate cancer

RP

radical prostatectomy

RTE

real-time elastography

INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. PATIENTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONFLICT OF INTEREST
  8. REFERENCES
  9. Supporting Information

Accurate staging is necessary to optimize the counselling and treatment of patients with biopsy-proven prostate cancer (PCa). Nomograms are commonly used to preoperatively predict pathological tumour characteristics [1]. Nomogram risk calculation is based upon different preoperative parameters, including DRE, preoperative PSA and biopsy Gleason score. Imaging techniques such as transrectal grey-scale ultrasonography (GSU) are used to receive more precise information in addition to DRE. Therefore, GSU is the most frequently used, cost-effective imaging technique in clinical practice to measure prostate size and, in addition, to detect cancer suspicious lesions and extracapsular extension (ECE) before surgery [2]. Studies have shown that hypoechoic lesions, which are considered cancer suspicious, are histologically confirmed as malignant only in 17–57% of cases [3]. Almost 60% of pT3 tumours are not detected preoperatively by GSU [4]. Hence, new imaging techniques are needed and should be scientifically compared with current GSU in clinical practice to investigate whether any improvement is achieved.

It was shown that malignant tissue of the prostate gland has a lower elasticity as a result of a higher cell density compared to benign tissue [5,6]. In recent years, substantial progress has been made in developing ultrasonography-based real-time elastography (RTE), a technique that enables areas of higher cell density to be distinguished from lower ones [7]. Compared with step section pathological analysis, sensitivity rates of RTE for PCa detection vary in different reports in the range 57–100%[8–11]. To date, the value of RTE has not been compared with conventional GSU.

The present study aimed to prospectively assess whether RTE improves GSU detection of intraprostatic lesions and ECE in men with biopsy-proven PCa. Imaging results were matched with histopathologically evaluated whole mount slides.

PATIENTS AND METHODS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. PATIENTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONFLICT OF INTEREST
  8. REFERENCES
  9. Supporting Information

Between August 2008 and July 2009, 229 consecutive patients with biopsy-proven PCa scheduled for radical prostatectomy (RP) were prospectively examined by a single investigator who was blinded to all preoperative data, particularly DRE and histology of prostate biopsy. All patients provided their informed consent and the study received local ethical review board approval. GSU as well as preoperative RTE were performed using an EUB-7500 HV with an V53W probe (Hitachi Medical, Tokyo, Japan) 1 day before surgery. Standardized examination in a left lateral position was performed using GSU first, followed by RTE. On GSU, hypoechoic lesions were classified as cancer suspicious (Fig. 1) and ECE was defined as positive when the capsule appeared bulgy or irregular. Prostate volume and all areas of abnormal echogenicity were recorded for both sides and each prostate sector (Fig. 2). All data were stored in a unified database (LinkMCS; Flexagilty, Hamburg, Germany).

image

Figure 1. Complete transrectal examination. Monochrome ultrasonography (left) showed a hypoechoic lesion in the right lobe. Elastography (middle) revealed stiffer regions in both lobes. The histopathological result (right) showed prostate cancer in both lobes, with an index lesion in the right lobe.

Download figure to PowerPoint

image

Figure 2. Zonal anatomy of the prostate. The peripheral zone is divided into prostate sectors at each side: base, mid-gland, apex. The inner gland was not involved in the evaluation.

Download figure to PowerPoint

Performing RTE, the prostate lobes were minimally compressed and decompressed free-handed with the transrectal probe. The frequency and the force of compressing the prostate were adjusted according to a visual indicator displayed on the video screen. This indicator was established to decrease the interobserver variability and facilitate the reproducibility of the images. RTE visualizes the tissue density in a colour scale from blue (hardest areas with minimum strain) to green (intermediate strain) and red (softest areas with greatest strain; Fig. 1). Systematic examination was performed in the transversal and longitudinal plane and takes ≈ 10–12 min for each patient. Malignancy criteria described by Konig et al.[12] were used to define cancer suspicious areas displayed as reproducible blue-coloured lesions (see Supporting information, Video S1). Localization of any suspicious lesion was registered according to the corresponding prostate sector (base, midgland, apex; n= 1374). The tumour focus with the largest diameter was assigned as the elastographic index lesion (IL). It was hypothesized that the size of the IL may correlate with the total tumour volume. Furthermore, we presumed a correlation between the relationship of the IL to the prostate capsule and the likehood of having ECE in this sector. To evaluate ECE, soft rim artefacts were triggered with the transrectal probe as described by Pallwein et al.[13]. Therefore, the probe induces a shift between the prostate capsule and the periprostatic fat layer, which causes a red-coloured border surrounding the prostate capsule. Discontinuances in the soft rim indicate ECE. To validate this hypothesis, the findings of GSU and RTE were compared with histopathological whole mount sections after RP.

After RP, surgical specimens were sectioned into 3-mm slides according to the Stanford protocol. After haematoxylin and eosin staining, microscopic evaluation was performed by a single dedicated pathologist, who was blinded to all clinical data. Localization of each PCa focus was documented on a prostate map form, which enables matching of the whole mount slides with the corresponding prostate sector on GSU or RTE. Total tumour volume was calculated by summing the single cancer volumes according to a superimposed grid for each 3-mm section. The largest cancer focus on the haematoxylin and eosin whole mount slides defined the histological IL. In addition, ECE was documented according to each sector. Tumour characteristics are shown in Tables 1 and 2.

Table 1.  Description of clinical baseline criteria of 229 patients
Patient characteristicOverall finding
Mean (range) age (years)64 (42–75)
Mean (range) PSA (ng/mL)10.3 (0.7–67)
Mean (range) prostate volume (mL)42.3 (11.2–142.6)
Table 2.  Description of histopathological findings in 229 patients
VariablePreoperative findingPostoperative finding
Clinical/pathological stage (n, %)
 1c 119 (52.0)0 (0)
 Organ confined92 (40.2)182 (79.5)
  2a31 (13.5)20 (8.7)
  2b38 (16.6)2 (0.9)
  2c23 (10)160 (69.9)
 Extracapsular extension18 (7.9)47 (20.5)
  3a12 (5.2)23 (10.4)
  3b6 (2.6)23 (10)
  40 (0)1 (0.4)
Gleason score prostate biopsy/radical prostatectomy specimen (n, %)
 ≤6139 (60.7)106 (46.3)
 775 (32.7)92 (40.2)
 >715 (6.6)31 (13.5)

For statistical analysis, we used unpaired Student’s t-tests (for numerical data) and the chi-square test (for categorical data) to compare the tumour characteristics between organ-confined cancers and those with ECE. The level of significance (two-tailed) was given with a P < 0.05. Regression analysis (Kendall tau-b) was used to determine the correlation between size of the index lesion and other parameters such as tumour volume, PSA level and Gleason score. Multivariate logistic regression models were used to determine new independent parameters to predict ECE.

Sensitivity, specificity, positive and negative predictive values, as well as the overall accuracy of both ultrasonography techniques, were calculated for the detection of tumour foci by prostate sector, localization of the IL and side of ECE.

Statistical tests were performed using IBM SPSS, version 17 (SPSS Inc., Chicago, IL, USA).

RESULTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. PATIENTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONFLICT OF INTEREST
  8. REFERENCES
  9. Supporting Information

TUMOUR CHARACTERISTICS

Mean (range) patient age was 64 (42–75) years. The mean PSA level was 10.3 ng/mL and mean prostate volume was 42.3 mL. A cancer suspicious palpable mass was assessed by DRE in 110 (48%) of 229 patients. The mean (range) time from biopsy to surgery was 30 (16–42) days. Clinical and pathological data are summarized in Tables 1 and 2.

In total, 1374 prostate sectors (six per patient) were investigated. PCa lesions were histopathologically verified in 894 sectors, with the highest frequency in the midgland (42.3%), followed by the prostate base (29.5%) and prostate apex (28.2%) not showing any side-specific difference.

Mean (range) tumour volume was 2.8 (0.05–24.2) cm3. Tumours where ECE was identified had a histopathologically higher tumour volume compared to organ-confined lesions (1.9 cm3 vs 6.1 cm3; P < 0.001). The diameter of the IL in histopathological specimens was significantly larger in tumours with ECE (21.6 mm) compared to organ-confined tumours (13.8 mm). Details are shown in Table 3.

Table 3.  Tumour characteristics of 229 patients after radical prostatectomy
VariableMean (sd) P value Organ-confined vs extraprostatic extension
OverallPathological organ-confined (n= 182)Pathological extraprostatic extension (n= 47)
Tumour volume (cm3) 2.8 (3.24) 1.9 (1.87) 6.1 (4.86)<0.001
Diameter of index lesion (mm)15.4 (8.89)13.8 (7.80)21.6 (10.12)<0.001
PSA (ng/mL)10.2 (9.9) 8.7 (7.74)16.2 (14.28)<0.001

Using regression analysis, we found aq positive correlation between the size of the histopathological IL and total tumour volume (r= 0.549, P < 0.001), preoperative PSA (r= 0.281, P < 0.001) and Gleason score (r= 0.397, P < 0.001) in all 229 cases. The index lesion diameter was a significant (P= 0.005) independent predictor for ECE in multivariate analysis.

COMPARISON BETWEEN GREY-SCALE AND REAL-TIME ELASTOGRAPHY WITH PATHOLOGICAL FINDINGS USING WHOLE MOUNT SECTIONS

Out of 894 histopathologically confirmed tumour foci, we detected 594 (66.4%) using RTE compared to only 215 (24%) areas with GSU. RTE could detect cancer lesions in a higher sensitivity at the apex compared to the base. This distribution was not identified using GSU. Although the histopathological distribution of the areas was almost balanced through the six prostate sectors, we found a different correlation for the preoperative findings with the whole mount analysis after RP. Whereas high statistical significant correlation (P < 0.001) could be shown for four of six sectors using RTE, only one sector correlated significantly (P= 0.05) with histopathological report using GSU. Overall sensitivity for RTE was 50.4% compared to 18.3% for GSU (Table 4). Whereas sectors at the prostate base showed a low sensitivity (left 29%, right 24.8%), better sensitivity was found for apical sectors (left 60.9%, right 62%) using RTE. This difference in sensitivity according to sector location was not shown for GSU. The detection rate increased with a higher Gleason grade. Detection accuracy for Gleason ≤6, 7 and >7 was 59.9%, 55.3% and 67.7% for RTE and 45.3%, 43.5% and 42.5% for GSU, respectively. The different intervals between biopsy and surgery did not influence the sensitivity of RTE or GSU.

Table 4.  Accuracy (%) of prostate cancer detection: grey-scale ultrasonography (GSU) vs real-time elastography (RTE)
SiteSensitivitySpecificityPositive predictive valueNegative predictive valueAccuracy
GSURTEGSURTEGSURTEGSURTEGSURTE
Left lobe          
 Base17.629.088.884.767.671.745.847.248.052.8
 Mid29.360.284.265.890.389.819.224.838.461.1
 Apex16.560.991.261.465.561.452.060.953.761.1
Right lobe          
 Base15.824.891.783.372.467.344.044.447.649.3
 Mid19.865.292.964.392.589.120.629.333.265.1
 Apex10.962.093.569.671.475.241.355.244.165.1
Total18.350.490.471.576.775.837.243.644.259.1

Out of 229 RP specimens, the histologically verified IL was detected using RTE in 198 cases (86.5%). The findings on RTE correlated significantly (P < 0.001) with histopathology. Details according to each sector are shown in Fig. 3. The best correlation was found for the midgland (59.1% vs 52.8%), followed by the apex (24.7% vs 13.5%). We could allocate the IL to the correct sector in 52.8% of cases (n= 121) and to the correct prostate lobe in 67.7% of cases (n= 155). The accuracy of IL detection by RTE is shown in Table 5. We found the best sensitivity for the right midgland and apex with 81.4% and 78.6%, respectively. By contrast, sensitivity was low for base sectors (37.5% and 40%). Overall accuracy was 87.2%, as shown in Table 5.

image

Figure 3. Distribution of index lesion in elastography and histopathology.

Download figure to PowerPoint

Table 5.  Accuracy (%) of real-time elastography for sides and regions of the prostate to predict the index lesion
SiteSensitivitySpecificityPositive predictive valueNegative predictive valueAccuracy
Left lobe     
 Base40.098.885.790.289.9
 Midgland65.390.669.688.884.3
 Apex50.093.536.896.190.4
Right Lobe     
 Base37.596.466.788.986.9
 Midgland81.483.567.691.382.8
 Apex78.689.736.798.288.9
Total58.892.160.592.387.2

ECE was histopathologically confirmed in 47 patients (20.5%). The sensitivity and specificity to detect ECE was 38.2% and 96.2% using RTE, vs 14.9% and 97.3% for GSU (Table 6).

Table 6.  Accuracy (%) of grey-scale ultrasonography and real-time elastography to predict extracapsular extension
TechniqueSensitivitySpecificityPositive predictive valueNegative predictive valueAccuracy
Greyscale ultrasonography14.997.358.381.680.3
Sonoelastography38.396.272.085.884.3

DISCUSSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. PATIENTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONFLICT OF INTEREST
  8. REFERENCES
  9. Supporting Information

Accurate staging is necessary to optimize the counselling and treatment of patients diagnosed with PCa. [14]. Tumour size as well as ECE are crucial factors for preoperative planning. Current imaging techniques are not sufficiently sensitive to provide accurate preoperative staging. Although GSU is the most commonly available and cost-effective imaging technique in daily practice, it is known that PCa lesions can appear hypo-, iso- or hyperechoic [15]. This dilemma explains the low sensitivity of this technique. However, in addition to DRE, GSU is widespread and often is used as preoperative standard procedure to evaluate clinical stage before RP in everyday practice.

This is the first prospective study to compare RTE and conventional GSU with final pathology. We correctly detected PCa lesions in a defined sector with a sensitivity of 50.4% using RTE compared to 18.3% using GSU. We allocated the IL to the correct prostate lobe (left and right) in 68% of cases, resulting in a sensitivity of 58.8% and a specificity of 92% using RTE. According to the varied rendering of the prostate capsule during GSU and RTE, we evaluated whether side-specific ECE can be predicted. In the present study cohort, RTE achieved a two-fold increased detection rate for ECE, with a sensitivity of 38.3% compared to 14.9% using GSU alone.

The present study shows that RTE does statistically improve cancer visualization by adding the attributes of tissue elasticity to conventional grey-scale imaging. Several studies have investigated RTE in patients with biopsy-proven PCa and compared these results with histology after RP. These studies reported sensitivities in the range 57–100%[16]. Salomon et al.[11] investigated 109 patients and improved the overall sensitivity and specificity for cancer detection to 75.4% and 76.6%, respectively. Tsutsumi et al. describe a variable sensitivity in the range 57–94% depending on the different sextants of the prostate [10]. In their study, different results are reported in the vertical axis of the prostate, with best detection in the anterior part of the gland; however, most studies differentiate within the horizontal axis, which includes the peripheral zone from base to midgland and apex. As a result of the transrectal approach, an increasing sensitivity and specificity for cancer detection was found for the prostate apex compared to the base [11,17]. A similar distribution was found in the present study, with the best sensitivity in the midgland and apex region (65–62%) and the lowest accuracy at the gland’s base (24.8%). The difference in sensitivity may depend on the prostate volume, which is larger at the glands base than at the apex. Therefore, compression and decompression of the tissue is facilitated at the apex. Even though RTE was found to improve PCa detection in the present study cohort, the observed sensitivity (50.4%) is too low to meaningfully impact clinical practice. Other studies report sensitivities in the range 71.4–100%[18,19]. Previous studies have suggested a better specificity of elastography in patients with cancerous lesions ≥5 mm [8,11,12]. False positive results as a result of inflammation, calcification and fibrosis may explain the reported sensitivity and specificity rates. The present study found a correlation between Gleason score and the sensitivity of elastographic PCa detection. To date, whether tissue elasticity is reduced in specimens with a higher Gleason grade remains controversial [11,13]. In the present study cohort, the accuracy of PCa detection was 67.7% in specimens with a pathological Gleason score ≥7 compared to 42.7% when a Gleason score <7 was present. This Gleason score specific detection was not found for GSU.

When evaluating a new imaging technique, the detection limit in regard to tumour size is often discussed [12]. It was reported that elastography can detect lesions <5 mm [16]. Tumour volume has been shown to be a predictor of pathological stage and to correlate with the probability of ECE [20]. If ECE was present in the present study, the histological measured mean tumour volume was significantly higher compared to organ-confined tumours (1.9 cm3 vs 6.1 cm3). Preoperative findings from neither DRE, nor GSU were able to predict tumour volume or the exact extent of PCa [18,21]. The present study investigated whether the IL, defined as largest tumour lesion, detected on imaging correlates with pathological total tumour volume. In the present study, the IL significantly correlated with tumour volume. In addition, we found significantly larger diameters of the IL when ECE was present (13.8 mm vs 21.6 mm). The correlation (P= 0.005) between RTE-IL and corresponding histological IL suggests that this parameter may be useful as predictor for ECE.

As a result of low sensitivity and specificity [22], transrectal ultrasonography is not recommended in the European guidelines for local PCa-staging [14] or the detection of ECE. The inhomogeneous appearance of prostate malignancies on GSU is the main reason not to incorporate this imaging technique. Pallwein et al.[8,13] noted the value of elastographic ‘soft rim artefacts’ as a predictor for ECE. This hypothesis has not yet been investigated. Studies determining ECE by GSU and DRE report variable sensitivity and specificity rates in the range 59–90% and 46–91%, respectively [23–25]. Although, in the present study cohort, RTE improved the detection of ECE by two-fold compared to GSU, a sensitivity of 38.3% does not justify implementing this technique within current guidelines.

RTE in addition to GSU provides a significant, cost-effective improvement of current ultrasonographic imaging, although more improvement is needed to justify implementation into daily practice. Adding contrast-enhanced ultrasonography to current RTE imaging may improve the sensitivity and specificity in the future. A ‘multiparametric approach’ is needed to enhance tumour detection and characterization on ultrasonography based imaging techniques.

In conclusion, the present prospective study shows a significant improvement when adding elastographic techniques to current GSU. Histopathologically confirmed lesions were identified significantly more often using RTE compared to GSU alone. ECE was detected by more than two-fold compared to GSU. Lesions with a higher Gleason grade had a higher likelihood of being identified using RTE. Further improvement is needed to achieve sensitivity values that can justify the implementation of these techniques within current clinical guidelines.

REFERENCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. PATIENTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONFLICT OF INTEREST
  8. REFERENCES
  9. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. PATIENTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONFLICT OF INTEREST
  8. REFERENCES
  9. Supporting Information

VIDEO S1 The video shows the real-time elastography side by side with grey-scale ultrasonography imaging in a transverse plane of the prostate. During gentle compression and decompression, a single blue-coloured lesion is visualized in the ventral zone of the left apex. Pathological analysis revealed a corresponding Gleason score 3 + 4 prostate cancer. Another blue-coloured but unsuspicious induration is visualized in the dorsal right part of the peripheral zone. These indurations are caused by calcifications registered by grey-scale ultrasonography.

FilenameFormatSizeDescription
BJU_10209_sm_VideoS1.avi58485KSupporting info item

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.