A positive real-time elastography is an independent marker for detection of high-risk prostate cancers in the primary biopsy setting

Authors


Abstract

Objective

  • To evaluate the performance of real-time elastography (RTE) in an initial biopsy setting.

Patients and Methods

  • In the period from February 2011 to June 2012, 127 consecutive patients were included in the study.
  • We used a Hitachi Preirus with Hi-RTE module, a prostate end-fire transrectal probe was used for RTE and for targeted biopsies, and a simultaneous biplane probe was used for the standard systematic biopsies.
  • The peripheral zone of the prostate was divided into six regions, and each biopsy obtained was referred to a specific region.
  • All patients were first examined with RTE and, if cancer was suspected, targeted biopsies were taken. A standard systematic 10-core biopsy was then taken in all patients.

Results

  • In all, 64 (50%) patients were diagnosed with prostate cancer in the initial biopsy setting. Three patients were diagnosed solely on RTE-targeted biopsies, 31 were found only in systematic biopsies, and 30 were correctly diagnosed with both methods.
  • In the RTE-positive group there was a significantly higher frequency of positive cores, a lower prostate volume, a higher Gleason score, and a higher fraction of cancer tissue in each core.
  • In a multiple regression model RTE was an independent marker for high-risk cancer.
  • The sensitivity of 42% for all prostate cancers increased to 60% for high-grade prostate cancers.
  • Similarly, the negative predictive value increased from 79% to 97%. An additional eight patients were diagnosed with prostate cancer during the study period.

Conclusions

  • A positive RTE is an independent marker for detection of high-risk prostate cancer, and a negative RTE argues against such.
  • RTE with targeted biopsies cannot replace systematic biopsies, but provides valuable additional information about the tumours.
Abbreviations
(N)(P)PV

(negative) (positive) predictive value

(RA)RP

robot-assisted radical prostatectomy

PZ

peripheral zone

RTE

real-time elastography

TZ

transitional zone

US

ultrasound/ultrasonography

Introduction

Prostate cancer is the most common cancer in men, accounting for 4299 new cases in Norway in 2009. The incidence was 110 per 100 000 (world standard), and is increasing. The principal tools for detection of prostate cancer are serum PSA level and DRE. PSA and DRE have low specificity, and they do not differentiate between aggressive and indolent disease. Currently the diagnostic standard of care is to perform B-mode TRUS-guided systematic biopsy of the prostate [1, 2]. According to European Association of Urology guidelines there is a need for at least two series of biopsies with at least 10 cores in the first series and 12 cores in the repeat biopsy series to exclude prostate cancer the cause of an elevated PSA level [3]. Even after two series of biopsies there will still be men with undetected but significant prostate cancer in the group. On the other hand, due to the low specificity of PSA testing, many men will have to undergo unnecessary prostate biopsies. There is a definite need for improvement in the diagnostic tools in prostate cancer and several new methods are emerging. Real-time elastography (RTE) is an ultrasound (US) method that can be helpful in detecting prostate cancer [4, 5]. The advantage of RTE is the possibility for the operator to place the biopsy needle into a suspicious area at the same session without the need for additional assessments or the use of contrast enhancement.

In the present study, our primary goal was to compare the ability of RTE-targeted biopsy with the standard 10-core systematic biopsy for detection of clinically significant prostate cancer in the initial biopsy setting, and to evaluate whether RTE increases the detection rate of high-risk cancers.

Patients and Methods

The study was performed in the period from February 2011 to June 2012. The 127 patients gave their oral and written consent to participate in the study after receiving written and oral information about the study. The study protocol was approved by the Regional Committee for Medical and Health Research Ethics (REC) in Western Norway. The patients' characteristics are given in Table 1. The patients were included using active inclusion and the inclusion criteria were:

  1. PSA level of 3.0–25.0 ng/mL and/or pathological DRE
  2. Age <75 years
  3. No prior biopsies within the last 5 years
  4. Amenable for radical treatment
Table 1. The patients' characteristics
VariableValue
Mean (sem; median, range): 
Age, years64.2 (0.6; 65, 38–76)
PSA level, ng/mL9.2 (0.4; 7.2, 2.2–24.4)
N (%): 
Reason for referral: 
Elevated PSA only68 (54)
Positive DRE by GP (only)3 (2)
Family history of prostate cancer5 (4)
Symptoms51 (40)
DRE evaluation by Urologist: 
Normal/ BPH87 (69)
Suspicion of cancer32 (25)
Inconclusive8 (6)
Mean (sem; median, range)prostate volume, mL61.4 (2.8; 53, 23–186)
Positive (reproducible) RTE (low strain–hard lesions), n (%)86 (68)

All patients were referred by GPs, either because of an elevated PSA level or because of a suspicious DRE. Based on the referral from the GPs, consecutive patients meeting the inclusion criteria were offered inclusion in the study. In all, 127 patients were subsequently included. Very few patients (less than five) refused to participate in the study.

US, RTE, Targeted Biopsies, Standard Biopsies and Report Forms

All patients were examined in the left lateral decubitus using a Hitachi Preirus US machine. The Hitachi Preirus was equipped with an RTE module (Hi-RTE). A V53W transrectal end-fire probe was used for RTE and targeted biopsies. A CC531 transrectal simultaneous biplane probe was used for standard systematic biopsies. RTE displays a colour-coded strain map called an elastogram, which is superimposed on the B-mode images in real-time. RTE visualises strain in the tissue using the extended combined autocorrelation method (ECAM) [6, 7].

The examination was performed using the default settings of the elastography software (E-dyn 4, frame rejection 6, noise rejection 4, smoothing 2, and persistence 3). Minimal compression and decompression of the prostate performed by the transrectal probe produced the RTE images. The machine gives feedback on the screen about the quality of the compression/decompression cycles. With training it is not hard to get consistent elastograms. The elastograms were presented simultaneously with the B-mode US images on a split-screen monitor. Hi-RTE software displays elastograms using a scale from red (highest strain; soft), through green (average strain; intermediate), to blue (low strain; hard). Figure 1 shows an elastogram suspicious of cancer.

Figure 1.

A typical elastogram (left) and the B-mode image (right). It shows a hard lesion (blue) in the left PZ at the apex. The lesion is marked for measurement (+ signs), and it is ≈10 × 11 mm. The red area is the urethra; the anterior aspect of the prostate cannot be evaluated by RTE (the blue area anteriorly). Three targeted biopsies were obtained from the lesion in the left PZ and all three revealed a Gleason grade 5 + 4, score 9 tumour.

The prostate was divided into six peripheral zone (PZ) regions excluding the transition zone (TZ) from the investigation [8]. Each region was examined for cancer suspicious lesions with both methods starting with B-mode. Prostate volume was calculated after measuring height, width, and length of prostate using the Hitachi software. The whole gland was then examined using RTE. If there were any suspicious (blue) areas, the examiner (Y.N.) would check their reproducibility. All reproducible suspicious areas would then undergo targeted biopsies. A maximum of five targeted biopsies were obtained. Irreproducible areas or fluctuant areas were considered inconclusive and would not undergo targeted biopsies. In the statistical analysis they were considered benign.

In patients with pathological RTE the location of the suspicious areas and the number of the targeted biopsies were registered. A standard 10-core biopsy was then taken in all patients. The 10-core systematic biopsy consisted of TRUS-guided standard sextant biopsy, supplemented with four lateral cores from the mid-prostate and the apex. For the biopsy scheme see Fig. 2. The urologist (S.A.H.) performing the standard biopsies did not have any knowledge of the results from the RTE examination. Hence, the patients served as their own control group for the performance of RTE and targeted biopsies vs systematic 10-core biopsies.

Figure 2.

The different prostatic zones used in the study (posterior view) and the standard biopsy pattern used in the study (posterior view).

A standardised clinical report form was used during the procedure and kept for later analysis. Prostate volume, DRE findings, and RTE findings were marked on the clinical report form. Individual biopsy cores were numbered and assigned to a prostate region before being examined by a uro-pathologist (O.J.H., K.G.).

All patients received antibiotic prophylaxis with ciprofloxacin (1 g orally) before the procedure. In all patients, 6 mL lidocaine (10 mg/mL) was administered as periprostatic infiltration anaesthesia.

Patients with a negative primary biopsy were followed up either because of urinary symptoms or because of a persistent, high suspicion of prostate cancer. A repeat biopsy was performed in patients with persisting indication for biopsy.

The biopsy and radical prostatectomy (RP) specimen results were retrieved from the pathology reports prepared by two study uro-pathologists (K.G., O.J.H.). The total core length, length of cancer tissue in each biopsy, and biopsy Gleason grade and score were recorded. In patients with several positive cores with different Gleason scores the highest Gleason score was used. In patients treated with robot-assisted RP (RARP) the information from the final pathology reports, based on whole-mount section histopathology and detailed in separate schemes, were used for the final analysis of the RTE performance.

Statistics

Standard descriptive statistics were used. Mean values are presented as the mean (sem). Sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and accuracy were calculated on a per-region basis. As gold standard we employed RTE-targeted biopsies, systematic 10-core biopsies, and repeat biopsies of 12 systematic cores including two cores from the TZ. In the 27 cases where patients underwent RARP, we also used final pathology if it provided additional information.

Comparisons between different groups were performed by cross-tables and exact chi-square test, Mann–Whitney U-test, and t-test for categorical, ordinal and continuous data, respectively.

The multiple logistic regression models were performed, without pre-selection of the variables, in a backward Likelihood Ratio test manner. A P < 0.05 was considered to indicate statistical significance.

Results

Prostate cancer was diagnosed in 64 (50%) of the 127 patients during the initial biopsy session. As all patients had a systematic 10-core biopsy series, the total number of cores was 1270. In all, 86 patients had suspicious lesions on RTE and were biopsied with targeted biopsies; a median of three cores was obtained. A total of 287 targeted biopsy cores were obtained. In only three patients was prostate cancer detected solely by RTE-targeted biopsies. In all, 30 of the 64 prostate cancers detected at initial biopsy were diagnosed on both RTE-targeted biopsies and systematic biopsies, and 31 of 64 on systematic biopsies alone. However, the frequency of positive cores was significantly higher in the RTE-targeted biopsies than in the standard systematic biopsies (80/287 vs 236/1270, P < 0.001). There was a trend towards a higher fraction of cancer in the targeted biopsies, with an average of 42% of the total core length compared with 33% in the standard biopsies (P = 0.087). The primary biopsy results are shown in Table 2 and the zone distribution of positive cores of the different biopsy methods is given in Table 3. Table 4 shows the tumour characteristics found on biopsy.

Table 2. Initial biopsy results
VariableValue
  1. *The number of patients that underwent RTE-targeted biopsies was 86, All 127 patients underwent systematic biopsies.
Total number of RTE-targeted biopsies*287
Total number of systematic biopsies1270
Median (range) number of RTE-targeted biopsies3 (0–5)
Median number of systematic biopsies10
N (%): 
Patients with prostate cancer64 (50)
Patients with prostate cancer in RTE-targeted biopsies only3 (5)
Patients with prostate cancer in systematic biopsies only31 (48)
Patients with prostate cancer in both RTE-targeted AND systematic biopsies30 (47)
Table 3. The zone distribution of positive biopsies
 ApexMid-prostateBase>1 regionTotal
N (%):     
 RTE-targeted biopsy series:2 (6)4 (12)4 (12)4 (12)14 (42)
Right8 (24)4 (12)2 (4)2 (6)16 (48)
Left2 (6)001 (3)3 (9)
Bilateral12 (36)8 (24)6 (9)7 (21)33 (100)
Total     
 Systematic 10-core biopsy series:     
Right3 (5)5 (8)4 (7)8 (13)20 (33)
Left3 (5)1 (2)2 (3)5 (8)11 (18)
Bilateral2 (3)1 (2)027 (44)30 (49)
Total8 (13)7 (11)6 (10)40 (66)61 (100)
Table 4. Tumour characteristics and treatment
VariableTotalRTE-targeted biopsies resultsSystematic 10-core biopsies results
n = 64n = 33n = 61
N (%)   
Clinical tumour stage (cT)   
T1c28 (46)  
T2a-b16 (25)  
T2c10 (16)  
T3a7 (11)  
Gleason Score   
5 (3+2)1 (2)01 (2)
6 (3+3)20 (31)13 (39)19 (31)
7a (3+4)20 (31)6 (18)24 (39)
7b (4+3)10 (16)7 (21)6 (10)
8 (4+4)6 (9)3 (9)4 (7)
9 (4+5)5 (8)3 (9)5 (8)
9 (5+4)2 (3)1 (3)2 (3)
D'Amico risk stratification:   
High risk27 (42)  
Intermediate24 (38)  
Low risk13 (20)  
Treatment   
RARP23 (36)  
External beam radiation therapy29 (45)  
Active surveillance12 (19)  

According to the D'Amico criteria [9] 42% of the patients were in the high-risk group, 38% were in the intermediate-risk group, and 20% had low-risk prostate cancer.

A subgroup analysis was performed between the cancers detected on both RTE-targeted and systematic biopsies (group I, n = 30). The prostate cancers only found on systematic 10-core biopsies constituted group II (n = 31). In the systematic biopsies, the fraction of cancer of the total core length was 41% in group I and 26% in group II (P = 0.01). In group I, the mean (sem) number of systematic positive cores was 5.1 (0.5) vs 2.7 (0.3) in group II (P < 0.001). Furthermore, the biopsy Gleason score was significantly higher within group I when comparing with the systematic biopsies only. If the biopsy results from the targeted biopsies were added to group I and the highest Gleason score was used, the differences became even more pronounced (Table 5). Furthermore, Group I had a significantly lower mean prostate volume than group II, at 44.0 (2.6) vs 68.0 (3.5) mL (P < 0.001).

Table 5. Impact of RTE-targeted biopsies on Gleason score in 61 patients detected by systematic 10-core biopsy
Gleason scoreRTE-positive patients (Group I) Only systematic biopsy results (n = 30)RTE-positive patients (Group I) Systematic and targeted biopsy results (n = 30)RTE-negative patients (Group II) Systematic biopsy results (n = 31)
  1. *Median Gleason Score was 7a (3+4) in both groups; however, the rank distribution within the groups are significantly different (P = 0.007, Mann–Whitney U-test). **Median Gleason Score was 7b (4+3) in group I and 7a (3+4) in group II. The rank distribution within the groups are significantly different (P = 0.001, Mann–Whitney U-test).
N (%):   
5 (3+2)001 (3)
6 (3+3)5 (17)4 (13)14 (45)
7a (3+4)14 (47)*10 (33)10 (32)**
7b (4+3)2 (7)6 (20)**4 (13)
8 (4+4)3 (10)4 (13)1 (3)
9 (4+5 or 5+4)6 (20)6 (20)1 (3)

A multiple regression model was used to identify markers for high-risk cancers. Five different parameters were entered for analysis; prostate volume, positive RTE (low strain), symptoms, study period, and age. The PSA level and DRE findings could not be included in such a model, as these are key criteria in D'Amico's risk stratification [9].

A positive RTE and low prostate volume were independent markers for high-risk cancers. Details are shown in Table 6.

Table 6. Logistic regression analyses for detection of high-risk prostate cancer in the primary biopsy setting
VariablesSimpleMultiple
UnadjustedFully adjustedFinal model
OR 95% CI*P**OR 95% CI*P**OR 95% CI*P**
  1. High-risk cancer of the prostate according to the risk stratification proposed by D'Amico [9]; threshold for dichotomy placed at median volume 53.1 mL; **P values are calculated by the Likelihood Ratio (LR) test, while the CI* are calculated by use of the Wald coefficient under an assumption of a normal distribution. The contradictory results, in regard to statistical significance, given by the CI and the P value for the RTE variable, are explained by these different methods of calculation. However, the LR-test should be regarded as the most accurate.
Age (continuous)1.01 (0.94,1.08)0.8601.03 (0.96,1.11)0.376  
Study period (second vs first half)0.77 (0.33, 1.81)0.5451.07 (0.41,2.77)0.891  
Symptoms (yes vs no)1.03 (0.43, 2.45)0.9441.23 (0.47,3.23)0.673  
Positive RTE (yes vs no)5.89 (1.31, 26.4)0.0054.56 (0.95,21.8)0.0304.20 (0.89,19.6)0.038
Prostate volume (≥53.1 vs <53.1 mL)7.69 (2.44, 25)<0.0016.67 (2.04,20)<0.0016.25 (2.04,20)<0.001

During the 6-month period after primary biopsy, four patients underwent TURP for symptomatic BPH with a benign histology, and 36 patients had a repeated systematic TRUS-guided 12-core biopsy, of which two cores were sampled from the TZ. At histology, another eight patients were diagnosed with prostate cancer. Six of these patients were low-risk cancers and two were intermediate risk. One of the intermediate-risk prostate cancers was localised at the base of the prostate; this was detected after multiparametric MRI-guided TRUS biopsy. Four of these patients were treated with RARP, and four were followed according to an active surveillance protocol. In all, 72 (57%) of 127 patients were diagnosed with prostate cancer.

For calculations of sensitivity, specificity, PPV, NPV and accuracy of RTE, we divided the prostate into six regions. In all, 762 regions in 127 patients were analysed. All these regions were analysed for all cancers and for high-grade cancers, using the information from the systematic biopsies, the RTE-guided biopsies, repeat biopsies, and pathology reports from patients treated by RARP. The sensitivity increased from 42% for all prostate cancers to 60% for high-grade prostate cancers. Further results are shown in Table 7.

Table 7. Sensitivity, specificity, NPV, PPV and accuracy of RTE for the detection of prostate cancer
 Sensitivity, %Specificity, %NPV, %PPV, %Accuracy (%)
All prostate cancers (all Gleason scores)4283794972
High-grade prostate cancer (Gleason 7b–10)6080972078

Discussion

One of the main findings of the present study was the capability of RTE to identify high-risk prostate cancer. Cancers identified with RTE showed a significantly higher Gleason score than those with a negative RTE. This is consistent with the findings of Nelson et al. [10] and Aigner et al. [11], while Brock et al. [12] were not able to find an association between RTE and Gleason score. This discrepancy may be explained by different distributions of Gleason scores in the patient series. In the series of Brock et al. [12], most of the patients were Gleason score 4–6. In the present study most patients were, on the contrary, intermediate- or high-risk cancers. Nelson et al. [10] found a positive association between RTE and intermediate- and high-grade cancers.

Another important result in the present study was the relation of both positive RTE and prostate volume with prostate cancer. The patients with RTE-positive cancers had significantly smaller prostates (mean 44 vs 68 mL). In patients for whom there was a suspicion of prostate cancer, the combination of a small prostate and a positive RTE did lead to an increased risk of being diagnosed with high-risk prostate cancer. In RTE-positive cancers we found significantly more positive cores (5.1 vs 2.7), and the fraction of cancer tissue in the cores was significantly higher in this group (41% vs 26%). The reference standard for the diagnostic performance of RTE consisted of all the biopsy results including repeat biopsy within 6 months. This reference standard is close to the true prevalence of prostate cancer in the study cohort. The sensitivity of a 42% detection of prostate cancer on a per-region basis is higher than the 24% reported by Taverna et al. [13] and 31% in the study of Cochlin et al. [14], but lower than the reported sensitivity of 61% in the study by Brock et al. [12]. This discrepancy in sensitivity may be explained by differences in technology, study design, and study population, as well as well-known inter-observer variability. Although the skilled examiner usually achieves higher sensitivity, a novice achieves valid results after some 50 examinations when he is properly trained [15]. As RTE is a real-time examination there will be differences in the interpretation of the images produced. As the present study is a single examiner study, the inter-observer variability could not be addressed.

In our experience RTE does not seem to be a good method to identify cancer lesions in large glands, which demonstrates one of the challenges for RTE. In glands with a large TZ, RTE shows a lower sensitivity. In patients with BPH the inner gland compresses the PZ. The PZ then appears thinned out and difficult to examine. The normal RTE pattern of BPH in large prostates is unevenly inelastic in most cases, producing hard lesions indistinguishable from cancer [16].

The low sensitivity precludes the use of RTE-targeted biopsy alone, and at present it has to be combined with systematic biopsies. However, in the present study RTE showed a high specificity of 83%, and a high NPV of 79%, for detection of prostate cancer. Although RTE showed too low a sensitivity to be used alone, RTE is of clinical value because it showed high specificity. The specificity remained high (80%) in the present study also for high-grade tumours. Furthermore, in a high-grade setting, the sensitivity increased to 60%. However, most important was the NPV of 97%. In our opinion, if verified in later studies, this could make a place for RTE routinely used as a component of multiparametric US assessment for excluding high-risk prostate cancer in a screening setting.

In all, 50% of the patients were diagnosed with prostate cancer at initial biopsy using systematic biopsies and RTE-guided biopsies. This is comparable to the 51% detection rate published by Brock et al. [12] in their study of combining RTE-targeted and systemic biopsies. They found a significantly higher cancer detection rate using RTE-guided systematic biopsies vs B-mode TRUS-guided systematic biopsies (51% vs 39%).

In the present series, the number of patients with high-risk cancer (42%) was also higher than in most cohorts. In a population-based study of initial management of prostate cancer in Norway in 2004, Hernes et al. [17] found 42% high-risk cancers among 1650 patients amenable for curative treatment. In Norway a population-based screening programme has not been implemented, but an increased use of erratic screening has led to an increased number of PSA-detected cancers. As many patients have only measured their PSA level once, there might be more patients with larger tumours than in heavily screened populations.

We have evaluated RTE with targeted biopsies and compared it with standard systematic biopsies. We did both examinations in all patients. Some studies have investigated RTE with targeted biopsies and compared it with standard biopsies. Aigner et al. [11] reported similar detection rates using RTE and targeted biopsies compared with 10-core standard systematic biopsies, despite the use of fewer cores.

We found that the cores sampled by RTE-targeted biopsies frequently had more cancer, and also a higher fraction of cancer tissue in each biopsy. This will for some patients make a difference in treatment options. For instance, in one patient we found, by standard systematic biopsies, a Gleason score 6 (3+3) tumour in one of 10 biopsies; the RTE-guided biopsies revealed a Gleason score 8 (grade 4+4) tumour. This patient was treated with RARP. Without the RTE-guided biopsies he would most likely have be recommended for active surveillance.

The study is limited by being a single-centre, single examiner study and by the number of patients. The RTE method is observer dependent. Although there is a learning curve for RTE, valid results may be achieved after ≈50 examinations [15].

In conclusion, a positive RTE is an independent marker for detection of high-risk prostate cancer and a negative RTE argues clearly against the presence of such. The relationship of RTE and prostate size in the patients with prostate cancer may imply a better diagnostic performance of RTE in more normal-sized prostates. RTE with targeted biopsies cannot replace systematic biopsies, but RTE provides valuable additional information in the initial biopsy setting.

Acknowledgement

The authors acknowledge statistician Geir Egil Eide at the Centre for Clinical Research, Haukeland University Hospital for his help in the preparation of the manuscript.

Conflict of Interest

The authors have nothing to disclose, and the study was carried out with funding from the institutions mentioned on the title page and by a grant from the URO-BERGEN research foundation. The authors alone are responsible for the content and writing of the paper.

Ancillary