Real-Time Elastography of the Prostate

Palpation of organs is one of the oldest clinical examination techniques, for instance, if you think of the palpation of the breast or the digital rectal examination of the prostate, where hard palpable regions are suspicious for cancer. This is the basic principle of real-time elastography, an ultrasound technique, which is able to visualise tissue elasticity. Since prostate cancer features an increased stiffness due to the higher cell and vessel density than the normal surrounding tissue, real-time elastography has been used for several years for prostate cancer detection. This review introduces the different techniques of ultrasound elastography and furthermore summarises its limitations and potentials.


Introduction
Prostate cancer (PCa) is one of the most common cancers in men in the western world. The American Cancer Society estimates 241,740 new cases of PCa and 28,170 cases of death related to PCa in the USA for the year 2012 [1]. Elevated PSA serum values and/or positive findings in digital rectal examination (DRE) are suspicious for PCa. Histopathological evaluation of systematic biopsy (SB) cores is used to confirm or rule out cancer. In SB, the conventional transrectal ultrasound (TRUS) is primary utilised for biopsy guidance and not for cancer detection, because suspicious hypoechoic areas represent cancer in only 9-53% of cases in the grey-scale technique [2].
Despite the low specificity of PSA testing and the low sensitivity of SB, these techniques remain the standard of care for PCa diagnosis, mainly because of the high availability and low costs [3][4][5]. Nevertheless, on the one hand, this strategy misses significant PCa in a high percentage of patients and, on the other hand, detects many insignificant PCa, which leads to overdiagnosis and overtherapy [6].
Because of encouraging technical innovations and developments in prostate imaging, it is now possible to visualise PCa with high sensitivity [7,8]. Beside PCa localisation, modern imaging modalities are capable of providing information about tumour volume, local staging, and cancer aggressiveness, which may be helpful for choosing the most appropriate therapy [8,9].
Multiparametric MRI (mpMRI) and modern TRUS techniques are currently used for PCa detection [7,10,11]. In contrast to mpMRI and contrast-enhanced TRUS, real-time elastography (RTE) is a cheap and noninvasive tool, which can be performed by both radiologists and urologists.

Techniques
Ophir et al. described the principles of elastography in 1991 [12]. Krouskop et al. found a significant difference in stiffness between normal and neoplastic prostate and breast tissue [13]. To reduce the need for time-consuming calculations, Pesavento developed a fast cross-correlation technique. This technique became the basis for RTE under real-time conditions [14]. Currently, there are 2 different techniques which are used to demonstrate tissue elasticity of the prostate: strain elastography (SE) and shear-wave elastography (SWE).

Strain Elastography (SE)
. SE is the most widely used technique to estimate tissue stiffness of the prostate. Elastograms are obtained by slight compression and decompression of the prostate, which is manually induced with the transrectal probe by the operator. To visualise the distribution of tissue elasticity, images are colour-coded, for example, from red (soft tissue) to blue (hard tissue). The elastography box should cover the entire gland and the surrounding tissues but avoid the bladder [15]. (SWE). In contrast to SE, SWE does not require a manually induced compression of the prostate. The tissue elasticity is calculated by measuring the velocity of shear-waves, which is different in soft and hard tissues of the prostate. Similar to SE, SWE can colour-code the distribution of tissue elasticity. Furthermore, this technique is able to generate an absolute value of tissue stiffness in kilopascal. The image can cover half of the gland in axial planes; thus, each side of the prostate should be examined separately [15,16].

Quantification.
Obtaining a measurable value indicating the degree of tissue stiffness has been shown to be useful for differentiating between benign and malignant hard lesions, for example, in the breast [17]. To possibly reduce false positive findings of RTE of the prostate, Zhang et al. investigated the strain ratio index of SE, which is a semiquantitative measurement tool comparing the strain value of the target lesion with the background tissue [18] (Figure 1). They reported a significant difference of peak strain index values between benign and malignant lesions ( < 0.01). Furthermore, the cut-off of 17.44 peak strain index value yielded the highest sensitivity (74.5%) and specificity (83.3%) with an AUC of 0.90 for discriminating PCa.
In contrast to SE, SWE can provide absolute values in kilopascals. Regarding the negative predictive value, cut-off values of 35 and 37 kPa were reported for SWE, providing the best performance with a sensitivity, specificity, PPV, and NPV of 63%, 91%, 69.4%, and 91%, respectively [15,16,19].  [20]. They included 60 PCa patients before radical prostatectomy (RPE). All men were examined by a novice and an expert observer. Results of the novice observer were validated with those of the expert. The study group demonstrated valid results after 30 examinations for the 8sector model when being trained by an expert. Regarding precision, more examinations enhanced the quality of the results as they could use the 16-sector model. Overall, they concluded that RTE of the prostate can be learned quickly by an unexperienced examiner and the examination itself is not very time-consuming.
However, the manual compression and decompression are relatively uncontrolled elements of SE with wide variability. Thus, images affected by artefacts have been found in up to 32% of cases [21]. Standardised compression using real-time balloon inflation has improved results, showing sensitivity and specificity up to 72.5% and 97.7%, respectively [22].  [23]. To rule out pathologies, the study group included patients with no signs of inflammation, normal findings in the DRE, total PSA serum values <4 ng/mL, PSA ratio >0.18, PSA density <0.04, no structural alterations in the peripheral zone (PZ) in the conventional TRUS, and prostate volumes between 20 cc and 100 cc. The normal PZ of the prostate was of intermediate elasticity ( Figure 2), while the inner gland showed more heterogeneity. Furthermore, they demonstrated that elasticity patterns of the prostate which are investigated with SE depend on physiological changes, demonstrating an increasing stiffness with growing age and volume, especially in the inner gland ( Figure 3). The absence of the soft rim artefact may indicate extracapsular extension (ECE) ( Figure 4). These findings are in line with those of Correas et al., who reported for SWE that the entire prostate in young healthy patients is of homogeneous soft appearance and that the PZ remains soft and homogeneous, while the inner gland becomes heterogeneous, in the case of benign prostate hypertrophy (BPH) [24].

False Positive Findings.
Benign entities like prostatitis, fibrosis, atrophy, adenomyomatosis, and BPH may be associated with increased tissue stiffness and therefore may be difficult to distinguish from PCa. This can be responsible for low positive predictive values to the point of only 39%, as reported by our group when studying men with PSA serum values <4 ng/mL [25] ( Figure 5).

False Negative Findings. Langer et al. divided cancers
into those with dense architecture on histology and into those with sparse architecture [26]. In particular, cancers with predominant Gleason pattern 3 (e.g., 3 + 3 or 3 + 4) may be of sparse architecture consisting of a mixture of normal tissue and cancerous tissue or possessing glands with dilated lumina and are therefore soft. For imaging, in general, this histological kind of tumour composition may be a problem. For diffusion-weighted MRI (DWI), Langer et al. demonstrated that all "invisible" tumours in DWI had predominant Gleason pattern 3 and showed sparse architecture on histology [26]. For RTE, we revealed similar results, since RTE visualises cancer due to the higher cell density compared to the normal surrounding tissue [13,27]. Nearly all cancers missed on RTE in our study population with a significant tumour volume were also of sparse architecture and had predominant Gleason pattern 3. Nevertheless, sensitivity of PCa with a Gleason score ≥4 + 3 was very high, since a predominant Gleason pattern 4 or 5 indicates dense tumours [27] (Figure 6).

RTE Targeted Biopsy
Many study groups have shown the usefulness of guiding the biopsy needle directly into lesions of the prostate which are suspicious for PCa on RTE (Figure 7). In a series of 230 men, Pallwein et al. compared a 5-core RTE targeted biopsy with a 10-core SB and demonstrated a sensitivity of 84% for RTE. The PCa detection rate per patient was not significantly higher for the targeted approach compared to the systematic approach (30% versus 25%; McNemar's test: = 0.134), but an RTE targeted core was 2.9-fold more likely to be cancer positive than a systematic core (McNemar's test: < 0.001) [28].
Similar results were obtained in a study of Aigner et al., which included only patients with PSA serum levels <4 ng/mL [25]. Targeted biopsies with a maximum number of 5 cores were performed in PCa suspicious areas on RTE, followed by 10-core SB. The PCa detection rate per patient was nearly equal (21.3% versus 19.1%), whereas the cancer detection rate per core was 4.7-fold greater for targeted compared to systematic biopsy. Furthermore, a high number of patients were PCa positive (27 of 94; 28.7%), even in this low PSA group, which may suggest that RTE detects PCa independently of PSA serum values. An interesting RTE targeted biopsy scheme was performed in a cohort of 178 men by Brock et al., who divided the prostate into sectors and limited the targeted approach to a 10-core biopsy to avoid additional cores and to reduce the overall number of biopsy cores [29]. If RTE did not visualise a suspicious area in a sector, the biopsy core was taken systematically; otherwise, an RTE targeted biopsy was performed. In their randomised prospective study, they compared the findings of these 178 patients with another study group of 175 men undergoing a 10-core conventional TRUS-guided biopsy. They showed a significantly higher PCa detection rate in the RTE group than in the control group (51.1% versus 39.4% ( = 0.027)). Overall sensitivity and specificity of PCa detection were 60.8% and 68.4%, respectively, for the RTE targeted approach. A recently published study from Salomon et al. underlines the potential of RTE targeted biopsy [11]. They analysed 1024 patients who consecutively underwent a 4-core RTE targeted biopsy in addition to a randomised 10-core conventional TRUS-guided biopsy. PCa detection rate of the combined approach, of the 10-core SB, and of the 4-core RTE biopsy was 46.2% ( = 473), 39.1% ( = 400), and 29.0% ( = 297), respectively. RTE targeted biopsy also detected a further 72 patients with PCa who were not detected by systematic 10core biopsy. Therefore, the additional use of RTE led to an incremental value of 18.3%. Of vital importance is the fact that 34 patients harboured significant PCa (primary Gleason pattern 4 or 5) diagnosed by RTE biopsy only.
Targeted biopsy under RTE guidance seems to have the advantage of the reliable detection of significant disease [30]. According to Nygård et al. the addition of RTE targeted cores to standard SB raised the negative predictive value from 79% to 97% for high-risk PCa. They concluded that a positive RTE is an independent marker of the detection of high-risk PCa, and a negative RTE argues against such a diagnosis. Nevertheless, both Salomon et al. and Nygård et al. did not recommend RTE targeted biopsy alone, since this approach missed a high proportion of patients with PCa in their studies [11,30].
Despite these promising results, there are heterogeneous results in the literature for the general benefit of RTE targeted biopsy for PCa diagnosis. For instance, Taverna et al. found a relatively low sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) of 24.4%, 65.7%, 21.9%, and 68.6%, respectively, in 102 men with PSA serum values ranging from 2.5 ng/mL to 10 ng/mL for RTE. They concluded that RTE cannot be recommended for standard PCa diagnosis [31]. Therefore, prospective multicentre studies would be desirable to show the true value of RTE targeted biopsy. A first big multicentre study was already initiated by 10 European centres (ESUI 12/01 HI-RTE); the first results are expected in the coming months.
Nevertheless, nearly all studies have shown that the RTE targeted approach detects high-risk PCa more reliably than SB, requires a reduced number of cores for PCa detection, and enhances the overall sensitivity in the combined biopsy setting.

RTE before Radical Prostatectomy (RPE)
Since SB may miss PCa, it serves as an incomplete gold standard to calculate the overall diagnostic accuracy for other imaging modalities [32]. A comparison with RPE specimens allows for the exact localisation of PCa, but the investigator might be biased by his knowledge of the patient's disease.

PCa Index Lesion.
One of the key requirements of imaging is to visualise the PCa index lesion, since this may determine the clinical prognosis [38]. Walz

Tumour Size and Tumour Volume.
Our study group investigated PCa detection rates for RTE with regard to tumour size and tumour volume in a whole-mount step section analysis [27]. Concerning the tumour size, RTE detected 6 out of 62 cancer lesions with a maximum diameter of 0-5 mm (9.7%), 10 out of 37 with a maximum diameter of 6-10 mm (27%), 24 out of 34 with a maximum diameter of 11-20 mm (70.6%), and 14 out of 14 with a maximum diameter of >20 mm (100%). Concerning the tumour volume, RTE detected 40 out of 48 cancer lesions with a tumour volume ≥0.2 cm 3 (83.3%) and 31 out of 34 with a tumour volume ≥0.5 cm 3 (91.2%). Roethke et al. found similar results for endorectal MRI; therefore, we may conclude that imaging of PCa is generally influenced by tumour size and tumour volume [40].

Local Staging.
For the prediction of ECE, Brock et al. showed a sensitivity of 38% and a specificity of 96% for RTE ( Figure 6). This represents an improvement in comparison to the grey-scale TRUS but still remains inadequate for clinical practice [35]. In contrast, Pelzer et al. found a sensitivity of 79% and a specificity of 89% for the detection of ECE by RTE [41]. Zhu et al. showed a sensitivity for PCa detection confined to the capsule of 51.6% and of 79.3% with ECE and stated that this might be related to the fact that tumours with larger diameters, which were more likely to be detected by RTE, were more likely to extend throughout the capsule [42].

RTE in Comparison to Multiparametric MRI (mpMRI)
It is of great interest to compare RTE with MRI for PCa diagnosis, since MRI is a very promising imaging technique for the prostate and is also considered a gold standard (Figure 8).

Detection Rates and Localisation.
In a small series, Aigner et al. investigated the potential of RTE in comparison with T2-weighted endorectal MRI (T2w-MRI) for PCa detection using a biopsy specimen as the reference standard and showed similar results [43]. RTE showed sensitivity rates and negative predictive values per patient of 84.6% and 86.7%, respectively, a T2w-MRI of 84.6% and 83.3% for MRI. Sumura et al. used whole-mount step sections as the reference standard and reported a superiority of RTE detection rates to those of MRI (74.1% versus 47.4%) and nearly equal detection rates for RTE alone on both the anterior side (75.0%) and the posterior side (73.7%) of the prostate [44]. Furthermore, a higher tumour detection rate for RTE was observed for tumours with a higher Gleason score and larger tumour volume.
While the study conducted by Sumura et al. used T2w-MRI and contrast-enhanced MRI (ceMRI) only, Pelzer et al. used a multiparametric approach including T2w-MRI, DWI, and ceMRI following the recommendations from the European Society of Urogenital Radiology (ESUR) [41,46]. Summarising their results, RTE showed advantages in apical and middle parts of the prostate for PCa detection, whereas mpMRI provided advantages in the glands' base and transitional zone (TZ). Both RTE and mpMRI had limitations, particularly in basal and ventral parts. Furthermore, most of the undetected tumours were of low tumour volume and Gleason score.
These data are in line with a study by Junker et al., who also compared RTE findings with those of mpMRI in a whole-mount step section analysis [45]. Histopathology revealed 61 cancer lesions ≥0.2 cm 3 . RTE detected 78% cancer lesions in the PZ and 18.2% in the TZ, while mpMRI detected 90% and 72.2%, respectively. Significant differences between both modalities were found for the TZ and anterior parts in prostates with a volume >40 mL ( < 0.05). Detection rates for high-risk PCa (Gleason score ≥ 4 + 3) and cancer lesions with volumes >0.5 cm 3 were 93.8% and 80.5%, respectively, for RTE, and 87.5% and 92.7%, respectively, for mpMRI. The authors concluded that RTE and mpMRI detected high-risk PCa with high sensitivity, but mpMRI seems to have advantages in tumour volume assessment and in the detection of PCa in the TZ and in anterior-localised PCa within prostate glands with volumes of >40 mL.

Considerations for Focal Therapy.
Curiel et al. assessed the volume of HIFU lesions in prostate glands and found that elastography generally underestimated tumour volume in comparison with MRI [47]. These findings are in agreement with our observations: Spearman's rank correlation was 0.72 for mpMRI and only 0.46 for RTE when comparing volume measurements of cancer lesions with histopathology [45]. The explanation might be the usage of an endfire probe for RTE in our population, which scans the prostate at different angles compared to the whole-mount step sections or to mpMRI (Figures 9 and 10). When additionally keeping in  mind the results for detecting the index lesion and for lowrisk PCa, which may be candidates for focal therapy, it seems that RTE alone does not have the potential for this approach [11,38,39,48].

Summary
Currently, PCa diagnosis is in a paradigm shift. On the one hand, imaging was and is still used to improve the PCa detection rate. On the other hand, some urologists have more recently indicated the desire to detect clinically significant cancers only in order to prevent overdiagnosis and overtreatment, which is also associated with higher costs and complications. RTE may be particularly suitable for this purpose, since RTE is of limited value in the detection of small cancers and may miss carcinomas with a primary Gleason pattern of 3. In the end, it is up to the clinician to decide which approach is to be favoured. Therefore, we do not summarise advantages and disadvantages of RTE but report the characteristics only: (xi) it has low evidence level due to the lack of multicentre studies.
The mentioned characteristics have led to the following protocol for PCa diagnosis at our institution: in the primary biopsy setting we combine RTE targeted and systematic biopsy. RTE targeted cores are taken only in case of cancersuspicious lesions. In the case of a prior negative biopsy with the combined approach but ongoing cancer suspicion, we investigate the prostate with mpMRI at 3T. If there are cancer-suspicious lesions on mpMRI, patients undergo an ultrasound/mpMRI image fusion guided biopsy ( Figure 13).