Ghulam Nabi, Department of Urology, Population Sciences Division, Medical Research Institute, University of Dundee, Dundee DD1 9SY, UK. e-mail: email@example.com
What's known on the subject? and What does the study add?
Cancer tissue is stiffer than normal tissue, a fact known for many years. This has been measured using ultrasound (US) technology and is termed as elastosonography (ES). There have been reports of this technique being used in the detection of prostate cancer; however, no definite guidelines for its clinical application exist.
The present review, for the first time synthesises published data of transrectal ES (TRES) using diagnostic review methodology. TRES increases prostate cancer detection as compared with grey-scale US. Also, the study highlights limitations and strengths of data in this area and includes recommendations for future research.
To assess the diagnostic performance of transrectal elastosonography (TRES) for the detection of prostate cancer. Two reviewers independently extracted the data from each study. Quality was assessed with a validated quality assessment tool for diagnostic accuracy studies. Diagnostic accuracy of TRES in relation to current standard references (transrectal ultrasonography [TRUS] biopsies and histopathology of radical prostatectomy [RP] specimens) was estimated. A bivariate random effects model was used to obtain sensitivity and specificity values. Hierarchical summary receiver operating characteristic (HSROC) were calculated. In all, 16 studies (2278 patients) were included in the review. Using histopathology of the RP specimen as reference standard, the pooled data of four studies showed that the sensitivity of TRES ranged between 0.71 to 0.82 and the specificity ranged between 0.60 to 0.95 (pooled diagnostic odds ratio [DOR] 19.6; 95% confidence interval [CI] 7.7–50.03). The sensitivity varied from 0.26 to 0.87 and specificity varied from 0.17 to 0.76 (pooled DOR 2.141; 95% CI 0.525 to –8.737) using TRUS biopsies (minimum of 10) as a reference standard. The quality of most studies was modest. SROC estimated 0.8653 area under the curve predicting high chances of detecting prostate cancer. There were no health economics or health-related quality of life of the participants reported in the studies and all the studies used compressional technique with no reported standardisation. The TRES technique appears to improve the detection of prostate cancer compared with systematic biopsy and shows a good accuracy in comparison with histopathology of the RP specimen. However, studies lacked standardisation of the technique, had poor quality of reporting and a large variation in the outcomes based on the reference standards and techniques used.
Prostate cancer is the most common cancer in men, its incidence has increased over the past decades mainly due to improved awareness, increased life expectancy and seemingly increased uptake of opportunistic screening in countries where screening is not a part of health policy [12,28]. The diagnosis of prostate cancer is suspected in men with raised PSA levels and/or abnormal DRE. Histopathology remains the mainstay of confirming the diagnosis in these men and is obtained by examining biopsies of the prostate gland guided by ultrasonography (US). TRUS with grey-scale imaging is widely practised technique for guiding prostate biopsies . However, US imaging as it exists cannot reliably differentiate between normal and cancerous tissue in up to 50% of cases and its use is limited to visualising the prostate, measuring its volume and placing biopsy needles into the desired region as per the biopsy protocol [10,28]. This drawback of current US prostate imaging forms the basis of continuing technological research in the US area; designing of multiple protocols to improve sampling of gland and use of pre-biopsy multiparameteric MRI in the diagnostic evaluation of prostate cancer. Recently, there has been an interest in using US-contrast agents to increase diagnostic accuracy . Multiple other advances including use of Doppler to identify prostate cancer foci within the prostate gland have been reported with some promising results [23,30].
Cancer tissue is known to have higher tissue density than normal and therefore it projects differently when compressed and decompressed allowing differentiation between it and normal healthy tissue on elastosonography (ES). This basic principle gave rise to the development of various techniques, e.g. compression-strain imaging, vibration ES, acoustic radiation force generated by an US pulse and real-time shear wave velocity . Transrectal ES (TRES) is a relatively new investigative tool that maps out the relative tissue stiffness of the prostate gland and was first described in 1987 . The tissue elastography, also called strain imaging was popularised by Ophir et al.  in 1991.
In elastography, images are obtained from the target tissues by stressing, vibrating or compressing them. It is based on assumptions that if force is applied to the tissue (stress), relative displacement of points (strain) will be proportional to the force and is represented by well-described Young's modulus . ES assesses the strain in the tissues under examination by applying controlled stress with the hypothesis that soft tissues deform more than the stiff tissue, and these differences can be quantified by imaging. The technological developments in ES can be classified into two main areas depending upon the means of applying force (manual compression or non-operator dependant shear wave) and detection of strain (low-frequency amplitude or high-frequency amplitude).
Many foreseeable advantages exist in imaging of prostate cancer using TRES, as the technique seems to be an extension of well-established DRE. In fact, recent studies have shown that TRES can detect up to 90% of prostate cancers with a specificity of 80% . Whilst the available data suggests better performance of TRES as compared with TRUS for the detection of prostate cancer, no conclusive evidence exists as to whether TRES should be part of routine clinical practice. Possible barriers to the dissemination of ES for clinical practice have been the lack of reproducibility and quantification of imaging data. However, with the introduction of shear-wave elastography; there are reports from breast cancer imaging of improved reproducibility and better quantification of data . The ongoing pilot study at authors' institute has shown some promising results (Fig. 1). However, it remains unclear whether improved imaging of the prostate using TRES could selectively target cancer foci within the prostate gland and thereby reduce the need of multiple biopsies. This question has great clinical implications as use of TRES could reduce morbidity and the cost of TRUS biopsies.
With these uncertainties, we undertook a systematic review to compare the diagnostic performance of TRES in the detection of prostate cancer in comparison to two reference standards, i.e. routine TRUS-guided prostate biopsy or histopathology of radical prostatectomy (RP) specimen.
SEARCH STRATEGY AND STUDY SELECTION
The systematic review and meta-analysis was performed according to the Cochrane diagnostic accuracy reviews guidelines. The search strategy was conducted to find relevant studies from the following databases:
1MEDLINE (1966 to March 2011)
2EMBASE (1980 to March 2011)
3Cochrane Central Register of Controlled Trials – CENTRAL (in The Cochrane Library– Issue 1, 2011)
4CINAHL (1872 to March 2011)
7Individual urological journals
Terms and Medical Subject Headings (MeSH) phrases used included: ‘prostate sonoelastography’, ‘prostate elastography’, ‘prostate cancer elastography’, ‘Prostate [MeSH] AND Elasticity Imaging Techniques [MeSH]’, and ‘Elasticity Imaging Techniques [MeSH]) AND Biopsy [MeSH]’.
Reference lists of previous reviews and previous trials were included, papers in languages other than English were included, references of searched papers were evaluated for potential inclusion, and recently published versions were included if the publication was duplicated. Authors of the included studies were contacted whenever the data was not available or not clear; however, only published data was included in the meta-analysis.
Studies that appeared to fit the inclusion criteria were identified for full review by two reviewers (O.A. and G.N.). The inclusion criteria were:
1Any study which reported TRES in men suspected of prostate cancer.
2Studies comparing diagnostic accuracy of TRES with one of the reference standards.
Studies reporting experimental laboratory research and animal studies were excluded.
Each reviewer independently selected studies for inclusion in the review. Disagreement between the two extracting authors was resolved by consensus involving a third party (one of the other members of the research team).
DATA EXTRACTION AND STATISTICAL ANALYSIS
Studies reporting on evaluation of TRES compared with conventional TRUS biopsies or histopathology of RP specimens were included. The following variables were extracted from each study: publication year; country of origin of the study; study design; study population demographics; the TRUS-guided biopsies protocol (number of biopsies); TRUS machine (manufacturer); type of TRES technology and the type of reference standard (systematic biopsies or histopathology of RP specimens). The data on true positive, false positive, true negative and false negative was calculated for each study. The pooled diagnostic odds ratio (DOR) was estimated. DOR is the ratio of the odds of positivity of TRES in the prostate cancer relative to the odds of positivity in normal tissue. Alternatively, DOR is a ratio of the odds of cancer in TRES-positive imaging to the odds of cancer in TRES-negative imaging. Summary receiver operating characteristic curves (SROC) were constructed using the random effects DerSimonian–Laird model. We used Review Manager (RevMan 5.0.23) and the MetaDiSc softwares (ftp://ftp.hrc.es/pub/programas/metadisc/Metadisc_update.htm) for all statistical analyses .
The methodological quality of each included study was assessed by two reviewers (O.A., G.N.) independently using the STARD checklist . Each study was assessed for recruitment methodology, test assessment and statistical methods. Also assessed were the various biases in the introduction and discussion sections.
The literature search yielded 471 studies, of which 419 were excluded for various reasons based on titles and abstracts (Fig. 2). In all, 52 studies were then retrieved for further assessment, of which 16 were included in the review [2,4,8,13,14,16,17,19,24–27,29,31–33]. All the included studies were published between 2002 and 2010, reflecting recent introduction of the prostate TRES technology. In all, 10 of the included studies assessed diagnostic accuracy of TRES compared with routine TRUS biopsies as reference standard [1,4,8,13,14,16,17,19,21,22] and six studies compared ES findings with the histopathology of the RP specimens of patients undergoing RP [26,27,29,31–33]. Two studies also compared TRES with Doppler US [15,23] and one study compared TRES with TRUS, Doppler US, and MRI . Whenever data was not available in the reports or there was not enough clarification, lead authors were contacted to get the raw data or further information. The response to such individual requests was three of 10. Many corresponding authors had left the institutions where the published study came from. Sumura et al.  provided unpublished data of TRUS vs TRES biopsies, which could not be used for meta-analysis due to schematic biopsy protocol (six or eight biopsies) used in the study.
Several studies [4,13,14,16,17,19,29] could not be included in the meta-analysis based on the variation in the reference standard to the current clinical practice. The reference standard has changed over the past few years and the current practice is to offer at least 10 TRUS biopsies for the detection of prostate cancer .
ES with TRUS biopsy allows a promising ability to identify prostate cancer index lesion.
Most of the studies (50%; eight of 16) were performed in the Europe; whereas a few were performed in Japan (37.5%; six of 16) or in the USA (12.5%; two of 16). In all, 2278 patients were included in the present review. The patient population in the included studies was very heterogeneous. Men suspected of prostate cancer had varying PSA concentrations ranging from 0.2 to 200 ng/dL with or without abnormal feeling prostates on DRE (Table 1). The age ranged between 35 and 88 years. Most (eight of 16) of the studies used EUB 8500 Hitachi US machine with 7.5 MHz probe , whereas others (three of 16) used EUB 6500, one of the 16 used EUB 7500, one used a Toshiba Vision 6000, two used Voluson 730 , and only one study did not mention which machine was used (Table 2). The time scale of the studies ranged between 3 and 23 months with an average of 13 months (Table 2)
METHODOLOGICAL QUALITY OF THE INCLUDED STUDIES
Overall, the quality of the reported studies was extremely variable (Fig. 3). All the included studies may be subject to spectrum bias as their method of recruitment of patients was limited to target group (patients suspected of prostate cancer) rather than applying the index and reference test to an unselected patient population (patients with and without high PSA concentrations or abnormal DRE). However, this may be contested by many as index test might not be applicable to the unselected population in real-life clinical practice. Most studies may be subject to review bias as it was unclear in all 16 studies whether the investigators who used the reference test (TRUS-imaging findings) were the same who performed TRES. It was also unclear whether the investigators performing the index test (TRES) were ‘blinded’ to the results of the TRUS findings and vice versa.
DEFINITION OF THE REFERENCE STANDARD TEST
Six studies used sextant TRUS biopsies as the reference standard, which is not the current standard of care and hence the data from these studies was not included into the meta-analysis [4,13,14,16,17,19,29]. Data sufficient to calculate sensitivity and specificity were available for eight studies (Table 1) [2,8,21–23,29,31,32]. The remaining two studies did not have sufficient data for calculations of parameters for diagnostic accuracy and attempts in contacting the corresponding authors remained unsuccessful [27,33]. All the included studies did not report whether biopsies were reported by uropathologists or general pathologists. Also, none of the included studies explored correlation between Gleason grade and ES images.
DIAGNOSTIC VALUE OF TRES
The results were grouped into three categories based on the reference standard used in the included studies:
Studies using histopathology of RP specimen as reference standard
Four studies [26,29,31,32] used histopathology of RP specimen as reference standard. The sensitivity of studies ranged from 0.71 to 0.82 and the specificity ranged from 0.60 to 0.95 with a pooled DOR of 19.643 (95% CI 7.712–50.034; Fig. 4A). The DOR implies that patients with positive TRES on imaging have 20-times more odds of having prostate cancer than those with negative findings on TRES. The corresponding ROC curves are shown in Fig. 4B. The ROC curve assesses the accuracy of the TRES to discriminate between diseased and non-diseased tissue . The area under curve (0.865) clearly estimated a higher discriminatory power of TRES for the detection of prostate cancer.
Studies using TRUS biopsies as a reference standard (patient level analysis)
Six studies [2,8,24,25,31,32] used TRUS biopsies as the reference standard at the patient level. The sensitivity ranged from 0.26 to 0.87 and specificity ranged from 0.17 to 0.76 with a pooled DOR of 2.1 (95% CI 0.53–8.7; Fig. 5).
Studies using TRUS biopsies as reference standard (core biopsy level analysis)
Two studies [8,24] used analysed at the core biopsy level in comparison to TRUS biopsies (≥10) as the reference standard. The sensitivity varied from 0.36 to 0.69 and specificity ranged from 0.89 to 0.93 with a pooled DOR of 12.1 (95% CI 5.08–29.02; Fig. 6).
The results of the analyses based on reference standard varied in the studies. Whilst the histopathological analysis of RP specimens is considered as the best available standard, several studies have compared ES with the TRUS biopsies, which has its own limitations.
Several studies using <10 biopsies reported a better prostate cancer detection rate using TRES; however, they were excluded from analysis. The location of tumour also influenced the TRES imaging. Four studies [2,21,22,27] reported that TRES detects apical tumours better than basal; one study  reported a better detection for mid-glandular tumours and two studies [29,31] reported better performance of TRES in anteriorly located cancers.
TRES appears to perform better than random systematic biopsies using conventional B mode US for the detection of prostate cancer. Data using the RP specimen as reference standard showed a higher diagnostic accuracy of TRES. TRES has a better sensitivity and specificity; however, there are large variations between the studies such as: patients' population (PSA concentration and DRE findings), technique of TRES image acquisition, study protocols (number of biopsies); use of reference standards; methodology and design of studies, reporting and use of different machines with different settings. The standardisation of the TRES technique in the included studies remains a challenge. Most of the reported studies used the compression technique, which is prone to inter-operator variations. No data were reported on the learning curve and reproducibility of the technique. This certainly has implications for the external validity of the research findings. Furthermore, there was no long-term reporting on the effect of technique on the number of negative biopsies with persistently high PSA concentration.
Based on conclusions reported by the authors of the primary studies, seven of the included studies reported that TRES had a higher prostate cancer detection rate than B-mode US [2,14,16,21,22,27,31]. Three studies reported TRES has a higher detection rate than DRE [16,17,29], and one study reported TRES has a higher detection rate than both colour Doppler US and MRI . However, five studies reported that combined TRES with TRUS had a higher detection rate than TRES alone [13,16,19,31–33]. TRES remains a safe procedure; three studies mentioned that no complications occurred due to addition of TRES to conventional TRUS examination [8,19,22]. In one study, TRES was reported to reduce the number of core biopsies needed for the detection of cancer .
FINDINGS OF THE SYSTEMATIC REVIEW IN CONTEXT TO PREVIOUS STUDIES
In the past, several studies have reported modifications of TRUS biopsy protocols to improve its diagnostic accuracy characteristics and increase cancer detection rates, centres have increased their biopsy sample numbers; however, this led to increased complication rates as well as poor patient tolerance, with no significant increase in prostate cancer detection [2,4,19]. Other studies have attempted to use colour Doppler US to assess increased vascularity in the prostate area, which might represent cancerous tissue; however, these studies remain inconclusive [4,19]. MRI using an endorectal coil has also been used to detect prostate cancer with significantly higher accuracy than standard methods, but it has not been recommended for routine evaluation of patients suspected of prostate cancer before histological diagnosis [8,32]. Moreover, this approach seems to be less cost-effective.
With the technological development of TRES and its use in other areas, e.g. detection of breast cancer, numerous studies have shown that ES could increase the detection rate of prostate cancer and reduce the number of core biopsies needed [2,21,22,32]. This has apparent benefits for the patients and healthcare organisations.
LIMITATIONS OF THE CURRENT TRES TECHNIQUE FOR THE LOCATION OF PROSTATE CANCER
Location of tumour foci within the prostate gland can influence detection rate using TRES. Four studies reported TRES-detected tumours more commonly in the apex followed by mid-gland and least at the base [21,22,26,27]. One study reported TRES-detected prostate cancers at the mid-gland level more frequently than the apex and least at the base . Two other studies found that TRES detects anterior tumours more frequently than posterior tumours [29,32]; however, Tsutsumi et al.  reported that posterior tumours were better detected on DRE and B-mode US and therefore combining with TRES could potentially enhance the cancer detection rate. Solomon et al.  also reported that TRES detects tumours on the right side more frequently than the left; however, this could be due to the fact that the patients were lying on the left lateral position and therefore this might alter the compressibility of the prostate compared with the opposite side. This highlights the operator dependency of the technique, a significant challenge in the TRES technique using the compression method. For elastosonographic detection of clinically significant and high-grade prostate cancers, 10 studies [2,13,14,19,21,22,27,29,31] reported that the TRES detection rate was higher for prostate cancer with a higher Gleason score, whereas Tsutsumi et al.  found that TRES detected low-staged tumours more frequently than high-grade tumours. Two studies found that the TRES detection rate increased the higher the PSA concentration [16,17].
Major limitations of TRES are that it is operator-dependant, as the person doing the procedure is manually compressing the prostate therefore repetition and training is needed to get reproducible results [16,29]. Pallwein et al.  reported that examiners need a minimum of 3–6 months training before achieving reproducible results. Manual compression can be a cause of artefact images . Different US machines (Table 2) were used in the included studies, which could cause variations in the outcomes. None of the studies included in the review provided or reported any information on the sponsorship by the manufacturers of the US machines.
STRENGTHS OF THE PRESENT STUDY
The present study is based on a comprehensive review of the literature, based on an extensive search strategy and significant multidisciplinary input. A considerable effort, including contacting authors for missing data, was made to synthesise the data and bring together all available primary studies that were relevant for our research question. A robust quality assessment instrument recommended for diagnostic accuracy studies was used. TRES, a new technique, has already been reported in >2000 patients. The summary of evidence in the present study highlights the need for urgent attention to improve methodological quality and reporting of literature in this field.
LIMITATIONS OF THE PRESENT STUDY
The validity of the results of systematic review depends on the quality of included cohort studies, including selection of participants and inclusion criteria. The quality of reporting in the primary studies was often poor with a lack of important information. The impact of variability of patients' population (size of prostate gland, presenting symptoms, PSA concentration) on test performance parameters could not be assessed. This limits the external validity of the present study, as we cannot adjust results to clinical variations of patients' population. The baseline age, PSA concentration and DRE findings varied in the review and this may introduce spectrum bias. The other limitation of the present review is related to application of various different reference standards. In the detection of prostate cancer, TRUS biopsies (≥10) are considered as reference standard. However, standard schematic TRUS biopsies perform poorly in the location of prostate cancer when compared with histopathology of the RP specimen . Moreover, issues could be related to choosing ‘target condition’ and in prostate cancer arguments can be made for testing new detection techniques to differentiate clinically significant from clinically insignificant disease. Hence, true diagnostic accuracy of the index test (TRES in the present study) in these situations remains unknown. Review bias is another limitation as none of the studies reported that the pathologists or persons performing the reference tests were ‘blinded’ to the TRES findings. This could potentially over or under estimate the test performance. The methods of histological preparation and core lengths for reference and index test were not reported in any of the included studies.
IMPLICATIONS FOR PRACTICE
TRES-guided biopsy (compared with the current standard of care) increases the detection of prostate cancer and may reduce the number of core biopsies required. However, as this is an emerging technique, practitioners should be trained in its application, technique should be standardised and references standards should be agreed upon. This may be a step towards localisation of prostate cancer foci within the gland and aid in directing focal therapy.
IMPLICATIONS FOR RESEARCH
Future research efforts should be concentrated on higher quality, more rigorous evaluation of ES. As a minimum, these should use predefined ideally standardised US parameters, agreed reference standards and studies should be sufficiently powered to estimate the relevant outcomes. Studies should be protocol driven, preferably peer reviewed before the start. Outcome measures must include outcomes assessed by patients and ideally health economic outcome measures. The impact of TRES detection on the Gleason pattern of prostate cancer, location of disease, size of lesions, need for repeat biopsies, correlation with multiparametric MRI findings need to be examined.
The technology of ES has improved and recent introduction of shear-wave elastography needs to be evaluated, especially considering the emerging evidence in other conditions, e.g. breast and liver cancer. The claimed advantages of operator independence (radiation force is produced by probe) and quantitative measurement of stiffness needs to be seen in prostate cancer.
The TRES techniques appears to improve the detection of prostate cancer compared with systematic biopsy and shows a good accuracy compared with histopathology of RP specimen; however, studies lacked standardisation of the technique, had varied reference standards, methodology and quality of reporting. Further studies with better design are needed to assess the role of TRES in the detection of prostate cancer.
CONFLICT OF INTEREST
None declared. Source of funding: Endowment funds were provided by NHS Tayside.