Yes we scan: how MRI was trained on the prostate for detection and diagnosis
Author:
Simon Crompton
Date of publication: 05 October 2022
Last update: 05 October 2022
Introduction
Magnetic resonance imaging (MRI) of the prostate has, in the second decade of the 21st century, become integral to ‘gold standard’ pathways for the early detection of prostate cancer. MRI reduces the need for invasive and potentially damaging prostate biopsy to confirm the presence of cancer, usually after an initial blood test has indicated raised levels of PSA (prostate specific antigen).
Because it can accurately identify clinically significant prostate cancer, MRI also reduces the risk of overdiagnosis and overtreatment of prostate cancers that are unlikely to do harm – that is, cancers that were less likely to be identified and unnecessarily treated, with patients needlessly running the risk of harms such as impotence and incontinence.
MRI as a tool in the early diagnosis of prostate cancer was first explored in European centres in the 1990s, and the combined efforts of urologists and radiologists, many of them in centres in Europe, resulted in accurate and standardised scanning techniques and trials demonstrating superiority to ultrasound-guided biopsy after PSA testing.
Recently, in 2020, MRI was recommended in the prostate cancer diagnostic guidelines by both the European Association of Urology and American Urological Association. The change in their guidance came after two important European trials: PROMIS (Prostate MR imaging study) and PRECISION (Prostate evaluation for clinically important disease: sampling using image guidance or not).
The PROMIS trial found that using MRI to triage men referred on suspicion of prostate cancer might allow more than one in four patients to avoid a primary biopsy and might detect 18% more cases of clinically significant cancer than standard pathways using transrectal ultrasound (TRUS) biopsy. A linked study found that an MRI-first strategy is cost-effective.
PRECISION found that using MRI before biopsy, and then guiding biopsy with MRI if a confirmatory biopsy was required, led to diagnosis of significantly more harmful prostate cancers and significantly fewer harmless cancers than standard pathways using TRUS biopsy.
The impact of sophisticated imaging in the diagnosis of prostate cancer should not be underestimated, said Mark Emberton, consultant urologist at University College London Hospitals, and professor of interventional oncology at University College London (UCL), who led initiatives to investigate the effectiveness of prostate MRI.
“The first prostatectomy was carried out in 1904, so we’ve been treating prostate cancer for more than 100 years – and we’ve been treating it for most of that time without being able to see it,” he said. “When you compare that to other cancers, it is extraordinary because cancers like breast cancer and lung cancer have long been defined by our ability to see them on scans.
“So I think it’s hard to overemphasise the transformative moment of going from an inability to see the disease that we were treating to an era where we can actually see the cancer’s location, volume, morphology and multiplicity.”
Biopsy – a standard with complications
Biopsy of the prostate had been a standard clinical procedure for identifying cancer since the early 20th century. A biopsy was prompted by finding an abnormality during a digital rectal examination (DRE) or raised serum levels of PSA. A transrectal or transperineal needle biopsy was guided by digital palpation through the rectum.
In the 1980s, TRUS biopsy became the standard procedure following a raised PSA reading. It involves inserting an ultrasound probe into the rectum and then, guided by the ultrasound images, firing a fine needle along the probe through the rectum wall and into the prostate, to remove a tissue core. This happens many times – usually 12 – as the doctor takes samples from different prostate areas. It is essentially a ‘blind’ procedure, because prostate cancer cannot be identified on ultrasound images due to poor soft tissue resolution.
TRUS biopsy runs the risk of complications. Pain, urinary retention and blood in the urine and semen occur commonly, and serious infections such as septicaemia occur in 1% to 4% of men. It is sometimes associated with the development of prostatitis, which about 2% of men experience after biopsy. A recent study indicated that 1 in 20 men regret having a biopsy.
Furthermore, there was growing evidence of the limitations of prostate biopsies in identifying the presence of cancer. Another recent paper found that even repeat 12-core biopsies can be negative, despite the patient having cancer.
MRI enters the picture
Against this background, MRI was evolving and producing clearer images of soft tissues. The first study of the prostate using MRI was published in 1982 by physicians at the University of Aberdeen, Scotland. John Steyn, a urologist, and Francis Smith, a radiologist, concluded there was potential for MRI in the management of prostate cancer.
This is also notable because Aberdeen was one of the leading centres in developing MRI for clinical diagnosis, including building the world’s first whole body scanner, and Smith ran the world’s first clinical trial of MRI in 1980 and set up the world’s first diagnostic MRI service at Aberdeen Royal Infirmary in 1981.
However, Steyn and Smith found that the quality of prostate scans was not as good as the computed tomography (CT) technology at that time (and in any case, CT cannot provide clear or any images of cancer in organs such as the prostate). But they had little doubt that MRI would have technical developments that would improve the diagnosis and planning of treatment of cancer, particularly in organs such as the prostate.
That promise was realised with several techniques that have enhanced quality in prostate MRI – first, T1 and T2 weighted imaging (T1 and T2 concern the timing of radiofrequency pulse sequences), and then dynamic contrast-enhanced (DCE) imaging, and diffusion weighted imaging (DWI). Spectroscopic MRI is another technique that showed much promise but DWI is now preferred as it takes less time and is easier to perform.
It was in the 1990s that groups in London and Nijmegen, Netherlands, began to explore the potential of these new techniques in prostate cancer.
In London, Emberton and his colleagues first started to use MRI when investigating a technique called photodynamic therapy to treat recurrent prostate cancer at the UCL National Medical Laser Centre, which was set up in the UCL medical school in 1986.
“Of course we wanted to see whether we were making any difference in the prostate,” Emberton said. “And at the experimental stage we looked around and saw there were MRI scanners that were being used for imaging the spinal column. So we thought why don’t we use them? We obtained two slots a week where we were permitted to look at the prostate of our subjects. I remember one of the first images of a prostate I saw – it was from one of the individuals in the study – and his prostate was peppered with these light cannulas that we were using. It wasn’t a very good image.” (The findings were reported at a meeting of the British Oncological Association, Association of Cancer Physicians and, Royal College of Radiologists in Nottingham.)
For Emberton and his team, what followed was a succession of studies using MRI to look at the prostate – but not at first for diagnosing the disease.
“We came at it initially as a means of measuring treatment response, and used it to better understand what we were doing to the prostate. And then, as you did more and more, you could use it as a staging tool – in other words, to tell whether prostate cancer had broken through the capsule covering the prostate gland. And then later it was used as a detection tool.
“Most discoveries are made in concert when technologies are right. We started to learn from brain cancer, where gadolinium was being used as a contrast agent to see vascularised tumours. And then slowly it grew informally, as things did then, starting to try to sneak in a few prostate MRIs – because MRI was a very valuable resource at the time.
“Of course once we could see the cancer we could start thinking about treating it differently. MRI opened up the era of using information on tumour location to do radiation differently, to operate differently and indeed to treat the tumour plus a margin and try and preserve the prostate.”
By 2006, improvements in MRI technology meant that a potential role in the diagnosis of prostate cancer was now feasible – though limited. As Emberton and imaging experts Alexander Kirkham and Clare Allen reported: “The sensitivity of MRI has improved to the point that it has potential in several new areas: targeting of biopsies, monitoring of disease burden both during active surveillance and after focal therapy, and exclusion of cancer in patients with a raised prostate-specific antigen level.” However: “Specificity is not yet good enough to consider the use of MRI in screening.”
The ‘multiparametric’ MRI equation develops
Meanwhile, in the Netherlands, Jelle Barentsz, professor of radiology at Radboud University Medical Center in Nijmegen, was exploring ways in which MRI imaging might be combined to produce the best possible images of the prostate. Since his PhD thesis on bladder cancer and MRI in 1990, he had been motivated to find ways of using MRI to help patients with urological problems.
“From the year 2000, my group were convinced we could make a difference in prostate cancer,” he said. “MRI was being used to detect how far the tumour extends out of the capsule to improve the surgeon’s resection margins. You’re talking about one cell layer, a microscopic level, so the results were disappointing. We tried to solve that with spectroscopy and using an endorectal coil, but it was all very uncomfortable and it could take hours. So I said we have to look into what is really needed. We don’t actually need a very good tool to predict endocapsular disease but we do need a tool to detect a significant cancer.
“Around 2008, the first results of the ERSPC (European Randomised Study of Screening for Prostate Cancer) were coming out and showed that PSA testing for prostate cancer is good but – and it was a big ‘but’ – it led to too many biopsies and there was overdiagnosis. We needed to test too many men with too many needles. And I thought we could address that with MRI. OK, so we have an idea. Now what should we do?”
At the same time, members of his group and others in Europe were demonstrating the value of new types of MRI such as DCE imaging and DWI. One by one, the constituent pieces for what was to become ‘multiparametric’ MRI were emerging. The new imaging techniques could be added together and combined into a high definition view of the prostate.
Multiparametric MRI (mpMRI) creates a more detailed picture of the prostate than a standard MRI scan by combining four techniques noted earlier: T1- and T2-weighted images, DCE imaging, and DWI.
In 2009, Barentsz and his Nijmegen colleagues proposed a prostate cancer screening test using mpMRI) that reduces the number of unnecessary prostate biopsies and which discriminates between more and less aggressive forms of prostate cancer.
“Multiparametric MRI has a high specificity for prostate cancer detection and provides information about prostate cancer aggressiveness,” they wrote. “PSA in combination with multiparametric MRI performed at 1.5 Tesla appears to be a fairly accurate screening test. Due to its high costs and limited availability, multiparametric MRI is not suitable as a primary screening test. However, it could serve as a subsequent screening test if the PSA has increased above threshold values.”
The London group was thinking along similar lines. In 2009, Emberton and his team published an article, Is it time to consider a role for MRI before prostate biopsy?. It set a new agenda given the technical advances in MRI and its increasing accuracy at detecting and staging prostate cancer.
“We propose an increased use of MRI, not only in those with a diagnosis of prostate cancer but also for men before a prostate biopsy,” wrote the team. “The use of MRI before a biopsy can serve as a triage test in men with raised serum prostate-specific antigen levels, in order to select those for biopsy with significant cancer that requires treatment. This strategy could avoid biopsy, and hence unnecessary treatment, in those with no disease or insignificant cancer.”
It was, said Emberton, “one of our first flags in the sand. We were pretty sure that there was a role.”
A need to standardise: enter PI-RADS
Late in 2009, the potential of MRI as a tool for prostate cancer diagnosis was discussed at a European consensus conference among radiologists and urologists. Attending were the teams from UCL, Radboud and also Lille University Hospital, France, where Arnauld Villers had also been researching the value of MRI.
A paper from Villers and his team, MRI and prostate cancer: a paradigm shift, had just been published, recommending that “to promote acceptance by clinicians and increased access to patients, the indications for prostate MR must be better defined (and provide useful data to urologists), the cost must be reduced, and results must be more reproducible and standardized”.
These were key themes taken up at the consensus conference, held at the Royal College of Surgeons, London. It was all very well that evolving technology had opened up new potential for accurate diagnosis, but if wasn’t used in a rigorous, standardised way, it was of little use to patients.
“The meeting was really at the instigation of Louise Dickinson, a urologist at UCL, who wanted urologists and radiologists to get together,” said Barentsz. “We were saying, ‘OK, what is wrong with prostate MRI at the moment,’ and we wanted to make recommendations on a standardised method for the conduct, interpretation and reporting of prostate mpMRI so that we could detect significant cancer and localise it.
“We needed urologists to point us in the right direction for what was needed. But it was specialist radiologists who could identify what the problem was and how it could be rectified.”
The problem, as Villers had pointed out, revolved around standardisation. “What we concluded after that meeting is that there is no consistency. Everybody looked at the images in a different way. Everybody had a different quality.”
The solution was articulated in the meeting’s conclusions: “Consensus was reached on a number of areas related to the conduct, interpretation and reporting of mpMRI for the detection, localisation, and characterisation of prostate cancer. Before optimal dissemination of this technology, these outcomes will require formal validation in prospective trials.”
The Nijmegen group took the lead on finalising the new standards. Their work resulted in new European Society of Urogenital Radiology guidelines on prostate MRI, published in 2012. More significantly, it yielded a standardised reporting structure which made the widespread use of mpMRI practicable and cost-effective. It is called PI-RADS (Prostate Image Reporting and Data System) and has been the global standard ever since.
“Each lesions is assigned a PI-RADS assessment category of 1–5, based on the likelihood that findings on T2W, DWI and DCE imaging correlate with the presence of a clinically significant cancer at a particular location,” said Barentsz.
“The moment p PI-RADS came out in 2012 many institutions started to work with it. Then to my surprise, within 10 years, it was in the international standards for diagnosis of prostate cancer. Urologists are now using it everywhere. They don’t do a diagnosis without MRI any more, and that was unthinkable in 2006 or 2007.”
In 2014, PI-RADS version 2 was released, and then in 2019 an updated version, PI-RADS v2.1, was published, in collaboration with the American College of Radiology and the European Society of Urogenital Radiology. “I went to the Americans and said, please let’s make it better,” said Barentsz. “My aim was to make it global.”
Barentsz said that it wasn’t so much technical developments in MRI that enabled it to become so widely used as a diagnostic tool. “It was standardisation. You don’t need a very sophisticated MRI machine to do a good prostate MRI, you just need good gradients and a good sequence. The mpMRI sequence used is a regular sequence; not that difficult.”
Evidence to overcome resistance
The decade-long journey from the consensus conference to widespread acceptance of mpMRI as an essential part of the diagnostic process was not, however, an easy one.
As Emberton said: “We were using MRI, and we were sure it worked but it needed nine randomised controlled trials and then our two multicentre studies, PROMIS and PRECISION, to change practice. I think Europe drove that change – with the pivotal multicentre trials and many of us lobbying, lecturing, debating and writing a substantial number of papers. We worked quite hard over these 10 years to persuade people.”
The PROMIS study, evaluating mpMRI in the diagnosis and characterisation of prostate cancer, was conceived in a coffee break during an international prostate cancer meeting in 2010.
“At the time I was having lots of conversations about imaging with other influential urologists and many of us were starting to think that screening with PSA was just not going to work,” said Emberton. “At the conference, Fritz Schroeder (the influential urologist in the Netherlands, who was leading efforts to evaluate the value of PSA testing for prostate cancer screening) was reporting on the latest findings of ERSPC, and we were hearing the data and thinking ‘this isn’t right’ because we knew there was a much better test around the corner. Fritz, to his credit, saw the limitations of PSA testing too. So over coffee, Chris Parker, clinical oncologist at the [Royal Marsden](@ royal-marsden), and I designed PROMIS because there wasn’t enough data in the public domain showing that MRI was particularly good at that point.”
PROMIS tested the effectiveness of mpMRI in diagnosing prostate cancer by asking patients to undergo mpMRI followed by template prostate mapping and transrectal ultrasound biopsy. It started enrolling through UK centres in 2012, recruiting 740 men by 2015.
“It was the first formal attempt to evaluate the performance characteristics of MRI in a very controlled setting, so all the tests were independent of each other, men had 5 mm template sampling, and you kind of played battleships with their prostates – and then saw whether the MRI predicted positivity or negativity against TRUS biopsy.”
The growing interest in MRI in prostate cancer diagnosis was met with considerable resistance from other urologists, particularly in the US. “In a fee-for-service environment, using mpMRI meant the start of the diagnostic pathway going from urologists to radiologists, so a loss of income to urologists,” said Emberton. “So there was a lot of resistance and it held up MRI in the US for a long time. And unfortunately if the Americans hold things up then everybody else can hide behind that.
“The insurers didn't want to pay for mpMRI either, of course. Even though it was cost-effective, and they would spend less overall, it was a new cost and expensive.”
Barentsz recalls the response to a TEDx talk – ‘Yes we scan’ – he gave in 2012. “Urologists were metaphorically throwing fruit at me after that talk – they didn’t like my statements about them doing biopsies blindfold if they weren’t guided by MRI. They thought I’d said a terrible thing.”
But the evidence from European trials began to exert an irresistible force. Reporting in 2017, the PROMIS team concluded: “TRUS-biopsy performs poorly as a diagnostic test for clinically significant prostate cancer. MP-MRI, used as a triage test before first prostate biopsy, could identify a quarter of men who might safely avoid an unnecessary biopsy and might improve the detection of clinically significant cancer.”
Even Emberton was amazed by the results. “The sensitivity of MRI was 93% compared with a sensitivity for existing standard biopsy of 49%. So there was a 100% difference in performance. It indicated that for 50 years, we had been using a test that was missing over half of all the clinically significant cancers.”
The trial proved influential in gathering support for the use of mpMRI in prostate cancer diagnosis. It also provided impetus for PRECISION, another major multicentre study – Europe-wide this time, but again overseen by UCL – comparing mpMRI as an alternative to traditional transrectal ultrasound-guided biopsy for prostate cancer detection.
“PRECISION was supported by an EAU [European Association of Urology] grant and it had its badging which was quite important,” said Emberton. “The study was done on a wing and a prayer actually, accomplished largely through goodwill by the European participants. People just wanted to do it, to transition from TRUS biopsy to MRI. They wanted to accelerate that process, and by landing a trial in which half the participants would be randomised to MRI, they forced their institutions to provide resources to do it. It was the right time, people wanted to change, and that allowed the study to recruit really quickly.”
Five hundred men with clinical suspicion of prostate cancer but without prior biopsy were recruited to the randomised controlled trial. The conclusions, published in the New England Journal of Medicine in 2018, were pivotal: “In men with a clinical suspicion of prostate cancer, we found that a diagnostic pathway including risk assessment with MRI before biopsy and MRI-targeted biopsy in the presence of a lesion suggestive of cancer was superior to the diagnostic pathway of standard transrectal ultrasonography guided biopsy.”
The resistance breaks
By 2019, the substantial body of evidence could no longer be ignored. A Cochrane review concluded that “the MRI pathway is better than systematic biopsies in making a correct diagnosis of clinically significant prostate cancer” – though further research was required.
The evidence was sufficient for the National Institute for Health and Care Excellence (NICE), the health technology assessment body, to recommend that mpMRI should be offered as the first-line investigation for people with suspected clinically localised prostate cancer in England and Wales.
Almost simultaneously, the European Association of Urology changed its guidelines for the diagnosis of prostate cancer to recommend MRI before biopsy. The American Urological Association followed suit a year later.
Emberton said the story exemplifies how change happens in health services. “The world changes when the world is ready, or when you see it is impossible to hold the opposite view any longer. Everybody says, ‘We can’t change because there’s not enough evidence.’ And then, when the world changes, you look back and see there was plenty of evidence. I’m sure the delay in implementation resulted in many men having diagnoses missed. People were being very cautious. Not all randomised studies pointed in the right direction but the majority did and they were the best quality ones.
“You could argue that the big studies weren’t actually necessary; the data was there. But what these big studies do is they act as persuaders. It’s not the data: it’s the publicity they get. You can’t ignore something that’s in the New England Journal of Medicine.”
Legacy and challenges
Barentsz looks back on what has been achieved as uniquely European: a result of close collaboration in diverse settings that encourage innovation.
“For some reason, if you have multiple countries in Europe, with multiple cultures, together it’s a very creative environment. That’s certainly what happened with PI-RADS – the intellects, the creation, the development and modification – even the cost-effectiveness is now being looked at across Europe.”
This is not to say there are not organisational and financial barriers to the universal introduction of MRI into prostate cancer diagnosis.
“We know that it can save you a lot of money and increase men’s quality of life,” said Barentsz. “But it’s not yet the case that every man who needs an MRI gets one. Now that it’s in the guidelines, we have the biggest hurdle yet to come – to change healthcare systems so that the use of prostate MRI becomes universal.
“MRI is the best biomarker we have for clinically significant prostate cancer. It will improve, and there will be PI-RADS versions 3, 4 and 5. We now need to focus on quality control. Artificial intelligence will help, allowing us to combine information on genomics, molecular pathology and clinical information, and in the end we will have a nice personalised diagnostic tool with MRI as the basis.”
Just how rapidly things have changed is indicated by the fact that mpMRI is now being seriously considered across Europe not just as the basis of individual diagnosis but as a central component of national prostate cancer screening programmes.
At the time of writing in July 2022, the European Commission was expected to announce new cancer screening proposals, as part of its Beating Cancer Plan, recommending that all member states establish organised prostate cancer screening programmes with mpMRI scans (as opposed to biopsy) as a follow-up to PSA tests. This would be part of a risk-stratified approach to reduce the risks of overdiagnosis and overtreatment.
MRI has, in effect, changed the balance of risk and benefit associated with prostate cancer screening. It nullifies the overdiagnosis risk that Schroeder and others were so concerned about with PSA.
“Things are changing in Europe,” said Barentsz. “This will be a continuation of the dream of Fritz Schroeder, who showed that with screening you have an enormous gain – if only you can reduce the number of biopsies and the risk of overdiagnosis. And that’s what mpMRI can achieve.”
For more on the evolution of MRI of the prostate with original images and references to studies in the timeline above see this article by Mark Emberton and colleagues.
Early 20th century
First prostatectomy carried out in 1904 and biopsy became a standard clinical procedure
1963
First transperineal ultrasound examination of the prostate but poor image quality
1971
First clinical application of 3.5 MHz transrectal ultrasound (TRUS), which becomes the standard by the 1980s despite complications and limitations in identifying cancer
1980-1981
World’s first clinical trial of MRI and first MRI diagnostic service at Aberdeen Royal Infirmary/University of Aberdeen
1982
First MRI prostate study by John Steyn and Francis Smith at Aberdeen and potential of MRI noted
1983-1985
Pelvic MRI imaging study with T1 and T2 weighted imaging allowed accurate volumetric assessment of malignancy but unable to distinguish cancer from prostatitis; another study reported appearance of cancer but the equipment was too expensive for screening. A study found T2-weighted images better visualised prostate cancer and invasion of seminal vesicles. Optimal pulse sequencing was investigated and ability of MRI to distinguish cancer from benign prostatic hyperplasia
1986
MRI used by researchers at University College London (UCL) to investigate photodynamic therapy in recurrent prostate cancer
1987
First descriptive MRI study of the appearance of the prostate gland and periprostatic structures
1993
First report on use of dynamic contrast-enhanced (DCE) imaging in prostate cancer staging; since then there has been rapid development in the use of a contrast medium in prostate MRI
2002
Application of diffusion weighted imaging (DWI) first reported, and other studies follow supporting its use in diagnosis
2006
MRI technology improved enough for potential role in diagnosis
2009
Prostate screening test before biopsy proposed using multiparametiric MRI (mpMRI), which combines T1/T2, DCE imaging, DWI and MRI discussed at a European consensus conference
2012
PI-RADS (Prostate Image Reporting and Data System) published that sets a global standard
2017
PROMIS (Prostate MR imaging study), a randomised trial, found a quarter of men could avoid a biopsy and 18% better cancer detection than TRUS
2018
PRECISION trial published, which confirms superiority of MRI
2018-2020
The National Institute for Health and Care Excellence (NICE), European Association of Urology and American Urological Association issue guidance recommending mpMRI for first-line investigation for people with suspected clinically localised prostate cancer.