The PROMIS study is a UK-based study designed to confirm earlier data suggesting that having a multiparametric MRI (mpMRI) scan could really help to decide which men needed a prostate biopsy because they really did or didn’t have prostate cancer.
The primary objectives of the PROMIS trial were to establish
- The proportion of men who could safely avoid biopsy altogether
- The proportion of men correctly identified by mpMRI who actually had clinically significant prostate cancer.
Ahmed et al. have just reported the results of this trial in The Lancet, and we were able to receive and review an advance copy of the paper that has just been published today (as of midnight UK time). The full text of this article is available on line, as is some extremely detailed supplementary data.
Between May 17, 2012, and November 9, 2015, the research team enrolled > 700 patients who were scheduled for a prostate biopsy for risk for prostate cancer (but who had never received a prior biopsy) and who agreed, initially, to participate in this trial.
The trial design was such that the patients first received a 1.5 T mpMRI and then were given both a TRUS-guided biopsy and a transperineal template prostate mapping TPM biopsy in a combined biopsy procedure (CBP). The TPM biopsy was used because it is believed to have the highest capability of any type of biopsy to actually be able to identify the presence of prostate cancer (and it has to be carried out under anesthesia, which is why it is not commonly used as a standard form of biopsy). The TRUS-guided biopsy procedure was used because it is the most common type of biopsy used in the diagnosis of prostate cancer.
It is very important to note that no patient received an MRI/TRUS fusion-guided biopsy in this trial. After their MRI, they all received the standard TRUS-guided biopsy and the TPM biopsy (the combined biopsy procedure) without the biopsying physicians being aware of the results of the MRI scans.
The findings of the study are as follows (and they are complex):
- A total of 740 men were enrolled and initially agreed to participate in this study at 11 different centers in the UK.
- 164 men did not complete the trial for one or more reasons (the most common being a very large prostate)
- 576/740 men actually completed all phases of the study, including an mpMRI and the two types of biopsy.
- Based on the TRUS-guided biopsy alone
- 124/576 men (22 percent) were diagnosed with clinically significant prostate cancer.
- 452/576 men (78 percent) were diagnosed with no prostate cancer at all or no clinically significant prostate cancer
- Based on the TPM biopsy data alone
- 230/576 men (40 percent) were diagnosed with clinically significant prostate cancer
- 346/576 men (60 percent) were diagnosed with no prostate cancer at all or no clinically significant prostate cancer
- 56/576 men (10 percent) had a Gleason score of ≥ 4 + 3 = 7
- 220/576 men (38 percent) had a Gleason score of ≥ 3 + 4 = 7
- 174 men could be classified as having clinically significant disease solely on the basis of core length even though they might have had a Gleason score of 3 + 3 = 6 or 3 + 4 = 7.
- 13/576 patients (2 percent) were found to have clinically significant prostate cancer on their TRUS biopsy but not on their TPM biopsy.
- The sensitivity of mpMRI for clinically significant cancer was 93 percent, with a negative predictive value of 89 percent.
- The specificity of mpMRI for clinically significant cancer was 41 percent with a positive predictive value of 51 percent.
- 158/576 men (27 percent) had a negative mpMRI.
- 17 of these men had clinically significant cancer on TPM biopsy.
- All 17 men had a Gleason score of ≤ 3 + 4 with core lengths from 6 to 12 mm.
- 119 significant cancers were found on TPM biopsy but missed by TRUS biopsy.
- 13/119 were Gleason 4 + 3
- 99/119 were Gleason 3 + 4
- 7/119 were Gleason 3 + 3
- mpMRI was more accurate than TRUS biopsy in terms of sensitivity (93 vs 48 percent) and negative predictive value (89 vs 74 percent).
- TRUS biopsy was more specific that mpMRI in terms of specificity (41 vs 96 percent) and positive predictive value (51 vs 90 percent).
The authors state that, based on these data:
- Using mpMRI to triage men might allow 27 percent of patients to avoid a primary biopsy and diagnosis of 5 percent fewer clinically insignificant cancers.
- If subsequent TRUS biopsies were directed by mpMRI findings, up to 18 percent more cases of clinically significant cancer might be detected compared with the standard pathway of TRUS biopsy for all.
- mpMRI, used as a triage test before first prostate biopsy, could reduce unnecessary biopsies by about 25 percent.
- mpMRI can reduce over-diagnosis of clinically insignificant prostate cancer and improve detection of clinically significant cancer.
So the first thing to note here is that, on average, for the vast majority of men at risk of prostate cancer, having an mpMRI prior to a TRUS-guided prostate biopsy would make a lot of sense (if that mpMRI is being evaluated by a urologic radiologist who really knows how to interpret these types of scan and if the MRI data are then used as a component in the conduct of the TRUS-guided biopsy).
But there are also problems, starting with the fact that 2 percent of patients were found to have clinically significant prostate cancer on a standard TRUS biopsy but not a transperineal TPM biopsy. Frankly, we are not quite sure what to make of that finding.
It also appears that having a very large prostate was a problem at several stages in this trial. One therefore has to assume that the results may not be applicable to men with very large prostates.
The other question that a lot of people are going to ask is going to be, “Is a 1.5 T mpMRI good enough? Why not a 3 T MRI?” This is a fair question.
There are some data suggesting that, in the hands of experienced radiologists, it really doesn’t seem to make a lot of difference whether the type of MRI is a 1.5 T or a 3 T mpMRI. But if one is going to implement mpMRI as a triage methodology for all men thought to be at risk for localized prostate cancer then (a) that question needs to be answered and (b) the relevant type of MRI scanning needs to be easily available to all patients.
From a US-based perspective, the two biggest obstacles to implementation of mpMRI scanning of all men at risk for prostate cancer prior to any type of biopsy are:
- Access to a qualified and appropriately equipped mpMRI scanning center
- The high cost of mpMRI scanning in the US compared to other countries
What the PROMIS trialists have done, it seems to us, is that they have justified the concept of using mpMRI triage prior to any type of biopsy for the majority of men who don’t have “obvious” evidence of prostate cancer. It may be possible to start to implement mpMRI triage in some countries with nationalized health care systems and low-cost MRI availability based on these data.
Here in the US, the situation is going to be very different because of the two problems mentioned above. We would encourage patients today to always ask about the possibility of an mpMRI prior to a prostate biopsy so that the results of that mpMRI can be taken into account in giving any subsequent biopsy (if the physician and the patient still feel that a biopsy is appropriate). However, we suspect it will be a while before our health care system can be persuaded that every patient should be encouraged to have an mpMRI prior to a prostate biopsy.
We have little doubt that the principle of having an MRI scan prior to any type of prostate biopsy is now well established and confirmed by the PROMIS data. The practicalities of how patients and the medical profession (and the health care insurance system here in the US) need to act on the basis of these data are likely to take some time to work out.
The investigators are to be congratulated on the completion of a large and complex trial that certainly helps us to have a better understanding of the value of mpMRI scans in the diagnosis of prostate cancer and the necessity for biopsies in certain types of patient.
Editorial comment: We thank Mr. Hashim Ahmed for providing us with an advance copy of the full text of this paper. (Surgeons in England are always called Mr., not Dr.)
Filed under: Diagnosis, Management, Risk | Tagged: biopsy, mpMRI, MRI, multiparametric, PROMIS, trial |
THE "NEW" PROSTATE CANCER INFOLINK 
In a smaller study (191 lesions in 108 patients), mpMRI and TRUS biopsy found 74% of significant lesions: mpMRI biopsy identified 74% of those, TRUS biopsy identified 61%. So, mpMRI-biopsy found a total of 55%. The reason for the misses seemed to be geographical for mpMRI — dorsolateral and apex lesions.
What equipment should a scanning center have in order to be considered appropriately equipped?
Minor note: The word “multiparametric” is an unnecessary adjective to describe prostate MRI. All MRIs are multiparametric whether it be prostate, brain, abdomen, knee, or spine.
Dear Mark:
The question as to what equipment a scanning center should have is going to depend to a very large extent on the answer to the issue of whether 3 T systems do or do not offer meaningfully superior data on which more accurate clinical decisions can be made.
With regard to your “minor note”, I am not a radiologist. The term “multiparametric” has been applied to the types of MRI used to scan the prostate since this use of MRI started to be explored. You may be correct, and it may well be redundant today, but I’d want that confirmed by an expert in uroradiology.
Dear Mark:
If you look at the information headed “Test 1: MP-MRI (index test)” in the section headed “Procedures” in the full text of the article, it will tell you exactly what equipment and MRI capabilities were used in this study, which the authors define as a “standardized” multiparametric MRI that is “compliant with European Society of Uro-Radiology guidelines.” I assume that there is a reason for their use of such specific language.
Mark:
You are mistaken that all MRIs are multiparametric. Most MRIs scan using a single parameter, usually T1 (not to be confused with 1.5 or 3 Tesla (T) which indicates magnetic field strength). If your urologist feels a lump and wants a closer look, or if your radiation oncologist is doing your planning, that is usually the kind of MRI you will get. T1 or T2 MRIs only detect tissue-type differences, but cannot detect areas suspicious for cancer. The multiparametric MRIs used for prostate cancer detection usually include three parameters: T2, DCE, and DWI. Other parameters (e.g., MR spectroscopy or RSI) are less often included.
Gentlemen,
All MRI scans are multiparametric. Ask any radiologist.
T1 is a pulse sequence. T2 is a pulse sequence. Diffusion/ADC map is pulse sequence. Post-intravenous contrasted images are T1s. All MRI scans have T1 and T2 images, often with STIR and gradient echo sequences.
Of course this is a very minor point. MRI of the prostate has found a beneficial place for prostate cancer imaging evaluation, but we should not get enamored by fancy names.
Not sure how the word “multiparametric” got tagged onto MRI of the prostate.
Mark:
It’s silly to argue over facts. I assure you that when I had an MRI last week, it was T1 only, and took about 15 minutes for image acquisition. That is the most common parameter in use. Other parameters are used for specific reasons. A full multiparametric MRI for the prostate takes about an hour because of all the different image acquisitions they must do to collect data from all the parameters collected. If you would like to understand more about this, read p.22-28 of this document on PIRADS v.2
When MRI first came out in the USA in the ealy 1980s, the only sequences avaiable on the scanners (parameters) were T1- and T2-weighted sequences and nearly all scans were of the brain and spine. Shortly thereafter, in the mid and late 1980s, researchers developed new MRI sequences that gave additional information about the diseases that were being imaged. For example, gradient echo sequences are good at looking blood in the brain. STIR images show fluid areas well. Diffusion images show certain disease processes, such as acute stroke, well. Since the late 1980s, all diagnostic MRI scans, whether of the brain, spine, pelvis, abdomen, or extremity, consist of more than just T1- and T2-weighted images sequences (parameters).
Dear Allen and Mark:
Without in any way wishing to be rude, I think this is something of a pedantic discussion. From my purely editorial perspective the term that is used to describe the type of MRI scan currently being used/investigated in the diagnosis and monitoring of prostate cancer is referred to in all of the clinical literature that deals with this issue as a “multiparametric” MRI (unless, as in some cases, and most particularly in the radiology literature, the authors actually describe the precise type of scan in full). Thus, that is what patients who don’t appreciate all the nuances of development of MRI scanning over the past 40 years need to be calling it too, so that they are on the same page as their clinicians (who are usually urologists or radiation oncologists, not radiologists, and who also usually don’t understand all of the nuances either).
Thus, although I appreciate both your perspectives, I intend to simply go on calling them multiparametric MRI scans until the urology community decides to refer to them as something else.
I want to congratulate Sitemaster for producing such a lucid summary of this rather difficult paper. I am still trying to get my head around it. I will just make one observation.
It seems that the results of a study like this depend a lot on (1) how you define “clinically significant”, (2) the method of scoring the MRI results (i.e., what threshold of suspicion do you set), and (3) what you take as your reference (i.e., “gold standard”) for “true” versus “false” findings.
This paper defines “clinically significant” to mean Gleason 4 + 3 or higher, or having any core with cancer core length 6 mm or longer. This seems a bit curious to me. For example, if almost all the cores were positive with Gleason 3 + 4, I would be inclined to call it significant even if no core had 6 mm of cancer.
In this study, the radiologists reading the MRI results at the 10 participating centers all underwent 2 days of centralized training to ensure that they would all be scoring the MRIs in the same way.
An example of a study where some rather different choices seem to have been made is the 2014 paper by Panebiancho et al., which reported that mpMRI for detection of prostate cancer had sensitivity of 86% and specificity of 94%. So, one way or another, the latter study arranged things to have a lot fewer false positives.