Imaging in Prostate Cancer

There are several different imaging modalities used in prostate cancer. These include:


  1. Ultrasound - primarily for biopsy
  2. Bone scan
  3. Computerized Tomography (CT) scan
  4. Magnetic Resonance Imaging (MRI)


Imaging plays different roles depending on what the intended purpose of the scan is and not all modalities are necessary or even helpful in the different 'states' of prostate cancer. The results of any imaging test must be interpreted in the context of the complete clinical picture and should not be used in isolation.


Use of Imaging in Screening for Prostate Cancer


Imaging remains incapable of making a diagnosis of prostate cancer with enough certainty to justify definitive management. MRI, however, has emerged as a powerful tool (in select circumstances) for further risk stratification. At present, MRI (nor any other currently available test) can exclude prostate cancer with absolute certainty. The first step in determining a man's risk of harboring prostate cancer is a integrated risk assessment using age, race, family history, PSA, finger exam findings and prior biopsy results (if available).


Use of Imaging in the Diagnosis of Prostate Cancer

Prostate biopsy remains the only test capable of establishing a diagnosis of prostate cancer. There is no imaging test which can make the diagnosis.


If prostate cancer is suspected based on the screening evaluation, ultrasound guided biopsy is the gold standard for establishing a diagnosis. The purpose of the ultrasound is to ensure that the biopsy samples are representative of the prostate gland, rather than clustered in just one area. Ultrasound is poor for identifying tumors, however.


The use of MRI is establishing itself as a useful tool in select circumstances. There are no current recommendations for MRI as a first-line screening tool. There are 2 features of prostate cancers that can help in identifcation of tumors:

  1. Low signal intensity on T2 weighted images
  2. Restricted diffusion of water

Neither of these features can establish a diagnosis of prostate cancer and biopsy is still required in cases where the risk of harboring a tumor justifies the risks of a biopsy. There is a spectrum of findings on MRI and therefore a uniform approach has been proposed for the interpretation of these scans. It is critical that the 'protocoling' and 'reading' of these scans be performed by physicians who are knowledgeable in this area otherwise the MRI may be useless.


The utility of any test depends on the pre-test probability (focus on the information on likelihood ratios) Without this contextual knowledge, the interpretation of the test result may be highly inaccurate. For this reason, it is important to consider a man's risk of harbouring prostate cancer based on established risk modeling for prostate cancer screening/diagnosis in order to establish a pre-test probability (or some other assessment to see the applicability of a test to specific circumstances). In addition to established risk modelling tools, there are an important number of very specific clincal features that will bear on the pre-test probability of finding cancer which are not taken into account by models of risk including:


  1. Prior biopsy history: number of biopsies, time and time-interval between biopsies, size of gland, location of biopsies
  2. Prior cancer diagnosis: especially location
  3. Specifics of the digital rectal examination


The PROMIS Study: MRI in Prostate Cancer Screening


Clinical studies allow us to interpret a test within specific contexts. The use of MRI in the context of screening was investigated by the PROMIS study (multi-Parametric MRI and TRUS biopsy in Prostate cancer - Lancet January 2017). This study was done in men without prior biopsy history who were felt to be at 'increased' risk for a variety of reasons. Biopsies were done 2 ways - either with a template (usually 20-30 samples under a general anesthetic in the OR taking about one hour) AND with a standard TRUS biopsy (which was done under the same anesthetic for convenience but which is almost alwasy done with local anesthetic).


It is important to understand that every man has some risk of having prostate cancer so it's important to see what the risk was for men in this study to determine the pre-test probability. This helps determine the applicability of the study findings in other circumstances. Men who were enrolled in this study had these characteristics:


  1. PSA 'elevated': up to 15 mcg/L within 3 months; mean 7.1 (SD 2.9); assuming normal distribution, 95% of men had PSA between 1 and 14; 68% between 4 and 10
  2. DRE: any abnormality at all
  3. Family History: any


Based on these inclusion criteria there are a range of pre-test probabilities of having cancer, the ends of which are bracketed by:


  1. On the low end: 55 year old man could ostensibly have had a biopsy with a PSA of 1 mcg/L and a family history of prostate cancer or abnormal DRE. 
  2. on the high end: 65 year old man with a PSA of 16, abnormal prostate exam and a negative family history could also have been advised to have a biopsy. 


These are very, very different patients with a pre-test probability of having high grade cancer of 0.6% and 48%, respectively (using the ERSPC Model 6 calculator). If you understand how the pre-test probabilty affects the utility of a test you'll understand that the exact same MRI result  would have very different implications in these 2 examples.


Note that the extent of the DRE abnormality, family history or specifics of PSA were NOT specified in the PROMIS results - the results of the study pertain to the group AS A WHOLE - i.e. to some fictional 'average' patient. Subgroup analysis would be required to make more specific statements and will hopefully be forthcoming. The consequence of this is that your risk of having cancer if you proceed with a biopsy may be much higher or much lower than reported in the pooled analysis results.


There a few very important caveats in interpreting the results of this study (in addition to the absence of a refined assessment of a man's pre-test probability of cancer). These have the effect of OVERSTATING the performance of MRI as a diagnostic test as well as OVERSTATING the false-negative rate of routine prostate biopsy (i.e. the chances that a cancer was 'missed'). One would need to sift through the methods of the study and supplemtary material to determine the magnitude of these effects.


  1. A more lax definition of 'significant cancer' was applied to MRI-detected cancer than biopsy detected cancer 
    1. MRI detected cancers were significant if Gleason 3+4 or higher (grade group 2/5) compared with biopsy detected cancers which needed to be Gleason 4+3 or higher (grade group 3/5). This is important because some grade group 2/5 cancers can be considered for active surveillance (observation) where as most grade group 3/5 cancers would not (i.e. would be treated with surgery or radiation)
    2. MRI detected cancers could be of lower volume than biopsy cancers - greater than or equal to 4 mm vs. biopsy detected cancers which needed to be over 6 mm in core length
  2. Cancers detected on routine transrectal ultrasound were NOT considered if they were not also detected on a templated biopsy. Stated another way, if the tumour wasn't seen on a templated biopsy it didn't count even if a standard TRUS biopsy detectected the tumour. This assumption is irrational as it should not matter how the cancer is detected - if it was found, it's there no matter how it was found.


Lastly, before looking at the results one needs to determine how and why MRI might be used. Is it being used to avoid the risks of a biopsy or to reinforce the risks of a biopsy? Is it to facilitate a targeted (fusion) biopsy for someone who has already made a decision? There are many different applications but fundamentally all are compromises - with every 'avoided biopsy' there is the risk that a clinically significant tumour will be missed. For every biopsy done for a 'high risk' finding on MRI there is the chance that there is no cancer or that the MRI missed a significant cancer.


The baseline risk (pre-test probability) of patients as a whole entering the PROMIS study cannot be calculated by the provided data since the proportion of men with abnormal prostates was not provided. With the data that is provided (average age of 64 years, 87% caucasian, 22% family history and a mean PSA of 7.1 - Data Appendix):


The pre-test probability of patients enrolled in PROMIS is between 12% and 25% using the ERSPC risk calculator. A fair estimate is that the risk is between 15-20%. This is quite a high pre-test probability of cancer.


The PROMIS study used a slighly different grading system then the PIRADS v1 or PIRADS v2 systems which is the way that all MRI findings are currently being reported. For the sake of simplicity, you can take the MRI level in PROMIS to be similar to that in the PIRADS system. With that aside, here are the results:


MRI Report (highest = 5 = highest risk Chance of cancer on Template Biopsy
  Gleason 3+4 or higher (grade group 3)
'Non-suspicious MRI' 1+2 combined 11%
1 4%
2 11%
'Suspicious MRI' 3+4+5 combined 51% overall
3 21%
4 58%
5 81%


Some other numbers derived from the primary data using a definition of cancer found by BOTH types of biopsy (templated and routine TRUS):


  • If you have clinical suspicion plus suspicious MRI, what are the chances of the MRI-guided biopsy is falsely negative? 10%
  • If you have a clinical suspicion of cancer and TRUS biopsy is negative, what is the chance of having:
    • Gleason ≥ 3+4 cancer? A: 112/556 = 19%
    • Gleason ≥4+3 cancer? A: 13/576 = 2%
  • If you have a clinical suspicion of cancer and MRI is negative, what is the chance of having a Gleason ≥ 3+4 cancer? A: (38+13)/556 = 9%
    • A: How about if we only count templated biopsy cancers as 'real' cancers: 38/556 = 7%


In conclusion, MRI can be very helpful under certain circumstances. If you pre-test probability is lower than the average risk for PROMIS patients (see above) then these results will tend to OVERSTIMATE your risk; if you are at higher risk then these results will tend to UNDERESTIMATE your risk. In applying these results to an individual, one needs to consider much more the the results of the PROMIS study as a whole.



CT and bone scan have no use for the diagnosis of localized prostate cancer (but are useful in staging - see below). If there is a strong suspicion that the cancer has spread then CT and bone scan may be used to make a tentative diagnosis - e.g. in the setting of a very high PSA a bones scan suspicious for spread is usually enough to start treatment.


Use of Imaging for Biopsy Proven Prostate Cancer


If a diagnosis of prostate cancer has been made on biopsy, imaging may be used for further staging. NOT ALL CANCERS REQUIRE ADDITIONAL IMAGING. In many cases the combination of PSA, digital rectal exam findings and the biopsy results will provide all the required information. Imaging is often over-utilized and may provide no additional information or in some cases false information.


MRI may be used for assessment of lymph nodes as well as for operative planning - especially if it is unclear if nerve-sparing surgery to preserve erections may compromise cancer control.


CT scan can be useful for assessing lymph node involvement. Note that not all enlarged lymph nodes are cancer - some may be 'reactive' to the biopsy or infection. The larger and 'rounder' the lymph node, the more likely it is to be cancer.


Bone scan can be useful for assessing spread to bones. Note that prior trauma and osteoarthritis can cause a 'positive' bone scan. Prostate cancer has an affinity for the large and axial skeletal bones - that is, the pelvis, spine and large bones of the legs. The pattern of spread and the likelihood that a patient might have spread to the bones is important in interpretation. For example, a bone scan showing uptake in the leg of a patient with a low risk of spread (e.g. low PSA and grade of tumour) is likely reactive where in a patient with a high chance of spread (e.g. high PSA and grade) is more likely to have actual cancer.


Note that the because of the small amounts of cancer which have spread outside the prostate may not be detectable, the absence of spread on CT or bone scan does not guarantee that there is no spread. However, if the Ct scan or bone scan suggest that the disease has spread it is unlikely that the disease has not spread.


PET Scan


BCCA Vancouver Consent

BCCA Vancouver Information on PET Study