Diagnosing Extraprostatic Extension (EPE)

Extraprostatic Extension (EPE) or "Stage T3a" means the cancer has eaten through the edge of the prostate and is penetrating into the tissue outside. It can be difficult to diagnose before a prostatectomy. Sometimes, it can be felt using a digital rectal exam (DRE) as a bulge or irregular texture, but that is an exception rather than the rule. More often, it is seen on an mpMRI or ultrasound image.

It is important because its presence is a strong predictor of recurrence after treatment. It is one of three risk factors used in the NCCN definition of "high-risk" prostate cancer. That definition is based on the AJCC staging criteria (see this link), which means that it is strictly only based on DRE. DREs almost always fail to detect EPE. If the bulge is so big that one can feel it through the rectum, it adds significantly to the risk. But how much does it add to the predicted risk if it can only be seen on a powerful MRI?

Reisaeter et al. evaluated the Mehralivand EPE Grading System and found it was somewhat more sensitive in clinical practice as the commonly used Likert EPE scoring system. (The interested patient may also wish to read this editorial by Peter Choyke)

The Likert score looks at certain imaging abnormalities (tumor contact length with the prostate capsule, irregularity, bulging, gross extension, and loss of rectoprostatic angle) and summarizes them using five categories:

• 1 = criterion not present
• 2 = probably not present
• 3 = uncertain if present
• 4 = probably present
• 5 = definitely present.

The Mehralivand System uses three grades:

• Grade 1 refers to tumors with a contact length of 1.5 cm or greater or contour bulge or irregularity. 
• Grade 2 refers to tumors with a contact length of 1.5 cm or greater and contour bulge or irregularity, 
• Grade 3 refers to gross visible extension beyond the prostate.

Both systems require very well-trained radiologists - interobserver agreement is only fair.

Mehralivand compared the predictions of the EPE detection system to what was actually detected after the same patients had prostatectomies. Even when the Mehralivand System assigned Grade 3 to a suspected EPE, a third of them were false (positive predictive value (PPV) = 66%). False positives may be caused by inflammation, tumor scar tissue, or biopsy scar tissue. Contact with the capsule may be wholly inside, and a bulge may be wholly contained within the capsule. 

What's worse, the Mehralivand System incorrectly predicted there would be no EPE in 18% of cases where EPE was eventually found (negative predictive value (NPV) = 82%). False negatives are caused by tumors below the size where MRI can detect them. 

• The PPV was 41%, 48%, and 66% for Grade 1, 2, and 3, respectively
• The NPV was 90%, 88% and 82% for Grade 1, 2, and 3, respectively

Since DREs are so bad at detecting EPE, and MRIs are little better, what can be done to better predict EPE, and is better prediction necessary?

Is better prediction always necessary?

It has been found that focal EPE (extensions of less than 3 mm) and EPE comprising low Gleason score tumor tissue are not predictive of treatment failure. In this Johns Hopkins study, 10 year biochemical recurrence-free survival was 76% among men with focal EPE (post-prostatectomy) and 59% among those with more extensive EPE. A surgeon discovering a focal EPE may simply cut wider to get it all. GS 6 tumors have low metastatic potential (see this link). However, a patient who learns in advance that the surgeon will "cut wide" thereby increasing his risk of incontinence or impotence may opt instead for radiotherapy.

mpMRI-targeted transprostatic biopsy

It may be possible to detect clinically significant EPE by detecting suspicious sites using an mpMRI and following up with a real-time ultrasound fusion-targeted biopsy. Some pathologists have argued that needle-biopsy cores that show close proximity to the prostate margin, admixture with skeletal muscle at the apex, or admixture with adipose or other peri-prostatic soft tissue predict for EPE. This suggests that clinically significant EPE may be diagnosed with transprostatic needle-biopsy cores. This is an unusual procedure. Of course, as with any needle biopsy, it may miss the site, and several cores from the suspicious site should be taken. A periprostatic nerve block is required (which imho should be required on all needle biopsies) to prevent any additional pain. There is also some risk of extra bleeding if a blood vessel is nicked. It is worth discussing with the biopsy urologist. It is also important that the designated cores be evaluated by an experienced pathologist like Jonathan Epstein at Johns Hopkins.

(update June 2022) Moroianu et al. reported a "deep learning" algorithm that is better at detecting EPE from an mpMRI than a radiologist.

• Model sensitivity was 80% vs. 50% for radiologists (model predicted more true positives)
• Model specificity was 28% vs. 77% for radiologists (model predicted more false positives)

A combined computational/radiologist approach may be best.