Performance of PI-FAB and TARGET scoring systems in post focal therapy mpMRI surveillance

Imagine your body is a vast, bustling city, and sometimes, a few troublemakers decide to set up shop in a very specific neighborhood: the prostate. For a long time, when these troublemakers (prostate cancer cells) showed up, the standard plan was to send in the bulldozers and tear down the entire neighborhood to make sure none were left. But that's a lot of destruction for a small problem. In recent years, doctors have started using "focal therapy," which is like sending in a precision laser team to zap only the bad houses while leaving the rest of the city intact. This saves the city's important functions, like plumbing and power, but it leaves a tricky question: How do we know the bad guys didn't sneak back in?

To answer that, doctors use a special kind of camera called an MRI, which takes super-detailed pictures of the city. But here's the catch: the laser zap leaves behind a messy construction site—swelling, inflammation, and scar tissue—that looks a lot like the troublemakers trying to return. The old rulebook for reading these pictures (called PI-RADS) was designed for a clean city, not a construction zone, so it often gets confused. To fix this, two new rulebooks were recently invented: PI-FAB and TARGET. These are like specialized reference guides designed specifically to help detectives tell the difference between a harmless construction site and a real criminal hiding in the rubble. But nobody knew which reference guide was actually better at solving the case.

This paper is the story of two detectives (radiologists) who decided to test these two new reference guides against each other. They gathered a team of 90 patients who had already undergone the precision laser treatment and then went through the standard check-up process: an MRI scan followed by a biopsy (a tiny sample check) to see if the cancer had truly returned. The detectives looked at the MRI scans using both the PI-FAB and TARGET rulebooks, completely blind to what the biopsy results said, to see if their "guesses" matched the reality found in the tissue samples.

The results were surprisingly even. It turns out that both reference guides are roughly equally good at the job. When the detectives used PI-FAB, they correctly identified about 67% of the patients who had cancer returning (sensitivity) and correctly said "all clear" for about 82% of those who didn't (specificity). When they switched to the TARGET system, the numbers were almost identical: 67% sensitivity and 79% specificity. In fact, when the researchers ran the math to see if one system was statistically superior, the answer was a flat "no difference." Both systems were pretty good at ruling out cancer if the score was low (a high "negative predictive value" of around 88-89%), meaning if the reference guide said "no cancer," it was very likely true. However, neither system was perfect; they both missed some cases and sometimes raised false alarms.

One interesting twist in the story was how well the two detectives agreed with each other. When they both used the TARGET system, they agreed on the diagnosis about 45% of the time (a "moderate" agreement), which was slightly better than when they used PI-FAB, where they only agreed about 35% of the time (a "fair" agreement). This suggests that while the TARGET system might be slightly easier to use consistently, both systems still leave room for human interpretation to vary.

The bottom line of this study is that neither PI-FAB nor TARGET is the clear winner yet. They perform similarly, acting as helpful but imperfect guides in the messy aftermath of focal therapy. The authors suggest that because these scoring systems aren't perfect on their own, doctors might need to combine them with other clues, like how fast a patient's PSA levels are rising, to get the full picture. Until then, these two new rulebooks are both valid tools in the detective's kit, but the case of the perfect post-treatment scan isn't quite closed yet.