PSA Half Life and Length of Abstenence Period Prior to Test

I was reading a post where a guy was asking the time-honored question of how long one should abstain from sex, DREs, bicycle riding, etc. prior to having blood drawn for a PSA test, and noting that the times he was suggesting seemed longer than necessary, when it occurred to me that simply saying "that seems too long" was insufficiently sciency and I wanted some numbers to dazzle him with.

I found a nice little article in "The Australian Prescriber" (whatever that is) from 2011 that seemed to have the information I wanted.

Here's a link: Prostate specific antigen

At six years old its not new information but the part I was interested in -- the half-life of PSA in the body -- has been pretty well established for a quarter of a century and so a six-year old article that was so simple that even I could read it seemed perfect.

I read through the relevant parts several times, did a bit more Googling to fill in some gaps and have come to the conclusion that the question is more complicated than I thought. For a while it was one of those the-more-I-read-the-less-I-know things but, through dogged perseverance, I have managed to battle my way back to a level of understanding that falls only slightly short of where I started.

Here are a few things that I think I know. (Feel free to point out the places where I am wrong.)

PSA is released into the blood stream by a number of things but prostate cancer is special in that it releases PSA gradually -- more or less constantly -- while everything else (including an enlarged prostate) tends to release PSA in occasional spurts when the gland is jostled or otherwise bothered.

The purpose of PSA in semen is to bind with proteins that make semen (in the seminal vesicles) quite viscous and make it less viscous so the sperm can swim unimpeded. The concentration of PSA in semen is a million times the levels found in the bloodstream.

When it is released into the blood PSA can be divided into two types: PSA that is active and able to combine with suitable proteins, and PSA molecules that are either immature or damaged and are inactive and unable to bind. I am not sure of the proportion of active vs inactive PSA released, I suspect it is somewhere around 50/50.

There are three processes that occur with PSA released into the blood that effect its concentration over time. The first is that "active" PSA will bind with proteins in the blood. I think this happens quickly and by the time you can measure anything it has already happened. Any remaining unbound (free) PSA is of the "inactive" variety that will never bind. The second process is the body metabolizing free PSA. This process has a half life of about two hours -- that is, after two hours half of the initial free PSA will be gone, then after the next two hours half of that free PSA will be gone, and so forth. The body finds and eliminates free PSA at a pretty brisk rate. 48 hours would give a 16 million-fold reduction in free PSA. The third (slowest) process is elimination of PSA that is bound to proteins in the blood. This takes longer. It has a half life of about two days.

Because free PSA doesn't last long in the body the percentages of free vs bound PSA can be used to get some sense of the average age of the PSA one is looking at (the time since release into the blood stream). If you see lots of free PSA (as a percentage) that means that it has been released fairly recently and the man probably does not have prostate cancer since the PSA has been dumped into the blood all at once (which prostate cancer doesn't tend to do).

Most of the data that has been collected on the speed with which PSA is eliminated from the body has been collected from men who have had their prostates removed. You start with the value of total PSA just after surgery and the measure it every few hours for the first couple of days then every day for a few weeks and you can compute a half-life... mostly... Most studies have seen a bit of a dog-leg in their data early in the process where there is a higher clearance rate initially and then it drops off later to that two day half life figure. This is generally attributed to some amount of free PSA being cleared out quickly followed by the slower elimination of the bound PSA. It might be worth speculating that the population of men having their prostates removed might tend to over-represent men with prostate cancer (who have relatively low levels of free-PSA) and the inclusion of more normal men in the sample might give a more pronounced initial drop off.

A study of men engaging in long-distance bicycle riding found (in very round numbers) that a ride of between 35 and 100 miles, measured five minutes after completing the ride, increased a man's total PSA by around 10 percent. (Here is a link). To put this in context, if your "correct" PSA reading is, say, 4.0 ng/dL and you live 100 miles from the lab then if you ride you bike to your appointment and immediately have blood drawn when you get there you could expect something like 4.4. If your lab is backed up two hours (as mine sometimes is) then half of the free part of that extra 0.4 ng/dL would be gone and a smidgen of the extra, bound PSA as well, so you might see 4.29 ng/dL. Wait two more hours and three quarters of the extra free PSA is gone so you're at around 4.25 ng/dL.

So, to circle back to the question: how long before your blood draw for a PSA test should you avoid the bull riding machine at your local redneck bar? 48 hours? Four days? A week? I guess it sorta depends how neurotic you want to be. The extra PSA will have a high initial clearing rate (while your body eliminates the unbound PSA that goes quickly) and then will clear more slowly, about half every two days. But, since that extra PSA will only start at, say, 10 percent of your expected reading, that means a man who ought to get a 2.0 ng/dL would get a 2.2 if he has sex with his venipuncturist right before she pokes him. If he waits something like 12 hours that initial high clearance rate should have dropped his total PSA down to, maybe, 2.1. Then, if he waits another 32 hours he should have cleared half of the bound PSA and be down to 2.05. That's probably good enough. PSA is far from a precise measurement (for men who still have their prostates) and a couple of percent just gets lost in the noise.

10% and fast half-life. So it really doesn't matter.

So my 59 could have been a 54 if Lee hadn't been frisky the prior evening. What a difference that would have made.

Thanks for that explanation. To take it a little further, and perhaps too far...

PSA is first produced as a precurser form (proPSA) with 7 extra amino acids. They must be snipped off to render the PSA active. The activated PSA gets attached to a protein called ACT. The non-activated PSA is “free” or unattached. Some of the free PSA is due to BPH, some is due to PC, and some comes from other sources. The Free PSA that results from PC still has 2 extra amino acids attached to it. It is called [-2] Pro PSA or p2PSA. It is never produced from BPH or prostatitis. Beckman-Coulter found that the most discriminating use of this marker is the prostate health index or PHI.

PHI= (p2PSA/free PSA ) x √PSA

Risk of PC goes up as PHI goes up, as p2PSA and PSA go up, and as free PSA goes down. The square root discounts the effect of PSA that may get high because of prostatitis or BPH. (In some studies they use log PSA to accomplish the same thing, or they use wide ranges (0-10, 11-20, 21+) to create risk categories because small differences in PSA are inconsequential prior to primary therapy).

PSA is one of a class of proteins called "kallikreins." There is a test, called 4Kscore, that detects kallikreins that are more specific to prostate cancer. There is another new test called isoPSA that does much the same thing. All 3 have about the same diagnostic accuracy, but PHI is cheaper and covered by insurance, so I recommend that one.

And.... math! I feel all warm inside.

Worried Guy,

Here is a recent study (from Brazil) where they measured total, free, and complexed PSA prior to-, and at one, five, 24 and 72 hours after sending the men into a little room with a sample bottle to collect semen. (The researchers didn't do any tests on the semen, it simply served as a man's ticket to get into the study group as opposed to the control group.)

Effect of ejaculation on Serum Prostate-Specific Antigen concentration

Their numbers are roughly similar to the cycling results I linked. At one hour after ejaculation they found a 10.52±1.64 percent increase in total PSA, a 9.91±3.00 percent increase in free PSA, and only a 4.26±2.08 percent increase in complexed PSA. These effects lingered at five hours at slightly lower levels and were mostly gone at 24 hours. (All the error bars at 24 hours include a zero value.)

Of course this study was only guys whacking off in the back of the lab. To get a result that better matches your anecdotal report one would need to examine the PSA kinetics of men whose names appear second in the credits of gay pörno films, at one, five, 24 and 72 hours after a performance.

While you are here, and since Redwing is Jonesing for more math, can you think of a way to put some math behind my intuitive argument that since the delivery of PSA to the blood by processes other than cancer is bursty, it has the potential to show higher proportions of the faster-eliminated free PSA (from a recent burst) whereas the slow accumulation typical of prostate cancer cannot achieve the same proportions.

My start on the notion would be to model PSA as a series of discrete impulses, each adding a particular amount of PSA to the blood at a particular time. On could then define a function that maps from a set of impulses in the recent past to the expected current concentrations of free and complexed PSA. The steady release of PSA that is typical of prostate cancer would be approximated by frequent, evenly spaced impulses each contributing a small amount of PSA. The limit as the time between impulses goes to zero should equal the steady state equilibrium levels you would get using differential equations (and, actually, would be the same thing.) For conditions other than cancer the model would characterize different sets of impulses -- with semi-random distributions of amount of PSA and interval between impulses -- and try to determine the probability distribution for various values of free PSA ratio...

TurpT1A,

That is great data!!!
My mantra is "If you want consistent results you yourself need to be as consistent as possible."
I use the same lab, the same time (morning) on the same day of the week. I go on an empty stomach and abstain from sex - even though I don't have a prostate!!!

I even eat at the same restaurant after the test. I consider that follow-through.

Peter DA,
I like the idea of your model You split between pulses that decay with a certain half life. That simulates sex, bicycle riding, DRE, and the cancer production. Why do you consider the latter a fast series of pulses? I picture it as a steady state rate that increases at rate that increases with the PSADT. We don't know whether the production is a series of pulses that decay and "average out " to a level or if it is the level. I'm not sure if there is a significant difference between the two. If not, the steady state would be easier to model.