The Prostate Health Index (phi test) and the need for biopsy


It will come as no great surprise to regular readers that your Sitemaster has been less than enthusiastic to date about the potential of the Prostate Health Index or phi test as a tool for assessment of risk for prostate cancer. However, new data is starting to give him slightly greater hope regarding the value of this test.

In a paper by de la Calle et al. just published in the Journal of Urology, the authors used prospective data from two independent patient cohorts  in an attempt to assess the ability of the phi test to differentiate between risk for aggressive prostate cancer and risk for indolent or no cancer in a biopsy-naive population. We shall be blunt in saying that the involvement of two highly respected researchers (Drs. Rubin and Sanda) in the development and implementation of this study gives it a stature that has not been quite so evident in some of the earlier research.

The authors used a primary cohort of 561 biopsy-naive patients from one clinical trial site (Cohort A) and another 395 biopsy-naive patients from a secondary trial site (Cohort B)  to build and validate their hypothesis. Specifically, they then set out to compare the diagnostic specificity of the phi test to the specificity of total and %free PSA data by using prostate biopsy results, where aggressive prostate cancers were classified as those with a Gleason score of 7 or higher as opposed to indolent prostate cancers (with a Gleason score of 6 or lower) or no prostate cancer on biopsy.

Here are the core study findings from Cohort A:

  • Higher phi values were significantly associated with a Gleason score of 7 or higher.
  • The area under the curve (AUC) for detection of prostate cancer was 0.815.
  • At 95 percent sensitivity,
    • The specificity of the phi test was 36.0 percent.
    • The specificity of the total PSA level was 17.2 percent.
    • The specificity of the %free PSA level was 19.4 percent.
  • At 95 percent sensitivity for the detection of aggressive prostate cancer
    • The optimal phi cut-point was 24.
    • A cut point of 24 predicted avoidance of 41 percent of unnecessary biopsies.
  • The phi cut-point of 24 allows avoidance of 36 percent of all biopsies, with very few aggressive cancers being missed.
  • All these results were confirmed by repeated studies using Cohort B

The authors conclude that the phi test

detects aggressive prostate cancer with a better specificity than total PSA and percent free PSA in a biopsy-naïve population, and could be a useful tool to decrease unnecessary prostate biopsies.

However, we would want to be very clear that we are not overly excited by these data. Use of the phi test in this manner appears to us to result in

  • A failure to accurately identify 59 percent of all patients who (arguably) may not need a biopsy
  • A failure to accurately identify all patients with potentially aggressive disease (i.e., Gleason 4 + 3 = 7 or higher)

The fact that this test is better than a PSA test and a %free PSA test is certainly something, but other new tests may be able to do better than this. Perhaps it is fair to say that the phi test offers us an improvement over what we have had before, but The “New” Prostate Cancer InfoLink isn’t going to be enthusiastic until we have some sort of test that can eliminate something like 75 percent (or more) of the currently unnecessary biopsies and do that with a 95 percent or higher specificity for the identification of Gleason 4 + 3 = 7 prostate cancer and higher.

18 Responses

  1. I have a very different take on PHI. I think it is certainly the best value for the money of any test out there that can be used to decide if a biopsy is warranted, and I heartily recommend it to all patients who have to decide on a biopsy.

    First, let me say that it has had great names in research attached to it from the start. As you can see in <a href=http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3140702/the 2011 validation study by Catalona et al. that was used to get FDA approval, it has names attached to it like Catalona, Partin, Sanda, Marks, and Slawin (the inventor of the free PSA test).

    It is not so much that PHI is a great test — it’s not — but it’s better than the other two most common tests — PSA and %free PSA and costs less.

    For values between 2.0 ng/ml and 10.0 ng/ml, the range most of us were sent to biopsy with, PSA has an overall accuracy of only 0.53. That means that it’s no better than flipping a coin at predicting whether we will have cancer on biopsy or not. Even worse, it doesn’t have a cut-off that simultaneously gives good sensitivity (low rates of false negatives) and good specificity (low rates of false positives). For example, at a cut-off of 1.0 ng/ml, it will miss 17% of the cancers, but will send 61% of men to unnecessary biopsies. But if we raise the cut-off to, say, 4.0 ng/ml, it will miss 80% of the cancers, but only send 4% to unnecessary biopsies. There is no cut-off value that does a good job with both.

    %Free PSA raises the overall accuracy to 0.65 for men with PSAs between 2.5 and 10.0 ng/ml. However, it suffers the same lack of optimal cut-off that PSA suffers from. If the cutoff for %free PSA is less than 25 percent, it will find almost all the cancers, but 87% of the biopsies will be unnecessary. If only men with less than 10% are biopsied, it will miss more than half of the cancers, but almost all the biopsies will be positive. But most men will have %free PSA in the range of 10-25% (median is 15%). %Free PSA is also related to prostate volume and BPH.

    PHI includes both PSA and free PSA and a third biomarker, -2ProPSA. Its overall accuracy for men with PSAs between 2 and 10 is 0.72 — better, but not great. However, it does have an optimal cut-off. At a cut-off of 35, it sensitivity and specificity are both 65 percent (see <a href=http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3140702/table/T2/Table 2). Would I prefer them to be 90%? Of course! But given my alternatives, I’ll take it! Also, it is less affected by BPH and prostatitis than the other two tests.

    Perhaps the best part is that it only costs $80, and it is covered by Medicare and most insurance. When you get it, you also get PSA and %free PSA as part of it. I can’t see any downside, and only upside to getting it instead of a PSA test. Because it is less affected by BPH and prostatitis, it makes sense to track progression using PHI. There have been suggestions that it may be better for that purpose in the setting of active surveillance too.

    I also looked at other pre-biopsy tests. PHI performs similarly to another test that looks at kallikreins, the 4KScore (see <a href=http://www.sciencedirect.com/science/article/pii/S0302283814007520Nördstrom et al.), which costs five times as much as PHI. 4KScore is not covered by Medicare or most private insurance. PCA3 is only covered after a previous negative biopsy, and costs $200-$300. Because many cancers do not express PCA3, it is a low sensitivity test, but has good specificity. It is a good complement to PHI. Similar considerations apply to the TMPRSS2:ERG fusion gene test. MLabs at the University of Michigan has recently made available a combination of PCA3 and TMPRSS2:ERG gene fusion and calls it Mi-Prostate Score (MiPS). It costs $750, but is not covered by Medicare or insurance. Other genetic tests are in the $3,000-$4,000 range and really only have value (if then!) post biopsy for use in deciding whether active surveillance is a good alternative.

    You can see why I endorse PHI until we have a better, low cost test to screen prospective biopsy patients with.

  2. Yes, this seems like an incremental step forward but with success tested against a low bar. The bar is low because it consists of just PSA and total PSA. These days, thoughtful clinicians are looking at more than that, especially the PSA trend, if there is one, or pattern (such as up and down), as well as DREs, PSA density roughly based on DREs, age, and race, and perhaps other newer investigational tests (to name some of the more prominent touchstones). Comparing the phi test to such a more realistic clinical evaluation would have been ideal, but much messier in research terms and perhaps impractical. All this said, I’m glad the research team took up this hypotheses and published their results.

    I’m wondering whether the phi test can add substantial value to an algorithm for sorting out the wolves and sheep. I’m wondering if it makes sense to track trends or patterns in phi results as we do with PSA.

  3. Jim:

    I always recommend tracking with PHI. I don’t know whether insurance companies will push back eventually, though. Prostate volume has been found to improve the predictive accuracy of PHI in multivariate models, but I’m not sure that “PHI density” is a good parameter because of the mathematics involved. PSA kinetics have been found to be utterly useless for initial diagnosis and have been rejected for that purpose. PHI, which is less affected by all the other causes of increased PSA, is a much more suitable parameter for tracking. Some have suggested it may be more useful for active surveillance as well.

  4. Allen:

    How readily available is the phi test today?

    When you say it is best value for money, how does it compare to regular and ultrasensitive PSA tests in cost, and will insurers cover it?

  5. Any registered physician can order it through Innovative Diagnostic Laboratories. If your physician is not registered with them, he can call them to get registered and receive the proper supplies for drawing blood and sending it in, or you can call the lab for a list of physicians in your area who are already registered and visit one of them. It is FDA approved, so Medicare and most insurers fully cover it. It is comparable in cost to a regular PSA test. I don’t know how much an ultrasensitive PSA costs, but that is a terrible idea for someone who has not had a biopsy or who is on active surveillance.

  6. Chiming in on Allen’s comment to Rick’s question about ultrasensitive testing

    Amen! I’ve been convinced for 14 years that ultrasensitive testing is the way to go for much monitoring of post-treatment PSA and for managing ADT when PSA is below conventional test limits, but it doesn’t make sense before diagnosis. In fact, it is optimized for low-range sensitivity and gives up some capability for measuring high PSA. (At least, that used to be the case.)

  7. While I absolutely agree with Jim and Allen about ultrasensitive testing, I see more often that many urological centers appear to only offer the ultrasensitive test now. For example, at UCSF I think that is the case, and I believe it is true at other NCCN and NCI hospitals. Probably less true at community hospitals and urology offices. My buddy with a PSA in the 100’s gets an ultrasensitive test …?

    The most useful purpose of the ultrasensitive test is in the first 18 months after RP and initially on recurrence at low levels — but once PSA is > 1.0 ng/ml, I would argue it makes little difference.

    The reason I asked the original question that way is because the ultrasensitive test is more expensive than a regular PSA, and I wanted a frame of reference. It was not intended to be a recommendation for men pre-diagnosis.

  8. Hi Rick. We are all on the same page. I can see at least one justification for using an ultrasensitive test when the PSA is above the lower limit of conventional test kits: comparability with prior results to get a better idea of trend. That has been done at times in my own case when I passed the threshold of conventional detectability while my PSA was increasing during the vacation from ADT as part of my IADT program.

    Also, ultrasensitive tests give third digit reliability, which may be worthwhile in interpreting trends for some patients with low PSAs that are above the conventional test threshold. Generally, though, my primary oncologist has used conventional PSA tests when I have been in the conventional range. At present (and hopefully in the future) I will be using ultrasensitive tests after radiation in 2013 as my PSAs have not exceeded 0.02, which is lower than I expected at 1 year out from the end of adjuvant ADT.

  9. What do you make of a PSA of 11 followed up by a PHI test of 41?

  10. Dear Greta:

    I think that anyone would need a good deal more information about the patient before they could express an opinion about whether or not he should have a biopsy. Other factors are important here, such as the following:

    — What is his age?
    — Has he had any symptoms of lower urinary tract issues?
    — What is the size/volume of his prostate?

    It really isn’t possible (or appropriate) for anyone to make any sort of suggestions about whether he should or shouldn’t have a biopsy unless one is a physician who has actually examined this patient. However, what I can tell you is that if I was in this patient’s position and I was trying to avoid having a biopsy, I would want to make quite sure that I got a multiparametric MRI to see what that showed before I made any final decisions. In general, most men who have a PSA of 11 and who are otherwise healthy and have a life expectancy of 15 or more years would be wise to have an MRI and then a biopsy if the MRI showed anything suspicious (or they could just go get the biopsy done and worry about what to do when they got the biopsy results).

  11. This patient is my mentally disabled brother. He is 73 years old. He is a Type I diabetic, has diabetic neuropathy, is incontinent of urine, and has an indwelling catheter. He also has chronic bronchitis, hypertension, and diabetic kidney disease. A urologist after doing the PSA and PHI has recommended a biopsy next.

    Given all of his other problems as Medical POA I am just wondering if this biopsy is really necessary or should I just take a wait and see approach. The urologist could find no growths after doing a digital exam. My brother has not complained of any symptoms.

  12. Dear Greta:

    Thank you for the additional information.

    So EVEN IF we were to make the (obviously incorrect) assumption that your brother was in PERFECT health, and then use the Male Life Expectancy Survey for prostate cancer patients on the web site of the Memorial Sloan-Kettering Cancer Center, then we would find that it predicted the following (based on the apparently correct idea that he has a clinical stage of T1c and if I make the assumption that he might be diagnosed with a Gleason score of 4 + 3 = 7):

    If he did absolutely nothing about getting either a biopsy or later treatment, then

    — At 10 years, only 7/100 men like him would have died of untreated prostate cancer, but 35/100 men would have died of something else.
    — At 15 years, only 11/100 men like him would have died of untreated prostate cancer, but 58/100 men would have died of something else.

    But your brother obviously has multiple, significant co-morbidities, vastly increasing the likelihood that he might die of something else other than prostate cancer (if he even has it) in the next 10 to 15 years.

    I think you have two options here:

    (1) Use the survey mentioned above yourself to get a more accurate estimate of your brother’s risk for death from prostate cancer as opposed to other things over the next 10 and 15 years (his clinical stage seems to be T1c, and you can choose the Gleason score of 4 + 3 = 7). My suspicion is that your brother has a very high risk of death from all sorts of things over the next 10 to 15 years that are much more significant that any risk of death from prostate cancer.

    (2) Go and get a second opinion from a specialist in urologic oncology who is a known expert in the diagnosis and management of prostate cancer (who I THINK would tell you that worrying about a diagnosis of prostate cancer in a patient like your brother is probably not the best idea in the world).

    I say this while also making it absolutely clear that IF your brother was unlucky enough to be diagnosed with Gleason 8 to 10 disease (which is possible, albeit rather less likely), then the risk of him having a serious case of prostate cancer that could lead to his death if untreated in the next 10 to 15 years goes up considerably.

    Decisions like this for people like you are never easy. All that I can actually TELL you is that IF I was your brother and I had all the comorbid conditions that you describe (except the mental deficit), I would look at you and say, “Are you nuts!? Forget the biopsy! I don’t even wanna know about that possible problem on top of everything I’m already dealing with!”

  13. Thank you so much for your information. You are echoing a great deal of what I have been thinking!

  14. I am looking for any information I can find. My 42-year-old son has some urinary symptoms, a PSA of 6.2, and a PHI of 32.4. The DRE was negative. My husband, his father, was diagnosed at age 50 with prostate cancer, Gleason 7. He had a prostatectomy (at Mayo, Rochester) and 18 years later he is fine.

    A lot has changed since my husband went through this, but I’m wondering why they didn’t do an ultrasound. My son’s doctor thinks waiting 6 months and running tests again is the best option. I’ve urged my son to get a second opinion. Any thoughts would be greatly appreciated.

  15. I would wait for about a year and then retest. Those figures though abnormal are not THAT out of range.

  16. Mary,

    I can certainly understand why you are worried for your son. It is rare (but not unheard of) for a 42-year-old to be diagnosed with prostate cancer. Certainly he has some genetic risk, but his PHI is below the cut-off that would be a red flag. His urinary issues are more likely caused by non-cancerous causes, like BPH, prostatitis, or urinary obstruction. Those causes should be investigated first before a biopsy is given. I do think an ultrasound is a good idea to measure his prostate size, and many urologists do them routinely. That could help diagnose BPH, and if found, 6 months of Proscar or Avodart, could aid the diagnosis. Sometimes 6 months of waiting will be enough to allow any prostatitis to remit on its own (antibiotics are frowned upon). Waiting 6 months will also provide a second data point to see which way the PSA is heading. Waiting 6 months is not likely to make any difference, even if it is prostate cancer,

  17. Dear Mary:

    Your concern about your son’s condition is entirely understandable. However, I also suspect that your son is also concerned about the potential slippery slope into over-treatment for something that could be all sorts of things other than prostate cancer (as suggested above by Allen Edel).

    If I was your son, I think what I would want to do is go get a second opinion from someone who is a recognized specialist in the diagnosis and management of prostate cancer (as opposed to any old urologist) and ask about getting retested in 3 to 6 months time and getting an ultrasound and a multiparametic MRI scan as well as a repeat PSA test if they seem appropriate 3 or 6 months from now — based on what happens to his PSA and his symptoms in the interim.

    These decisions are always difficult for any man, and they are also difficult for mothers like you who have already been through this all once before. However, I am sure either you or your husband (or both) have already give your son sound guidance based on your prior experiences, and sometimes one has to just take a deep breath and let the patient make his own decisions about what he wants to do.

    As Allen has already noted, a delay of 6 months before anyone has to “do something” given the current circumstances is probably not going to be clinically important, and it is possible that the problem could resolve without any need for invasive intervention. Sometimes the best decision is just to take that deep breath and say you’ll look again in a few months time.

  18. Thanks to all of you for your insight. It does bring some peace of mind. I hope he does get a second opinion though.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: