Genetics, PSA, and steps toward a better prostate cancer risk management strategy


A new analysis of a large data set compiled by Kaiser Permanente suggests that combining PSA data with genetic risk data may offer a much more effective way to assess risk for clinically significant prostate cancer.

Now it is probably fair to say that many American consumers would have been dubious about the chances that a large managed care organization would actually want to make sure that it was finding men at significant risk for prostate cancer with accuracy by using the PSA test. Surprise!

This new study by Hoffman et al., which is available as a full-text article in Nature Communications, is based on findings from a so-called “genome-wide association study” (GWAS) of PSA levels in a total of > 45,000 men (28,503 in the Kaiser Permanente database and another 17,428 in so-called “replication cohorts” that could be used to check the initial findings). PSA data on these 45,000+ men include nearly 500,000 PSA measurements going back to the early 1990s.

The research team (which comprised staff from both Kaiser Permanente and the University of California, San Francisco) describe the following key findings:

  • Detection of 40 genome-wide significant (P < 5 × 10−8) single nucleotide polymorphisms (SNPs):
    • 19 novel SNPs
    • 15 SNPs that had previously been identified for PSA (14 of which were also prostate cancer-associated)
    • 6 SNPs previously identified for prostate cancer only
  • At least half of the 40 SNPs are PSA-associated independent of prostate cancer.
  • The 40 SNPs explain 9.5 percent of PSA variation in non-Hispanic whites.
  • The remaining GWAS SNPs explain an additional 31.7 percent of PSA variation.
    • This percentage is higher in younger men, supporting the genetic basis of PSA levels.

So what does any of this mean in terms of evaluation of men’s risk for prostate cancer?

We know that PSA tests on their own are not good indicators of prostate cancer risk. They simply aren’t sensitive enough, and they lead to a lot of men having unnecessary biopsies and then unnecessary treatment too, which comes with more risk (for complications and side effects) than benefit (in terms of extended survival and avoidance of metastatic disease). But it has also been well understood that not testing for risk of prostate cancer increases risk for not diagnosing sufficiently early the forms of aggressive prostate cancer that really do need to be treated.

Because of the unusually detailed data that Kaiser Permanente has collected on its members over the past 30 years and longer, according to one of the co-senior authors, it was possible for the research team

to link every man in the study to our electronic clinical data and determine not just that they had a test, but also the level and frequency of testing over many years.

Now the current study was not sensitive enough to identify the specific genetic loci responsible for the additional variation is risk introduced by the genetic data. And, in fact, the results further suggest that additional genetic predictors of PSA levels are likely to be uncovered in further research. But

If we can gain increased understanding of how a given patient’s genetic predisposition is related to his elevated PSA, this could help clinicians to better evaluate PSA test results and so better predict the patient’s actual risk of prostate cancer — maybe by “normalizing” the results of PSA testing based on each individual’s natural PSA levels or maybe by adjusting the threshold(s) used to determine if a test result should trigger further testing (inclusive of prostate biopsies or at least multiparametric MRI scans when necessary).

This process isn’t “ready for prime time” as yet, but we continue to make progress on how best to combine data categories from differing types of medical information resource to move toward a more sophisticated paradigm for deciding who is at real and definable risk for clinically significant prostate cancer and who really isn’t. And if we can refine this paradigm significantly, we should be able to become much, much better at treating the men who need to be treated and sparing the men who are at zero risk of clinically significant, progressive prostate cancer.

8 Responses

  1. than benefit (in terms of extended survival

    You continue to misrepresent that there is an extended survival benefit associated with PSA screening. PSA screening has no benefits and many motivated harms.

    1. No trial has ever shown an improvement in overall survival (a.k.a extended survival benefit). This is by far the most important fact, far more important than anything about disease specific survival.

    2. The one randomised “controlled” trial that claimed a disease specific survival benefit used a flawed statistical analysis that failed to account for clustering (into countries). When clustering is accounted for, there is no statistically significant disease specific survival benefit.

  2. Using previously published genomic studies that were available in 2013 and their own saliva test, 23andMe® rated prostate cancer as my #1 disease risk. I was impressed because I had already completed my treatment in 2009 after ordinary biopsy evaluations showed the presence of intermediate-risk prostate cancer. However, the FDA subsequently stopped 23andMe® from disclosing disease risks, saying customers could misuse the results to order their medical providers to refer them for possibly unnecessary treatments. I know that I’m just a single case, but for me, this inexpensive test validated my other PSA and Gleason scores.

  3. Dear Mr. O’Neill:

    I am not sure where you are getting the idea that we believe there is a survival benefit associated with PSA screening (let alone “misrepresent that there is an extended survival benefit” associated with such screening). In fact, for the best part of the past 20 years, we have argued that this is unproven. We were critical of the very design of the PLCO trial back in the 1990s. We have constantly stated that the ERSPC trial was not really a trial at all, but rather a meta-analysis of seven different (mostly not well conducted) national trials. And we have always distinguished extremely carefully between the potential value of individualized PSA testing (based on risk factors) and the lack of proven benefit of mass, population-based screening.

    If you can find anywhere on this web site where we have stated that the value of mass PSA screening has been demonstrated, please do let me know! We certainly don’t say this or even imply it in the report above.

  4. Dear Rob:

    I have great respect and interest in what 23andMe has been doing and tried to do over the past 10 to 15 years. However, I have always believed that they “jumped the gun” when they started telling people that they were at risk for specific disorders because initially they did so on the basis of epidemiological averages as opposed to on the basis of validated clinical data. As in your case, if one uses epidemiological averages, you are certain to be right some of the time, … but you are also certain to be wrong a lot of the time.

    It is my understanding that 23andMe has now been working closely with the FDA for quite a while to be able to provide such information on a much more scientific and clinically validated basis, which will make the value of their projections a great deal more useful (and less risky) for individuals in the future.

  5. Dear Sitemaster:

    You are right, of course. That is how they derived those original risk assessments. They are currently working closely with the FDA but each disease assessment is now separately reviewed and officially approved before they provide or update a customer report. Understandably, it’s a slow process. They are also doing a lot of non-medical research and even though I got my initial results back in 2013, they are constantly updating old data and even creating new reports. It’s a gift to me that just keeps on giving.

  6. Nice job in summarizing the results of a very detailed and complicated study. One finding caught my eye because it has become a pet peeve of mine.

    In looking at the PSA data for the men in the study and making correlations with genetic data, the authors transformed PSA test results to base 10 logrithmic values. This is because without this transformation, PSA values are not on a true ratio scale and cannot be compared with other biological values that do have this property; for example, you cannot compute a coefficient of variation on data which correspond to an interval scale. This is why PSA data should always be graphed on a semi-log scale, so that the data do not have to be transformed to log values in order to be visually analyzed.

    I realize that this will seem like an esoteric point to some people. However, if you want to work in the word of natural science, you must be able to compare apples to apples and this means dealing with data sets that have the same mathematical properties.

    Readers interested in understanding the differences between ratio and interval scales can go to Wikipedia and read the examples provided.

  7. @Chris O’Neill ….. while PSA screening may not show any epidemiological benefits, tell that to the 30,000 to 40,000 men who are diagnosed annually with high-risk disease. Most all are asymptomatic and, without screening, their survival would certainly be shortened.

    Sitemaster and I have respectfully argued this point for many years, so we appreciate each other’s position — there is no need to belabor either of our points of view. But when a new name tells me there is no benefit to PSA screening, I say — clearly you were not diagnosed with high-risk disease as a result of a PSA test!

  8. Eighteen years ago, at age 50, my husband had a prostate full of cancer (Gleason 7) with a PSA of 1.7. It was the DRE that saved him. He had an ultrasound and biopsy at the same visit and a prostatectomy 2 months later at the Mayo Clinic in Rochester, MN. He is alive and well today.

    His father was diagnosed at age 78, also with a normal PSA. I don’t know his Gleason score, but he was treated with hormones and died 12 years later from pulmonary issues.

    So, in these two cases with probable hereditary links, the PSAs were normal. I guess what I’m trying to say is the PSA is merely one small piece to the puzzle. It didn’t save my husband’s life, his medical team did. Human beings are unique and complex, and not just numbers and statistics.

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: