Just how accurate are current genetic/genomic tests in the diagnosis and prognosis of prostate cancer?

As we have pointed out regularly over the years, all prostate cancers are by no means the same. And a recently published paper in European Urology makes the inherent diversity in the subtypes of prostate cancer even clearer — even within any one specific patient.

Wei et al. took tissue specimens from the prostates of four consecutive patients, all of whom had been treated by radical prostatectomy at Roswell Park Cancer Institute between June and July 2014. These patients all, originally, had met NCCN criteria for intermediate- or high-risk prostate cancer after their initial biopsy, and none of them received any type of neoadjuvant therapy prior to their surgery.

The research group then used high quality DNA and RNA sequencing techniques and related tests to investigate the genetic and genomic diversity of the forms of the cancers present in the tissue specimens from all four patients. And the key point here is that the genetic/genomic sequencing was not carried out on just one core from one tissue sample. Rather, for each patient, they sequenced tissue from three samples within the index lesion and also from one sample from each of three other, smaller lesions.

This is in contrast to what commonly happens clinically today. If your biopsy tissues are sent for genetic/genomic testing today (using a test like the Oncotype DX, Prolaris, and Decipher assays), what happens is that a single pre-treatment biopsy sample is tested, and that tissue sample has almost invariably been taken from what is thought to be the index lesion at the time of biopsy.

When the research team looked at many post-surgical tissue samples like this — as opposed to just a single pre-treatment biopsy sample, they found (perhaps less than surprisingly) that there was significant variation in the genetic and genomic findings from the different samples — even within individual patients — and that these findings had implications for the risks from the cancers for each patient. Why? Because the potentially most aggressive subtypes of the cancer (based on the genetic subtypes and how they were expressed in the tumors) were not necessarily found in the index lesions at all and because, even within an individual’s index lesion, there could be variation in the subtypes of the cancers. To quote the authors precisely:

There was considerable variability in genomic alterations among [prostate cancer]  cores, and between RNA- and DNA-based platforms. Heterogeneity was found in molecular grouping of individual [prostate cancer] foci and the activity of gene sets underlying the assays for risk stratification and androgen receptor activity, and was validated in independent genomic data sets. Determination of the implications for clinical decision-making requires follow-up studies.

What this really means is that if you get a test like the Oncotype DX, Prolaris, or Decipher assay today, and that test is being carried out on a single tissue sample from a single biopsy core, no one can be sure that that test has actually been carried out on the tissue sample that carries the greatest risk for an aggressive form of prostate cancer.

In their “Patient summary” as part of the abstract of this paper, the authors also state that:

We examined the molecular composition of individual cancers in a patient’s prostate. We found a lot of genetic diversity among these cancers, and concluded that information from a single cancer biopsy is not sufficient to guide treatment decisions.

This paper is also addressed in a media release from the Cleveland Clinic, as well as in an editorial by Lamb et al. in European Urology. (which we have not been able to access).

Now we need to treat this information with a degree of caution. It does not necessarily mean that having a genetic or genomic test carried out on tissue from a biopsy sample from an index lesion is a bad idea. What it means is that the result of that test is not going to be able to provide you with any type of “absolute truth” about risk for an individual patient.

As the authors note in their paper — and as the media release from the Cleveland Clinic repeats:

the use of genomic analysis to personalize treatment plans is in its infancy and … many more large studies will be required to develop next-generation prognostic tools that can be relied on to guide treatment selection and planning for men with prostate cancer.

On top of that, the genetic makeup of the cancer cells is not the only crucial factor. Individuals have differing levels of immunobiological ability to suppress expression of cancers, and the precise biochemical microenvironment in which specific cells are growing can also be profoundly important to their ability to grow and metastasize. Sorting all of this out is almost certainly going to take decades.

16 Responses

  1. “On top of that, the genetic makeup of the cancer cells is not the only crucial factor. Individuals have differing levels of immunobiological ability to suppress expression of cancers, and the precise biochemical microenvironment in which specific cells are growing can also be profoundly important to their ability to grow and metastasize. Sorting all of this out is almost certainly going to take decades.”

    Your closings is probably more profound than the lead.

    To use a rather tasteless sports metaphor; “Defense wins games.”

    Should genomic testing be focusing on defense first rather, than offense?

    If each genetic fault requires a specific genetic cure then the odds for a generic cure are slim. Odds of a broader acting genetic defense just seem higher.

  2. This confirms similar studies that found wide genetic heterogeneity within a single tumor (see this link), that found that tumors within a single prostate may arise independently (see this link), and that a lethal metastatic cell can arise from a small, relatively low-grade cancer focus in the prostate, and not from the index tumor (see this link).

    This has implications for focal treatment as well. The guiding principle of focal ablation is that it doesn’t matter that prostate cancer is multifocal because all the smaller foci spawn off from a single index lesion. According to that theory, ablating the index tumor renders the cancer non-lethal in that patient. Clearly the theory is wrong, at least in some patients. This makes me increasingly wary of focal ablation.

  3. This is absolutely no surprise to me. I see many advocating these tests to assess treatment possibilities. We discuss at SWOG the value of molecular testing being applied to treatment assessment and pretty much all agree to say ~ maybe one day we can. But not today! In another group of physicians I have worked with the discussion of getting BRCA tests was met with the same answer. Maybe we can suggest that a certain BRCA positive Gleason 6 candidate should not consider active surveillance?

    SWOG and other research co-ops are currently searching their bio-depositories and trying to connect outcomes to see where the hypothesis can be made to lead to clinical trials to prove the possibilities that one day we can connect the dots and arrive at treatment decisions. SWOG has great depositories from SELECT and PCPT and other trials. But the work will take quite some time before trial suggestions come about. But all of SWOG’s trials are now gathering tissue for DNA and RNA sequencing. Same also with other research groups.

    Interestingly, Wei et al. gives us a glimpse of the heterogeneity of early prostate cancer. But Dr. Pete Nelson gives us a glimpse of the homogeneity of metastatic prostate cancer in this document

    This makes sense. I have heard it said many times that prostate cancer is two diseases. The slow version seems to be made up of several genetic signatures. Once the disease becomes fast and aggressive it seems to largely maintain it’s genetic make-up in the metastatic lesions that pop up and spread.

    Fascinating! To say the least.

  4. Tony, Alan, Sitemaster and others:

    The Wei and Nelson studies may NOT be inconsistent …

    … Wei’s observations are based on men with early intermediate- and high-risk disease, while Nelson’s work examines metastatic patients with multiple lesions.

    I suspect that as the disease morphs it may become much more homogeneous. Thoughts …

  5. Rick:

    When cells undergo EMT and metastasize there are certainly characteristic changes. You are certainly correct that metastatic cells are less well differentiated than their progenitor cells. The lack of differentiation means that many genetic systems get turned off as the cell metastasizes. Metastatic prostate cancer is certainly a different disease from non-metastatic prostate cancer. There is no question about that, and we’ve known that for about a century. We’ve known for a long time that certain genes get turned off, particularly tumor suppressor genes (like p53) and ones involved in apoptosis, and other genes that code for growth factors get turned on. But the question is not about metastatic tumors being different from pre-metastatic tumors.

    The question that Nelson asked was whether there was genetic diversity in the metastases of different men (there was, and all previous studies affirm that) and whether there was genetic diversity in the metastases of a single man (there was not, and this is controversial). Nelson’s work supports a clonal origin for all tumors. He looked at certain genes (e.g., E2F1, RB1, FA-complex genes, ATM, etc.) and androgen receptor activity and found them to be similar across metastases within the same man (but different between men). Some other studies, perhaps looking at different genes or setting a different bar for what qualified as a meaningful difference in a gene’s DNA, had different findings.

    This has implications. If metastases have a clonal origin from some mother tumor, it supports the index lesion theory and focal therapy; it supports debulking by removing the primary source; and it supports personalized medicine (because all of a given man’s tumors could be treated with the same therapy).

    If different metastases have independent origins in the same man (as some of the other studies show), then the above therapies will not work well.

    It may be that all the studies are true. Some metastases are spread clonally from the “mother ship,” and some are not. There are also known epigenetic factors that change genes depending on the environment they find themselves in. The “mixed model” hypothesis is supported by: (1) the fact that some (but not all) men do seem to be cured by focal therapy, and (2) early retrospective studies that show an association between debulking the primary and longer survival (but not complete remission). The implication for personalized medicine is that a single remedy will not eliminate the cancer burden for any given man, although it may reduce it and lengthen his survival.

  6. Can prostate cancer be cured?

  7. Octogenarians are genetically abnormal as well. Superior genetic ability to control various cancers. What percentage of cancer research is spent studying their abnormalities?

  8. Dear Vincent:

    Some prostate cancers can quite certainly be cured … although knowing for certain that they have been cured may take many years.

    Other prostate cancers can only very rarely be “cured”, and knowing whether they have been cured is even harder.

    And then there are prostate cancers that we have no idea how to cure … but some of those forms of prostate cancer can occasionally go into long-term remission for reasons we don’t understand at all, leaving the patient apparently “cured” and with no prostate cancer.

    In other words the simplest answer is “Yes. Quite often. And particularly in men who have localized prostate cancers that are confined to the prostate and very nearby tissues.” On the other hand, curing metastatic prostate cancers that have spread throughout a patient’s body is a very rare event indeed.

  9. Dear Michael:

    We are all “genetically abnormal” since no two people have exactly the same genetic make-up (not even identical twins as I understand this).

    The question as to why some people live significantly longer than others, however, is not just a matter of genetics. It is based on a whole range of factors that we really don’t understand very well at all.

    I am not aware of any cancer research that is specifically focused on the genetics of the long-lived (octogenarians or even older too). I am not even sure that it would be easy to tease out specific information about why such patients either don’t get or can the resist the development of varied forms of cancer. I am aware that cancer is extremely rare in my family, and that living into one’s late 80s and even late 90s has been commonplace for the best part of the last 200 years. Thus, I have long suspected that members of my family tend to have very efficient immunological systems … but that is about the limit of what I know on this topic.

  10. Thanks so much for your comprehensive answer Alan. …

    The question Len Sierra and I have been discussing is whether a prostate biopsy from a multi-metastatic man falls within Nelson homogeneity hypothesis within the same man.

    For a man that has his prostate intact, a prostate biopsy can be a much simpler procedure than a bone or visceral biopsy … but is it still representative of most tumors for that man’s disease?

  11. Dear Rick:

    I think the answer to your and Len’s question is a resounding “Maybe”. But I’ll be interested in what Allen has to say.

    What I mean is that for some men the answer may be “Yes” and for others it will certainly be “No”. The problem is that one can’t tell unless one does both biopsy types and the prostate biopsy is inclusive of multiple cores.

    The costs associated with doing these types of study are considerable … and from a practical point of view in the community setting they would be impossible.

  12. Sitemaster,

    My mother is 97, I have paternal aunts with a 95 and a 95 and going year track record. Now, either the three in their nearly 300 hundred years never had a precancerous cell or they possess rather superior genetics able to detect and destroy precancerous cells. I on the other hand at the age of 58 was diagnosed with bladder, prostate, and squamous cell skin cancer. I apparently did not inherit those detect and destroy genetics.

    On the surface it would appear my tree has polar opposite cancer genetics. Neither set of genes is likely to yield a simple answer in understanding the role of genetics in cancer. However, between the two extremes, logic says the successful gene set is more likely to yield a more useful understanding than studying my fractured set of defective genes. Simpler in that the superior is able to detect a wider range of precancerous cells. At least, to myself, 300 cancer-free years would indicate that to be factual. Yet, my gene set is more likely to be studied.

    In part that is simply because doctors study the ill by training and the ill’s needs are the more immediate. In part it is also that the money is in the cure not in the prevention. I do not expect private money to shift focus, it would be wise, in my view, if government research dollars worked on the successful genetics as well as the defective genetics.

  13. Michael:

    And I would entirely agree with your suggestion. However, getting there will require political and practical will at the NIH.

  14. Rick,

    I agree with the Sitemaster. It depends on whether the biopsy hits the place that is the progenitor tumor of most of the metastases. How can one know? Actually, biopsies of metastases can be even easier than a biopsy of the prostate, if the metastases are large enough and accessible enough. When they do that, they are usually looking for the phenotype (via staining and microscope) rather than the genotype (via DNA sequencing). Metastases that include rare cancer types (e.g., small cell) may require different kinds of chemotherapy and may be resistant to radiation and/or hormone therapy.

    There is a large project that NCI is now running called NCI-MATCH. It will DNA-sequence a vast array of tumors and cancers (including prostate cancer) to see if there is a genetic mutation in the cancer for which a medicine is already available. They require a recent tissue sample. For more details about it , see this link and this link.

  15. Since here is some evidence that metastatic tumors can overtake the primary site, it is feasible a primary site could morph into a homogeneous mass consistent with metastatic lesions.

    Having known too many men who underwent biopsies of metastatic lesions, I would suggest that in most cases a prostate biopsy is an easier procedure to tolerate. All too often metastatic sites are small and hard to access, not to mention dangerous. Invariably biopsies are now looking for genomic information.

    Thanks to Paul Hobson in Oz for bringing this article to our attention during tonight’s Answer Cancer recurrent/high-risk virtual support call.

  16. Dear Rick:

    Looking at the data from these four patients, one is tempted to conclude that there may be significant differences between those patients with relatively local metastasis (e.g., in the iliac crest within the pelvis) and those with truly distant metastasis (e.g., in the shoulder).

    What this research team has shown is that such “reverse metastasis” back to the primary organ can happen, which is not exactly a surprise. A much more important question is how often this happens and how long patients have been metastatic before this becomes evident.

    The question of the meaningfulness of genetic analysis of biopsy data from any single site in men with metastatic disease is not answered in any way by this paper.

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: