How many “different” types of prostate cancer are there?

This, of course, is a rhetorical question. We really have no good idea how many “different” types of prostate cancer there are. And so what we are really dealing with here is how we are going to sub-classify prostate cancers in the future in order to treat them most appropriately at the earliest relevant stage of development.

At present, there are three basic ways to think about the “types” of prostate cancer:

  • According to their cellular type, e.g., as adenocarcinoma of the prostate (by far the most common biological type or prostate cancer) or small cell carcinoma, etc.
  • According to their clinical risk categorization, using the D’Amico or the National Comprehensive Cancer Network risk categories (low-, favorable intermediate-, unfavorable intermediate-, and high-risk disease)
  • According to their genetic make-up based on the expression of specific types of marker

What no one has managed to do yet is to pull all this information together in a coherent manner to tell us — in a really helpful way — how best to manage specific categories of patient who meet certain specific criteria that address all of the above. However, a new article by Ross-Adams et al. in the online journal EBioMedicine may be helpful in this regard (see also this report on the ScienceDaily web site). If you click here, the entire article by Ross-Adams et al. is also freely available.

Basically, Ross-Adams et al. argue that prostate cancers can be classified into five genetically different categories and that this classification system may be helpful to doctors (and presumably their patients too) in deciding on the best course of treatment for each individual patient, based on the characteristics of each patient’s individual tumor. More specifically, here is what they did and found.

  • They used tissue specimens from 259 men diagnosed with primary prostate cancer in the UK.
  • From those tissue specimens, they isolated 482 different samples of cancerous tissue, benign tissue, and so-called “germline” tissue.
  • By analyzing copy number alterations (CNAs) and methods, they identified genomic loci affecting expression levels of messenger RNA (mRNA) in a quantitative manner and then they were able to stratify the patients into subgroups associated with future clinical effects and behaviors of the cancers.
  • They were able to describe five separate patient subgroups with distinct genomic alterations and expression profiles.
  • Using these five subgroups, they were able, consistently, to predict future biochemical relapse after first-line treatment with a high degree of accuracy.
  • They were able to validate their predictions using a different cohort of patients.
  • They confirmed alterations in six genes previously associated with prostate cancer (MAP3K7, MELK, RCBTB2, ELAC2, TPD52, ZBTB4), and identified 94 genes other not previously linked to prostate cancer progression that would not have been detected using either transcript or copy number data alone.
  • They confirmed several known molecular changes associated with high-risk prostate cancer (including MYC amplification, and NKX3-1, RB1, and PTEN deletions, as well as over-expression of PCA3 and AMACR, and loss of MSMB in tumor tissue).
  • They showed that use of the subset of 100 genes described above significantly outperforms established clinical predictors of poor prognosis (PSA, Gleason score) and previously published gene signatures (p = 0.0001).

Perhaps most importantly, Ross-Evans et al. also describe how their molecular profiling methodology can be used for the early detection of aggressive prostate cancer in a clinical setting, and thus inform treatment decisions.

Now let us be very clear that being able to do this sort of study and analysis at the types of major academic institution where the authors of this study work is by no means the same as being able to translate the authors’ methodology into routine clinical practice. For that to happen, there are going to need to be a number of additional steps, of which the most important are the following:

  1. The study data published by Ross-Adams et al. will need independent validation.
  2. The ability to “scale up” this type of gene profiling such that routine biopsy samples can be screened for expression of 100 genes will be essential, and it must be possible to do this in a timely manner.
  3. The test must be available at a cost we can afford and must provide data with a high enough level of accuracy to justify any increase in costs over the available methods.

What this study does help to further demonstrate is the potential of genetic testing to the application of “personalized” medicine. What it also shows us is that there must actually be thousands of different “types” of prostate cancer (based only on the numbers of ways that 100 different genes might be being expressed in an individual patient). Whether the five categories of cancer (described by Ross-Adams et al. as iCluster 1 through iCluster 5, based on the mixtures of genes found in specific groups of patients) is “the best” way to categorize patients based on their genetic profiles will take time to work out.

In the interim, what Ross-Adams et al. appear to suggest is that the value of this type of molecular profiling will be highest when used in conjunction with current clinical risk profiling. To that extent, one might want to look at the molecular profiling protocol they describe as a much enhanced version of the Prolaris and Oncotype tests currently available.

One Response

  1. This may be of interest to you.

Leave a Reply

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

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

Google photo

You are commenting using your Google 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 )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.

%d bloggers like this: