Monday, December 23, 2013

PCa Genetic Tests: Do They Make Any Sense?

The recent report on such a cancer prognostic model such as Oncotype DX by Knezvic et al is a putatively prognostic method used in prostate cancer. Fundamentally what they do is examine cancer cells for the expression of various genes and examine three sets; baseline expressions, excess expressions and reduced expression. They use the baseline to set levels for excess and reduced. They then use the excess or reduced in a one dimension expression to determine a prognostic measure. This seems to be in contrast with work we reported on a few months ago[1].

Like PSA measures, CA125, CEA, and the like, they try to reduce everything to a single number. We argue here that such an approach is problematic at best. Furthermore they fail totally to demonstrate any internal pathway influence. There is no predictive basis for their approach predicated upon the actual dynamics of the cell. It is purely correlative and there may be substantial confounders involved. This approach is an example of what we fell to be the poorer aspects of genomics applied to cancer prognostics.

In a recent study the authors develop a score called the GPS score which is based upon know malignant PCs cells and then argue that then score has significant prognostic value. The authors state:  
The Oncotype DX Prostate Cancer Assay has been clinically validated, demonstrating that the GPS, assessed in diagnostic biopsy tissue, can predict the likelihood of the presence of adverse pathology (high-grade and/or high-stage disease), and that it complements existing pre-treatment risk assessment tools such as PSA levels, Gleason Score, and clinical stage. The assay is intended to help guide treatment decisions in early-stage prostate cancer, including the decision between immediate therapy and active surveillance. As evidence that the analytical assay was designed well for its intended use to test RNA from small biopsies, in a clinical validation study, valid GPS results were generated for more than 95% of samples requiring 1 mm and 30 microns of tumor tissue…

They continue:  

Optimization of the Oncotype DX platform has enabled the development and analytical validation of the Oncotype DX Prostate Cancer Assay for use with prostate biopsy specimens. This RT-PCR assay has been clinically validated to predict the risk of high grade and/or non-organ confined disease at radical prostatectomy using biopsy samples containing as little as 1 mm of tumor tissue. The Oncotype DX Prostate Cancer Assay complements traditional clinical and pathologic diagnostic features and will assist clinicians to discriminate patients with indolent prostate cancer from aggressive prostate cancer to help make the most appropriate treatment decisions. 

The approach is as follows as shown in the Figure below. Basically take the malignant cells and measure the expression of certain genes via their RNA using a baseline reference gene expression level.


Now the genes they have selected are categorized as follows. They have four categories related to PCa and one category for the purpose of setting a reference level.

Stromal Gene
Cellular Organization Group
Androgen Group
Proliferation Group
Reference Genes
BGN
FLNC
FAM13C
TPX2
ARF1
COL1A1
GSN
KLK2

ATP5E
SFRP4
TPM2
AZGP1

CLTC

GSTM2
SRD5A2

GPSI




PGK1

Now based upon the levels of expression of these genes against the gene reference level they have proposed a metric which they term the GPS metric which is a measure of prognostic value related to the aggressiveness of the cancer. The GPS metric is given by:

The higher the GPS measure the arguably the greater the virulence of the cancer. The internal value above is given by:

Finally the specific value calculations by class are given by: 

We summarize this below:
Group
Genes
Weight
Stromal Gene



BGN
0.527

COL1A1
0.457

SFRP4
0.156
Cellular Organization Group



FLNC
0.163

GSN
0.504

TPM2
0.421

GSTM2
0.394
Androgen Group



FAM13C
0.634

KLK2
1.079

AZGP1
0.642

SRD5A2
0.997
Proliferation Group



TPX2
1.00
Reference Genes



ARF1


ATP5E


CLTC


GPSI


PGK1


We present graphically below thevalues as shown above, namely the weights:

From an earlier Press Release there was reported the results of a study stating[2]:

Results showed that the test, developed in collaboration with UCSF and Cleveland Clinic, strongly predicted disease aggressiveness (p=0.002) offering information beyond currently available clinical factors, such as PSA and biopsy Gleason Score, to help physicians and their prostate cancer patients confidently choose the most appropriate treatment based on an individualized risk assessment.

Furthermore, this first-of-its-kind, multi-gene test has been validated to guide treatment decisions using the prostate needle biopsy sample taken before the prostate is removed -- thereby providing the opportunity for low risk patients to avoid invasive treatments such as radical prostatectomy or radiation.

"The results of our study showed that the individual biological information from the Oncotype DX prostate cancer test tripled the number of patients who can more confidently consider active surveillance and avoid unnecessary treatment and its potential side effects. The test also identified a smaller number of patients who, despite seemingly low-risk clinical factors, had more aggressive disease and, would suggest that they consider immediate treatment," said Peter Carroll, M.D., MPH, professor and chair, Department of Urology, UCSF and principal investigator of this validation study.

"With these new study results, I believe we may be able to significantly increase the use of active surveillance, which has been limited to some extent by the absence of a validated genomic tool to more accurately distinguish low and high risk disease at the time of biopsy." Active surveillance is a treatment plan that employs careful and consistent monitoring of the cancer in a man's prostate without removing it. Under active surveillance, patients have regular check-ups and periodic PSA blood tests, clinical exams and potential biopsies to closely monitor for signs of prostate cancer progression.

The Oncotype DX prostate cancer test measures the level of expression of 17 genes across four biological pathways to predict prostate cancer aggressiveness. The test results are reported as a Genomic Prostate Score (GPS) that ranges from 0 to 100 and is combined with other clinical factors to further clarify a man's risk prior to treatment intervention.

Now there are many significant issues in this analysis.

1. The weights are arguably chosen to maximize the risk of missing an aggressive PCa. However I have not yet seen adequate clinical evidence to that effect.

2. Prior proposed genes and the ones included herein are shown below, one from the study currently in discussion and the other from a prior study of a Myriad genetic profile:

Target Genes
Oncotype DX
Housekeeping Genes
Oncotype DX
Target Gene
Myriad

Housekeeping Gene
Myriad
AZGP1
ARF1
ASF1B
CLTC
BGN
ATP5E
ASPM
MMADHC
COL1A1
CLTC
BIRC5
MRFAP1
FAM13C
GPSI
BUB1B
PPP2CA
FLNC
PGK1
C18orf24
PSMA1
GSN

CDC2
PSMC1
GSTM2

CDC20
RPL13A;LOC728658
KLK2

CDCA3
RPL37
SFRP4

CDCA8
RPL38
SRD5A2

CDKN3
RPL4
TPM2

CENPF
RPL8
TPX2

CENPM
RPS29


CEP55
SLC25A3


DLGAP5
TXNL1


DTL
UBA52


FOXM1



KIAA0101



KIF11



KIF20A



MCM10



NUSAP1



ORC6L



PBK



PLK1



PRC1



PTTG1



RAD51



RAD54L



RRM2



TK1



TOP2A


It should be obvious that these two tests are dramatically different. Yet they claim similar results. The question is; what genetic expression has gone astray? Why, for example, do we see such a massive disparity? Frankly, other than CLTC we see no other commonality. What causes these disparate expressions? The answers are left hanging. At least with PSA we have some clear cause and effect. Here, at best, we have some correlative values.

With such disparate sets of genes one wonders why and how can these tests be compared if at all. Or, are these results just suggestive and are neither causative or resulting from the lesions.

3. In the current test under discussion the cells used for extraction are arguably from the prostate biopsy. The Myriad appear to be more wide spread.

4. Are these tests worth anything? Furthermore, groups are offering tests to assess risks based upon genetic profiles. As stated[3]:

Myriad also rolled out new tests. In September, the company launched its myRisk Hereditary Cancer™ test, a 25-gene panel covering eight major cancers (breast, colorectal, endometrial, gastric, melanoma, ovarian, pancreatic, and prostate) at an average selling price of $3,700. In October the company introduced myPlan Lung Cancer, which carries a $3,400 list price; followed in November by myPath Melanoma, which has an average selling price of $1,500. By 2015, Myriad has said, it expects to discontinue several current tests, including the BRACAnalysis test at the center of the Supreme Court case.

Just because some genetic profile may have some correlative relationship the genetic profile is not causative. Tests like these can be costly and of yet to be fully justified clinical value. Take a melanoma, if one has a suspect pigmented lesion then a simple excision and competent path study should suffice. That is an order of magnitude less than the genetic profile. In fact if one were to do a profile it should be of the melanocytes and not of the cells in general.

References

Knezvic, D., et al, Analytical validation of the Oncotype DX prostate cancer assay – a clinical RT-CR assay optimized for prostate needle biopsies, BMC Genomics 2013, 14:690.