“Hopes rise for a personalised test for prostate cancer,” according to the Daily Mail. The newspaper says that the blood test routinely used to spot signs of the cancer can be made more accurate if it is used in conjunction with a man’s genetic information.
This news is based on research that looked at improving the predictive power of the prostate-specific antigen (PSA) test commonly used to help detect prostate cancer. When used alone the test can be unreliable as PSA levels, which may indicate cancer, can be raised by a number of factors, such as benign prostate growth or medication use. Equally, not all prostate cancer leads to raised PSA levels. Bearing in mind the limitations of the PSA test, the researchers performed a number of genetic analyses to identify mutations linked to high PSA and prostate cancer. They found that combining genetics with PSA results was more accurate than relying on the test alone.
This type of study is a useful foundation for improving the performance of the PSA test. Further research in this area would need to optimise the performance of the test and assess its ability to reduce prostate cancer deaths before it could be widely used as a screening tool.
Where did the story come from?
The study was carried out by researchers working for deCODE genetics, a private company in Iceland, and collaborators from universities in Cambridge, Spain, Romania, USA and the Netherlands. No funding sources were reported. It was published in the peer-reviewed journal Science Translational Medicine.
The research was covered well by the Daily Mail, which reflected the current problems with the PSA test well and highlighted the preliminary nature of this research.
What kind of research was this?
The prostate specific antigen (PSA) is a protein released by cells in the prostate gland. It can be used to test for prostate cancer as some men with prostate cancer have raised PSA levels. However, while some have suggested that a PSA test could be used as a mass screening tool (given to all men regardless of the presence of symptoms), the issue is controversial as the test only has moderate accuracy. This is because PSA levels naturally vary between men, and PSA is not a very specific marker for prostate cancer, as levels may rise following benign changes to the prostate, some medications or inflammation. This means that in a considerable proportion of men the PSA test fails to detect the disease and in others it gives false positive results.
The researchers report that around 40% of the variation in PSA levels is due to inherited factors. In this research they sought to look at the DNA of a large group of men to see whether they could identify SNPs (single ‘letter’ variations in their genetic code) that were associated with high or low PSA levels. They hoped that any variants identified could be used to adjust PSA test results to account for the inherited variation in PSA levels, making it a better predictor of which PSA increases were specifically due to cancer.
What did the research involve?
The researchers had access to information on PSA values from 15,757 Icelandic men who had been tested from 1994 to 2009, and did not have prostate cancer. They also had similar samples from the Prostate Testing for Cancer and Treatment trial, which was carried out in the UK. This included data for:
- 524 men with PSA values greater than three nanogrammes (ng)/ml who were diagnosed with prostate cancer after a needle biopsy of their prostate
- 960 men with PSA values between 3 and 10ng/ml prostate cancer who were confirmed as not having prostate cancer after they were given a biopsy
- 454 men with PSA values less than 3ng/ml who had not undergone biopsy
There is no consensus on the best threshold PSA level above which men should be given a biopsy to test for prostate cancer, but PSA levels in the range of 2.5–4ng/ml are commonly used.
With the data from the Icelandic men, the researchers performed a genome-wide association study to look for small variations in the genetic sequences of the men’s DNA, which they could then relate to each man’s PSA values. They then looked at whether any SNPs were associated with having a negative prostate biopsy result in 3,834 men who had biopsies. This was to determine whether men with raised PSA levels due to their genetic make-up were having biopsies that turned out to be unnecessary.
They also looked at whether the SNPs identified were also associated with risk of prostate cancer, by looking at their presence in 5,325 prostate cancer cases and 41,417 unaffected control subjects from Iceland, the Netherlands, Spain, Romania and the United States.
Finally, they used the genetic variations they identified to determine what PSA level was “normal” for each individual and whether accounting for genetics would improve the ability of the PSA test to distinguish between men with and without prostate cancer. They also looked at whether adding genetic information about 23 genetic variants associated with prostate cancer in other studies would also improve the ability of the PSA test to distinguish between men with and without prostate cancer.
What were the basic results?
In the genome-wide analysis they found that variations in six regions of DNA were associated with men’s PSA levels. They found that the strongest association was for variations in a region of DNA containing the gene that encodes the PSA protein (a site called KLK3). These variations were estimated to account for about 4.2% of the variability in PSA levels in the Icelandic sample, and 11.8% of the variability in the UK sample.
Among 3,834 men who underwent a prostate biopsy, they found that three of these variations were also associated with having a negative biopsy for prostate cancer. The researchers calculated an odds ratio between 1.15 and 1.27, meaning that if a man had a DNA variant in these regions associated with high PSA he would be 15 to 27% more likely to have a negative biopsy result than men who did not.
The researchers then compared the presence of the six variations associated with higher PSA levels in men with and without prostate cancer. They found that four of the variations were also associated with a higher likelihood of prostate cancer. The other two variations were associated with higher PSA levels only.
The researchers then used various models to look at how adding a person’s genetic information might potentially improve the PSA test’s ability to distinguish between men with and without prostate cancer. They found that taking into account just the six variants they had linked to PSA levels improved the performance of the PSA test, but not by much. A model that combined an adjustment for genetic variations associated with high PSA levels and genetic variations associated with prostate cancer risk was the most accurate.
How did the researchers interpret the results?
The researchers said that they have identified variations in six DNA regions associated with PSA levels. They said that, of the four models that they produced in order to predict biopsy outcome in men who had high PSA levels, the greatest improvement in prediction accuracy was seen when both the genetic factors associated with high PSA and with increased prostate cancer risk were taken into account.
They said that “for a screening test as important and as widely used as the PSA test, having a better way to interpret the measured serum PSA levels is likely to improve substantially the clinical usefulness of the test”.
This well-conducted research found that it is possible to increase the predictive power of the PSA test by taking into account genetic factors associated with higher levels of PSA and increased risk of prostate cancer. This is a useful step towards improving the performance of the PSA test for identifying prostate cancer. Use of the PSA test alone produces a high rate of false positive and false negative rates, leading to some men undergoing unnecessary biopsies and some cases of prostate cancer being undetected.
The researchers highlighted that they had based their analysis mostly on data from two populations, from Iceland and the UK, and further large prospective studies with mixed populations would be needed to see whether these findings could be applied generally.
Lastly, the models used in this study did not include other factors that may have influenced the results, such as age, ethnicity and family history of the disease. These too would ideally be tested for inclusion in a model aimed at improving how well the PSA test identifies prostate cancer in individuals.
Once they have been optimised, such models would need to be tested in clinical trials to determine whether they have the power to reduce deaths from prostate cancer.
Links to the headlines
Hopes rise for a personalised test for prostate cancer after genetic breakthrough. Daily Mail, December 16 2010
Links to the science
Gudmundsson J, Besenbacher S, Sulem P et al. Genetic Correction of PSA Values Using Sequence Variants Associated with PSA Levels. Science Translational Medicine, 15 December 2010: Vol. 2, Issue 62, p. 62ra92
Witte JS. Personalized Prostate Cancer Screening: Improving PSA Tests with Genomic Information. Science Translational Medicine, December 2010: Vol. 2, Issue 62, p. 62ps55