Can you help us improve NHS Choices?

Take part in our survey to share your views

DNA test for prostate cancer relapse

Behind the Headlines

Tuesday May 8 2012

Genetic changes may predict prostate cancer relapse

“A blood test for men with prostate cancer could detect if the illness is likely to be life-threatening,” the Daily Mail has reported. In some cases prostate cancer can be benign, but in others it can be life-threatening and require treatment such as surgery or chemotherapy.

The news is based on research that looked at whether scanning for genetic abnormalities in various types of tissue could be used to predict whether prostate cancer would rapidly relapse after treatment. To conduct the research, scientists looked at samples of blood, prostate tumours and healthy prostate tissue from 238 men who had undergone surgery to remove their whole prostate gland. They compared genetic differences to the different outcomes the men experienced.

The researchers found that prostate cancer patients had high numbers of genetic mutations called "copy number variations", in which sections of DNA are abnormally repeated or missing. Particular repeats and deletions were more common in patients who went onto have a rapid relapse, and the size of copy number variations tended to be larger in these patients as well. The researchers then used this information to create prediction models based on the DNA in different types of tissue samples.

This study is exciting because it raises hope that one day a test could help predict the outlook for patients with prostate cancer, which could in turn aid treatment decisions. However, these findings will have to be further validated, and a simple, inexpensive test developed and tested before this is a realistic option for doctors.


Where did the story come from?

The study was carried out by researchers from the University of Pittsburgh and was funded by the American Cancer Society, the US National Cancer Institute and the University of Pittsburgh Cancer Institute. The study was published in the peer-reviewed American Journal of Pathology.

The Daily Mail’s coverage of the story was generally accurate, although it may imply that the research could help discern benign prostate tumours from life-threatening ones before treatment. However, the research was conducted in patients after a round of treatment where they had already had surgery to remove all or part of the prostate. Also, the Mail’s headline suggests that the test has been developed and is soon to be implemented, which is not the case. The story correctly stated that “it is likely to be several years before the academics are able to develop a blood test to predict the speed prostate cancer will spread.”


What kind of research was this?

Prostate cancer is not typical of most cancers, as men can live with the condition for many decades without any symptoms or the need for treatment. This is because in most cases the cancer progresses very slowly, at such a rate that they will probably die of other causes or even old age before it causes ill effects on their health. However, a minority of men with the disease experience a more rapid progression of their cancer that requires treatment, such as a prostatectomy (surgery to remove all or part of the prostate).

This cohort study of men with prostate cancer examined whether the presence of genetic abnormalities in various types of tissue could predict whether a person with treated prostate cancer would relapse. It looked at the amount of genetic abnormalities in tumour tissue, normal tissue adjacent to a tumour and in blood.

The specific type of genetic abnormality the researchers looked at is called a copy number variation. This occurs when duplications or deletions in the genetic code cause an abnormal number of copies of a region of DNA. The researchers initially looked at samples from 238 men who had had radical prostatectomy (a surgical procedure where part or all of the prostate gland is removed), and three cell lines (a type of isolated cell used in lab research). They then validated their findings on an additional 25 samples.

This study was appropriately designed to see whether patients who have different outcomes have differences in copy number variation. However, before this technique can be used as a test, it will have to be trialled on a much larger cohort of people, so that researchers can get a clearer picture of its use in clinical settings. For example, researchers will need to know how often the test might miss patients that are likely to relapse, and also how often the test incorrectly suggests a person’s cancer is likely to relapse, which could lead them to have unnecessary further treatment. Also, as the authors note, the techniques used in this study need high-quality DNA, so may be difficult and expensive to perform.


What did the research involve?

The researchers looked at prostate cancer samples, samples of healthy tissue taken from next to a tumour, and blood from men who had had a radical prostatectomy to treat prostate cancer. The researchers extracted DNA from these samples, and then looked at their genome, the entire genetic code they contained. The researchers then compared the results from this analysis to the outcomes experienced by the prostate cancer patients:

  • Approximately one-third of the patients had a relapse soon after surgery, with a median time to progression of 1.9 months.
  • One-third had a relapse but much more slowly, with a median time to progression of 47.4 months.
  • One-third of patients in the cohort were free of cancer for at least five years.

Based on the associations they found, the researchers developed an algorithm for predicting whether a patient would relapse, and how quickly they would relapse. This was based on whether the genetic code at specific locations was repeated or deleted, or on the size of copy number variation found across a person’s genome. They then tested their prediction model on an additional 25 samples.


What were the basic results?

The researchers found that the prostate cancer samples had a large number of genetic abnormalities. Deletions of specific regions occurred at high frequency, and amplification (abnormal repetitions) of other regions occurred in a subset of samples. Healthy tissue adjacent to a tumour also had similar amplification and deletion patterns. The blood of patients with prostate cancer also contained copy number variations, and some of these variations occurred in the same locations within the DNA as they had in the prostate cancer samples.

The researchers then developed a tool to predict whether a cancer would relapse based on DNA regions that had a significant proportion of amplification or deletion in prostate tissue samples (both cancerous and healthy) from patients who relapsed, but not in patients who did not relapse.

  • The prediction model looking at cancer tissue samples could predict a relapse correctly 73% of the time. It had a 75% accuracy for predicting rapid relapse (relapse with a short PSA doubling time).
  • The prediction model based on examining healthy tissue samples could predict a relapse 67% of the time. It had a 77% accuracy for predicting rapid relapse.

For blood samples, looking at amplification and deletion of specific regions did not distinguish patients who relapsed from non-relapsed patients. Instead, the researchers looked at the size of copy number variations throughout the genome, as they had observed that the size of variations in patients who relapsed was significantly larger.

This blood-based prediction model had an accuracy of 81% for predicting relapse, and a 69% accuracy for predicting rapid relapse.

The researchers then validated their models on a further 25 samples:

  • The cancer tissue analysis tool had an accuracy of 70% for predicting relapse, and 80% for rapid relapse.
  • The healthy tissue sample tool had an accuracy of 70% for relapse and rapid relapse.
  • The blood sample tool had an accuracy of 100% for relapse and 80% for rapid relapse.


How did the researchers interpret the results?

The researchers said that copy number variation analysis of blood, normal prostate tissue or tumour tissues “holds promise to become a more efficient and accurate way to predict the behaviour of prostate cancer”.



This novel study looked at genetic abnormalities in patients who had prostate cancer and whether they could be used to predict whether the cancer would relapse and, if so, how soon. It found that prostate cancer patients had high numbers of “copy number variations”, where there is an abnormal number of copies of a particular DNA region. These regions were also often large.

By comparing particular amplifications and deletions and the size of copy number variants, the researchers were able to build a prediction model to identify patients who would relapse and those who would relapse quickest (as demonstrated by a rapid increase in PSA levels). The prediction models could be used on DNA extracted from prostate cancer tissue, normal prostate tissue or blood.

This study is exciting because it raises hope that a test may one day predict the prognosis of patients with prostate cancer, which could in turn aid treatment decisions. However, this will only be possible once these findings have been further validated and a simple, inexpensive test developed and tested.

Analysis by Bazian

Analysis by Bazian

Edited by NHS Choices

Links to the headlines

Blood test for missing or damaged DNA could provide hint of a prostate cancer relapse. Daily Mail, May 7 2012

Links to the science

Yu YP, Song C, Tseng G et al. Genome abnormalities precede prostate cancer and predict clinical relapse. The American Journal of Pathology, 2012;180:2240-2248


How helpful is this page?

Average rating

Based on 2 ratings

All ratings

Add your rating