Doctors
General Public
Medical Students
Main Page

New technique to detect ovarian cancer

About Dr Ambreen Farrukh

An article in The Times reports on a promising new technique for detecting ovarian cancer, involving the identification of a 'protein barcode' in the blood. The article accurately reflects the findings of a study published in the Lancet but evaluation of the technique in the wider population is needed.

The Times  reports that by using powerful computers, researchers can detect the signature 'barcode' of ovarian cancer. The article also reports that the new technique could be used to spot the disease in its earliest and most treatable stages, making it more sensitive than any other ovarian cancer test in the world.

The study in the Lancet initially compared blood samples from 50 women known to have ovarian cancer to 50 unaffected women in order to identify the 'protein barcode'. Subsequently, masked blood samples from 50 women known to have ovarian cancer and 66 unaffected women were analysed and compared to the barcode. The test was able to identify all of the cases of ovarian cancer, but three samples in unaffected women were incorrectly classified as cases of ovarian cancer.

In general, the article accurately reflects the research findings. However, as the test was only compared with one ovarian cancer test (CA125) and only 18 of the samples came from women with early stage ovarian cancer, the claims that the technique could be used to spot the disease in its earliest stages and that it is more sensitive than any other ovarian cancer test are overstated.
The technique appears promising but evaluation in the wider population is required before its true accuracy can be determined. Evaluation of the evidence base for the use of proteomic patterns in serum to identify ovarian cancer

What were the authors' objectives?
To develop a bioinformatics tool and use it to identify proteomic patterns in serum that distinguish neoplastic from non-neoplastic disease within the ovary.
What was the nature of the evidence? The research was conducted in two phases. In phase I, a preliminary set of blood samples from women at high-risk of developing ovarian cancer (50 women with biopsy-proven cancer and 50 unaffected women) were used to identify the proteomic pattern (a discriminating pattern formed by a small key subset of proteins thought to reflect the underlying pathological state of an organ) for ovarian cancer. In phase II, a masked set of blood samples from 99 women at high-risk of developing ovarian cancer (50 women with biopsy-proven cancer and 49 unaffected women), and 17 unaffected women from the general population were analysed.
Women in the high-risk group were self-referred under at least one of a number of eligibility criteria including a genetic predisposition to cancer, or a family or personal history of cancer. All women received a yearly ultra-sound and measurement of the concentration of the mostly widely used biomarker for ovarian cancer, cancer antigen 125 (CA125).

What were the factors of interest?
In phase I, the optimum proteomic pattern was found to be defined by a cluster of five proteins. Using a genetic algorithm (used to generate a best pattern and classify diagnostically unknown samples) the masked samples were analysed and matched to the pattern identified in phase I. Each unknown sample was classified into three possible categories: cancer, unaffected, or new cluster.

What were the findings?
Analysis of the masked blood samples, correctly classified 63 out of 66 (95%) of the controls as not cancer, including correct classification of all 17 non-cancer disease controls taken from the general population. Twenty-two out of 24 (92%) of the true 'normals' were correctly classified, and all 50 cancer samples were correctly classified as malignant.
The results yielded 100% sensitivity (proportion of disease positives who are test positive) and 95% specificity (proportion of disease negatives who are test negatives). The positive-predictive value (the probability that a patient who is test positive actually has the disease) of the test was 94%, compared to 35% for the CA125 test.

What were the authors' conclusions?
These findings justify a prospective population-based assessment of proteomic pattern technology as a screening tool for all stages of ovarian cancer in high-risk and general populations.

How reliable are the conclusions?
The sample size was sufficiently large, and it is very unlikely that the results of the test arose by chance alone. However, this is a pilot study of a technology in its infancy. The authors themselves acknowledge that the origin and full identity of the discriminating proteins are currently under investigation. The true accuracy of the technique will not be known until it has been evaluated in one or more independent studies and at this stage no claims can be made about its usefulness in routine clinical practice.
The authors' conclusions are fair and further investigation and validation of the technique in a prospective trial seems warranted.