Researchers developed a test that may be able to detect ovarian cancer up to two years earlier

0
227

Researchers from Queen’s University Belfast have developed a test that may be able to detect ovarian cancer up to two years earlier than current approaches.

The researchers discovered that the presence of four proteins together, known as a biomarker panel, indicates the likelihood of Epithelial Ovarian Cancer (EOC), a type of ovarian cancer.

Using these biomarkers the researchers then developed a screening test that initial studies suggest may be able to detect ovarian cancer up to two years before current detection tests.

The research was carried out in partnership with the University of New South Wales Australia, University of Milan, University of Manchester and University College London.

The study, published in British Journal of Cancer, involved the analysis of blood samples from 80 individuals across a seven-year period.

Dr. Bobby Graham from the School of Biological Sciences at Queen’s University Belfast and lead author of the study explains:

“Firstly, we discovered that the presence of the biomarker panel will enable us to detect EOC.

We then developed a screening test to detect this biomarker panel, making this a relatively simple diagnostic test.

“The algorithm designed will screen the blood sample and flag any abnormal levels of the proteins associated with the cancer.

The screening test identifies ovarian cancer up to two years before the current tests allow.”

Most ovarian cancers are epithelial ovarian cancers, which is a cancer that forms in the tissue covering the ovary.

In females in the UK, ovarian cancer is the sixth most common cancer. In 2016, 4227 deaths were reported as a result of EOC.

If diagnosed at stage one of EOC, there is a 90 percent chance of five-year survival compared to 22 percent if diagnosed at a stage three or four.

Dr. Graham added: “The results of this study are encouraging, however, we now want to focus on testing it in a wider sample set so that we can use the data to advocate for an ovarian cancer screening programme.”

Dr. Rachel Shaw, Research Information Manager at Cancer Research UK, said: “Around half of ovarian cancer cases are picked up at a late stage, when treatment is less likely to be successful. So developing simple tests like these that could help detect the disease sooner is essential.

“At Cancer Research UK, we’re working hard to find new ways to detect cancer early and improve the tests already available. It’s really exciting to see these encouraging results for this type of ovarian cancer.”


Discussion

Ovarian cancer is a heterogeneous disease with tumour histology identifying multiple distinct types with differing prognosis, genetic markers and treatment regimens.23 

It is possible that some of these ovarian cancers are so aggressive they are essentially untreatable and their successful identification in a screening trial may not reduce mortality.

The challenge is to increase the identification of those treatable cases that do exist and to do so before they progress and become untreatable.

In this study of prospectively collected and retrospectively analysed samples, the tool presented above demonstrates an improved performance over CA125 interpreted by ROCA alone for early detection of ovarian cancer in the studied sample set.

The four protein model was able to offer up to a 1–2-year lead time in diagnosis over that achieved for the cases in the UKCTOCS trail for both Type I and Type II cases.

A direct comparison to the ROCA algorithm applied in the UKCTOCS trial and UKFOCSS study, is not possible retrospectively and the smaller sample set used here is likely to have resulted in an overestimation of the achievable PPVs.

However, the signal identified in the protein panel is certainly real and warrants further investigation in independent sample sets both in Type II and Type I EOCs.

The simulated screening programme shows the potential for relatively subtle dysregulation, against patient baseline, of multiple biomarkers in combination to return a sensitive and specific assessment of EOC risk.

Since either up- or down-regulation of each marker contributed to risk estimation it may be better to think of these makers not being directly mechanistically linked to the initiation and progression of disease, but as independent polls on dysregulation of several physiological systems.

The degree and direction of dysregulation being determined by multiple interacting feedback mechanisms.

There is a general lack of correlation between biomarker scores indicating their independence whilst the cases with low CA125 scores may still have high EOC probability scores due to contributions from other biomarkers showing how that independence contributes to sensitivity.

A caveat to the current model that should be borne in mind is that the number of controls (141 samples from 31 women) may underestimate the variability of expression of these proteins in women without ovarian cancer.

Higher variability would reduce the specificity of the test, reduce the PPV of risk estimates and erode the confidence in diagnosis. Specificity is critical at two points in the analysis. The first is where dysregulation from baseline is scored by specificity thresholds derived from the controls.

The second is in the simulated screening programme where PPV thresholds are set to classify subjects as normal, intermediate, elevated or severe. One likely reason that this study has underestimated variation is that there may be non-EOC conditions, much more frequently occurring than EOC, but sufficiently rare as to have been excluding from this control set, that affect PROZ, LCAT and CRP levels just endometriosis and the menstrual cycle do for CA125.24,25 

Analysis of the longitudinal behaviour of these proteins in a much larger cohort of control women is required to obtain more precise estimates of specificity and PPV. In mitigation of this, however, the most concerning member of the panel in this regard (CRP) returns only intermediate risk if dysregulated alone.

This indicates robustness to confounding conditions. Despite the important caveats above the investigation of the relationship between possible combinations of dysregulation score and risk estimate indicate that elevation of risk on a single biomarker, even CA125, does not return the highest risk estimates.

The strategy of developing a biomarker panel in the context of a simulated screening (Figs. 2 and 3) programme based on the UKCTOCS trial has focused analysis on the clinically critical positive predictive value, which has further enabled risk estimates from the tool to be related to potential care pathways.

The panel and associated model developed pursuing this strategy is able to detect EOC in samples drawn from women 1–2-years prior to their diagnosis in the course of the UKCTOCS trial. This time period is expected to contain a high proportion of cases at a treatable stage where patients may benefit from a 90% rather than a 22% 5-year survival rate which would be of substantial clinical utility.

This analysis from the concept of taking markers as independent polls on physiological dysregulation (indicative of disease), through to a simulated screening programme, provides not just an abstracted statistical relationship between marker and disease status; but a tangible worked example accessible to oncologists.

This study alone is not sufficient to justify deployment as a screening tool given the limitations of the analysis we have highlighted. The new panel does, however, have sufficient potential to justify a larger scale validation study and confirms the presence of EOC detecting signal from protein biomarker panels in early asymptomatic EOC.


More information: Matthew R. Russell et al. Diagnosis of epithelial ovarian cancer using a combined protein biomarker panel, British Journal of Cancer (2019). DOI: 10.1038/s41416-019-0544-0

ournal information: British Journal of Cancer
Provided by Queen’s University Belfast

LEAVE A REPLY

Please enter your comment!
Please enter your name here

Questo sito usa Akismet per ridurre lo spam. Scopri come i tuoi dati vengono elaborati.