Quantcast

Industry news that matters to you.  Learn more

Discovery of Diagnostic and Prognostic Prostate Cancer Serum Biomarkers Guided by Cancer Genetics

Reading time: 2 – 4 minutes

An interdisciplinary team of researchers from ETH Zurich, University Hospital Zurich and the Cantonal Hospital of St. Gallenhas has defined biomarkers in patients’ blood serum that indicates the presence of prostate cancer. The method used has the potential to be applied to other types of tumors.

The study, published in the journal PNAS, describes a two-stage strategy to define novel signatures in patient serum that can report the presence or absence of cancer, in particular, prostate cancer. Current diagnostic methods that detect tumor antigens in the blood are not always reliable and often deliver false positives, which is not only expensive but can lead to unnecessary biopsies.

Stage one: candidate biomarker identification and validation was performed in prostatic Pten deficient PbCre4-Pten fl/fl (Pten cKO) mice. Researchers selectively evaluated N-glycosylated proteins to maximize detectability in serum as well as to focus on a “subproteome” that is enriched for validated serum biomarkers (30 of the 38 protein biomarkers currently being used in the clinic are glycosylated). Candidate biomarkers were analyzed based on three criteria: Pten dependency, prostate specificity and detectability in serum.

Stage two: researchers tested whether PTEN inactivation in human prostate cancer is associated with a specifc serum signature by FISH and immunohistochemistry analysis of human tissue samples followed by targeted quantitative mass spectrometry of serum samples from patients with BPH and localized prostate cancer to identify 49 candidate protein biomarkers.

The researchers then used the random forest (RF) classifier algorithm to build predictive models for the discrimination between normal and aberrant PTEN status. They identified a four protein serum signature that could correctly predict 78% of cases belonging to patients having aberrant or normal PTEN status with a sensitivity of 79.2% and specificity of 76.7%: thrombospondin-1 (THBS1), metalloproteinase inhibitor 1 (TIMP-1), complement factor H (CFH), and prolow-density lipoprotein receptor-related protein 1 (LRP-1).

A four protein serum signature for prostate cancer diagnosis was also determined by machine learning analysis applied to a quantitative data set derived from selected reaction monitoring (a technique for targeted quantitative proteomics by mass spectrometry): hypoxia up-regulated protein 1 (HYOU1), asporin (ASPN), cathepsin D (CTSD), and olfactomedin-4 (OLFM4). This signature discriminated between BPH and prostate cancer with an area under the receiver operating characteristic (ROC) curve almost equal to that of PSA alone. Together with PSA, the five proteins increased the area under the curve by 15.7%. The resulting sensitivity was 85% with a specificity of 79%.

According to the authors, the method can be applied to other cancer types:

This study provides a general framework for rational cancer biomarker discovery. The underlying concept is that activation of cancer-signaling pathways caused, for example, by the inactivation of a defined tumor-suppressor gene is associated with specific protein signatures that can be measured in serum and potentially used to detect disease at an early stage or to derive information about the tumor grade and thus guide treatment decisions.

Study: Cancer genetics-guided discovery of serum biomarker signatures for diagnosis and prognosis of prostate cancer

Source: ETH Zurich