Predictive Biomarkers

Chapter 1 - Definition: Translational and Personalised Medicine, Biomarkers, Pharmacodynamics

A predictive biomarker is usually measured before treatment and provides information on the probability of response to a particular therapy. As depicted graphically on the right, the probability of survival depends on treatment in those patients who express the biomarker associated with response to a particular therapy. For patients who are biomarker negative, there is no difference in survival between treated and untreated patients. An example of a predictive biomarker in breast cancer is expression of the HER2/neu protein. Approximately 15% to 20% of patients with invasive breast cancer have increased expression of the HER2/neu protein (a member of the EGFR family of transmembrane receptors) that is associated with response to anti-HER2-targeted agents such as trastuzumab, pertuzumab, and trastuzumab-emtansine. Another example is expression of the oestrogen receptor in breast cancer, which correlates with sensitivity to hormonal agents used in the adjuvant and metastatic settings.

In metastatic non-small cell lung cancer, patients with tumours harbouring either EGFR exon 19 or 21 gene mutations or the EML4-ALK fusion protein can be offered oral targeted therapies against EGFR (gefitinib or erlotinib) or ALK (crizotinib), respectively, which are more effective than cytotoxic chemotherapy. These somatic genetic alterations are only identified in a subset of metastatic non-small cell lung cancer patients (EGFR mutations in 15%-20% and EML4-ALK translocation in 3%-5%). The development of robust, clinically applicable biomarkers identifying these alterations has enabled the selection of only those patients who would derive benefit from targeted therapy.

In colorectal cancer, KRAS is a frequently mutated oncogene and is a predictive biomarker for resistance to anti-EGFR monoclonal antibody therapy. Approximately 40% of colorectal cancers have mutations in KRAS. This is routinely tested in the metastatic setting, as only tumours with wild-type KRAS derive benefit from EGFR inhibitors such as cetuximab and panitumumab. More recently, it has been shown that additional mutations in other RAS genes are observed in 5%-10% of colorectal cancers and are also associated with resistance to anti-EGFR therapy. KRAS mutations were initially identified retrospectively after large clinical trials were completed in unselected patients. This example highlights the importance of tumour tissue collections in larger clinical trials in order to facilitate the identification of clinically useful biomarkers in the future.