CANCER AND TUMOUR BIOMARKERS
Biomarkers are defined by the FDA as ‘A characteristic that is objectively measured and evaluated as an indicator of normal biologic processes, pathogenic processes, or pharmacologic responses to a therapeutic intervention’. The leading candidates for biomarkers are membrane proteins, secreted proteins or extracellular matrix proteins that are more likely to be released circulation than intracellular proteins. Perhaps the most familiar tumour biomarker is prostate-specific antigen (PSA) – a protease also called kallikrein 3. The detection of elevated blood levels of PSA would prompt the recommendation of a prostate biopsy. However, seven in ten men with raised PSA levels will not have cancer and more than one in five of those with prostate cancer will have normal circulating levels of PSA.
A second example is the glycoprotein mucin 16, otherwise known as cancer antigen 125 (CA125), a trans-membrane glycoprotein thought to mediate adhesion. It is normally expressed in a variety of cell types but is up-regulated in ovarian cancer cells. Blood levels of CA125 are raised in about 50% of these cancers at an early stage and in about 90% of advanced cases. The decrease in the circulating level of CA125 has been associated with the response of ovarian cancers to treatment.
These examples illustrate the current cancer predicament over biomarkers. The PSA test has FDA approval but the high false-positive and false-negative rates indicate its limitations. Similarly, CA125 has approval for monitoring the response to treatment of ovarian cancers but is severely limited as a diagnostic assay. In pursuit of potential markers for the major cancers hundreds of compounds have been investigated and occasionally something of promise emerges. Thus, for example, the FDA has recently approved epididymal protein 4 (encoded by WFDC2 and also known as HE4) both for detecting stage 1 ovarian cancer and for monitoring treatment response. A broader approach is to identify combinations of genes (say up to five showing coherent responses) that collectively provide a more sensitive indicator than single biomarkers. Nevertheless, there are currently no serum biomarkers for cancer that approach an ideal specification and optimism that this position will change rests heavily on the development of metabolomic analyses.
An alternative approach to tumour biomarkers is through genomic analyses that, in the first instance, have been used to quantify tumour burden. Thus, for example, the levels of BCR-ABL1 mRNA in blood from CML patients can follow responses to imatinib treatment. A decrease of 1,000-fold in response to therapy is associated with prolonged progression-free survival and these findings indicate the potential for this type of approach both to monitor response and to provide an early indication of relapse.
The detection of transcripts of the BCR-ABL1 translocation is, of course, a signature of some forms of leukaemia. In addition to leukaemic cells, circulating tumour cells (CTCs) – cells that have detached from a primary tumour and entered the bloodstream – can also be detected in blood, as can free, tumour-derived nucleic acids from a range of solid cancers. These have revealed a variety of genetic alterations including point mutations (e.g. in KRAS, MYC and EGFR) and epigenetic changes. The level of tumour-derived DNA may be as little as 0.01% of the total of normal fragments in the circulation but, if mutations specific to a primary tumour are known, mutant DNA in plasma samples may be used to follow response to therapy. Because most solid tumours carry gene rearrangements, resembling those driving leukaemias, that are specific to each of a tumour, they can be identified by whole-genome sequencing. The use of the polymerase chain reaction with primers spanning the breakpoints provides an exquisitely sensitive method for the detection of these ‘personalised biomarkers’ that reflect tumour burden and hence patient response to therapy. These methods are currently unable to provide initial tumour detection screens but this limitation may be relieved as mutation signatures for at least the major sub-types of cancers are defined and become utilised for individual screening.
One of the most exciting recent developments has been the application of silicon chip technology to the detection of CTCs (Fig. 1).
1. Tumour cell isolation from whole blood by a CTC-chip.
The chips, which are the size of a microscope slide, have about 80,000 microscopic columns etched on their surface that are coated with an array of antibodies that recognise antigens expressed on the surface of epithelial cells. CTCs are present in tiny amounts in circulating blood – one in a billion normal cells – but the combination of antibody specificity and the flow properties of the cell are such that about 100 CTCs can be captured from a teaspoon (5 ml) of blood.
This remarkable technology may offer both the most promising way to early tumour detection and of determining responses to chemotherapy. It also provides a bridge between proteomic and genomic technologies because DNA, extracted from the captured cells, can be used for whole-genome sequencing. Thus, for example, specific EGFR mutations, sequencing tumour biopsies, have been detected in DNA isolated from CTC-chips in over 90% of one set of samples. If this system is able to capture cells from most major types of a tumour it provides a rapid route from early detection through genomic analysis to tailored chemotherapy without the requirement for tumour biopsies.
The upshot of the sequencing revolution is that we will soon have a complete view of the cancer scene in that essentially all mutations will be known. This will shift the focus to the problem of developing treatments: how to produce agents that can specifically target dominant oncoproteins and how to re-activate lost tumour suppressors. For the first category, another revolution is underway. It is now possible to generate huge chemical libraries of hundreds of thousands of related compounds. In addition, high throughput screening systems now mean that the time from making a compound to showing that it has potency has been reduced.
Hitherto a major problem in identifying potential anti-cancer agents has been the absence of effective methods for their identification without carrying out expensive and time-consuming tests in animals. One solution is emerging in collaboration between the Cancer Genome Project at the Wellcome Trust Sanger Institute (UK) and the Center for Molecular Therapeutics, Massachusetts General Hospital Cancer Center (USA). This has established over 1,000 genetically characterised human cancer cell lines that can be used to screen new agents for toxicity and anti-proliferative effects that can be correlated with the genetic profile of individual cell lines. The rationale behind this approach is that, although a cell line will acquire further mutations as it is grown in the lab – so it is not identical to the human tumour of origin – in general, a closely similar mutational pattern is retained. This represents a step towards tailoring therapy programmes to genetically defined sub-sets of cancers. Agents that pass this test are still a long way from becoming clinically useful but these technical developments are greatly accelerating the rate at which new drugs are becoming available. This programme has already generated substantial amounts of data that can be viewed at the Catalogue of Somatic Mutations in Cancer (COSMIC) website.
Other initiatives have also been established to integrate genomics with drug production and screening, for example, the Cancer Target Discovery and Development Network (CTD2N) of the National Cancer Institute. Progress is illustrated by the relative rapidity with which a specific inhibitor of BRAF became clinically available following the identification of the activating mutation – a period of about eight years by comparison with the decades for drugs targeting BCR-ABL1 or mutant forms of the EGFR. This, in turn, has been eclipsed by the three years taken for the ALK kinase inhibitor crizotinib to be used for the treatment of non-small cell lung cancer from the discovery of the activating translocation. It is evident therefore that these developments designed to expedite translational research are already beginning to yield results.
At this point, we should remind ourselves of a feature of cancer that we have encountered several times in the story: namely those tumours are heterogeneous. That is, not merely is the complete mutational pattern revealed by the whole genome sequence unique to each cancer, but that it represents an average from a mixed population of cells – a reflection of all the clones that comprise that tumour. The implications of this picture are immense because it means that any a tumour is made up of groups of genetically distinct cells that are moreover continually evolving, i.e. mutating. Therapeutic strategies are therefore confronting not only mixed targets but moving targets. At first sight, the concept of a moving therapeutic target comprising multiple cancer propagating cells is somewhat daunting because, intuitively, it would seem easier to have to eliminate just one set of identical cells. However, all may not be lost if the shifting mutational pattern mainly reflects different components in key signal pathways. If that were true then the treatment that neutralised the oncogenicity of a given pathway would work provided its target was downstream of an activating mutation. In other words, if drug cocktails can be produced that target the critical ‘driver’ mutations detected in a tumour they may still be sufficiently effective at controlling the disease even though major components of the mutational signature come from different cells – i.e. different clones that comprise the bulk of a tumour.