Companion diagnostics: the key to personalized treatment of breast and ovarian cancer

April 20, 2015

The development of powerful new companion diagnostics is creating a new model of personalized medicine. Companion diagnostics is accelerating scientific discoveries that will provide physicians with new tools and clinically objective information to help select which medicines and treatments will work best for their patients.

Genomics and companion diagnostics

The mid-1990s saw the development of genomics as a concept. The excitement generated by the promise of this new field led to significant public and private investment in DNA sequencing and mRNA characterization. The following decade saw the realization of aspects of this potential with the completion of the Human Genome Project, but it was accompanied by significant disappointment—as the anticipated avalanche of novel drug targets failed to materialize. There were no seismic shifts in the efficiency of drug development. Despite a similar enthusiastic detour into proteomics, in retrospect, it can be argued that the real beneficiary of these efforts was the emergent research-based molecular diagnostics industry.

The discovery of the BRCA1 and BRCA2 hereditary breast and ovarian cancer genes was an important event in the development of this new field, and it helped to create a genetic testing environment that supported a significant healthcare advancement for individuals and families challenged by this specific genetic heritage. January 2015 saw a noteworthy milestone with simultaneous U.S. Food and Drug Administration (FDA) approval of Lynparza (olaparib), the first-in-class PARP inhibitor, and a germline BRCA test for use as a companion diagnostic for maintenance therapy of high grade serous ovarian cancer. This marked the FDA’s first approval of a laboratory developed test (LDT) through the premarket approval process and the first approval of an LDT companion diagnostic. Behind this radical advance lies an intimate dissection of the molecular mechanisms that underlie DNA repair in healthy normal and tumor cells.

Mechanisms of DNA repair

Multiple biological mechanisms underlie nature’s efforts to preserve the integrity of DNA sequence and thus avoid the generation and propagation of germline and somatic mutations. We now understand cancers to be essentially genetic diseases. Breast and ovarian cancers, in particular, commonly have BRCA1, BRCA2, and an emerging set of other inherited gene defects as a base for tumor development. Many of these genes play a role in the repair of double strand (DS) breaks, and when compromised, the cell employs a less faithful DNA repair pathway involving poly-ADP ribose polymerases (PARPs). In a seminal study, it was shown that tumor cell lines that carried germline mutations in BRCA2 were especially sensitive to PARP inhibitor compounds, laying the groundwork for many more years of preclinical and clinical development for pharmaceutical companies and the academic sector.

Germline BRCA mutations are found in about 14 percent of high grade serous ovarian cancer patients and approximately 18 percent of patients with triple negative breast cancer (TNBC). To attempt to expand the patient groups liable to be treated with mechanism-based therapies, two partially overlapping scientific approaches were pursued to create novel companion diagnostic tests.

The first was to expand our understanding of additional somatic mechanisms that could underlie tumor-level BRCA1 or BRCA2 inactivation and add to the germline burden. In both ovarian and TNBC, somatic BRCA mutations and methylation of the BRCA1 promoter are relatively common and add to the potential PARP inhibitor responder epidemiology. This enhancement of the potential patient responder group has been recognized in Europe with CE Marking of either germline or tumor BRCA sequence status as the companion diagnostic for Lynparza. As mentioned above, additional dysfunctional non-BRCA DS repair genes may also encode response to PARP inhibitors and other agents disrupting the DS DNA repair process, but their lower frequencies in patients mean that there is a lack of statistically robust data on relationship with responder status.

Recognizing this difficulty in characterizing DS repair functional status by examination of multiple genes with potentially variable effect sizes, integrative biomarkers were sought that would potentially encompass multiple genes and multiple dysfunctional mechanisms. It was logical to imagine that since BRCA1 and BRCA2 were prototypical DS DNA repair genes, it should be possible to discern global patterns of changes in the DNA of tumors that would not only include the observed BRCA deficiencies but would extend beyond known territory to encompass functionally similar causative defects. An alternative successful approach was to search for a DNA-based signature that reflected platinum response, since limited data suggested enhanced platinum responses in BRCA mutant tumors. All these approaches employed array-based interrogation of single nucleotide polymorphisms (SNPs) across the tumor genome, using sophisticated bioinformatics analyses coupled with discovery and validation study designs.

Measuring the DNA “scar”

Three measures were shown to robustly measure the DNA “scar” resulting from double-stranded DNA repair deficiency. These were originally termed loss of heterozygosity (LOH) score, telomeric allelic imbalance (TAI) score, and large scale state transition (LST) score. The phenomenon that each of these scores attempts to measure is now termed homologous recombination deficiency (HRD), being the biological state generated by loss of two functional copies of a homologous recombination gene such as BRCA1 or BRCA2. Comparison of the individual scores within ovarian and TNBC tumor sets showed a high degree of correlation, suggesting the individual scores measure a central facet of DNA instability with each score measuring some unique variance. The unweighted sum of the individual scores correlates most highly with BRCA deficiency, and this has led to the idea that this should be the most robust biomarker measurement for clinical use. The combined measurement of LOH, TAI, and LST is now termed the “HRD score.”

While initial clinical studies focused on the correlation of HRD score with BRCA (and RAD51C) deficiency and survival metrics in HGSOC and TNBC, multiple studies have now been completed showing that HRD score predicts the preoperative (neoadjuvant) response to platinum-containing chemotherapy regimens in TNBC. Importantly, approximately 50 percent of HGSOC and TNBC patients’ tumors carry either a deleterious or suspected deleterious mutation in BRCA1 or BRCA2 and/or an HRD score of equal to or greater than 42, which is the predetermined and validated threshold that corresponds to a functional inability to repair double-stranded DNA breaks.

The clinical laboratory assay has been hardened so that robust results are obtained on formalin fixed paraffin embedded (FFPE) pathology tissue with BRCA1 and BRCA2 exon sequences and genome wide quantitative SNP data generated by next generation sequencing. The resulting assay is high-throughput and suitable for future clinical use. Studies of the use of tissue-based BRCA genotyping and HRD score estimation for prediction of platinum or PARP inhibitor responses are ongoing and may lead to profound personalized medicine treatment opportunities in the future. By exploiting a basic understanding of the epidemiology of tumor DNA repair, scientists have been able to impose biomarker-guided precision on what have up to now been considered generalized chemotherapies such as platinum and potentially anthrocycline drugs.


Companion diagnostics for oncology medicines is poised to revolutionize the precision delivery of powerful drugs to patients in greatest need. A number of remarkable drugs accompanied by appropriate biological marker tests are already well established in the market, and the number of these combinations is growing rapidly. Biological complexity is both a blessing and a curse as it relates to disease mechanism discovery, but careful attention to the clues left behind as a tumor develops can ultimately be exploited for a patient’s benefit.