Whole-Genome Sequencing: Title and subTitle BreakA Step Closer to Personalized Medicine

The past 60 years have witnessed remarkable progress in genetics and genomics from the description of the DNA double helix by Watson and Crick¹ to the release of the first draft sequence of the human genome in 2001^{2 - 3} and the successful completion of the human genome project in 2003.⁴ From that time, there has been increasing hope and expectation that, as soon as the cost of sequencing the whole genome could become affordable, the promise of personalized medicine would be fulfilled.

No field of medicine has benefited more from advances in genomics and the application of genetic testing than oncology. These advances have had a substantial influence on cancer risk assessment, determination of prognosis, and choice of treatment. Clinical applications of novel genetic tools include sequencing and analysis of germline genomic rearrangements at key cancer genes like BRCA1, BRCA2, and TP53⁵ ; mismatch repair genes such as MLH1, MSH2, MSH6, and PMS2⁶ ; development and widespread use of clinical karyotyping for hematologic malignancies⁷ ; analysis of ERBB2 overexpression in breast cancer⁸ ; KRAS gene mutations in colorectal cancer⁹ —and even gene expression analysis in breast cancer as a form of molecular pathology.¹⁰

In many cases, these genetic data have not only prognostic value but also important therapeutic implications for the patient.^{8 - 10} With the advent of more comprehensive, genome-wide analyses of large tumor sets by projects like the Cancer Genome Atlas,¹¹ the number of clinically relevant molecular tests is expected to continue to increase. Yet even these tests of commonly mutated cancer genes have limits and will always lack sensitivity to atypical mutations or gene expression patterns. The complexities of any individual's cancer genome, and the treatment implications therein, are unlikely to be defined by single genes. For this reason, the latest advances in genomic technology have excited so many in the field.

Over the past few years, high-throughput, short-read DNA sequencers have revolutionized the field of genomics and have accelerated the pace of discovery in cancer research.¹² The new technologies simultaneously read millions of short, 50- to 200-nucleotide DNA sequences from a pool of randomized genomic fragments in a single experiment. The process, which would have taken months with older technologies, is finished in a few days. Prior to the completion of the Human Genome Project, these short reads would have been difficult to interpret, as their genomic origin would have been unknown. However, the human reference sequence is now used to computationally map these sequences to the genome and to identify polymorphisms and novel mutations in a patient's DNA.

The ability to sequence an individual's entire genome as well as the patient's tumor genome is now a feasible enterprise at a cost and speed that was unthinkable even 5 years ago. In less than 3 years, DNA sequencing costs have decreased by more than 100-fold.¹³ The rate of improvement exceeds the advancement in computational power over the same time period, which predicts an oncoming wave of genomic data. Today, sequencing a tumor genome is still expensive and requires an infrastructure that is incompatible with a clinical setting, but the trend suggests that we are a lot closer to cost-effective, clinical genomics than most physicians realize.

Two articles^{14 - 15} in this issue of JAMA are remarkable examples of the power that these genomic data hold for patients with a diagnosis of cancer. In one report, Link and colleagues,¹⁴ from Washington University, St Jude Children's Research Hospital, and University of Chicago, performed whole-genome sequencing on skin and leukemic cells from a woman with suspected cancer susceptibility syndrome based on the early onset of several primary tumors. The patient had breast cancer at age 37 years, had ovarian cancer at age 39 years with recurrence at age 42 years, and developed treatment-related acute myeloid leukemia (t-AML) 6 months later. Genetic testing for mutations of the BRCA1 and BRCA2 genes was unrevealing, and in keeping with guidelines for assessment of high risk for familial cancer, no additional targeted genetic testing was obtained. However, whole-genome DNA sequencing showed that the patient was heterozygous for a novel deletion of 3 exons of the TP53 gene and that the intact copy of TP53 had been lost in the leukemic cells due to uniparental disomy. This mutation in TP53 would not have been discovered without whole-genome sequencing. Although this discovery did not help save the patient's life, the implications for her children who may have inherited this mutation are immediate. Indeed, current guidelines from the National Comprehensive Cancer Network⁵ recommend annual screening mammograms, breast magnetic resonance imaging, or both starting at the age of 20 to 25 years. Furthermore, risk-reducing mastectomy is an effective option to prevent the development of breast cancer in this high-risk group of individuals. The high frequency of asymptomatic cancers in healthy individuals with Li-Fraumeni syndrome has been highlighted in a recent study assessing the usefulness of positron emission tomography scans in patients.¹⁶

The second article, by Welch and colleagues¹⁵ at Washington University, focuses on a 39-year-old woman with a diagnosis of AML of unclear subtype. Although the patient's clinical presentation was consistent with acute promyelocytic leukemia, a subtype of AML with a favorable prognosis, cytogenetic analysis revealed a different subtype associated with a poor prognosis and for which bone marrow transplantation in first remission is recommended.¹⁷ To resolve this therapeutic conundrum, Welch et al¹⁵ performed whole-genome sequencing of the DNA extracted from the original leukemic bone marrow and from a skin biopsy. The results identified a novel insertional translocation on chromosome 17 that produced a pathogenic PML-RARA gene fusion. This type of genetic event could not have been identified with classic cytogenetic techniques. The patient's genome was sequenced and analyzed in 6 weeks, and the results led to a change in therapy. First, the patient was eligible to receive treatment with retinoic acid, which significantly improves the overall prognosis of patients with acute promyelocytic leukemia.¹⁸ Second, the patient was no longer considered for bone marrow transplantation in first remission. The ability to generate comprehensive genomic data in a time frame that is clinically relevant for a patient is a remarkable achievement.

These cases of personalized genomic medicine are just some of the first examples of what will likely be commonplace in the near future. New innovations in DNA sequencing are expected to increase the speed of data collection and decrease the costs by another order of magnitude in the next 2 years. Clearly, the technology will no longer be the major impediment to widespread clinical use of these tools, and the main challenges will soon move to the clinical implementation and interpretation of genomic data. Numerous regulatory and reimbursement issues remain. The authors of the current studies have presented one framework for how this can be accomplished, with a “moving firewall” between the genome sequencing center and the physicians who deliver the genomic data to the patient, but other models also may need to be developed.

Meanwhile, it is clear that the first test bed for implementation of personalized genomic medicine is being established in oncology, where its use can have an immediate effect on patient care. There is also a great potential for the use of similar approaches for unbiased discovery of genetic features associated with sensitivity and resistance to a given therapy. Hence, recent advances in genomics are likely to change the molecular characterization of cancer rapidly and provide a path for the personalized treatment of patients with cancer.