January 1, 2011 (Vol. 31, No. 1)
George L. Gabor L. Gabor Miklos, Ph.D.
As DNA Sequencing Becomes Industrialized, Translation to Patient Care Is Paramount
Rapid industrial-strength nucleic acid sequencing technologies are highlighting the challenges faced by the research, medical, legal, ethical, and direct-to-consumer genetic testing communities in personalized medicine.
For example, who are the knowledgeable human beings who can explain to cancer patients—in a sensitive way—their best therapeutic option based on sequencing data from the churning genomic rubble of their life-threatening metastatic tumors? Who can confidently explain the disease risks faced by healthy couples, or their yet-to-be-conceived children, when those risks consist of an unknown summation of genetics and the myriad epigenetic effects of individual lifestyles?
Such challenges are not to be feared; we simply need to intelligently apply ourselves to understanding the benefits and limitations of such data, while discounting media reports, such as those in Science, that the National Cancer Institute has a plan to eliminate suffering and death from cancer by 2015. We need clinical reality, not the peddling of patently false hopes that mislead desperate cancer patients and their relatives who are vulnerable to such errant nonsense.
The commercial reality is that the powerhouses of semiconductor technology, pure chemistry, and raw computing power are making human genomes, their methylated derivatives (methylomes), and their RNA outputs (transcriptomes) a successful commodity. Sequencing factories are producing data at a speed that is unparalleled in the history of biology and generating a shockwave similar to the effects on the Vatican’s pigeons of the appearance of Galileo’s cat. Whether we like it or not, every one of us will ultimately be touched by sequence data and the acid test will be its usefulness, or lack thereof, in specific clinical settings.
While only a handful of sequenced human genomes were available a few years ago, by the end of 2011 American and European laboratories will likely finish sequencing 9,000 and 6,000 human genomes, respectively. In contrast, the Beijing Genomics Institute (BGI) in Shenzhen will complete somewhere between 10,000 and 20,000 genomes on its own.
Merck has signed a statement of intent to collaborate with BGI Shenzhen to explore opportunities in the healthcare space, and Twins UK and BGI Shenzhen will collaborate to sequence the methylomes of 5,000 twins to discover therapeutic targets. The U.K. government is set to examine whether an entire healthcare system is ready for genetic testing in the cancer sphere, with the National Health Service to begin mutation testing on as many as 12,000 cancer patients by 2011.
While this testing is with a very limited number of genes, it is inevitable that, as costs fall, whole-genome sequencing will become a reality throughout entire populations, and pharmaceutical and data-analysis companies will be major participants.
A proactive clinical approach to the sequencing deluge has already started at the Beth Israel Deaconess Medical Center, where clinicians, researchers, mathematicians, and software companies are creating and integrating new tools for sequence analysis and the training of next-generation molecular pathologists. While sophisticated noninvasive imaging technologies are de rigueur, it is molecular pathology enhanced by high-performance computing that will be the gateway to clinically actionable information from sequencing data.
Whole-genome sequencing has revealed the causative single-gene products altered in Miller syndrome and a form of Charcot-Marie-Tooth neuropathy in specific families. Some healthy couples can now determine whether they, their parents, or grandparents are carriers of highly penetrant and high expressivity variants that may manifest as a particular trait in their children. These examples illustrate a clinical reality; knowing about a condition, but being unable to treat it, can have long-lasting emotional consequences. Few therapies exist for most of these thousands of Mendelian conditions.
A more relevant clinical challenge concerns predictions on the future of healthy individuals in whom vascular disease, obesity, and dementia will develop in their lifetimes, where hundreds of gene products—under the epigenetic influences of nurture, diet, drugs, and stress—sculpt unique outcomes over many decades.
Precisely predicting any future outcome must inevitably involve analysis of methylomes and transcriptomes from clinically relevant samples (e.g., from normal coronary arteries, reproductive tissue, or deep brain regions), but these are unavailable from healthy individuals owing to profound unresolved ethical, legal, and clinical issues involving the healthy patient, the doctor, and the service provider.
Predictions on quantitative traits underpinned by degenerate networks are the most significant challenge to “bullet-train biotechnology”, and the jury is out on the extent to which there will be significant clinical impact in therapeutic contexts.
While most human diseases involve cells with diploid genomes, cancer is different. Solid tumors constitute 90% of all cancers, and they are exceedingly heterogeneous at the cellular level. This heterogeneity is apparent from DNA sequencing of solid tumors and their metastases, every one of which is unique in terms of its combination of variants and structural alterations. It is sobering that the billion-dollar blockbuster drug Herceptin, which is the poster child of breast cancer treatment, is not targeted to a mutation but to amplification of the normal Her2 gene product via segmental aneuploidy.
Personalized oncology is really personalized group treatment. In this group context, some variants in the KRAS, BRAF, and EGFR genes appear promising as patient stratifiers, but only Phase III clinical trials will determine their true clinical value.
In the area of mental illness, three Nobel laureates (James Watson, Eric Kandel, and Sydney Brenner) and colleagues surprisingly advocate the sequencing of 100,000 genomes from individuals with schizophrenia, autism, bipolar disorder, and depression as well as the use of mouse models to devise new treatment strategies for mentally ill patients. They further propose to “reset” schizophrenic and autistic brains on the basis of rodent data.
The ethical implications of attempting to reset an anatomically miswired human brain in childhood or adult life is clinically and legally hazardous. Each human brain is anatomically and functionally unique, as evidenced by the enormous variation in symptoms among schizophrenic individuals including different hallucinations, delusions, and dysfunctions in the formulation of thoughts and their expression in language.
Postmortem data from schizophrenic and autistic individuals reveals neuronal loss in different brain regions in different individuals, such as the subnuclei of the medial dorsal nucleus, cingulate cortex, hippocampal formation, entorhinal cortex, and parahippocampal gyrus. In addition, even if experimental animals experience hallucinations and delusions, they cannot report on such conditions.
Thus, introducing hundreds of mutated genes individually into mice where different strains have greatly divergent expression of well-known pharmaceutical drug targets in their brains and determining which of those perturbations are causal, which are innocent bystanders, and which are relevant to perturbations in the human brain, is the “ultimate endless staircase”—particularly since existing human data on these conditions reveal that it is large copy-number variants rather than single mutations that play a predominant role. It seems that when your only tool is a hammer, every problem looks like a nail.
In summary, multilevel data analysis from identical twins, discordant for quantitative traits, provides an excellent platform for testing clinical utility. Second, predictions on the outputs of variants in gene products that are involved in drug metabolism must be rigorously tested in clinical contexts. Third, only Phase III oncology trials, based on median overall survival, will reveal whether the stratification of cancer patients, based on particular variants in their tumors, will be of therapeutic utility.
Finally, healthy couples will certainly benefit from their personal sequencing information prior to starting a family. However it will be by diagnosing and filtering potentially devastating conditions, via in vitro fertilization as well as preimplantation transcriptome and genome sequencing and associated technologies, that bullet-train biotechnology will contribute significantly to preventive personalized medicine.