8.16 - Genome sequencing and personalised medicine

From sequences to medical knowledge

Sequencing provides information about DNA that must be interpreted before it can guide care.

A genome is an organism's complete genetic material. DNA sequencing determines the order of bases in DNA; whole genome sequencing examines DNA across the genome, including coding and non-coding regions. This differs from a DNA profile used to compare selected variable regions for identification. A profile is not a complete sequence or a report of all disease risks.

Large sequencing projects produce reference sequences, catalogues of variants (differences in DNA sequence) and linked health data. A reference is a comparison tool, not a genetically perfect person. Researchers compare sequences and health outcomes to identify variants associated with a disease or response to a medicine. Sequencing relatives can help establish whether a variant follows the inheritance of a condition. Studying a variant's effect on a protein or cell process helps test a possible mechanism.

The Human Genome Project provided a reference sequence on which later comparisons could build. The 100,000 Genomes Project linked sequences with information from people affected by rare conditions or cancer. Comparing tumour DNA with a patient's other cells helps distinguish acquired tumour changes from inherited variation, supporting diagnosis and selection of possible treatment targets. These projects supply reusable evidence and methods, rather than a universal drug. NHGRI: Human Genome Project, Genomics England: cancer outcomes.

This creates a chain from sequence to useful knowledge: identify a variant, investigate its biological effect, and test whether acting on that information improves outcomes. An association alone does not establish that the variant causes the effect: it may be inherited alongside the causal variant, and other factors may differ between groups. Functional studies and clinical trials strengthen the evidence.

Once a useful relationship is established, a patient's result can be compared with the evidence. A targeted test for a known variant or a panel of relevant genes may be sufficient; every decision does not require sequencing an entire genome. Most variants are harmless, some are harmful, and others remain difficult to interpret. Reading a sequence and understanding every base are different achievements. NHGRI: sequencing, human variation.

Matching treatment to a variant

The useful question is often how a variant changes the protein on which a medicine depends.

Pharmacogenomics studies how genomic variation affects responses to medicines. Variants may alter a drug's target protein or enzymes involved in activating or breaking down a drug. A medicine can therefore be ineffective or cause adverse effects in some people. Genetic information helps select among options, while age, other medicines, organ function and environment still matter. NHGRI: pharmacogenomics.

CFTR provides a concrete example of matching a treatment to a molecular defect. CFTR is a chloride channel in epithelial cell membranes. In airways, reduced CFTR function contributes to poorly hydrated mucus and impaired clearance. G551D is a CFTR variant whose protein reaches the cell surface but has a severe channel-opening, or gating, defect. Ivacaftor is a potentiator: it increases the opening probability of responsive CFTR channels already at the surface, increasing chloride transport. It does not replace the DNA sequence. FDA: ivacaftor mechanism, section 12.1.

The following patients are fictional. Their genotypes illustrate real biological differences, rather than supply prescribing instructions.

The same disease name hides different molecular problems. Some other CFTR medicines help folding and delivery to the surface and are used in combinations, illustrating why failure of ivacaftor alone does not mean no genotype-directed treatment exists. Primary CFTR experiments, folding-correction research.

In a real randomised, placebo-controlled trial, people with CF and at least one G551D allele showed improved lung function on average with ivacaftor. This supports a clinical benefit beyond a cell-based prediction. It does not promise identical improvement for every person or a cure: genotype-guided selection still requires evidence about benefits, adverse effects and the individual. Ramsey and colleagues, 2011.

Prevention without certainty

A genomic result can guide care before a disease appears, as well as after diagnosis.

For example, some inherited harmful variants in BRCA1 or BRCA2 increase susceptibility to certain cancers. These genes contribute to DNA repair; impaired repair can allow further DNA changes to accumulate. Detecting a well-established harmful variant may support earlier or more intensive surveillance and discussion of risk-reducing options with a specialist. Surveillance aims to detect disease earlier; it does not itself prevent a cancer from starting. Risk-reducing interventions may lower risk but also have burdens or harms. A predisposition is not a diagnosis or a prediction that cancer must occur. NCI: BRCA variants and risk.

Two kinds of uncertainty must be separated. First, the effect of a variant may be unknown. A variant of uncertain significance is a sequence difference for which there is insufficient evidence to classify its health effect. It is not a confirmed harmful variant. Second, even a well-established risk variant may not determine the outcome: other genes, environmental exposures and chance can influence whether disease develops. A negative result also cannot exclude every genetic cause or every non-genetic route to disease.

Sequencing tumour cells serves a different purpose. An acquired variant in a tumour may identify a target for treatment; it is not automatically inherited or present in the person's other cells. Two tumours in the same organ may have different targets. Conversely, finding a target does not ensure a lasting response, because not all cancer cells may share it and resistant cells may persist or evolve. NCI: biomarker testing.