PERSONALIZED MEDICINE VS GOOGLE

It is of public domain that Larry Page, the ambitious CEO of Google, has engaged in the exciting challenge of solving the biggest problems of medicine. The guys at Google are in fact betting on nanotechnology applications to win everlasting enemies like cancer. Page started this new adventure by opening a top-secret lab on Google's campus and by buying companies like Calico.

THREE IMPORTANT GENES IN LUNG CANCER

Molecular studies of lung cancer cells have been useful in the setting of advanced stage disease, whereas their usefulness in resectable disease is still under investigation. So far, molecular testing has been proved to be useful only for NSCLC (Non Small Cell Lung Cancer). Molecular testing can be done on samples from biopsies or on cytology specimens.

Amongst others, three genes have been frequently reèprted for showing mutations of diagnostic and predictive significance in lung cancer: EGFR, ALK and BRAF.

NLG: NATURAL LANGUAGE GENERATION

Natural Language Generation (NLG) is an informatic process based on algorithms that try to transform text data into sentences which can be understandable by a human. The ultimate application of NLG is scientific curation. A computer can be taught to read a scientific article, extract important data from it (e.g. evidences on genotype-phenotype correlations) and re-elabororate them in condensed statements that can be fast and easily understood by a scientist without him reading the original article. In theory. 

For instance, a computer could 'read' an article on EGFR gene mutations in lung cancer, extract the association between the L858R mutation and the sensitivity to the treatment with gefitinib and finally output a statement like 'the somatic mutation L858R in lung cancer cells increases the sensitivity to the treatment with gefitinib'. In theory.

WORKFLOW IN IDENTIFYING DRIVER MUTATIONS IN CANCER

As said in the article STRATEGIES IN IDENTIFYING DRIVER MUTATIONS IN CANCER, Whole-Exome-Sequencing (WES) is probably the most widespread technique used to identify driver mutations in cancer. Here below some steps to be undertaken in identifying driver mutations:

1. SIMULTANEOUS SEQUENCING OF TUMOR AND A MATCHED NORMAL SAMPLE 

Identifying driver mutations is first of all a matter of identifying and differentiating cancer somatic mutations from patient's germline mutations. Therefore, in the effort of identifying driver mutations it is always necessary to do simultaneous sequencing of both the tumor sample and a matched control of normal tissue from the same patient.

2. ASSESSMENT OF TUMOR PURITY

Each tumor sample is always an admixture of normal tissue sample and different cancer cell lines. Tumor purity is defined by the fraction of cancerous cells in the sample. Therefore to identify putative driver mutations it is necessary to proceed to an estimate of tumor purity. Such an estimate can be done by special algorithms which can be included in a variant calling software. For instance, with VarScan2 the user can enter an estimate manually, whereas software like MuTect and Strelka calculate this automatically based on allele frequencies observed. There are also software which are solely dedicated to the estimation of tumor purity: ABSOLUTE and ASCAT can infer tumor purity and tumor ploidy from SNP array data.

3. ASSESSMENT OF INTRATUMOR HETEROGENEITY

Although cancer development is a clonal process, not all cancer cells are identical. In any cancer it is actually possible to identify different subtypes of cancer cells. Also in this case bioinformatics comes to help. The following methods can be used to infer intratumor heterogeneity: (1) with a software (e.g. THetA: Tumor Heterogeneity Algorithm), (2) by next generation sequencing at ultra high coverage to identify different SNVs (i.e. differentiating clonal mutations, arisen at the beginning, from subclonal mutations, arisen later in tumor development) (3) by sequencing different tumor regions after multiple biopsies.

4. EVALUATION OF ALLELE FREQUENCY

As said in STRATEGIES IN IDENTIFYING DRIVER MUTATIONS IN CANCER, assessing the allele frequency is probably the most accepted strategy in identifying driver mutations, following the concept that the higher the allele frequency of a mutation, the higher the possibility it is a driver mutation.

To do so, it is important to determine if the putative driver mutation has an allele frequency which exceeds the background mutation rate (BMR), i.e. the probability that a given locus a passenger mutation arises (a passenger mutation is a mutation with no significance for cancer development). BMR depends on genomic context, mutation type, gene transcription rate and gene replication timing.

5. VERIFICATION BY FUNCTIONAL STUDIES

Allele frequency and in silico analysis may help the initial phase, but the final evidence that a mutation is a driver mutation must be obtained by functional assays in vitro, which may demonstrate that the mutation has the power to influence cell division, cell dedifferentiation, ecc.

Related subjects:

• DRIVER MUTATIONS AND PASSENGER MUTATIONS IN CANCER
• STRATEGIES IN IDENTIFYING DRIVER MUTATIONS IN CANCER
• PRECISION ONCOLOGY

STRATEGIES IN IDENTIFYING DRIVER MUTATIONS IN CANCER

Driver mutations are somatically acquired changes in the DNA of cancer cells which are driving the development of cancer. Identifying driver mutations is not an easy task, but it is necessary to better understand cancer biology and to design new potential treatments.

There are some strategies which have been used so far to identify driver mutations. The first one is based on allele frequency, i.e.: the higher the mutation frequency across different samples of the same tumor, the higher the probability that that is a driver mutation. In fact, if a mutation confers growth advantage to cancer cells, this will be perpetuated in the clonal proliferation and therefore will be present in all cells of the cancer. However, it should be mentioned that some driver mutations can be rare mutations (precision oncology is just based just on this: identifying rare driver mutations alongside with more common driver mutations).

Another strategy is to look first at known cancer genes, where driver mutations are more likely to arise, whereas passenger mutations (i.e. mutations with no particular effect on cancer development) are usually randomly distributed in the rest of the genome.

Actually, any putative driver mutation should be finally proved to be a real driver mutation by functional assays in vitro, which can demonstrate the power of that mutation on cell growth, cell dedifferentiation, etc. 

Whole-Genome-Sequencing (WGS) and Whole-Exome-Sequencing (WES) are the main techniques used in identifying driver mutations. WES is probably the most widespread because of its reduced costs and reduced amount of data to be elaborated in comparison to WGS. 

The detection of somatic mutation in cancer is also the aim of some collaborative projects like The Cancer Genome Atlas (TCGA) and the International Cancer Genome Consortium (ICGC).

Related subjects:

• DRIVER MUTATIONS AND PASSENGER MUTATIONS IN CANCER
• WORKFLOW IN IDENTIFYING DRIVER MUTATIONS IN CANCER
• PRECISION ONCOLOGY

PERSONALIZED MEDICINE

Personalized medicine is a term used to identify a personalized approach in diagnosis and treatment based on the individual genetic profile of a patient. In oncology, personalized medicine is essentially based on the analysis of the genetic profile of a tumor.

It is actually proved that some somatic mutations in cancer can confer susceptibility or resistance to certain drugs and that as consequence the patient's treatment can be tailored in order to obtain a better response (see also PHARMACOGENOMICS). Not only: some of these somatic cancer mutations are of help in narrowing the histological diagnosis as they can be typical of certain cellular types and/or cancer growth patterns (histomolecular correlations). As a collateral effect of their therapeutic and diagnostic relevance, some somatic mutations may also show a prognostic significance in terms of association with (1) progression-free survival (PFS) and/or (2) longer/shorter overall survival (OS). Some EGFR gene mutations, for instance, are known to increase sensitivity to the treatment with gefinitinb, whereas some other mutations in the same gene are known to give resistance to that same drug.

These cancer somatic mutations of crucial therapeutic, diagnostic and/or prognostic value are sometimes belonging to the pool of the so-called driver mutations. Properly, driver mutations are mutations which have a fundamental role in cancer genesis and development. It is because of these driver mutations that a cancer can thrive and invade the human body. Cancer cells also bear several other genetic mutations which are thought not to be as important as driver mutations and which are called passenger mutations. Not all driver mutations can be targetable by a specific drug and, on the other side, not all somatic mutations with a relevant diagnostic or prognostic significance can be proved to be driver mutations, i.e. mutations which drive the growth of the cancer.

Related articles:

• PRECISION ONCOLOGY
• PHARMACOGENOMICS

PRECISION ONCOLOGY

Precision oncology (or precision cancer medicine) can be considered a chapter of personalized medicine. The aim of precision oncology would actually be to tailor patient handling and treatment at a precision that goes beyond the few pharmacogenomic correlations known to date. The existing knowledge on the genetic associations of cancer drugs and biomarkers is still very limited. Only few mutations in few genes (pharmacogenomic markers) have been confirmed to always give resistance or sensitivity  to a certain drug. However, according to several authors, there is a huge (and unfortunately disorganized) amount of data regarding patient treatments and tumor genetic variants which could be analyzed and possibly decoded into complex interaction principles to help personalized medicine to become even more powerful. This would mean identifying additional common driver mutations, rarer driver mutations, epigenetic and environmental factor and characterizing cancer mutations activity on different germline substrata.

Some (big) efforts along the way to build the knowledge foundation necessary to develop precision oncology have been already initiated. See for instance the open-access knowledge base for precision medicine PMC Wiki and Knowledgebase (PMID: 25563458). Both public and private organizations have started to build their own databases and some of their projects are also going beyond genetics, involving the collection of many additional biochemical and clinical data (see for instance Flatiron Health).

Related artiles:

• PERSONALIZED MEDICINE
• PHARMACOGENOMICS