Molecular Profiling and Companion Diagnostics: Where is Personalized Medicine in Cancer Heading?

Mukesh Verma


Personalized Medicine. 2014;11(8):761-771. 

In This Article

Selected Examples in Personalized Medicine in Cancer, CoDxs & Challenges in the Field

Personalized medicine involves more than genetics. Social, family, behavioral, environmental and economic factors also contribute to disease development and resistance to therapy. Some examples follow. All-trans retinoic acid is considered highly effective in treating acute promyelocytic leukemia because of the identification of the PML–ARRA fusion gene.[78,79] Using a similar approach, the BCR–ABL fusion gene was found to be useful in treating chronic myelogenous leukemia and in developing imatinib.[80] Based on the identification and characterization of erbref-2/HER2 amplification, the anti-HER2 therapeutic trastuzumab was developed for use in treating breast and gastric cancers.[81–84] Other examples of personalized medicine in cancer include Herceptin used to treat female breast cancer patients who express higher levels of HER2, and Gleevec used in chronic myeloid leukemias (CMLs) to inhibit tyrosine kinase. Mutations in the tyrosine kinase gene affecting the expression of EGFR in lung cancer, BRAF gene mutations in melanoma and EML–ALK translocations in lung cancer have been reported and have implications in personalized medicine.[85–87] By contrast, some specific gene mutations in different cancers are not useful in cancer treatment as patients do not respond to therapy. Examples include KRAS mutations and antiepidermal growth factor therapy with cetuximab in colorectal cancer,[88] and high ERCC-expressing lung tumor treatment with platinating drugs in lung cancer.[89] In tumors of some patients with low expression of cytochrome P450 (CYP2D2 variant), tamoxifen response was low in estrogen receptor-positive breast cancer patients.[90] Furthermore, patients with high cytochrome P450 expression showed side effects of their treatment with tamoxifen. Researchers estimate that there are 286 known tumor suppressor genes and 33 known oncogenes; many of these genes are being targeted by the pharmaceutical industry.[1] Here it is emphasized that mutated KRAS is not targetable but actionable and should be utilized in personalized medicine because it impacts on treatment choice (being a routine biomarker).

A number of drugs can be used to treat cancer in the presence of specific biomarkers. Examples include irinotecan with UGT1A1 as a biomarker for colorectal cancer, cetuximab with EGFR and KRAS markers for colorectal and head and neck cancers, gentifinib with EGFR-TK mutations as a biomarker for NSCLC, busulfan with Ph+ as a marker for CML, denileukin diftitox with CD24+ as a marker for cutaneous T-cell lymphoma, imatinib with Ph+ and C-Kit as markers for CML and gastrointestinal stroma tumors, trastuzumab with EbR2 overexpression as a biomarker for breast and gastrointestinal cancers, mercaptopurine with TPMT as a biomarker for leukemia, decatinib with Ph+ as a biomarker for acute lymphoblastic leukemia and CML, thioguanine with TPMT as a biomarker for acute leukemia and chronic lymphocytic leukemia (CLL), erlotinib with EGFR+ as a biomarker for NSCLC and pancreatic cancer, nilotinib with Ph+ as a biomarker for CML, arsenic trioxide with PMAL and RAR-α as biomarkers for acute myeloid leukemia (AML), lapatinib with HER2+ as a biomarker for breast cancer and panitumumab with EGFR and KRAS as biomarkers for colorectal and breast cancers.

Personalized medicine involves not only tailoring the right treatment/drug for the right person but also evaluating predisposition to disease, sometimes several years before a disease is fully developed (e.g., before metastasis). Additional aspects of the infrastructure remain to be established before personalized medicine can be translated into practice. For example, the 'one-size-fits-all' style of treating cancer still is in use, regardless of whether the treatment comprises chemotherapy or radiation. An example is the genetics-based drug, ipilimumab, which is used in treating melanoma and has a response rate of only 11%. Another drug, PLX4032, also is ineffective for treatment.[21–23] Contradictory reports in colon cancer treatment with cetuximab-containing combination therapy were observed, making it difficult for physicians to make treatment decisions.[91]

The term 'companion biomarker' sometimes is used in the field of personalized medicine for biomarker-based tests that will assist a physician in making a treatment decision by selecting a drug that is associated with the particular test. The advantage of such an approach is that this study design requires only a small number of participants to follow the effects of treatment.

Challenges in personalized medicine include matching advanced technologies (genomics, proteomics, epigenomics) with in silico techniques, resolving tumor heterogeneity-associated problems in patients' molecular profiles, interpreting bioinformatics results and ethical issues associated with genetic testing and the use of results in therapy.[92,93] Currently, there is no validated technology that has been proven to efficiently integrate clinical and biomedical data from diverse domains and deliver targeted knowledge to doctors in a language that they can follow. Furthermore, there are no clear guidelines for practicing personalized medicine in cancer except in treating a few selected tumor types.

For personalized medicine to be completely successful, caregivers – including doctors, nurses and clinicians – should be trained in and familiar with the latest molecular profiling of cancer samples and interpreting results for implementation in clinical practice. Personalized medicine and point-of-care testing techniques must fulfill a number of constraints for real-world applicability. Social, ethical, legal and regulatory issues surrounding adopting a personalized medicine approach should be discussed by the health community.

Selected challenges in the CoDx field include implementing co-developed CoDx in today's clinical, ethical and logistic environment; uncertainty about the regulatory pathways for the therapeutics-associated test; and a weak business case to support in the field. Despite efforts made by the FDA, progress has been slow.[9]

Recently it has demonstrated that patients with or without PIK3 mutation behave differently are subjected to lifestyle changes and use common drugs, such as aspirin, which is not a targeted and expensive drug response differently for colon cancer treatment.[94,95] This group demonstrated that personalized medicine plays an important role in patient survival. This study included markers such as PTGS2, phosphorylated AKT, KRAS, BRAF, microsatellite instability, CpG island methylator phenotype (CIMP) and methylation of long interspersed nucleotide element-1. This study involves two prospective cohorts, Nurses Health Study and Health Professional Follow up Study, where more than 100,000 people were enrolled.

Personalized medicine becomes more significant when we realize that the individual tumor possesses its own unique characteristics in terms of molecular make up, tumor microenvironment and its interaction with host machinery. The Cancer Genomic Atlas[96] program of the NCI also emphasizes the importance of tumor characteristics and tissue samples in cancer genomics and our understanding about how cancer genomics are changing the way we approach cancer diagnosis and treatment. Attempts in single cell technologies may also help in characterizing intratumor heterogeneity. It is worth gathering data on tumor heterogeneity in cancer registries around the world. Prioritization of biomarkers which should be included in registries should also be done because the number of tumor biomarkers is so large that all cannot be included.

In association with international immunologists, Immunoscore methodology was recently established by Galon et al.[97,98] to assess in clinical practice the immune infiltrate. The type, the density and the location of immune cells within the tumor influence prognosis of cancer. Immunoscore may distinguish tumors isolated from different patients with the same stage. Tumor immunotherapy has potential in cancer treatment. By applying integrative tumor immunology approaches, a comprehensive view of the immune system evolution can be observed.

In immunotherapy, patient's own T cells (cells in the body that are capable of recognizing, attacking and destroying foreign invaders) are genetically engineered to kill cancer cells selectively.[99–102] This application of immunotherapy falls under personalized medicine. Attempts are being made to develop personalized vaccines.[103–105] Positive results in lymphocytic leukemia provide hope to thousands of patients. There is a potential in cancer genomics to identify specific tumor mutations in patients that may be used as targets in cancer vaccines to overcome problems linked to self-antigens.

With the advancement in technologies and their wide applications in cancer diagnosis and prognosis 'precision medicine' term is being used,[106–109] but for this article I have used 'personalized medicine' and all the discussion is based on the term personalized medicine.[110–116]