Holding onto the concept of PM as a genetic biomarker-driven treatment, an increasing number of cases have been accumulated in recent years. Of these, the most cited examples include the use of imatinib for BCR-ABL positive chronic myeloid leukemia or c-Kit positive gastrointestinal tumors [
7], anti-HER2 antibodies for HER2 amplified breast or gastric cancers [
8], anti-EGFR antibodies for non KRAS mutated colorectal cancers [
9], small molecule EGFR inhibitors for EGFR mutated lung cancers [
10,
11], BRAF inhibitors or combinations of BRAF and MEK inhibitors for melanoma patients bearing BRAF mutations in their tumors [
12] or ALK inhibitors for ALK or ROS translocations in lung cancer [
13]. Conventional Phase III trials have shown that when patients are stratified by the use of a genetic test known as “companion diagnostics” to receive targeted therapy, patients testing “positive” for the biomarker experience a superior clinical benefit in terms of progression free survival and/or overall survival as compared to those testing “negative” for the same biomarker. This approach has often allowed accelerated market approval for the corresponding drugs [
14]. It must be added that meta-analyses including a total of approximately 85,000 patients have confirmed that the genetic biomarker-driven patient selection is safe and has been associated with improvements in all outcome variables [
15‐
17]. In addition, it would be considerate that the increasing use of this molecular approach in both cancer research and clinical practice, bring to a higher expense for target drugs, which are compensated by a less overall costs for the Health System coming from the better patient outcome and reduction of hospital admissions.
However, strictly speaking, these examples of PM are not truly considered “personalized medicine” since they are not tailored to individual patients, but rather to subgroups of patients sharing only one of the several genetic alterations present in their tumors.
The genetic biomarker-driven concept of PM has been challenged by a series of facts and evidence. Firstly, the presence or absence of the specific biomarker does not always result in biological and clinical sensitivity to the corresponding drug. For example, a subset of lung cancer patients which do not bear activating EGFR mutations can achieve clinical responses to EGFR inhibitors [
18], or also a good proportion of BRAF mutated melanoma patients do not respond to BRAF inhibitors [
19]. The case of BRAF mutations is even more intriguing because activating oncogenic BRAF mutations are found in several other tumors, including colorectal, thyroid, lung cancer but in the majority of those cases they are not predictive of drug response to the same BRAF inhibitors as in melanoma. Mechanistic explanations to these findings are emerging and reside in the presence of additional genetic or epigenetic alterations which may create from case to case “favorable” or “unfavorable” contexts to the action of a specific target therapy. This brings us to the second line of evidence: tumors are in general highly heterogeneous and mutated in several driver genes.