[Clinical prediction in medical oncology]

Predictive factors (PF) are variables that give information about survival, treatment response or toxicity and future complications on cancer patients. The foremost utility of PF takes root in the possibility to furnish an individualized treatment schedule with higher succeeding options. They include clinical characteristics of the patient, tumour features, treatment administrated, classical pathological and new molecular data obtained from patients clinical samples. Clinical parameters comprise age, sex, underlying diseases and performance status among others, and in concurrence with tumour pathology and clinical stage (TNM) usually define the best treatment options. Also, chemotherapy response can modify natural history of several tumours, and thus is a PF. Modifications in evolving PF typically induce a variation in patient outcome. Hence, surgical tumour size reduction or neoadjuvant down-staging improve survival in several cancers. In the other side, treatment adjustment to steady PF should offers better outcome than "standard therapies". Recent advances on cancer research have generated a great deal of biological data that help us to search new treatment and diagnostic modalities. Biotechnology offers a great amount of possibilities in the next future and probably a true individualized therapy. Conversely, there are a small amount of molecular evidences that imply a creal variation in current clinical practice. Hence, more scientific and financial efforts are necessary to exploit to the full knowledge spurting up from basic science. In summary, the prediction in oncology is a hard task derived from clinical observation, tumour behaviour, treatment schedules and biological evidences that must offer realistic predictions on a concrete cancer patient. Oncologists have a duty to know all these variables to accomplish this thorny assignment. This review will focus on classical and recent biological PF in cancer.