Correlative Science
The Promise of Personalized Medicine

Correlative science is a term used to show the relationship between molecular biology (i.e. biomarkers such as genes and proteins) and clinical outcomes (i.e. disease progression). This is the promise of personalized medicine.

Research is conducted using tissue from patients and comparing it to normal tissue. This tissue is more correctly referred to as a biospecimen. Many people use the term "tissue" to refer to any type of biospecimen sample used for studies, but we will use it only if it is actually tissue and not another substance as described below.

Biospecimens are any material taken from the human body, such as tissue, blood (both serum & plasma), cells (DNA), spinal fluid, bone marrow and urine that can be used for cancer diagnosis and analysis.

When patients have a biopsy, surgery, or other procedure, often a small amount of the specimen removed can be stored and used for later research. These biospecimens are analyzed for patterns that predict:

• Which people are most likely to get cancer
• Which patients are most likely to live longest, without treatment
• Which patients are most likely to benefit from specific treatments
• Better ways to deliver drugs or agents to specific cells
• Methods used to identify how diseases progress and vary

Biorepositories (or biobanks) are "tissue libraries" where biospecimens are stored and made available for scientists to study for research purposes.

Image courtesy of the National Cancer Institute

Clinical Applications of Correlative Science

New technology combined with increased biological knowledge may make it increasingly possible to predict which patients are most likely to benefit or suffer severe side effects from a new treatment.

The collection of tumor tissues from serial biopsies, along with the determination of plasma drug levels and surrogate markers in non-tumor tissues, may be shown to correlate with clinical outcomes providing insight into the drug interaction with the target. These end points are often incorporated into the design of trials using new anticancer agents, and also may be relevant in their development.

This image shows patient outcome data of a biomarker (KRAS), which was discovered in correlative science studies. "Colon cancer patients with KRAS mutations have zero response to treatment with panitumumab (Vectibix, Amgen) and cetuximab (Erbitux, Bristol-Myers Squibb)" - Dr. Baselga told Medscape Oncology. Published in Cancer Research 2006; 66:3992-95

CISN Tip: If you are receiving Erbitux or Vectibix for the treatment of colorectal cancer, you should confirm with your physician that you have been tested for the KRAS mutation. If you have not been tested, based on published results you might want to discuss getting tested with your physician.

Positives elements of correlative science research

Using genomic classifiers to target treatment can greatly improve:

• The therapeutic ratio of benefit to adverse effects
• Smaller clinical trials, saving time and money
• Improved likelihood that a treated patient benefits (closer to personalized medicine)
• Economic benefit for society

Concerns about correlative science research

Tumor Biopsies

Little is known or published on patients', physicians', or institutional review board members' acceptance of and perceptions associated with mandatory, sequential, research-related tumor biopsies.

Biomarker Classifiers

Much of the conventional wisdom about how to develop and utilize predictive biomarker classifiers is thought to be flawed and may not lead to definitive evidence of treatment benefit for a well-defined population.

 

Research implications of poor sample collection and processing (garbage IN)

Irreproducible results leading to confusion and lost data from valuable patient samples. Misinterpretation of artifacts as biomarkers.

Clinical Implications of poor sample collection and processing (garbage OUT)

Potential for incorrect diagnosis Potential for incorrect treatment

 

Bottom line: garbage in - garbage out.