While biomarkers may be used to diagnose disease, the ups and downs of biological molecules may not depend entirely on whether a particular disease is present. Protein levels in the blood stream, for example, may vary with individual genetic, clinical, and lifestyle factors.
This observation isn’t exactly novel. Biomarkers for specific diseases have been linked with lifestyle and demographic factors. For example, in a study of diabetes that considered how well the disease could be predicted by assessing levels of a protein called sex hormone binding globulin (SHBG), relevant factors were found to include age, reproductive history, usage of exogenous estrogen, body mass index, physical activity, alcohol consumption, coffee intake, smoking, and various dietary factors.
When these findings appeared last year, study leader Simin Liu, M.D., a professor of epidemiology and medicine at Brown University, said, “This protein seems to capture the cumulative effect between the gene and our environment in reflecting a metabolic state of our body, particularly in the liver, ultimately affecting diabetes risk.”
If individual genetic and lifestyle factors can influence one biomarker, they can presumably influence many other biomarkers. Yet biomarkers used for diagnosing disease should preferably indicate variations in protein levels only for those individuals who are suffering from a particular disease. Nor should they vary for reasons that have nothing to do with the disease.
To address this issue in a systematic way, researchers from Uppsala University recently conducted a large-scale study of non-disease-related factors that can influence protein levels in the bloodstream. Specifically, the researchers, led by Stefan Enroth, Ph.D., analyzed 92 protein biomarkers for cancer and inflammation in a clinical study of 1,000 healthy individuals.
The results of the study appeared August 22 in Nature Communications, in an article entitled, “Strong effects of genetic and lifestyle factors on biomarker variation and use of personalized cutoffs.” According to this article, hereditary factors play a significant role for more than 75% of the proteins, and a detailed genetic analysis demonstrates 16 genes with a strong effect on protein levels.
“These results are important, as they show which variables are significant for variations in the measurable values,” explained Dr. Enroth. “If these factors are known, we have a greater possibility of seeing variations and we get clearer breakpoints between elevated values and normal values. By extension this may lead to the possibility of using more biomarkers clinically.”
According to the study, genetics and lifestyle together account in some cases for more than 50% of variations in protein levels among healthy individuals. This means that information about both genetic and lifestyle factors must be taken into account in order for protein biomarkers to be used effectively.
Dr. Enroth kindly responded to a few questions.
GEN: Could your investigation be regarded as an application of proteomics?
Dr. Enroth: Yes, in a broad sense, I would say so. But this would perhaps differ when you ask somebody else.
GEN: Was accounting for personal variations across dozens of proteins a data-intensive activity?
Dr. Enroth: In the first stage, such as reported in the paper, yes. But most of the CPU time has been used on the genetic associations, mainly because of the sheer numbers of variants that have been characterized in the human genome. After the models are built, not really. Calculating the residuals is fast.
GEN: How would personalized cutoffs be determined?
Dr. Enroth: We envision that studies such as ours, possible larger and encompassing also ethnicity (genetic markers such as SNPs have different baseline frequencies in different parts of the world), will determine factors that influence the circulating levels of the proteins. In a clinical setting, a biomarker level would be monitored in a patient, and this level would then be recalculated given your personal facts—your age, your weight, what medications you use, etc., and then compared to a predefined value to determine if your levels are deviating.