TABLE OF CONTENTS
- Population Study Biomarkers and Evaluation Criteria
- Laboratory/Clinical Perspectives on Biomarkers
- Observatinal Studies/Clinical Trials Experiences
- Integrating Biomarkers/Methodological Considerations
- Summary and Recommendations
- Workshop Members
Normal metabolic processes generate potentially deleterious reactive oxygen species, oxidative damage and inflammation, which increase with age and may contribute to senescence. Oxidative stress and inflammation have been implicated in many heart, lung, blood, and sleep (HLBS) disorders, including atherosclerosis, hypertension, asthma, COPD, acute lung injury, heart failure and sleep apnea. CRP and other measures of inflammation have been incorporated into many population studies. However, there are many components to these complex systems and it is unclear which are the most predictive of disease, and how these measures can be integrated with measures of oxidative stress. Both oxidative stress and inflammation biomarkers, as well as anti-oxidants and anti-inflammatory factors, need to be measured concurrently since the metabolic processes overlap and the lack of balance between these stressors and protectors may lead to development or progression of HLBS disorders. While there are many biomarkers of inflammation, oxidative stress is difficult to measure in vivo. Reliable surrogate biomarkers for use in large-scale population studies need to be identified. The NHLBI thus convened a workshop on August 18-19, 2004 with the following objectives: 1) assess the opportunities for incorporating measures of oxidative stress and inflammation into population studies; and 2) recommend future research directions for studying the role of these risk factors in the development of HLBS disorders.
The workshop began with recognition of the importance of studying oxidative stress in population studies despite the failure of antioxidant trials, and an overview of criteria to be considered in selecting biomarkers for population studies. Laboratory/clinical perspectives on several biomarkers and/or methods were next presented, followed by the experience to date in utilizing these biomarkers in several NHLBI observational studies and clinical trials. Integrating oxidative stress and inflammation biomarkers and methodological issues were also addressed. The workshop concluded with summary and recommendations on which biomarkers were currently suited for use in large scale population studies, and what additional data and/or resources were needed to continue work in this field.
Population Study Biomarkers and Evaluation Criteria
Dr. Nieto outlined issues regarding selecting biomarkers for large scale population studies. Biomarkers may measure internal exposure, host susceptibility, or early effects of disease. An individual biomarker may be involved in several physiologic pathways. Biological validity, sensitivity and specificity, standardization, and reproducibility, as well as the need for special collection methods, must be evaluated during planning. It is important to realize that strength of association does not always equal predictive value and thus the contribution of some of these novel markers to existing predictive equations based on traditional risk factors might be modest. Biomarkers might prove most valuable in furthering our understanding of the pathophysiology of complex conditions by selecting those that most accurately reflect the stage of pathophysiology or hypothesis to be studied.
Laboratory/Clinical Perspectives on Biomarkers
F2-isoprostanes: Dr. Morrow reviewed the biochemistry of F2-isoprostanes and their utility as oxidative stress biomarkers. Gas chromatography/mass spectrometry measurement of F2-isoprostanes has been shown to be robust and reliable using commercially available standards. Some researchers also report that immunoassay methods are acceptable although these antibodies were generated in individual investigator laboratories. Commercially available immunoassay kits may present difficulties with reproducibility and cross-reactivity issues. Evidence that this class of biomarkers reflects in vivo lipid peroxidation was presented. Some diurnal variation is seen in individuals but this is not significant at the population level. Specimens for isoprostane analysis must be analyzed immediately or stored at -70ºC to prevent artifact generation, which is primarily a concern for plasma specimens. F2-isoprostanes can be detected in all tissues and body fluids. Several studies in which F2-isoprostanes levels have been associated with cardiovascular risk factors were discussed. Weight and BMI in particular have been strongly associated with elevations in F2-isoprostanes. While vitamin E appears to be a relatively weak antioxidant, with sufficient time lapse, a decrease in F2-isoprostanes in a small study of hypercholesterolemic humans was eventually seen.
Reactive Oxygen and Nitrogen Species: Dr. Harrison discussed several reactive oxygen and nitrogen species. Loss of endogenous nitric oxide production is associated with hypercholestrolemia, diabetes, hypertension, aging and heart failure. In vascular cells, the reduced form of nicotinamide adenine dinucleotide phosphates, NADPH oxidases, are a major source of reactive oxygen species. Electron spin resonance measurement of xanthine and NADPH oxidases correlates with endothelial dysfunction in patients with coronary artery disease, but is too difficult to perform clinically or in population studies. Measurement of intracellular antioxidant thiols, such as glutathione, may be more feasible as they can be estimated using the free oxygen radical test (FORT) via spectrophotometric techniques on specimens obtained from fingersticks. FORT can be performed on frozen samples. Increased glutathione redox ratios and increased FORT values have been associated with increased IMT, microalbuminuria and depression.
Biomarkers in Exhaled Breath: Dr. Erzurum reviewed the biomarkers of systemic and pulmonary disease that can be measured non-invasively in exhaled breath and exhaled breath condensate using chemiluminescent analyzers. Increased exhaled nitric oxide is associated with upper respiratory tract infections, asthma and lung cancer, while decreased exhaled nitric oxide is associated with smoking, glucocorticoid steroids, pulmonary hypertension and cystic fibrosis. Collection of exhaled nitric oxide can be done in participants of all ages including children, and in a clinic visit or offsite via collection of breath into balloons that can be stored and shipped. There is good correlation among analyzer equipment and good reproducibility within subjects. Exhaled nitric oxide may be a good biomarker of lower airway inflammation and suitable for monitoring anti-inflammatory therapy in asthmatic patients or exacerbations in disease activity. Genetic variants in nitric oxide synthase have been identified. Relatively little is known about the reliability of exhaled breath condensate measurement.
Mass Spectrometry Methods: Dr. Hazen discussed the use of mass spectrometry to create molecular fingerprints of oxidized tyrosine species that result from NADPH and NOS pathways. These biomarkers are not affected by freeze/thaw cycles and assays can be performed on de-paraffinized tissues. The coefficient of variation is low for both intra- and inter-assays. Normal levels have been established for several gender, age, and race/ethnic groups in healthy subjects. Bromotyrosine is associated with FEV1 and airways obstruction in asthmatics. Nitrotyrosine is associated with endothelial dysfunction and coronary artery disease. Evidence of the antioxidant effects of statins on inhibiting formation of superoxide and oxidized tyrosines was presented. Myeloperoxidase was found to oxidize apoA1 in subjects with cardiovascular disease, and to be an independent predictor of future adverse outcome in troponin-negative patients. Methylated arginine species (ADMA), which are stable to storage and freeze/thaw cycles, have been associated with the metabolic syndrome.
Observational Studies/Clinical Trial Experience
Framingham Heart Study: Oxidative Stress and Inflammation: Dr. Benjamin began by reiterating the desired characteristics of a population study biomarker: measurable via an assay that performs well in the low range likely to be found in a community-based study; feasible on a large scale; and utilize a minimal amount of specimen. ELISA assays are often practical but do not have high throughput. Decisions must be made on whether to measure a biomarker in the entire cohort or within a case-referent study. The Framingham Heart Study has experience with several oxidative stress and inflammation biomarkers. HSP-70 assays were not fruitful. BMI was found to be an independent risk factor for urinary F2-isoprostane levels. CRP levels were significantly associated with hyperemia, but not flow-mediated dilation, after adjustment for other risk factors. The heritabilities of several inflammatory biomarkers were presented, and ranged from 14%-44% after multivariate adjustments. Challenges such as multiple-statistical testing and accessibility of data, particularly in light of negative publication biases, were discussed.
CARDIA Study: F2 -Isoprostanes: Dr. Gross described how endogenous and exogenous stressors result in oxidative stress, but need not lead to disease if the stressors are absorbed without damage, or a sufficient adaptive increase in antioxidant pools can be induced. Temporality versus causation is also an issue; the goal is to determine whether oxidative damage precedes the development of clinical events. F2-isoprostanes are likely to be a systemic biomarker of oxidative stress, and can be measured in either plasma or urine. The plasma assay requires a larger sample but has been validated more than the urine assay. Artifact formation can be avoided in both specimen types. Gas chromatography/mass spectrometry measurement is more costly than immunoassays, but more specific, sensitive, precise and reliable. Whole blood specimens were stable with slight hemolysis or single freeze-thaw cycles, but F2-isoprostanes levels increased with storage at -20ºC. In the CARDIA Study, elevated levels of F2-isoprostanes were associated with female gender, smoking, triglycerides, HDL, alcohol and BMI. Decreased levels were associated with blood antioxidant levels, exercise, education and African-American race.
DASH Studies: Dietary Patterns and Oxidative Stress: Dr. Miller reviewed the apparent paradox between the role of antioxidants in reducing oxidative stress and the failure of many antioxidant supplementation trials, and described the findings of the DASH and DASH-sodium trials. The Oxygen Radical Absorbing Capacity, or ORAC, antioxidant assay was utilized. ORAC values are primarily derived from thiol proteins and uric acid, with smaller contributions from vitamins A, C and E, folic acid and bilirubin. The DASH trials featured controlled nutrient intake and repeated measurements, allowing for longitudinal analyses of biomarkers and traditional CVD risk factors. Those participants following the DASH diet had increases in serum antioxidant and ORAC levels, and reduced breath ethane levels. Similarly in the DASH Sodium Trial, those on the DASH diet had increases in serum antioxidant and ORAC levels, lowered F2-isoprostane levels and increased antibodies to oxidized LDL. Thus modulation of oxidative stress biomarkers was demonstrated to be achievable with dietary changes.
San Antonio Family Heart Study: Genetic Influences on Oxidative Stress and Inflammation Biomarkers: Dr. Mahaney reported heritability estimates from the San Antonio Family Heart Study of eight biomarkers of oxidative stress and ten biomarkers of inflammation, along with the coefficients of variation for the assays used. Particularly high heritabilities were found for advanced glycation endproduct (ACE, 0.91) and paranoxase (PON, 0.84). Plasma nitrotyrosine levels did not differ in diabetics compared to non-diabetics, nor were the levels associated with fasting or 2-hour glucose, smoking or BMI. Levels were higher in women than men. Evidence for pleiotropy between plasma total antioxidant activity (TAS) and other plasma CVD risk factors, including fasting insulin and glucose, CRP, VCAM and fibrinogen, was presented. These data suggest a common genetic basis for interindividual variation in redox homeostasis and these other plasma CVD risk factors. Genotype-by-environment interactions were found for the biomarker genotype with sex and smoking. No significant evidence of chromosomal quantitative trait loci for TAS has been found to date, possibly due to the composite nature of the measure; however, some of the additive genetic effect appears due to a mitochondrial locus. Valid biomarkers are necessary for all types of studies; in statistical genetic studies accuracy is also desired, but precision is essential. Genetic association studies should only be done in populations in which there is some evidence of genetic influence on the phenotype/trait of interest.
Integrating Biomarkers/Methological Considerations
Integrating Oxidative Stress and Inflammation Biomarkers: Dr. Tracy discussed plasma biomarkers of inflammation related to atherosclerosis and how they represent the underlying physiology or pathophysiology in a graded and continuous manner; however, any individual biomarker may or may not be in the causal pathway of any particular health outcome. Inflammation biomarkers have been associated with multiple outcomes and may have different meanings across the age span. CRP, a systemic biomarker, is only weakly associated with ethnicity, gender, age, and hypertension and moderately associated with glucose tolerance status. However, CRP is strongly associated with obesity and a number of components of the metabolic syndrome; associations which begin early in life and persist throughout the life span. Interestingly, CRP is not consistently associated with IMT or coronary calcification, but has been shown to co-localize with the NADPH oxidase subunit p22phox in atherosclerotic plaque. Oxidized LDL, a marker of oxidative stress is also strongly associated with the metabolic syndrome and, like CRP and other inflammation markers, predicted incident MI in the Health ABC Study. It appears likely that inflammation and oxidative stress biomarkers may reflect different aspects of the underlying pathophysiology. The concept of "antagonistic pleiotropy" was introduced, in which genetics, early life "programming" and current physiological state may result in short-term benefit to an individual but, due to longer spans, long-term damage. Due to the way defense mechanisms are integrated with basic metabolism, larger panels of individual biomarkers or metabolic processes may need to be measured in population studies. These approaches; however, will result in large amounts of data to be interpreted. Thus, new conceptual frameworks and data analytic tools for understanding oxidative stress and inflammation, particularly in population studies, will be needed.
Methodological Considerations: Dr. Schisterman stated that there are a number of biomarkers available to measure different phases of oxidative stress status. However, less is known about appropriate choice of biomarkers to use in a study, and in particular, which combination of biomarkers might distinguish between those with and those without a disease such as coronary heart disease. Methods and issues discussed included: 1) use of ROC curves and AUC (area under the curve) values; 2) measurement error corrections; 3) pooling biomarkers, which is cost-effective and particularly useful in pilot studies; and 4) linear combination assessments. AUC can be used to evaluate discrimination ability of biomarkers. Five biomarkers of oxidative stress or antioxidant status were demonstrated to be independently associated with CVD, however; using AUC- linear combination assessments, TBARS were found to contribute the most discriminatory ability. Ultimately, the choice of biomarkers should be made based on biological relevance, not statistical relevance.
Summary and Recommendations
As evidenced in the workshop presentations, oxidative stress and inflammation biomarkers are associated with many HLBS conditions and risk factors in population and clinical studies including participants across a broad spectrum of age, race/ethnicity, gender, geographical regions. Genetic and environmental influences on the biomarker levels were also demonstrated. Clearly, this work is still in the early stages of investigation. The Workshop Members recognized that ideally, biomarkers for population studies need to meet criteria related to reproducibility, biological variability, analytic variability, sensitivity and specificity in a healthy population, as well as large-scale study feasibility. However, it was also recognized that the data necessary to fully evaluate these criteria are not yet available for most biomarkers. The Workshop Members are building on their own experience as well as compiling published information related to these evaluation criteria for several biomarkers, which will be presented in a forthcoming report.
While there are many useful inflammation biomarkers, there are few markers of oxidative stress. The most studied oxidative stress biomarkers to date are the F2-isoprostanes, which meet the criteria for use in large-scale population studies; however, other markers, in particular those found in exhaled breath, may also be suitable.
Additional biomarkers are certainly needed, as Workshop Members agreed that no one biomarker is likely to serve as a general marker of oxidative stress. The same biomarker measured in different specimens (i.e., plasma, urine, exhaled breath, etc) provides information on different aspects of the underlying physiology and pathophysiology. A combination of biomarkers will be needed to accurately reflect oxidative stress and inflammation status across the range of normal physiology and HLBS disorders. In addition, it was noted that many of the available assays are technically quite challenging and only a few laboratories have the resources and/or skills necessary to perform them. Stronger collaboration between laboratory scientists and epidemiologists is critical to further advance the application of new biomarker techniques to large-scale population studies.
The following recommendations were made:
- Biomarkers of oxidative stress status and inflammation have been associated with many HLBS conditions and risk factors; additional examination in population studies is important to elucidate which are most predictive of disease.
- Further standardization of specimen collection, storage, processing and assaying conditions would facilitate evaluation of the biomarkers for use in large scale studies and comparison of disease/risk factor associations across studies.
- If existing specimen resources in the ongoing or completed population studies are not sufficient or have not been collected appropriately to enable this evaluation and hypothesis testing, additional biological specimen repositories should be established, along with appropriate funding to enable their use.
- Less technically challenging assays that can be more easily performed in general research laboratories, particularly with high throughput, should be developed. For assays where this is not feasible, funding of core laboratories to serve multiple studies is encouraged.
- Methods for combining and integrating multiple or large panels of biomarkers of oxidative stress and inflammation into analyses should be developed.
- Dissemination of negative results, which are often not published but do provide useful information to other researchers, should be encouraged. Abstracts of these results could be posted on study websites, although peer-review is also important.
- While some intervention studies examining the effects of diet and exercise on biomarkers of oxidative stress and inflammation have been done, more studies are needed to better understand the underlying biology and help direct future development of therapies.
David G. Harrison, MD, Division of Cardiology, Emory University School of Medicine
F. Javier Nieto, MD, PhD, Department of Population Health Sciences, University of Wisconsin Medical School
Emelia J. Benjamin, MD, ScM, Framingham Heart Study, Boston University School of Medicine
Serpil C. Erzurum, MD, Pulmonary and Critical Care Medicine, Cleveland Clinic Foundation
Myron D. Gross, PhD, Department of Laboratory Medicine and Pathology, University of Minnesota
Stanley L. Hazen, MD, PhD, Department of Cell Biology, Cleveland Clinic Lerner Research Institute
Michael C. Mahaney, PhD, Department of Genetics, Southwest Foundation for Biomedical Research
Edgar R. Miller, MD, PhD, Departments of Medicine and Epidemiology, Johns Hopkins Medical University
Jason D. Morrow, MD, Department of Clinical Pharmacology, Vanderbilt University Medical Center
Enrique F. Schisterman, Epidemiology Branch, National Institute of Child Health and Human Development, NIH, DHHS
Russell P. Tracy, PhD, Departments of Pathology and Biochemistry, University of Vermont
Maurizio Trevisan, MD, PhD, Department of Social and Preventive Medicine, State University of New York at Buffalo
Last Updated June 2011