The non-selective nature along with metabolic specificity and various molecular windows available are advantageous characteristics of 1H NMR spectroscopy in metabonomics. Furthermore, the abundant information on various relevant biomolecules makes 1H NMR spectroscopy a well suited methodology for multi-parametric biochemical and clinical studies of body fluids [1]. We have developed a new metabonomics framework to visualize and interpret the serum data and to link the metabolic profiles to the underlying diagnostic and biochemical variables [2]. In general, our work demonstrates the diffuse nature of complex vascular diseases and the limitations of single ‘diagnostic’ biomarkers. However, it also promises cost-effective solutions via high-throughput analytics and advanced computational methods [1-3].

The talk will focus on the methogological issues as well as on the biomedical applicability and rationale of 1H NMR metabonomics. Our recent data and results will be shown to demonstrate the inherent suitability of 1H NMR spectroscopy of serum to identify subtle changes in lipoprotein subclass-related metabolism together with various other metabolites of high interest in clinical medicine. Also, associations of various 1H NMR-based multi-metabolic phenotypes with clinical diagnostics (e.g., micro- and macrovascular complications) as well as mortality will be discussed in type 1 diabetes [2]. A 1H NMR metabonomics study of the systemic metabolic phenotypes that relate to mild cognitive impairment and thereby to high risk for the Alzheimer’s disease will also be discussed [3].

I am grateful to all my colleagues and co-authors that share the scientific interest and endeavor for NMR metabonomics and the multidisciplinary field of ‘omics sciences. This work has been financially supported by the Academy of Finland Centre of Excellence program for 2006–2011.

• Ala-Korpela M. Potential role of body fluid 1H NMR metabonomics as a prognostic and diagnostic tool. Expert Rev Mol Diagn. 2007;7:761–773.
• Mäkinen VP, Soininen P, Forsblom C, Parkkonen M, Ingman P, Kaski K, Groop PH, Ala- Korpela M. 1H NMR metabonomics approach to the disease continuum of diabetic complications and premature death. Mol Syst Biol. 2008;4:167.
• Tukiainen T, Tynkkynen T, Mäkinen VP, Jylänki P, Kangas A, Hokkanen J, Vehtari A, Gröhn O, Hallikainen M, Soininen H, Kivipelto M, Groop PH, Kaski K, Laatikainen R, Soininen P, Pirttilä T, Ala-Korpela M. A multi-metabolite analysis of serum by 1H NMR spectroscopy: early systemic signs of Alzheimer's disease. Biochem Biophys Res Commun. 2008;375:356– 361.

The non-selective nature along with metabolic specificity and various molecular windows available are advantageous characteristics of 1H NMR spectroscopy in metabonomics. Furthermore, the abundant information on various relevant biomolecules makes 1H NMR spectroscopy a well suited methodology for multi-parametric biochemical and clinical studies of body fluids [1]. We have developed a new metabonomics framework to visualize and interpret the serum data and to link the metabolic profiles to the underlying diagnostic and biochemical variables [2]. In general, our work demonstrates the diffuse nature of complex vascular diseases and the limitations of single ‘diagnostic’ biomarkers. However, it also promises cost-effective solutions via high-throughput analytics and advanced computational methods [1-3].

The talk will focus on the methogological issues as well as on the biomedical applicability and rationale of 1H NMR metabonomics. Our recent data and results will be shown to demonstrate the inherent suitability of 1H NMR spectroscopy of serum to identify subtle changes in lipoprotein subclass-related metabolism together with various other metabolites of high interest in clinical medicine. Also, associations of various 1H NMR-based multi-metabolic phenotypes with clinical diagnostics (e.g., micro- and macrovascular complications) as well as mortality will be discussed in type 1 diabetes [2]. A 1H NMR metabonomics study of the systemic metabolic phenotypes that relate to mild cognitive impairment and thereby to high risk for the Alzheimer's disease will also be discussed [3].

• Ala-Korpela M. Potential role of body fluid 1H NMR metabonomics as a prognostic and diagnostic tool. Expert Rev Mol Diagn. 2007;7:761–773.
• Mäkinen VP, Soininen P, Forsblom C, Parkkonen M, Ingman P, Kaski K, Groop PH, Ala- Korpela M. 1H NMR metabonomics approach to the disease continuum of diabetic complications and premature death. Mol Syst Biol. 2008;4:167.
• Tukiainen T, Tynkkynen T, Mäkinen VP, Jylänki P, Kangas A, Hokkanen J, Vehtari A, Gröhn O, Hallikainen M, Soininen H, Kivipelto M, Groop PH, Kaski K, Laatikainen R, Soininen P, Pirttilä T, Ala-Korpela M. A multi-metabolite analysis of serum by 1H NMR spectroscopy: early systemic signs of Alzheimer's disease. Biochem Biophys Res Commun. 2008;375:356– 361.

Purpose/Introduction – Circulating lipoprotein particles, and the lipoprotein phenotype in serum, are key metabolic components affecting the individual risk for vascular complications (1).

Subjects and Methods – Recent experimentation will be presented to illustrate the flexible role of 1H NMR spectroscopy in lipoprotein research. The quantitative characteristics for isolated lipoprotein subclass particles as well as for human serum samples will be demonstrated and critically discussed in the case of extensive sample collections and accompanied biochemical and clinical data.

Results – Generally, 1H NMR spectroscopy appears to be a versatile methodology for quantitative studies of lipoprotein particles both in isolation and in biological mixtures. For example, the effects of phospholipase A2 on low density lipoprotein particles in a physiological environment can be non-destructively followed and at 800 MHz all major phospholipids, including lysophosphatidylcholine, in the particles can be identified and quantified (2). A newly developed metabonomics framework to visualize and interpret the serum data and to link the metabolic profiles to the underlying diagnostic and biochemical variables will be introduced (3). Associations of various NMR-based multi-metabolic phenotypes with clinical diagnostics as well as mortality will also be assessed for an extensive cohort of type 1 diabetic patients (3). A common feature for all 1H NMR spectroscopy applications in the field of lipoprotein research is clearly the need for advanced data analyses.

Discussion/Conclusion – The central role of lipoprotein subclasses in the risk assessment of atherothrombosis, diabetes, and other common diseases is currently well-established. Furthermore, the abundant information on other, potentially relevant biomolecules, makes 1H NMR spectroscopy of serum a well suited methodology for multi-parametric biochemical and clinical studies (1,3,4). Our recent applications demonstrate the diffuse nature of complex vascular diseases and the limitations of single diagnostic biomarkers. However, they also show potential for cost-effective clinical solutions via high-throughput analytics and advanced computational methods. It appears that 1H NMR metabonomics of serum is inherently suitable for identifying subtle changes in lipoprotein subclass-related metabolism together with several other metabolites of high interest in disease risk assessment and diagnostics.

• M. Ala-Korpela, 2008, Clin Chem Lab Med, 46, 27.
• P. Soininen, K. Öörni, H. Maaheimo, R. Laatikainen, P. T. Kovanen, K. Kaski, M. Ala-Korpela, 2007, Biochem Biophys Res Commun, 360, 290.
• V.-P. Mäkinen, P. Soininen, C. Forsblom, M. Parkkonen, P. Ingman, K. Kaski, P.-H. Groop, M. Ala-Korpela, 2008, Mol Syst Biol, 4, 167.
• M. Ala-Korpela, 2007, Expert Rev Mol Diagn, 7, 761.

The diagnostics, screening and prognostics of brain related diseases pose generally fundamental difficulties. Recently the application of body fluid proton ( 1H) nuclear magnetic resonance (NMR) metabonomics has received considerable interest also in this area [1]. A brief introduction to the methodological aspects and recent key findings will be given in the first part of the talk. Then the focus will move to our recent application of 1H NMR metabonomics of serum to study if it would be possible to identify metabolic phenotypes characteristic for mild cognitive impairment (MCI) [2]. If this were possible there might be means to identify individuals having an increased risk for Alzheimer's disease (AD). Importantly, recent findings point towards a fundamental role of cholesterol and lipoproteins in AD pathophysiology. Consequently, 1H NMR metabonomics offers an appealing new approach to investigate serum metabolism also in relation to MCI and AD [3,4]. We applied a combination of three molecular windows that leads to information on lipoprotein subclasses, various low-molecular-weight metabolites and also individual lipid molecules together with their degree of (poly)(un)saturation [2]. The data were analysed by self-organising maps, a methodology facilitating a solely data-driven overview on the metabolic phenotypes and their associations with the clinical characteristics [5]. In fact, MCI-related metabolic phenotypes were identified. Many of the known associations between vascular risk factors and MCI were revealed. It was also found that low high-density lipoprotein (HDL) cholesterol and high triglycerides, typical characteristics of the metabolic syndrome, as well as obesity and diabetes were clearly associated with MCI.

Acknowledgements. I am grateful to all my colleagues and co-authors, particularly in reference [2], who share the scientific interest and endeavour for NMR metabonomics and the multidisciplinary field of 'omics sciences. This work has been financially supported by the Academy of Finland Centre of Excellence program for 2006-2011.

• E. Holmes, T. M. Tsang, S. J. Tabrizi. The application of NMR-based metabonomics in neurological disorders. NeuroRx. 3, 358-372 (2006).
• T. Tukiainen, T. Tynkkynen, V.-P. Mäkinen, Hokkanen J, P. Soininen, Jylänki P, Vehtari A, Gröhn O, Hallikainen M, Soininen H, Kivipelto M, P.-H. Groop, R. Laatikainen, K. Kaski, Pirttilä T, M. Ala-Korpela. Characterisation of metabolic phenotypes associated with cognitive decline by proton NMR metabonomics of serum. In preparation (2008)
• M. Ala-Korpela. Potential role of body fluid 1H NMR metabonomics as a prognostic and diagnostic tool. Expert Rev Mol Diagn. 7, 761-773 (2007)
• M. Ala-Korpela. Critical evaluation of 1H NMR metabonomics of serum as a methodology for disease risk assessment and diagnostics. Clin Chem Lab Med. 46, 27-42 (2008)
• V.-P. Mäkinen, P. Soininen, C. Forsblom, M. Parkkonen, P. Ingman, K. Kaski, P.-H. Groop, M. Ala-Korpela. 1H NMR metabonomics approach to the disease continuum of diabetic complications and premature death. Mol Syst Biol. 4, 167 (2008)

Objective: Diabetic complications are the main cause of increased mortality in patients with type 1 diabetes [1]. Early detection of the highrisk phenotypes is challenging due to the gradual and multifactorial disease processes in the kidneys, retina and the vascular system. The goals of this study are to assess whether 1H NMR metabonomics of serum can characterize the metabolic phenotypes and how these phenotypes are associated with complications and their progression.

Materials and methods: At baseline, clinical characteristics were collected and 1H NMR data of serum were measured for 1,791 patients with type 1 diabetes. At followup (7.4 years on average), vitality status was checked from the Finnish population registry and the progression of diabetic kidney disease was determined from hospital records. The 1H NMR experiments were targeted at three molecular windows: i) a standard 1H NMR spectrum that contains pronounced signals from lipoprotein lipids and albumin, ii) a T2filtered spectrum of lowmolecularweight metabolites [2] and, for a subset of 331 patients, iii) a 1H NMR spectrum of serum lipid extracts. All data were measured at 500 MHz. The spectra were analyzed by linefitting, regression models and the selforganizing map to uncover the associations between the metabolic phenotypes, clinical characteristics and disease progression [2,3].

Results: The 1H NMR spectra of a single serum sample were able to characterize the continuous disease grading and the overlapping clinical categories at baseline, and indicated a subtle multivariate difference between microvascular complications and the metabolic syndrome. The highest 10year mortality of 51% CI95[29%,66%] for men and 36% CI95[23%,41%] for women was associated with an advanced kidney disease phenotype, as expected. The combination of the three molecular windows was able to distinguish stable patients from those that progressed to kidney disease. In particular, patients that were clinically the most uncertain (microalbuminuria) were characterized by the fatty acid composition and other metabolic traits, although the numbers were insufficient for assessing predictive performance.

Conclusions: 1H NMR metabonomics of serum offered a robust and reproducible tool for metabolic phenotyping of a large set of patients with type 1 diabetes. The combination of lipoprotein subclasses, lipid species and lowmolecularweight metabolites provided additional insight into the metabolic implications of diabetes.

• Mäkinen VP, et al. Diabetes. 2008; in press.
• Mäkinen VP, et al. Mol Syst Biol. 2008;4:167.
• AlaKorpela M. Clin Chem Lab Med. 2008;46:2742.

The non-selective nature along with metabolic specificity and various molecular windows available are advantageous characteristics of 1H NMR spectroscopy in metabonomics [1]. 1H NMR spectroscopy as a method to analyse lipoprotein subclasses has also received wide clinical interest [1–3]. A key strategic reason for this is the avoidance of their tedious physical isolation from plasma and the consequent potential for detailed studies of extensive populations. The central role of lipoprotein subclasses in the risk assessment of coronary heart disease, diabetes, and other diseases is currently wellestablished. Furthermore, the abundant information on other, potentially relevant biomolecules, makes 1H NMR spectroscopy of serum a well suited methodology for multi-parametric biochemical and clinical studies [1,4–6].

We have applied two molecular windows in metabonomic 1H NMR studies of serum [1,4–6]. The LIPO window represents the typical spectrum of serum showing broad overlapping resonances arising mainly from lipoprotein lipids [1–3]. The LMWM window takes the advantage of T2-relaxation to modify the detectable molecular information and enables improved detection of low-molecular-weight metabolites [1,2,4–6]. In addition, to obtain specific information on individual lipid molecules and their degree of (poly)(un)saturation, we have measured 1H NMR data on extracted serum lipids. Our recent application of this novel combination of three molecular windows in relation to mild cognitive impairment (MCI), a neuropsychological diagnosis with severely increased risk for Alzheimer's disease (AD), demonstrated the capability of the approach to characterise systemic metabolic phenotypes with respect to the neuropsychological and clinical diagnostics. The results underlined the association between MCI, the metabolic syndrome, obesity and diabetes [6].

We have developed a new metabonomics framework to visualize and interpret the serum data and to link the metabolic profiles to the underlying diagnostic and biochemical variables [5,6]. In general, our work demonstrates the diffuse nature of complex vascular diseases and the limitations of single 'diagnostic' biomarkers. However, it also promises costeffective solutions via high-throughput analytics and advanced computational methods [1,5,7].

The talk will focus on the methogological issues as well as on the biomedical applicability and rationale of the developed framework. Our recent data and results will be shown to demonstrate the inherent suitability of 1H NMR metabonomics to identify subtle changes in lipoprotein subclass-related metabolism together with various other metabolites of high interest in clinical medicine. Also, associations of various 1H NMR-based multi-metabolic phenotypes with clinical diagnostics (e.g., micro- and macrovascular complications) as well as mortality will be discussed in a cohort of over 4,000 patients with type 1 diabetes [5,7]. A 1H NMR metabonomics study of the systemic metabolic phenotypes that relate to MCI and thereby to high risk for AD will also be discussed [6]. All our collaborative studies indicate the value of combining all the various molecular data when applying 1H NMR metabonomics to disease risk assessment and diagnostics. Notably, use of the three molecular windows generates huge amount of metabolic data with affordable measurement time and efforts.

I am grateful to all my colleagues and co-authors that share the scientific interest and endeavor for NMR metabonomics and the multidisciplinary field of 'omics sciences. This work has been financially supported by the Academy of Finland Centre of Excellence program for 2006–2011.

• Ala-Korpela M. Potential role of body fluid 1H NMR metabonomics as a prognostic and diagnostic tool. Expert Rev Mol Diagn.2007;7:761–773.
• Ala-Korpela M. 1H NMR spectroscopy of human blood plasma. Progr Nucl Magn Reson Spectr. 1995;27:475–554.
• Ala-Korpela M, Lankinen N, Salminen A, Suna T, Soininen P, Laatikainen R, Ingman P, Jauhiainen M, Taskinen MR, Héberger K, Kaski K. The inherent accuracy of 1H NMR spectroscopy to quantify plasma lipoproteins is subclass dependent. Atherosclerosis. 2007;190:352–358.
• Mäkinen VP, Soininen P, Forsblom C, Parkkonen M, Ingman P, Kaski K, Groop PH, Ala-Korpela M. Diagnosing diabetic nephropathy by 1H NMR metabonomics of serum. Magn Reson Mater Phy. 2006;19:281–296.
• Mäkinen VP, Soininen P, Forsblom C, Parkkonen M, Ingman P, Kaski K, Groop PH, Ala-Korpela M. 1H NMR metabonomics approach to the disease continuum of diabetic complications and premature death. Mol Syst Biol. 2008;4:167.
• Tukiainen T, Tynkkynen T, Mäkinen VP, Jylänki P, Kangas A, Hokkanen J, Vehtari A, Gröhn O, Hallikainen M, Soininen H, Kivipelto M, Groop PH, Kaski K, Laatikainen R, Soininen P, Pirttilä T, Ala-Korpela M. A multi-metabolite analysis of serum by 1H NMR spectroscopy: early systemic signs of Alzheimer's disease. Submitted, 2008.
• Mäkinen VP, Forsblom C, Thorn LM, Wadén J, Gordin D, Heikkilä O, Hietala K, Kyllönen L, Kytö J, Rosengård-Bärlund M, Saraheimo M, Tolonen N, Parkkonen M, Kaski K, Ala-Korpela M, Groop PH. Metabolic phenotypes, vascular complications and premature deaths in a population of 4,197 patients with type 1 diabetes. Diabetes. In press, 2008.

Topics covered:

• Lipid molecules and lipoprotein particle structure
• Lipoprotein metabolism – physiology
• Cholesterol metabolism and atherosclerosis
• Lipoproteins and atherothrombosis – pathophysiology
• Magnetic resonance techniques and molecular characterisation of atherothrombosis
• Diabetes epidemic – micro- and macrovascular complications
• NMR metabonomics and metabolic phenotyping; Disease diagnostics, screening and prognostics – what is the difference?
• Future medicine – from treatment to prevention.

Risk assessment of diabetic complications is challenging due to the gradual disease processes and the lack of accurate metabolic phenotypes. Here the focus is on the phenotypes: to reveal the multi-variate biochemical features behind the clinical outcomes and premature deaths in patients with type 1 diabetes.

Baseline data were collected for 2176 males and 2121 females with type 1 diabetes from the Finnish Diabetic Nephropathy Study. The vitality status was obtained after an average of 6.5 years of follow-up (24589 patient years, 271 deaths). Clinical records and measurements of 19 biochemical variables (lipoprotein subfractions, apolipoproteins, serum and urine creatinine, inflammatory markers and urine albumin) were collected by standardized methods. The data were analysed by self-organizing maps (SOMs) of males and females separately.

Two sides of the complications cluster were revealed: a metabolic syndrome profile of insulin resistance, low HDL2C and abdominal obesity, and a high-adiponectin, high-LDLC, high HDL2C profile associated with microvascular complications and longer diabetes duration. An 11.4-fold [CI95%: 9.7-13.4] population adjusted risk of death in males and a 9.1-fold [6.9-11.3] risk in females was observed at the intersection of the two metabolic states, where also the prevalence of kidney disease was the highest. At sub-clinical level, this high-risk metabolic phenotype was associated with a 3.7-fold [1.3-6.0] risk in males with no kidney disease (normalbuminuria), and a 4.3-fold [0.5-7.8] risk in males with unclassified albuminuria.

The SOM enabled the dissection of albuminuria into a metabolic continuum between two characteristic phenotypes and we expect these results to open new insights to the complex mechanisms behind diabetic complications.

Objective – Subtle metabolic changes precede and accompany chronic vascular complications, which are the primary causes of premature death in diabetes. Here we assess the ability of proton NMR metabonomics of serum to characterize metabolic phenotypes, including lipoprotein subclasses, associated with vascular complications and premature death in type 1 diabetes.

Design and patients – An age- and sex-matched (against albuminuria) subset of type 1 diabetic patients from the multicentre Finnish Diabetic Nephropathy (FinnDiane) study: 613 patients, of which 250 (41%) were normoalbuminuric (AER < 20 µg/min), 139 (23%) had microalbuminuria (20 ≤ AER < 200 µg/min) and 224 (36%) had macroalbuminuria (AER ≥ 200 µg/min). Vitality status after an average of 8.2 years of follow-up (4972 patient years, 56 deaths) was used for prospective analyses.

Methods – The proton NMR experiments of serum were targeted at two different types of molecular windows – lipoprotein lipids and low-molecular-weight metabolites. These data were recorded at the physiological temperature of 310 K at 500 MHz using a double tube system facilitating absolute metabolite quantification. The lipoprotein subclasses were quantified computationally by Bayesian regression analyses specifically designed for this NMR protocol. Metabolic phenotypes were identified by self-organizing map analyses.

Results and conclusions – The analyses revealed a metabolic phenotype that was associated with a 7.8-fold increase in age and population adjusted risk of death. This high-risk phenotype shared features from diabetic kidney disease (elevated serum creatinine and urea) and from the metabolic syndrome (increased concentration of TG-rich lipoprotein particles and lactate, low HDL fractions). Our results illustrate how a single half-an-hour proton NMR protocol, that combines data from lipoprotein subclasses and low-molecular-weight metabolites, is able to identify the polydiagnostic metabolite manifold of type 1 diabetes and how its alterations translate to clinical phenotypes, clustering of micro- and macrovascular complications, and mortality during several years of follow-up. This work demonstrates the diffuse nature of complex vascular diseases and the limitations of single diagnostic biomarkers. However, it also promises cost-effective solutions through high-throughput analytics and advanced computational methods, as applied here in a case that is representative of the real clinical situation. Furthermore, our application of proton NMR metabonomics and statistical visualizations may improve the tracking of patients' progress in the diabetic disease continuum in a way not attainable by traditional approaches. Hence, it may become possible to re-route the multimetabolite path of a vulnerable patient away from adverse clinical endpoints and towards a more favourable phenotype before it is too late.

The non-selective nature along with metabolic specificity and various molecular windows available are advantageous characteristics of 1H NMR spectroscopy in metabonomics [1]. Recent results will be presented on the inherent suitability of 1H NMR spectroscopy to identify subtle changes in lipoprotein subclass-related metabolism together with several other metabolites of high interest in clinical medicine. Based on a single 1H NMR protocol, we have developed a new framework to visualise and interpret the data and to link the multi-metabolic phenotypes to the underlying diagnostic and biochemical variables [2]. Our work emphasises the diffuse nature of complex vascular diseases and the fundamental limitations of a single-biomarker approach. However, it also promises cost-effective clinical solutions via high-throughput analytics and advanced computational methods. Our studies indicate the importance of combining the macromolecular information with data from low-molecular-weight metabolites in serum for disease risk assessment and diagnostics. We also discuss how our application of 1H NMR metabonomics and statistical visualisations may improve the tracking of patients' progress in the diabetic disease continuum in a way not attainable by traditional approaches. Hence, it may become possible to re-route the multimetabolite path of a vulnerable patient away from adverse clinical endpoints and towards a more favourable phenotype before it is too late.

• M. Ala-Korpela. Expert Rev Mol Diagn. 2007;7:761.
• V.-P. Mäkinen, P. Soininen, C. Forsblom, M. Parkkonen, P. Ingman, K. Kaski, P.-H. Groop, M. Ala-Korpela. Mol Syst Biol. 2008;4:167.

Background: Lipoprotein phenotypes, beyond serum and LDL lipids, would currently be favored in the risk assessment for vascular complications. However, due to practical labor and cost issues, the Friedewald-based LDL cholesterol (C) is the prevailing measure in epidemiological studies and it also remains an essential clinical measure and treatment objective.

Methods: Artificial neural network (ANN) regression models, using serum-C, serum triglycerides (TG) and HDL-C as inputs (as in the Friedewald formula), were trained and cross-validated for various lipoprotein lipids and apolipoprotein (apo) A-I and B. Lipid measurements were available for 1,775 serum samples from which VLDL, IDL, LDL and HDL fractions were also isolated by ultracentrifugation. For HDL2-C and the apolipoproteins 343 and 247 samples were available, respectively.

Results: Good models were obtained for VLDL-TG (R for cross-validation 0.98), LDL-C (0.91), HDL2-C (0.93), apoA-I (0.92) and apoB (0.95). Only a semi-quantitative model (R = 0.77) was obtained for IDL-C. Due to the anticipated role of IDL-C for the risk of atherosclerosis, it was still pursued further in the clinical application. In comparison, the correlation between the traditional Friedewald LDL-C estimates and the measured IDL-C+LDL-C from the ultracentrifuged fractions was 0.91. The new estimates (including, e.g., HDL3-C=HDL-C–HDL2-C and apoB/apoA-I) were used to reveal the associations of multi-variate lipoprotein phenotypes with the vascular complications and premature deaths in a population-based set of 4,084 patients with type 1 diabetes. In this clinical context, the overall behavior of IDL-C versus that of LDL-C as well as that of HDL2-C compared to HDL3-C was notably different. This points to the differing (patho)physiological role of these ANN-estimates and supports their computation instead of the usual Friedewald LDL-C. The IDL-C and apoB were also among the best predictors of vascular complications and mortality.

Conclusions: The ANN-based Friedewald approach provides an opportunity to estimate rather advanced lipoprotein phenotypes with no additional cost if serum C, serum TG and HDL-C, i.e., the conventional Friedewald inputs, are available.

Sub-clinical risk assessment of diabetic kidney disease is challenging due to the gradual disease processes and the lack of accurate metabolic phenotypes. Here the focus is on the latter problem: to reveal the multi-variate biochemical features behind the clinical outcomes and premature deaths in a population-based set of Finnish patients with type 1 diabetes.

Baseline biochemical and clinical data were collected for 2173 men and 2024 women with type 1 diabetes (age of onset below 35 years, insulin treatment within a year of onset) from the Finnish Diabetic Nephropathy Study. The vitality status was obtained after an average of 6.3 years of follow-up (24,486 patient years, 295 deaths).

Clinical records and measurements of 20 biochemical variables (lipoprotein lipids, apolipoproteins, serum and urine creatinine, inflammatory markers and 24h-urine albumin) were collected by standardized questionnaires and laboratory methods. The data were analysed by a self-organizing map (SOM), and visualized separately for males and females with computational significance estimates and confidence intervals. The metabolic phenotypes revealed by the SOM were then compared against the observed all-cause mortality.

Two sides of the diabetic complications cluster were revealed: a metabolic syndrome profile of insulin resistance, low HDL2C, and abdominal obesity, and a high-adiponectin, high-LDLC, high-HDL2C profile associated with microvascular complications. A 10.1-fold [CI95%: 7.3-13.1] population adjusted risk of death in males, and a 10.8-fold [CI95%: 7.9-13.7] risk in females was observed at the intersection of the two metabolic states.

The SOM approach offers a visual way of incorporating multi-variate biochemical datasets into clinical decision making, as demonstrated in this study by the dissection of diabetic complications into a metabolic continuum between two characteristic phenotypes. We expect this type of computational medicine to be crucial not only for cultivating the scientific understanding of multi-factorial metabolic disorders, but also for enabling tailored treatments of complex diseases.

Alzheimer's disease (AD) is the most common dementia in the world. A serious health care challenge is ahead due to the increasing number of AD patients. The initiation and development of AD are poorly understood and there are no distinct biomarkers allowing for early detection and preventive treatment. Thus, the focus in AD research has shifted to elderly people with mild cognitive impairment (MCI), a condition in which cognitive skills are impaired more than typical to normal aging yet not as severely as in dementia. The MCI patients are at high risk of subsequently developing dementia. Thereby, if it were possible to identify metabolic phenotypes characteristic for MCI, there might be means to identify and treat individuals having an increased risk for AD.

Recent findings point towards a fundamental role of cholesterol and lipoproteins in AD pathophysiology. Consequently, 1H NMR metabonomics offers an appealing new approach to investigate serum metabolism in relation to MCI and AD. A novel combination of three molecular windows was applied in this study, leading to information on lipoprotein subclass profiles, various low-molecular-weight metabolites and also individual lipid molecules together with their degree of (poly)(un)saturation.

The 1H NMR data were run for a prospective set of 180 serum samples from 45 elderly individuals. Thirty percent of the samples were characterised as being related to MCI, while the rest were cognitively intact. The data were analysed by self-organising maps, a methodology facilitating a holistic and a solely data-driven overview on the metabolic phenotypes and their associations with the clinical characteristics.

The data set available was limited, yet it allowed an identification of distinct metabolic phenotypes related to MCI. Many of the known associations between vascular risk factors and MCI were also revealed by the data and the novel methodology used. It was found that low HDL cholesterol and high triglycerides, typical characteristics of the metabolic syndrome, as well as obesity and diabetes were clearly associated with MCI. The prevalence of apolipoprotein E ε4 alleles was high among the MCI patients. Notably, the associations were not inclusive and it appeared that the highest risk for the MCI was related to the clustering of unfavourable metabolic and lipid phenotypes.

The non-selective nature along with metabolic specificity and various molecular windows available are advantageous characteristics of 1H NMR spectroscopy in metabonomics [1]. 1H NMR spectroscopy as a method to analyse lipoprotein subclasses has also received wide clinical interest [1–3]. A key strategic reason for this is the avoidance of their tedious physical isolation from plasma and the consequent potential for detailed studies of extensive populations. The central role of lipoprotein subclasses in the risk assessment of coronary heart disease, diabetes, and other diseases is currently wellestablished. Furthermore, the abundant information on other, potentially relevant biomolecules, makes 1H NMR spectroscopy of serum a well suited methodology for multi-parametric biochemical and clinical studies [1,4–6].

We have recently applied two molecular windows, LIPO and LMWM as illustrated in the Figure below, in metabonomic 1H NMR studies of serum [1,4–6]. The LIPO window represents a typical spectrum of serum showing broad overlapping resonances arising mainly from lipoprotein lipids [1–3]. The LMWM window takes the advantage of T2-relaxation to modify the detectable molecular information, thus enabling improved detection of low-molecular-weight metabolites [1,2,4,5]. In addition, to obtain specific information on individual lipid molecules and their degree of saturation, we have measured 1H NMR data on extracted serum lipids [6].

Based on a single 1H NMR protocol, we have developed a new metabonomics framework to visualize and interpret the serum data and to link the metabolic profiles to the underlying diagnostic and biochemical variables [4–6]. Our work demonstrates the diffuse nature of complex vascular diseases and the limitations of single diagnostic biomarkers. However, it also promises cost-effective clinical solutions via high-throughput analytics and advanced computational methods [5,6]. During the talk recent results will be presented in order to convince the audience on the inherent suitability of 1H NMR metabonomics to identify subtle changes in lipoprotein subclass-related metabolism together with several other metabolites of high interest in clinical medicine. Also, associations of various 1H NMR-based multi-metabolic phenotypes with clinical diagnostics (e.g., micro- and macrovascular complications) as well as mortality during several years of follow-up will be discussed in a cohort of over 700 type 1 diabetic patients [5]. In addition, preliminary results are shown from a 1H NMR metabonomics study of the metabolic phenotypes that relate to cognitive impairment [6]. All these collaborative studies indicate the value of combining the macromolecular information with data from low-molecular-weight metabolites when applying 1H NMR metabonomics to disease risk assessment and diagnostics.

I am grateful to all my colleagues and co-authors that share the scientific interest and endeavor for NMR metabonomics and the multidisciplinary field of 'omics sciences. This work has been financially supported by the Academy of Finland Centre of Excellence program for 2006–2011.

• M. Ala-Korpela. Expert Rev Mol Diagn. 2007;7:761.
• M. Ala-Korpela. Progr Nucl Magn Reson Spectr. 1995;27:475.
• M. Ala-Korpela et al. Atherosclerosis. 2007;190:352.
• V.-P. Mäkinen et al. Magn Reson Mater Phy. 2006;19:281.
• V.-P. Mäkinen et al. 2007; submitted.
• T. Tukiainen et al. 2007; in preparation.

The central role of lipoprotein subclasses in the risk assessment of coronary heart disease (CHD), diabetes, and other diseases is currently well-established. 1H NMR spectroscopy as a method to analyse lipoprotein subclasses has received wide clinical interest since the experimental part is a fast, routine procedure [1,2]. It is notable that 1H NMR of serum can also provide data on other, potentially relevant biomolecules [2,3]. Thus, 1H NMR offers two concomitant approaches – to quantify individual metabolites and also to consider the spectra of all the NMR-detectable compounds as metabolic profiles that relate, e.g., to the risk of CHD. The latter holistic approach – metabonomics – suggests that we do not need to quantify each metabolite if there are means to classify the profiles in an appropriate manner. The quantitative approaches and the metabonomic methodologies are complementary. However, in extensive clinical studies aiming for risk profiling, the metabonomic approach may well be more appropriate to start [4]. We have recently applied two molecular windows, LIPO and LMWM, in metabonomic 1H NMR studies of serum [3]. The LIPO window represents a typical spectrum of serum showing broad overlapping resonances arising mainly from lipoprotein lipids [2]. The LMWM window takes the advantage of T2-relaxation to modify the detectable molecular information, thus enabling improved detection of low-molecular-weight metabolites [2,3]. In the talk, results that demonstrate the inherent suitability of 1H NMR metabonomics to identify subtle changes in lipoprotein subclass-related metabolism are shown [1,4]. Also, we illustrate how the multimetabolite data via 1H NMR reveal the continuum of diabetic complications in a cohort of over 700 type 1 diabetic patients [5]. We also show preliminary results from a 1H NMR metabonomics study of the metabolic characteristics in cognitive impairment [6]. This study also includes a third molecular window, namely data on extracted serum lipids. All these collaborative studies show the value of combining the macromolecular information with data from small molecules in clinical 1H NMR metabonomics.

• M. Ala-Korpela et al. Atherosclerosis. 2007;190:352.
• M. Ala-Korpela. Progr Nucl Magn Reson Spectr. 1995;27:475.
• V.-P. Mäkinen et al. Magn Reson Mater Phy. 2006;19:281.
• T. Suna et al. NMR Biomed. 2007 Nov;20(7):658-72.
• V.-P. Mäkinen et al. 2007; submitted.
• T. Tukiainen et al. 2007; in preparation.

A key challenge in metabonomics is to uncover quantitative associations between multidimensional spectroscopic data and biochemical measures used for disease risk assessment and diagnostics. Here we focus on clinically relevant estimation of lipoprotein subclasses by 1H NMR spectroscopy of serum. Some subclasses are more important than others with respect to the risk for atherosclerosis, but the current clinical risk assessment methodology cannot take this properly into account. 1H NMR enables a fast measurement of the lipoprotein profile, and several computational methods have already been developed for this purpose [1]. We have recently presented a novel Bayesian approach to quantify clinical variables and to determine their spectroscopic counterparts in 1H NMR data [2]. The analysis focused on lipid concentrations of major lipoprotein fractions, namely VLDL, IDL, LDL and HDL to test the automated Markov chain Monte Carlo (MCMC) Bayesian inference with a wellknown biochemical background and spectroscopic characteristics. The results illustrated a high-quality quantification ability of the presented Bayesian approach [2]. In this work we study if the MCMC Bayesian inference could also be used to quantify the lipoprotein subclasses from the 1H NMR spectra of serum. Using experimentally derived model signals for VLDL1, VLDL2, IDL, LDL1, LDL2, LDL3, HDL2b, HDL2a, HDL3a, HDL3b and HDL3c we simulated a representative set of spectra that contained extensive variation in the lipoprotein subclass profiles. The simulated spectra also included individual variations in the lipoprotein subclass signals. Our preliminary Bayesian MCMC analysis results for the quantitative performance in the case of all the abovementioned 11 lipoprotein subclasses are very good for the simulated data set. This suggests that the Bayesian approach, while computationally demanding, could provide a framework for quantitative modelling of metabonomic NMR data. In line with our earlier analysis [1], the quantification accuracy seems to be subclass dependent. Even though the preliminary results show very good generalisation abilities with the simulated data, it is possible that the simulation process itself may not optimally represent the inter-subject variance of the spectral features for a realistic population. Thus, the results must be verified with extensive real data; such preliminary data for an extensive set of serum samples with also NMR-independent biochemical measures for the lipoprotein subclasses will be presented together with critical considerations of the pros and cons of the methodology.

• M. Ala-Korpela, et al. Atherosclerosis. 2007, 190, 352–358.
• Vehtari A, et al. BMC Bioinformatics. 2007, 8(S2), S8.

Macrovascular disease in connection to diabetic nephropathy accounts for most of the premature deaths in type 1 diabetic (T1D) patients. Metabolic syndrome (MetS) is common in the subset that develop these serious complications, which suggests that the risk factors are similar to those in the general population. The definitions of MetS or other risk criteria, however, may not be well suited for T1D nor be accurate enough to detect subtle metabolic changes. For this reason, we have measured 1H NMR spectra of serum for a subset of T1D patients (n = 613) from the FinnDiane multi-center study. Our goal is to determine the multivariate metabolic characteristics of T1D that will serve as the baseline for the prospective phase of FinnDiane. Since the measurement data are very extensive and complex, we used self-organizing maps to investigate the statistical patterns of metabolism that incorporate the serum lipid profile, creatinine and albumin among others. Also, we did classification analysis to assess the descriptive power of 1H NMR spectra in this setting, and found strong links between specific spectral patterns, metabolic syndrome and diabetic nephropathy (see figure below; the map is colored according to the percentage of diabetic kidney disease within a given map region). Based on this cross-sectional study, 1H NMR metabonomics can produce a detailed and comprehensive picture of metabolic risk factors for cardiovascular disease – an advantage that is likely to have great prognostic value in the detection and treatment of diabetic complications.

Metabonomics is a complementary post-genomic technology that provides a holistic view of time-related metabolic responses of biological systems to (patho)physiological stimuli or genetic modifications. 1H NMR spectroscopy is the most common single methodology applied for the assessment of metabolic phenotypes. Recent studies have shown promise that with this technique metabolic fingerprints for different biochemical conditions can be detected. Here we present a novel holistic application to study cognitive decline by 1H NMR.

We applied three molecular windows; (i) a LIPO window (dominated by lipoprotein lipids), (ii) a LMWM window (low-molecular-weight metabolites), and (iii) serum lipid extracts. Together these data provide a wealth of quantitative biochemical information difficult to obtain by conventional methods. These data include the lipoprotein subclass profile, a bunch of other metabolites and individual lipids together with their degree of (un)saturation. In general, the analysis of these overwhelming data poses a problem but we present a group correlation spectroscopy approach which is a visual and straightforward way to study various metabolic associations within and between the different molecular windows (see the adjacent figure).

Here we present results from a pilot study of cognitive decline by 1H NMR metabonomics of serum and the application of group correlation spectroscopy to reveal the intra- and intermolecular dependencies within the data of 45 elderly subjects. During the follow-up of six years 17 patients showed symptoms of mild cognitive impairment while 28 remained cognitively normal. Serum samples and 1H NMR data via the three molecular windows were collected from each subject at four time points (0, 1, 3, and 6 years). This unique data set provides means to assess, e.g., as illustrated in the figure above, the role of lipoprotein profile in relation to the serum lipid content and (poly)(un)saturation in cognitive decline.

Identification of physiological conditions relating to increased risk of atherosclerosis would be advantageous to facilitate individual primary prevention. Metabolic syndrome, with characteristic lipoprotein subclasses profiles, is such a condition with an increasing prevalence in the general population. 1H NMR metabonomics offers an alternative approach to identify lipoprotein subclass profiles and can also provide a holistic overview of other serum metabolites. We have studied whether this spectroscopic approach would enable characterisation of lipoprotein subclass-related metabolism in a clinically relevant context. By using experimentally derived model signals for VLDL1, VLDL2, IDL, LDL1, LDL2, LDL3, HDL2b, HDL2a, HDL3a, HDL3b and HDL3c, two biochemically characteristic categories of spectra were simulated representing lipoprotein subclass profiles for a normolipidaemic and a metabolic syndrome status. We also generated a set of spectra representing a metabolic pathway between these two categories. The analysis of the 1H NMR spectra clearly identified the lipoprotein subclass profiles and their changes. The findings are supported by comparable analyses for a representative set of experimental 1H NMR spectra of 69 serum samples with a wide range of lipoprotein lipid concentrations. The results demonstrate the inherent suitability of 1H NMR metabonomics to identify subtle changes in lipoprotein subclass-related metabolism. Accordingly, 1H NMR can be seen as a new methodology for screening individuals at high risk for atherosclerosis and also being potentially useful in prospective assessment of individual 'health paths' in the borderline of, for example, normal lipoprotein metabolism and the metabolic syndrome.

Metabolic sydrome (MetS) is associated with disruptions in glucose metabolism and is a frequent finding among type 1 diabetic (T1DM) patients. The current definitions of MetS are designed for nondiabetic and T2DM patients and may not be well suited to characterise the T1DM population. For this reason, our aim is to study the overall metabolite composition of T1DM patients in relation to diabetic complications such as kidney disease. We measured 1H NMR spectra of serum from a subset of patients (n = 613) from the FinnDiane study, and applied selforganizing maps to study the connections between the spectra, traditional risk factors and clinical endpoints. As expected, we found strong associations between kidney indicators (creatinine and albumin) and MetS indicators like triglycerides and other lipids, but we also detected specific metabolite associations between glucose control and diabetic complications from the spectra. In addition, the metabonomic analysis provided a new comprehensive view of the clustering of these metabolic risk factors (Figure) that was not attainable by traditional markers alone.

Metabolic syndrome (MetS) is associated with disruptions in glucose metabolism and is a frequent finding amont type 1 diabetic (T1DM) patients. The current definitions of MetS are designed for non-diabetic and T2DM patients and may not be well suited to characterise the T1DM population. For this reason, our aim is to study the overall metabolite composition of T1DM patients in relation to diabetic complications such as kidney disease. we measured 1H NMR spectra of serum from a subset of patients (n = 463) rom the FinnDiane study, and applied self-organizing maps to study the connections between kidney indicators (creatinine and albumin) and MetS indicators like triglycerides and other lipids, but we also detected specific metabolite associations between glucose control and diabetic complications from the spectra. In addition, the metabonomic analysis provided a new comprehensive view of the clustering of these metabolic risk factors (Figure) that was not attainable by traditional markers alone.

Purpose/Introduction: The most severe complication of type 1 diabetes (T1DM) is diabetic nephropathy. It is associated with a high risk of cardiovascular complications and premature death and requires early detection to be efficiently treated. The clinical practice to diagnose diabetic nephropathy is also a non-optimal and tedious set-up being based on albumin excretion rate in multiple over-night or 24h-urine samples. Conversely, in this study, these independent diagnostic data are used to provide a realistic testing case for applying 1H NMR metabonomics of serum in a diagnostic fashion.

Subjects and Methods: 182 T1DM and 21 non-diabetic (non-T1DM) individuals were studied. The 1H NMR of serum at 500 MHz was targeted at two molecular windows - lipoprotein lipids and low-molecular-weight metabolites.

Results: T1DM and non-T1DM individuals were exclusively separated by 1H NMR. For diabetic nephropathy diagnosis in the T1DM patients 1H NMR data (and clinical biochemistry data) gave a sensitivity of 87.1% (83.9%) and a specificity of 87.7% (95.9%). The predictive values of positive and negative tests were 89.0% (95.5%) and 83.6% (79.2%), respectively.

Discussion/Conclusion: 1H NMR metabonomics clearly distinguishes metabolic characteristics of T1DM and appears approximately as good means to diagnose diabetic nephropathy from serum as an advanced set of biochemical variables.

Purpose/Introduction: 1H NMR metabonomics plays a key role in evaluating metabolic phenotypes for disease risk assessment and clinical diagnostics. Two typical simplifications often made in handling the data are the use of 'bins', i.e., summed spectral intensities over fairly broad regions, and the application of rather uninformative chemometric methods, such as principal component analysis. This often hampers the biochemical rationale of the results. Here we will illustrate and apply a straightforward and visual approach facilitating the use of full spectral resolution and allowing detection of intra- and inter-molecular dependencies in metabonomic data sets.

Subjects and Methods: The approach involves calculation of the (correlation and) covariance between all the data points in a data set and displaying the information in the form of colour-coded pseudo-2D NMR spectra as illustrated in the figure: The simulated spectra representing the lipoprotein part of human serum were constructed as a sum of experimentally obtained model signals for eleven lipoprotein subclasses - VLDL1, VLDL2, IDL, LDL1, LDL2, LDL3, HDL2b, HDL2a, HDL3a, HDL3b and HDL3c. By varying the levels of these subclasses, two clinically distinct categories of sum spectra were generated: a group of spectra corresponding to a normal lipoprotein profile (N) and another group corresponding to a lipoprotein profile in metabolic syndrome (MS), a common disturbance of lipoprotein metabolism related to increased risk for coronary heart disease. Internal heterogeneity was induced to these two categories resulting in 729 spectra for each category. In addition, a metabolic pathway (MP) consisting of 501 spectra combining linearly the N and MS categories was generated. Also, over 500 serum samples from two clinically distinct populations, type 1 diabetic and alcoholic patients, we studied by 1H NMR spectroscopy at 500.13 MHz.

Results: For example, the associated changes in VLDL (increase) and HDL2 (decrease) along the MP from N towards the MS were, even in the case of heavy signal overlap, immediately evident from the covariance pattern of the pseudo-2D spectrum; the statistical variation within N and MS groups produces completely different patterns. The data for the two patient groups revealed dissimilar metabolic relation of glucose and lipoproteins.

Discussion/Conclusion: The simulated data reliably illustrated the performance of the sample correlation spectroscopy and covariance patterns in the case of 1H NMR spectra of serum and thereby gave confidence for the new metabolic findings in the patient groups.

Abstract

Measuring metabolites is not new. For decades, clinicians have monitored chemistries in various body fluids, e.g., using glucose to track diabetes and cholesterol to screen heart disease. What is new in the metabo*omics approaches is that we are now attempting to collect an unbiased sample of metabolites that can serve as a snapshot of an organism's (patho)physiology. As an ultimate goal we wish to distinguish between an individual who is healthy and someone who has (diagnosis) – or might develop (risk ssessment) – a disease.

1. Introduction

Mass and nuclear magnetic resonance (NMR) have become the two key spectroscopies applied in metabo*omics applications. An appealing feature of NMR for metabonomic applications is its specific yet non-selective nature: using 1H NMR spectroscopy one can efficiently obtain information on a large number of metabolites in biological fluids like human serum [1]. Recently, a call for applying 1H NMR metabonomics to facilitate disease risk assessment and clinical diagnostics has also emerged [2-4]. A key issue in bringing metabonomics for clinical use will be to bridge the gap between biochemistry – as revealed by 1H NMR spectroscopy – and the relevant measures of current clinical practice.

2. Molecular Windows

As illustrated in Fig. 1 the 1H NMR experiments of serum samples can be targeted at two different molecular windows – lipoprotein lipids (LIPO) and lowmolecular-weight metabolites (LMWM). The assignments for the LIPO window resonances refer to fatty acids in triglycerides, cholesterol compounds and phospholipids in various lipoprotein particles, the cholesterol backbone –C(18)H3 and the –N(CH3)3 groups of surface phospholipids. The LMWM resonances marked gp are from glycoproteins. In a 1H NMR spectrum, one metabolite can manifest several peaks, and the signal intensities are both biochemically and (patho)physiologically related. Furthermore, the data sets are extensive but redundant: one measurement can yield tens of thousands of data points, but the effective dimensionality is much less due to a smaller number of NMR-visible compounds. There is also heavy overlap of metabolite resonances [1,5]. For these reasons there are also methodological challenges in trying to associate 1H NMR metabonomics data to diagnostically relevant biochemical variables. Two typical simplifications often made in handling the data are the use of 'bins', i.e., summed spectral intensities over fairly broad regions, and the application of rather uninformative chemometric methods, such as principal component analysis. This often hampers the biochemical rationale of the results. Here we will illustrate and apply a straightforward and visual approach facilitating the use of full spectral resolution and allowing detection of intra- and inter-molecular dependencies in metabonomic data sets. The approach involves calculation of the correlation and covariance between all the data points in a data set and displaying the information in the form of colour-coded pseudo-2D NMR spectra.

4. Diagnostic Role in Diabetes

Recent literature [3,4] as well as our own work [5,6] point towards a key role for 1H NMR spectroscopy in evaluating metabolic phenotypes for disease risk assessment and diagnostics. From our recent application [6] we will show how 1H NMR metabonomics data – obtained in two molecular windows in a set of 182 type 1 diabetic patients – can be associated with key clinical measures by the aid of statistically significant correlation patterns. Furthermore, using various regression analyses, we will illustrate the quantitative nature of the metabolite information present in a 1H NMR metabonomic data set of serum, not only for the well established case of lipoprotein lipids [1,5], but also for apolipoprotein components and for low-molecularweight metabolites [6].

5. Fuzzy Pathophysiology

In many disease processes the biological heterogeneity as well as the potentially slow development and progression of pathological conditions make the borderline between 'healthy' and 'diseased' fuzzy (Fig. 2). Atherosclerosis is also a diffuse systemic disease, characterised by the local build-up of lipid-rich plaques within the walls of large arteries. The atherothrombotic processes are multigenetic, being influenced also by dietary and environmental components, and are apparent as early as the second decade in life with an increased incidence in the elderly. Atherothrombosis involves inflammatory processes with an array of metabolic, molecular and cellular manifestations in tissues, e.g., those depicted within the arterial wall. A varying degree of these intimal processes are reflected by the biochemistry of serum. One option to approach the problem, as studied and discussed in this presentation, is 1H NMR metabonomics of serum equipped with a chemometric classifier, e.g., a self-organising map (SOM). On the left in Fig. 2 a hypothetical SOM is shown together with four overlapping clusters that are thought to represent the metabolic changes in the arterial intima. While definite classification as 'healthy' and 'diseased' may not be available by nature, the metabonomics approach with a holistic look at the multidimensional metabolic changes may prove useful in the assessment and follow up of individual 'health path' (represented by the light green line within the SOM) alongside the interplay between metabolic pathways and their consequences.

6. Conclusions

Development of metabonomic approaches, capable to visualise and interpret multidimensional metabolic influences rather than to try to find 'complete' classifications of health' and 'disease', is in our opinion a key for the favourable reception of 1H NMR metabonomics in the clinical field. Generally, the high analytical power of the combined molecular windows gives additional confidence for the metabonomics approach and suggests realistic potential for the usage of 1H NMR metabonomics also as an aid for disease diagnostics. Analysis approaches that indicate the biochemical rationale for the diagnostic outcome make the NMR metabonomics approach much more easily acceptable for the clinical arena as it would be without clear emphasis on the molecular biochemistry. The new analysis approaches and biomedical findings that will be presented are expected to stimulate discussions on the diagnostic potential of 1H NMR spectroscopy and the important role of new data analysis methods in paving the way for 1H NMR metabonomics into clinics.

• M. Ala-Korpela (1995) 1H NMR spectroscopy of human blood plasma. Progr Nucl Magn Reson Spectr 27:475-554.
• J. Griffin (2003) Metabonomics: NMR spectroscopy and pattern recognition analysis of body fluids and tissues for characterisation of xenobiotic toxicity and disease diagnosis. Curr Opin Chem Biol 7:648–54.
• T. Clayton, J. Lindon, O. Cloarec, H. Antti, C. Charuel, G. Hanton, J. Provost, J. Le Net, D. Baker, R. J. Walley, J. R. Everett, J. K. Nicholson (2006) Pharmaco-metabonomic phenotyping and personalized drug treatment. Nature 440:1073-7.
• H. L. Kirschenlohr, J. Griffin, S. C. Clarke, R. Rhydwen, A. A. Grace, P. M. Schofield, K. M. Brindle, J. C. Metcalfe (2006) Proton NMR analysis of plasma is a weak predictor of coronary artery disease. Nat Med 12:705-10.
• M. Ala-Korpela, N. Lankinen, A. Salminen, T. Suna, P. Soininen, R. Laatikainen, P. Ingman, M. Jauhiainen, M.-R. Taskinen, K. Héberger, K. Kaski (2006) The inherent accuracy of 1H NMR spectroscopy to quantify plasma lipoproteins is subclass dependent. Atherosclerosis 190(2):352-8.
• V.-P. Mäkinen, P. Soininen, C. Forsblom, M. Parkkonen, P. Ingman, K. Kaski, P.-H. Groop, M. Ala-Korpela (2006) Diagnosing diabetic nephropathy by 1H NMR metabonomics of serum. MAGMA. 19(6):281-96.