Metabolomica: Introduzione e review di alcune applicazioni in ambito clinico. Luigi Atzori (Università di Cagliari)

1. Metabolomica:
Introduzione e review di alcune applicazioni in ambito
clinico
Seminari CRS4
Workshop di Disseminazione
Luigi Atzori MD, PhD
Department of Biomedical Sciences
Clinical Metabolomics Unit
University of Cagliari
latzori@unica.it
03/06/2015

2. •What is metabolomics?
03/06/2015
•Why Metabolomics?
•Which tools?
•NMR or MS
•Multivariate analysis

3. What is metabolomics?
03/06/2015

4. The suffix “-ome” or “-omics” is often added
to an area of human biology, conveying the
impression that the field is supported by
hard science.
03/06/2015
-Omics

5. Allergenome; Triaolome, Connectome, Cytome,
Editome, Embryome, Envirome, Epigenome,
Exoposome, Exome, Foodome, Genome, Glycome,
Interferome, Interactime, Ionome, Kinome, lipidome,
Metabolome, Metagenome, Metallome, Obesidome,
Organome, Pharmacogenome, Phenome, Physiome,
Proteome, Regulome, Secretome. Transcriptome,
Toponome.
-omics topics in biology

6. Hot or not!?!?
• Established
– Genome, transcriptome, proteome,
metabolome
• Emerging
– Variome,epigenome, interactome, fluxome
• Aspiring
– Phenome, regulome, integrome, omnisciome
Nature (2013) 494:416-19
03/06/2015

7. 03/06/2015
•Economics 293,000,000
•Genomics: 27,000,000
•Proteomics 8,000,000
•Metabolomics 1,000,000
Google

8. Metabolomics-Pubmed
03/06/2015
0
200
400
600
800
1000
1200
1400
1600
1800
2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014
Years
Counts

9. Metabonomics
“…measurement of the dynamic
multiparametric metabolic response of
living systems to pathophysiological
stimuli or genetic modification…”
Nicholson et al., 1999
Metabolomics
“...the complete set of metabolites/low-
molecular-weight intermediates, which are
context dependent, varying according to the
physiology, developmental or pathological
state of the cell, tissue, organ or
organism…”
Oliver, 2002
03/06/2015

10. Metabolomics: quantitative measurement of
dynamic metabolic changes of living systems in
response to genetic modifications or
physiological stimuli, including nutrients and
drugs.
03/06/2015

11. 03/06/2015
Why Metabolomics

12. Metabolomics?
The new Clinical Chemistry
03/06/2015

13. • Discover new disease biomarkers for screening and
therapy progression
– A small short-list of metabolites can indicate an early
disease stage or predict a therapy efficiency (a priori
process)
• Associate metabolites (functions) with transcripts
(genes)
– Metabolites are downstream results of gene
expression and can be associated to
physiopathological mechanisms (long list of
metabolites)(post hoc process)
03/06/2015

14. Applications in the clinic
•Basic physiology and biochemistry
•Human disease
•Diagnosis of disease states
•Sub-classification of disease
•Tracking disease progression
•Measuring therapeutic or adverse response to
treatment
•Toxicology studies
•Selection of biomarkers
03/06/2015

15. Which tools?
03/06/2015

16. 03/06/2015
To think about….

17. Chemical properties metabolome
• Hydrophilic/hydrophobic
• Volatility
• Chemical reactivity
• Concentration
Select appropriate method
03/06/2015

18. • Metabolism is in constant flux
• Metabolomic experiment: a snapshot of
the metabolome
• The snapshot should represents the
metabolome at the sampling moment
03/06/2015

19. Common analytical techniques applied to metabolomics
Abbreviation Technique Relevant
GC-MS Gas chromatography mass spectrometry
GCxGC-MS 2 dimensional GC coupled to MS
LC-EC Liquid chromatography using an electrochemical array
HPLC-MS High performance LC-MS
UPLC-MS Ultra performance LC-MS
HILIC Hydrophobic interaction chromatography
CE-MS Capillary electrophoresis-MS
NMR Nuclear magnetic resonance
LC-NMR LC coupled to NMR
FT-ICR-MS Fourier transform ion cyclotron resonance MS
03/06/2015

20. Rachel Ruysch
03/06/2015

21. Comparing the relative sensitivities of various
metabolomic tools
03/06/2015

22. Analysis:
Open or closed
03/06/2015
Open analysis
•An analysis of the total detectable content of the sample
(e.g. an NMR spectrum of plasma)
•Primarily used for the detection of novel entities
Closed analysis
•An analysis focused onto a specific molecule or molecules
•Used for the measurements of known variables for a
model

23. 03/06/2015

24. • The 1H NMR-based metabolomic approach is usually rapid and
reproducible and can potentially provide large data sets that turn out to be
suitable for statistical interpretation.
• This approach, in particular, opens the possibility of using NMR spectral
data for the classification of samples without the use of chemical
information, allowing an unbiased chemically comprehensive comparison
to be made among different sample.
03/06/2015

25. In general, NMR Spectra do not show a single peak for
each functional group but show split peaks or ‘multiplets’
(singlets, doublets, triplets, etc.)
03/06/2015

26. 03/06/2015

27. 03/06/2015

28. NMR
Reductionistic Holistic
 Structure
 Interactions
 Functioning
mechanisms of
separated
elements
Metabolic
networks
Interconnection
among metabolic
processes
Metabolic
trajectories
03/06/2015

29. •Why MVA in omics science?
Multivariate Analysis in omics-
sciences
03/06/2015

30. •To obtain a holistic description of the systems under investigation.
• Hidden information can be extracted from large and noisy data sets.
• Model interpretation can be obtained by plots.
• MVA is suitable for hypothesis free approaches: it can be the starting point to build new
hypothesis to test
Why MVA in omics science?
03/06/2015

31. 03/06/2015
North Star
Ursa maior
Milky Way
Metabolite targeted analysis
Metabolic profiling
Metabolic fingerprinting

32. An introduction to multidimensional space
03/06/2015

33. 03/06/2015

34. "Science is facts; just as houses are made of stones, so is
science made of facts; but a pile of stones is not a house
and a collection of facts is not necessarily science
(Henry Poincaré)
03/06/2015

35. 03/06/2015
Is One Class?

36. Pattern recognition
•The analysis of a large number of biological samples
by any technique will usually produce an equally large
number of extremely complex datasets.
•This type of data consists of the measurements of a
range of metabolites (variables) for a number of
individuals (observations) and the identification and
quantification of analytes from raw data is often very
difficult.
03/06/2015

37. 03/06/2015
Important information is therefore more
likely to be found in correlation patterns
as opposed to individual signals.

38. It is very important
1) to include adequate samples sizes
without confounding variables, to avoid
excessive false discovery rate due to multiple
hypothesis testing
2) to use appropriate control
3) to exclude overfitting (sovradattamento)
(generally caused by the failure to perform
adequate validation and cross-validation).
Many studies fail to take these into account
03/06/2015

39. Modern analytical technologies allow for the identification
of patterns that confer significantly more information than
the measurement of a single parameter, much as a bar
code contains more information than a single number.
03/06/2015

40. • A basic tenant of these techniques is to calculate
a smaller number of factors which account for
the same amount of variation present in the
larger dataset.
• This reduces the dimensionality while minimising
loss of information.
03/06/2015

41. • Unsupervised techniques
• Require no information about class membership.
• Just look for inherent variation in the dataset.
Principal Components Analysis.
Hierarchical Cluster Analysis.
03/06/2015

42. Supervised techniques
Can correlate external variables (e.g. healthy or
not, age, etc) with the data.
– PLS partial least squares, PLS Discriminant
Analysis (PLS-DA), (Orthogonal)O-PLS-DA.
– Neural networks.
03/06/2015

43. Validation
• A common problem is the overfitting of data because there more
variables than there are samples when performing the statistical
analyses. The most common and easiest way to validate is cross-
validation in which the model is validated with the current data,
such as a leave-one-out method. While this validation is easy, it
often is not sufficient, especially when the model is to be used for
diagnostic purposes.
• The better validation option is to use an external dataset. This
new dataset will offer a more informative indication on how well
the model works.
• Using sensitivity and specificity allow for better comparison
between methods.
03/06/2015

44. Before starting.....
• Wide variety of techniques to choose from
• Be sure to:
– Pick the right one for your data
– Validate properly

45. Conclusions
Metabolomics represent a paradigm shift in metabolic research,
away from approaches that focus on a limited number of reactions or
single pathways, to approaches that attempt to capture the complexity
of metabolic networks.
It is reasonable to expect that the metabolomics approach,
together with functional genetics and proteomics, will have substantial
impact in clinical (personalized medicine) and environmental studies.
In addition to reducing times and costs of research and
experimentation with new drugs, metabolomics may predict new
indications for drugs already in production based on the individual
metabolic profile.
Finally, metabolomics is hypothesis-generating rather than
hypothesis-based. Therefore, one has to be really open-minded
about the results obtained.
03/06/2015

46. 03/06/2015
“It is much more important to know what sort of
patients has a disease, that what sort of disease
a patient has”
Sir William Osler

47. 03/06/2015
Thanks!!!
For Metabolomics:
Federica Murgia, Simone Poddighe,
Milena Lussu,, Cristina Piras, Maria Laura Santoru,
Lisa Marras, Sonia Liggi
University of Cagliari