Can metabolic profiling provide a new description of osteoarthritis and enable a personalised medicine approach?

Abstract

Osteoarthritis (OA) is a multifactorial disease contributing to significant disability and economic burden in Western populations. The aetiology of OA remains poorly understood, but is thought to involve genetic, mechanical and environmental factors. Currently, the diagnosis of OA relies predominantly on clinical assessment and plain radiographic changes long after the disease has been initiated. Recent advances suggest that there are changes in joint fluid metabolites that are associated with OA development. If this is the case, biochemical and metabolic biomarkers of OA could help determine prognosis, monitor disease progression and identify potential therapeutic targets. Moreover, for focussed management and personalised medicine, novel biomarkers could sub-stratify patients into OA phenotypes, differentiating metabolic OA from post-traumatic, age-related and genetic OA. To date, OA biomarkers have concentrated on cytokine action and protein signalling with some progress. However, these remain to be adopted into routine clinical practice. In this review, we outline the emerging metabolic links to OA pathogenesis and how an elucidation of the metabolic changes in this condition may provide future, more descriptive biomarkers to differentiate OA subtypes.

Full text

PERSPECTIVES IN RHEUMATOLOGY
Can metabolic profiling provide a new description of osteoarthritis
and enable a personalised medicine approach?
M. K. J. Jaggard
1,2
&C. L. Boulangé
2,3
&G. Graça
2
&U. Vaghela
4
&P. Akhbari
1,2
&R. Bhattacharya
1
&
H. R. T. Williams
2,5,6
&J. C. Lindon
2
&C. M. Gupte
1,6,7
Received: 18 February 2020 /Revised: 30 March 2020 /Accepted: 16 April 2020
#The Author(s) 2020
Abstract
Osteoarthritis (OA) is a multifactorial disease contributing to significant disability and economic burden in Western populations.
The aetiology of OA remains poorly understood, but is thought to involve genetic, mechanical and environmental factors. Currently,
the diagnosis of OA relies predominantly on clinical assessment and plain radiographic changes long after the disease has been
initiated. Recent advances suggest that there are changes in joint fluid metabolites that are associated with OA development. If this is
the case, biochemical and metabolic biomarkers of OA could help determine prognosis, monitor disease progression and identify
potential therapeutic targets. Moreover, for focussed management and personalised medicine, novel biomarkers could sub-stratify
patients into OA phenotypes, differentiating metabolic OA from post-traumatic, age-related and genetic OA. To date, OA bio-
markers have concentrated on cytokine action and protein signalling with some progress. However, these remain to be adopted into
routine clinical practice. In this review, we outline the emerging metabolic links to OA pathogenesis and how an elucidation of the
metabolic changes in this condition may provide future, more descriptive biomarkers to differentiate OA subtypes.
Keywords Metabolic profiling .Osteoarthritis .Personalised medicine .Metabonomics .Metabolomics
Metabolic profiling (also known as metabolomics,
metabonomics and metabolic phenotyping) is emerging as a
strategy to provide biomarker diagnostics capable of disease
prediction, monitoring and early detection. It is defined as “the
quantitative measurement of the dynamic multiparametric
response of a living system to pathophysiological stimuli or
genetic modification”[1]. Metabolites are the end points of
molecular biology and variations in their profile reflect a re-
sponse to the patient’s disease, environment, diet and lifestyle.
Metabolic profiling has already established itself as a robust
technique in oncology, epidemiology, gastrointestinal disease,
metabolic syndrome (MeS), diabetes and cardiovascular dis-
ease [2–8]. Its scope for impact on clinical practice is exem-
plified by studies in which biomarker analysis can enhance the
prediction of short term all-cause mortality, the profiling of
chronic kidney disease and assessment of cardiovascular risk
[9–11].
We can derive comprehensive descriptions of the thou-
sands of metabolites present in a biofluid or tissue through
complex analytical methods including but not exclusive to
the following: nuclear magnetic resonance (NMR) spectrom-
etry, liquid chromatography mass spectrometry (LC-MS), gas
chromatography mass spectrometry (GC-MS) and capillary
electrophoresis mass spectrometry (CE-MS) (Fig. 1).The
granular metabolic “fingerprint”, unique to specific disease
states, can be compared with other disease subgroups or
non-disease controls. Consequently, the technique can poten-
tially delineate OA disease traits, symptoms, outcomes or
*U. Vaghela
uv14@ic.ac.uk
1
Department of Orthopaedics & Trauma, Imperial College Healthcare
NHS Trust, London, UK
2
Department of Metabolism, Digestion and Reproduction, Imperial
College London, London, UK
3
Nestle Research Centre, Lausanne, Switzerland
4
School of Medicine, Imperial College London, South Kensington,
London SW7 2AZ, UK
5
Department of Gastroenterology, Imperial College Healthcare NHS
Trust, London, UK
6
NIHR Imperial Biomedical Research Centre, Imperial College
Healthcare NHS Trust, London, UK
7
Department of Surgery and Cancer, Imperial College London,
London, UK
Clinical Rheumatology
https://doi.org/10.1007/s10067-020-05106-3

Fig. 1 The step-wise process of metabolic profiling
Clin Rheumatol

treatment response. Analysis can be “untargeted”, where anal-
ysis does not target specific metabolites, or “targeted”,tofo-
cus on metabolite groups with associations of interests [1].
Nevertheless, interpreting metabolic data is challenging due
to the significant number of variables and confounding fac-
tors. This is where computational multi-variate statistics can
enable data interpretation and identify representative
metabolites.
How does metabolic profiling differ
from the other “big data”research methods?
The advantage of metabolic profiling is the “top down”rep-
resentation of a disease phenotype, encompassing the genetics
of disease in addition to environmental, lifestyle and dietary
factors [12].
Pathology in biological systems can be classified at the
genome (genomic), transcriptome (transcriptomic), proteome
(proteomic) and, ultimately, metabolome level (Fig. 2).
Notably, these entities are not mutually exclusive, with
cross-interactivity between the metabolome and other repre-
sentations of the phenome [13]. Inter alia, metabolic profiling
provides insights into the final stage of disease expression.
The strength of metabolic data is in the patterns and relative
abundance of small molecules.
Traditionally, clinicians have focused upon genomic and
proteomic disease descriptions. However, pathogenesis or
treatment response is modulated by both exogenous and host
factors: immunosuppression, exercise, toxins, radiation, con-
current disease, diet and gut microbiome. Under circum-
stances where these factors have a substantial role in disease
evolution, metabolic profiling, and its ability to account for
this, provides a more attractive disease description.
Search methods
In order to understand and research the metabolic influences
of OA, the following search of the EMBASE and MEDLINE
databases was performed:
1. “osteoarthritis”AND (“metabolic”OR “metabonomic”
OR “metabolomic”OR “metabolism”)
2. (“synovial fluid”OR “cartilage”OR “synovium”OR “se-
rum”OR “plasma”OR “urine”)AND(“NMR”or “Mass
Spectrometry”)
A total of 8851 abstracts were identified and 28 reports
contained suitable metabolic data.
What is metabolic syndrome and how does it
link to osteoarthritis?
MeS has a plethora of definitions and is associated with an
increased risk of cardiovascular disease. Common to all defi-
nitions are insulin resistance, visceral obesity, atherogenic
dyslipidaemia and hypertension [13]. In instituting the term
“metabolic OA”, the link with metabolic syndrome has been
pursued due to the shared mechanism of inflammation, oxida-
tive stress, common metabolites and endothelial dysfunction
in their aetiologies [14]. Therefore, metabolic syndrome un-
derpins the hypotheses of OA metabolic description, having a
Fig. 2 The methods of
representing a biosystem using
the “omic”techniques
Clin Rheumatol

large impact upon study design and execution of many
studies.
A twofold increase in the risk of developing MeS has been
demonstrated in patients suffering from OA [15].
Cardiovascular disease risk is elevated in female OA patients
[16,17]. Diabetes occurs more commonly in patients with
radiological but not necessarily symptomatic osteoarthritis
and confers a greater risk of developing bilateral hip osteoar-
thritis [18]. Inflammatory markers and pain scores are elevat-
ed in those suffering with OA and non-insulin-dependent di-
abetes mellitus compared with OA alone [19,20]. A study of
the serum of patients with OA showed an increase in total
cholesterol compared with suitably matched controls [21].
This wide body of evidence for phenotypic variations and
potentially more progressive osteoarthritis in the presence of
metabolic disturbance has now resulted in OA being incorpo-
rated into some definition of MeS [22]. Nevertheless, what
remains unclear from these predominantly retrospective stud-
ies is whether metabolic disturbances are a cause or effect of a
proinflammatory state.
Previous studies have suggested that treating these meta-
bolic disturbances, through dietary modification or direct-
acting drug therapies, could slow or even halt the progression
of OA [23]. Statin therapy in animal models showed promis-
ing reductions in catabolic OA cytokines and attenuated his-
tological OA progression. Moreover, a 5-year longitudinal
cohort study identified that atorvastatin significantly lowered
the risk of developing knee pain [24]. The mechanism by
which statins exert these effects remains unclear; however,
reductions in IL-1, IL-6 and IL-8 and upregulation of nitric
oxide synthase (NOS) have been postulated as potential mech-
anisms. These factors are intrinsically involved in the inflam-
matory pathogenesis of metabolic OA.
Why is metabolism important
in osteoarthritis?
OA is a multifactorial disease, which has a widely accepted
metabolic variant. How and which key metabolic components
establish and/or drive OA is not clear [14,23,25]. Broadly, it
is perceived that the overexpression of proinflammatory me-
diators in OA is triggered through components of the MeS,
including, for example, adipokines and advanced glycation
end-products, from dyslipidaemic and hyperglycaemic states
respectively.
Cohort studies suggest phenotypic variations in OA mani-
fest as differing responses of pain and OA progression, BMI,
sex and depression—all of which have been shown to differ
metabolically [26]. Although these factors influence systemic
phospholipid metabolism, a causal effect upon OA pathogen-
esis is yet to be established [27,28]. By elucidating the role of
metabolic stress in OA, we will gain insights into novel mo-
lecular targets for disease-modifying treatments.
Which tissue types to sample and why?
The goal of metabolic analysis is to identify metabolic differ-
ences occurring secondary to disease or an intervention. Thus,
in the context of disease analysis, the selection of body fluid or
tissue requires careful consideration.
Typical biofluids such as urine and blood are favoured as
advantageous due to a ubiquitous route of excretion, ease of
sampling and mirroring any future clinicalmeasurement of the
disease state. A key shortcoming is their heterogeneity with
many body systems resulting in spurious effects upon metab-
olite levels. Furthermore, in the context of OA, the relative
effect of local joint disease towards systemic metabolism may
be limited.
Local sampling of an affected tissue or fluid may prove
more representative of the altered metabolic state and subse-
quently better describe disease aetiology or mechanism.
However, sampling these fluids or tissues may be technically
and ethically challenging [8].
In OA, sampling the synovial fluid or tissue can enable the
investigation of specific changes in cartilage metabolism,
where disease is most fulminant and major structural changes
occur. Additionally, synovial fluid is solely responsible for
supplying nutrients and removing waste products from the
cartilage, thereby making it an attractive reservoir for the pas-
sage of OA-specific small molecules.
Alternatively, the synovium and intraarticular fat, as an
adipokine source, may provide more insights into pro-
inflammatory mediators of OA. Diseased cartilage may be
studied directly to understand the small molecule changes
associated with cartilage depolymerisation—nevertheless,
clinical applicability will be limited due to the destructive
nature of joint surface sampling.
The metabolic discoveries of osteoarthritis
There are several archetypal metabolic profiles associated
with OA. Synovial tissue culture from patients with varying
degrees of osteoarthritis showed depletion in the branch chain
amino acids (BCAA), amino acids (AA) and tricarboxylic
acid (TCA) cycle intermediates with the predominance of lac-
tate and pyruvate production [29]. This is consistent with the
notion that the energetic stresses of OA mediate a shift to an
anaerobic state [30–32]. It is worth noting, however, that this
metabolic state is not specific to OA [33,34]. Thus, clarifica-
tion of the timing or degree of this metabolic disturbance
compared with other disease states is necessary for OA-
specific inferences. A more promising biomarker is plasma
Clin Rheumatol

histidine, which is less concentrated in OA patients and as a
singular measure may have value as a point-of-care diagnostic
tool [34,35].
The effect of osteoarthritic disease upon intra-articular pro-
teoglycan destruction has been demonstrated in cartilage and
synovial fluid [27,30,36,37]. Whether this change is quan-
tifiable as a proxy measure of the magnitude of joint destruc-
tion is unclear and requires larger, well-designed studies with
more accurate descriptions of the intra-articular state. In prin-
ciple, the measurement of biological monomers generated by
the depolymerisation of structural molecules should reflect
joint destruction, albeit, perhaps, with a non-linear
relationship.
Lipid metabolism is an obvious target for metabolic study,
due to its pro-inflammatory properties. OA synovial fluid has
a marked increase in longer fatty acid chains [38].
Furthermore, with concurrent elevations in ketone bodies, this
indicates a switch to fatty acid metabolism [30,39,40]. The
results of the Canadian metabolome project indicate arginine
depletion in plasma alongside an alteration to lipid species in
OA [41]. Furthermore, the alterations in the lipid profile were
more pronounced in the presence of diabetes—suggesting it
could be a modality to isolate patients with metabolic OA
[27]. A description of osteoarthritis populations revealed a
separation of patients into lipid-specific classes, distinguished
by virtue of glycophospholipid and sphingomyelin levels
[42]. These lipid classes varied with age, obesity and response
to pain, factors which confound with ageing-related and me-
chanical OA, but could still be linked to metabolic OA.
Mesenchymal stromal cell (MSC) differentiation towards
cartilage anabolic cell lines has been shown to be affected in
OA. These cell lines have been shown to favour glycolysis
and uronic acid metabolism as disease progresses, specifically
pentose and glucuronic acid [43].
The majority of studies to date have examined the joint as a
biosystem in entirety. However, one must consider that mito-
chondrial DNA variants may drive metabolism and osteoar-
thritis disease. In addition to a disruption in oxidative phos-
phorylation is the effect upon macromolecule synthesis and
maintenance. Whether the presence of the mitochondrial
DNAvariantsissolelyresponsibleforOAdisease,metabolic
associations and macromolecule catabolism is not clear [44].
However, despite the generation of reactive oxygen species,
we do not believe this mechanism fully accounts for the in-
flammatory component of the disease or the altered lipid
metabolism.
What are the potential uses and outcomes
of the metabolic data?
Descriptive reporting of the metabolic OA phenotype has
identified associated metabolites. Nevertheless, to date,
variations within OA subtypes have not been demonstrated
or been able to explain the OA cohort variations in, for exam-
ple, depression, pain and disease severity. While this can be
attributed in the most part to inadequately powered studies, a
metabolic description of the OA subtypes is required to deter-
mine potential associations with these clinically observed phe-
notypic variations.
If the aforementioned limitations are resolved, by
characterising patients’metabolic profiles, there is the po-
tential to identify specific OA subtypes (Fig. 3). This model
of care, with metabolic profiling as a diagnostic tool, can
facilitate a paradigm shift from the present “one size fits all”
OA management approach, which is indifferent to trauma-
related or metabolic-induced OA. Metabolic profiling of
joint fluid may provide a personalised insight into subsets
of metabolic OA. Overall, this bespoke approach to treat-
ment can only benefit patients, by permitting more effica-
cious, cost-effective treatment allocation and even the abil-
ity to predict treatment response [12,45]. The latter can be
achieved with the ability of metabolic data to provide ad-
vanced disease descriptions and a window into the real-time
effects and interaction between treatment, lifestyle and di-
etary changes.
OA paent
Characterise phenotype
Determine metabolic joint
fluid profile
Subclassify into eg MeS,
inflammatory, degenerave
Apply a bespoke tailored
treatment
Fig. 3 The osteoarthritis diagnostics and therapeutic workflow enabled
by joint fluid metabolic profiling
Clin Rheumatol

Limitations of metabolic phenotyping
and the over-reporting of results
As metabolic profiling is all-encompassing, confounding fac-
tors, including lifestyle, diet and exercise, pose the greatest
limitation upon meaningful metabolic profile interpretation.
This is further complicated by modern polypharmacy, which
introduces additional metabolites and metabolism-altering
drugs. These obstacles can be controlled by study design,
sample numbers and sound statistical principles—no research-
er should be naïve about their presence or influence.
Detection of metabolites is limited by metabolite similarity
and instrument sensitivity. Instrument sensitivity continues to
improve. However, a single technique will always have selec-
tive limitations and thus, using multiple analytical methods
(NMR, LC-MS, GC-MS, CE-MS) is desirable.
A majority of multi-variate statistical models use linear re-
gression and thus, any variable which has a non-linear associa-
tion may be misrepresented—something frequently overlooked
by investigators. These statistical models while powerful con-
tribute to the exaggerated reporting of statistical significance.
For example, one can easily use partial least squares discrimi-
nant analysis (PLS-DA) models to generate plots showing good
pictorial separation of the sample classes; however, care must be
taken that models are not overfitted and are appropriately vali-
dated. Hence, if the quality of model is not assured by using test
sets or cross-validation, data can be misrepresented and generate
false positives. Unfortunately, in some clinical studies, these
errors have been carried through into publication.
“Personalised medicine”and the impact
of metabolic profiling
“Personalised medicine”is defined by NHS England as “a
move away from a ‘one size fits all’approach to the treatment
and care of patients with a particular condition, to one which
uses new approaches to better manage patients’health and
targets therapies to achieve the best outcomes in the manage-
ment of a patient’s disease or predisposition to disease.”[44].
This is a concept widely held to be a viable product of
metabolic profiling techniques and is being realised in the
evaluation of atherosclerosis and the risk of developing cardio-
vascular disease [46]. The newfound interest and role of the
metabolome in osteoarthritis make this goal achievable via a
more detailed description of OA. This has potential for a more
detailed and sub-categorised OA diagnosis with aetiology and
treatments to match, possibly with metabolic targets. The im-
portance of MSC subtypes and sub-cellular mitochondrial me-
tabolism is not established but these metabolic perturbations
are likely to be attractive targets for future study.
A great limitation to the success of these techniques in the
research of OA is the lack of population-based studies. In
order to address this, phenome centres have been created with
significant financial and academic commitment, allowing
centralisation of expertise and resources. There are now facil-
ities capable of providing phenome data to rigorous standards
for population-based studies. The geographical locations of
these phenome centres lend themselves to make population
comparisons. Furthermore, the possibility for metadata com-
parison and collation is much greater, due to agreed standards
for experimental methods. To date, osteoarthritis research has
not exploited these resources.
The power of “big data”tools like genomics and metabolic
profiling comes through combined analysis, which effectively
multiplies the descriptive power of one’s available data.
Genome-wide associated studies (GWAS) have provided
some OA genetic targets, primarily involved in gene regula-
tion. Links have been demonstrated to genes involved in car-
tilage turnover (cartilage oligomeric matrix protein, COMP)
and inflammation (transcription growth factor alpha, TGFA)
[47,48]. Marrying these findings to their downstream meta-
bolic effects can allow a greater understanding of OA subtype
aetiologies and widen the scope for therapeutic targets.
Presently, in OA research, merged genotyping and advanced
phenotyping are lacking.
Data analysis in using “big data”methods is challenging.
The biggest limitation is often the data itself, due to a failure of
collection or addressing the clinical question. Multi-variate
linear regression models have become the standard analysis
methods in interpreting metabolic profiling. Novel computa-
tional and machine learning algorithms have become more
powerful and attracted sizeable investment. Limited metabolic
profiling reports have applied these methods and one would
expect an expansion going forward, with uncertain impact but
hopeful advancement.
One of the barriers to clinical applications of metabolic phe-
notyping is its requirement for extensive and specialist analyt-
ical methods and equipment. The engineering of solutions ca-
pable of providing a complex, albeit, specific pattern of metab-
olite composition is evolving at pace. Handheld engineering
solutions are now capable of sampling organic metabolites.
Microchip technology allows multiple metabolite analysis and
a coded result with the potential for robust, affordable and non-
invasive testing [49]. In OA, these technologies could enable
cost-effective early detection in asymptomatic individuals and
sub-stratification in symptomatic individuals. Hence, we can
envisage a future where a simple handheld breath or urine test
can characterise disease and permit personalised treatment.
Summary
The benefits of using the metabolic profile to provide an ad-
vanced description of osteoarthritis are numerous in both re-
search and clinical practice. A detailed metabolic description
Clin Rheumatol

of the OA patient can enable targeted therapy, ultimately re-
duce costs of treatment and allow predictions of one’streat-
ment journey. Drug development and treatment can be fur-
thered by a greater understanding of an individual’sresponse
to disease and therapy.
It is still not possible to accurately and reliably sub-stratify
patients into OA phenotypes and identify metabolic OA.
Nevertheless, robust studies are now being undertaken to in-
form our knowledge of OA subtypes and to suggest associated
metabolites. Progress will require a population-based study
with comprehensive clinical measures and in-depth metabolic
measurements, with a view to describing the observed disease
cohorts that are seen. Phenome centres will be pivotal in pro-
viding access to population data and expertise. This will move
us away from a “one size fits all”model for OA treatment and
deliver the promise of metabolic profiling.
Author contributions MKJJ drafted the manuscript. MKJJ, PA, UV,
CLB and GG were involved in the research and interpretation of the
material. MKJJ, JCL and CMG designed the manuscript. RB, HRTW,
JCL and CMG conceived the project. All authors revised the manuscript
and approved the final version for publication.
Compliance with ethical standards
Disclosures None.
Open Access This article is licensed under a Creative Commons
Attribution 4.0 International License, which permits use, sharing, adap-
tation, distribution and reproduction in any medium or format, as long as
you give appropriate credit to the original author(s) and the source, pro-
vide a link to the Creative Commons licence, and indicateif changes were
made. The images or other third party material in this article are included
in the article's Creative Commons licence, unless indicated otherwise in a
credit line to the material. If material is not included in the article's
Creative Commons licence and your intended use is not permitted by
statutory regulation or exceeds the permitted use, you will need to obtain
permission directly from the copyright holder. To view a copy of this
licence, visit http://creativecommons.org/licenses/by/4.0/.
References
1. Lindon JC, Holmes E, Nicholson JK (2003) So what’s the deal with
metabonomics? Anal Chem 75:384A–391A
2. Dumas ME, Maibaum E, Teague C, Ueshima H, Zhou B, Lindon J,
Nicholson J, Stamler J, Elliott P, Chan Q, Holmes E (2006)
Assessment of analytical reproducibility of 1H NMR spectroscopy
based metabonomics for large-scale epidemiological research: the
INTERMAP study. Anal Chem 78:2199–2208
3. Gu Y, Lu C, Zha Q, Kong H, Lu X, Lu A, Xu G (2012) Plasma
metabonomics study of rheumatoid arthritis and its Chinese medicine
subtypes by using liquid chromatography and gas chromatography
coupled with mass spectrometry. Mol BioSyst 8(5):1535–1543
4. Hodavance MS, Ralston SL, Pelczer I (2007) Beyond blood sugar: the
potential of NMR-based metabonomics for type 2 human diabetes,
and the horse as a possible model. Anal Bioanal Chem 387:533–537
5. Odunsi K, Wollman R, Ambrosone C, Hutson A, McCann S,
Tammela J, Geisler J, Miller G, Sellers T, Cliby W, Qian F, Keitz
B, Intengan M, Lele S, Alderfer J (2005) Detection of epithelial
ovarian cancer using
1
H-NMR-based metabonomics. Int J Cancer
113:782–788
6. Qiu Y, Cai G, Su M, Chen T, Liu Y, Xu Y, Ni Y, Zhao A, Cai S, Xu
LX, Jia W (2010) Urinary metabonomic study on colorectal cancer.
J Proteome Res 9(3):1627–1634
7. Zhang Z, Qiu Y, Hua Y, Wang Y, Chen T, Zhao A, Chi Y, Pan L,
Hu S, Li J, Yang C, Li G, Sun W, Cai Z, Jia W (2010) Serum and
urinary metabonomic study of human osteosarcoma. J Proteome
Res 9(9):4861–4868
8. Hagberg G (1998) From magnetic resonance spectroscopy to clas-
sification of tumors. A review of pattern recognition methods.
NMR Biomed 11:148–156
9. Fischer K, Kettunen J, Wurtz P, Haller T, Havulinna AS, Kangas
AJ, Soininen P, Esko T, Tammesoo ML, Magi R, Smit S, Palotie A,
Ripatti S, Salomaa V, Ala-Korpela M, Perola M, Metspalu A
(2014) Biomarker profiling by nuclear magnetic resonance spec-
troscopy for the prediction of all-cause mortality: an observational
study of 17,345 persons. PLoS Med 11(2):e1001606. https://doi.
org/10.1371/journal.pmed.1001606
10. Holmes E, Loo RL, Stamler J, Bictash M, Yap I, Chan Q, Ebbels T,
De Iorio M, Brown I, Veselkov K, Daviglus M, Kesteloot H,
Ueshima H, ZhaoL, Nicholson J, Elliott P (2008) Human metabolic
phenotype diversity and its association with diet and blood pressure.
Nature 453:396–400
11. Yu B, Zheng Y, Nettleton JA, Alexander D, Coresh J, Boerwinkle
E (2014) Serum metabolomic profiling and incident CKD among
African Americans. Clin J Am Soc Nephrol 9(8):1410–1417.
https://doi.org/10.2215/CJN.11971113
12. Nicholson JK, Lindon JC (2008) Systems biology: Metabonomics.
Nature 455(7216):1054–1056
13. Ding X, Hu J, Li J, Zhang Y, Shui B, Ding Z, Yao L, Fan Y (2014)
Metabolomics analysis of collagen-induced arthritis in rats and in-
terventional effects of oral tolerance. Anal Biochem 458(0):49–57
14. Li H, George DM, Jaarsma RL, Mao X (2016) Metabolic syndrome
and components exacerbate osteoarthritis symptoms of pain, de-
pression and reduced knee function. Ann Transl Med 4(7):133.
https://doi.org/10.21037/atm.2016.03.48
15. Gkretsi V, Simopoulou T, Tsezou A (2011) Lipid metabolism and
osteoarthritis:lessons from artherosclerosis. Prog Lipid Res 50(2):
133–140
16. Philbin EF, Ries MD, Groff GD, Sheesley KA, French TS, Pearson
TA (1996) Osteoarthritis as a determinant of an adverse coronary
heart disease risk profile. J Cardiovasc Risk 3:529–533
17. Singh G, Miller JD, Lee FH, Pettitt D, Russell MW (2002)
Prevalence of cardiovascular disease risk factors among US adults
with self-reported osteoarthritis: data from the third NationalHealth
and nutrition examination survey. Am J Manag Care 8:S383–S391
18. Berenbaum F (2011) Diabetes-induced osteoarthritis: from a new
paradigm to a new phenotype. Ann Rheum Dis 70:1354–1356
19. Cimmino MA, Cutolo M (1990) Plasma glucose concentration in
symptomatic osteoarthritis: a clinical and epidemiological survey.
Clin Exp Rheumatol 8:251–257
20. Hart DJ, Doyle DV, Spector TD (1995) Association between met-
abolic factors and knee osteoarthritis in women: the Chingford
study. J Rheumatol 22(6):1118–1123
21. Borman P, Seckin U, Yucel M (1999) Dyslipidaemia in patients
with rheumatoid and osteoarthritis. Phys Med 2(3):5–9
22. Zhuo Q, Yang W, Chen J, Wang Y (2012) Metabolic syndrome
meets osteoarthritis. Nat Rev Rheumatol 8(12):729–737. https://
doi.org/10.1038/nrrheum.2012.135
23. Farnaghi S, Prasadam I, Cai G, Friis T, Du Z, Crawford R, Mao X,
Xiao Y (2017) Protective effects of mitochondria-targeted antioxi-
dants and statins on cholesterol-induced osteoarthritis. FASEB J
31(1):356–367. https://doi.org/10.1096/fj.201600600R
Clin Rheumatol

24. Baker JF,Walsh P, MulhallKJ (2011) Statins: a potential role in the
management of osteoarthritis? Joint Bone Spine 78:31–34
25. de Munter W, van der Kraan PM, van den Berg WB, van Lent PL
(2016) High systemic levels of low-density lipoprotein cholester-
ol: fuel to the flames in inflammatory osteoarthritis?
Rheumatology (Oxford) 55(1):16–24. https://doi.org/10.1093/
rheumatology/kev270
26. Rockel JS, Kapoor M (2018) The Metabolome and osteoarthritis:
possible contributions to symptoms and pathology. Metabolites
8(4). https://doi.org/10.3390/metabo8040092
27. Zhang WD, Sun G, Likhodii S, Aref-Eshghi E, Harper PE, Randell
E, Green R, Martin G, Furey A, Rahman P, Zhai GJ (2016)
Metabolomic analysis of human synovial fluid and plasma reveals
that phosphatidylcholine metabolism is associated with both osteo-
arthritis and diabetes mellitus. Metabolomics 12(2):24 ARTN 24.
https://doi.org/10.1007/s11306-015-0937-x
28. Gierman LM, Wopereis S, van El B, Verheij ER, Werff-van der Vat
BJ, Bastiaansen-Jenniskens YM, van Osch GJ, Kloppenburg M,
Stojanovic-Susulic V, Huizinga TW, Zuurmond AM (2013)
Metabolic profiling reveals differences in concentrations of
oxylipins and fatty acids secreted by the infrapatellar fat pad of
donors with end-stage osteoarthritis and normal donors. Arthritis
Rheum 65(10):2606–2614. https://doi.org/10.1002/art.38081
29. Adams SB Jr, Setton LA, Kensicki E, Bolognesi MP, Toth AP,
Nettles DL (2012) Global metabolic profiling of human osteoar-
thritic synovium. Osteoarthr Cartil 20(1):64–67
30. Damyanovich AZ, Staples JR, Chan ADM, Marshall KW (1999)
Comparative study of normal and osteoarthritic canine synovial
fluid using 500 MHz
1
H magnetic resonance spectroscopy. J
Orthop Res 17:223–231
31. Adams Jr. SB, Jones LC, Haile A, Miller SD, Gutyon GP, Schon
LC (2012) Inflammatory cytokine composition and metabolic pro-
file of post-traumatic ankle joint arthritis. In: Transactions of the
Orthopaedic Research Society
32. Marshall KW, Damyanovich AZ, Staples RJ (2000) A comparative
study of deproteinized human synovial fluid in early and late stage
osteoarthritis using 500 MHz 1H-NMR spectroscopy. Paper pre-
sented at the 46th Annual Meeting, Orthopaedic Research Society,
Orlando, Florida, March 12-15
33. Naughton DP, Haywood R, Blake DR, Edmonds S, Hawkes GE,
Grootveld M (1993) A comparative evaluation of the metabolic
profiles of normal and inflammatory knee-joint synovial fluids by
high resolution proton NMR spectroscopy. FEBS Lett 332(3):
221–225
34. Sitton NG, Dixon JS, Bird HA, Wright V (1986) Serum and syno-
vial fluid histidine: a comparison in rheumatoid arthritis and osteo-
arthritis. Rheumatol Int 6:251–254
35. Zhai G, Wang-Sattler R, Hart DJ, Arden NK, Hakim AJ, Illig T,
Spector TD (2010) Serum branched-chain amino acid to histidine ratio:
a novel metabolomic biomarker of knee osteoarthritis. Ann Rheum Dis
69(6):1227–1231. https://doi.org/10.1136/ard.2009.120857
36. Shet K, Siddiqui SM, Yoshihara H, Kurhanewicz J, Ries M, Li X
(2012) High-resolution magic angle spinning NMR spectroscopy of
human osteoarthritic cartilage. NMR Biomed 25:538–544
37. Damyanovich AZ, Staples RJ, Marshall KW (2000) The effects of
freeze/thawing on human synovial fluid observed by 500 MHz 1H
magnetic resonance spectroscopy. J Rheumatol 27(3):746–752
38. Duffy JM, Grimshaw J, Guthrie DJ, McNally GM, Mollan RA,
Spedding PL, Trocha-Grimshaw J, Walker B, Walsh E (1993)
1H-nuclear magnetic resonance studies of human synovial fluid in
arthritic disease states as an aid to confirming metabolic activity in
the synovial cavity. Clin Sci 85(3):343–351
39. Damyanovich AZ, Staples JR, Marshall KW (1999)
1
H NMR in-
vestigation of changes in the metabolic profile of synovial fluid in
bilateral canine ostroarthritis with unilateral joint denervation.
Osteoarthr Cartil 7:165–172
40. Mickiewicz B, Heard BJ, Chau JK, Chung M, Hart DA, Shrive NG,
Frank CB, Vogel HJ (2015) Metabolic profiling ofsynovial fluid in
a unilateral ovine model of anterior cruciate ligament reconstruction
of the knee suggests biomarkers for early osteoarthritis. J Orthop
Res 33(1):71–77. https://doi.org/10.1002/jor.22743
41. Zhang W, Sun G, Likhodii S, Liu M, Aref-Eshghi E, Harper PE,
Martin G, Furey A, Green R, Randell E, Rahman P, Zhai G (2016)
Metabolomic analysis of human plasma reveals that arginine is
depleted in knee osteoarthritis patients. Osteoarthr Cartil 24(5):
827–834. https://doi.org/10.1016/j.joca.2015.12.004
42. Zhang W, Likhodii S, Zhang Y, Aref-Eshghi E, Harper PE, Randell
E, Green R, Martin G, Furey A, Sun G, Rahman P, Zhai G (2014)
Classification of osteoarthritis phenotypes by metabolomics analy-
sis. BMJ Open 4(11):e006286. https://doi.org/10.1136/bmjopen-
2014-006286
43. Rocha B, Cillero-Pastor B, Eijkel G, Fernández-Puente P, Paine
MRL, Ruiz-Romero C, Heeren RMA, Calamia V, Blanco FJ
(2020) Integrative metabolic pathway analysis reveals novel thera-
peutic targets in osteoarthritis. Mol Cell Proteomics 19:574–588
44. Blanco FJ, Valdes AM, Rego-Pérez I (2018) Mitochondrial DNA
variation and the pathogenesis of osteoarthritis phenotypes. Nat
Rev Rheumatol 14(6):327–340. https://doi.org/10.1038/s41584-
018-0001-0
45. Andrew Clayton T, Lindon JC, Cloarec O, Antti H, Charuel C,
Hanton G, Provost J-P, Le Net J-L, Baker D, Walley RJ, Everett
JR, Nicholson JK (2006) Pharmaco-metabonomic phenotyping and
personalized drug treatment. Nature 440(7087):1073–1077 http://
www.nature.com/nature/journal/v440/n7087/suppinfo/
nature04648_S1.html
46. Mahmood SS, Levy D, Vasan RS, Wang TJ (2014) The
Framingham heart study and the epidemiology of cardiovascular
disease: a historical perspective. Lancet 383(9921):999–1008.
https://doi.org/10.1016/s0140-6736(13)61752-3
47. Styrkarsdottir U, Helgason H, Sigurdsson A, Norddahl GL,
Agustsdottir AB, Reynard LN, Villalvilla A, Halldorsson GH,
Jonasdottir A, Magnusdottir A, Oddson A, Sulem G, Zink F,
Sveinbjornsson G, Helgason A, Johannsdottir HS, Helgadottir A,
Stefansson H, Gretarsdottir S, Rafnar T, Almdahl IS, Braekhus A,
Fladby T, Selbaek G, Hosseinpanah F, Azizi F, Koh JM, Tang NLS,
Daneshpour MS, Mayordomo JI, Welt C, Braund PS, Samani NJ,
Kiemeney LA, Lohmander LS, Christiansen C, Andreassen OA, Arc
O, Magnusson O, Masson G, Kong A, Jonsdottir I, Gudbjartsson D,
Sulem P, Jonsson H, Loughlin J, Ingvarsson T, Thorsteinsdottir U,
Stefansson K (2017) Whole-genome sequencing identifies rare geno-
types in COMP and CHADL associated with high risk of hip osteo-
arthritis. Nat Genet 49(5):801–805. https://doi.org/10.1038/ng.3816
48. Castano-Betancourt MC, Evans DS, Ramos YF, Boer CG,
Metrustry S, Liu Y, den Hollander W, van Rooij J, Kraus VB,
Yau MS, Mitchell BD, Muir K, Hofman A, Doherty M, Doherty
S, Zhang W, Kraaij R, Rivadeneira F, Barrett-Connor E,
Maciewicz RA, Arden N, Nelissen RG, Kloppenburg M, Jordan
JM, Nevitt MC, Slagboom EP, Hart DJ, Lafeber F, Styrkarsdottir
U, Zeggini E, Evangelou E, Spector TD, Uitterlinden AG, Lane
NE, Meulenbelt I, Valdes AM, van Meurs JB (2016) Novel genetic
variants for cartilage thickness and hip osteoarthritis. PLoS Genet
12(10):e1006260. https://doi.org/10.1371/journal.pgen.1006260
49. Patil SB, Dheeman DS, Al-Rawhani MA, Velugotla S, Nagy B,
Cheah BC, Grant JP, Accarino C, Barrett MP, Cumming DRS
(2018) An integrated portable system for single chip simultaneous
measurement of multiple disease associated metabolites. Biosens
Bioelectron 122:88–94. https://doi.org/10.1016/j.bios.2018.09.013
Publisher’snoteSpringer Nature remains neutral with regard to jurisdic-
tional claims in published maps and institutional affiliations.
Clin Rheumatol