Metabolic profile changes in serum of migraine patients detected using 1H-NMR spectroscopy

DOI: https://doi.org/10.1186/s10194- 021- 01357- w

Abstract-Summary Our aim was to identify, using proton nuclear magnetic resonance (1H-NMR) spec- troscopy, metabolites in serum that are associated with lifetime and active migraine by comparing metabolic profiles of patients and controls.

A total of 100 signals representing 49 different metabolites were detected in 289

cases (of which 150 active migraine patients) and 1360 controls.

We were able to identify profiles consisting of 6 metabolites predictive for life-

time migraine status and 22 metabolites predictive for active migraine status.

We estimated with subsequent regression models that after correction for age, sex, BMI and smoking, the association with the metabolite profile in active migraine remained.

This study indicates that metabolic profiles, based on serum concentrations of several metabolites, including lipids, amino acids and metabolites of glucose metab- olism, can distinguish active migraine patients from controls.

Extended: A total of 73,728 data points covering a spectral width of 12,019 Hz

were collected using 16 scans.

A total of 100 metabolite signals were detected in the JRES projection and quan-

tified in the 1H-NMR spectra [259].

To validate the findings from the previous analysis we performed analysis of variance (ANOVA) in which we compared the performance of the full model with the identified scores for migraine with the performance of a model containing only information on age, sex, BMI and smoking.

A replication study, to validate our findings, should be performed.

Introduction Although much progress has been made with unravelling its (non) genetic disease mechanisms [260], a diagnosis of migraine is still made by interview and physical examination or questionnaire, as no diagnostic biomarker is available.

Validated biomarkers can greatly improve diagnosis, prognosis and assessing effectivity of treatment of patients, as was already shown for several diseases other than migraine [261, 262].

To search for migraine metabolite profiles in serum we used an untargeted, hypothesis-free, approach and performed high-throughput proton nuclear magnetic resonance (1H-NMR) spectroscopy.

We analysed metabolite profiles in serum samples of migraine patients and con- trols from the Erasmus Rucphen Family population, a large Dutch population-based family study from the Southwest of the Netherlands in which we previously had identified migraine cases [263].

284

2 Mechanisms

We set out to investigate whether metabolites identified by 1H-NMR spectros- copy are associated with migraine by comparing metabolic profiles of migraine patients and controls in a “real-life variation” cohort.

Material and Methods To reduce the dimensionality of the data and due to possible correlations between the parameters, elastic net regression was used to select a subset of the most infor- mative signals for: (1) lifetime migraine diagnosis, and (2) a diagnosis of active migraine (defined as having at least one severe migraine in the last 12 months).

In this cross-validation step we validated the selection of the signals by perform- ing our regression analysis on 50 randomly chosen samples of our study population. To determine whether we had to correct our logistic regression model we inde- pendently assessed the influence of sex, age, body mass index (BMI) and smoking status on the “weighted metabolite score”, by visually inspecting stratification plots and performing a linear model, where the “weighted metabolite score” was mod- elled as a function of migraine status.

Results Elastic net regression analysis of all 289 migraine patients and 1360 controls for all 100 signals identified six 1H-NMR signals as the best prediction subset.

Subsequent logistic regression analysis showed support for association (odds ratio (OR) = 2.72; 95% confidence interval (CI) 1.97–3.75; p = 1.28 × 10− 9) explain- ing 3.9% of the variance in migraine status (Nagelkerke R2).

We performed an elastic net regression on all 150 active migraine patients and

1360 controls for all 100 signals.

The regression showed support for association between the signal data and active migraine status (OR = 2.72; 95% CI 2.09–3.54; p = 1.35 × 10− 13) explaining 8.5% of the variance, this association remained after correction for sex, age, BMI and current smoking status (OR = 1.84; 95% CI 1.34–2.53; p = 1.64 × 10− 4) with a total explained variance of 12.3% (Nagelkerke R2).

Discussion A recent meta-analysis encompassing 2800 migraine patients and 7353 controls from eight Dutch cohorts, using a different 1H-NMR metabolomics platform in a systematic approach, also showed alterations in HDL metabolism, in that study defined by a decreased level of lipoprotein A1 and a decreased free cholesterol to total lipid ratio in small HDL subspecies [264].

One study has investigated the lactic and pyruvic acid levels in the plasma of the migraine patients it was shown that both were significantly higher in migraine patients than in normal controls [265].

In a small study of 37 migraine patients and 40 controls elevated levels of isoleu-

cine in blood serum were found [266].

This LC-MS study found decreased levels of leucine, isoleucine and methionine

in migraine patients compared to controls.

2.2 Biology

285

Acknowledgement A machine generated summary based on the work of Harder, Aster V. E.; Vijfhuizen, Lisanne S.; Henneman, Peter; Willems van Dijk, Ko; van Duijn, Cornelia M.; Terwindt, Gisela M.; van den Maagdenberg, Arn M. J. M. 2021 in The Journal of Headache and Pain.

Increased nitroglycerin-mediated vasodilation in migraineurs without aura in the interictal period