DNA-methylation and immunological response in medication overuse headache

Abstract

Objective

To investigate whether medication-overuse headache patients have differential DNA-methylation pattern.

Methods

We collected blood samples from 120 medication-overuse headache-patients, 57 controls (29 episodic migraine patients and 28 healthy controls) in a hypothesis-generating cross-sectional case-control pilot study; 100 of the medication-overuse headache-patients were followed for six months and samples were collected at two and six months for the longitudinal methylation analyses. Blood cell proportions of leucocytes (neutrophils, NK-cells, monocytes, CD8+ and CD4+ T-cells, and B-cells) and the neutrophile-lymphocyte ratio were estimated using methylation data as a measure for immunological analysis and a cell type-specific epigenome wide association study was conducted between medication-overuse headache-patients and controls, and longitudinally for reduction in headache days/month among medication-overuse headache-patients.

Results

We found a higher neutrophile-lymphocyte ratio in medication-overuse headache-patients compared to controls, indicating a higher immunological response in medication-overuse headache-patients (false discovery rate (adjusted p-value)<0.001). Reduction in headache days/month (9.8; 95% CI 8.1–11.5) was associated with lower neutrophile-lymphocyte ratio (false discovery rate adjusted p-value = 0.041).

Three genes (CORIN, CCKBR and CLDN9) were hypermethylated in specific cell types in medication-overuse headache-patients compared to controls. No methylation differences were associated with reduction in headache days in medication-overuse headache-patients after six months.

Conclusion

This pilot study was consistent with higher immunological response in medication-overuse headache-patients which decreased with a reduction in headache days in longitudinal analysis. medication-overuse headache-patients exhibited differential methylation in innate immune cells but did not exhibit longitudinal differences with alterations in headache days. Our study creates hypotheses for further biomarker searches.

ClinicalTrials.gov Identifier: NCT02993289

Keywords

Biomarkers chronic headache overuse of analgesics

Introduction

Medication-overuse headache (MOH) is a chronic headache occurring at least 15 days/month in individuals with pre-existing headache (1). Globally, MOH affects 63 million people (2). It is generally believed to be secondary to >3 months overuse of acute migraine medication or analgesics (overuse defined as ≥10 days/month with triptans, ergotamine, combination-analgesics or opioids, or ≥15 days/month with paracetamol or non-steroidal anti-inflammatory drugs [NSAIDs]) (1). The medication overuse contributes to a worsening of the pre-existing headache and increases headache frequency. Most patients with MOH are offered a combination of withdrawal therapy for the overuse and preventive headache medication as essential parts of the initial management (3,4). This approach results in remission from MOH in up to 90% of patients, meaning that medication overuse is ended, or patients revert to episodic headache (<15 headache days/month), or both. Thus between 65–75% of patients revert to episodic headache, and the headache frequency is reduced by approximately 40% (5 –7).

However, MOH diagnosis is solely based on the patient’s history, and the pathophysiology of MOH remains unknown. Approximately 25% of patients still have chronic headache after successful treatment, and 10–40% relapse within a year (6,8,9). Moreover, it is not possible to predict who will benefit from the treatment described above, or who will relapse. Identification of diagnostic and prognostic biomarkers for patients with MOH is highly warranted, but only few clinically relevant biomarkers have yet been identified (3).

In the search for such biomarkers, epigenetic analyses may be of considerable help, as seen in several other research fields (10 –12). Epigenetic analyses may be used to assess the cellular state of gene regulation, thus providing detailed information about the background for a specific phenotype. Altered regulation of the gene expression has previously been speculated to be involved in pathophysiology of several pain conditions, migraine, chronification of headache and MOH (13 –16). One the epigenetic modification is DNA-methylation, that can either be associated with a decreased or increased gene expression, depending on the location on the genome (i.e., promotor region or gene body). DNA-methylation changes can occur during lifetime due to environmental factors, and is in some cases even inheritable (i.e. genetic) (17).

We hypothesized that MOH-patients presented with a different DNA-methylation pattern, and that changes in DNA-methylation are associated with reduced headache frequency. The aim of this hypothesis-generating pilot study was to investigate whether MOH-patients had a different DNA-methylation pattern compared to healthy controls and patients with episodic migraine without medication overuse in a case-control study. Furthermore, we aimed to examine if treatment related changes in headache frequency were associated with altered DNA-methylation in a prospective longitudinal study.

Methods

Study population

From November 2016 to January 2019, 120 MOH-patients, 29 controls with episodic migraine (EM) and 28 healthy controls (HC) were included consecutively. At two months, 108 MOH-patients were followed up, while it was 100 after six months (Figure 1). All participants were at least 18 years of age, had no considerable comorbid pain, physical or psychiatric disorders, no alcohol or drug addiction, and were not pregnant or breastfeeding. The study was approved by the regional ethical committee (H-16029763) and all participants signed informed consent before inclusion.

Figure 1.

Flowchart for the two parts of the study.

MOH-patients

At the Danish Headache Center (DHC), all patients diagnosed with MOH and migraine and/or tension-type headache as pre-existing headache according to the International Classification of Headache Disorders 3rd beta version (ICHD-3beta) (18) were screened for participation. The diagnoses were given by headache specialists based on detailed history taking or at least one month filled out headache calendar. Patients should be considered eligible for outpatient treatment (without daily use of opioids or barbiturates).

MOH-patients daily registered occurrence of headache in a headache calendar throughout the study period, that included planned visits at treatment start (baseline), two-month and six-month follow-up (Figure 1). The original study design of the longitudinal part was a randomized controlled trial comparing three treatment strategies for MOH: withdrawal therapy and preventive headache medication; preventive headache medication (without withdrawal) and; withdrawal therapy (with delayed start of preventives) (19). Since no major differences between the treatment strategies in reducing headache days/month were found, we decided to pool all patients from the three groups in the longitudinal design and analyze their reduction in headache days (at baseline (BL) the groups were similar).

Control groups

The control groups were gender- and age-matched to the MOH-patients on group level. EM-controls were diagnosed according to the ICHD-3beta and had maximum six headache days/month without any history of chronic migraine. EM-controls could use preventive medication (e.g. anti-hypertensive or anti-epileptic drugs; CGRP monoclonal antibodies were unavailable in Denmark in the study period) if they were stable on the dose. HC-controls had no history of migraine and maximum two headache days/month. Use of analgesics was allowed up to maximum six days/month in both control groups, but any history of medication overuse was not allowed. Both control groups were recruited as volunteers via DHC, forsoegsperson.dk or the Danish blood donors.

Baseline characteristics

Detailed information about headache and analgesic-use was collected from all participants at baseline. Continuous variables are shown as mean with standard deviation (SD). Categorical variables as number with percentages in brackets. P-values <0.05 were considered significant.

Blood samples

Blood samples were collected from all participants at baseline, and for MOH-patients at two-month and six-month follow-up for longitudinal analyses (Figure 1). Blood was drawn from the cubital vein in 9 ml EDTA glass and stored as whole-blood in 2 ml plastic tubes at −25 degrees. All samples were analyzed at the same time to avoid batch-effect.

DNA methylation analysis

Detailed description of the methods is described in the online supplementary material.

DNA extraction and genome-wide quantification of DNA methylation

DNA was extracted and bisulfite-converted using the EZ-96 DNA Methylation Kit (Zymo Research, Irvine, California, USA) and analyzed with the Infinium Human Methylation EPIC Beadkit (Illumina, San Diego, California, United States). Each EPIC slide contained eight samples that were randomized for case (MOH), control (HC and EM) status. All time-point samples from same individual (MOH) were run on the same array and the sequence randomized.

Quality control and data preprocessing

Data quality at sample level was assessed by detection P-value <0.01 and bead count ≥3 sample call rate (CR) >0.98 and median log2 intensity signal of methylation (M) and unmethylation (U) signals >10. All samples passed the quality control criteria. Similarly, data quality was assessed at probe level by call frequency (CF) >0.99. Probes overlapping a single nucleotide polymorphisms (SNP) (20) and crossed hybridizing (21,22) were identified, and non-autosomal probes were filtered out by EPIC manifest annotation. This resulted in 717,501 CpG sites included in our analysis. Locations of the CpG sites are obtained using the Homo sapiens genome assembly GRCh37.

Data was normalized by a quantile normalization method (23), and the methylation proportion beta-values (M/[M + U]) were calculated. Normalized beta-values were batch corrected for two batch parameters, namely slide and array (rows). Seven principal components of the internal EPIC control probes (PCCP) and six components relating to non-reference based cellular heterogeneity, obtained while adjusting for gender and age, were used as covariates in the batch correction (24). This resulted in a dataset of adjusted methylation proportions (AdjBeta), that were filtered, normalized and finally adjusted for technical batch effects and cellular heterogeneity.

Statistical analyses

Immunomethylomic analysis

The DNAm based estimated blood cell proportions (eBCPs) were obatined with an adult blood DNAm reference panel from neutrophils, monocytes, NK-cells, B cells, CD4+ and CD8+ T cells. We determined the eBCPs from the most significantly cell type differentiating CpGs sites (IDOL set), using the CIBERSORT method for DNAm data (25,26). We calculated the neutrophil-to-lymphocyte ratio (NLR), a measure of systemic inflammatory response, from the eBCPs as NLR = neutrophils/(CD8+ T cells + CD4+ T cells+ B cells + NK-cells).

The eBCPs and NLR were first compared between the two control groups, i.e. EM and HC, using a t-test. Since no difference were found (see result section), the combined control group including EM and HC was compared to MOH, also using the t-test. In the MOH-group we also investigated the association between eBCPs and NLR and change in monthly headache days longitudinally (i.e. from baseline to two- and six-month follow-up) using linear mixed model (LMM). Resulting p-values were corrected for multiple testing for the six cell types using the false discovery rate (FDR) adjusted p-value and considered significant when FDR adjusted p-value <0.05. Statistical analyses were performed in R (v4.0.2).

Epigenome-wide association analysis

Cell type-specific epigenome wide association studies (EWAS), for each of the eBCP (described above) was conducted, identifying cell type-specific differentially methylated positions (CT-DMPs) between MOH-patients and controls (EM and HC) (Figure 1). The CT-DMPs analyses were performed by linear regression of all samples at baseline, with AdjBeta as the outcome and an interaction term between each of the eBCPs and MOH case/control status as the independent variables, with eBCP, sex, age, sampling year and month as covariates.

We obtained test statistics for each cell type-specific association at a genome-wide scale and performed 30 genome-wide random permutations of each model by randomizing case control status (27). This was done to empirically assess and adjust the test statistics with a random background, as p-values from the interaction model tend to be inflated for some cell types. We furthermore ran a standard EWAS for bulk tissue associations, only obtaining test-statistics for the association with MOH (no interaction term), while including the same covariates as above.

Empirical p-values were obtained by computing the empirical cumulative distribution function for all permuted p-values in each cell type and bulk tissue and obtaining the corresponding percentile of each test p-value, which is interpreted as its empirical p-value. Empirical p-values were corrected for multiple testing using the false discovery rate, resulting in FDR_emp adjusted p-values.

Furthermore, we identified differentially methylated regions with Comb-P, for bulk tissue (DMRs) and each cell-type (CT-DMRs). The CT-DMRs and DMRs were assessed by combining all (non-empirical) p-values with parameters previously suggested with the one exception of the p-value cutoff – seed parameter which was set to 1. To combine all proximal p-values and obtain comparative statistic from all tests and permutations, we used these to calculate FDR_emp adjusted p-value as previously described in the CT-DMP/DMP analysis for bulk tissue and each cell type. While this approach relies on combining all spatially neighboring methylation sites – regardless of their association with MOH, we confirmed our results by combined sites with the suggested p-value cutoff (seed 0.05) – which is unsuitable for obtaining an empirical background, but only combines regions association with MOH.

Of interest, DMPs refer to methylation on single CpG sites, while DMRs refer to simultaneously methylated proximal CpG sites. DMRs and DMPs could be associated with different biology, and DMRs are less likely to be spurious findings.

Methylation in a prospective longitudinal design

The association between reduction in headache days among MOH-patients and CT-DMPs/DMPs was assessed. We used a linear mixed model to assess variations in headache days with variations in the AdjBeta. Similar to the previous analysis we used AdjBeta as the outcome and the interaction between headache days and eBCPs as the independent variable, we controlled for the correlated data structure in the longitudinal design with the individual ID as random effect.

The model was adjusted for eBCPs and sampling time point in the study. Similar to the method described in the baseline model, we empirically adjusted the p-values with random permutations and combined CT-DMPs and DMPs into CT-DMRs and DMRs.

Power calculation

For EWAS, the statistical power was estimated with expected difference in methylation (Δβ) of 0.1 and 0.5 (online Supplementary Figure 1A), while for the paired-sample design, the power was estimated as a paired t-test using standard deviations varying from 0.5–5 within pairs (online Supplementary Figure 1B) (28,29) for difference in methylations. The estimated sample size was 100 patients with MOH, and 30 individuals in each control group.

Results

Study population

Baseline characteristics for the three study groups (MOH, EM and HC) are presented in Table 1. There was no significant difference in the age and gender distribution between the three groups, but a significant difference in sampling time (month and year) was found (p = 0.02 and <0.001, respectively, data not shown). After two months treatment in the MOH-group, the mean reduction in headache days/month was 7.3 days (95% CI 5.8–8.8 days) and after six months it was 9.8 days (95% CI 8.1–11.5 days).

Table 1.

Baseline characteristics.

	MOH(n = 120)	EM(n = 29)	HC(n = 28)	p-value
Age (years)	42.5 (12.2)	43.1 (10.9)	42.4 (14.1)	0.79
Gender (women)	96 (80.0%)	23 (79.3%)	19 (67.9%)	0.37
Headache frequency (days/month)	24.8 (5.7)	2.9 (1.7)	0.3 (0.5)	<0.001*
Acute medication (days/month)	22.6 (6.39)	2.0 (1.5)	0.3 (0.5)	<0.001*
Pre-existing headache
Chronic Migraine	62 (51.7%)	–	–
Episodic Migraine and TTH	38 (31.7%)	–	–
Chronic TTH	20 (16.7%)	–	–
Duration of overuse (years)	4.40 (7.66)	–	–
Type of preventives for headache
None	–	26 (89.7%)	–
Candesartan	–	2 (7.0%)	–
Magnesium	–	1 (3.4%)	–
Type of medication overuse
Simple analgesics	34 (28.3%)	–	–
Triptans	22 (18.3%)	–	–
Combination analgesics	38 (31.7%)	–	–
Opioids	1 (0.8%)	–	–
Combinations of medication overuse^†	1 (0.8%)	–	–

MOH, patients with medication-overuse headache; EM, control group with episodic migraine; HC, healthy controls, TTH, tension-type headache; NSAID, Non-Steroid Anti-Inflammatory Drug; ASA, Acetylsalicylic Acid.

Age, headache frequency, acute medication-use and duration of overuse are presented as mean (SD). Gender, pre-existing headache, preventive headache medication and medication overuse are shown as n (%). *p < 0.05.

^†ASA + Caffeine and triptans: 10 (8.3%); Paracetamol and triptans: 3 (2.5%); NSAID and triptans: 2 (1.7%); Paracetamol and ASA + Caffeine: 2 (1.7%); NSAID and ASA + Codeine: 2 (1.7%); NSAID and ASA + Caffeine: 1 (0.8%); Triptans and ASA + Codeine: 1 (0.8%); Paracetamol, NSAID and triptans 1 (0.8%); Paracetamol, NSAID and ASA + Caffeine: 1 (0.8%); Paracetamol, NSAID and ASA + Codeine: 1 (0.8%); Poly-overuse (paracetamol, NSAID and triptans): 1 (0.8%).

Immunomethylomic analysis

No difference in eBCPs or NLR between EM and HC were found, so to increase power both control groups were combined (Table 2). The proportion of neutrophils in MOH-patients was significantly higher than in the controls, while the proportions of NK-cells, monocytes and CD8+ T cells were significantly lower in the MOH-patients than in the controls (Figure 2A–G). A higher NLR was found in MOH-patients compared with the controls. Time of sampling (month and year) were included in the analyses.

Table 2.

Blood cell proportions and neutrophile:lymphocyte ratio.

	MOH(n = 120)	EM(n = 29)	HC(n = 28)	FDRHC vs. EM	FDRControl groups vs. MOH
CD4+ T-cells	0.13 (0.05)	0.14 (0.03)	0.14 (0.03)	0.81	0.52
CD8+ T-cells	0.08 (0.04)	0.11 (0.04)	0.10 (0.03)	0.81	0.001*
NK-cells	0.03 (0.02)	0.04 (0.02)	0.04 (0.02)	0.81	<0.001*
B-cells	0.05 (0.03)	0.05 (0.02)	0.05 (0.02)	0.88	0.78
Monocytes	0.06 (0.03)	0.07 (0.02)	0.08 (0.04)	0.81	0.042*
Neutrophils	0.65 (0.09)	0.59 (0.07)	0.60 (0.08)	0.88	<0.001*
NLR	2.49 (1.02)	1.86 (0.58)	1.96 (0.63)	0.81	<0.001*

MOH, Medication overuse headache; EM, Episodic migraine patients; HC, Healthy controls; FDR, False discovery rate adjusted p-value; Control groups, Episodic migraine patients and healthy controls; NLR, Neutrophile:lymphocyte ratio.

Blood cell proportion and NLR are presented as mean (SD). *FDR adjusted p-value <0.05.

Figure 2.

Blood cell proportions and neutrophile:lymphocyte ratio in MOH-patients and controls. The boxplots illustrate NLR (a) and blood cell proportions ((b) Neutrophils, (c) NK-cells, (d) CD8+ T-cells, (e) CD4+ T-cells, (f) Monocytes and (g) B-cells) for MOH-patients and the control group.

The reduction in headache days/month (7.3 and 9.8 days above), was accompanied by a change in eBCPs and NLR. We found that reduction in headache days/month was associated with a higher proportion of CD4+ T cells and a lower proportion of neutrophils (FDR adjusted p-value = 0.041 and 0.029, respectively), as well as a lower NLR (FDR adjusted p-value =0.041).

Epigenome-wide association study

Three CT-DMPs and two CT-DMRs were found to be significantly associated when comparing MOH-patients with controls (EM and HC), as summarized in Table 3.

Table 3.

Results of the epigenome-wide association study.

Analysis	Cell type	Marker	Chromosome	Position	Gene	β	SE	FDR_emp
CT-DMP	NK-cells	cg04066968	4	47,708,231	CORIN	0.838	0.147	0.017
CT-DMP	NK-cells	cg06497682	8	49,196,492	–	1.000	0.242	0.017
CT-DMP	Neutrophils	cg05721382	15	38,519,181	–	0.096	0.015	0.033
CT-DMR	Neutrophils	–	11	6,290,951–6,294,024	CCKBR	–	–	0.033
CT-DMR	NK-cells	–	16	3,061,835–3,064,377	CLDN9	–	–	0.000

CT-DMP, Cell type specific differentially methylated positions; CT-DMR, Cell type differentially methylated regions; NK-cells, Natural killer cells; FDR_emp, False discovery rate adjusted p-value.

The table presents CT-DMPs and CT-DMRs in MOH-patients compared to controls (episodic migraine patients and healthy controls).

Directly associated with a gene, we found a significant higher methylation level of a CpG site in NK cells, cg04066968 (chr4:47708231, FDR_emp adjusted p-value = 0.017) located in an intron of CORIN. MOH-patients had a higher methylation level in neutrophiles at chr11: 6,290,951–6,294,024 (FDR_emp adjusted p-value = 0.033), located in the gene body and 3′UTR of CCKBR, while in NK-cells, MOH-patients had a higher methylation level at chr16: 3,061,835–3,064,377 (FDR_emp adjusted p-value = 0, i.e., a p-value smaller than any p-value obtained in the 30 genome wide permutations) located in the gene body and 5′UTR transcription start site of CLDN9.

Association of reduction in headache days with methylation

The change in headache days (%) was not associated with significant genome-wide changes in methylation levels within the bulk tissue nor any of the cell types (neither DMPs nor DMRs). Furthermore, none of the previously associated CT-DMPs or CpG sites within the CT-DMRs were even marginally associated with headache days in the same or alternative cell types.

Discussion

Knowledge on the pathophysiology of MOH is limited, as well as the ability to predict treatment outcome. Therefore, it is of high clinical importance to search for diagnostic and prognostic biomarkers, and results from this pilot study may help in generating hypotheses for future research. We found that MOH-patients had DMPs compared to controls (i.e. EM and HC), but these were associated with specific cell types and not detected in bulk tissue. These CT-DMPs did not change over time in relation to treatment (i.e. reduction in headache days). Unanticipated, we also found a difference in several DNAm based eBCPs and NLR between MOH-patients and controls, which changed with reduced headache frequency potentially due to MOH treatment.

Potential role of the immune system in the MOH pathogenesis

The higher NLR in MOH-patients compared to controls could be consistent with potential involvement of the immune system in the pathogenesis and probably an increased inflammatory state. The proportion of neutrophils in the MOH-groups was higher, while the proportions of NK-cells, monocytes and CD8+ T cells were lower, compared to both control groups. Furthermore, the increase in CD4+ T cells and reduction in neutrophils and NLR were associated with reduced headache frequency after MOH-treatment.

Based on our results, it could be hypothesized that a potential increased immunological response is a counteraction of the constant suppression of the immune system, caused by a frequent use of antipyretic and anti-inflammatory drugs. Approximately 60% of our MOH-patients overused paracetamol and/or NSAIDs, while the other 40% overused triptans; all of which are known to affect the immune system (30 –32).

Previous studies have reported changes in the immune cell levels among MOH-patients. A higher lymphocyte count was found in MOH-patients compared to episodic migraine patients, but no significant difference in the lymphocyte count between MOH-patients and patients with chronic migraine was seen (33). Though we only have estimated cell-type proportions and not data on count level, disabling direct comparisons, we found a contradictory trend with lower lymphocyte proportions in MOH-patients versus controls. Another study reported a change in leukocyte function during treatment of MOH in gene expression levels (e.g., NK cell-mediated cytotoxicity, and lymphocytic cell signaling) (34).

It has been hypothesized that part of the initiation of a migraine attack could be an inflammatory response in the meningeal regions (35), and NLR and leukocyte count were reported to be elevated during the specific migraine attack (36). One of the clinical features of MOH is that the threshold for a new headache attack is lowered, and headache attacks become more migraine-like, mimicking a chronic migraine condition (37). Thus, it may be hypothesized that a more constantly activated immune system may contribute to decreasing the threshold for initiation of migraine-like attacks, resulting in more frequent attacks. Further investigations are needed to clarify the role of the immune system in MOH.

DNA methylation in patients with medication overuse headache

Since blood cell proportions were associated with MOH, we assessed cell type specific differences in methylation level between MOH patients and controls. We identified three genes with genome-wide significant cell type specific differential methylation in their markers/regions; methylation levels in CCKBR were increased for MOH-patients in neutrophils, and methylation levels in CLDN9 and CORIN were increased for MOH-patients in NK-cells. Noteworthy, it has been reported that triptans may specifically affect NK-cells and neutrophils (32).

CCKBR, encoding a G protein-coupled receptor for cholecystokinin and gastrin, has an important role in neurotransmission in the central nervous system, e.g., regulating dopamine activity. It has been shown to be associated with morphine tolerance and cocaine abuse (38,39). Functional cerebral changes in the dopamine circuit have previously been reported in imaging studies of patients with MOH (40), which supports our findings in the blood profiles of patients. However, it is still debated and controversial to view MOH as an addiction disorder.

CLDN9, a member of the claudin family, encodes a protein in tight junctions and thereby regulates cell permeability. Claudins are involved in the blood-brain barrier, but the specific function for claudin 9 has not yet been revealed. It could be hypothesized that an affected permeability could be involved in the pathophysiology of developing medication overuse. It has previously been speculated whether the permeability of the blood-brain barrier increases during an attack of migraine with aura, but this hypothesis has been challenged by an imaging study on humans showing an intact blood-brain barrier during attacks (41). To our knowledge, it has not been investigated if a change in the permeability in the blood-brain barrier is involved in MOH so further research is needed.

CORIN, encoding a serine peptidase, plays a role in cardiac hormonal functioning regulating blood volume and pressure. Interestingly, 21% of our patients reported hypertension as a co-morbidity at baseline (data not shown) despite a mean age on 42.5 years. Furthermore, anti-hypertensive drugs (e.g., beta-blockers, candesartan, lisinopril) as preventive headache treatment is very common and, for many patients, effective in reducing the headache frequency.

To date, no studies have been performed investigating methylation in MOH, however, one study investigated methylation in leukocytes in chronic migraine (15). Likely due to limited power, they were unable to find epigenome-wide significant methylated markers, but they did suggest roles for synaptic plasticity, glutamatergic activity, calcium ion binding and cell adhesion. Our study did not support these observations.

Methylation alters DNA accessibility over a relatively short time affecting the level of gene expression and, for example, have been seen in response to changes in environment or other diseases, such as cancer (10). We, therefore, hypothesized that methylation changes may occur during treatment of MOH and would be able to explain why some patients do not recover from medication overuse and/or return to episodic migraine patients. However, we did not find any changes in methylation to follow the reduction in headache days observed in the patients during the six months treatment of MOH. This raises the question, whether the methylation differences we find in patients with MOH is not a consequence of the medication overuse, but instead a risk factor for development of MOH, e.g. an inherited predisposition or sensitivity towards overuse. Alternatively, manifestation of methylation changes may take more than six months as assessed here.

Strengths and limitations

Our study was performed on a carefully phenotyped cohort, consisting of a representative group of patients with overuse of over-the-counter analgesics and/or prescription medicine, as well as two well-characterized control groups. We obtained phenotypes and blood samples using both a case-control design and a longitudinal design, which is a highly demanding task requiring a structured and organized set-up. As the study was part of a randomized controlled trial, the patients underwent three distinct pain management strategies. Though no effect was found on the outcome of interest, this may have affected the epigenetic profiles. The sample size was, however, too small to analyze separately. Inclusion of EM-controls without medication overuse is a strength since up to 90% of MOH-patients formerly had episodic migraine without medication overuse before developing chronic headache and MOH; and approximately 70% revert to episodic migraine without medication overuse after treatment. However, as all MOH-patients had chronic headache, we are unable to disentangle whether the methylation signal and immune response were caused by the chronic headache condition or the medication overuse. In future studies, patients with chronic headache without medication overuse should be included as an additional control group.

Generally, MOH-patients are a heterogenous group, with different type of medication overuse. The heterogeneity in the MOH-group may well influence the results and could even explain the larger variance of the NLR in the MOH-group. The current study lacks power to address this notion. Due to ethical considerations and feasibility in a tertiary center we allowed a stable use of preventive medication in our migraine control group; however only 3 patients used preventatives, so it is unlikely that it have affected the outcome.

No changes in methylation during the six months treatment of MOH were found. This may be due to statistical limitations and lack of power. Moreover, we investigated blood samples instead of nervous tissue due to practical limitations in human studies. Though overlap between methylation in blood and brain tissue is not complete, there may be relevant overlap in the methylation markers between the tissues (42). While we only found significant associations in the cell type specific analysis, these will need to be reproduced in sorted cell types to confirm their differential methylation status in MOH.

Conclusion

This hypothesis-generating pilot study indicated that MOH-patients had a higher level of inflammatory response compared to patients with episodic migraine and healthy controls. Moreover, the inflammatory response was reduced in association with a lower headache frequency, following intervention treatment of MOH. In MOH-patients, we found three interesting methylation markers/areas in cell-types active in the innate immune response. Even though replication studies are needed to confirm findings, this study contributes with new and very promising hypothesizes to be explored in the MOH field.

Article highlights

Patients with MOH presented with higher neutrophile-lymphocyte ratio (NLR) in MOH-patients compared to controls (patients with episodic migraine and healthy controls), indicating a higher immunological response in MOH-patients.

A lower NLR was associated with reduction in headache frequency after treatment of MOH.

Three genes (CORIN, CCKBR and CLDN9) were higher methylated in MOH-patients than in controls.

Supplemental Material

sj-jpg-1-cep-10.1177_03331024221147482 - Supplemental material for DNA-methylation and immunological response in medication overuse headache

Supplemental material, sj-jpg-1-cep-10.1177_03331024221147482 for DNA-methylation and immunological response in medication overuse headache by Louise Ninett Carlsen, Christine Søholm Hansen, Lisette J. A. Kogelman, Thomas Mears Werge, Henrik Ullum, Jonas Bybjerg-Grauholm, Thomas Folkmann Hansen and Rigmor Højland Jensen in Cephalalgia

Supplemental Material

sj-pdf-2-cep-10.1177_03331024221147482 - Supplemental material for DNA-methylation and immunological response in medication overuse headache

Supplemental material, sj-pdf-2-cep-10.1177_03331024221147482 for DNA-methylation and immunological response in medication overuse headache by Louise Ninett Carlsen, Christine Søholm Hansen, Lisette J. A. Kogelman, Thomas Mears Werge, Henrik Ullum, Jonas Bybjerg-Grauholm, Thomas Folkmann Hansen and Rigmor Højland Jensen in Cephalalgia

Footnotes

Acknowledgement

The authors express their gratitude to medical doctors Mia Nielsen and Ida Engelstoft, nurses Mette Bisgaard, Malene Danø, Annette Rasmussen, and Annette Jonasson for inclusion of patients and data collection. We thank Mette Munk and Dea Adamsen for excellent laboratory assistance.

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

We thank TrygFonden and Danish Capital Region for funding.

ORCID iDs

Louise Ninett Carlsen

Lisette J. A. Kogelman

References

Headache Classification Committee of the International Headache Society (IHS). The International Classification of Headache Disorders, 3rd edition. Cephalalgia 2018; 38: 1–211.

GBD. Global, regional, and national incidence, prevalence, and years lived with disability for 301 acute and chronic diseases and injuries in 188 countries, 1990–2013: a systematic analysis for the Global Burden of Disease Study 2013. Lancet (London, England) 2015; 386: 743–800.

Diener

Dodick

Evers

, et al. Pathophysiology, prevention, and treatment of medication overuse headache. Lancet Neurol 2019; 18: 891–902.

Diener

Antonaci

Braschinsky

, et al. European Academy of Neurology guideline on the management of medication-overuse headache. Eur J Neurol 2020; 27: 1102–1116.

Tassorelli

Jensen

Allena

, et al. A consensus protocol for the management of medication-overuse headache: Evaluation in a multicentric, multinational study. Cephalalgia 2014; 34: 645–655.

Carlsen

Munksgaard

Jensen

, et al. Complete detoxification is the most effective treatment of medication-overuse headache: A randomized controlled open-label trial. Cephalalgia 2018; 38: 225–236.

Kristoffersen

Straand

Vetvik

, et al. Brief intervention by general practitioners for medication-overuse headache, follow-up after 6 months: a pragmatic cluster-randomised controlled trial. J Neurol 2016; 263: 344–353.

Katsarava

Muessig

Dzagnidze

, et al. Medication overuse headache: rates and predictors for relapse in a 4-year prospective study. Cephalalgia 2005; 25: 12–15.

Munksgaard

Bendtsen

Jensen

RH.

Detoxification of medication-overuse headache by a multidisciplinary treatment programme is highly effective: a comparison of two consecutive treatment methods in an open-label design. Cephalalgia 2012; 32: 834–844.

10.

Taby

Issa

JP.

Cancer epigenetics. CA 2010; 60: 376–392.

11.

Urdinguio

Sanchez-Mut

Esteller

Epigenetic mechanisms in neurological diseases: genes, syndromes, and therapies. Lancet Neurol 2009; 8: 1056–1072.

12.

Hamm

Costa

FF.

The impact of epigenomics on future drug design and new therapies. Drug Disc Today 2011; 16: 626–635.

13.

Eising

van den Maagdenberg

, et al. Epigenetic mechanisms in migraine: a promising avenue? BMC Med 2013; 11: 26.

14.

Montagna

The primary headaches: genetics, epigenetics and a behavioural genetic model. J Headache Pain 2008; 9: 57–69.

15.

Winsvold

Palta

Eising

, et al. Epigenetic DNA methylation changes associated with headache chronification: A retrospective case-control study. Cephalalgia 2018; 38: 312–322.

16.

Kwiatkowska

Bacalini

Sala

, et al. Analysis of epigenetic age predictors in pain-related conditions. Front Public Health 2020; 8: 172.

17.

Heyn

The Role of the Genetic Code in the DNA methylation Landscape Formation. In: Fraga

MFFAF

(ed) Epigenomics in Health and Disease. Elsevier, Ch. 1, 2015.

18.

IHS. The International Classification of Headache Disorders, 3rd edition (beta version). Cephalalgia 2013; 33: 629–808.

19.

Carlsen

Munksgaard

Nielsen

, et al. Comparison of 3 treatment strategies for medication overuse headache: a randomized clinical trial. JAMA Neurol 2020; 77: 1069–1078.

20.

H. ZWLPS. Comprehensive characterization, annotation and innovative use of Infinium DNA Methylation BeadChip probes, https://zwdzwd.github.io/InfiniumAnnotation (2017. accessed 01 November 2017)

21.

Pidsley

Zotenko

Peters

, et al. Critical evaluation of the Illumina MethylationEPIC BeadChip microarray for whole-genome DNA methylation profiling. Genome biology 2016; 17: 208.

22.

McCartney

Walker

Morris

, et al. Identification of polymorphic and off-target probe binding sites on the Illumina Infinium MethylationEPIC BeadChip. Genom Data 2016; 9: 22–24.

23.

Touleimat

Tost

Complete pipeline for Infinium(®) Human Methylation 450K BeadChip data processing using subset quantile normalization for accurate DNA methylation estimation. Epigenomics 2012; 4: 325–341.

24.

Rahmani

Zaitlen

Baran

, et al. Sparse PCA corrects for cell type heterogeneity in epigenome-wide association studies. Nat Methods 2016; 13: 443–445.

25.

Koestler

Jones

Usset

, et al. Improving cell mixture deconvolution by identifying optimal DNA methylation libraries (IDOL). BMC Bioinformatics 2016; 17: 120.

26.

Newman

Liu

Green

, et al. Robust enumeration of cell subsets from tissue expression profiles. Nat Methods 2015; 12: 453–457.

27.

Pedersen

Schwartz

Yang

, et al. Comb-p: software for combining, analyzing, grouping and correcting spatially correlated P-values. Bioinformatics 2012; 28: 2986–2988.

28.

Graw

Henn

Thompson

, et al. pwrEWAS: a user-friendly tool for comprehensive power estimation for epigenome wide association studies (EWAS). BMC Bioinformatics 2019; 20: 218.

29.

Dupont

Plummer

Jr.

Power and sample size calculations. A review and computer program. Control Clin Trials 1990; 11: 116–128.

30.

Bertolini

Ferrari

Ottani

, et al. Paracetamol: new vistas of an old drug. CNS Drug Rev 2006; 12: 250–275.

31.

Bolay

Durham

Pharmacology. Handbook Clin Neurol 2010; 97: 47–71.

32.

Puente

Jaque

Carrasco

, et al. Triptan drugs, natural killer cell cytotoxicity, and neutrophils pro-matrix metalloproteinase-9 secretion. Headache 2008; 48: 1482–1489.

33.

Forcelini

Dantas

Luz

, et al. Analysis of leukocytes in medication-overuse headache, chronic migraine, and episodic migraine. Headache 2011; 51: 1228–1238.

34.

Hershey

Burdine

Kabbouche

, et al. Genomic expression patterns in medication overuse headaches. Cephalalgia 2011; 31: 161–171.

35.

Ashina

Hansen

, et al. Migraine and the trigeminovascular system-40 years and counting. Lancet Neurol 2019; 18: 795–804.

36.

Karabulut

Egercioglu

Uyar

, et al. The change of neutrophils/lymphocytes ratio in migraine attacks: A case-controlled study. Ann Med Surg (Lond) 2016; 10: 52–56.

37.

Munksgaard

Bendtsen

Jensen

RH.

Modulation of central sensitisation by detoxification in MOH: results of a 12-month detoxification study. Cephalalgia 2013; 33: 444–453.

38.

Dourish

O'Neill

Coughlan

, et al. The selective CCK-B receptor antagonist L-365,260 enhances morphine analgesia and prevents morphine tolerance in the rat. Eur J Pharmacol 1990; 176: 35–44.

39.

Crespi

The role of cholecystokinin (CCK), CCK-A or CCK-B receptor antagonists in the spontaneous preference for drugs of abuse (alcohol or cocaine) in naive rats. Methods Find Exp Clin Pharmacol 1998; 20: 679–697.

40.

Ferraro

Grazzi

Muffatti

, et al. In medication-overuse headache, fMRI shows long-lasting dysfunction in midbrain areas. Headache 2012; 52: 1520–1534.

41.

Hougaard

Amin

Christensen

, et al. Increased brainstem perfusion, but no blood-brain barrier disruption, during attacks of migraine with aura. Brain 2017; 140: 1633–1642.

42.

Walton

Hass

Liu

, et al. Correspondence of DNA methylation between blood and brain tissue and its application to schizophrenia research. Schizophr Bull 2016; 42: 406–414.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.08 MB

0.12 MB