Identifying oxidized lipid mediators as prognostic biomarkers of chronic post-traumatic headache.

Abstract

Chronic Post Traumatic Headache (PTH) is among the most common and disabling sequelae of traumatic brain injury (TBI). Current PTH treatments are often only partially effective and have problematic side effects. We previously showed in a small randomized trial of patients with chronic non-traumatic headaches that manipulation of dietary fatty acids decreased headache frequency, severity, and pain medication use. Pain reduction was associated with alterations in oxylipins derived from n-3 and n-6 fatty acids, suggesting that oxylipins could potentially mediate clinical pain reduction. The objective of the present study was to investigate whether circulating oxylipins measured in the acute setting following TBI, could serve as prognostic biomarkers for developing chronic PTH. Participants enrolled in the Traumatic Head Injury Neuroimaging Classification Protocol provided serum within 3 days of TBI and were followed up at 90 days post-injury with a neurobehavioral symptom inventory (NSI) and satisfaction with-life-survey (SWLS). Liquid chromatography tandem mass spectrometry methods profiled 39 oxylipins derived from n-3 docosaheaxaenoic acid (DHA), and n-6 arachidonic acid (AA) and linoleic acid (LA).Statistical analyses assessed the association of oxylipins with headache severity (primary outcome, measured by headache question on NSI) as well as associations between oxylipins and total NSI or SWLS scores. Among oxylipins, 4-hydroxy-DHA and 19,20-epoxy-docosapentaenoate (DHA derivatives) were inversely associated with headache severity, and 11-hydroxy-9-epoxy-octadecenoate (an LA derivative) was positively associated with headache severity. These findings support a potential for DHA-derived oxylipins as prognostic biomarkers for development of chronic PTH.

Introduction

Traumatic Brain Injury (TBI) is among the most common causes of disability in young adults world-wide [10,26,53]. Post-traumatic headache (PTH), a secondary headache disorder that develops within 7 days of a TBI and often resembles a migraine or tension type headache [15], is a common sequelae of TBI. PTH resolves within 90 days in most cases, however about 40% of PTH sufferers continue to have headaches persisting >3 months, thus meeting the criteria for chronic PTH [16,20], Chronic PTH is consistently rated as one of the most consequential and disabling sequelae of TBI, with major adverse impacts on social activities, function and life satisfaction [29,47,51,53,55]. Effective treatments for PTH are lacking, in part due to incomplete understanding of the mechanisms linking acute TBI to chronic PTH [7,11,58]. Identification of prognostic and/or pharmacodynamic biomarkers for PTH could be useful for identifying susceptible subpopulations and targeting/tailoring interventions. Biomarkers associated with PTH can also potentially provide insights into the underlying mechanisms and ultimately, could suggest new targets and, approaches for treatment.

Oxylipins derived from n-3 and n-6 fatty acids as pain mediators

The discovery that non-steroidal anti-inflammatory drugs inhibit the synthesis of prostaglandins from arachidonic acid (AA) [54] identified a specific mechanism linking oxidized lipids (oxylipins) to pain. Advances in mass spectrometry have enabled detection of novel lipids that are present in tissues at nMol concentrations, including several families of oxylipins, derived from omega-6 (n-6) and omega-3 (n-3) polyunsaturated fatty acids (PUFA), that regulate inflammation and pain in preclinical models [14,35,48]. Since humans cannot synthesize double bonds in the n-3 and n-6 position of fatty acids, targeted manipulation of PUFAs by diet or supplementation could potentially alter oxylipins in an analgesic manner. In 2013, we reported, in a small randomized trial of chronic non-traumatic headache patients, that targeted manipulation of dietary n-3 and n-6 PUFA decreased headache frequency and severity, while reducing acute pain medication use [39]. Pain reduction was accompanied by alterations in oxylipins and their precursor n-3 and n-6 PUFA, suggesting that certain oxylipins could potentially mediate clinical pain reduction [38]. These findings were followed up more recently where we demonstrated that circulating concentrations of an oxylipin derived from linoleic acid (LA - a dietary n-6 PUFA) negatively correlated with clinical pain reductions [37]. Additionally, we reported that these oxylipins were elevated in inflamed human tissue, and were involved in nociception in preclinical experiments [37].

Recognizing the growing prevalence of TBI and PTH and the large gaps in the understanding of PTH pathogenesis, we aimed to investigate a potential link between oxylipins and the development of chronic PTH. We hypothesized that oxylipins derived from docosahexaenoic acid (DHA, an n-3 PUFA found in fatty fish), which have been linked to antinociception in preclinical models [30,60,64], would be inversely associated with PTH severity. We measured oxylipins in serum collected during the acute phase following TBI and examined the relationship between oxylipin concentrations and PTH severity 90 days post-TBI. Results from this analysis identified three oxylipins as candidate biomarkers for chronic PTH development.

Methods

Subject Recruitment and Enrollment

This report analyzed a subset of subjects who were enrolled in the Traumatic Head Injury Neuroimaging Classification (“THINC”, NCT01132937). “THINC” is an IRB approved prospective study at 3-trauma centers in the Washington, DC metro area over a 2.5-year period. Patients who presented to the emergency department (ED) following an acute head trauma were enrolled into THINC if they met the following criteria: (1) high clinical suspicion of non-penetrating acute TBI, as determined by the ED physician and for which a head computed tomography (CT) scan was performed, (2) age ≥18 years, (3) interval between time of injury and enrollment <48h, (4) and ability to obtain informed consent from the subject or a legally authorized representative. Data and samples were collected by the investigators of “THINC”-NCT01132937. No research procedures were performed prior to written informed consent. All research subjects included in this analysis agreed to share their data and biospecimens for future research. Authors of this work do not have access to identifiers, and the results of this analysis cannot be linked to specific research participants. Investigators on the collection protocol are aware of this secondary research, have agreed to of the use of the data/sample, but were not scientifically engaged in the collaboration and are, therefore, not listed as authors. We have appropriately acknowledged their contribution.

Blood samples and neurobehavioral symptom inventory (NSI) were collected from eligible patients. Additionally, Glasgow Coma Scale (GCS), potential amnesia, loss of consciousness, relevant history, and clinical diagnoses were entered into standardized case report forms by the THINC team. All patients (or their legally authorized representatives) gave written informed consent for the use of their clinical data and blood samples in accordance with a protocol approved by the Institutional Review Board of the National Institute of Neurological Disorders and Stroke (NINDS). Both blood samples and de-identified demographic and clinical data were provided to the Center for Neuroscience and Regenerative Medicine (CNRM) Biorepository and approval was obtained for use for this project. THINC protocol follow-up included administration of the NSI and the Satisfaction with Life Scale (SWLS). For this report we obtained available serum samples from subjects that had (1) post-injury serum sample collected within 3 days of injury and (2) completed follow-up appointments 90 days post-injury (n=121, Figure 1).

Figure1. Experimental design for the study primary objective and the biosynthetic pathways for oxidized lipid mediators measured by liquid chromatography-tandem mass spectrometry (LC-MS/MS).

(A) Serum samples from patients who suffered from a TBI was analyzed with LC-MS/MS to determine which oxidized lipid mediators were associated with headache severity at 90-days post TBI. (B) the chemical structures of 37 of the 39 oxidized lipid mediators measured by our LC-MS/MS assay (9-hydroxy-octadecatrienoic acid and 18-hydroxy-eicosapentaenoic acid are omitted from the figure). Oxidized lipid mediators are synthesized from n-6 (top panel) or n-3 (bottom panel) PUFA (LA, AA, ALA, EPA, DHA). An arrow connecting lipids symbolizes that one of these lipids can be synthesized from the other either enzymatically or non-enzymatically, with the direction of the reaction indicated by the direction of the arrow. Only oxidized lipids that are measured by our LC-MS/MS assay are depicted with their chemical structures.

TBI, traumatic brain injury; LC-MS/MS, Liquid Chromatography-Tandem Mass Spectrometry; NSI, Neurobehavioral Symptom Inventory; SWLS, Satisfaction with Life Survey; n-6, omega-6; n-3, omega-3; PUFA, polyunsaturated fatty acid; LA, linoleic acid; AA, arachidonic acid; ALA, alpha-linolenic acid; EPA, eicosapentaenoic acid; DHA, docosahexaenoic acid; 5-HETE,5-hydroxy-eicosatetraenoic acid; 12-HETE, 12-hydroxy-eicosatetraenoic acid; 15-HETE, 15-hydroxy-eicosatetraenoic acid; 5-oxoETE, 5-keto-eicosatetraenoic acid; LTB4, Leukotriene B4; LXA4, Lipoxin A4; PGE2, Prostaglandin E2; PGF2, Prostaglandin F2; 8-isoPGF2, 8-iso-Prostaglandin F2α; TXB2, Thromboxane B2; 4-HDHA, 4-hydroxy-docosahexaenoic acid; 7-HDHA, 7-hydroxy-docosahexaenoic acid; 10-HDHA,10-hydroxy-docosahexanoic acid; 14-HDHA, 14-hydroxy-docosahexaenoic acid; 17-HDHA, 17-hydroxy-docosahexaenoic acid; 16-EDP, 16-epoxy-docosapentaenoic acid; 19,20-EDP, 19-epoxy-docosapentaenoic acid; MAR1, Maresin1; RVD1, Resolvin D1; PDX, Protectin Dx; PD1, Protectin D1; 9-HODE, 9-hydroxy-octadecenoic acid; 13-HODE, 13-hydroxy-octadecenoic acid; 9-oxoODE, 9-oxo octadecenoic acid; 13-oxoODE, 13-oxo-octadecenoic acid; 9-epOME, 9-epoxy-octadeecnoic acid; 12-epOME, 12-epoxy-octadecenoic acid; 9,10-diHOME, 9,10-dihydroxy-octadecenoic acid; 12,13-diHOME, 12,13-dihydroxy-octadecenoic acid; 9,10,11-triHOME, 9,10,11-trihydroxy-octadecenoic acid; 9,12,13-triHOME, 9,12,13-trihydroxy-octadecenoic acid; 9,10,13-triHOME, 9,10,13-trihydroxy-octadecenoic acid; 9-H-12-E-LA, 9-hydroxy-12-epoxy-octadecenoic acid; 11-H-12-E-LA, 11-hydroxy-12-epoxy-octadecenoic acid; 11-H-9-E-LA, 11-hydroxy-9-epoxy-octadecenoic acid; 13-H-9-E-LA 13-hydroxy-9-epoxy-octadecenoic acid; 9-K-12-E-LA, 9-oxo-12-epoxy octadecenoic acid; 5-HpETE, 5-hydroperoxy-eitosatetraenoic acid; 12-HpETE, 12-hydroperoxy-eitosatetraenoic acid; 15-HpETE, 15-hydroperoxy-eitosatetraenoic acid; 9-HpODE, 9-hydroperoxy-octadecenoic acid; 13-HpODE, 13-hydroperoxy-octadecenoic acid; 4-HpDHA, 4-hydroperoxy-docosahexanoic acid; 7-HpDHA, 7-hydroperoxy-docosahexanoic acid; 14-HpDHA, 14-hydroperoxy-docosahexanoic acid; 17-HpDHA, 17-hydroperoxy-docosahexanoic acid;

Neuroimaging

All subjects obtained non-contrast head CT scans as part of usual clinical care. The clinical CT scans were interpreted by staff radiologists at the participating hospitals, upon hospital admission. Reports were reviewed by study personnel and findings related to trauma were recorded.

TBI Classification

TBI classification was based off of the American Congress of Rehabilitation Medicine definition of mild TBI [19,49]. TBI was classified into 4 categories (silent, mild, moderate, severe) based on a combination of GCS, amount of time the patient was unconscious, amount of post-traumatic amnesia time. A subject was classified as having severe TBI if they met any of the following criteria: 1) GCS between 3 and 8; or 2) >24 hours of both post-traumatic amnesia and >24 hours lost consciousness. If a subject did not meet any criteria for severe TBI they were classified as having a moderate TBI if they presented with either 1) GCS between 9 and 12; 2) >24 hours of post-traumatic amnesia; or 3) 0.5-24 hours of lost consciousness. If a subject did not meet any criteria for a moderate TBI they were assessed as having a mild TBI if they presented with either 1) GCS of 13-15; 2) <24 hours of post-traumatic amnesia; or 3) less than 30 mins of lost consciousness. A patient was classified as having a silent TBI if they did not meet any criteria for severe, moderate or mild TBI.

Blood Sampling

Average time from injury to blood collection was 16 hours and all samples were collected within 61.5 hours after injury. Blood was collected from an 18G antecubital vein catheter into serum separator tubes and stored at 4 °C until serum isolation.

Serum Processing

The serum processing, solid phase extraction (SPE) and LC-MS/MS methods have been described in detail [37,52,63]. Briefly, patient serum (200 μL) was thawed and aliquoted into 500 μL ice cold methanol containing 0.02% butylated hydroxy toluene (BHT) and 0.02% ethylenediaminetetraacetic acid (EDTA) and 10 ng/mL of deuterated internal standards. Samples were then incubated at −80°C for 1 hour, to precipitate proteins, and then centrifuged at 17000 x g for 10 min at 4°C to create a protein pellet. The supernatant was collected and stored under N₂ gas at −80°C overnight.

Solid Phase Extraction

Methanol extracts of serum samples were diluted in 6 mL of ice cold, ultrapure water and loaded onto SPE columns (Strata™-X 33 μm Polymeric Reversed Phase, 200 mg, Phenomenex, CA) that had been conditioned with methanol, followed by water. Once the samples were loaded onto SPE columns, the columns were cleaned with 10% methanol and allowed to dry for 2 min. Oxidized lipids were eluted, with methanol containing 0.0004% BHT, into glass culture tubes containing 10 μL of 30% glycerol in methanol. Samples were then concentrated by completely evaporating the methanol under N₂ gas and reconstituting into 40 μL HPLC grade methanol. The whole sample was then transferred into a 250 μL glass insert that was fit into a 2 mL amber vial (Agilent, CA) for LC-MS/MS analysis.

Liquid Chromatography Tandem-Mass Spectrometry (LC-MS/MS)

Detailed descriptions of the LC-MS/MS conditions and parameters have previously been described by our group [63]. Data processing was performed with Analyst Software™ (version 1.6.3), with the limit of quantitation (LOQ) for each analyte assessed by manual peak inspection in explorer mode of the Analyst Software™. The LOQ was defined as the lowest point of the standard curve, or lowest sample with a signal to noise ratio (S/N) of greater than 5. The LOQ was first established by using the Analyst Software calculated concentration of the sample (i.e. finding the lowest concentration sample with S/N >5). Next samples with analyte peak areas lower than that of the sample defining the LOQ were manually inspected to confirm a S/N ≥5. For analytes with an 18-carbon backbone, samples with calculated concentrations or peak areas lower than the lowest standard within the linear range of our calibration curve were classified as below the limit of quantitation (BLQ) and the LOQ was defined as the concentration of the lowest standard curve sample within the linear range. Calibration curves tended to be linear below the LOQ for analytes with a backbone greater than 18 carbons. Samples determined to be BLQ were imputed to have a concentration of ½ LOQ. After chromatogram inspection we determined that the conditions of our LC-MS/MS were unstable during 12 sample runs (an approximately 12-hour window that occurred sequentially): these samples exhibited large shifts in peak retention times, that were not predictable by the shifting of the internal standard peaks. Due to these unpredictable shifting retention times we were not able to identify analytes of interest with confidence. Therefore, samples that were run during the abnormal LC-MS/MS conditions were classified as missing completely at random and excluded from the analysis. This designation—missing completely at random—indicates a situation in which the missing values are unlikely to be related to outcome variables [24].

Neurobehavioral Symptom Inventory and Satisfaction with Life Scale

The primary outcome was headache severity which was extracted from the NSI. The NSI is a 22-item questionnaire used by clinicians to quantify TBI symptom severity [5]. Each NSI question refers to a TBI related symptom (e.g., “Headaches”) and asks the patient to rate how that symptom has affected their life on a 0-4 scale (0 = “Rarely if ever present; not a problem at all” and 4 = “Almost always present and I have been unable to perform at work, school or home due to this”). Our primary endpoint and our exploratory endpoints were headache severity (headache question of the NSI) and the total NSI score, respectively, with both assessed at 90 days post TBI. The headache question of the NSI has been reported to be well correlated with the Headache Impact Test (HIT-6) [23]. We also explored the relationship between oxylipins following TBI and life satisfaction at 90-days post TBI using the SWLS. SWLS is a 5-item questionnaire that is used to assess global life satisfaction [8]. Each item on the SWLS was rated by the participant on a scale of 1 (strongly disagree) to 7 (strongly agree) with a total score between 26 and 30 indicating the participant was satisfied with their life. The SWLS has been shown to be appropriate for a wide range of applications [33], including evaluation of quality-of-life in TBI populations [2].

Using archival samples to investigate whether candidate biomarkers are pharmacodynamic

To examine the potential for targeted dietary (n-3 and n-6 PUFA) manipulations to modify candidate biomarkers identified in this report, we measured concentrations of oxylipins, using the above described methods, in archived baseline and post-intervention plasma samples (n=44) collected from the chronic daily headache trial (NCT01157208). The methods (including CONSORT diagram) and results of this small randomized trial have been published elsewhere [38-40]. Briefly, a high n-3 plus low n-6 dietary PUFA intervention decreased headache frequency and severity, while increasing precursor n-3 and n-6 fatty acids and selected oxylipins. Here we quantified an expanded list of n-3 and n-6 PUFA derived oxylipins to investigate whether circulating levels of these candidate biomarkers are modifiable by diet and to explore whether diet-induced changes correlate with clinical pain reduction.

Statistics

We excluded 12 subjects with abnormal MS as missing completely at random. Missing oxylipin values were imputed with ½ of the limit of quantitation. We investigated the effect of each serum post-injury oxylipin on each outcome at day 90 using three types of models: ordinal regression for NSI headache severity, negative binomial regression for NSI total score, and linear regression for the total SWLS. Each model was adjusted for age, gender, race, ethnicity, observed CT scan abnormalities that were consistent with TBI, THINC TBI rating (silent, mild, moderate, or severe), and time from injury to blood draw as well as the log sum of oxylipins from each precursor fatty acid (LA, DHA, and ARA) excluding the oxylipin in question. Model variables were chosen a priori. Analyses were performed in Stata 15 (StataCorp. 2017. Stata Statistical Software: Release 15. College Station, TX: StataCorp LLC.).

Results

Population Characteristics

Demographics and clinical characteristics for our TBI study population are shown in Table 1. Patients enrolled in this study were 69% male and 74% White. Most of the injuries were classified as a mild TBI (71%), with only a single incidence of severe TBI. Demographics and clinical characteristics of the chronic daily headache cohort have been previously published [39].

Table 1.

Population Demographics and Clinical Characteristics (n=121)

Characteristic	Mean or Count	SD or %
Years of Age	48	18
Male	84	69%
Race
Black	27	22%
White	89	74%
Asian	4	3%
Native American	1	1%
Latino or Hispanic	19	16%
Hours from Injury to Blood Draw	16	12
TBI Found on CT Scan	38	31%
THINC TBI Rating
Silent	26	21%
Mild	86	71%
Moderate	8	7%
Severe	1	1%

SD, standard deviation; TBI, Traumatic Brain Injury; CT, Computed Tomography; THINC, Traumatic Head Injury Neuroimaging Classification

Post-Traumatic Headache severity decreases with time

Median headache severity declined over time (Figure 2a). In the acute phase following TBI, 78% of participants reported having any headache, while only 40% did at 90 days post-injury.

Figure 2. Three oxylipins are associated with chronic headache severity.

(A) Immediately following TBI, median headache severity was 3 out of 4 and it declined over time. (B) Three oxylipins were significantly associated with headache severity at 90 days post-TBI. There was a higher probability of having no headaches at 90 days with higher serum concentrations of (C) 4-HDHA and (D) 19,20-EDP (p=0.005 and p=0.019, respectively) and lower concentrations of (E) 11-H-9-E-LA (p=0.031). Standardized coefficients (B) and probabilities (C) are based on ordinal logistic regression of day 90 NSI headache question on each respective serum post-injury mediator, adjusted for the sums of LA, DHA, and AA mediators, age, gender, race, ethnicity, traumatic brain injury, and time from injury to blood draw. This analysis excludes 12 subjects with abnormal mass spectrometry (n=109).

IQR, interquartile range; NSI, neurobehavioral symptom inventory; TBI, Traumatic Brain Injury; V. Severe, very severe; 4-HDHA, 4-hydroxy-docosahexaenoic acid; 19,20-EDP, 19-epoxy-docosapentaenoic acid; 11-H-9-E-LA, 11-hydroxy-9-epoxy-octadecenoic acid; ALA, alpha-linolenic acid; LA, linoleic acid; EPA, eicosapentaenoic acid; DHA, docosahexaenoic acid; AA, arachidonic acid;

Concentrations of oxidized lipid mediators in serum

We profiled for 39 oxylipins, 27 of which were detected in a sufficient number of samples to include in our analysis (Figure 1). These include several classic AA-derived lipid mediators of pain and inflammation (i.e., PGE2, TXB2, LTB4), two DHA-derived oxylipins (14-HDHA and 17-HDHA) that are pathway precursors for specialized pro-resolving lipid mediators with anti-nociceptive properties in preclinical models, and several LA-derived oxylipins that are reported to have pro-nociceptive properties in preclinical models.

Serum oxylipin concentrations are listed in Supplementary Table 1. In general, oxylipins derived from n-6 PUFA (AA or LA) were more highly concentrated than n-3 PUFA derived oxylipins (Supplementary Table 1). Of the classic mediators of pain and inflammation that we were able to measure, TXB2 was approximately 30 and 60-fold more concentrated in serum compared to LTB4 and PGE2, respectively (Supplementary Table 1). Notably, 8-iso Prostaglandin F2α (8-IsoPGF2a), a marker of non-enzymatic peroxidation, was not detected in any sample.

Oxylipins associated with headache severity following TBI

Only three of the oxylipins measured were significantly associated with headache severity at 90 days post-TBI (Figure 2b). Two DHA-derived oxylipins, 4-HDHA and 19,20-EDP (Figure 2c and d), were inversely associated with headache severity (p≤0.019). One LA-derived oxylipin, 11-H-9,10-E-LA, was positively associated with headache severity (p<0.031 Figure 2e). To put this in context of the serum concentrations observed, a one standard deviation increase in 4-HDHA and 19,20-EDP was associated with a 70 and 60% decrease, respectively, in the odds of having headache at 90 days, while a one standard deviation increase in 11-H-9,10-E-LA was associated with a 90% increase.

Oxylipins associated with overall traumatic brain injury symptoms and life satisfaction

Associations between serum oxylipins after TBI and total NSI score, which measures TBI symptom severity, 90 days after TBI were weak (figure 3). 13-HODE (p=0.037) and 4-HDHA (p=0.048) were the only oxylipins that were associated with NSI score at a p<0.05 level [β (95% CI) = −0.38 (−0.73, −0.02) and −0.33 (−0.66, −0.00), respectively).

Figure 3. Association of post-injury oxylipins and NSI total score at day 90 (n=109).

Acute serum concentration of 1 oxylipin (13-HODE) was inversely associated with total NSI. Incident ratios were estimated using negative binomial regression of the day-90 NSI score on each respective serum post-injury oxylipin, adjusted for age, gender, race, ethnicity, traumatic brain injury (THINC and CT scan) and time from injury to blood draw. 12 subjects with abnormal MS were excluded. Coefficients are adjusted for log sums of LA, DHA and ARA derived oxylipin concentrations, excluding the mediator in question. 5-HETE,5-hydroxy-eicosatetraenoic acid; 12-HETE, 12-hydroxy-eicosatetraenoic acid; 15-HETE, 15-hydroxy-eicosatetraenoic acid; 5-oxoETE, 5-keto-eicosatetraenoic acid; LTB4, Leukotriene B4; PGE2, Prostaglandin E2; TXB2, Thromboxane B2; 4-HDHA, 4-hydroxy-docosahexaenoic acid; 10-HDHA,10-hydroxy-docosahexanoic acid; 14-HDHA, 14-hydroxy-docosahexaenoic acid; 17-HDHA, 17-hydroxy-docosahexaenoic acid; 19,20-EDP, 19,20-epoxy-docosapentaenoic acid; 9-HODE, 9-hydroxy-octadecenoic acid; 13-HODE, 13-hydroxy-octadecenoic acid; 13-oxoODE, 13-oxo-octadecenoic acid; 9,10-epOME, 9-epoxy-octadeecnoic acid; 12,13-epOME, 12-epoxy-octadecenoic acid; 9,10-diHOME, 9,10-dihydroxy-octadecenoic acid; 12,13-diHOME, 12,13-dihydroxy-octadecenoic acid; 9,10,11-triHOME, 9,10,11-trihydroxy-octadecenoic acid; 9,12,13-triHOME, 9,12,13-trihydroxy-octadecenoic acid; 9,10,13-triHOME, 9,10,13-trihydroxy-octadecenoic acid; 11-H-12-E-LA, 11-hydroxy-12-epoxy-octadecenoic acid; 11-H-9-E-LA, 11-hydroxy-9-epoxy-octadecenoic acid; 13-H-9-E-LA, 13-hydroxy-9-epoxy-octadecenoic acid.

In another exploratory analyses, post-injury 12,13 DiHOME and 14-HDHA were associated [β (95% CI) = −0.39 (−0.67, −0.11) and 0.36 (0.06, 0.65), respectively) with SWLS (figure 4) at 90 days post-injury (p=0.007 and 0.018, respectively).

Figure 4. Association of post-injury lipid mediators and satisfaction with life score (SWLS) at day 90 (n=109).

Standardized coefficients were estimated using negative linear regression of the day-90 SWLS score on each respective serum post-injury oxylipin, adjusted for age, gender, race, ethnicity, traumatic brain injury (THINC and CT scan) and time from injury to blood draw. 12 subjects with abnormal MS were excluded. Coefficients are adjusted for log sums of LA, DHA and ARA derived oxylipin concentrations, excluding the mediator in question. 5-HETE,5-hydroxy-eicosatetraenoic acid; 12-HETE, 12-hydroxy-eicosatetraenoic acid; 15-HETE, 15-hydroxy-eicosatetraenoic acid; 5-oxoETE, 5-keto-eicosatetraenoic acid; LTB4, Leukotriene B4; PGE2, Prostaglandin E2; TXB2, Thromboxane B2; 4-HDHA, 4-hydroxy-docosahexaenoic acid; 10-HDHA,10-hydroxy-docosahexanoic acid; 14-HDHA, 14-hydroxy-docosahexaenoic acid; 17-HDHA, 17-hydroxy-docosahexaenoic acid; 19,20-EDP, 19,20-epoxy-docosapentaenoic acid; 9-HODE, 9-hydroxy-octadecenoic acid; 13-HODE, 13-hydroxy-octadecenoic acid; 13-oxoODE, 13-oxo-octadecenoic acid; 9,10-epOME, 9-epoxy-octadeecnoic acid; 12,13-epOME, 12-epoxy-octadecenoic acid; 9,10-diHOME, 9,10-dihydroxy-octadecenoic acid; 12,13-diHOME, 12,13-dihydroxy-octadecenoic acid; 9,10,11-triHOME, 9,10,11-trihydroxy-octadecenoic acid; 9,12,13-triHOME, 9,12,13-trihydroxy-octadecenoic acid; 9,10,13-triHOME, 9,10,13-trihydroxy-octadecenoic acid; 11-H-12-E-LA, 11-hydroxy-12-epoxy-octadecenoic acid; 11-H-9-E-LA, 11-hydroxy-9-epoxy-octadecenoic acid; 13-H-9-E-LA, 13-hydroxy-9-epoxy-octadecenoic acid.

Analysis of archived serum samples from 2013 dietary headache intervention trial

The primary results of the chronic daily headache diet trial dietary headache were published in 2013 [38,39]. Briefly, a 12-week intervention altering n-3 and n-6 PUFA led to clinically meaningful reductions in headache impact and severity [39]. Here we found that these interventions significantly increased plasma concentrations of 4-HDHA (p=0.03) and 19,20-EDP (p=0.002) (figure 5). Additionally, diet-induced changes in both 4-HDHA and 19,20-EDP were inversely associated with headache hours per day, headache days per month and headache impact test (HIT-6) (figure 5, p<0.007). The results of this analysis for 11-H-9,10-E-LA have been published elsewhere [37]. Briefly, diet did not significantly alter circulating 11-H-9,10-E-LA and this oxylipin was not associated with any headache outcome measured [37]. With respect to other oxylipins measured in this analysis, we observed that DHA derived oxylipins were inversely associated with headache impact and severity while only 17-HDHA was appreciably altered by diet (Supplementary Table 2). Notably, 8-isoPGF2a was detected in only 5 samples and median 4-HDHA and 19,20-EDP concentrations were similar to those of our PTH cohort (0.04 and 0.05 ng/ml, respectively) indicating substantial lipid peroxidation was not occurring in these samples.

Figure 5. Analysis of archival serum samples from 2013 dietary headache trial indicate that 4-hydroxy docosahexanoic acid (4-HDHA) and 19,20-epoxy docosapentaenoic acid (19,20-EDP) are modifiable by diet and support their relation to headache severity.

(A) Serum concentrations of 4-HDHA and 19,20-EDP were increased (p=0.03 and 0.002, respectively) after 12 weeks where participants were fed a low omega-6 PUFA diet, with or without increased omega-3 PUFA. After 12-weeks on their respective diet, participants (B) headache hours per day, (C) headache days per month and (D) Headache Impact Test (HIT-6) scores were inversely associated with serum oxylipin concentrations (p<0.007).

Discussion

Summary and context:

Chronic PTH is among the most common and disabling sequelae of TBI. Mechanisms underlying the transition from acute TBI to chronic PTH are poorly understood and prognostic, mechanism-based and pharmacodynamic biomarkers for chronic PTH are lacking. Since oxylipins derived from n-3 and n-6 PUFA play key roles in the amplification and resolution of inflammatory responses and nociception [14,48] and can be manipulated by diet [41], circulating oxylipins are intriguing candidates for non-invasive, mechanism-based, pharmacodynamic biomarkers with predictive power for the development and resolution of chronic pain. In the present study, we identified two oxylipins derived from DHA (4-HDHA and 19,20-EDP) measured in serum during the acute phase following TBI that were strongly, inversely associated with risk of developing chronic PTH, and one LA derived oxidized lipid mediator (11-H-9-E-LA) that was positively, but relatively weakly, associated with headache severity. These oxylipins can potentially serve as biomarkers in a clinical setting to identify TBI patients that may develop chronic PTH.

Mechanism-based biomarker candidates

The main objective of this work was to identify biomarker candidates for chronic PTH following TBI. However, plausible biological mechanisms could potentially help explain these observed associations between oxidized derivatives of DHA and LA and chronic PTH. In preclinical models, oxidized derivatives of DHA—including DHA-derived mono-epoxides [57], resolvins [30,60], protectins [46], and maresins [3,64]—are reported to have potent anti-inflammatory and anti-nociceptive effects. By contrast, in preclinical models oxidized derivatives of LA—including HODEs, DiHOMEs, and hydroxy-epoxides—are reported to have pro-nociceptive properties [12,31,32,37,65]. The exact cause of chronic PTH is unknown, however, the findings reported here indicate that the relative levels of n-3 and n-6 PUFA derived oxylipins, immediately after injury, may play a role in disease development. Future work should investigate the biological effects of oxylipins in TBI.

Pharmacodynamic biomarkers (responsive to diet)

We previously showed in a small randomized trial of patients with chronic non-traumatic headaches that targeted manipulation of dietary PUFA—increasing n-3 EPA + DHA while decreasing LA—decreased headache frequency and severity [39]. Here, using archived samples from that small trial, we expand upon those findings by demonstrating that 4-HDHA and 19,20-EDP are pharmacodynamic (i.e. altered by the targeted substrate manipulation) and that diet-induced changes in these two oxylipins were closely correlated with clinical pain reduction. These findings provide further support for these two compounds as candidate prognostic biomarkers of pain reduction. In an RCT in 60 chronic migraine sufferers, coadministration of an EPA and DHA supplement with amitriptyline (a tricyclic antidepressant with analgesic properties) decreased the number of headache days per month compared to placebo and amitriptyline [50]. However, in a larger RCT in patients with episodic migraines, fish oil supplementation provided no benefit compared to placebo [36]. Thus, collective findings from randomized trials testing dietary or supplemental provision of EPA + DHA on headache outcomes are mixed and may differ depending on accompanying pharmacological interventions or dietary changes. Two larger RCTs testing the effects of targeted substrate manipulation—increasing EPA+DHA with or without LA lowering—on pain and related endpoints in patients with frequent episodic migraines (ClinicalTrials.gov Identifier: NCT02012790 [25]) and chronic (unresolved) PTH (ClinicalTrials.gov Identifier: NCT03272399) will help clarify the biochemical and clinical consequences of these interventions and the potential utility of these oxylipins as pharmacodynamic, mechanism-based biomarkers for these conditions.

Effects of TBI on lipid metabolism

It has been reported that lipid metabolism is impacted by TBI [1,44]. In rodents, after a TBI cortical concentrations of esterified pools of DHA have been reported to decrease acutely and then gradually increase in a manner suggestive of a role of DHA containing phospholipids in injury healing and recovery [13]. Unesterified DHA is believed to be the form of DHA that most abundantly enters the brain [4] and in humans, acutely after an injury this form of DHA is increased in the plasma and CSF [9,34] suggesting an increased uptake of DHA into the brain. Interestingly, studies in mice have found that increasing plasma DHA concentrations after an injury is associated with improved TBI recovery [22] suggesting a role for DHA in treating TBI. Despite this preclinical evidence, use of n-3 PUFA to treat TBI remains mostly anecdotal [21]. Our findings that DHA derived oxylipins were associated with less headache severity and greater life satisfaction in humans support this hypothesis, however, we cannot speculate on the cause of elevated oxidized DHA derived lipids in patients with better outcomes (i.e. diet/supplement vs. genetics).

TBI has been shown to alter lipid peroxidation and oxidized lipid concentrations in blood and CSF [9,17,59,61,62]. Previous reports in humans have indicated that in CSF oxylipins derived from AA tend to be increased after TBI [9,56,59]. Results from our study do not provide support for the use of circulating AA derived oxylipins (i.e. prostaglandins, leukotrienes and mono-oxygenated AA derivatives) as prognostic biomarkers of TBI recovery. It should be noted, however, that most reports linking AA-derived oxylipins to TBI measured oxylipins in CSF of TBI cases vs controls making it impossible to compare to this report, where we analyzed oxylipins solely in serum of TBI patients.

To our knowledge this is the first report to profile LA-derived oxylipins in TBI using LC-MS/MS. A recent report found that serum LA was lower in TBI patients with cognitive impairment compared to those without cognitive impairment [61] but did not measure LA-derived oxylipins. In the present study, higher concentrations of the LA-derived oxylipins 11-H-9-E-LA and 12,13-DiHOME were associated with chronic PTH severity and lower life satisfaction recovery, respectively. However, higher concentrations of another LA-derived oxylipin (13-HODE) were associated with improved TBI recovery. Given the number of statistical comparisons, it is possible that these discordant associations are simply due to chance. Future studies are needed to better understand the potential roles of LA-derived oxylipins in TBI and PTH.

It should be stressed that oxylipins were identified as potential biomarkers of chronic PTH development, and the observational design of our study did not allow us to draw conclusions on their bioactivity. Particularly, we were unable to conclude if these oxylipins were related specifically to brain injury or general tissue injury, since TBI is often accompanied by damage to other tissues [6]. Traumatic (non-brain specific) injury can increase synthesis of oxylipins [18,43], and expression of genes encoding enzymes that catalyze oxylipin synthesis were associated with complications from traumatic injury [28]. We did not collect information relating to extent of non-brain injuries in our participants nor did we analyze serum of non-TBI patients, therefore, we are unable to conclude on the relationship of TBI with oxylipin concentrations. Additionally, oxylipins were measured acutely after TBI and not at the time of our clinical endpoint (3 months post-TBI), therefore, future study is required to draw conclusions about the bioactivity or mechanism of action of these oxylipins in TBI and headache pathophysiology. Nevertheless, we identified 2 oxylipins (4-HDHA and 19,20-EDP) acutely after TBI that were associated with the development of chronic headache. Identifying signatures that predict the transition of acute to chronic pain is of great importance clinically and scientifically, and the need for this research is reflected in the fact that an NIH Common Fund Program (The Acute to Chronic Pain Signatures Program) is dedicated to this research [27].

Limitations

Several limitations of the present work require consideration. Since this is an observational study, associations reported here should not be interpreted as causal. Our measure for headache severity was extracted from a larger questionnaire (NSI) designed to evaluate total overall brain injury symptoms and not specifically headache. However, the NSI headache question has been reported to be significantly correlated with HIT-6 (r=0.8) and to adequately capture headache severity [23]. Future work assessing this hypothesis should utilize a questionnaire specifically designed and validated to assess headache severity. Additionally, only 8% of the TBI cases in this report were rated moderate or severe. Though the majority of TBI cases are of mild severity [10], future work should aim to select for moderate and severe TBI cases. Samples were collected in a clinical setting; thus it is possible that delays between sample collection and processing to serum could have impacted oxylipin results. Previously published reports indicate that oxylipins are stable in blood stored at 4°C for up to 4 hours [42,45]. Though we were unable to record blood processing times, oxylipins generated via non-enzymatic lipid peroxidation (i.e. 8-isoPGF2a) were BLQ (LOQ = 0.1ng/ml) in all PTH cohort sample indicating that substantial auto-oxidation likely did not occur. Additionally, median oxylipin concentrations in our PTH cohort and in our archival samples were remarkably consistent with a previous report from our group that measured oxylipins in a research setting [52] indicating that blood processing or storage did not significantly impact oxylipin concentrations. Limitations that are common to mass spectrometry assays—including lack of resolution of isomers and related compounds, could potentially influence results. Finally, these results require confirmation in larger studies.

In summary, we identified two DHA-derived oxylipins as mechanism-based, pharmacodynamic biomarker candidates for predicting the development of chronic PTH following TBI. Future studies are needed to validate these findings.