Elevation of CSF Tumor Necrosis Factor α Levels in New Daily Persistent Headache and Treatment Refractory Chronic Migraine
Objective: To determine if patients with new daily persistent headache (NDPH) have elevated levels of tumor necrosis factor α (TNF α) in the CSF.
Background: NDPH is considered one of the most treatment resistant of all headache syndromes. This reflects a lack of understanding of its pathogenesis. As a certain percentage of NDPH patients have their headaches start after an infection, the possibility of a persistent state of systemic or CNS inflammation comes into question. TNF α is a proinflammatory cytokine involved in brain immune and inflammatory activities, as well as in pain initiation. The goal of this study was to look at TNF α levels in the CSF of NDPH patients, to determine if CNS inflammation may play some role in the pathogenesis of this condition.
Methods: CSF TNF α levels were studied in 38 patients: 20 with NDPH and a control population of 16 patients with chronic migraine (CM), and 2 with post-traumatic headache (PT).
Results: CSF TNF α levels were elevated in 19 of 20 NDPH patients, 16 of 16 CM patients, and both PT patients. Serum TNF α levels were normal in most of the study subjects.
Conclusion: An elevation of CSF TNF α levels was found in almost all NDPH patients and suggest a role for TNF α in the pathogenesis of this condition. Surprisingly, all CM and PT patients tested had elevated CSF TNF α levels. In most patients with elevated CSF levels, serum TNF α levels were normal. All of these syndromes may be manifestations of CNS inflammation. As most of the positive-tested patients showed minimal to no improvement during aggressive inpatient treatment, persistent elevation of CSF TNF α levels may be one of the causes of treatment refractory CDH.
New daily persistent headache (NDPH) is a recognized form of chronic daily headache. It is marked by a daily headache from onset, typically in individuals with no prior headache history. NDPH is considered one of the most treatment resistant of all headache syndromes. This reflects a lack of understanding of its pathogenesis. As about 30% of NDPH patients have their headaches start after an infection or illness, the possibility of a persistent state of systemic or CNS inflammation comes into question. Tumor necrosis factor (TNF) α is a proinflammatory cytokine involved in brain immune and inflammatory activities, as well as in pain initiation. Elevation of CSF TNF α levels has been found in a select number of neurologic conditions, suggesting CNS inflammation in those disorders. The goal of this study was to look at TNF α levels in the CSF of NDPH patients, to determine if CNS inflammation may play some role in the pathogenesis of this condition.
CSF TNF α levels were collected from NDPH patients admitted to an inpatient headache unit. This was a consecutive series of patients admitted over a 1-year time period. Patients fulfilled the International Classification of Headache Disorders-2 criteria for NDPH. All patients were deemed to have primary NDPH, as secondary causes were ruled out by neuroimaging, including brain MRI with and without gadolinium, MRA of the intracranial and extracranial circulation, and a MR venogram. CSF TNF α levels were also tested in a control group, which consisted of an unselected population of chronic daily headache (CDH) patients, admitted during the same time period, and who required a lumbar puncture as part of their evaluation. All patients had a headache at the time of CSF collection. Levels of TNF α in the CSF and serum were performed at ARUP laboratories (Salt Lake City, UT). A chemiluminescent immunoassay was utilized. CSF and serum TNF α levels less than 8.2 pg/mL are recognized as normal by ARUP laboratories. ARUP laboratories established and validated normal reference ranges for TNF α levels after screening 36 healthy volunteers (personal communication with ARUP laboratories). Thirty-eight patients were studied in all 20 patients with NDPH, 16 patients with chronic migraine (CM), and 2 patients with post-traumatic headache (PT). Patient demographics are provided in Table 1 . All patients experienced head pain on a daily basis. Medication overuse was noted in 4 NDPH patients, 11 CM patients, and 2 PT patients. All patients in analgesic rebound were using opiates on a daily basis. In those patients not in medication overuse, analgesics (NSAID, triptan, butalbital, opiate) if used were taken 2 or less days per week. Lumbar punctures were carried out in a lateral decubitus position. Spinal fluid opening pressures were measured and CSF analysis included cell count, total protein, glucose, gram stain, cryptoccocal antigen, and TNF α levels. Serum TNF α levels were collected in most patients. In the NDPH subgroup, 13 of 20 patients were able to recognize a triggering event to their headache. Nine patients stated an infection or illness was present at the time their headache began, while stress, a surgical procedure, immediately postpartum, and during a basketball game were noted in 4 other patients.
Student t test was used to compare CSF TNF α levels between different subpopulations: NDPH patients with different triggering events, study subjects in medication overuse versus those not overusing medication, and patients with CDH duration of 2 or more years versus those with daily headache of less than 2 years duration
CSF TNF α levels were elevated (>8.2 pg/mL) in 19 of 20 NDPH patients (range 8.8 to 16.7 pg/mL, mean 11.1 ± 2.7 pg/mL).. Serum TNF α levels were tested in 14 NDPH patients and only 3 had elevated levels (one of those patients had normal CSF levels). Sixteen of 16 CM patients had elevated CSF TNF α levels (range 10.2 to 16.1 pg/mL, mean 12.0 ± 2.3 pg/mL). Serum TNF α levels were tested in 5 CM patients and none were elevated. CSF TNF α levels were elevated in both PT patients (mean 12.0 pg/mL). Serum levels were tested in 1 patient and were normal. No patient demonstrated an abnormal CSF cell count, total protein, or glucose level. Cultures were negative. Seven patients had elevated spinal opening pressures (>20 cm H2O), but none were deemed to have intracranial hypertension, with normal fundoscopic examinations and no improvement of head pain with CSF removal. In regard to response to therapy, which consisted of multiple intravenous protocols (magnesium, promethazine, corticosteroids, dihydroergotamine, chlorpromazine, diphenhydramine, and divalproex sodium), 32 of the 38 study patients (84%) had none or only minimal improvement with inpatient therapy. Only 5 of 20 NDPH patients had more than minimal improvement of their pain with hospitalization, while only 1 of 16 of the CM patients and none of the PT patients showed any significant improvement in their headaches. The lone NDPH patient with high serum TNF α levels, but normal CSF levels, became headache-free with intravenous magnesium.
When looking at various subpopulations, there were some slight differences in CSF TNF α levels, but none statistically significant. In the NDPH subgroup, there was a difference in TNF α levels if the patients had NDPH triggered by an infection or illness (mean CSF TNF α level 9.8 pg/mL) versus if the patients had no known triggering event (mean CSF TNF α level 10.1 pg/mL). This approached but did not reach statistical significance (P = .063). When looking at all the study patients as a whole (NDPH plus control group), CSF TNF α levels were higher in those who were in medication overuse with opiates (mean CSF TNF α level 12.4 pg/mL) versus those who were not in medication overuse (mean CSF TNF α level 11.0 pg/mL), but this was not statistically significant (P = .59). When looking at duration of daily headache, individuals with daily headache of less than 2 years duration had a mean CSF TNF α level of 11.0 pg/mL, while patients with 2 years or longer duration of daily headache had a mean CSF TNF α level of 12.3 pg/mL. This was also not statistically significant (P = .12).
The pathogenesis of NDPH is unknown. The role of proinflammatory cytokines in the evolution of NDPH needs to be considered, because a significant number of patients will start having their headaches after an infection or illness, which are known stimulators of cytokine production. There are no prior studies looking at levels of TNF α in the serum or CSF of NDPH patients. Based on our findings, it appears that elevated levels of TNF α in the CSF may play some role in the pathogenesis of NDPH. Serum TNF α levels were not elevated in most NDPH patients, thus this does not appear to be a disorder derived from systemic inflammation, but rather inflammation involving the CNS.
Proinflammatory cytokines in the CSF produce and enhance pain in animal models. Up until recently, spinal cord glia were considered the sole source of CSF cytokines including TNF α. A new investigation, however, has suggested that cells in the meninges adjacent to the CSF space also produce proinflammatory cytokines that enter into the spinal fluid. Little is known about what triggers the meningeal cells to produce inflammatory cytokines. However, in laboratory animals, recognized triggers of glial cell activation include infection, stress, and surgical procedures.[4-6] Tying this into NDPH, in addition to infection, the 2 other recognized inciting events for NDPH in the literature are stressful life events and surgical procedures.
How elevated TNF α levels can produce head pain can only be postulated. There is recent evidence that TNF α will induce calcitonin gene-related peptide (CGRP) production. TNF α receptors are located on rat trigeminal ganglia neurons and CGRP release is increased after cultured trigeminal neurons are treated with TNF α. CGRP is a known factor in the migraine pathogenesis cascade. It is believed to cause dilation of meningeal and cerebral arteries, release of inflammatory compounds from mast cells and be involved in the transmission of nocioceptive signals from cerebral blood vessels and the trigeminal sensory system. Blockade of CGRP can lead to pain relief in migraine. As of yet, there have been no studies looking at CGRP levels in NDPH, but based on the fact that CGRP is believed to be involved in the pathogenesis of most trigeminal-based head pain syndromes (migraine, tension-type headache, cluster headache, and paroxysmal hemicrania), one could assume a role for CGRP in NDPH pathogenesis. Persistent elevation of TNF α could lead to persistent elevation of CGRP, and thus daily head pain.
Surprisingly, all members of our control group showed elevated CSF TNF α levels. This suggests that TNF α in the CSF may play a role in the genesis of NDPH, CM, and PT headaches. Persistence of elevated TNF α levels in the CSF may create a treatment-resistant headache condition, as all of our study patients failed aggressive outpatient therapy and then showed minimal to no improvement after aggressive inpatient treatment. In general, 75% of our inpatients have at least moderate to substantial improvement after treatment. Elevated levels of TNF α in the CSF may explain why a certain percentage of our hospitalized patients do not improve with standard headache therapy.
Several studies have looked at serum TNF α levels in patients with episodic migraine. One study found an ictal increase in TNF α in jugular venous blood, while another noted elevated serum levels during migraine attacks. Our study suggests CSF TNF α levels are also elevated in chronic migraine, at least the form that is treatment-resistant. Medication overuse is one of the main causes of CM. In the laboratory, chronic exposure to opiates is a trigger of glial activation. Thus, the elevation of CSF TNF α levels in our CM patients, of which 11/16 were on chronic opiates, may have been caused by opiate-induced glial activation. In those individuals not on chronic opiates, one could hypothesize that the stress of daily pain could have activated microglia with TNF α production. Both PT patients were also on chronic opiates at the time of the study, which may explain the elevation of CSF TNF α levels in these study subjects. Importantly, only 4 NDPH patients were on chronic opiates; thus, for this unique CDH subtype, something other than chronic opiate exposure is causing TNF α elevation in the CSF.
From this investigation, it appears that TNF α levels are elevated in various forms of CDH. CSF TNF α levels did vary somewhat when looking at different subpopulations of patients, but the differences were overall slight when present and not statistically significant. In the NDPH subgroup, there was no statistically significant difference in TNF α levels if the patients had NDPH triggered by an infection or if the patients had no known triggering event, although levels were higher in the no known triggering event subgroup. Individuals on daily opiates had slightly increased TNF α levels versus those patients not in medication overuse with opiates or other agents, while TNF α levels were slightly higher in those patients with CDH 2 years or longer in duration versus those with shorter duration daily headache. From these subpopulation findings, it appears that opiates alone and/or duration of illness alone cannot explain the elevation of TNF α in the CSF in all CDH patients.
This study is the first to look at CSF TNF α levels in patients with NDPH. This is an initial observation, which must be substantiated by future studies. There are certainly weaknesses to this investigation. As not all of our patients had serum TNF α levels, we cannot fully comment on the dissociation between serum and CSF levels in this patient population. However, other neurologic conditions such as frontotemporal dementia and acute traumatic brain injury have shown normal serum levels of TNF α in the presence of elevated CSF levels. It is possible that TNF α plays no role in the genesis of treatment refractory headache and that the elevation of CSF TNF α levels in our hospitalized patients was caused by the parenteral medications given. There are no published studies looking directly at the effect of parenteral headache treatments on TNF α levels especially in the CSF, but there is some literature, mostly from animal studies, which would not suggest a role for medication-induced elevation of TNF α. Fetal levels of TNF α in humans was not effected by predelivery administration of magnesium sulfate. In a rat model of endotoxic shock, chlorpromazine reduced serum levels of TNF α. Diphenhydramine was shown to reduce TNF α levels in an animal model of septic shock, while hydrocortisone had no effect on TNF α levels.[13,14] It appears that none of the agents that are typically given to our hospitalized patients will cause an elevation of TNF α levels. In reality, they either have no effect or in most instances reduce levels. Thus, the elevation of TNF α in the CSF appears to be independent of medication affect.
Finally, as almost all of our patients (NDPH and control group) had elevated CSF TNF α levels (thus no "real" control group), we would have to look at a new control population of either episodic headache patients or treatment-responsive CDH patients to determine the true role of TNF α in the pathogenesis of CDH subtypes. If CSF TNF α levels, for example, were normal in an episodic migraine population, that would lend support to the hypothesis that elevated CSF TNF α levels are a contributor in the transformation of episodic migraine to chronic migraine and are an essential element to the production of CDH from onset, as in NDPH. There is literature to suggest that CSF TNF α levels are not elevated in nonheadache neurologic patients and in normal healthy controls. A recent investigation of CSF TNF α levels in patients with amyotrophic lateral sclerosis did have a neurologic control population in which CSF TNF α levels were analyzed. This control population included patients with Parkinson disease, supranuclear palsy, hydrocephalus, spinocerebellar ataxia, neuropathy, Alzheimer disease, Lewy body disease, and chorea. In this control group, CSF TNF α levels were undetectable. In addition, CSF TNF α levels were also not found to be elevated in healthy exercising males. The headache history of these control groups is not known, but the fact that CSF TNF α levels were not elevated suggests that an increase in TNF α levels in the CSF may play a true role in the pathogenesis of CDH.
Elevation of CSF TNF α levels were noted in 19 of 20 NDPH patients and suggests a role for TNF α in the pathogenesis of NDPH. Surprisingly, all CM and PT patients tested had elevated CSF TNF α levels. In most patients with elevated CSF levels, serum TNF α levels were normal. All of these syndromes may be manifestations of CNS inflammation. TNF α may be the prime generator of NDPH and may also explain the mechanism by which episodic migraine may transform to chronic migraine in the face of overuse of opiates. As most of the positive-tested patients showed minimal to no improvement during aggressive inpatient treatment, we hypothesize that persistent elevation of CSF TNF α levels is one of the causes of treatment refractory CDH. Based on these findings, specific TNF α inhibitors may have an important role in the treatment of NDPH and refractory chronic migraine.
Todd Rozen, Michigan Head Pain and Neurological Institute - Neurology, 3120 Professional Drive, Ann Arbor, MI, 48104.
Headache. 2007;47(7):1050-1055. ©2007 Blackwell Publishing