Abstract
Eighty-five brain tumour patients were examined for further characteristics of brain tumour-associated headache. The overall prevalence of headache in this population was 60%, but headache was the sole symptom in only 2%. Pain was generally dull, of moderate intensity, and not specifically localized. Nearly 40% met the criteria of tension-type headache. An alteration of the pain with the occurrence of the tumour was experienced by 82.5%, implying that the preexisting and the brain tumour headaches were different. The classic characteristics mentioned in the International Classification of Headache Disorders (worsening in the morning or during coughing) were not found; this might be explained by the patients not having elevated intracranial pressure. Univariate analysis revealed that a positive family history of headache and the presence of meningiomas are risk factors for tumour-associated headache, and the use of β-blockers is prophylactic. Pre-existing headache was the only risk factor according to logistic regression, suggesting that patients with pre-existing (primary) headache have a greater predisposition to develop secondary headache. Dull headache occurs significantly more often in patients with glioblastoma multiforme, and pulsating headache in patients with meningioma. In our study, only infratentorial tumours were associated with headache location, and predominantly with occipital but rarely frontal pain.
Introduction
A secondary headache in patients with a brain tumour is defined in The International Classification of Headache Disorders (ICHD-II) (1) in subchapter 7.4. This headache type is attributed to a neoplasm if it improves after effective treatment, if it develops in close temporal relation with the diagnosis of the tumour, or if there is good evidence that the intracranial disorder can cause headache. Thus, the presence of a space-occupying lesion, hydrocephalus or carcinomatous meningitis has to be proven and, moreover, the headache has to deteriorate with advancing disease or resolve after removal of the neoplasm or treatment with corticosteroids.
It is not known whether secondary headache associated with an intracranial neoplasm has a unique clinical picture or, especially, whether there is a predisposition to develop it in the presence of a brain tumour. Furthermore, there are no clear pathophysiological concepts explaining how secondary headaches in patients suffering from an intracranial tumour are related to primary headaches. In the ICHD-II definition, headache characteristics are associated with nausea and vomiting, worsening in the morning, coughing, or with the Valsalva manoeuvre. Reports in the literature have characterized brain tumour headache so far mainly by prevalence, severity, duration, character, location, association with trigger factors, and pathology and location of the associated tumour; however, the results of these studies are inconsistent (2–5).
We investigated the headache and tumour history of 85 patients with a primary or secondary brain tumour to determine their intrapersonal risk factors for a tumour-associated headache. A pre-existing headache, if present, was compared with the tumour-associated headache. Furthermore, we assessed whether there was a correlation between the pain quality and the pathology of the tumour as well as between headache and tumour location. Finally, we also determined the typical features of brain tumour-associated headache and compared the results with those in the literature.
Methods
All patients were recruited in 2002 from a large neurosurgical department in a university hospital in Southern Germany. Ninety-seven in-patients with primary or secondary brain tumours agreed to participate in the study and gave their written informed consent. Twelve had to be excluded: one due to an incomplete questionnaire and 11 because no definite information on the pathology of the tumour was obtained. Thus, 85 patients took part. All were interviewed about their personal details (age, gender, family history of headache) and were asked to fill in a standardized questionnaire, which included information on their headache. The focus was on the pre-existing headache (presence and classification, if known by the patient), current headache, or on a change in quality and quantity of the headache (characteristics, location, intensity and frequency). Current headache was categorized according to frequency (continuous, daily, >15/month, weekly, less than weekly), location (frontal, temporal, parietal, occipital, hemispheric, unilateral, bilateral), intensity [on a nominal analogue scale (NAS) from 1 to 10 with 1 for little and 10 for worst pain], quality of pain (dull, stinging, pulsating other), duration of the event (<1 h, 1–4 h, 4 h to 3 days, 3–7 days, >7 days, no data), association with nausea, vomiting, photophobia, lacrimation, or flickering of the vision, and trigger factors. The diagnostic criteria for the primary headaches (i.e. duration of attack, pain characteristics and associated symptoms) in the ICHD-II (1) were used in an attempt to categorize the tumour-associated headache as migraine-like, tension-type-like, or cluster headache-like. Each patient underwent a standardized neurological and neuropsychological examination.
Information on the neoplasm was obtained from the patient's chart (histopathology of the tumour) and from neuroradiological imaging (magnetic resonance imaging or contrast-enhanced computed tomography) as regards tumour size, location and surrounding oedema. Tumour size was defined as the product of the two largest dimensions (in cm) and was used as a measure to differentiate small (<10 cm2), medium (10–20 cm2) and large tumours (>20 cm2). Multifocal tumours were assigned to a special category. The surrounding oedema was estimated subjectively on a scale of 0–3, from no oedema (0) to extensive oedema (3). The main location of the tumour was chosen from the following list: frontal, temporal, parietal, occipital, hemispheric or infratentorial. Tumour histopathology was metastasis, glioblastoma multiforme (GBM), astrocytoma (World Health Organization grade 1–3), meningioma, or other.
Statistical analysis was performed using SPSS 12.0G for Windows (SPSS Inc., Chicago, IL, USA). Categorical and continuous variables were analysed using χ2 and two-sample t-test, respectively. The level of significance was set at 0.05. Binary logistic regression including all probably relevant variables (with P < 0.15) was used to determine the set of variables most likely to predict the occurrence of headache in association with a brain tumour. The probability of correct prediction with these variables was assessed with the Hosmer–Lemeshow test.
Results
Study population
The mean (± SD) age of the 85 study patients was 54.2 ± 13.6 years. The majority (54.1%) were female, about one-quarter had a positive family history of headache and 40 (47.1%) remembered having had a headache in their medical history prior to the diagnosis of brain tumour (Table 1). Fifty-one (60%) complained of a current headache syndrome. Additional symptoms found in 54.1% included focal neurological (hemiparesis, unsteady gate, visual field defect, etc.) and in 37.6%, neuropsychological (memory deficit, dysphasia, change of personality, etc.) signs. Symptomatic epileptic seizures occurred in 27.1%. In the study population vertigo, blurred vision, nausea/vomiting and tinnitus were also frequent (36.5%). Only one patient (2%) complained of headache as the sole presenting symptom (data not shown).
Selected characteristics of the study population
SD, Standard deviation; NA, not available; GBM, glioblastoma multiforme.
In general, pain control was achieved with non-opioid pain medication. Steroids and antiepileptic drugs were frequently administered (45.9 and 29.4%, respectively). Arterial hypertension was the most frequent cardiovascular risk factor (22.4%); it was treated in eight cases with β-blockers and in 19 with other medication. The histopathology of the brain tumours showed a nearly equal distribution of 22–26% among the four entities of meningioma, astrocytoma, GBM and metastasis. The primary tumours of the 20 metastases were breast (20.0%), colon (10.0%) and lung (10.0%) cancer, renal carcinoma (15.0%), and other or unknown (45.0%).
Brain tumour-associated headache: univariate analysis
Table 2 lists all 85 patients with a brain tumour: 51 complained of headache, 34 did not. These two groups were analysed further for differences in the distribution of various intrapersonal or tumour characteristics. Gender, age, tumour size, tumour location and extent of the surrounding oedema were equally distributed in the two groups. A positive family history of headache and pre-existing headache were significantly associated with secondary headache (P < 0.01 in both). Brain tumour patients complained less often of headache if they were on β-blockers for blood pressure control (P = 0.03). A trend towards a higher frequency of headaches was found in association with meningiomas (P = 0.06).
Univariate analysis of 85 patients with brain tumours as regards risk factors for headache
CI, Confidence interval; NA, not available; SD, standard deviation; GBM, glioblastoma multiforme.
Brain tumour-associated headache: binary logistic regression analysis
Potential candidates for relevant variables were defined by a P-value <0.15 and were analysed further by binary logistic regression. These variables were age, positive family history of headache, individual pre-existing headache, medium and large size of surrounding oedema, meningioma as tumour pathology and the intake of β-blockers. Table 3 demonstrates the adjusted statistical values. A strong trend toward less headache after regularly using β-blockers was confirmed; however, this was not statistically significant. A history of pre-existing headache was the only factor to have a significant influence on the occurrence of brain tumour-associated headaches (P < 0.01). Hosmer–Lemeshow testing was done on all seven variables: 78.4% were correctly predicted to belong to the headache group and 73.5% to the non-headache group (total correct prediction 76.5%, data not shown).
Binary logistic regression analysis of 85 patients with brain tumours as regards risk factors for headache
CI, Confidence interval.
Comparing pre-existing headache with current headache
Forty patients recalled having had a pre-existing headache. Of these, nine (22.5%) thought or were told they had had migraine and six (15%) tension-type headache (data not shown). Thirty-three (82.5%) reported that the headache had changed around the time of the tumour diagnosis (Fig. 1). Quality (character and location) and quantity (severity and frequency) of the symptomatology changed in about two-thirds. Six patients with a history of headache (18.2%) reported a marked relief of the headache in association with the brain tumour (headache lost). Only 17.5% still had the headache they were familiar with. A new onset of headache was reported by 17 of those tumour patients who had no history of headache.
Stratification of all brain tumour patients (pt) according to the presence or absence of pre-existing headache. A further division was made according to the presence of tumour-associated headache; this resulted in alteration or no alteration of the known headache behaviour. Alteration was specified by location, severity, character or frequency and new or lost headache (multiple entries possible). The respective number of patients is shown in the boxes.
Typical brain tumour headache
Brain tumour-associated headache projected with nearly equal frequency to the frontal, parietal or occipital lobes (Table 4). The pain was most often hemispheric and was as likely to be bilateral (49.0%) as unilateral (21.6% or 29.4%). It was of medium intensity (median at NAS 6) and seldom very strong (90th percentile <NAS 9). Tumour-associated headache in about two-thirds was described as dull and in only 17.6% or 15.7%, respectively, as stinging or pulsating. In about half of all patients a headache event did not last for >4 h, in six cases (11.8%) headache duration was >7 days. A headache attack occurred daily in 15 cases (29.4%), less than once weekly in 43.1%, and 11.8% suffered from continuous pain. Association with nausea and vomiting was found in 17.6%, with flickering of the vision in 7.8% and with photophobia and lacrimation in 3.9%. These features were combined to see whether the current headache met the diagnostic criteria of the primary headache syndromes. The criteria for tension-type headache were met in 20 patients (39.2%) and for migraine or cluster headache in none (data not shown). Most patients could not specify any trigger of the headache (78.4%). The ‘classic’ association with horizontal posture or coughing (9.9%) was as frequent as with strain and exercise (11.7%).
Brain tumour-associated headache, N = 51
N/V, Nausea and vomiting; NAS, numerical analogue scale.
Headache quality and tumour pathology
The relationship between headache character and tumour histopathology is addressed in Table 5. Only statistically significant or strong tendencies are shown. The most malignant tumour, GBM, occurred to a significantly higher degree in patients with dull as opposed to non-dull headache (odds ratio 8.6). The GBM and metastases were seldom associated with pulsating pain. This difference, however, was not significant. Meningioma showed a strong tendency to be associated with pulsating pain. No association was found between the occurrence of tension-type-like headache and tumour pathology (data not shown).
Association between headache quality and tumour histopathology (only relevant data shown with P < 0.15)
CI, Confidence interval; GBM, glioblastoma multiforme.
Location of headache and tumour
Generally, there was no association between the location of the headache and the tumour (Table 6). Only infratentorial tumours were found to be significantly more often associated with occipital and less often with frontal headache (P = 0.03 and 0.02, respectively). The only temporal headache was associated with an occipital tumour (P < 0.01). If headache was unilateral, then headache and tumour were on the same side in only 61.5% of cases (data not shown).
Association between headache and tumour location (only relevant data shown with P < 0.15)
CI, Confidence interval.
Discussion
In this study, 85 patients with primary or secondary brain tumours were examined for concurrent headache. They were not diagnosed to have ‘headache attributed to intracranial neoplasms’ according to the ICHD-II 7.4 definition, for two reasons: they did not have a highly elevated intracranial pressure (see below), and a longitudinal study was not performed. The data presented here therefore fit best to headache associated with brain tumours. The prevalence of 60% in our patients was similar to the literature (3, 4, 6, 7) reporting 31–71%. Only 2% had headache without any other symptoms; this confirms the finding in a prospective study of 183 patients (5), which reported that the first and isolated clinical manifestation of brain tumours was headache in 8.2% and, at the time of diagnosis, even <1%.
Pain was of moderate intensity, requiring non-opioid analgesics for sufficient control, as reported elsewhere (2, 4). In contrast to the findings of Forsyth et al. (2), who localized tumour-associated headache to the front, the pain of our patients was not restricted to a specific region of the head. Its character was mainly dull, in agreement with other reports (2, 4, 6), and rarely stinging or pulsating, in contrast to the findings of Pfund et al. (3), who described it as having a mainly throbbing quality. To the best of our knowledge, the finding that nearly two-fifths of the tumour-associated headaches are tension-type-like is described here for the first time.
Univariate analysis of intrapersonal or tumour-associated risk factors for headache revealed that a positive family history of headache and pre-existing headache are significant intrapersonal risk factors. Meningiomas are a histopathology-associated risk factor. In contrast, gender, age, tumour size, tumour location and the surrounding oedema had no influence on the presence of headache. This is congruent with findings of Vazquez-Barquero et al. (5) as regards age, sex and neoplasm location and with data of Forsyth et al. (2) on tumour size. The findings that mainly infratentorial tumours (3, 4, 8), metastases and astrocytomas (3), or larger tumours (2) are associated with headache could not be confirmed. We also did not find support for the hypothesis that meningiomas are less likely to cause headache because of their slow growth (4). Our finding that only infratentorial tumours were associated with headache location, and predominantly with occipital but rarely frontal pain, agrees with findings of Forsyth et al. (2) and Suwanwela et al. (4), but contradicts those of Pfund et al. (3). No other localizing or lateralizing value for headache as regards tumour location was found. We also found that the intake of β-blockers was significantly less associated with headache, suggesting they have a possibly protective effect. Headache was independent of the intake of steroids. The anti-oedematous mechanism of steroids might have been ineffective in preventing headache in our patients, since there were no signs of elevated intracranial pressure clinically or in the radiological findings. This is supported by the independence of the pain from the size of the surrounding oedema. However, a certain false-low estimation of the overall prevalence of headache in our study population, with 45.9% being pretreated with steroids, could not be completely ruled out. The absence of elevated intracranial pressure might be important for the interpretation of the finding that headache is not associated with the classic triggers of worsening in the morning, coughing, or Valsalva manoeuvre. This is contrary to the ICHD-II (1), Suwanwela et al. (4) and Forsyth et al. (2). The latter authors distinguished between headache in brain tumour patients with and without increased intracranial pressure. They found less association with these worsening factors if intracranial pressure was not elevated. This agrees with our findings, which favoured other pathogenic mechanisms for the occurrence of headache instead of elevated intracranial pressure.
The only significant risk factor according to binary logistic regression was pre-existing headache. Forsyth et al. (2) also found that many patients with pre-existing headache complained of tumour-associated headache (78%). This association raises the question of whether the mechanisms that induce headache in brain tumour patients are in some way related to the mechanisms also involved in primary headache. This hypothesis is supported by the report that patients with prior headaches also had a greater risk of postlumbar puncture headache (9). Furthermore, the trend toward less headache during treatment with β-blockers might indicate they have a prophylactic effect for tumour-associated headache, as they do for migraine. This suggests that similar mechanisms act in both headache types. The importance of the classic ‘traction on intracranial pain-sensitive structures’ (10–14) for the development of brain tumour headache remains unclear in our study.
To the best of our knowledge, the findings of higher frequency of dull headache in GBM and a tendency for pulsating pain in meningioma are presented here for the first time. The latter pain might be due to the pronounced vasculature of meningeal tumours, which are innervated by the trigeminal nerve. Both vasculature and trigeminal nerve are parts of the trigeminovascular system, which is thought to be of pathophysiological relevance in the primary pulsating headache, migraine (e.g. Goadsby et al. (15)).
Conclusion
Although headache is a common feature of brain tumours, it very seldom occurs as a single symptom. A new or altered pre-existing headache should therefore encourage the physician to make a thorough neurological and neuropsychological evaluation. Headache associated with brain neoplasms has no unique clinical picture besides a similarity to tension-type headache in two-fifths of patients. The pain is mainly dull, but can also be stinging or pulsating, depending on the histopathology. The most important risk factor for developing headache while suffering from a brain tumour is having had a pre-existing headache. This implies that there is a predisposition for secondary headache and that a common pathophysiological mechanism is involved.
Acknowledgements
We thank J. Benson for her help in preparing the manuscript. The study was supported by a grant of the BMBF (Chronic headache, project D1).