Abstract
Dear Sir We very much appreciate the compliments expressed by Jes Olesen, one of the leading living experts on the headache response induced by nitroglycerin administration to migraineurs and healthy subjects.
Most of the questions raised by Olesen can be answered on the basis of our very careful selection of both patients and controls. This took several years, precisely because it was our intention to form: (i) a consistent group of controls presenting no personal or family history of migraine or cluster headache, and only occasional headaches (infrequent tension-type headache, fever headache, etc.); (ii) a group of ‘pure’ migraine patients and a group of migraineurs who also suffered from tension-type headache (mostly episodic), but who were perfectly able to distinguish between the two types of headache from the very onset of their attacks; and (iii) a group of ‘pure’ cluster headache patients.
In the light of this, the low response rate in our control group may be explained by the fact that our healthy controls were also controlled for family history of primary headache. Previous data by Sicuteri (1) had indeed shown that nitroglycerin is able to induce a migraine headache in healthy subjects with a positive family history of migraine. In this regard, it was particularly enlightening that a more thorough investigation of family history (direct interview with the relatives) revealed a positive family history of migraine in both the controls who developed a specific response to nitroglycerin test. In addition, differences in terms of percentage of response in healthy subjects may be ascribed to the different methodology used: continued, sustained (20 min) i.v. infusion of nitroglycerin in the Danish group (2, 3) vs. single, sublingual dose in our study (4).
Instead, the possibility that we may have overlooked ‘mild headaches’ can be scientifically ruled out, given that an observer – an expert physician blind to the diagnosis and to the type of subject (patient/control) – was instructed to remain continuously with the subjects throughout the first 2 h of the test and to record on the diary card the occurrence of any type of headache of whatever intensity, as well as its duration, associated symptoms and possible side-effects. In any case, it must be noted that this paper focused on describing a response ‘specific’ for migraine or cluster headache; the non-specific headache response in the control groups thus has little relevance to the purposes of the present investigation.
Age correlation was not informative since the age ranges of both the controls and patient groups were very narrow (controls 33.4 ± 3.2 years, migraine 34.1 ± 1.23 years, cluster headache 37.5 ± 1.5 years). Meanwhile, gender was not found to have any influence on the presence, absence, or type of response.
As regards delayed headache, it is incorrect to state that this type of response was lacking in our study. For evaluation and validation purposes, we chose to describe the headache response as immediate/delayed and specific/non-specific. Most of the specific responses reported after the first hour of observation fit the description of delayed headache according to Sicuteri (1): they appear 60 min after nitroglycerin administration, are indistinguishable from the spontaneous headache, and are long-lasting.
It is also imprecise to state that the biphasic response was not reported. Actually, 40% of our migraine patients developed a non-specific immediate response followed by a specific response in the subsequent 7-h interval. Given that the non-specific immediate response is short-lived, these patients definitely qualify for the biphasic response described in the seminal work by Sicuteri (1). The novel aspect of our report is the identification of a subtype of headache response within Sicuteri's ‘long-lasting alike spontaneous headache’ (LLASH) (1): the rapid-onset (immediate) specific response (recorded in nearly 40% of the migraineurs who reported a specific response to nitroglycerin). This response consists of a migraine-like headache that begins within 1 h of nitroglycerin administration and develops fully over the following hours.
It is obvious that if one represents the headache response in terms of average pain intensity over time, the shape of the final curve may turn out to be biphasic because of a nadir resulting from the short duration of the immediate non-specific headaches (which in our migraine population were reported in nearly half of the subjects). The idea that differences in response may be attributable to a slower onset of action resulting from the route of administration used would appear to be highly improbable: the pharmacokinetics of nitroglycerin show that the onset of action of both sublingual and intravenous nitroglycerin occurs within 1–5 min, with a peak at 5–15 min in the case of the sublingual route, and a peak that depends on the duration of the infusion in the case of the i.v. route. In addition, the C max concentration of sublingual nitroglycerin occurs at 2–5 min (source: Thomson Healthcare Micromedex: http://www.micromedex.com).
Our migraine patients had a mean of 9.1 ± 4.3 headache days per month (minimum 2, maximum 30). This figure also includes days with tension-type headache. The number of headache days was > 15/month in only 24 of our migraineurs. None of our patients suffered from chronic migraine according to the criteria of the International Classification of Headache Disorders (ICHD-II) (5), while chronic or episodic tension-type headache was reported in 22 and 14 migraine subjects, respectively. None of our patients fulfilled ICHD-II criteria for Medication Overuse Headache. Having said that, Olesen's suggestion that a lower threshold to nitroglycerin may exist in subjects with a higher number of monthly headache days is an intriguing hypothesis, partly supported by the demonstration that migraineurs with a higher monthly frequency of attacks were more likely to develop a specific response to nitroglycerin (see Fig. 2 of the paper), and by the observation that 87% of migraineurs with > 15 headache days/month reported a specific response to nitroglycerin, as opposed to only 70% of migraineurs with ≤ 15 headache days/month.
As regards the low frequency of attacks of migraine with aura evoked by nitroglycerin in the migraine with aura group, we do not believe that a percentage of 36.5% in a population of 22 subjects (4) can really be considered statistically different from a percentage of 50% in a population of 12 subjects (6). On the other hand, it is noteworthy that Christiansen et al. failed to detect any aura phenomena, while three of our migraine with aura patients developed a typical migraine aura following nitroglycerin administration. Therefore, we definitely agree with Olesen regarding the need for further studies on larger populations of patients suffering from migraine with aura.
Although nitroglycerin administration consistently induced a mild to moderate transient reduction in blood pressure, the paucity of side-effects is real and not related to lack of adequate monitoring: side-effects were indeed reported by only 2% of subjects, and consisted of symptomatic hypotension associated with prefainting symptoms which occurred within 30 min of nitroglycerin administration and promptly disappeared upon leg raising. The fact that the patients were kept in the supine position for 1 h after nitroglycerin administration may have contributed to the low rate of side-effects.
We do not recommend the nitroglycerin test ‘unconditionally’ for diagnostic purposes, as historical data obtained from patients are frequently sufficient to allow the rapid formulation of a diagnosis of migraine or cluster headache. As indicated throughout the discussion, the test is indicated in specific cases (lack of adequate clinical history, atypical forms) and in the presence of specific needs (diagnostic support in legal or work-related settings, objective and witnessed confirmation of ‘subjective’ reports). In this effort to validate a test for what is a mostly subjective disorder – a huge undertaking – we decided to report only the data relating to performance of the test in primary neurovascular headaches (migraine and cluster headache), where the attacks are characterized by the presence rather than by the absence (as in the case of tension-type headache) of clinical features (see ICHD-II criteria) and in some cases by the presence of signs in addition to symptoms. This decision was also prompted by the observation that the non-specific response frequently fits the diagnostic criteria for tension-type headache, which would have been a confounding variable. In addition, in the case of migraine, we decided to evaluate the typical patient seeking help at a Headache Centre, i.e. the patient presenting a moderate to high frequency of monthly attacks (in our population the monthly frequency ranged from two to 13 per month, mean 6.7 ± 5.5). The validity of the test in subjects with a lower frequency of migraine attacks is debatable, but so is its usefulness for practical purposes.
We are very familiar with the available literature on the use of nitroglycerin as a diagnostic test: it is cited in the Introduction and indeed inspired our work. However, these previous studies were conducted on limited and differently selected populations, using a different methodology and conducted for purposes other than evaluating the reliability of the test. Therefore, we felt that a comparative evaluation was not warranted and did not fall within the scope of our paper.
As regards the question of the resources needed to perform the test, in our study the patients were monitored for an 8-h period. This long monitoring period was obviously related to the aim of the study: validation of a method. On the basis of our results, we suggest that in daily practice patients might be kept under observation for the first 2–3 h, as this has been identified as the time span during which the majority of nitroglycerin-induced attacks develop and side-effects can occur. Subsequently, the patients may be discharged having first been shown how to fill in the diary card over the subsequent 5–6 h, and how to treat the potential headache attack. The day after the test the patients will need to return to the Centre for a short evaluation of their clinical condition and diary records.
We wish to thank Jes Olesen for his constructive criticism and valuable suggestions, which will undoubtedly lead to future studies by our group, and hopefully by other research groups too. We firmly believe that the application of a strict and rigorous methodology is the only way to improve our knowledge of and our approach to primary headaches.