Abstract
Dear Sir First, a big compliment to the authors (1). This is an important study done with care and including a huge amount of material. The results largely make sense in relation to the previous literature on nitroglycerin provocation but also raise a number of questions for which answers require more information.
It is surprising that so few control patients developed headache after such a high dose of nitroglycerin. In our studies of normal experimental subjects, we most often have preselected subjects according to their response to 0.5 mg nitroglycerin. Our experience is that only one in five has to be excluded because they do not develop any headache. We have also studied an unselected group where 80% developed headache. Is it possible that in the paper by Sances et al. mild headaches have been overlooked?
Age and sex correlation with the developed headache both in the patient groups and in the controls would be interesting.
The second surprising finding is the lack of delayed headache after nitroglycerin. In their original publication, Sicuteris’ group found a clear separation into immediate and delayed headache after sublingual administration. In every study we have done, including also with other substances such as cacitonin gene-related peptide and histamine, there has been a biphasic headache response in migraine sufferers. Sublingual nitroglycerin has a much slower onset of action than intravenous and perhaps a longer duration of action, which might tend to smooth out the headache response. Another possibility is the high attack rate of 6.7 attacks per month. How many migraine days did the patients have per month?
Along this line, I am surprised at the low frequency of tension-type headache in this group. We have previously shown (2) that the amount of tension-type headache (phenomenologically speaking) increases with increasing migraine attack rate and is very prevalent as soon as patients have more than two attacks a month. In other words, I suspect that these patients must have had almost daily headache with migraine attacks on top and thus might represent the severe end of the migraine spectrum. This could perhaps explain a constantly low migraine threshold and therefore the early onset of migraine attacks after nitroglycerin. The low frequency of provoked attacks in migraine with aura is rather surprising in the light of the results of Christiansen et al. (3), who studied a small group of patients with pure migraine with aura and found a relatively high incidence of migraine without aura attacks after nitroglycerin. Sances et al. also had a low number of patients, so that statistical fluctuation may play a role. Future studies should definitively establish whether migraineurs with aura are less sensitive to nitroglycerin.
Side-effects are very superficially described. If this test is to be suggested for wider use, then a much more thorough description of the side-effects is necessary. Nitroglycerin 0.9 mg is a high dose and one would expect to see more cases of hypotension and/or vasovagal attacks. For how long were patients kept supine? Since they were asked to walk stairs in order to determine the frequency of a specific response, I presume that they were taken up quite early after nitroglycerin administration.
The authors rather unconditionally recommend the nitroglycerin test, as used in their study, for diagnostic purposes. There are, however, several issues to be discussed here. First, they have demonstrated a high sensitivity only in patients with very frequent migraine, as mentioned above. There is no knowledge of the sensitivity and specificity in patients with fewer attacks. Furthermore, comparison has been made only with a normal control material without headache. The real problem in headache differential diagnosis is to distinguish between different kinds of headache, primarily mild attacks of migraine without aura and attacks of tension-type headache. This should be pointed out in the paper and future studies in such groups are necessary before the test can be generally recommended. Ideally, the test should also be used in a material selected from the general population. There is fairly extensive, old literature on the use of nitroglycerin as a diagnostic test, but this literature has not been discussed.
The last issue is a practical one. How much does it take in terms of resources to do this test? For how long would the doctor need to be next to the patient, and would it really be necessary for the patient to stay in hospital for more than an hour or two? To keep patients in hospital for 8 h is probably beyond the capacity of most headache centres. After i.v. infusion we observe patients for 3 h, but our experience has been that side-effects always occur within the first hour. After 3 h we have sent patients home and they have recorded their headaches in a diary similar to the one used by Sances et al. Such a modification would save a lot of resources.
Despite the criticisms, the study is an impressive effort and I look forward to further development and validation of this test.