J. J. Murphy, S. Heptinstall, J.R.A. Mitchell, The Lancet, July 23, 1988

Department of Medicine, University Hospital, Nottingham

Summary: The use of feverfew (Tanacetum parthenium) for migraine prophylaxis was assessed in a randomised, double-blind, placebo-controlled crossover study. After a one-month single-blind placebo run-in, 72 volunteers were randomly allocated to receive either one capsule of dried feverfew leaves a day or matching placebo for four months and then transferred to the other treatment limb for a further four months. Frequency and severity of attacks were determined from diary cards which were issued two months; efficacy of each treatment was also assessed by visual analogue scores. 60 patients completed the study and full information was available in 59. Treatment with feverfew was associated with a reduction in the mean number and severity of attacks in each two-month period, and in the degree of vomiting, duration of individual attacks was unaltered. Visual analogue scores also indicated a significant improvement with feverfew. There were no serious side-effects.

Zusammenfassung: Die Anwendung von Mutterkraut (Tanacetum parthenium) in der Migräne-Prophylaxe wurde mit einer randomisierten, doppelblind angelegten und placebo-kontrollierten Crossover-Studie geprüft. Nach einer einmonatigen einfachblinden Placebo-Verabreichung, wurden 72 Freiwillige zufällig in zwei Gruppen aufgeteilt, die entweder vier Monate lang täglich eine Kapsel Mutterkraut-Blätter oder Placebo erhielten. Danach wurde jeder Patient für weitere vier Monate der jeweils anderen Behandlung unterworfen. Die Häufigkeit und Schwere der Anfälle wurde über 2 Monate täglich festgehalten. Die Wirksamkeit der beiden Behandlungsmethoden wurde nach Befragung der Patienten numerisch gewichtet. 60 Patienten beendeten die Studie, von 59 konnten alle Befunde ermittelt werden. Die Behandlung mit Mutterkraut war begleitet mit einer Verringerung der Zahl und der Schwere der Anfälle im jeweiligen 2-Monatszeitraum. Nicht verändert waren das Ausmaß der Übelkeit und die Dauer der einzelnen Attacken. Auch die Übertragung der Befunde in numerische Werte ergab eine signifikante Verbesserung nach Behandlung mit Mutterkraut. Besondere Nebenwirkungen wurden nicht beobachtet.

Introduction

The aromatic plant known as feverfew (Tanacetum parthenium) was used in monastic times as an antipyretic¹. Over the past decade it has regained its former status as a medicinal herb and the leaves are a popular lay remedy for migraine prevention. The plant is rich in a family of compounds known as sesquiterpene lactones, principally parthenolide². The intensity of interest in feverfew in the UK was shown when 25.000 replies were received in response to an offer of further information in a national newspaper³. Both fresh and dried leaves are ingested and several commercial preparations are available.

The pathogenesis of migraine remains unknown, but abnormal platelet behaviour has been implicated⁴. During an attack platelets release serotonin⁵ and a pathogenic role for this process is supported by the value of serotonin antagonists in migraine prevention (pizotifen⁷, methysergide⁸) and of ergotamine in acute treatment. In-vitro studies have shown that feverfew extract inhibits serotonin release from platelets⁹, if this response occurs in vivo, it might explain the clinical benefit reported in migraine¹⁰.

There have been few clinical studies of feverfew. In a double-blind study, Johnson et al¹¹ examined the effects of withdrawal of feverfew from regular users who felt its use to be beneficial. The deterioration in migraine in those changed to placebo provides indirect evidence of efficacy, but, by virtue of the self-selection of these patients, the effect of feverfew on migraineurs generally could not be assessed. We now report the results of a prospective double-blind placebo-controlled trial of feverfew in migraine prophylaxis.

Methods

Patients

An open meeting, organised in association with the local migraine self-help group and publicised by the media, was attended by 150 people. From this group and from those responding directly to the media publicity, 190 individuals volunteered for the study. 119 satisfied the entry criteria and 76-23 with classical and 53 with common migraine-agreed to participate. All patients gave informed consent and the protocol was approved by the hospital ethical committee. Those eligible for the study were adults with migraine (as defined by Blau¹²) of more than two years duration with at least one attack a month. Patients were excluded if they were receiving treatment for conditions other than migraine except for women taking oral contraceptives, who were included if their treatment had been unchanged for at least three months. Women of childbearing age not receiving adequate contraception were excluded.

The study group consisted of 20 men and 56 women whose ages ranged from twenty-four to seventy-two years (mean forty-six). 70 (92%) had previously consulted a general practitioner about their migraine and 19 (25%) had been referred to a hospital consultant. 37 (47%) had already tried conventional migraine prophylaxis and 39 (51%) had received ergotamine; only 17 (22%) had received neither form of therapy. 18 (23%) had previously tried feverfew, of whom 11 thought it helpful. On entry to the study 5 were still taking feverfew, 4 propranolol, and 3 pizotifen. All migraine-related drugs were stopped at the beginning of the trial.

Trial Design

After a one-month single-blind placebo run-in period, patients were randomly allocated to receive either one capsule of feverfew daily or matching placebo for four months. They were then transferred, with no intervening washout period, to the other treatment limb for the final four months.

Assessment of Patients

Patients were assessed every two months for the double-blind phases of the study. A diary of migraine symptoms was provided for each of these periods in which patients recorded the number and duration of individual attacks, the severity of headache, and any associated features. The severity of headache was graded as follows: 0 = no pain; 1 = mild, not interfering with daily activities; 2 = severe, reducing working capacity; 3 = very severe, requiring rest in bed. Nausea and vomiting were also recorded: 0 = no nausea or vomiting; 1 = nausea only; 2 = vomited once; 3 = vomited repeatedly. Details of visual or focal neurological features and working days lost through migraine were noted.

The patients’ overall impressions of each two-month treatment period were assessed in two ways. A 10 cm visual analogue scale was used to represent an individual`s range of migraine, with "worst ever" and "best ever" as the two extremes. Measurement in millimetres from the "worst ever" end produced a score ranging from 0 to 100, with higher scores indicating a better period for migraine. Patients also graded each treatment period by choosing one for five reponses: A = much worse; B = worse; C = same; D = better; E = much better. at the end of the study patients were asked to indicate which of the two four-month periods of treatment they regarded as more effective.

Table I - COMPARISON OF MEAN (SEM) NUMBER AND DURATION OF ATTACKS IN EACH TWO-MONTH PERIOD

NS = Not significant

The presence of mouth ulceration, reported to be a side-effect of feverfew¹³, was elicited by direct questioning at each visit. Other side-effects were elicited by recording the response to a standard question: "Have you noticed any problems with the treatment?" Urinalysis for blood and protein (BM-Test-5L, Boehringer Mannheim) and routine haematological and biochemical tests were done during the run-in and at the end of each treatment phase; blood pressure and pulse rate were recorded at the same visits. Blood pressure (right arm, Hawksley random-zero sphygmo-manometer, all treasurements in duplicate) was measured supine after resting for 5 min, and erect after standing for 2 min; pulse rate was determined by radial palpation for 30 s after each blood pressure measurement.

Treatment

Feverfew plants were propagated from original plants which had been selected for their high anti-secretory activity (see below). They were grown under glass, watered daily, and fed weekly with a high nitrogen, high potassium feed. The plants were kept at 20-24°C and received sixteen hours of light a day from fluorescent tubes. Leaves were gathered weekly throughout the year and were taken from the shoots of non-flowering plants.

Table II - ATTACKS OF STATED SEVERITY ON EACH TREATMENT

Leaves were washed in clean cold water, soaked in a solution of 0-5% (v/v) sodium hypochlorite for 10 min to reduce bacterial contamination, and rinsed in copious amounts of fresh cold water. They were then blot-dried on large sheets of absorbent paper and transferred to a drying cabinet at 37°C. After four days, the dried leaves were placed in polythene containers and stored at 4°C until required. Before the capsules were prepared, samples of each batch of leaves were analysed for anti-secretory activity⁹ by quantifying the inhibitory effect of chloroform extracts of the laeves on ¹⁴C-serotonin release from human platelets.

Hard capsulets were prepared from finely powdered leaves in the hospital pharmacy in accordance with standard guidelines¹⁴ and stored at 4°C until dispensed (or discarded after five months if unused). The amount of feverfew powder used per capsule varied with the strength of the preparation, as judged by its anti-secretory activity. Capsule weights ranged from 70 to 114 mg (mean. 82 mg) and contained the equivalent of 2.19 (SD 0.63) µrnol parthenolide per capsule.

Table III - ATTACKS ACCOMPANIED BY NAUSEA AND VOMITING ON EACH TREATMENT

One capsule therefore corresponded to about two medium-sized leaves. Placebo capsules contained dried cabbage leaves which were similarly prepared; the cabbage did not possess any anti-secretory activity.

Statistical Analysis

Analysis was restricted to patients who completed the study. Unless otherwise stated, mean values, and standrad errors are reported. Analysis of variance was used to compare the number and duration of attacks, the visual analogue scores, blood pressure, and pulse rate. The homogeneity of variance was assessed with Cochraná test and logarithmic transformation of the data was done as necessary. Contingency tables were analysed by c ² , with Pearson´s coefficient of contingency to quantify differences between treatments.

Results

60 of the 76 patients (79%) completed the study. 4 withdrew during the placebo run-in and 12 after randomisation (6 on feverfews, 6 on placebo), 5 of the withdrawals after randomisation were because of side-effects (2 feverfew, 3 placebo); 4 chose to withdraw because of failure to improve (3 feverfew, 1 placebo) and 2 after development of serious illness unrelated to the trial (both on placebo); and 1 patient defaulted from follow-up. Of the 60 remaining patients who completed the study, 1 lost her migraine diary on two separate occasions, so the results refer to the remaining 59.

There was a 24% reduction (95% confidence interval 14-34%) in the number of attacks during feverfew treatment but no significant alteration in the duration of individual attacks (table I). there was a non-significant tendency (p = 0.06) towards milder headaches with feverfew (table II) and a significant reduction (p < 0.02) in nausea and vomiting accompanying the attacks (table III). 68 working days were lost through migraine during the feverfew phases vs 76 with placebo.

Both of the global assessments of efficacy showed that feverfew was better than placebo. The visual analogue score (higher scores indicate improvement) was 74 (2) on feverfew vs 60 (3) on placebo (p < 0.0001).

Fig 1 compares the grading (five-point scale) of each two-month period. The reponses on feverfew and placebo differed

significantly (p < 0.001): 36% of all feverfew periods were graded as "much better" for migraine and only 1% as "much worse" compared with placebo values of 21% and 10%, respectively.

Fig 1 - Overall grading of the two-month treatment periods according to the five-point scale, expressed as a percentage of the total number of periods on each treatment.

After completing the study (and stiff "blind" to therapy) 35 (59%) patients reported that a feverfew period was more effective whereas only 14 (24%) chose placebo (p < 0.001). 10 patients (17%) felt there was no difference between treatments.

Fig 2 - Effect of each treatment on the number of attacks throughout the study.

* = significant difference between feverfew and placebo, p< 0.05.

When the 17 patients with classical migraine were considered separately, feverfew reduced the number of attacks by 32% from 4.3 (0.5) to 2.9 (0.4) (95% confidence interval 11-53%, p < 0.05), and increased the visual analogue score from 57 (5) to 78 (4) (p < 0.01). There was no difference in the proportion of attacks associated with an aura. For the 42 patients with common migraine, feverfew reduced the number of attacks by 21 %, from 4.9 (0.4) to 3.9 (0.3) (95% confidence interval 10-33%, p = 0.06) and increased the visual analogue score from 61 (3) to 72 (2) (p < 0.01).

Of the 59 patients completing the study, 42 had never previously taken feverfew. When this group was considered separately, feverfew reduced the number of attacks from 4.5 (0.4) to 3.5 (0.3), a reduction of 23% (p < 0.05, 95% confidence interval 10-35%). This improvement was reflected in the mean visual analogue score, which rose from 64 (3) with placebo to 74 (2) with feverfew (p < 0.05).

Figs 2 and 3 show the numbers of attacks and visual analogue score in the whole group throughout the nine months of the study. The results suggest that the observed differences are attributable to improvement with feverfew rather than to deterioration on its withdrawal.

There were no differences between the mean erect and supine blood pressures or heart rate on either treatment.

Fig 3 - Effect of each treatment on visual analogue score throughout the study.

* p < 0.05; ** p < 0.01.

Adverse Effects

Table IV shows the reported side-effects with each treatment. Mouth ulceration was more common during treatment with placebo, and there was no overall excess of side-effects on feverfew. Analysis of variance was used to compare laboratory investigations on each treatment; no change was seen in haematological tests, urea, creainine, electrolytes, blood sugar, or tests of liver function. There was no difference in urinalysis on the two treatments.

Table IV - SIDE-EFFECTS ON EACH TREATMENT

*Indicates withdrawal of a patient because of side-effects

Discussion

Treatment with feverfew was clearly associated with a reduction in migraine frequency, and in the vomiting associated with attacks; there was also a trend towards a reduction in migraine severity. 3 patients withdrew from the study during a feverfew phase because of failure to improve compared with 1 during a placebo phase, but this finding would not account for the observed differences between the remaining patients. Stopping regular feverfew has been associated with deterioration in migraine symptoms¹¹, and with our crossover design, which did not include a washout period between treatments, differences could have resulted from feverfew; this possibility is not supported by the data shown in figs 2 and 3.

The sesquiterpene lactones (chiefly parthenolide) in feverfew are responsible¹⁴ for the reduced secretory activity of platelets and white cells in vitro⁹, possibly by neutralisation of sulphydryl groups¹⁶. It is unclaer, however, whether these observations have any bearing on the therapeutic effect of feverfew.

Our findings have wider implications. In common with other herbal remedies, feverfew is exempt from the preliminary tests of safery required in the UK by the 1968 Medicines Act. Although no serious adverse effects have been recorded in long-term users of feverfew^11,13,17, these studies were retrospective and the patients were therefore self-selected. Our approach has been pragmatic in that we were asking our volunteers to do no more, under close medical supervision, than they were prepared to do without such supervision. Patients who take feverfew on their own initiative are likely to be seen with increasing frequency by general practitioners, physicians, and neurologists and further information is therefore required both on the potentail benefits and on safery associated with its use. In our study, treatment with feverfew did not produce any adverse effects but the period of treatment (four months) was short, especially since lifelong treatment may bei required for migraine. Information is also needed on the quality of commercial preparations of the herb. In one study¹⁸, wide differences in the anti-secretory activity were measured in different commercial products; activity was even undetectable in some brands, and, as one would expect, in all homoeopathic preparations. Moreover, the ability of feverfew to modify the natural history of migraine serves as a reminder that active agents are still readily available through health-food shops.

We thank Mrs. S. Green for preparing the capsules and for her help with analysing the results, Miss W. Antoinette Groenewegen for helping with the assay for anti-secretory activity, and Mr. P. H. Reilly for his advice on statistics. We also thank Dr. Brian Power and staff of the Department of Botany at Nottingham University for cultivating the feverfew plants, and the Nottingham and District Migraine Self-Help Group.

Correspondence should be addressed to J.J. M., Department of Medicine, University Hospital, Queen´s Medical Centre, Nottingham NG7 2UH.

References

1. Berry MI. Feverfew faces the future. Pharm J. 1984; 232:611-14.

2. Bohlmann F, Zdero C. Sesquiterpene lactones and other constituents from.Tanacetum parthenium, Phytochemistry 1986: 21:2543-49

3. Editiorial, Feverfew-a new drug or an old wives´remedy? Lancet 1985; ii: 1084

4. Hannington E, Jones RJ, Amess JAL, Wachowicz B. Migraine: a platelet disorder. Lancet 1981; ii:720-23.

5. Foward JR. Serotonin, migraine and platelets. In: Van Zwieten PA, Schonbaum E, eds. Progress in pharmacology, vol 4, no 4. Stuttgart: Gustav Fischer, 1982: 135-36.

6. Sjaastad O, Stersrad P. Appraisal of BC-105 in migraine prophylaxis. Acta Neurol Scand 1969; 45:594-600.

7. Penderson E, Moller CE. Methysergide in migraine prophylaxis. Clin Pharmacol Ther 1966; 7:520-26.

8. Friedmann AP, Merritt HH. Treatment of headache. JAMA; 163:1111-17.

9. Heptinstall S, White A, Williamson L, Mitchell JRA. Extracts of feverfew inhibit granule secretion in blood platelets and polymorphonuclear leucocytes. Lancet 1985; i: 1071-74.

10. Johnson ES, Patients who chew chrysanthemum leaves. MIMS Magazine 1983; May 15: 32-35.

11. Johnson ES, Kadam NP, Hylands DM, Hylands PJ. Efficacy of feverfew as prophylactic treatment of migraine. Br. Med J 1985: 291:569-73.

12. Blau JN. Towards a definition of migraine headache. Lancet 1984; i: 444-45.

13. Johnson ES. Feverfew. London: Sheidon Press, 1984:71-86.

14. Department of Health and Social Security. Guide to good pharmaceutical manufacturing practice. London: HM Stationery Office, 1977.

15. Groenewegen WA, Knight DM, Heptinstail S. Compounds extracted from feverfew that have anti-secretory activity contain an a-methylene butyrolactone unit. J Pharm Pharmacol 1986; 38: 709-12.

16. Heptinstail S. Groenewegen WA, Spangenberg P., Loesche W. Extracts of feverfew may inhibit platelet behaviour via neutralization of sulphydryl groups. J. Pharm Pharmacol 1987; 39: 459-65.

17. Johnson ES, Kadam NP, Anderson D, Jenkinson PC, Dewdney RS, Blower SD. Investigation of possible genotoxic effects of feverfew in migraine patients. Hum Toxicol 1987; 6: 533-34.

18. Groenewegen WA, Heptinstail S. Amounts of feverfew in commercial preparations of the herb. Lancet 1986; i: 44-45.