512
an abnormal glucose-tolerance test by conventional criteria.4
Fajans and Conn define normal glucose tolerance as a one-
hour blood-sugar concentration <160 and a two-hour <110
mg/dl and require an elevation of both to >160 and >120
mg/dl, respectively, for the diagnosis of diabetes.4 Others have
considered either a one-hour "true" blood-sugar concentration
> 1605 or a two-hour >130 mg/das diabetes. When the
blood-sugar concentration by the Nelson-Somogyi method
used by Fajans and Connwas converted to serum-glucose con-
centration by the glucose-oxidase method used in this study,2&middot;3
the Fajans and Conn values of 160 and 120 mg/dl became 166
and 125 mg/dl, respectively. By these criteria, one of the four
excluded patients, with a one-hour serum-glucose concentra-
tion of 217 and a two-hour of 175 mg/dl, had unequivocal dia-
betes. Of the other three, two had an elevated serum-glucose
concentration at one hour of 175 and 179 mg/dl, respectively,
and one at two hours of 142 mg/dl, suggesting that these pa-
tients may also have had diabetes.56 None of the other patients
or controls had any elevated values in the glucose tolerance
test.
You also state that "the restradiol/testosterone (E/T) ratios
were significantly higher in the atherosclerotic than in the con-
trol subjects". As noted in the paper, the mean E/T value was
not significantly higher in the patients. This point is discussed
on p. 1732 of my paper.3
You argue for the hypothesis that hyperinsulinaemia is ath-
erogenic. However, glucose intolerance, hypercholesterolaemia,
hypertriglyceridaemia, and hyperoestrogenaemia2as well as
hyperinsulinx-mia appear to be associated to a high degree
with coronary heart-disease.7 Of these, a cause-and-effect rela-
tionship has been demonstrated only between cestrogen and
myocardial infarction in man.8 The possibility arises that the
glucose, insulin, and lipid abnormalities may even be inciden-
tal to the development of myocardial infarction, since my
study suggests that these abnormalities are secondary to an in-
crease in E/T, which tends to accompany hyperoestrogenxmia.3
To single out hyperinsulinaemia and to attribute to it "a causal
role" in atherosclerosis, as your editorial does, largely on the
basius of an increase in the lipid content of arteries in animals
given insulin,9 may be unwarranted.
College of Physicians and Surgeons
of Columbia University,
The Roosevelt Hospital,
428 West 59th Street,
New York, New York 10019, U.S.A. GERALD B. PHILLIPS
CIMETIDINE AND MENTAL CONFUSION
Dr Grimson (April 16, p. 858) of two
SiR,&mdash;The report by treatment
cases of confusion after with cimetidine (Tagamet)
prompts us to report a recent case.
A 74-year-old man presented with epigastric pain, radiating
through to his back. Thirty years previously he had undergone
partial gastrectomy for duodenal ulcer. His general practi-
tioner had assumed that his pain was due to indigestion and
started him on cimetidine at the recommended dose of 200 mg
three times a day and 400 mg at night z0 g a day). Within
a few days he became extremely confused, agitated, restless,
and unmanageable at home. Physical examination revealed no
abnormality, but a chest radiograph showed slight inflamma-
tory shadowing in the right upper zone, which had resolved a
few days later. Within twenty-four hours of the withdrawal of
4. Fajans, S. S., Conn, J. W. Ann. N.Y. Acad. Sci. 1959, 82, 208.
5. Marble, A. in Diabetes Mellitus (edited by A. Marble, P. White,m R. F.
Bradley, L. P. Krall); p. 202. Philadelphia, 1971.
6. Wld Hlth Org. techn. Rep. Ser. 1965, no. 310.
7. Tzagournis, M., Chiles, R., Ryan, J. M., Skillman, T. G. Circulation, 1968,
38, 1156.
8. J. Am. med. Ass. 1970, 214, 1303.
9. Stout, R. W. Br. med. J. 1970, iii, 685.
cimetidine the confusion disappeared and he had become quite
rational.
The two patients previously reported had had doses of cime-
tidine greater than those recommended by the manufacturers.
This patient did not, but, at the age of 74, it is likely that renal
function was impaired and the serum level of the drug was
high. Nevertheless, whatever the explanation, it seems that
mental confusion should be added to the list of possible side-
effects of cimetidine.
Whiston Hospital,
Prescot, J. C. DELANEY
Merseyside L35 5DR M. RAVEY
MIGRAINE, THROMBOCYTOPENIA, AND
SEROTONIN METABOLISM
SIR,-The unexpected onset and disappearance of migraine
headaches in a patient with haemolytic anaemia, coinciding
with the fall and rise of platelet-count, deserves attention.
Platelets are important stores of serotonin, and changes in ser-
otonin metabolism have been implicated in the pathogenesis of
migraine.
A 35-year-old right-handed female had been in excellent
health until October, 1975, when she had respiratory infection
followed 5 days later by severe vaginal bleeding, generalised
petechiae, and skin bruises. She was admitted to hospital. Her
red-blood-cell count was 3-74x 10%1.1, Hb 9.4 g/dl, haonatocrit
28%, and platelet-count 30 000/1. Hw-molytic anxmia was
diagnosed and steroids were prescibed. Shortly after admission
she had severe headaches preceded by blurred vision and mov-
ing scotomas in both eyes which lasted about 15 min. The pain
was severe, sharp, and throbbing, starting in the left frontal re-
gion and spreading to the occipital region of the same side.
Nausea, occasional vomiting, and phonophobia and photo-
phobia accompanied the pain. The headaches lasted about 2 h
and appeared twice daily. Bright illumination and/or flickering
lights triggered the headaches. The patient had never experi-
enced similar symptoms and she had no family history of
migraine.
On Nov. 4 the patient was transferred to University Hospi-
tals in Iowa City. Her platelet-count was 79 000 and continued
to rise for the next 4 days. At the same time the frequency
of headaches fell, and by Nov. 8 they were gone. On Nov. 8
the platelet count was almost normal. The patient left hospital
on Nov. 14, with a normal platelet-count and headache-free.
The discharge diagnosis was autoimmune hxmolytic anxna
with autoimmune thrombocytopenia.
In late 1976 she again bruised easily, and by January, 1977,
her headaches had reappeared; their frequency was twice a
week but some attacks lasted 17 h. In February the platelet-
count was 110 000 but it subsequently fell and she was re-
admitted on March 2, with a platelet-count of 13 000. The
headaches were by then a cause of major distress.
She was again given steroids; the platelet-count rose slowly,
and the headaches became less severe. By March 11 the head-
aches had gone, and the patients platelet-count was 123 000.
As in 1975 the Hb and hxmatocrit values remained below nor-
mal.
During both admissions the patient was thoroughly invest-
gated by neurologists but the only abnormality was that
electroencephalograms during headache periods showed occa-
sional mixed theta and delta discharges, more noticeable from
the left temporal leads, an electrophysiological finding fre-
quently associated with migraine. An E.E.G. during a head-
ache-free period on March 24, 1977, showed no abnormalities;
at that time the platelet-count was 241 000.
This patients headaches fit the classic description of
migraine, but the fact that they started in clusters, an unusual
early presentation for typical idiopathic migraine, suggests an
associated disease process of which migraine is the expression.
The patient had migraine only during periods of thrombocy-

topenia and anaemia. However, the anaemia was never fully
controlled, unlike the thrombocytopenia. The only accompani-
ment to both the onset and the improvement in the headache
was a change in platelet-count: the coincidence was striking.
Migraine headaches have been associated with serotonin
metabolism since Sicuteri et at. found an increase in the uri-
nary excretion of 5-hydroxyindoleacetic acid (5-H.I.A.A.) dur-
ing migraine attacks. This finding has been confirmed by
somebut not by others.3&deg; The disparity may be linked to the
heterogeneity of patient types usually included in studies of
migraine. Furthermore, plasma-serotonin decreases during a
migraine attack, and during such attacks the urinary excretion
of serotonin is increased even though 5-H.I.A.A. excretion
remains normal. Since serotonin constricts large arteries and
veins and dilates of arterioles, and capillaries, a fall in circulat-
ing serotonin could explain the extracranial vasodilatation
found during attacks and the capillary constriction which
causes the typical pallor of the migrainous patient. But what
causes the fall in serotonin?
One important store of serotonin is the platelets, and the
platelets have received special attention in studies of migraine.
They aggregate more readily in the presence of serotonin; their
serotonin content falls during a migraine attack and free sero-
tonin decreases. It has been suggested that free serotonin is
more rapidly picked up from the circulation, more quickly
metabolised, and excreted at an accelerated rate.8 It has also
been proposed that during migraine attacks there is a primary
depression of platelet monoamine-oxidase activity which might
be linked to abnormal metabolism of serotonin, in the absence
of changes in platelet-count.9 On the other hand, ultrastruc-
tural studies of platelets of migraine patients have not demon-
strated any abnormality. 10
In thrombocytopenia changes in serotonin content and
metabolism can be expected-indeed patients do have a de-
creased serum-serotonin when platelet-counts fall below
100 000.11 It is likely that circulating serotonin levels in my pa-
tient became very low during the thrombocytopenic crisis and
possibly abruptly. I suggest that the rate of lowering of sero-
tonin level may also be a factor in the sudden onset of a
migraine attack. Sudden changes in platelet-count might be
related to variation of serotonin level and thus contribute to
the production of migraine, in the absence of other clinical
signs.
Department of Neurology,
University of Iowa Hospitals and Clinics,
Iowa City, Iowa 52242, U.S.A. HANNA DAMASIO
THE ANION GAP
SiR,-Your editorial on the anion gap (April 9, p. 785) drew
attention to and condensed the valuable review by Emmett
and Narins.12
Because Dr Patel and Dr Wright (May 14, p. 1054) found
an increased anion gap in 51% of 94 patients aged 65-90 they
suggested that anion-gap results were less valuable in the
1. Sicuteri, F., Testi, A., Anselmi, B. Int. Archs Allergy, 1961, 19, 55.
2. Curan, D. A., Hinterberger, H., Lance, J. W. Brain, 1965, 88, 999.
3. Curzon, G. et al. J. Neurol. Neurosurg. Psychiatry, 1966, 29, 85.
4. Anthony M., Lance, J. W. in Modern Topics in Migraine (edited by J.
Pearce); p. 107. London, 1975.
5. Anthony, M. Paper read at sixth international Migraine Symposium, held in
London, in 1974.
6. Hilton, B. P., Cumings, J. N. J. Neurol. Neurosurg. Psychiat. 1972, 35, 505.
7. Anthony M., Hinterberger, H., Lance J. W. Res. clin. Stud. Headache, 1969,
2, 29.
8. Sommerville, B. W. Neurology, 1976, 26, 41.
9 Sicuteri, F., Buffoni, F., Anselmi, B., Del Bianco, P. L. Res. Clin. Stud.
Headache, 1972, 3, 245.
10. Grammeltvedt, A., Headache, 1975, 14, 226.
11. Bigelow, F. S. J. Lab. clin. Med. 1954, 43, 759.
12. Emmett, M., Narins, R. Medicine, 1977, 56, 38.
513
elderly. However, we found an increased anion gap in 31 % of
a hundred consecutive acutely ill inpatients aged 65-93 years
and in 20% of fifty consecutive outpatients aged 65-86 years.
Males constituted 60% of both our groups of patients all of
whom were patients of a large general teaching hospital. We
suggest that the high frequency of increased anion gap in the
series investigated by Dr Patel and Dr Wright is more apparent
than real, since they calculated the anion gap by the formula
(Na++K+) - (Cl-+HCO ), which has a reference range of
11-19 meq/l,"but they interpreted their results according to
the reference range of 8-16 meq/1 which belongs to the other
formula, (Na+) - (Cl-+HCO3-), as cited in your editorial.
Experience of more than 10 years routine reporting of the
anion gap and discussion with clinical colleagues at the bedside
has convinced the staff of this laboratory of the value of this
calculation in patient management.
Division of Clinical Chemistry,
Institute of Medical and Veterinary
A. PRIOR
Science,
Frome Road, Adelaide, P. J. PHILLIPS
South Australia 5000 R. W. PAIN
Hb G NORFOLK ASSOCIATED WITH MALIGNANT
MYELOSCLEROSIS
SIR,-It is now usual to perform haemoglobin electrophor-
esison patients presenting with a haematological malignancy.
We have investigated an 81-year-old man who presented with
malignant myelosclerosis. On electrophoresis an abnormal
haemoglobin band was detected which migrated between HbA
and HbS. The HbF level was less than 1% and the HbA2 level
was 2.5%. The abnormal haemoglobin was identified as HbG
Norfolk &agr;85(F6) Asp-Asn, and a slowly migrating band Gz
was also present (04%).
The association between leukxmia and abnormal hoemoglo-
bin production is established. However, this association is most
commonly seen in erythraemic myelosis.2&middot;3 The most common
abnormality is an imbalance in the synthesis of either the a or
the j3 chains. However, structural haemoglobin variants have
been found (White, J. M., personal communication). Also,
there are occasional changes in the HbA2 and HbF levels.
These are not confined to leukaemia, but are commonly found
in many conditions associated with dysh&aelig;mopoiesis.4
The finding of an abnormal haemoglobin in this patient sug-
gested that the abnormality resided within the erythroid line.
However, when it was found to be HbG Norfolk, it was rea-
lised that this was a fortuitous finding. Family studies con-
firmed this. Both of the affected relatives are hsematologically
normal.
Three cases of HbG Norfolk have been described. The first
was found during a survey to assess the frequency of the fast
moving Hb Norfolk (&agr;2Gly&rarr;Asp&bgr;2).5 The characterisation of
HbG Norfolk in the other two cases
was described by Lorkin
et al.6 One of these cases was in an English child from Swindon
who had leukaemia, and it was also present in the father unas-
sociated with any other haematological abnormality.
We thank Prof. H. Lehmann, F.R.S., for identifying the abnormal
haemoglobin as G Norfolk.
Kings College Hospital Medical School, D. R. HIGGS
University of London,
Department of H&aelig;matology, B. FROST
Denmark Hill, London SE5 8RX J. C. SHARP
13. Thomas, D. W., Pain, R. W., Duncan, B. M. Lancet, 1973, ii, 848.
1 Lubin, J., Rosen, S., Rylwin, A. M. Arch. intern. Med. 1976,136, 141.
2. White, J. C., and others. Br. J. H&oelig;mat. 1960, 6, 171.
3. Rosenzweig, A. I., and others. Acta h&oelig;mat. 1968, 39, 91.
4. Bradley, T. B., Ranney, H. M. Progr. H&oelig;mat. 1974, 8, 77.
5. Huntsman, R. G., Hall, M., Lehmann, H., Sukumaran, P. K. Br. med. J.
1963, i, 720.
6. Lorkin, P. A., Huntsman, R. G., Ager, J. A. M., Lehmann, H., Vella, F.,
Darbre, P. D. Biochim. biophy. Acta. 1975, 379, 22.