6

13 patients from whom anticoagulants were withdrawnhad a recurrence of their T.I.A. between two and eightweeks after withdrawal was begun. At the time T.i.A.srecurred the prothrombin level in each patient was outsidethe therapeutic range. Over the same period only 2 of the13 patients in whom treatment was continued had T.I.A.S.This difference is significant at the 5% level (one-tailtest). The patients from whom anticoagulants were beingwithdrawn had 25 T.i.A.s in all during this period, whereasthose who continued on anticoagulants had only 4. Thepatients in the withdrawal group who had further T.I.A.Shad effective therapy restored immediately; they and theremaining patients were kept under observation for fourmonths from the beginning of the trial, but no furtherT.I.A.s occurred. The 5 patients in the withdrawal groupwho had no T.I.A.s remained without therapy after theconclusion of the trial at four months.

Equally striking was the distribution of the recurrencesaccording to the duration of therapy before withdrawal.In the carotid group, all 5 patients who had been treatedfor twelve months or less had a recurrence, whereas the 2who had been treated for longer had no recurrence of theirattacks. The same tendency was present, though not soclearly, in the vertebrobasilar group.

Discussion

The recurrence of the T.I.A.s after the gradual with-drawal of anticoagulants was most striking. Observationof the current control group left no doubt about the signi-ficance of the increased incidence of the attacks in the

patients from whom anticoagulants were being withdrawn,particularly since during the period of therapy morepatients in the control group had been liable to haveT.I.A.S. Moreover, the precautions taken to obviate anyconfusing psychological disturbances made it clear thatthe increase was directly related to the alteration in theprothrombin level.A possible contribution to the increased incidence of

T.i.A.s by a " rebound " response has to be taken into

account. Had this, been significant, however, it shouldhave been operative without regard to whether the patienthad been treated for a shorter or longer period before thewithdrawal of anticoagulants. The fact that recurrencewas practically confined to those patients who had beentreated for twelve months or less makes it extremelyunlikely that this was solely a

" rebound " response. Itseems probable that these patients were at a stage whenthey were still liable to T.i.A.s—a liability from which theywere being protected by anticoagulants. The patients whohad been treated for more than twelve months had

apparently passed this stage in the evolution of theircerebrovascular disease.These observations confirm the value of anticoagulant

therapy in reducing the incidence of T.LA.S. They suggestthat in practice a patient should be treated initially for atleast twelve months. Thereafter, anticoagulants may begradually withdrawn whilst the patient is kept under closeobservation. In the event of recurrence, treatment shouldbe restored for a further protracted period.

SummaryWithdrawal of anticoagulant therapy from 13 patients

who had previously had transient ischsemic cerebrovascularattacks was followed by a recurrence of attacks in 8. Inthe same period only 2 out of 13 patients who continuedon anticoagulants had a recurrence of attacks. Patientswho had been treated for less than a year were most liableto recurrence.

We should like to thank Prof. P. Armitage for advice about thestatistical design and analysis; the British Heart Foundation, forfinancial support; and Evans Medical for the supply of ’ Dindevan ’tablets.

REFERENCES

Baker, R. N., Broward, J. A., Fang, H. C., Fisher, C. M., Groch, S. N.,Heyman, A., Karp, H. R., McDevitt, E., Scheinberg, P., Schwartz, W.,Toole, J. F. (1962) Neurology, Minneapolis, 12, 823.

Hill, A. B., Marshall, J., Shaw, D. A. (1960) Quart. J. Med. 29, 597.Marshall, J. (1963) Circulation, 28, 329.- (1964) Quart. J. Med. 33, 309.

Poller, L., Thomson, J. (1964) Lancet, ii, 62.Report of the Veterans Administration Cooperative Study of Atherosclerosis:

Neurology Section (1961) Neurology, Minneapolis, 11, 132.Siekert, R. G., Millikan, C. H., Whisnant, J. P. (1961) J. Amer. med. Ass.

176, 19.

LONG-TERM ANTICOAGULANT THERAPYIN TRANSIENT CEREBRAL

ISCHÆMIC ATTACKS

J. M. S. PEARCEM.B. Leeds, M.R.C.P.

LATE REGISTRAR

S. S. GUBBAYM.B. Adelaide, M.R.A.C.P.

RESEARCH ASSOCIATE

JOHN N. WALTONM.D. Durh., F.R.C.P.

NEUROLOGIST

REGIONAL NEUROLOGICAL CENTRE, GENERAL HOSPITAL,NEWCASTLE UPON TYNE

ANTICOAGULANT therapy has been recommended fortransient cerebral ischasmic attacks involving either thecarotid or the vertebrobasilar arteries because there isstrong evidence that most, if not all, of the ischaemicattacks are due to recurrent microembolism, often arisingfrom mural thrombosis in large vessels. The repetitivepattern of the attacks is explained by laminar flow in thecerebral circulation. If, as has been presumed, mostsuch microemboli consist of agglomerations of platelets,anticoagulant therapy might well be beneficial, butcholesterol emboli (Balla et al. 1964) could be uninfluencedby this treatment. Millikan et al. (1958), and Siekertet al. (1961) reported a striking reduction in the numberof transient ischsemic attacks and a diminished incidenceof subsequent cerebral infarction in patients treated withanticoagulants, but in neither of these studies were theresults adequately controlled. Fisher (1958) also describedbenefit from anticoagulant therapy, but his series includedonly 14 patients who presented with the syndrome oftransient ischsemic attacks. In a national cooperative

, study reported from the U.S.A. by Fisher (1961) 17

patients with transient ischaemic attacks who receivedanticoagulant therapy were contrasted with 15 controls.The conclusion was that the number of attacks wasreduced by anticoagulants, but 7 of the 15 controls hadfour or fewer than four further attacks, and 2 of the other8 accounted for most of the total of five or six’hundredattacks recorded in the control group. Because cerebralinfarction happened so seldom in either the treated casesor the controls in Fisher’s series, no definite conclusionsabout the benefit of long-term treatment could be drawn.Since the value of anticoagulants seemed still in doubt, wedecided to embark upon a controlled trial of treatment.

Patients and Methods

All patients with cerebrovascular disease admitted to theregional neurological centre between Oct. 1, 1962, and March31, 1964, were assessed independently by two of us as to theirsuitability for inclusion in the trial. Several otherwise suitable

patients were excluded solely because the clinical and radio-logical investigations revealed carotid stenosis, which was latertreated surgically. A total of 37 patients were eventually

7

included in the trial, and each fulfilled the following criteria:1. Age less than 70 years.2. Each patient had had one or more attacks of transient

cerebral dysfunction that had lasted less than 24 hours, andwas attributable to atherosclerotic cerebrovascular insuffi-ciency. In most instances, the attacks lasted much less thanan hour. Patients with a residual neurological deficit persist-ing for longer than 24 hours were excluded.

3. None of the patients had evidence of serious inter-current disease or of any contraindication to anticoagulanttherapy (these included such disorders as malignant hyper-tension’and chronic peptic ulceration).Each. patient was examined clinically. This examination

TABLE I-AGE-DISTRIBUTION AND SEX-DISTRIBUTION

included a basal blood-pressure recording, a full blood-count,measurements of the erythrocyte-sedimentation rate, electro-cardiography, and X-rays of skull and chest. To lessen the

possibility of including cases of cerebral haemorrhage in thetrial, the cerebrospinal fluid was examined in all patients whowere admitted within a week of an attack. Most of the patients,except those whose symptoms indicated that the insufficiencywas limited to the distribution of the vertebrobasilar system weresubjected to carotid angiography. Vertebral angiography wasnot performed because of the hazards of this technique inpatients with cerebrovascular disease.

Carotid angiography was performed in 29 patients-in 17bilateral. Atheromatous changes were detected in 17 patientsbut there were no changes in the other 12. None of the patientsin the trial had significant narrowing of the extracranial portionsof the internal carotid arteries, but 2 of them displayednarrowing in the region of the carotid siphons.

Patients who were accepted for the trial were randomlyallocated to high-dosage and low-dosage groups. Patients inthe former group were treated with phenindione (50 mg.tablets) and, after stabilisation in hospital at the therapeuticlevel of between 10% and 25% prothrombin activity (Quick one-

TABLE II-CLINICAL DIAGNOSIS

stage method, using a saline extract of human brain as thesource of thromboplastin), they were discharged and werereviewed in a clinic at weekly to monthly intervals for estima-tion of prothrombin activity and for regulation of dosage. Ateach visit any relevant symptom which had been recorded, asinstructed, by the patients was noted. Cases in the control

group were discharged from hospital on treatment withphenindione in ineffective dosage (1 mg. tablets of identicalappearance, once daily) and were similarly reviewed in theclinic at four-weekly to six-weekly intervals. To the best ofour knowledge, no patient knew whether he or she was

receiving the higher or the lower dosage.The design of the trial was similar to that adopted by Hill

et al. (1960).

TABLE III-INCIDENCE OF HYPERTENSION

Results

Table I shows that the age-distribution and sex-distri-bution were roughly comparable in the high-dosage andlow-dosage groups; the average age in the high-dosagegroup was 58 years and in the low-dosage group 54 years.The diagnosis in each patient was transient. cerebralischasmia of either the carotid or the vertebrobasilar

system; when the symptoms were mixed, the vascularterritory predominantly affected by the attacks was usedin classifying the cases (table 11). The distinction betweenvertebrobasilar and carotid insufficiency was not alwaysclear-cut; hence the classification was in some waysartificial. Nevertheless, 25 patients (14 in the high-dosage group and 11 in the low-dosage group) had hadattacks which, indicated that the carotid circulation waspredominantly affected, and 12 (3 in the high-dosagegroup, and 9 in the low-dosage group) had had attacks

TABLE IV-LENGTH OF TREATMENT

which pointed to predominantly vertebrobasilar ischsemia.Table ill gives the incidence of hypertension, and table ivsummarises the duration of treatment, the mean duration

being 11-1 months in the high-dosage group and 10-6months in the controls.

Tables v, vi, and vil show that anticoagulant therapydid not significantly influence the occurrence of furtherattacks of transient cerebral ischaemia—41 % of patientsin the high-dosage group and 55% in the low-dosagegroup had no further attacks. In the carotid-artery-insufficiency group, 50% of the treated cases and 36% ofthe controls were free from attacks; but the differencewas not significant. A more quantitative method ofanalysing the results was obtained by estimating the

proportion of patient-months in which there were noattacks. It was impracticable to count the total number ofattacks, because some patients had only one further

TABLE V-EFFECT OF TREATMENT ON INCIDENCE OF FURTHER ATTACKS

IN ALL PATIENTS

8

TABLE VI-EFFECT OF TREATMENT ON INCIDENCE OF FURTHER ATTACKS

IN PATIENTS WITH CAROTID-ARTERY INSUFFICIENCY

attack and others had a hundred or more attacks whichcould be assessed only very roughly. Tables vm-xsummarise the effect of treatment on the proportion ofpatient-months free from attacks; and, again, no benefitis shown to accrue from anticoagulant therapy. At first

sight table ix seems to indicate some slight benefit fromanticoagulant therapy in patients with carotid-arteryinsufficiency, but 12 of the 20 patient-months in whichfurther attacks occurred in the control group involveda single patient.

Table xi shows that only 1 patient in each group sus-tained a non-fatal completed stroke, and 1 control patientdied from a stroke within an hour of becoming hemiplegic.2 other control patients died from cardiac causes. 1of them died suddenly after an attack of very severechest pain, and the other died suddenly after becomingextremely breathless and cyanosed. In no instance was

TABLE VII-EFFECT OF TREATMENT ON INCIDENCE OF FURTHER ATTACKS

IN PATIENTS WITH VERTEBROBASILAR INSUFFICIENCY

a necropsy obtained. No patients had any major haemor-rhagic complications, and no instance of minor bleedingwas definitely attributable to anticoagulant therapy. 2

patients in the high-dosage group had drug rashes fromphenindione, but in both cases this drug was immediatelyreplaced by warfarin sodium with resolution of the

hypersensitivity reactions.f Discussion

Patients in the high-dosage group occasionally escapedfrom control for a week or more at a time, and some of thetransient cerebral ischxmic attacks may have come on

during these phases. Escape from control could possiblyhave had some bearing on the absolute number of attacks,but this factor should largely have been eliminated in the

TABLE VIII-EFFECT OF TREATMENT ON TOTAL NUMBER OF MONTHS IN

WHICH ALL PATIENTS WERE FREE FROM ATTACKS

quantitative assessments (tables vill-x). Still, this is a

well-recognised shortcoming of anticoagulant therapy,particularly in outpatients and, even with most painstakingcontrol, it can hardly be eliminated.Only 37 patients took part in this trial and some of the

results cannot be regarded as valid, because of the smallnumbers in some subgroup. Fortuitously, only 3 patientswith vertebrobasilar ischaemia received high-dosagetherapy; hence, it was impossible to judge the efficacy ofanticoagulant therapy in this subgroup.

In roughly half the control patients with recurrent

TABLE IX-EFFECT OF TREATMENT ON TOTAL NUMBER OF MONTHS IN

WHICH PATIENTS WITH CAROTID-ARTERY INSUFFICIENCY WERE FREE

FROM ATTACKS

* 12 of these 20 patient-months with further attacks occurred in a singlepatient. ,

transient cerebral ischasmic attacks -(even as many as twoor three daily) the attacks ceased spontaneously duringthe period of initial observation, which was usually abouta month. This tendency to spontaneous " cure " musthave influenced many clinicians in the past in theirfavourable view of anticoagulant therapy.

Marshall (1964), in an investigation of the natural

history of transient ischaemic cerebrovascular attacks,remarked that only 1 of 61 patients who first came underhis observation after experiencing only transient ischsemic

TABLE X-EFFECT OF TREATMENT ON TOTAL NUMBER OF MONTHS

IN WHICH PATIENTS WITH VERTEBROBASILAR-ARTERY INSUFFICIENCY

WERE FREE FROM ATTACKS

attacks, subsequently had a major episode of infarction.These patients were followed up for an average period ofabout 2 years. This good prognosis reinforces the

impression that anticoagulants are unlikely to give addi-tional protection against further attacks. Although therewere remarkably few complications during our trial, it iswell-established that anticoagulant therapy is not withoutrisk; moreover, the treatment is often inconvenient for thepatient and time-consuming for medical and laboratorystaff. After reviewing the investigations by various

groups of workers on the role of anticoagulant therapyin cerebrovascular disease, Shaw (1962) concluded that

9

TABLE XI-INCIDENTS OTHER THAN ISCHAMIC ATTACKS ARISING DURING

THE TRIAL

the scope and application of anticoagulants in neurologyhad tended to diminish with the passage of time. Ourobservations confirm this statement.

We appreciate the defects of this trial: the number ofpatients was small, the length of treatment was com-paratively short, and the distribution of carotid andvertebrobasilar cases in the high-dosage and low-dosagegroups was not as even as we would have wished-a not

unimportant disqualification, since there is evidence thatthe prognosis of carotid-artery insufficiency is lessfavourable than that of vertebrobasilar-artery insufficiency(Marshall 1964). We cannot claim from our results tohave proved that anticoagulant therapy has no applicationin recurrent cerebral ischaemia of all types; but plainly itsvalue, if any, must be limited, and the indications forprescribing this form of treatment remain uncertain.Cases of recurrent cerebral ischsemia fulfilling the rigiddiagnostic criteria already described are uncommon;and a large enough series of cases is unlikely to be collectedfrom a single centre for the type of long-term trial whichis clearly needed. Pending the results of a trial in a seriessubstantially larger than ours, we can state only that wehave failed to demonstrate any significant benefit fromanticoagulant drugs, and that the current widespreadassumption that recurrent cerebral ischxmic attacks arecontrolled by this form of treatment must be regarded as" not proven ".

I

SummaryA controlled trial of anticoagulant therapy was under-

taken in 37 patients presenting with transient cerebralischxmic attacks. 17 patients received phenindione ineffective dosage over an average period of ll’l months;20 control patients were followed for an average of 10-6months.

The results suggest that long-term anticoagulanttherapy is of no significant benefit to these patients as awhole; the number of cases was too small to judge theeffect of this treatment in vertebrobasilar-artery insuffi-ciency. Though not absolutely conclusive, our resultscast doubt on the value of anticoagulant treatment intransient cerebral ischaemic attacks.

We thank the staff of the department of hxmatology, NewcastleGeneral Hospital, for carrying out the estimations of prothrombinactivity; Dr. D. J. Newell for his statistical advice; and Dr. D. A.Shaw for helpful suggestions. The 1 mg. tablets of phenindione weresupplied by Weddell Laboratories Ltd.

REFERENCES

Balla, J. I., Howat, J. M. L., Walton, J. N. (1964) J. Neurol. Psychiat. 27, 144Fisher, C. M. (1958) Neurology, Minneapolis, 8, 311.

— (1961) ibid. 11 (no. 4, pt. 2), 119.

Hill, A. B., Marshall, J., Shaw, D. A. (1960) Quart. J. Med. 29, 597.Marshall, J. (1964) ibid. 23, 309.Millikan, C. H., Siekert, R. G., Whisnant, J. P. (1958) J. Amer. med. Ass.

166, 587.Shaw, D. A. (1962) in Modern Trends in Neurology (edited by Denis

Williams); p. 108. London.Siekert, R. G., Millikan, C. H., Whisnant, J. P. (1961) J. Amer. med. Ass.

176, 19.

PHENFORMIN IN RHEUMATOID ARTHRITIS

A FIBRINOLYTIC APPROACH

G. R. FEARNLFYM.D. Lond., F.R.C.P.

PHYSICIAN

R. CHAKRABARTIM.B. Calcutta

RESEARCH REGISTRAR

ELIZABETH D. HOCKINGM.B. Cantab., M.R.C.P.E.

MEDICAL REGISTRAR

GLOUCESTERSHIRE ROYAL HOSPITAL, GLOUCESTER

With the technical assistance ofJOHN EVANS

THE finding that blood fibrinolytic activity tended to belower in patients with rheumatoid arthritis and other

inflammatory conditions than in healthy controls and wasincreased by treatment with corticosteroids (Chakrabartiet al. 1964) suggested that inadequate fibrinolysis might bea factor in maintaining chronic inflammation, and that theinfluence of corticosteroids on fibrinolysis could be a

component of their anti-inflammatory action. The conceptof imbalance between coagulation and fibrinolysisadvanced for thrombosis (Astrup 1956, Fearnley 1961)was thus extended to inflammation-another area wherethe two systems appear to be linked by the deposition anddissolution of fibrin. The hypothesis would receive

convincing support if a substance, devoid of any knownanti-inflammatory properties but,capable of increasing thefibrinolytic activity of the blood, were found to exert

favourable effects in a chronic inflammatory disease. Onesuch compound, the antidiabetic agent phenformin, hasbeen shown to increase blood fibrinolytic activity in

patients with occlusive vascular conditions (Fearnley andChakrabart 1964a and b), but has no prior claim to beregarded as an anti-inflammatory drug. We here reportthat phenformin produces clinical and laboratory evidenceof improvement in patients with rheumatoid arthritis.

Patients and Methods

16 patients (5 men and 11 women) with unequivocal rheuma-toid arthritis were admitted to hospital. After a control periodduring which their blood fibrinolytic activity, erythrocyte-sedimentation rate (E.S.R.), plasma-fibrinogen level, strength ofgrip, and clinical status were measured to obtain a baseline,they were given phenformin by mouth and the measurementswere continued. In 8 of the patients who responded to treat-ment phenformin was discontinued, and in 7 of these the drugwas replaced by a placebo. Phenformin was given again to5 of these 7 because of the relapse of their symptoms. It wasthus possible to confirm that the clinical and laboratory changesobserved were due to phenformin and not simply to rest inhospital. At first a placebo of dissimilar appearance was given;latterly, dummy capsules of identical appearance, supplied byBayer Products, were used.Clinical Assessment

The hand-grips were measured daily at 11-11.30 A.M. with aspecial bag connected to a mercury sphygnomanometer andinflated to a pressure of 20 mm. Hg. The results are given asthe means of the grips of both hands. All these patients hadactive disease of the joints of the hands. Daily notes were madeof symptoms, joints affected, and mobility as judged by theusual criteria-confined to bed, sitting in a chair, walking, &c.

Laboratoi-3, AssessmentBlood-samples were obtained at 8.30 A.M. in the fasting,

resting state, daily for the measurement of fibrinolytic activity,and twice to thrice weekly for estimation of the E.S.R. and

plasma-fibrinogen level. Blood fibrinolytic activity was

estimated in duplicate by the dilute blood-clot-lysis timemethod of Fearnley et al.(1957) with reading of the end-pointmodified as described by Fearnley (1964). When necessary,