828

V-segment usage in T cell receptors from cytotoxic T lymphocytes specific for a major epitope of lymphocytic chromomeningitis. J Virol 1991; 64: 5919-26. 28. Brandle D, Burki K, Wallace V, et al. Involvement of both T cell receptor V&agr; and V&bgr; variable region domains and &agr; chain junctional region in viral antigen recognition. Eur J Immunol 1991; 21: 2195-202. 29. Via CS, Morse HC, Shearer GM. Altered immuno-regulation and autoimmune aspects of HIV infection: relevant murine models. Immunol Today 1990; 11: 250-55. 30. Clouse KA, Cosetine L, Weih KA, et al. The HIV envelope has the intrinsic capacity to stimulate monokine secretion. J Immunol 1991; 147: 2892-901. 31. Rieckmann P, Poli G, Fox C, Kehrl J, Fanci AS. Recombinant gp120 specifically enhances TNF production and Ig secretion. J Immunol 1991; 147: 2922-27. 32. Di Giorgi L, Habeshaw JA, Povey S, Matossian Rogers A. Reduction of

GVHD in neonatal Fl hybrid mice. Clin Exp Immunol 1990; 79: 130-34. 33. Knulst AC, Bazuin CB, Benner R. Prevention of lethal GVHD by a single low dose infection of anti-T cell monoclonal antibody to allograft receptor. Eur J Immunol 1991; 21: 103-07. 34. Howell M, Winters S, Olee T, Powell H, Carlo D, Brostoff S. Vaccination against experimental allergic encephalomyelitis with T cell receptor peptides. Science 1989; 246: 668-70. 35. Heney D, Norfolk DR, Bailey CC, Lewis J, Barnard DL. Thalidomide treatment for GVHD. Br J Haematol 1991; 78: 23-27. 36. Kion T, Hoffman G. Anti-HIV and anti anti-MHC antibodies in alloimmune and autoimmune mice. Science 1991; 253: 1138-41. 37. Stott J. Anti cell antibody in Macaques. Nature 1991; 353: 393. 38. Dalgleish AG. Viruses and cancer. Br Med Bull 1991; 47: 21-46. 39. Schattner A, Zisman BR. Virus induced autoimmunity. Rev Infect Dis

1990; 12: 204-22.

Conservative treatment of mild/moderate cervical

dyskaryosis: long-term outcome

There is some controversy about the management of women with mild or moderate dyskaryotic cervical smears. To assess the strategy of an established cervical cytology screening programme (Grampian region, northeast Scotland) we identified 500 women who had had mild or moderate cervical dyskaryosis in 1978 or 1979, and 500, matched by age, who had had a normal smear at that time. Follow-up smear results and any subsequent

investigation by colposcopy, cone biopsy, or hysterectomy, with biopsy result were recorded. Of the 500 women who initially had an abnormal smear, 300 (60%) had a smear that was normal or inflammatory at their last visit (after seven years’ median follow-up). 184 (37%) had undergone biopsy, 97 (19%) of whom were cervical intraepithelial neoplasia grade III or worse. Survival curves for time to biopsy and ten-year biopsy rates show that women with an abnormal smear before their baseline year were the most likely to have a subsequent biopsy. Older women had a biopsy less often and at biopsy were more likely to have minor

abnormalities. Mild or moderate dyskaryotic smears should not be an indication for immediate referral for colposcopy, since under a conservative management policy most women return to normal without needing treatment. Nevertheless, the increased risk associated with abnormal smears justifies rigorous surveillance. Introduction The management strategy for women with mild or moderate cervical dyskaryosis varies widely. Fox1 suggested that a mildly dyskaryotic lesion can progress rapidly to invasive disease without an intermediate stage. However, if all women with abnormal smears were referred, as Campion et aP have proposed, a huge increase in colposcopy services

would be needed; apart from the substantial cost, such a policy would entail women undergoing treatment that might not be necessary if mild lesions are transient. Moreover, referral for colposcopy can be more traumatic than once thought.3 In Grampian region, northeast Scotland, referral of women for further investigation has generally been confined first to those in whom cytology of cervical smears points to cervical intraepithelial neoplasia (CIN) grade III, and second to women who have persistent cell changes. Of 4899 women referred for biopsy between 1984 and 1990, 3137 (64%) were CIN III, 550 (11%) were CIN II, 354 (7%) were CIN I, 793 (16%) were benign, and 65 (2%) were microinvasive. There has been a substantial decrease in invasive cancer rates in this region from 30 per 100 000 (probably an underestimate) in the 1960s to 13-6 per 100 000 in 1989.4 Without a randomised trial of alternative referral procedures, analysis of the full screening records of a centre with an established screening programme can be used to assess the effects of a particular strategy. By following up a cohort of women who had a mildly or moderately dyskaryotic smear in 1978 or 1979 and who had not previously had a biopsy, we have assessed the management strategy of the Aberdeen programme, which allows better follow-up of patients than has been possible in other studies.5,6Additionally we have used statistical methods for analysis of the full screening experience of each woman.

Subjects and methods Patients From January, 1978, the daybooks at the central pathology laboratory, Grampian region, were used to identify 500 consecutive women (286 from 1978, 214 from 1979) who had an "atypical" cervical smear-eg, mild or moderate dyskaryosis-and who had not been treated previously. We chose an age-matched control who ADDRESSES: MRC Biostatistics Unit, Cambridge (A. J. Kirby, PhD, D.J Spiegelhalter, PhD, Prof N. E Day, PhD), and Department of Pathology, Harris Birthright Research Centre, University of Aberdeen, Aberdeen (L. Fenton, MRCPath, K Swanson, E. M F. Mann, MB, J. E. Macgregor, FRCPath), UK. Correspondence to Prof N E. Day, MRC Biostatistics Unit, 5 Shaftesbury Road, Cambridge CB22BW, UK.

829

Fig 2-Probability of biopsy for the four initial status groups.

Fig 1-Outcome of controls (top) and patients (bottom). *232

(46%

of total

patients)

negative or inflammatory and

so

had a previous smear that was also could be classified as "regressed"

had a normal smear on or around the same date by taking the next previously untreated woman in the daybook with the same year of birth but a negative smear. The entire smear history of each of these women was recorded both retrospectively and prospectively and any biopsies done were also recorded. In Grampian region, all smears are assessed at one laboratory, so any abnormal smear in the region should be found. Of the 1000 women, 792 (79%) received a smear after 1982; therefore, in view of the standard five-year screening interval, they could be regarded as being followed up within the programme. Women who could be traced were invited to attend for an up-to-date smear. Overall, 362 women had a smear in 1988 or later, 198 (55%) of whom were from the patient group. During 1978-88 the policy has been to refer women for a diagnosis on colposcopy or cone biopsy if severe dyskaryosis is observed, whereas mild/ moderate dyskaryosis leads to recall in a month to a year depending on the cell changes observed; further referral only takes place if abnormalities persist. The last known smear report of all the women was used to find the proportion who had indergone a biopsy and, of those, women that had CIN IIIor worse.

Statistical methods The date of the abnormal smear used to identify each woman has been taken as the baseline time zero. We calculated median times from baseline either to biopsy or to last smear. The variable length of follow-up (median eight years) should be taken into account when obtaining an actuarial estimate of time to biopsy, and of time until a finding of CIN III or worse. The standard statistical technique would be to exclude a woman from the study at the time

of her last smear. However, in our study this procedure could lead to serious bias, which would require a more sophisticated approach, because women with abnormalities were followed up much more intensively than were normal women; for example, in the five years after the smear which entered a woman into the study, controls had an average of 1.08 smears whereas patients had 286. Thus, according to standard methods, women with negative smears would tend to be excluded from the study earlier than those with abnormalities. Thus, by removing women with a last negative smear from the cohort, an unduly pessimistic estimate of treatment rate would be obtained. At the other extreme, it would be optimistic to assume that such women remained negative up until the full ten-year intended follow-up. The essential difficulty is that the final known status of the woman is informative (to a certain extent) about her prognosis, and this information should be used to avoid unnecessary bias. The best course seems to be to consider that after a negative smear a woman remains negative for three years, the approximate mean inter-smear interval for control women, by which time if she has not had another smear she is regarded as having defaulted and is removed from the cohort. If she returns to be screened later she is treated as a new individual who enters the cohort at the date of her return. Such analysis is straightforward within the epidemiological statistical package EGRET (Statistics and Epidemiological Research Corporation, Seattle). Time to biopsy (as measured from initial smear in 1977/78) is considered separately for controls (initial status 1) and for patients, the patients being divided into three groups depending on whether the abnormal smear was the first ever smear (initial status 2) or whether there have been previous normal (initial status 3) or abnormal smears (initial status 4). We used a Cox proportional hazards model’ to compare biopsy rates by group and by age category. We also analysed the detection rate of biopsy by group and by age category, and used a logistic regression model for statistical

comparison.

Results The outcome of the patients and controls is summarised in fig 1. The time from first abnormal smear to treatment for the 184 women referred was less than three years (76 women, 41%), three to five years (36, 19%), five to seven years (25, 14%), seven to nine years (27, 15%), and more than nine years (20, 11 %). All 5 women with invasive cancer had been part of the screening programme (when a cervical smear is assessed the patient’s entire smear history is known to the reader). 1 of the patients with clinically invasive cancer, aged 43 years, had 8 smears over fourteen years taken by her general practitioner (GP); she had never had a positive smear and presented clinically with a stage IIB cancer. Clinically, her cervix was "intact" throughout and the smears contained mainly superficial cells. The other patient, aged 37, had a previous negative smear, she complained of infertility, and the smear taken at the infertility clinic was moderately/

830

TABLE II--DETECTION RATE OF CIN III OR WORSE AT BIOPSY BY INITIAL STATUS AND AGE

No of biopsies done m parentheses 1 = control, 2 patient, no previous smears, 4 parent, previous abnormal smear

Fig 3-Probability of CIN

III

or worse

for the four initial status

groups.

severely dyskaryotic; an immediate repeat was requested, and further investigation with cone biopsy was not indicated on the basis of a single abnormal smear. 3 smears during six years were then taken by her GP; all were negative. She returned to the infertility clinic, had a severely dyskaryotic smear, was referred and found to have a stage IB cancer. The taking of smears and examination of the cervix were therefore inadequate in these 2 patients with invasive disease. 1 patient with microinvasive cancer, aged 40 years, had 14 smears over fourteen years with 4 mildly dyskaryotic smears scattered throughout. A biopsy sample was taken immediately after a severely dyskaryotic smear. The second patient aged 31 had 4 previous negative smears before a positive smear led to her referral. The third patient, aged 38, had 2 unsatisfactory smears and underwent biopsy on her third smear, which was abnormal. The slides of these cases have been carefully reviewed. Action might have been taken sooner in 2 of these patients, in which misdiagnosis was due to cell content of the smear rather than cytological

interpretation.8,9 The actuarial curves for time to biopsy in the four groups shown in fig 2. Women who had hysterectomies for reasons other than cervical-cell abnormalities have been regarded as exclusions. The rates are highest for women who had an abnormal smear before 1978/79 (initial status 4) with a rate of 80% (95% confidence interval [CI] 68-90) after ten years. The other patients have similar estimated ten-year biopsy rates of 42% (33-53) for those of initial status 2 and 40% (31-50) for those of inital status 3 whereas for controls it is 4% (2-7). (Under the pessimistic assumption that all women are excluded at their last smear, the value for those of initial status 3 is 46% [instead of 40%] whereas under the very optimistic assumption that all untreated women are still negative after ten years this estimated biopsy rate decreases to 32%.) Similar curves are are

TABLE I-ESTIMATED TEN-YEAR RATES OF CIN III OR WORSE ON BIOPSY BY INITIAL STATUS AND AGE

SE

in parentheses, proportions 1 = control, 2 patient, no previous

4=patient,

previous abnormal

smear

smears,

3 patient, previous normal

smear

3 = patient, previous normal

smear,

shown in fig 3 for the time to CIN IIIor worse on biopsy. As expected these rates are lower but again have a similar pattern. Table I shows the ten-year rate of detecting CIN III or worse on biopsy (as calculated from an actuarial survival curve) by initial status and age ( < 27, 27-35, and > 35 years) (to give roughly similar numbers of women in the three age groups). The actuarial rates are higher than the simple calculation of number treatednumber at risk, which is equivalent to the optimistic assumption that all untreated women lost to follow-up are still negative after ten years. Within an initial status group the rates are very similar for the two youngest age categories but much lower for the older women. The highest rates are in the women who had abnormal smears before 1978/79 (although the numbers are small). A Cox proportional hazards model shows that the effects of initial status and age are significant (p < 0001), although there is no evidence of an interaction between initial status and age. The proportions of those biopsied that are found to be CIN III or worse (the detection rate) are shown in table 11. The rates for all age groups apart from the oldest are similar, being between 52% and 71 %. The range is greater in the oldest age group but the numbers are small; a logistic regression confirms that age is significant (p < 0-01) whereas initial status is not. The fact that the detection policy rate is smaller in older women may reflect a more cautious policy in referring such women for biopsy, or it may point to a higher prevalence of CIN III in the younger age groups.

Discussion There have been two other recent studies to assess in women with mild or moderate cervical dyskaryosis. Robertson et all analysed two-year follow-up data on 1347 women who initially had a mildly dyskaryotic smear between 1965 and 1984. Fletcher et al6 considered 666 women with first-time, borderline, mild, or moderate dyskaryosis on routine screening and followed them for four and a half years. Robertson et al’s biopsy rate of 54% (722) within two years, reflects a more aggressive policy than the actuarial estimate of 14% (95% CI 12-15%) at four and a half years reported by Fletcher et al. Our four-and-a-halfyear biopsy rates for women without a previous abnormal smear at entry to the study are 19% (12-28) for those of initial status 2 and 26% (17-37) for initial status 3. Thus, our cohort has been subject to a similar, though possibly marginally less conservative, policy than that of Fletcher et al. In Robertson et al’s study, 205 biopsy samples were grade CIN III or worse (4 had invasive cancer), giving an estimated two-year detected incidence of 15 %; Fletcher et al give an estimated four-and-a-half-year detected incidence of 11 %. Our four-and-a-half year estimate of 11% (5-21) for initial status 2 and 13% (6-27) for initial status 3 strongly supports the findings of Fletcher et al. Our overall detection outcome

831

REFERENCES

rate of 55% for CIN III lies between two extremes reported

by

these

investigators (Robertson 29%, Fletcher 80%),

which again reflects an intermediate strategy. Our study accords with previous work,5,6 which suggests that a mild or moderate dyskaryotic smear should not be an indication for immediate referral for colposcopy, and that cytological surveillance alone allows many patients to return to normal and avoid unnecessary treatment. Nevertheless, the findings underscore the increased risk associated with repeated abnormal smears, and reaffirm the need for a rigorous surveillance programme. Our study does not support a pattern for the natural history of the disease in which there is steady progression from mild to moderate to severe dyskaryosis. Additionally, failure to detect disease may be due more often to inadequate cell content than to misinterpretation of smears. This research was

supported by the Harris Birthright Research Fund.

1. Fox H. Cervical smears: new terminology new demands. Br Med J 1987; 294: 1307-08. 2. Campion MJ, McCance DJ, Cuzick J, Singer A. Progressive potential of

mild cervical

atypia: prospective cytological, colposcopic, and virological study. Lancet 1986; ii: 237-40. 3. Posner T, Vessey M. Prevention of cervical cancer: the patient’s view. London: King’s Fund, 1988. 4. Grampian Health Board Annual Report. Aberdeen: Grampian Health Board. ISBN 0-95173614. 1990. 5. Robertson JJ, Woodend BE, Crozier EH, Hutchinson J. Risk of cervical cancer associated with mild dyskaryosis. Br Med J 1988; 297: 18-21. 6. Fletcher A, Metaxas N, Grubb C, Chamberlain J. Four and a half year follow up of women with dyskaryotic cervical smears. Br Med J 1990; 301: 641-64. 7. Matthews DE, Farewell V. Using and understanding medical statistics. Basel: Karger, 1985. 8. Woodman CBJ, Williams D, Yates M, Tomlinson K, Ward K, Luesley D. Indicators of effective cytological sampling of the uterine cervix. Lancet 1989; ii: 88-90. 9. Macgregor JE. What is an adequate cervical smear? Br J Obstet Gynaecol 1991; 98: 6-7.

Effect of calcitonin-gene-related peptide in patients with delayed postoperative cerebral ischaemia after aneurysmal subarachnoid haemorrhage EUROPEAN CGRP IN SUBARACHNOID HAEMORRHAGE STUDY GROUP

The finding that the carotid vascular beds are sensitive to the potent vasodilator calcitonin-generelated peptide (CGRP) suggested that the drug might help to prevent ischaemic deterioration after surgery for aneurysmal subarachnoid haemorrhage (SAH). The results of a preliminary study were encouraging, so we have carried out a randomised multicentre single-blind comparison of CGRP and standard best management in patients with ischaemic deficits after surgery for ruptured intracranial aneurysms. Patients aged 18-70 years in whom a focal neurological deficit developed or who had a reduction of 2 or more points on the Glasgow coma scale (GCS) after surgery entered the study after computed tomography had excluded non-ischaemic causes for the neurological deficit. 62 patients were randomly assigned an infusion of 0·6 µg/min CGRP for 4 h, then up to a maximum of 10 days, and 55 patients standard best management (controls). GCS and haemodynamic variables were assessed during the hospital stay, and all patients were followed up at 3 months by an independent investigator, who was unaware of their treatment. Outcome, measured on the Glasgow outcome scale, at 3 months was good in 66% of those treated with CGRP and 60% in the controls; the relative risk of a poor outcome in CGRP-treated patients was 0·88 (95% confidence interval 0·60 to 1 ·28). Hypotension was a common side-effect of the CGRP infusion. 66% of the CGRP group did not complete treatment because of adverse events (19 patients), lack of improvement at 4 h (17 patients) or later (4 patients), or patient’s request (1 patient).

Although we could not show a significant beneficial effect of CGRP in this trial, the wide confidence interval for the risk of a poor outcome and the fact that only a third of patients completed treatment mean that a clinically useful benefit cannot yet be ruled out.

Introduction of death after a haemorrhage from a cerebral include recurrent bleeding from the aneurysm, aneurysm be which can prevented by early surgical intervention, and ischaemic deterioration, of which one cause is cerebral vasospasm. Early surgical intervention can increase the likelihood of delayed ischaemic deterioration. There is, therefore, a difficult choice between early surgery, which prevents rebleeding but has a high ischaemic morbidity, and late surgery, which has a lower ischaemic morbidity but higher overall mortality. More than a hundred pharmaceutical agents had been used in unsuccessful attempts to reverse cerebral vasospasm1 when the calciumchannel antagonist nimodipine was shown to reduce the incidence of cerebral ischaemia after subarachnoid haemorrhage (SAH) and to improve outcome. However, death and severe disability occur in 20% of nimodipinetreated patients,2and the drug impairs cerebrovascular reactivity. The search continues for other agents to prevent ischaemic deterioration after SAH. Analysis of the calcitonin gene pointed to the existence of calcitonin-gene-related peptides (CGRPs) in man.3-5 Two The

causes

ADDRESS: Correspondence to Prof B. A. Bell, Atkinson Morley’s Hospital, Copse Hill, Wimbledon, London SW20 ONE, UK