A PROSPECTIVE, CONTROLLED TRIAL OF SIX FORMS OF HORMONE REPLACEMENT THERAPY GIVEN TO POSTMENOPAUSAL WOMEN BY
T. LIND,E. C. CAMERON, W. M. HUNTER,C.LEON,P.F. MORAN,A. OXLEY,J. GERRARD and U. C. G. LIND
SUMMARY
This study was undertaken to investigate the effect of various forms of hormone replacement therapy (HRT) upon postmenopausal women while controlling as many variables as possible. It was felt that the age, duration of amenorrhoea and the general health of the patients should be as comparable as possible and that each patient should provide her own pretherapy and posttherapy control data. In addition, it was felt that any placebo effect should be investigated and the patients were therefore randomly allocated to placebo tablets or one of six available forms of HRT. The age/sex registers of two large general practices were scrutinized and all women between 49 and 54 years of age were asked to cooperate; for a variety of reasons only 56 women were suitable and willing to take part in the project, yielding 8 women for each of the seven possible therapy groups. Blood samples were taken at 7day intervals three times before therapy was given and the mean of the three values was used as the control value. The women returned on day 21 of each subsequent therapy cycle for six consecutive months and finally three months after discontinuing therapy. From the data the following broad conclusions can be drawn: (i) some women have classic symptoms of hot flushes and sweating des-
pite high endogenous oestrogen concentrations; (ii) vaginal cytology is a relatively poor indicator of endogenous oestrogen status; (iii) while follicle stimulating hormone (FSH) and luteinizing hormone (LH) concentrations are reduced on HRT neither is decreased to anywhere near premenopausal values while prolactin is unaffected; (iv) plasma cholesterol levels are reduced on HRT, the pulse rate is slower and both systolic and diastolic blood pressure are reduced to a small but significant extent; (v) there is no adverse effect upon blood clotting; and (vi) most women experience significant or complete relief of symptoms on all forms of HRT as do some women taking a placebo. The combined preparations containing an oestrogen and progestogen produced vaginal bleeding in only 80 per cent of the women. Thus protection by regular endometrical shedding may not be afforded to all women. As vaginal bleeding is unacceptable to most women if they can achieve the same symptomatic relief without inducing menstruation, it is suggested that women have a low dose oestrogen preparation prescribed cyclically for 6 to 12 months. If therapy is to be maintained for a longer time, uterine curretage should be undertaken at regular intervals to exclude the possibility of endometrial carcinoma developing.
2 INTRODUCTION Possible benefits and drawbacks of hormone replacement therapy (HRT) for postmenopausal women have been widely debated. A comprehensive review of all of the available literature would be difficult but the reports of international symposia held in London during 1975, La Grande Motte in France during 1976 and Sheffield during 1978 demonstrate the spectrum of current opinion (Campbell, 1976; Van Keep et al, 1976; Cooke, 1978). Difficulties can arise in the intecpretation of data from HRT studies: sometimes synthetic oestrogen compounds are compared with ‘natural’ oestrogens (e.g. Jones et al, 1977); women of widely different ages may be compared as in the study of 60 women aged 49 to 91 years reported by Chakravarti et a1 (1976); when a ‘no-therapy’ control group is used, the women may be different to those given therapy rather than the same women before therapy (e.g. Stangel et al, 1977). If symptomatic responses are being investigated, the motivation of women taking part must be carefully considered; women who spontaneously attend an open ‘menopause clinic’ must be regarded as highly selected. The present study was undertaken to investigate the effect of various forms of HRT upon postmenopausal women, with control of as many variables as possible. It was felt that the age, duration of amenorrhoea and general health of the patients should be as comparable as possible and that each patient should provide her own pretherapy and post-therapy control data. In addition the patients should be randomized as to whether they received a placebo tablet or one of the six possible forms of HRT detailed below.
PATIENT SELECTION
Age It seemed unreasonable to compare the responses of women of markedly different ages to HRT; a woman of 70 years of age with perhaps 20 years amenorrhoea and a woman aged 5 5 years and five years amenorrhoea are both ‘postmenopausal’ yet are unlikely to demonstrate the same degree of response to therapy. Our patients were between 49 and 54 years of age.
Duration of amenorrhoea Similar constraints seemed necessary with regard to the duration of amenorrhoea; only women with amenorrhoea for not less than six months and not more than 48 months were accepted.
Reasons for exclusion Women already taking hormone preparations, those suffering from established, chronic illnesses, and those with a known psychiatric history were excluded. Women who had had a hysterectomy were also excluded because of the need to determine the proportion of women in whom menstruation was induced by HRT.
Method ofpatient recruitment The age/sex registers of two local general medical practices were scrutinized and women were selected by their date of birth as falling within the chosen age range; whether they had reached the menopause was not usually known and a questionnaire was sent to each enclosing a prepaid, addressed envelope for reply. A letter was enclosed explaining the purpose of the research, and it was made clear that if a more detailed explanation would be helpful, a female member of staff was willing to visit them at home at a convenient time and date. A car service would be provided to collect them from home and return them on every occasion. In this way, it was hoped to achieve maximum patient cooperation particularly as current publicity in the newspapers, womens’ magazines and on television had all been favourable to the use of HRT. It was requested that the questionnaire be returned even if the person could not help, as a check that everyone had been successfully contacted. The reponse was interesting; in one practice there were 382 women of the appropriate age but only 232 (61 per cent) replied. In the second practice, of 1 16 potentially suitable patients only 61 (53 per cent) replied (Table I). It could be argued that the questionnaire was not sufficiently clear in explaining matters but each person was asked to return the questionnaire in the prepaid envelope yet this was not used to ask for more information or even to indicate that the woman was unwilling or unable to help. Many women of
3 this age have returned to full time working and this could be a deterrent to helping with research; however, of those ultimately recruited, 54 per cent were working full-time, 18 per cent were working part-time and 28 per cent were housewives. Of the 232 women who replied from the first practice, 100 (43 per cent) were still menstruating and therefore unsuitable for our study; of these, 76 would have been willing to help and 24 would not. From the remaining 132 who were postmenopausal, 45 (1 9 per cent) declined to help, 38 (16 per cent) were willing to help but their duration of amenorrhoea fell outside our defined limits, leaving only 49 (21 per cent) suitable volunteers from the whole group. These data and the equivalent figures for the second practice are given in Table I, showing that only 65 suitable and cooperative patients were recruited from the total of 498 women contacted. For domestic reasons nine of these women were unable to take part in the study when the full scheme was explained, leaving a final group of 56.
METHODS
Standardization of conditions It was hoped that seeing the patients in a fasted or ‘basal’ state on each occasion would reduce within-patient and between-patient variability. Patients therefore attended the unit between 0900 and 0930 hours having fasted overnight for at least I2 hours. The majority were collected from their homes by a staff member; others came in their own cars. After the patient had been sitting quietly at rest for 20 to 30 minutes, pulse and blood pressure were recorded twice on a recording sphygmomanometer (Erlag) after which blood was drawn from an antecubital vein for the various determinations. Each woman had a blood sample taken at 7-day intervals three times before being given any therapy and the mean of these three values was used as the control value. They returned on day 21 of each subsequent therapy cycle for six consecutive months and finally three
TABLEI
Criteria used inpatient selection, the numbers in each medicalpractice who were of the appropriate age range, and !he responses of the women who replied Duration of amenorrhoea: 6-48 months Aged: 49-54 years Exclusions: those taking hormones, having chronic illnesses, previous hysterectomy or psychiatric history Practice 1 Approached: Replied: (a) Still menstruating: No Yes
382 232(61%) 24 76
Practice 2 116 61 (53%) 4 17
100 (b) Post-mens; but outside range No Yes
(c) Acceptable Total replies
498 293 28 93
21 12 12
45 38
Both
121 57 50
83
24
I07
49
16
65
232
61
293
‘Yes’ and ‘No’ refer to willingness or otherwise to cooperate in the study.
4 months after discontinuing all therapy (see details below).
Laboratory determinations The factors which were measured at each visit are listed in Tables I1 and 111. The methods used were as follows: Luteinizing hormone (LH) and follicle stimulating hormone (FSH). Highly specific antisera (F 87 and M 93 respectively) were kindly supplied by Professor W. R. Butt of the Birmingham and Midland Hospital for Women. Assay methods were adapted for this study from those described by Hunter and Bennie (1979) to yield working ranges of 2.7 to 280 (LH) and 4.0 to 160 (FSH) mU/1 serum in four hours total incubation time at room temperature followed by 16 hours incubation at 4OC for the double antibody separation. This separation was performed
by an automated system developed by Professor K. D. Bagshawe's laboratory at Charing Cross Hospital, London. The standards used were from the National Institute of Biological Standards and Control: for LH 68/40 assumed 77 U/ampoule and for FSH 69/104 assumed 24 U/ampoule. The within-batch coefficients of variation (CV) were 4.9 per cent for LH and 6.5 per cent for FSH (n=10) and the between-batch CV were 8.1 per cent and 7.8 per cent respectively (n=10). Oestrone and oestradiol. The radioimmunoassay for oestrone used anti-E,-6-(0carboxymethyl) oximino-bovine serum albumin (BAS) serum with a (2,4,6,7-3H,)-E tracer (Specific activity 90 Ci/mmol; Radiochemical Centre, Amersham, England). The within-batch CV was 9.2 per cent and the between-batch CV was 11.3 per cent. Oestradiol was measured by a method modified from Cameron and Jones
TABLEI1
The observations made and samples taken at each visit; the measurements determined on the blood sample are listed in Table III Visits Pretreatment Observations made and samples taken Weight Pulse Blood pressure Blood sample Urine (culture) Vaginal smear
1
2
v v v v v v
v
v
Treatment
Post-treatment
3
4
5
6
7
8
9
10
v v v v v
v v v v v v
v v v v v v
v v v v v v
v v v v *
v v v v *
v v v v * v
v v v v v v
* Urine only cultured if previous samples infected. TABLE111
Laboratory measurementsperformed on each blood sample FHS LH Oestradiol Oestrone Progesterone Progestogen *Prolactin
Cholesterol Triglyceride Urea Creatinine Fasting blood glucose
Full blood count (Coulter)
* Not on
every patient. Progestogen: Norethisterone or norgestrel. Partial thromboplastin time with Kaolin. PTTK : FDP: Fibrinogen degradation products. Platelet Agg: Platelet aggregation studies.
Prothrombin time PTTK FDP *Anti-thrombin 111 Fibrinogen Factors V, VII, VIII and X Platelet count Platelet Agg (ADP and collagen)
5 (1972), using an anti-E,-6-(0-~arboxymethyl) oximino-bovine serum albumin (BSA) serum with a (2,4,6,7-3H,>E, (specific activity, 84 Ci/mmol; Radiochemical Centre, Amersham, England). The within-batch CV was 8.3 per cent and the between-batch CV was 10.9 per cent. Progesterone was determined by the radioimmunoassay method described by Dighe and Hunter (1974) using a solid phase antibody and (1 ,2,6,7-3H) progesterone (specific radioactivity 80-110 Ci/mmol). The between-batch CV was 9 per cent (n= 10). Norethisterone was measured by the method of Morris and Cameron (1975) using an antinorethisterone- 1la-bovine serum albumin serum and a (15,l 6-3H,)-norethisterone tracer (specific activity 20 Ci/mmol; NEN GmbH, Frankfurt am Main, Germany). The within-batch CV was 8 per cent and the between-batch CV was 13.6 per cent. L-norgestrel was assayed by the method of Cameron and Cameron (in preparation) using an anti-norgestrel-1la-BSA serum and a (15163H,)-norgestrel tracer (specific activity 5 7 Ci/mmol; a gift from Scherling AC, Berlin, Germany. The within-batch CV was 6.4 per cent and the between-batch CV was 10 per cent. Prolactin was determined by the immunoassay method of Hwang et al (1971) modified for use with the Kemtech 3000 automated radioimmunoassay system described by Bagshawe (1975). The reagents, originally obtained from Professor Friesen, were made available to us by Dr A. S . McNeilly (Edinburgh). The specific activity of the labelled hormone was 150 pCi/pg, the reference range 25 to 360mU/1 and reference standard MRC 77/222. The within batch CV was 5.5 per cent (n=lO) and the between-batch CV was 13.5 per cent ( n = l 1). Cholesterol and triglyceride. The blood was withdrawn as rapidly as possible by single venepuncture with minimum stasis and was put into a plain glass tube which had been kept in an ice bath. The tube was returned to the ice bath until a firm clot had formed and the resultant serum was assayed the same day for cholesterol and triglyceride by the method of Eggstein and Kreutz (1966) using reagents supplied by Boehringer Corporation (London) Ltd. Urea and creatinine were determined the day the samples were taken. Urea was measured by
Chaney and Marbach’s (1962) modification of Berthelot’s reaction; creatinine was measured by a kinetic technique using an LKB 8600 reaction rate analyser as described by Bierens de Haan ( 1 972). Plasma glucose was measured on the same day by a modification of the glucose oxidaseperoxidase enzyme technique (Trinder, 1969). Antithrombin III. Because of a ‘bewildering variety of techniques’ (Leading article, 1976) which can be used to determine this factor, three assay methods were used for each sample: (i) Lane et a1 (1 974); (ii) Howie et al. (1973); (iii) a single radial immunodiffusion technique using antithrombin 111 plates commercially available from Hoechst Pharmaceuticals. Fibrinogen was determined by the heat precipitation method of Foster et a1 (1959). Prothrornbin time (PT). The one stage prothrombin time was measured by a technique based on the original method of Quick (1957). Partial thromboplastin time with kaolin (PTTK). This was measured by the method described by Biggs (1976) using Folch’s phospholipid as platelet substitute. Fibrinogen degradation products (FDP) were measured using the standard BurroughsWellcome Thrombo-WellcoTest. Factor analysis. In the case of Factors V, VII and X activity, a one-stage assay was used in which a normal control plasma was added to a substrate plasma, deficient in the factor being determined, and compared to the effect of the sample under review on the same deficient plasma using the Quick prothrombin time. Deficient plasma was obtained from American Hospital Supplies. Factor VIII activity was measured using the one-stage method of Breckenridge and Patnoff (1962). The Factor VIII standard was obtained from Immuno-Diagnostica and contained one unit of Factor VIII activitylml. Platelet count. This was measured on a Thrombocounter (Coulter Electronics Ltd.) Platelet aggregation studies. These were performed on an EEL 169 platelet aggregation meter connected to a Servoscribe S potentiometric recorder. Aggregation with adenosine diphosphate (ADP) was by the method published in Broadsheet 83 (1975) of the Association of Clinical Pathologists after the method of Born (1962). The collagen for
6
aggregation studies was obtained from HormonChemie, Miinchen and used at a final concentration of 4p/ml. Vaginal cytology. After passing a bi-valve speculum, two vaginal smears were obtained by drawing a disposable wooden spatula down the full length of each lateral vaginal wall. After fixation, these slides were stained by a Romanowsky staining method. Approximately 500 cells were counted per slide and the basal, precornified and cornified cells expressed as a percentage of the total number counted. Urine culture. All of the mid-stream urine samples were collected after the patients had been fully instructed in the correct technique and after a complete vulva1 wash with soap. Aerobic and anaerobic cultures were undertaken on every sample and these were initiated within three hours of collection.
Clinical observations Visit 1 . At the first visit each patient had a blood sample taken under the basal conditions described. A detailed case history was then taken together with a full physical examination of all major systems including weight, blood pressure, pulse rate, breast palpation, a mid-stream specimen of urine (MSU) for aerobic and anaerobic culture, and cervical and vaginal cytology smears in duplicate. The duration of amenorrhoea was noted together with details of the number of flushing and sweating attacks occurring per day or per week. The patient’s own assessment of her general energy and emotional status relative to the last five menstrual years was also noted. Visit 2 (one week later). A blood sample only was obtained under the same basal conditions and the laboratory determinations detailed above were measured. Visit 3 (a further week later). In addition to the blood sample a further MSU was obtained and the vaginal smears were repeated in duplicate. The data from these three visits defined the pretherapy status of each individual and the women were then randomized into one of seven possible therapy groups (detailed below). They were carefully instructed to take their tablets at breakfast time each day for 21 days and then to have seven days off therapy. Each package contained
sufficient tablets for 21 days and a new supply was only given at the next visit. The appointments were arranged for day 21 of each cycle, namely the day the last pill was taken; the time that the pill was taken at home was noted by the patient on her own watch and the time the blood was obtained was also noted on the same watch. In this way, while the actual time of day might not have been accurate, the interval between pill ingestion and blood sampling was estimated with reasonable accuracy. Visits 4 to 9. These were six consecutive monthly visits. As indicated above, the patients came on day 21 of each therapy cycle and had thus taken the last tablet of that particular course just before leaving home. The blood sample was always obtained under the same basal conditions and the pulse rate and blood pressure were determined in the rested state. Patients were weighed and an MSU obtained during visits 4, 5 and 6; if the urine was consistently sterile no further samples were taken. If infected and therapy prescribed, an MSU was taken at every visit. Further vaginal smears were obtained in duplicate during visits 4,5,6 and 9. Breakfast was provided following these tests and the patients were then questioned about menstruation, hot flushes, any side-effects of therapy and their opinion with regard to their general energy and emotional status. After six months (visit 9) all therapy was stopped and the patient was asked to return three months later. Visit 10. After three months off all therapy, every test and examination described was repeated. The tests and observations made at each visit are summarized in Table 11. THERAPY GROUPS At the time this study began there were four major forms of hormone replacement therapy: conjugated equine oestrogens (Premarin, 1.25 mg, Ayerst); piperazine oestrone sulphate (Harmogen, 1.5 mg, Abbott); oestradiol valerate (Progynova, 2 mg, Schering) and oestradiol valerate plus norgestrel, 0.5 mg, (Cycloprogynova, Schering). Only one firm (Schering) was producing a commercially available combined pill with an oestrone and progesterone effect but, at the suggestion of the
7
firms producing Harmogen (Abbott) and Premarin (Ayerst), norethisterone acetate, 5 mg, was added to their products making a total of seven possible therapy groups: placebo; conjugated equine oestrogens (EO), 1.25 mg; EO, 1.25 mg, plus norethisterone acetate (NA), 5 mg; piperazine oestrone sulphate (E,S), 1.5 mg; E,S, 1.5 mg, plus NA, 5 mg; oestradiol valerate (E,) 2 mg; and E,, 2mg, plus norgestrel (Ng), 0.5 mg. CLINICAL FINDINGS In Table IV, the patients have been divided according to the duration of amenorrhoea, type of therapy prescribed and whether they completed all 10 test visits. Of the nine patients having amenorrhoea for less than 12 completed months, the durations were 6, 7, 9 , 9 , 10, 10, 11, 11 and 11 months respectively. Seven patients withdrew before completing the whole trial, Their data are given below. Patient 1. Amenorrhoea had been present for 12 months; for the first two months on E 0 a vaginal blood loss occurred in the seven days off therapy. These losses were described as being like ‘normal periods’. Factor VIII activity was reported as 50 per cent of normal at the end of the second course of tablets. During the third month, irregular vaginal losses occurred while on therapy and Factor VIII was reported as only 25 per cent of normal. The loss in the seven days off therapy was described as ‘flooding’. The patient remained off therapy and after four weeks, Factor VIII activity had returned to 70 per cent. Due to a misunderstanding the patient then re-started E 0 and ‘flooding’ began to occur within a few days. Repeated determinations of Factor VIII activity were always in excess of 80 per cent of the normal value but all therapy was discontinued. All of the other measured indices of clotting function had remained normal on all occasions. Dilatation and curettage of the uterus under general anaesthesia during the first month off therapy failed to obtain any endometrial curettings. The patient has subsequently remained asymptomatic and well but her hot flushes and sweating, relieved by therapy, have returned. Patient 2. Amenorrhoea had been present for 30 months. After the first month of treatment (EO) her menopausal symptoms were improved
but before starting the next course of tablets the patient rang to say ‘her family wished her to discontinue the pills as she was so irritable and nervous’. Patient 3. Amenorrhoea had been present for 10 months. During the first month on treatment (E,S) her menopausal symptoms of hot flushes decreased but her ‘migraine’ headaches increased. During the third month, abdominal swelling was experienced and this symptom plus the headaches outweighed the benefits from the reduction in hot flushes and sweating. All symptoms subsided off therapy. Patient 4 . Amenorrhoea had been present for 24 months. During the first month on treatment (E,S+ NA) the patient felt unwell with indigestion. flatulance, and a general feeling of tiredness and depression. During the second month she stated that her husband found her irritable and aggressive and wished her to discontinue therapy. At follow-up she claimed to feel much better off therapy. Patient 5. Amenorrhoea had been present for nine months. During the first month of therapy (E,) there was complete freedom from hot flushes. No bleeding occurred during the seven days off therapy but during the first week of the second course of tablets a ‘torrential’ vaginal bleed occurred and lasted for one week. After therapy was stopped the bleeding continued and the patient was admitted to hospital. No curettings could be obtained with formal curettage under general anaesthesia but the bleeding gradually ceased over the following two days. At follow-up three months later, the patient had not had any further bleeding. Patient 6. Amenorrhoea had been present for 60 months. This patient had never had hot flushes or sweating attacks but wished to take part in the trial to see if it would help her tiredness and mild depression. Menses were induced by therapy (E,+Ng) and as she did not experience any symptomatic improvement after two months on HRT, she decided to withdraw from the study. Patient 7 . Amenorrhoea had been present for 12 months. With therapy (E,+Ng) regular vaginal bleeding occurred which the patient was prepared to accept but she also developed renal angle pain and frequency of micturition. She wished to maintain her part in the trial and continued with HRT for five months. Despite the
TABLEIV Patients divided according to their duration of amenorrhoea, the type of therapy prescribed and whether or not thev cornplered all 10 visits Duration of amenorrhoea (months) No. of patients
Treatment Under 12
12-17
18+
E0 2 1
Completed Withdrew E 0 + NA Completed Withdrew
5
-
EIS
Completed Withdrew EIS + NA Completed Withdrew
E*
6
2 8 -
6 -
7
6 I
7
I
1
Completed Withdrew
-
5
7 I
E, + Ng Completed Withdrew
5
6
1
2
29 3
41 1
-
7
8 -
36 3
49 7
All HRT Completed Withdrew
6
2
6 2
Placebo Completed Withdrew
All patients Completed Withdrew
I 2
6 2
EO, Conjugated equine oestrogen; NA, Norethisterone acetate; EIS, Piperazine oestrone sulphate: E,, Oestradiol valerate: Ng, Norgestrel; Placebo, Calcium lactate.
fact that MSUs cultured aerobically and anaerobically failed to grow organisms, she continued to experience dysuria and frequency. Pus cells were never found and cultures for Mycobacterium tuberculosis were initiated; results were negative and an IVP was reported as normal. After therapy was stopped, all symptoms referable to the genito-urinary system disappeared but her hot flushes and sweating returned. The type and duration of therapy taken by these patients are given in Table V. In the results which follow only the data from the 49 patients who completed the trial will be given. However,
with the exception of the one fall in Factor VIII described, the data from the seven omitted patients were similar to those from the other women in the same therapy groups during the equivalent visits; they would not have affected the overall conclusions in any way.
Blood pressure There was a small but statistically significant fall in the mean systolic and diastolic blood pressure in the 41 women taking HRT. After therapy was completed, both the systolic and diastolic pressures returned to the pretherapy
9
values (Table VI). The fall in systolic pressure occurred in 30 of the 41 treated women and the diastolic pressure fell in 3 1 of the 4 1 patients. The patients given a placebo had higher average diastolic pressures than the patients given HRT. The mean pretreatment levels were 79,79,83,89, 100, 101, 126 and 136 mmHg for the eight patients in this group. While some individuals in the HRT group had similar values, they were too few to affect the mean values to the same degree. The patient with a diastolic pressure of 136 mm Hg had a systolic pressure of 235 mm Hg and was treated with methyldopa before being given HRT.
Pulse rate Pulse rate also decreased by a small but significant amount in the patients given HRT; the rate increased again when HRT was discontinued. No change was noted in the patients taking placebo (Table VI). Weight and weight for height There was no convincing evidence of a change in either absolute weight or weight expressed as a percentage of ideal body weight for height. Induced uterine bleeding None of the women given a placebo bled on any occasion and women who did have ‘menses’ while using HRT invariably ceased having bleeds at the end of six months of therapy. The relation between the type of HRT taken and the induction of uterine bleeding was, however, variable. The women taking an oestrogen-only preparation rarely had bleeding induced; those completing the trial and who received E,S and E, did not bleed at
all and only 3 of the 6 patients receiving E0 had bleeding. The pattern of bleeding can be seen from Table VII where the concept of a ‘theoretical maximum number of menses’ has been used. This is the number of patients in any group multiplied by six (the number of therapymonths during which a bleed could have been induced). Thus the six women receiving E,+Ng would yield 36 ‘menstrual periods’ as their theoretical maximum. Only five women actually had menses induced but they yielded 29 bleeds hence each achieved a period virtually every month for the six months on therapy. A combined preparation, whatever the type, caused a higher incidence of induced bleeding; every women on EO+NA bled as did 6 out of 7 on E,S+NA and 5 out of 6 on E,+Ng. The actual number of bleeds caused fell below the theoretical maximum in each group (Table VII).
Vaginal cytology The changes in vaginal cytology were difficult to interpret. While, in general, women on HRT showed a decrease in the percentage of basal cells and an increase in the proportion of cornified cells, this was not invariable. Sometimes the two slides made from the same vagina on the same occasion would yield different cytological scores. The impression gained was that, cytologically, the vaginal mucosa did not respond homogenously over its entire surface. It was equally obvious that any ‘improvement’ in a given woman on HRT was not necessarily progressive. Thus a ‘better’ cytological picture could be found after say three months on HRT than after five or six months. A few women showed no significant ‘improvement’ at all.
TABLEV Details of seven patients whofailed to complete all 10 visits Patient
Therapy
E0 E0
I 2 3 4 5 6
EIS EIS + NA E, E, 4- Ng E, + Ng
I
Visits attended* 1-6 1-4 1-6 1-5 1-4 1-5 1-8
and and and and and
10 10
10 10 10
and 10
Reason for stopping therapy Heavy vaginal bleeding Irritable and nervous Headaches, abdominal swelling Abdominal pain, depression Heavy vaginal bleeding Abdominal pain, indigestion Tiredness, depression
~~
* 10 - indicates the patient was seen three months after the last month on therapy despite not completing the whole course
TABLEVI Blood pressure and pulse rate (mean
+ SD)for the 49 women during the three major phases of the srudv according 10
whelher
they received H R T or a placebo Paired t-test Pretreatment (A)
Treatment (B)
Post-treatment (C)
AvB
BvC
AvC
41 women given HRT Systolic blood pressure (mmHg) Diastolic blood pressure (mmHg) Pulse rate (beats/minute)
1 3 6 2 17 84 f 9 7 9 f 12
129 17 81 f 8 75 f 10
137 f21 8 6 k 11 78 16
*
0.001 0.001 0.001
0.001 0.001 0.05
NS NS NS
8 women given placebo Systolic blood pressure (mmHg) Diastolic blood pressure (mmHg) Pulse rate (beatdminute)
152 f 36* 98 f 22 7 7 * 15
142 f 29 95 & 18 75 7
138k23 9 7 k 16 7 6 5 10
0.05 NS NS
NS NS NS
0.05 NS
*
*
NS
NS = Not significant. * One patient was found to have essential hypertension with a systolic blood pressure of 235 mmHg before HRT and was given methyldopa. TABLEVII Incidence and pattern of uterine bleeding according to the type oftherapy taken. (See text f o r details) Therapy group
No. of patients
No. reporting menses*
Placebo E0 E 0 + NA EIS E,S + NA E* E, f Ng
Theoretical maximum number of mensest
48 339 48 42 42 42 36
Actual number of menses$ 0 15
42 0 28 0 29
* No. of patients who had a bleed on some occasion.
t No. of patients in the whole group times 6 (months on therapy). $ Total number of recorded uterine bleeds per group. 9 On 3 occasions presence or absence of a bleed not reported While there was some correlation between the clinical assessment of the vaginal mucosd status and the cytological picture, there were several exceptions in which an apparently moist and healthy vagina displayed a very atrophic cytological picture. There was no obvious correlation between the plasma levels of oestrone and oestradioi achieved and the cytology.
Urine culture Only four women in the whole series had positive cultures and these were found in the pretreatment samples in each case; the organisms had presumably been present for some time but all of the patients were asymptomatic. No chemotherapy was given for the first three
months on HRT to see if replacement therapy alone might be helpful; this was not the case and antibiotics were prescribed according to the sensitivity of the organisms concerned. Patient A . The MSU contained an average of 250 pus cells/bl and cultured Escherichia coli. An IVP was reported as normal and the patient responded to ampicillin. Patienr B. There were 33 pus cellslpl on average and E. coli was cultured. An IVP was normal and the patient responded to septrin. Patient C. There were 10 pus cells/$ on average: Streptococcus viridans was cultured on one occasion and coagulase negative staphyiococci on others. This patient withdrew from the trial before further investigations could be instigated.
11 attacks began while on HRT. The other patients were asked to grade the number of flushing attacks as none, fewer, the same or more, relative to their frequency before therapy. The same system was used to describe their status at the last visit three months after therapy had been discontinued. There were no obvious differences in response between the various HRT groups so the patients were grouped according to whether they had, or had not, experienced hot flushes before treatment and whether they received HRT or placebo. Of the 30 women receiving HRT and who had experienced flushing attacks, most obtained symptomatic relief. As each woman was questioned each month there were therefore 6 x 30 or 180 ‘women occasions’ on which their flush status could be recorded. On only 12 of these occasions were hot flushes reported as ‘the same’ or ‘worse’ than before taking HRT. However, improvement also occurred on 22 of the 42 ‘women occasions’ reported by the seven women taking a placebo who had previously had hot flushes (Table VIII). This was largely due to three women who reported an improvement on every occasion.
Patienr D. There were 30 pus cells/pl on average; bacteria were cultured in profusion and identified as lactobacilli on each occasion. These were assumed to be contaminants despite a careful collection technique but in view of their persistence and the presence of pus cells, a catheter specimen of urine was obtained with full aseptic technique and again lactobacilli were cultured. After a course of ampicillin the urine became sterile but two months later the organisms were cultured again. An IVP was normal and the patient appears to have a chronic tendency to carry these organisms in her bladder without these affecting her general health or causing symptoms. Hot flushes
The degree to which women experienced discomfort from ‘hot flushes’ was extremely variable. Some had no such sensations at all, others experienced a feeling of flushing on two or three occasions per week without perspiring, while others could suffer six or more flushes per day and perspire profusely. The complaint which most obviously affected the quality of life was that of night perspiration. This could be so severe that some patients had to change their bed linen at least once each night. Because of this variability we have simply recorded whether the patients experienced hot flushes at all and, if so, whether they improved or not with HRT. Those who did not have such sensations at the beginning of the trial could only continue to report ‘none’ or perhaps ‘more’ if
LABORATORY FINDINGS In the following tables the pretreatment data are the mean of the three tests before HRT, the treatment values the mean of the six tests performed while on therapy and the post-treatment values are derived from the single test occasion three months after HRT had been dis-
TABLEVIII Incidence of ‘hotflushes’ on HR T or placebo. ‘YES’and ‘NO’ refer io whether each patient did, or did not, experience ,ylushes’ prior to the trial*
Incidence of hot flushes Three months off therapy
During therapy Therapy group All HRT
Placebo
Before therapy YES (n = 30) NO ( n = 11)
YES(n=7) NO (n = 1)
None
Fewer
Same
More
None
124
45
8
65
-
-
3 1
10
6 6
16 -
16 -
-
4
-
6
1
Fewer -
I
-
Same
More
4
20
-
I
4
2
-
-
* Each patient made a response during each of the six therapy months and once when off therapy for three months (e.g. 30 patients would yield 6 x 30 possible responses on therapy = 180 and 30 x 1 responses off therapy).
12
caused a greater decrease in gonadotrophin levels than the oestrogen-only agents but the numbers in each individual therapy group were too small for statistical testing. However, combining the data from the women taking oestrogens only into one group and comparing them with the data from the women on an oestrogen-progestogen preparation as a singie group revealed that the decrease in serum concentration of both FSH and LH was significantly greater in the women taking the combined preparations (Table XI). The ratio of FSH/LH changed during the time HRT was taken from 1.14 (+ 0.34) to 0.96 (+ 0.28) due to a proportionally greater suppression of FSH (Table XII).
continued. Mean values and standard deviations are given for each therapy group in the tables but the number of patients in the individual groups is too small for reliable analysis. The patients who received HRT are therefore combined as one group for the three occasions ‘pretreatment’, ‘treatment’ and ‘post-treatment’ for statistical analysis. The placebo group is shown separately. FSH and L H The levels of FSH decreased significantly in women receiving HRT, the values returning to the pretreatment concentrations after therapy had been discontinued (Table IX). The same was true for LH (Table X). The decrease in FSH on treatment was shown by 40 of the 4 1 women and in LH concentration by 36 of the 40 women. It appears that the combined preparations
Oestrone Of the 8 patients randomized into the group
TABLEIX
Serum FSH levels (Mean f SD) for each treatment group before, during and after HR T or placebo Therapy group
-
Paired t-test
Serum FSH levels (U/1)
Pretreatment (A)
Treatment
(B)
Post-treatment (C)
66 & 19(6) 66 f 18 (8) 72 f 20 (7) 65 f 15 (7) 78 f 14(7) 69 f 31 (6) 69 19(41) 85 122 (8)
38 1 4 (6) 28 1 14 (8) 58 f 14(7) 39 f 9 (7) 55 15(7) 39 & 17 (6) 43 f 24 (41) 83 124 (8)
77 f 30 (5) 79 f 25 (7) 85 f 20 (7) 69 1 14 (7) 77 f 15 (6) 77 f 27 (6) 77 f 21 (38) 82 f 26 (8)
E0 E 0 + NA E,S EIS + NA E* E, + Ng All HRT Placebo
AvB
BVC
AVC
Number of patients in parentheses. TABLEX
Serum LH levels (Mean f SD) for each treatment group before, during and after HRT or placebo ~~~
~
Paired t-test
Serum LH levels (U/1) Therapy group Pretreatment (A)
Treatment (B)
Post-treatment (C)
70 f 25 (6) 67 f 15 (7) 60 f 11 (7) 58 & 16 (7) 73 f 17(7) 72 f 40 (6) 67 f 21 (40) 56 1 8(8)
55 & 18(6) 34 f 15 (8) 55 f lO(7) 41 f 9(7) 65 f 12(7) 48 f 23 (6) 49 & 17 (41) 52 f 07 (8)
55 f 14 ( 5 ) 77 & 20 (7) 6 0 f 15(7) 61 & lO(7) 70 f 22 (7) 58 f 32 (6) 64 f 20 (39) 55 f 9(8)
E0 E 0 + NA EIS E,S + NA E2
E, + Ng All HRT Placebo Number of patients in parentheses.
AvB
BvC
AvC
13 TABLE XI Decrement in serum levels of FSH and LH (pretreatment value minus treatment value) according fo whether the women received oestrogens only or an oestrogen-progestogenpreparation
LH W/I)
FSH (U/I) Oestrogens only Number Decrement Mean SD
Oestrogen-progestogen
Oestrogens only
Oestrogen-progestogen
20
21
20
20
21.3 16.8
31.5 14.2
9.3 13.7
24.1 17.2
* p < 0.01
* p < 0.05
* Significance of difference between the two therapy groups. TABLE XI1 Ratio of serum FSH to L H levelsf o r each treatment group before, during and after H R T or placebo Paired t-test
Ratio of serum FSH to LH levels
~-
Therapy group Pretreatment E0 E 0 + NA EIS E,S + NA E2 E2 + Ng All HRT Placebo
Treatment (B)
Post-treatment
(A)
1.01 2 0.37 (6) 1 . 1 1 f 0.28 (7) 1.25 f 0.37 (7) 1.20+ 0.39(7) 1.13 k 0.29 (7) 1.09 k 0.44 (6) 1.14 k 0.34 (40) 1.57 k 0.47 (8)
0.79 f 0.28 (6) 0.89 k 0.14 (8) 1.08 f 0.16 (7) 1.12+0.18(7) 0.93 k 0.27 (7) 0.96 k 0.50 (6) 0.96 k 0.28 (41) 1.70 5 0.50 (8)
1.46 & 0.72 ( 5 ) 1.02 0.20(7) 1.47 k 0.44 (7) 1.18&0.42(7) 1.34 5 0.67 (6) 1.42 f 0.44 (6) 1.30 f 0.49 (38) 1.51 k 0.45 (8)
AvB
B 11 C
(C1
-
0.00 I NS
Number of patients in parentheses.
taking placebo, 7 had had amenorrhoea for at least 18 months. Their mean endogenous oestrone values were low with a small standard deviation over the 10 months of the trial and probably represent the true state when spontaneous ovarian function has ceased (Table X111). While the mean pre-treatment values were about the same in the HRT groups there were some individuals who had a wide range of values, indicating that in a few patients endogenous ovarian function was still occurring. One woman allocated to the group to be given oestrone sulphate had pretreatment serum oestrone levels of 491, 767 and 255 pm/l respectively over the three weeks before therapy, presumably indicating spontaneous follicular activity. The effect on the standard deviation can be seen in Table XIII. By the time of the last test, three months after all therapy had been discontinued,
the endogenous levels were again low but now the standard deviations for each group were equivalent to those found in the placebo group. The standard deviations of serum oestrone values were very large in each group while taking HRT. This will be commented upon below. However, a11 41 patients receiving HRT showed a rise in serum oestrone levels relative to their own pretreatment values.
Oestradiol The same comments on oestrone are also true for oestradiol. The patients taking placebo had a low average value but this was not the case for the pretreatment concentrations in several of the HRT groups. In the patient referred to above, endogenous oestradiol values for the three weeks before therapy were 1 138, 1742 and 465 pmol/I
14 TABLEXI11
Serum oestrone levels (Mean? SO)f o r each treatment group before, during and after H R T or placebo Serum oestrone levels (prnol/l)
Therapy group
E0 E0 +NA EIS E,S + NA
E, E,
+ Ng
All HRT Placebo
Pretreatment (A)
Treatment
170f 86(6) 155 2 66(8) 164 1I56 (7) 131 f 47(7) 95 f 18(7) 122 f 12(6) 1 4 0 2 80(41) 1 2 7 2 37(8)
554 78(6) 522 f 222 (8) 755 k 271 (7) 527 f 206 (7) 944 1 597 (7) 533 k 169(6) 641 +332(41) 1 4 6 f 38(8)
(B)
+
Paired t-test
Post-treatment (C)
____ A B B I’ C 12
A 1’ C
112+30(5) 152 f 43 (7) 104 1 4 1 (7) I 1 1 2 32 (7) 135 2 30 (7) 94 2 31 (5) 120 & 39 (38) 144 33 (7)
*
Number of patients in parentheses.
TABLE XIV
Serum oestradiol levels (Mean fS D )for each treatment group before, during and after H R T or placebo Serum oestradiol levels (pmol/l)
Paired t -test
Therapy group
E0 E 0 + NA EIS E,S + NA E, E, + Ng All HRT Placebo
Pretreatment (A)
Treatment
184 f 146 (6) 130 f 124 (8) 224 393 (7) 156 f 184(7) 8 3 2 18(7) 98 f 51 (6) 146 f 191 (41) 88 f 35(8)
332 5 72(6) 290 & 103 (8) 2 4 6 f 48(7) 221 f 76(7) 367 f 151 (7) 255 f 117 (6) 285 f I07 (41) 85 f 30(8)
+
(B)
Post-treatment (C)
A
11
B
B C 11
A
I’
C
* 25844(5) (7)
71 178 & 93 _+ 78 _+ 54+ 69+ 92k 99 _+
70(7) 45(7) 13(7) 18(6) 117(39) 63(8)
Number of patients in parentheses.
respectively. This endogenous activity appears to have ended by the time the last test was performed in all but the group receiving EO+NA (Table XIV). The standard deviations of oestradiol values for all of the groups receiving H R T were quite large and this will be discussed in more detail later; however, 38 of the 41 women showed an increase in serum oestradiol concentration while on therapy relative to her own average pretreatment value. As with FSH and LH concentrations, it appears that the average serum concentrations of both oestrone and oestradiol achieved when a combined oestrogen and progestogen is taken are somewhat lower than when an oestrogen alone is taken, but it must be emphasized that the pre-
therapy serum concentrations are too variable to test the reality of this observation. Ratio of oestradiol to oestrone Irrespective of whether the hormonal preparation contained conjugated equine oestrogens, piperazine oestrone sulphate or oestradiol valerate. the effect was to increase the level of circulating oestrone more than oestrudiol. This can be seen in Table X V where the ratio of E2/E, decreased in each therapy group other than those receiving the placebo; this was always caused by the proportionately greater rise in serum oestrone concentration. Progesterone None of the patients had significantly different
15 TABLE XV Ratio of serum oestradiol to oestrone levelsf o r each treatment group before, during and after HR T or placebo Ratio of serum oestradiol to oestrone levels Therapy group
E0 E 0 + NA EIS E,S + NA E2
E, + Ng All HRT Placebo
Pretreatment (A)
Treatment (B)
Post-treatment
0.97 k 0.29 (6) 0.75 k 0.34 (8) 0.98 f 0.57 (7) 1.06 k 0.68 (7) 0.96 k 0.27 (7) 0.77 k 0.27 (6) 0.91 f 0.43 (41) 0.71 0.18 (8)
0.64 f 0.25 (6) 0.69 k 0.20 (8) 0.37 k 0.12 (7) 0.48 f 0.23 (7) 0.45 k 0.12 (7) 0.54 f 0.23 (6) 0.53 k 0.22 (41) 0.61 k 0.21 (8)
0.66 f 0.38 ( 5 ) 1.17 f 1.60 (7) 1.04 f 0.79 (7) 0.74 k 0.40 (7) 0.41 k 0.12 (7) 0.75 k 0.34 ( 5 ) 0.80 f 0.80(38) 0.69 k 0.39(7)
Paired t-test ___A vB B I* C ArC
(C)
-
-
-
-
-
-
-
-
-
-
-
-
0.001 NS
0.05 NS
NS NS
Number of patients in parentheses. TABLE XVI Serum progesterone levels (Mean k SD) f o r the women receiving HR T or placebo before, during, and afier rherapv togelher with serum norethisterone and norgestref levels (Mean 5 SD)f o r those women receiving these particular preparafions Period of sampling Therapy group All HRT Placebo E 0 + NA EIS + NA E2 + Ng
Serum progestogen Progesterone (pmolil) Norethisterone (nmol/l)
Pretreatment
During treatment
Post-treatment
124 k 99(41) 1 2 6 5 4 4 (8) -
89 k 33 (39) 1 0 4 f 4 4 (8)
-
108 f 71 (41) 109 k 4 1 (8) 61.8 f 21.2 (8) 35.2 f 11.9 (7)
-
13.7 k 5.3 (6)
-
Norgestrel (nmol/l)
-
Number of patients in parentheses.
levels of endogenous progesterone before, during or after therapy. Thus the few patients with high endogenous values for oestrone and oestradiol must have had residual follicular activity but not ovulation. The patient quoted above had, during the three pretreatment weeks, oestradiol values of 1138, 1742 and 465 pmol/l, and her progesterone concentrations were only 100, 100 and 140 pg/ml respectively. The mean values for all the HRT groups and the placebo group are given in Table XVI.
Norethisterone and norgestrel The average serum concentrations achieved in those women receiving these preparations are given in Table XVI. The levels of norethisterone achieved in conjunction with conjugated equine oestrogens were significantly greater than those
achieved with piperazine oestrone (p
T. LIND,E. C. CAMERON, W. M. HUNTER,C.LEON,P.F. MORAN,A. OXLEY,J. GERRARD and U. C. G. LIND
SUMMARY
This study was undertaken to investigate the effect of various forms of hormone replacement therapy (HRT) upon postmenopausal women while controlling as many variables as possible. It was felt that the age, duration of amenorrhoea and the general health of the patients should be as comparable as possible and that each patient should provide her own pretherapy and posttherapy control data. In addition, it was felt that any placebo effect should be investigated and the patients were therefore randomly allocated to placebo tablets or one of six available forms of HRT. The age/sex registers of two large general practices were scrutinized and all women between 49 and 54 years of age were asked to cooperate; for a variety of reasons only 56 women were suitable and willing to take part in the project, yielding 8 women for each of the seven possible therapy groups. Blood samples were taken at 7day intervals three times before therapy was given and the mean of the three values was used as the control value. The women returned on day 21 of each subsequent therapy cycle for six consecutive months and finally three months after discontinuing therapy. From the data the following broad conclusions can be drawn: (i) some women have classic symptoms of hot flushes and sweating des-
pite high endogenous oestrogen concentrations; (ii) vaginal cytology is a relatively poor indicator of endogenous oestrogen status; (iii) while follicle stimulating hormone (FSH) and luteinizing hormone (LH) concentrations are reduced on HRT neither is decreased to anywhere near premenopausal values while prolactin is unaffected; (iv) plasma cholesterol levels are reduced on HRT, the pulse rate is slower and both systolic and diastolic blood pressure are reduced to a small but significant extent; (v) there is no adverse effect upon blood clotting; and (vi) most women experience significant or complete relief of symptoms on all forms of HRT as do some women taking a placebo. The combined preparations containing an oestrogen and progestogen produced vaginal bleeding in only 80 per cent of the women. Thus protection by regular endometrical shedding may not be afforded to all women. As vaginal bleeding is unacceptable to most women if they can achieve the same symptomatic relief without inducing menstruation, it is suggested that women have a low dose oestrogen preparation prescribed cyclically for 6 to 12 months. If therapy is to be maintained for a longer time, uterine curretage should be undertaken at regular intervals to exclude the possibility of endometrial carcinoma developing.
2 INTRODUCTION Possible benefits and drawbacks of hormone replacement therapy (HRT) for postmenopausal women have been widely debated. A comprehensive review of all of the available literature would be difficult but the reports of international symposia held in London during 1975, La Grande Motte in France during 1976 and Sheffield during 1978 demonstrate the spectrum of current opinion (Campbell, 1976; Van Keep et al, 1976; Cooke, 1978). Difficulties can arise in the intecpretation of data from HRT studies: sometimes synthetic oestrogen compounds are compared with ‘natural’ oestrogens (e.g. Jones et al, 1977); women of widely different ages may be compared as in the study of 60 women aged 49 to 91 years reported by Chakravarti et a1 (1976); when a ‘no-therapy’ control group is used, the women may be different to those given therapy rather than the same women before therapy (e.g. Stangel et al, 1977). If symptomatic responses are being investigated, the motivation of women taking part must be carefully considered; women who spontaneously attend an open ‘menopause clinic’ must be regarded as highly selected. The present study was undertaken to investigate the effect of various forms of HRT upon postmenopausal women, with control of as many variables as possible. It was felt that the age, duration of amenorrhoea and general health of the patients should be as comparable as possible and that each patient should provide her own pretherapy and post-therapy control data. In addition the patients should be randomized as to whether they received a placebo tablet or one of the six possible forms of HRT detailed below.
PATIENT SELECTION
Age It seemed unreasonable to compare the responses of women of markedly different ages to HRT; a woman of 70 years of age with perhaps 20 years amenorrhoea and a woman aged 5 5 years and five years amenorrhoea are both ‘postmenopausal’ yet are unlikely to demonstrate the same degree of response to therapy. Our patients were between 49 and 54 years of age.
Duration of amenorrhoea Similar constraints seemed necessary with regard to the duration of amenorrhoea; only women with amenorrhoea for not less than six months and not more than 48 months were accepted.
Reasons for exclusion Women already taking hormone preparations, those suffering from established, chronic illnesses, and those with a known psychiatric history were excluded. Women who had had a hysterectomy were also excluded because of the need to determine the proportion of women in whom menstruation was induced by HRT.
Method ofpatient recruitment The age/sex registers of two local general medical practices were scrutinized and women were selected by their date of birth as falling within the chosen age range; whether they had reached the menopause was not usually known and a questionnaire was sent to each enclosing a prepaid, addressed envelope for reply. A letter was enclosed explaining the purpose of the research, and it was made clear that if a more detailed explanation would be helpful, a female member of staff was willing to visit them at home at a convenient time and date. A car service would be provided to collect them from home and return them on every occasion. In this way, it was hoped to achieve maximum patient cooperation particularly as current publicity in the newspapers, womens’ magazines and on television had all been favourable to the use of HRT. It was requested that the questionnaire be returned even if the person could not help, as a check that everyone had been successfully contacted. The reponse was interesting; in one practice there were 382 women of the appropriate age but only 232 (61 per cent) replied. In the second practice, of 1 16 potentially suitable patients only 61 (53 per cent) replied (Table I). It could be argued that the questionnaire was not sufficiently clear in explaining matters but each person was asked to return the questionnaire in the prepaid envelope yet this was not used to ask for more information or even to indicate that the woman was unwilling or unable to help. Many women of
3 this age have returned to full time working and this could be a deterrent to helping with research; however, of those ultimately recruited, 54 per cent were working full-time, 18 per cent were working part-time and 28 per cent were housewives. Of the 232 women who replied from the first practice, 100 (43 per cent) were still menstruating and therefore unsuitable for our study; of these, 76 would have been willing to help and 24 would not. From the remaining 132 who were postmenopausal, 45 (1 9 per cent) declined to help, 38 (16 per cent) were willing to help but their duration of amenorrhoea fell outside our defined limits, leaving only 49 (21 per cent) suitable volunteers from the whole group. These data and the equivalent figures for the second practice are given in Table I, showing that only 65 suitable and cooperative patients were recruited from the total of 498 women contacted. For domestic reasons nine of these women were unable to take part in the study when the full scheme was explained, leaving a final group of 56.
METHODS
Standardization of conditions It was hoped that seeing the patients in a fasted or ‘basal’ state on each occasion would reduce within-patient and between-patient variability. Patients therefore attended the unit between 0900 and 0930 hours having fasted overnight for at least I2 hours. The majority were collected from their homes by a staff member; others came in their own cars. After the patient had been sitting quietly at rest for 20 to 30 minutes, pulse and blood pressure were recorded twice on a recording sphygmomanometer (Erlag) after which blood was drawn from an antecubital vein for the various determinations. Each woman had a blood sample taken at 7-day intervals three times before being given any therapy and the mean of these three values was used as the control value. They returned on day 21 of each subsequent therapy cycle for six consecutive months and finally three
TABLEI
Criteria used inpatient selection, the numbers in each medicalpractice who were of the appropriate age range, and !he responses of the women who replied Duration of amenorrhoea: 6-48 months Aged: 49-54 years Exclusions: those taking hormones, having chronic illnesses, previous hysterectomy or psychiatric history Practice 1 Approached: Replied: (a) Still menstruating: No Yes
382 232(61%) 24 76
Practice 2 116 61 (53%) 4 17
100 (b) Post-mens; but outside range No Yes
(c) Acceptable Total replies
498 293 28 93
21 12 12
45 38
Both
121 57 50
83
24
I07
49
16
65
232
61
293
‘Yes’ and ‘No’ refer to willingness or otherwise to cooperate in the study.
4 months after discontinuing all therapy (see details below).
Laboratory determinations The factors which were measured at each visit are listed in Tables I1 and 111. The methods used were as follows: Luteinizing hormone (LH) and follicle stimulating hormone (FSH). Highly specific antisera (F 87 and M 93 respectively) were kindly supplied by Professor W. R. Butt of the Birmingham and Midland Hospital for Women. Assay methods were adapted for this study from those described by Hunter and Bennie (1979) to yield working ranges of 2.7 to 280 (LH) and 4.0 to 160 (FSH) mU/1 serum in four hours total incubation time at room temperature followed by 16 hours incubation at 4OC for the double antibody separation. This separation was performed
by an automated system developed by Professor K. D. Bagshawe's laboratory at Charing Cross Hospital, London. The standards used were from the National Institute of Biological Standards and Control: for LH 68/40 assumed 77 U/ampoule and for FSH 69/104 assumed 24 U/ampoule. The within-batch coefficients of variation (CV) were 4.9 per cent for LH and 6.5 per cent for FSH (n=10) and the between-batch CV were 8.1 per cent and 7.8 per cent respectively (n=10). Oestrone and oestradiol. The radioimmunoassay for oestrone used anti-E,-6-(0carboxymethyl) oximino-bovine serum albumin (BAS) serum with a (2,4,6,7-3H,)-E tracer (Specific activity 90 Ci/mmol; Radiochemical Centre, Amersham, England). The within-batch CV was 9.2 per cent and the between-batch CV was 11.3 per cent. Oestradiol was measured by a method modified from Cameron and Jones
TABLEI1
The observations made and samples taken at each visit; the measurements determined on the blood sample are listed in Table III Visits Pretreatment Observations made and samples taken Weight Pulse Blood pressure Blood sample Urine (culture) Vaginal smear
1
2
v v v v v v
v
v
Treatment
Post-treatment
3
4
5
6
7
8
9
10
v v v v v
v v v v v v
v v v v v v
v v v v v v
v v v v *
v v v v *
v v v v * v
v v v v v v
* Urine only cultured if previous samples infected. TABLE111
Laboratory measurementsperformed on each blood sample FHS LH Oestradiol Oestrone Progesterone Progestogen *Prolactin
Cholesterol Triglyceride Urea Creatinine Fasting blood glucose
Full blood count (Coulter)
* Not on
every patient. Progestogen: Norethisterone or norgestrel. Partial thromboplastin time with Kaolin. PTTK : FDP: Fibrinogen degradation products. Platelet Agg: Platelet aggregation studies.
Prothrombin time PTTK FDP *Anti-thrombin 111 Fibrinogen Factors V, VII, VIII and X Platelet count Platelet Agg (ADP and collagen)
5 (1972), using an anti-E,-6-(0-~arboxymethyl) oximino-bovine serum albumin (BSA) serum with a (2,4,6,7-3H,>E, (specific activity, 84 Ci/mmol; Radiochemical Centre, Amersham, England). The within-batch CV was 8.3 per cent and the between-batch CV was 10.9 per cent. Progesterone was determined by the radioimmunoassay method described by Dighe and Hunter (1974) using a solid phase antibody and (1 ,2,6,7-3H) progesterone (specific radioactivity 80-110 Ci/mmol). The between-batch CV was 9 per cent (n= 10). Norethisterone was measured by the method of Morris and Cameron (1975) using an antinorethisterone- 1la-bovine serum albumin serum and a (15,l 6-3H,)-norethisterone tracer (specific activity 20 Ci/mmol; NEN GmbH, Frankfurt am Main, Germany). The within-batch CV was 8 per cent and the between-batch CV was 13.6 per cent. L-norgestrel was assayed by the method of Cameron and Cameron (in preparation) using an anti-norgestrel-1la-BSA serum and a (15163H,)-norgestrel tracer (specific activity 5 7 Ci/mmol; a gift from Scherling AC, Berlin, Germany. The within-batch CV was 6.4 per cent and the between-batch CV was 10 per cent. Prolactin was determined by the immunoassay method of Hwang et al (1971) modified for use with the Kemtech 3000 automated radioimmunoassay system described by Bagshawe (1975). The reagents, originally obtained from Professor Friesen, were made available to us by Dr A. S . McNeilly (Edinburgh). The specific activity of the labelled hormone was 150 pCi/pg, the reference range 25 to 360mU/1 and reference standard MRC 77/222. The within batch CV was 5.5 per cent (n=lO) and the between-batch CV was 13.5 per cent ( n = l 1). Cholesterol and triglyceride. The blood was withdrawn as rapidly as possible by single venepuncture with minimum stasis and was put into a plain glass tube which had been kept in an ice bath. The tube was returned to the ice bath until a firm clot had formed and the resultant serum was assayed the same day for cholesterol and triglyceride by the method of Eggstein and Kreutz (1966) using reagents supplied by Boehringer Corporation (London) Ltd. Urea and creatinine were determined the day the samples were taken. Urea was measured by
Chaney and Marbach’s (1962) modification of Berthelot’s reaction; creatinine was measured by a kinetic technique using an LKB 8600 reaction rate analyser as described by Bierens de Haan ( 1 972). Plasma glucose was measured on the same day by a modification of the glucose oxidaseperoxidase enzyme technique (Trinder, 1969). Antithrombin III. Because of a ‘bewildering variety of techniques’ (Leading article, 1976) which can be used to determine this factor, three assay methods were used for each sample: (i) Lane et a1 (1 974); (ii) Howie et al. (1973); (iii) a single radial immunodiffusion technique using antithrombin 111 plates commercially available from Hoechst Pharmaceuticals. Fibrinogen was determined by the heat precipitation method of Foster et a1 (1959). Prothrornbin time (PT). The one stage prothrombin time was measured by a technique based on the original method of Quick (1957). Partial thromboplastin time with kaolin (PTTK). This was measured by the method described by Biggs (1976) using Folch’s phospholipid as platelet substitute. Fibrinogen degradation products (FDP) were measured using the standard BurroughsWellcome Thrombo-WellcoTest. Factor analysis. In the case of Factors V, VII and X activity, a one-stage assay was used in which a normal control plasma was added to a substrate plasma, deficient in the factor being determined, and compared to the effect of the sample under review on the same deficient plasma using the Quick prothrombin time. Deficient plasma was obtained from American Hospital Supplies. Factor VIII activity was measured using the one-stage method of Breckenridge and Patnoff (1962). The Factor VIII standard was obtained from Immuno-Diagnostica and contained one unit of Factor VIII activitylml. Platelet count. This was measured on a Thrombocounter (Coulter Electronics Ltd.) Platelet aggregation studies. These were performed on an EEL 169 platelet aggregation meter connected to a Servoscribe S potentiometric recorder. Aggregation with adenosine diphosphate (ADP) was by the method published in Broadsheet 83 (1975) of the Association of Clinical Pathologists after the method of Born (1962). The collagen for
6
aggregation studies was obtained from HormonChemie, Miinchen and used at a final concentration of 4p/ml. Vaginal cytology. After passing a bi-valve speculum, two vaginal smears were obtained by drawing a disposable wooden spatula down the full length of each lateral vaginal wall. After fixation, these slides were stained by a Romanowsky staining method. Approximately 500 cells were counted per slide and the basal, precornified and cornified cells expressed as a percentage of the total number counted. Urine culture. All of the mid-stream urine samples were collected after the patients had been fully instructed in the correct technique and after a complete vulva1 wash with soap. Aerobic and anaerobic cultures were undertaken on every sample and these were initiated within three hours of collection.
Clinical observations Visit 1 . At the first visit each patient had a blood sample taken under the basal conditions described. A detailed case history was then taken together with a full physical examination of all major systems including weight, blood pressure, pulse rate, breast palpation, a mid-stream specimen of urine (MSU) for aerobic and anaerobic culture, and cervical and vaginal cytology smears in duplicate. The duration of amenorrhoea was noted together with details of the number of flushing and sweating attacks occurring per day or per week. The patient’s own assessment of her general energy and emotional status relative to the last five menstrual years was also noted. Visit 2 (one week later). A blood sample only was obtained under the same basal conditions and the laboratory determinations detailed above were measured. Visit 3 (a further week later). In addition to the blood sample a further MSU was obtained and the vaginal smears were repeated in duplicate. The data from these three visits defined the pretherapy status of each individual and the women were then randomized into one of seven possible therapy groups (detailed below). They were carefully instructed to take their tablets at breakfast time each day for 21 days and then to have seven days off therapy. Each package contained
sufficient tablets for 21 days and a new supply was only given at the next visit. The appointments were arranged for day 21 of each cycle, namely the day the last pill was taken; the time that the pill was taken at home was noted by the patient on her own watch and the time the blood was obtained was also noted on the same watch. In this way, while the actual time of day might not have been accurate, the interval between pill ingestion and blood sampling was estimated with reasonable accuracy. Visits 4 to 9. These were six consecutive monthly visits. As indicated above, the patients came on day 21 of each therapy cycle and had thus taken the last tablet of that particular course just before leaving home. The blood sample was always obtained under the same basal conditions and the pulse rate and blood pressure were determined in the rested state. Patients were weighed and an MSU obtained during visits 4, 5 and 6; if the urine was consistently sterile no further samples were taken. If infected and therapy prescribed, an MSU was taken at every visit. Further vaginal smears were obtained in duplicate during visits 4,5,6 and 9. Breakfast was provided following these tests and the patients were then questioned about menstruation, hot flushes, any side-effects of therapy and their opinion with regard to their general energy and emotional status. After six months (visit 9) all therapy was stopped and the patient was asked to return three months later. Visit 10. After three months off all therapy, every test and examination described was repeated. The tests and observations made at each visit are summarized in Table 11. THERAPY GROUPS At the time this study began there were four major forms of hormone replacement therapy: conjugated equine oestrogens (Premarin, 1.25 mg, Ayerst); piperazine oestrone sulphate (Harmogen, 1.5 mg, Abbott); oestradiol valerate (Progynova, 2 mg, Schering) and oestradiol valerate plus norgestrel, 0.5 mg, (Cycloprogynova, Schering). Only one firm (Schering) was producing a commercially available combined pill with an oestrone and progesterone effect but, at the suggestion of the
7
firms producing Harmogen (Abbott) and Premarin (Ayerst), norethisterone acetate, 5 mg, was added to their products making a total of seven possible therapy groups: placebo; conjugated equine oestrogens (EO), 1.25 mg; EO, 1.25 mg, plus norethisterone acetate (NA), 5 mg; piperazine oestrone sulphate (E,S), 1.5 mg; E,S, 1.5 mg, plus NA, 5 mg; oestradiol valerate (E,) 2 mg; and E,, 2mg, plus norgestrel (Ng), 0.5 mg. CLINICAL FINDINGS In Table IV, the patients have been divided according to the duration of amenorrhoea, type of therapy prescribed and whether they completed all 10 test visits. Of the nine patients having amenorrhoea for less than 12 completed months, the durations were 6, 7, 9 , 9 , 10, 10, 11, 11 and 11 months respectively. Seven patients withdrew before completing the whole trial, Their data are given below. Patient 1. Amenorrhoea had been present for 12 months; for the first two months on E 0 a vaginal blood loss occurred in the seven days off therapy. These losses were described as being like ‘normal periods’. Factor VIII activity was reported as 50 per cent of normal at the end of the second course of tablets. During the third month, irregular vaginal losses occurred while on therapy and Factor VIII was reported as only 25 per cent of normal. The loss in the seven days off therapy was described as ‘flooding’. The patient remained off therapy and after four weeks, Factor VIII activity had returned to 70 per cent. Due to a misunderstanding the patient then re-started E 0 and ‘flooding’ began to occur within a few days. Repeated determinations of Factor VIII activity were always in excess of 80 per cent of the normal value but all therapy was discontinued. All of the other measured indices of clotting function had remained normal on all occasions. Dilatation and curettage of the uterus under general anaesthesia during the first month off therapy failed to obtain any endometrial curettings. The patient has subsequently remained asymptomatic and well but her hot flushes and sweating, relieved by therapy, have returned. Patient 2. Amenorrhoea had been present for 30 months. After the first month of treatment (EO) her menopausal symptoms were improved
but before starting the next course of tablets the patient rang to say ‘her family wished her to discontinue the pills as she was so irritable and nervous’. Patient 3. Amenorrhoea had been present for 10 months. During the first month on treatment (E,S) her menopausal symptoms of hot flushes decreased but her ‘migraine’ headaches increased. During the third month, abdominal swelling was experienced and this symptom plus the headaches outweighed the benefits from the reduction in hot flushes and sweating. All symptoms subsided off therapy. Patient 4 . Amenorrhoea had been present for 24 months. During the first month on treatment (E,S+ NA) the patient felt unwell with indigestion. flatulance, and a general feeling of tiredness and depression. During the second month she stated that her husband found her irritable and aggressive and wished her to discontinue therapy. At follow-up she claimed to feel much better off therapy. Patient 5. Amenorrhoea had been present for nine months. During the first month of therapy (E,) there was complete freedom from hot flushes. No bleeding occurred during the seven days off therapy but during the first week of the second course of tablets a ‘torrential’ vaginal bleed occurred and lasted for one week. After therapy was stopped the bleeding continued and the patient was admitted to hospital. No curettings could be obtained with formal curettage under general anaesthesia but the bleeding gradually ceased over the following two days. At follow-up three months later, the patient had not had any further bleeding. Patient 6. Amenorrhoea had been present for 60 months. This patient had never had hot flushes or sweating attacks but wished to take part in the trial to see if it would help her tiredness and mild depression. Menses were induced by therapy (E,+Ng) and as she did not experience any symptomatic improvement after two months on HRT, she decided to withdraw from the study. Patient 7 . Amenorrhoea had been present for 12 months. With therapy (E,+Ng) regular vaginal bleeding occurred which the patient was prepared to accept but she also developed renal angle pain and frequency of micturition. She wished to maintain her part in the trial and continued with HRT for five months. Despite the
TABLEIV Patients divided according to their duration of amenorrhoea, the type of therapy prescribed and whether or not thev cornplered all 10 visits Duration of amenorrhoea (months) No. of patients
Treatment Under 12
12-17
18+
E0 2 1
Completed Withdrew E 0 + NA Completed Withdrew
5
-
EIS
Completed Withdrew EIS + NA Completed Withdrew
E*
6
2 8 -
6 -
7
6 I
7
I
1
Completed Withdrew
-
5
7 I
E, + Ng Completed Withdrew
5
6
1
2
29 3
41 1
-
7
8 -
36 3
49 7
All HRT Completed Withdrew
6
2
6 2
Placebo Completed Withdrew
All patients Completed Withdrew
I 2
6 2
EO, Conjugated equine oestrogen; NA, Norethisterone acetate; EIS, Piperazine oestrone sulphate: E,, Oestradiol valerate: Ng, Norgestrel; Placebo, Calcium lactate.
fact that MSUs cultured aerobically and anaerobically failed to grow organisms, she continued to experience dysuria and frequency. Pus cells were never found and cultures for Mycobacterium tuberculosis were initiated; results were negative and an IVP was reported as normal. After therapy was stopped, all symptoms referable to the genito-urinary system disappeared but her hot flushes and sweating returned. The type and duration of therapy taken by these patients are given in Table V. In the results which follow only the data from the 49 patients who completed the trial will be given. However,
with the exception of the one fall in Factor VIII described, the data from the seven omitted patients were similar to those from the other women in the same therapy groups during the equivalent visits; they would not have affected the overall conclusions in any way.
Blood pressure There was a small but statistically significant fall in the mean systolic and diastolic blood pressure in the 41 women taking HRT. After therapy was completed, both the systolic and diastolic pressures returned to the pretherapy
9
values (Table VI). The fall in systolic pressure occurred in 30 of the 41 treated women and the diastolic pressure fell in 3 1 of the 4 1 patients. The patients given a placebo had higher average diastolic pressures than the patients given HRT. The mean pretreatment levels were 79,79,83,89, 100, 101, 126 and 136 mmHg for the eight patients in this group. While some individuals in the HRT group had similar values, they were too few to affect the mean values to the same degree. The patient with a diastolic pressure of 136 mm Hg had a systolic pressure of 235 mm Hg and was treated with methyldopa before being given HRT.
Pulse rate Pulse rate also decreased by a small but significant amount in the patients given HRT; the rate increased again when HRT was discontinued. No change was noted in the patients taking placebo (Table VI). Weight and weight for height There was no convincing evidence of a change in either absolute weight or weight expressed as a percentage of ideal body weight for height. Induced uterine bleeding None of the women given a placebo bled on any occasion and women who did have ‘menses’ while using HRT invariably ceased having bleeds at the end of six months of therapy. The relation between the type of HRT taken and the induction of uterine bleeding was, however, variable. The women taking an oestrogen-only preparation rarely had bleeding induced; those completing the trial and who received E,S and E, did not bleed at
all and only 3 of the 6 patients receiving E0 had bleeding. The pattern of bleeding can be seen from Table VII where the concept of a ‘theoretical maximum number of menses’ has been used. This is the number of patients in any group multiplied by six (the number of therapymonths during which a bleed could have been induced). Thus the six women receiving E,+Ng would yield 36 ‘menstrual periods’ as their theoretical maximum. Only five women actually had menses induced but they yielded 29 bleeds hence each achieved a period virtually every month for the six months on therapy. A combined preparation, whatever the type, caused a higher incidence of induced bleeding; every women on EO+NA bled as did 6 out of 7 on E,S+NA and 5 out of 6 on E,+Ng. The actual number of bleeds caused fell below the theoretical maximum in each group (Table VII).
Vaginal cytology The changes in vaginal cytology were difficult to interpret. While, in general, women on HRT showed a decrease in the percentage of basal cells and an increase in the proportion of cornified cells, this was not invariable. Sometimes the two slides made from the same vagina on the same occasion would yield different cytological scores. The impression gained was that, cytologically, the vaginal mucosa did not respond homogenously over its entire surface. It was equally obvious that any ‘improvement’ in a given woman on HRT was not necessarily progressive. Thus a ‘better’ cytological picture could be found after say three months on HRT than after five or six months. A few women showed no significant ‘improvement’ at all.
TABLEV Details of seven patients whofailed to complete all 10 visits Patient
Therapy
E0 E0
I 2 3 4 5 6
EIS EIS + NA E, E, 4- Ng E, + Ng
I
Visits attended* 1-6 1-4 1-6 1-5 1-4 1-5 1-8
and and and and and
10 10
10 10 10
and 10
Reason for stopping therapy Heavy vaginal bleeding Irritable and nervous Headaches, abdominal swelling Abdominal pain, depression Heavy vaginal bleeding Abdominal pain, indigestion Tiredness, depression
~~
* 10 - indicates the patient was seen three months after the last month on therapy despite not completing the whole course
TABLEVI Blood pressure and pulse rate (mean
+ SD)for the 49 women during the three major phases of the srudv according 10
whelher
they received H R T or a placebo Paired t-test Pretreatment (A)
Treatment (B)
Post-treatment (C)
AvB
BvC
AvC
41 women given HRT Systolic blood pressure (mmHg) Diastolic blood pressure (mmHg) Pulse rate (beats/minute)
1 3 6 2 17 84 f 9 7 9 f 12
129 17 81 f 8 75 f 10
137 f21 8 6 k 11 78 16
*
0.001 0.001 0.001
0.001 0.001 0.05
NS NS NS
8 women given placebo Systolic blood pressure (mmHg) Diastolic blood pressure (mmHg) Pulse rate (beatdminute)
152 f 36* 98 f 22 7 7 * 15
142 f 29 95 & 18 75 7
138k23 9 7 k 16 7 6 5 10
0.05 NS NS
NS NS NS
0.05 NS
*
*
NS
NS = Not significant. * One patient was found to have essential hypertension with a systolic blood pressure of 235 mmHg before HRT and was given methyldopa. TABLEVII Incidence and pattern of uterine bleeding according to the type oftherapy taken. (See text f o r details) Therapy group
No. of patients
No. reporting menses*
Placebo E0 E 0 + NA EIS E,S + NA E* E, f Ng
Theoretical maximum number of mensest
48 339 48 42 42 42 36
Actual number of menses$ 0 15
42 0 28 0 29
* No. of patients who had a bleed on some occasion.
t No. of patients in the whole group times 6 (months on therapy). $ Total number of recorded uterine bleeds per group. 9 On 3 occasions presence or absence of a bleed not reported While there was some correlation between the clinical assessment of the vaginal mucosd status and the cytological picture, there were several exceptions in which an apparently moist and healthy vagina displayed a very atrophic cytological picture. There was no obvious correlation between the plasma levels of oestrone and oestradioi achieved and the cytology.
Urine culture Only four women in the whole series had positive cultures and these were found in the pretreatment samples in each case; the organisms had presumably been present for some time but all of the patients were asymptomatic. No chemotherapy was given for the first three
months on HRT to see if replacement therapy alone might be helpful; this was not the case and antibiotics were prescribed according to the sensitivity of the organisms concerned. Patient A . The MSU contained an average of 250 pus cells/bl and cultured Escherichia coli. An IVP was reported as normal and the patient responded to ampicillin. Patienr B. There were 33 pus cellslpl on average and E. coli was cultured. An IVP was normal and the patient responded to septrin. Patient C. There were 10 pus cells/$ on average: Streptococcus viridans was cultured on one occasion and coagulase negative staphyiococci on others. This patient withdrew from the trial before further investigations could be instigated.
11 attacks began while on HRT. The other patients were asked to grade the number of flushing attacks as none, fewer, the same or more, relative to their frequency before therapy. The same system was used to describe their status at the last visit three months after therapy had been discontinued. There were no obvious differences in response between the various HRT groups so the patients were grouped according to whether they had, or had not, experienced hot flushes before treatment and whether they received HRT or placebo. Of the 30 women receiving HRT and who had experienced flushing attacks, most obtained symptomatic relief. As each woman was questioned each month there were therefore 6 x 30 or 180 ‘women occasions’ on which their flush status could be recorded. On only 12 of these occasions were hot flushes reported as ‘the same’ or ‘worse’ than before taking HRT. However, improvement also occurred on 22 of the 42 ‘women occasions’ reported by the seven women taking a placebo who had previously had hot flushes (Table VIII). This was largely due to three women who reported an improvement on every occasion.
Patienr D. There were 30 pus cells/pl on average; bacteria were cultured in profusion and identified as lactobacilli on each occasion. These were assumed to be contaminants despite a careful collection technique but in view of their persistence and the presence of pus cells, a catheter specimen of urine was obtained with full aseptic technique and again lactobacilli were cultured. After a course of ampicillin the urine became sterile but two months later the organisms were cultured again. An IVP was normal and the patient appears to have a chronic tendency to carry these organisms in her bladder without these affecting her general health or causing symptoms. Hot flushes
The degree to which women experienced discomfort from ‘hot flushes’ was extremely variable. Some had no such sensations at all, others experienced a feeling of flushing on two or three occasions per week without perspiring, while others could suffer six or more flushes per day and perspire profusely. The complaint which most obviously affected the quality of life was that of night perspiration. This could be so severe that some patients had to change their bed linen at least once each night. Because of this variability we have simply recorded whether the patients experienced hot flushes at all and, if so, whether they improved or not with HRT. Those who did not have such sensations at the beginning of the trial could only continue to report ‘none’ or perhaps ‘more’ if
LABORATORY FINDINGS In the following tables the pretreatment data are the mean of the three tests before HRT, the treatment values the mean of the six tests performed while on therapy and the post-treatment values are derived from the single test occasion three months after HRT had been dis-
TABLEVIII Incidence of ‘hotflushes’ on HR T or placebo. ‘YES’and ‘NO’ refer io whether each patient did, or did not, experience ,ylushes’ prior to the trial*
Incidence of hot flushes Three months off therapy
During therapy Therapy group All HRT
Placebo
Before therapy YES (n = 30) NO ( n = 11)
YES(n=7) NO (n = 1)
None
Fewer
Same
More
None
124
45
8
65
-
-
3 1
10
6 6
16 -
16 -
-
4
-
6
1
Fewer -
I
-
Same
More
4
20
-
I
4
2
-
-
* Each patient made a response during each of the six therapy months and once when off therapy for three months (e.g. 30 patients would yield 6 x 30 possible responses on therapy = 180 and 30 x 1 responses off therapy).
12
caused a greater decrease in gonadotrophin levels than the oestrogen-only agents but the numbers in each individual therapy group were too small for statistical testing. However, combining the data from the women taking oestrogens only into one group and comparing them with the data from the women on an oestrogen-progestogen preparation as a singie group revealed that the decrease in serum concentration of both FSH and LH was significantly greater in the women taking the combined preparations (Table XI). The ratio of FSH/LH changed during the time HRT was taken from 1.14 (+ 0.34) to 0.96 (+ 0.28) due to a proportionally greater suppression of FSH (Table XII).
continued. Mean values and standard deviations are given for each therapy group in the tables but the number of patients in the individual groups is too small for reliable analysis. The patients who received HRT are therefore combined as one group for the three occasions ‘pretreatment’, ‘treatment’ and ‘post-treatment’ for statistical analysis. The placebo group is shown separately. FSH and L H The levels of FSH decreased significantly in women receiving HRT, the values returning to the pretreatment concentrations after therapy had been discontinued (Table IX). The same was true for LH (Table X). The decrease in FSH on treatment was shown by 40 of the 4 1 women and in LH concentration by 36 of the 40 women. It appears that the combined preparations
Oestrone Of the 8 patients randomized into the group
TABLEIX
Serum FSH levels (Mean f SD) for each treatment group before, during and after HR T or placebo Therapy group
-
Paired t-test
Serum FSH levels (U/1)
Pretreatment (A)
Treatment
(B)
Post-treatment (C)
66 & 19(6) 66 f 18 (8) 72 f 20 (7) 65 f 15 (7) 78 f 14(7) 69 f 31 (6) 69 19(41) 85 122 (8)
38 1 4 (6) 28 1 14 (8) 58 f 14(7) 39 f 9 (7) 55 15(7) 39 & 17 (6) 43 f 24 (41) 83 124 (8)
77 f 30 (5) 79 f 25 (7) 85 f 20 (7) 69 1 14 (7) 77 f 15 (6) 77 f 27 (6) 77 f 21 (38) 82 f 26 (8)
E0 E 0 + NA E,S EIS + NA E* E, + Ng All HRT Placebo
AvB
BVC
AVC
Number of patients in parentheses. TABLEX
Serum LH levels (Mean f SD) for each treatment group before, during and after HRT or placebo ~~~
~
Paired t-test
Serum LH levels (U/1) Therapy group Pretreatment (A)
Treatment (B)
Post-treatment (C)
70 f 25 (6) 67 f 15 (7) 60 f 11 (7) 58 & 16 (7) 73 f 17(7) 72 f 40 (6) 67 f 21 (40) 56 1 8(8)
55 & 18(6) 34 f 15 (8) 55 f lO(7) 41 f 9(7) 65 f 12(7) 48 f 23 (6) 49 & 17 (41) 52 f 07 (8)
55 f 14 ( 5 ) 77 & 20 (7) 6 0 f 15(7) 61 & lO(7) 70 f 22 (7) 58 f 32 (6) 64 f 20 (39) 55 f 9(8)
E0 E 0 + NA EIS E,S + NA E2
E, + Ng All HRT Placebo Number of patients in parentheses.
AvB
BvC
AvC
13 TABLE XI Decrement in serum levels of FSH and LH (pretreatment value minus treatment value) according fo whether the women received oestrogens only or an oestrogen-progestogenpreparation
LH W/I)
FSH (U/I) Oestrogens only Number Decrement Mean SD
Oestrogen-progestogen
Oestrogens only
Oestrogen-progestogen
20
21
20
20
21.3 16.8
31.5 14.2
9.3 13.7
24.1 17.2
* p < 0.01
* p < 0.05
* Significance of difference between the two therapy groups. TABLE XI1 Ratio of serum FSH to L H levelsf o r each treatment group before, during and after H R T or placebo Paired t-test
Ratio of serum FSH to LH levels
~-
Therapy group Pretreatment E0 E 0 + NA EIS E,S + NA E2 E2 + Ng All HRT Placebo
Treatment (B)
Post-treatment
(A)
1.01 2 0.37 (6) 1 . 1 1 f 0.28 (7) 1.25 f 0.37 (7) 1.20+ 0.39(7) 1.13 k 0.29 (7) 1.09 k 0.44 (6) 1.14 k 0.34 (40) 1.57 k 0.47 (8)
0.79 f 0.28 (6) 0.89 k 0.14 (8) 1.08 f 0.16 (7) 1.12+0.18(7) 0.93 k 0.27 (7) 0.96 k 0.50 (6) 0.96 k 0.28 (41) 1.70 5 0.50 (8)
1.46 & 0.72 ( 5 ) 1.02 0.20(7) 1.47 k 0.44 (7) 1.18&0.42(7) 1.34 5 0.67 (6) 1.42 f 0.44 (6) 1.30 f 0.49 (38) 1.51 k 0.45 (8)
AvB
B 11 C
(C1
-
0.00 I NS
Number of patients in parentheses.
taking placebo, 7 had had amenorrhoea for at least 18 months. Their mean endogenous oestrone values were low with a small standard deviation over the 10 months of the trial and probably represent the true state when spontaneous ovarian function has ceased (Table X111). While the mean pre-treatment values were about the same in the HRT groups there were some individuals who had a wide range of values, indicating that in a few patients endogenous ovarian function was still occurring. One woman allocated to the group to be given oestrone sulphate had pretreatment serum oestrone levels of 491, 767 and 255 pm/l respectively over the three weeks before therapy, presumably indicating spontaneous follicular activity. The effect on the standard deviation can be seen in Table XIII. By the time of the last test, three months after all therapy had been discontinued,
the endogenous levels were again low but now the standard deviations for each group were equivalent to those found in the placebo group. The standard deviations of serum oestrone values were very large in each group while taking HRT. This will be commented upon below. However, a11 41 patients receiving HRT showed a rise in serum oestrone levels relative to their own pretreatment values.
Oestradiol The same comments on oestrone are also true for oestradiol. The patients taking placebo had a low average value but this was not the case for the pretreatment concentrations in several of the HRT groups. In the patient referred to above, endogenous oestradiol values for the three weeks before therapy were 1 138, 1742 and 465 pmol/I
14 TABLEXI11
Serum oestrone levels (Mean? SO)f o r each treatment group before, during and after H R T or placebo Serum oestrone levels (prnol/l)
Therapy group
E0 E0 +NA EIS E,S + NA
E, E,
+ Ng
All HRT Placebo
Pretreatment (A)
Treatment
170f 86(6) 155 2 66(8) 164 1I56 (7) 131 f 47(7) 95 f 18(7) 122 f 12(6) 1 4 0 2 80(41) 1 2 7 2 37(8)
554 78(6) 522 f 222 (8) 755 k 271 (7) 527 f 206 (7) 944 1 597 (7) 533 k 169(6) 641 +332(41) 1 4 6 f 38(8)
(B)
+
Paired t-test
Post-treatment (C)
____ A B B I’ C 12
A 1’ C
112+30(5) 152 f 43 (7) 104 1 4 1 (7) I 1 1 2 32 (7) 135 2 30 (7) 94 2 31 (5) 120 & 39 (38) 144 33 (7)
*
Number of patients in parentheses.
TABLE XIV
Serum oestradiol levels (Mean fS D )for each treatment group before, during and after H R T or placebo Serum oestradiol levels (pmol/l)
Paired t -test
Therapy group
E0 E 0 + NA EIS E,S + NA E, E, + Ng All HRT Placebo
Pretreatment (A)
Treatment
184 f 146 (6) 130 f 124 (8) 224 393 (7) 156 f 184(7) 8 3 2 18(7) 98 f 51 (6) 146 f 191 (41) 88 f 35(8)
332 5 72(6) 290 & 103 (8) 2 4 6 f 48(7) 221 f 76(7) 367 f 151 (7) 255 f 117 (6) 285 f I07 (41) 85 f 30(8)
+
(B)
Post-treatment (C)
A
11
B
B C 11
A
I’
C
* 25844(5) (7)
71 178 & 93 _+ 78 _+ 54+ 69+ 92k 99 _+
70(7) 45(7) 13(7) 18(6) 117(39) 63(8)
Number of patients in parentheses.
respectively. This endogenous activity appears to have ended by the time the last test was performed in all but the group receiving EO+NA (Table XIV). The standard deviations of oestradiol values for all of the groups receiving H R T were quite large and this will be discussed in more detail later; however, 38 of the 41 women showed an increase in serum oestradiol concentration while on therapy relative to her own average pretreatment value. As with FSH and LH concentrations, it appears that the average serum concentrations of both oestrone and oestradiol achieved when a combined oestrogen and progestogen is taken are somewhat lower than when an oestrogen alone is taken, but it must be emphasized that the pre-
therapy serum concentrations are too variable to test the reality of this observation. Ratio of oestradiol to oestrone Irrespective of whether the hormonal preparation contained conjugated equine oestrogens, piperazine oestrone sulphate or oestradiol valerate. the effect was to increase the level of circulating oestrone more than oestrudiol. This can be seen in Table X V where the ratio of E2/E, decreased in each therapy group other than those receiving the placebo; this was always caused by the proportionately greater rise in serum oestrone concentration. Progesterone None of the patients had significantly different
15 TABLE XV Ratio of serum oestradiol to oestrone levelsf o r each treatment group before, during and after HR T or placebo Ratio of serum oestradiol to oestrone levels Therapy group
E0 E 0 + NA EIS E,S + NA E2
E, + Ng All HRT Placebo
Pretreatment (A)
Treatment (B)
Post-treatment
0.97 k 0.29 (6) 0.75 k 0.34 (8) 0.98 f 0.57 (7) 1.06 k 0.68 (7) 0.96 k 0.27 (7) 0.77 k 0.27 (6) 0.91 f 0.43 (41) 0.71 0.18 (8)
0.64 f 0.25 (6) 0.69 k 0.20 (8) 0.37 k 0.12 (7) 0.48 f 0.23 (7) 0.45 k 0.12 (7) 0.54 f 0.23 (6) 0.53 k 0.22 (41) 0.61 k 0.21 (8)
0.66 f 0.38 ( 5 ) 1.17 f 1.60 (7) 1.04 f 0.79 (7) 0.74 k 0.40 (7) 0.41 k 0.12 (7) 0.75 k 0.34 ( 5 ) 0.80 f 0.80(38) 0.69 k 0.39(7)
Paired t-test ___A vB B I* C ArC
(C)
-
-
-
-
-
-
-
-
-
-
-
-
0.001 NS
0.05 NS
NS NS
Number of patients in parentheses. TABLE XVI Serum progesterone levels (Mean k SD) f o r the women receiving HR T or placebo before, during, and afier rherapv togelher with serum norethisterone and norgestref levels (Mean 5 SD)f o r those women receiving these particular preparafions Period of sampling Therapy group All HRT Placebo E 0 + NA EIS + NA E2 + Ng
Serum progestogen Progesterone (pmolil) Norethisterone (nmol/l)
Pretreatment
During treatment
Post-treatment
124 k 99(41) 1 2 6 5 4 4 (8) -
89 k 33 (39) 1 0 4 f 4 4 (8)
-
108 f 71 (41) 109 k 4 1 (8) 61.8 f 21.2 (8) 35.2 f 11.9 (7)
-
13.7 k 5.3 (6)
-
Norgestrel (nmol/l)
-
Number of patients in parentheses.
levels of endogenous progesterone before, during or after therapy. Thus the few patients with high endogenous values for oestrone and oestradiol must have had residual follicular activity but not ovulation. The patient quoted above had, during the three pretreatment weeks, oestradiol values of 1138, 1742 and 465 pmol/l, and her progesterone concentrations were only 100, 100 and 140 pg/ml respectively. The mean values for all the HRT groups and the placebo group are given in Table XVI.
Norethisterone and norgestrel The average serum concentrations achieved in those women receiving these preparations are given in Table XVI. The levels of norethisterone achieved in conjunction with conjugated equine oestrogens were significantly greater than those
achieved with piperazine oestrone (p
