British Journal of Obstetrics and Gynaecology Vol. 82 No. 11
NEW SERIES
NOVEMBER 1975
THE DIAGNOSIS OF EARLY PREGNANCY FAILURE BY SONAR BY
HUGH P. ROBINSON Glasgow University Department of Midwifery Queen Mother’s Hospital, Glasgow G3 8SH Summary In a series of 425 consecutive patients examined by sonar in the first half of pregnancy 176 ultimately aborted. On analysis of the sonar and post-abortion findings it was found that the aborted pregnancies fell into five clearly defined groups ; blighted ova or anembryonic pregnancies, missed abortions, hydatidiform moles and early and late live abortions. The blighted ova and the missed abortions comprised by far the largest and the early live abortions the smallest groups. Strict diagnostic sonar criteria of abnormality, independent of menstrual or clinical histories, were established for the first three of the groups, and an absolute diagnosis could be made at the time of the first examination in all cases of missed abortion and hydatidiform mole and in just over half of the cases of blighted ovum, the remainder requiring a second and occasionally a third examination. In the first half of the study the majority of the patients were allowed to abort spontaneously but with increasing confidence in the techniques patients were offered termination whenever the diagnosis of an abortive pregnancy was made. Anticipation of fetal death in utero or impending abortion of a live fetus proved to be a much more difficult problem, and in only those patients who aborted a live fetus before the tenth week of pregnancy did the sonar examination reveal any significant abnormality. Possible aetiological backgrounds to these groups of abortions are discussed in the light of the sonar findings.
SPONTANEOUS abortion or early pregnancy failure is one of the most frequent complications of pregnancy, but great difficulty is still experienced in reliably anticipating which pregnancies will terminate in abortion and in many cases a definitive diagnosis is only made when the cervix is found to be opening. This situation is particulady distressing to those patients with a history of threatened or recurrent abortion who abort despite resting in hospital for many weeks. There is, therefore, a need for techniques which will allow an early diagnosis to be made in these patients, preferably with such certainty that a more active line of management can be pursued if so desired. In recent years assays of plasma placental
lactogen (Niven et al, 1972), serum alphafetoprotein (Seppala and Ruoslahti, 1972), and urinary oestrogen, pregnanediol and human chorionic gonadotrophin (Brown et al, 1970), have all been shown to be useful guides to the outcome of threatened abortions. But the overlap between their normal and abnormal ranges does not permit their use as sole arbitors of which pregnancies should be terminated. In addition, as the significance of a result may depend on the maturity of the pregnancy, uncertain or mistaken “dates” could lead to a result being falsely interpreted. With respect to plasma placental lactogen (Niven et al, 1972), and urinary oestrogen and pregnanediol assays (Brown et al, 1970), however, levels below the 849
32
850
ROBINSON
lower limit of normal for any gestational age indicate an abnormal pregnancy irrespective of maturity. A very different approach to the diagnosis of early pregnancy failure has been that of diagnostic sonar. Donald et al (1972) reporting on 131 cases of threatened abortion in the first trimester of pregnancy found abnormal ultrasonic appearances in 57 of the 66 patients who subsequently aborted and in 10 of the 75 patients whose pregnancy continued. The false positive and false negative rates were therefore 13.3 per cent and 13.6 per cent respectively. The sonar features which they considered abnormal included loss of definition of the gestation sac, a “small-for-dates’’ gestation sac, failure of growth of the sac over a period of one to two weeks, a low position of the sac, and absence of fetal echoes. Using almost identical criteria Hellman et al(1973) in New York in a series of 140 patients, reported very similar false positive and false negative rates of 10.5 per cent and 13.6 per cent respectively. While these results represent a considerable advance in the recognition of abnormal early pregnancy the significant false positive rate precludes the use of the above criteria as a means of reliably indicating which pregnancies merit termination. Furthermore, all but two of the criteria of abnormality are subjective in nature and depend on the experience of the observer. Recently, however, three techniques of a more objective nature have been described for the sonar assessment of early pregnancy : the detection of fetal heart movements (Robinson, 1972) and the measurement of fetal crown-rump length (Robinson, 1973), both from seven weeks of amenorrhoea, and the estimation of “gestation sac” volume from five weeks of amenorrhoea (Robinson, 1975). The present investigation was designed to apply these techniques to the evaluation of early pregnancy and its complications, to establish criteria for the diagnosis of pregnancies which are already dead or abnormal, and if possible to predict which living pregnancies will ultimately die or abort. In clinical terms, the primary objective was to ensure that the criteria of abnormality were sufficiently stringent to allow a pregnancy diagnosed as being abnormal to be terminated on sonar evidence alone without fear of error.
METHODS AND DEFINITIONS This study was confined to those patients who gave no history of having passed tissue and who, on vaginal examination, were found to have a closed cervix, thereby excluding patients with incomplete or complete abortions. The sonar apparatus used was a standard Nuclear Enterprises Diasonograph NE4102. During the course of the examination the fetus, if present, was measured (Robinson, 1973), the presence or absence of fetal heart movements determined (Robinson, 1972) and the volume of the gestation sac estimated (Robinson, 1975). In the event of an abortion careful note was made as to the presence or absence of a fetus, its size, and the degree of maceration or infarction of the fetus and trophoblast. The aborted material was subsequently examined histologically. The type of abortion was categorized according to the sonar and post-abortion findings as defined below. Blighted ovum or anernbryonic pregnancy. The terms blighted ovum and anembryonic pregnancy were used synonymously, and their diagnosis was restricted to those pregnancies in which a gestation sac could be defined by sonar but which could not be demonstrated to contain a fetus, either by sonar or subsequently on examination of the aborted products of conception. Missed abortion. A diagnosis of missed abortion was made when a fetus could be clearly demonstrated and measured by sonar within the gestation sac but no fetal heart movements could be detected. Pathological confirmation was obtained if a very macerated fetus was identified in the products of conception, although this was not considered to be essential in establishing the diagnosis. Live abortion. An abortion was considered to be “live” if fetal heart movements had been clearly demonstrated by sonar within the few days prior to spontaneous abortion, and or if the fetus at the time of abortion showed no evidence of maceration and had a crown-rump length compatible with the period of amenorrhoea. Pregnancies were designated as early and late live abortions depending on whether abortion occurred before or after the 12th week. Hydatidiform mole. A sonar diagnosis of hydatidiform mole was made in those preg-
SONAR DIAGNOSIS OF PREGNANCY FAILURE
nancies in which the uterus was seen to be filled with echoes of similar size and amplitude, and within which no gestation sac or fetus could be seen. A further sonar feature was the ease with which the intrauterine echoes disappeared with decrease in the transmitter power output as described by MacVicar and Donald (1963). RESULTS A total of 425 consecutive patients with singleton pregnancies was included in this study, of whom 176 subsequently aborted. These latter were classified according to the criteria given TABLE I Classification of early pregnancy failure and rhe number ofpatients in each group Blighted ovum Missed abortion Live abortion
{$ :;
Hydatidiform mole Miscellaneous Total
69 69 6 18 10 4 176 -
(Table I). The principal indications for the first sonar scan in both the normal and abnormal pregnancies are shown in Table 11. Blighted ova or anembryonic pregnancies (69 patients) Of the three sonar techniques used in this
851
investigation to assess an early pregnancy only that of gestation sac volume estimation was relevant to blighted ova, as the principal abnormality in these pregnancies was the absence of a fetus. The gestation sac volumes of 65 of the 69 pregnancies are shown in Figure 1. The remaining four patients with a blighted ovum were beyond 16 weeks of amenorrhoea and each had a sac volume of between 30 and 90 mi. It can be seen from this graph that all but seven of the pregnancies had a gestation sac volume which fell below the second standard deviation of normality at the time of the first examination, of whom three were beyond seven weeks of amenorrhoea and should therefore have contained a recognizable fetus (Robinson, 1972). Thus, if a sac is seen by sonar to be “small-fordates” and or empty, then a diagnosis can be made with a reasonable degree of certainty at the first examination. In this investigation, however, there were several patients whose pregnancies were suspected of being blighted because the gestation sac was found to be very small-for-dates, but who on subsequent examinations were shown to have normal continuing pregnancies, the problem being simply one of mistaken dates. Before a diagnosis of blighted ovum can be accepted it is thus of the utmost importance to eliminate the possibility of the gestation sac being small simply because of mistaken dates, and the fetus being absent simply because it is too small to be seen.
TABLEI1 Principal indication in order of importance for the first sonar scan in 425 consecutive pregnancies
Indications for sonar
Pregnancies
No which continued Past 28 Weeks
Outcome Early pregnancy failures Blighted ovum
Missed abortion
Live abortion
Hydatidiform mole
12 9 -
9 1 -
~
Threatened abortion Recurrent abortion Large for dates Small for dates Irregular cycle or unsure of dates Hyperemesis gravidarum Confirmation of normal pregnancy
182 90 6 10
56 64 5 9
47 10
-
56 7 -
-
1
50 5
43 5
4 -
2 -
82
67
8
3
3
-
Totals
425
249
69
69
24
10
Miscellaneous
852
ROBINSON
Gestation sac volume (ml)
I
BLIGHTED OVA
FIG1 Gestation sac volumes in 65 of the 69 patients with a blighted ovum.
To achieve this objective the clinical and menstrual history should be ignored and the state of the pregnancy should be assessed on sonar findings alone. The first step, therefore, was to define sonar criteria of normality. On analyzing the findings in 249 pregnancies which progressed beyond 28 weeks it was found that the fetus and fetal heart movements had been detected in all those in which the gestation sac had a volume of 2.0 ml or more. It was also found that all pregnancies with a gestation volume of less than 5 ml increased in volume over a period of one week by more than 100 per cent. A generous allowance was always made to obviate the possibility of underestimating the sac volume in a normal pregnancy because of an error in measurement technique. The potential error in estimating gestation is 1 1 0 per cent (Robinson, 1975). A blighted ovum was therefore redefined as a pregnancy which had a
gestation sac volume of 2.5 ml or over at any single examination but in which no fetus could be identified by sonar, or, if the sac was less than 2 - 5 ml in volume, one which failed to increase in size by at least 75 per cent over a period of one week. Applying these stricter criteria to the 69 anembryonic pregnancies in this series it was found that in 37 a definitive diagnosis could be made at the first sonar examination. Of the 32 remaining patients, 17 aborted or were terminated during the following seven days and 15 were re-examined by sonar one week later. At this second examination a definitive diagnosis of blighted ovum was made in 11, the remaining 4 requiring a third examination. On no occasion was a firm diagnosis of a blighted ovum made in a pregnancy which was subsequently shown to be normal. Of the 69 patients, 34 aborted spontaneously and 35 had their pregnancies electively terminated. In no case did a macroscopic or microscopic examination of the aborted products of conception reveal the presence of a fetus. Missed abortions (69 patients). The sonar diagnosis of missed abortion is comparatively simple. If a fetus can be identified within the sac yet no fetal heart movements can be detected on a thorough examination, then the diagnosis can be made with complete confidence irrespective of any menstrual or clinical history. Thus 43 of the 69 cases of missed abortion were electively terminated on the grounds of the sonar findings and in all but nine a grossly macerated fetus was found. It is probable that in these latter nine cases the fetus had been broken up beyond recognition by the instrumentation at the time of the evacuation of the uterus. Of the 26 pregnancies which aborted spontaneously a macerated fetus was found in all but one. There were no patients in whom a diagnosis of a missed abortion was made who were subsequently shown to have a continuing pregnancy. While the fetus was invariably found to be in an advanced stage of maceration it was frequently noted that the trophoblast was macroscopically fresh either in whole or in part, thereby accounting for the common finding of a positive urine pregnancy test despite the fetus being unequivocally dead. That the trophoblast continues to function is further demonstrated in Figure 2
SONAR DIAGNOSIS OF PREGNANCY FAILURE Gestation sac volume (ml)
’7 150-
I
140-
130-
MISSED
120-
ABORTIONS
110100-
90-
8070-
605040-
3020-
100’ Menstrual age (weeks)
FIG2 Gestation sac volumes of the 14 patients with missed abortions who were examined on more than one occasion. In each case there was no doubt that the fetus was dead at the time of the examinations.
which shows the gestation sac volumes of the 14 patients who were examined on more than one occasion. In all but two of these pregnancies the gestation sac initially increased in volume after the death of the fetus, albeit at a slower than normal rate and in several instances spontaneous abortion was preceded by a fall in the sac volume after most of the trophoblast had presumably also died. There were 10 patients in this group in whom fetal heart movements had been clearly demonstrated by sonar at least once in the earlier part of pregnancy but in whom a subsequent sonar examination had shown the fetus to be dead. On reviewing the fetal heart rates in these pregnancies no consistent abnormality was found although there was a tendency for the rates to be faster than normal (Robinson and
853
Shaw-Dunn, 1973). In addition, the heart rate of one fetus was shown to be irregular, having a “pulsus trigeminus” pattern, but as this was the only arrhythmia found in the total series its significance is uncertain. Analysis of the fetal crown-rump lengths and gestation sac volumes showed no significant deviation from normality, and in particular three fetuses who died between successive weekly scans had apparently grown at a normal rate up to the time of death. If this latter finding holds true for missed abortions in general then the crown-rump length of a fetus can be used to estimate the gestational age at which it died (Robinson and Fleming, 1975). In this way the mean time of estimated fetal death was found to be 84 weeks with 54 of the 69 fetuses (78 per cent) dying between six and nine weeks of menstrual age. The maximum age of fetal death in this series was 14 weeks. In the group of 26 patients who aborted spontaneously the mean interval between the estimated time of fetal death and the actual abortion was 32 days, with a range of 10 to 73 days. Very few patients gave a history of bleeding within a period of one week before or after the estimated time of fetal death. Live abortions (24 patients). Of the 24 patients in this category, six aborted on or before the tenth week of menstrual age and 18 aborted at 14 weeks or later. This clear division of live abortions into early and late groups suggests different aetiological backgrounds, and for this reason they have been considered separately. All but three of the patients in the late live abortion group were examined by sonar on at least two occasions in the first trimester of pregnancy, and analysis of the findings revealed no abnormal trends or results in any of the three parameters studied. In the second trimester, the pregnancies were monitored by biparietal cephalometry and confirmation of fetal life using a Doppler apparatus, and again no significantly abnormal trends were noted even in the period immediately prior to abortion. Of the 18 patients 12 had clinical features which were suggestive of an incompetent cervix. In all cases the fetus was fresh at the time of abortion and its size was compatible with the given menstrual age or the maturity as determined by an earlier crownrump length measurement. Five of the six patients in the early live abortion
854 ROBINSON deviation of normality. This low “sac/fetus ratio” was found to be present in less than 5 per cent of those pregnancies which progressed normally. Hydatidiforrri mole (10 patients). An immediate diagnosis was made in all of the 10 patients in this group at the first examination, although in two an alternative diagnosis of a long-standing missed abortion could not be excluded. The sonar findings in these latter patients were atypical in that there were clear areas within the uterus, which, in retrospect, were probably attributable to blood clot, both patients having had considerable vaginal bleeding. In no patient, however, was there any doubt that the pregnancy
group were examined within two days of spontaneous abortion, the interval in the sixth being four days. At the time of abortion a fresh fetus with a crown-rump length very close to that determined by sonar was found in the products of conception in four cases. From the sonar findings in each of the six patients (Fig. 3), it can be seen that the fetal heart rates were all within normal limits with no tendency for the values to be either high or low. The remaining two parameters, however, show a consistently unusual pattern when considered together, in that, while the crown-rump lengths were all on or above the mean, the gestation sac volumes were all on or below the second standard
160 ‘Gettatton sac volumelmi)
I
1
lbeatrfmin) 180
140
170. 160150-
1301 120
140. 130-
110-
120-
100-
110-
I
90-
100-
800
6
7
8
9
10 11 12 Gestation (weeks)
13
14
15
70-
W90 Crown-Rump length lmrnl
5080
4070
3060
2050
1040
0’ 30
2c
ia C
6
7
8
9 10 11 12 Gertation(weeks1
13
14
FIG3 The fetal heart rates, crown-rump lengths and gestation sac volumes of the six patients who aborted “live” fetuses before 10 weeks of menstrual age. The patient represented by the “square” aborted four days after her examination, the remainder aborting within two days.
SONAR DIAGNOSIS OF PREGNANCY FAILURE
was not a continuing one, and all ten were subsequently terminated. Histology confirmed the diagnosis in all patients. Miscellaneous (4 patients). The four patients in this group were either lost to follow-up or aborted at home. In each case a provisional diagnosis of blighted ovum had been made. DISCUSSION This investigation has shown that given a knowledge of menstrual history it is possible to diagnose virtually all cases of blighted ovum, missed abortion and hydatidiform mole by a single sonar examination, but that if such sonar diagnoses are to be made with certainty then the menstrual history must be ignored and the normality or otherwise of the pregnancy assessed on the sonar evidence alone. This exclusion of the menstrual history when making a sonar assessment of an abnormal pregnancy is of great importance in practical terms if the possibility of a false positive diagnosis due to misleading or mistaken menstrual dates is to be avoided. On this basis and using the criteria defined in this study it has been shown that half of the cases of blighted ovum can then be diagnosed with confidence at the first examination, the remainder requiring a second and occasionally a third examination. While the number of firm diagnoses at the first examination is markedly decreased by this constraint, those pregnancies which have been diagnosed as blighted can then be managed, or terminated, without fear that an error may have been made. The diagnosis of a missed abortion however is not constrained by any knowledge or lack of knowledge of the menstrual history, for in these pregnancies a diagnosis is based on the simple premise that if a fetus can be seen yet no heart movements can be detected then the fetus is dead. The recognition of a hqdatidiform mole is similarly independent of the menstrual maturity. The diagnostic problem in these patients is a more academic one in that it can on occasion be difficult to differentiate a hydatidiform mole from a long-standing missed abortion and vice versa, especially when the placenta of the missed abortion has undergone hydropic degeneration. In either case, however, the pregnancy can confidently be labelled as noncontinuing.
855
The primary object of this investigation was to formulate criteria for the sonar diagnosis of abnormal pregnancies such that these diagnoses could be applied prospectively and with complete reliability in the active management of established early pregnancy failures. Before applying these criteria as a basis for intervention, however, clinical confidence in their accuracy had to be established, and in the earlier part of the series most pregnancies were allowed to take their natural course and abort spontaneously or were only terminated following several confirmatory examinations at weekly intervals. With increasing confidence, however, and an appreciation that no false positive results were being obtained it was felt that patients could be offered elective termination of their pregnancies whenever a firm diagnosis had been established. While there were incidental clinical grounds to support the sonar diagnosis in the majority of cases, termination of pregnancy was performed on sonar grounds alone in eight patients who had given no history of bleeding and whose urine pregnancy tests had been persistently positive. Any other worker wishing to provide a similar service for the diagnosis and management of early pregnancy failure must be prepared to establish the techniques in his own hands for an initial “non-intervention” period with good lines of communication with the obstetrician who is managing the patient. It may be argued that these pregnancies will all ultimately abort spontaneously and that early surgical intervention is simply needless interference. However, the great majority of patients with spontaneous abortions will ultimately come to theatre and the early diagnosis by sonar allows what may have been an emergency procedure to be carried out electively with increased safety for the patient. Further considerations are the psychological stress to the patient of awaiting a spontaneous abortion, and the cost of hospitalization if kept under specialist care. The diagnosis of blighted ova, missed abortion and hydatidiform moles can be made with accuracy but a much more difficult problem is the diagnosis or anticipation of fetal death or abortion in a living pregnancy. In the group of ten patients who were scanned before their pregnancies became missed abortions the sonar
856
ROBINSON
findings when the fetus was alive showed no significant departure from normal apart from the one fetus whose heart rate was irregular. Thus it is not yet possible to anticipate fetal death by sonar. These pregnancies, however, are to be distinguished from those in which the fetus is alive up to the time of spontaneous abortion. In the latter group sonar examination consistently showed the gestation sac volumes to be abnormally small in relation to the crownrump length of the fetus, a feature which was very uncommon in pregnancies which progressed normally. While this low “sac-fetus ratio” in the first ten weeks of pregnancy is not diagnostic of impending abortion it may serve as a warning that all may not be well. In contrast in those pregnancies which resulted in the abortion of a live fetus after the fourteenth week, the sonar findings were of no value in predicting outcome. The blighted ovum and missed abortion groups form the major portion of the abnormal pregnancies in this series. If the miscellaneous group and the patients with hydatidiform moles are excluded (most of them were referred from other hospitals) then the incidence of each of the above groups is seen to be 42.6 per cent, the equivalent figure for the early and late live abortions being 11 . I per cent and 3 . 7 per cent respectively. The relative incidence of each of the groups in this series is similar to that found by Fujikura et a1 (1966), who, in a pathological study of 327 aborted pregnancies found 5 . 5 per cent of their specimens to have totally empty intact sacs and a further 23.5 per cent ruptured but complete sacs with no evidence of a fetus or cord. Of their 72 (22 per cent) incomplete specimens it is probable that a significant proportion were anembryonic and if half of this group were added to those above then the incidence of blighted ovum in the two studies would be remarkably close. The remaining 50 per cent all had an accompanying fetus or there was evidence of a cord stump, and of these 31 (9.5 per cent) showed no evidence of maceration and might therefore be considered as “live” abortions. It is of note that there were no fetuses in this “live” group whose crown-rumps length was equivalent to an age of between 10 and 14 weeks, a finding which exactly parallels that in this investigation. All of the remaining fetuses were macerated to a variable degree and the
majority would therefore have been missed abortions. It is clear from these investigations that missed abortions are common and should not be confused with the more uncommon condition of “carneous mole”. This investigation has shown that early live abortions occur infrequently and that their natural history as illustrated by sonar is very different from that of missed abortions and blighted ova. A defect at the embryonal or fetal level such as a chromsomal abnormality (Carr, 1963 and 1965) would be the logical aetiology of blighted ova and missed abortions, while abnormality of the trophoblast may be the cause of early live abortions as the formation of amniotic fluid (largely a function of the trophoblast at this stage) has been shown to be markedly reduced. Conversely, the aetiology of late live abortions can usually be ascribed to cervical incompetence. In terms of the clinical management of a patient with a history of recurrent abortion a knowledge of the types of her previous abortions could be of considerable value. For example, there would be little point in exercising particular care in patients who had had missed abortions or blighted ova, while early live abortions might be due to hormonal deficiency and a recurrent abortion might be prevented by suitable supportive therapy.
ACKNOWLEDGEMENTS I am grateful to Professor Ian Donald for his advice and comments during this investigation and I thank the obstetric consultants of the Queen Mother’s Hospital whose patients were included in this study, and Mr J. Devlin and his department for preparing the illustrations. This work was performed under the provisions of a Medical Research Council Programme Grant in the University of Glasgow, with Professor Ian Donald as the Programme Leader. REFERENCES Brown, J . B., Evans, J. H., Beischer, N. A., Campbell, D. G., and Fortune, D. W. (1970): Journal of Obstetrics and Gynaecology of the British Commonwealth, 77, 690. Carr, D. H. (1963): Lancet, 2, 603. Carr, D. H. (1965): Obstetrics and Gynecology, 26, 308. Donald, I., Morley, P., and Barnett, E. (1972): Journal of Obstetrics and Gynaecology of the British Commonwealth, 79, 304.
SONAR DIAGNOSIS OF PREGNANCY FAILURE
Fujikura, T., Froehlich, L. A., and Driscoll, S. G . (1966): American Journal of Obstetrics and Gynecology, 95, 902. Hellman, L. M., Kobayashi, M., and Cromb, E. (1973): American Journal of Obstetrics and Gynecology, 115, 615. MacVicar, J., and Donald, I. (1963): Journal of Obstetrics and Gynaecology of the British Commonwealrh, 10, 387. Niven, P. A. R., Landon, J., and Chard, T. (1972): British Medical Journal, 3, 799.
857
Robinson, H. P. (1972): British Medical Journal, 4, 466. Robinson, H. P. (1973): British Medical Journal, 4, 28. Robinson, H. P., and Shaw-Dunn, J. (1973): Journal of Obstetrics and Gynaecology of the British Commonwealth, 80, 805. Robinson, H. P. (1975): British Journal of Obstetrics and Gynaecology, 82, 100. Robinson, H. P., and Fleming, J. E. E. (1975): British Journal of Obstetrics and Gynaecology, 82,102. Seppala, M., and Ruoslahti, E. (1972): British Medical Jortral, 4, 769.
NEW SERIES
NOVEMBER 1975
THE DIAGNOSIS OF EARLY PREGNANCY FAILURE BY SONAR BY
HUGH P. ROBINSON Glasgow University Department of Midwifery Queen Mother’s Hospital, Glasgow G3 8SH Summary In a series of 425 consecutive patients examined by sonar in the first half of pregnancy 176 ultimately aborted. On analysis of the sonar and post-abortion findings it was found that the aborted pregnancies fell into five clearly defined groups ; blighted ova or anembryonic pregnancies, missed abortions, hydatidiform moles and early and late live abortions. The blighted ova and the missed abortions comprised by far the largest and the early live abortions the smallest groups. Strict diagnostic sonar criteria of abnormality, independent of menstrual or clinical histories, were established for the first three of the groups, and an absolute diagnosis could be made at the time of the first examination in all cases of missed abortion and hydatidiform mole and in just over half of the cases of blighted ovum, the remainder requiring a second and occasionally a third examination. In the first half of the study the majority of the patients were allowed to abort spontaneously but with increasing confidence in the techniques patients were offered termination whenever the diagnosis of an abortive pregnancy was made. Anticipation of fetal death in utero or impending abortion of a live fetus proved to be a much more difficult problem, and in only those patients who aborted a live fetus before the tenth week of pregnancy did the sonar examination reveal any significant abnormality. Possible aetiological backgrounds to these groups of abortions are discussed in the light of the sonar findings.
SPONTANEOUS abortion or early pregnancy failure is one of the most frequent complications of pregnancy, but great difficulty is still experienced in reliably anticipating which pregnancies will terminate in abortion and in many cases a definitive diagnosis is only made when the cervix is found to be opening. This situation is particulady distressing to those patients with a history of threatened or recurrent abortion who abort despite resting in hospital for many weeks. There is, therefore, a need for techniques which will allow an early diagnosis to be made in these patients, preferably with such certainty that a more active line of management can be pursued if so desired. In recent years assays of plasma placental
lactogen (Niven et al, 1972), serum alphafetoprotein (Seppala and Ruoslahti, 1972), and urinary oestrogen, pregnanediol and human chorionic gonadotrophin (Brown et al, 1970), have all been shown to be useful guides to the outcome of threatened abortions. But the overlap between their normal and abnormal ranges does not permit their use as sole arbitors of which pregnancies should be terminated. In addition, as the significance of a result may depend on the maturity of the pregnancy, uncertain or mistaken “dates” could lead to a result being falsely interpreted. With respect to plasma placental lactogen (Niven et al, 1972), and urinary oestrogen and pregnanediol assays (Brown et al, 1970), however, levels below the 849
32
850
ROBINSON
lower limit of normal for any gestational age indicate an abnormal pregnancy irrespective of maturity. A very different approach to the diagnosis of early pregnancy failure has been that of diagnostic sonar. Donald et al (1972) reporting on 131 cases of threatened abortion in the first trimester of pregnancy found abnormal ultrasonic appearances in 57 of the 66 patients who subsequently aborted and in 10 of the 75 patients whose pregnancy continued. The false positive and false negative rates were therefore 13.3 per cent and 13.6 per cent respectively. The sonar features which they considered abnormal included loss of definition of the gestation sac, a “small-for-dates’’ gestation sac, failure of growth of the sac over a period of one to two weeks, a low position of the sac, and absence of fetal echoes. Using almost identical criteria Hellman et al(1973) in New York in a series of 140 patients, reported very similar false positive and false negative rates of 10.5 per cent and 13.6 per cent respectively. While these results represent a considerable advance in the recognition of abnormal early pregnancy the significant false positive rate precludes the use of the above criteria as a means of reliably indicating which pregnancies merit termination. Furthermore, all but two of the criteria of abnormality are subjective in nature and depend on the experience of the observer. Recently, however, three techniques of a more objective nature have been described for the sonar assessment of early pregnancy : the detection of fetal heart movements (Robinson, 1972) and the measurement of fetal crown-rump length (Robinson, 1973), both from seven weeks of amenorrhoea, and the estimation of “gestation sac” volume from five weeks of amenorrhoea (Robinson, 1975). The present investigation was designed to apply these techniques to the evaluation of early pregnancy and its complications, to establish criteria for the diagnosis of pregnancies which are already dead or abnormal, and if possible to predict which living pregnancies will ultimately die or abort. In clinical terms, the primary objective was to ensure that the criteria of abnormality were sufficiently stringent to allow a pregnancy diagnosed as being abnormal to be terminated on sonar evidence alone without fear of error.
METHODS AND DEFINITIONS This study was confined to those patients who gave no history of having passed tissue and who, on vaginal examination, were found to have a closed cervix, thereby excluding patients with incomplete or complete abortions. The sonar apparatus used was a standard Nuclear Enterprises Diasonograph NE4102. During the course of the examination the fetus, if present, was measured (Robinson, 1973), the presence or absence of fetal heart movements determined (Robinson, 1972) and the volume of the gestation sac estimated (Robinson, 1975). In the event of an abortion careful note was made as to the presence or absence of a fetus, its size, and the degree of maceration or infarction of the fetus and trophoblast. The aborted material was subsequently examined histologically. The type of abortion was categorized according to the sonar and post-abortion findings as defined below. Blighted ovum or anernbryonic pregnancy. The terms blighted ovum and anembryonic pregnancy were used synonymously, and their diagnosis was restricted to those pregnancies in which a gestation sac could be defined by sonar but which could not be demonstrated to contain a fetus, either by sonar or subsequently on examination of the aborted products of conception. Missed abortion. A diagnosis of missed abortion was made when a fetus could be clearly demonstrated and measured by sonar within the gestation sac but no fetal heart movements could be detected. Pathological confirmation was obtained if a very macerated fetus was identified in the products of conception, although this was not considered to be essential in establishing the diagnosis. Live abortion. An abortion was considered to be “live” if fetal heart movements had been clearly demonstrated by sonar within the few days prior to spontaneous abortion, and or if the fetus at the time of abortion showed no evidence of maceration and had a crown-rump length compatible with the period of amenorrhoea. Pregnancies were designated as early and late live abortions depending on whether abortion occurred before or after the 12th week. Hydatidiform mole. A sonar diagnosis of hydatidiform mole was made in those preg-
SONAR DIAGNOSIS OF PREGNANCY FAILURE
nancies in which the uterus was seen to be filled with echoes of similar size and amplitude, and within which no gestation sac or fetus could be seen. A further sonar feature was the ease with which the intrauterine echoes disappeared with decrease in the transmitter power output as described by MacVicar and Donald (1963). RESULTS A total of 425 consecutive patients with singleton pregnancies was included in this study, of whom 176 subsequently aborted. These latter were classified according to the criteria given TABLE I Classification of early pregnancy failure and rhe number ofpatients in each group Blighted ovum Missed abortion Live abortion
{$ :;
Hydatidiform mole Miscellaneous Total
69 69 6 18 10 4 176 -
(Table I). The principal indications for the first sonar scan in both the normal and abnormal pregnancies are shown in Table 11. Blighted ova or anembryonic pregnancies (69 patients) Of the three sonar techniques used in this
851
investigation to assess an early pregnancy only that of gestation sac volume estimation was relevant to blighted ova, as the principal abnormality in these pregnancies was the absence of a fetus. The gestation sac volumes of 65 of the 69 pregnancies are shown in Figure 1. The remaining four patients with a blighted ovum were beyond 16 weeks of amenorrhoea and each had a sac volume of between 30 and 90 mi. It can be seen from this graph that all but seven of the pregnancies had a gestation sac volume which fell below the second standard deviation of normality at the time of the first examination, of whom three were beyond seven weeks of amenorrhoea and should therefore have contained a recognizable fetus (Robinson, 1972). Thus, if a sac is seen by sonar to be “small-fordates” and or empty, then a diagnosis can be made with a reasonable degree of certainty at the first examination. In this investigation, however, there were several patients whose pregnancies were suspected of being blighted because the gestation sac was found to be very small-for-dates, but who on subsequent examinations were shown to have normal continuing pregnancies, the problem being simply one of mistaken dates. Before a diagnosis of blighted ovum can be accepted it is thus of the utmost importance to eliminate the possibility of the gestation sac being small simply because of mistaken dates, and the fetus being absent simply because it is too small to be seen.
TABLEI1 Principal indication in order of importance for the first sonar scan in 425 consecutive pregnancies
Indications for sonar
Pregnancies
No which continued Past 28 Weeks
Outcome Early pregnancy failures Blighted ovum
Missed abortion
Live abortion
Hydatidiform mole
12 9 -
9 1 -
~
Threatened abortion Recurrent abortion Large for dates Small for dates Irregular cycle or unsure of dates Hyperemesis gravidarum Confirmation of normal pregnancy
182 90 6 10
56 64 5 9
47 10
-
56 7 -
-
1
50 5
43 5
4 -
2 -
82
67
8
3
3
-
Totals
425
249
69
69
24
10
Miscellaneous
852
ROBINSON
Gestation sac volume (ml)
I
BLIGHTED OVA
FIG1 Gestation sac volumes in 65 of the 69 patients with a blighted ovum.
To achieve this objective the clinical and menstrual history should be ignored and the state of the pregnancy should be assessed on sonar findings alone. The first step, therefore, was to define sonar criteria of normality. On analyzing the findings in 249 pregnancies which progressed beyond 28 weeks it was found that the fetus and fetal heart movements had been detected in all those in which the gestation sac had a volume of 2.0 ml or more. It was also found that all pregnancies with a gestation volume of less than 5 ml increased in volume over a period of one week by more than 100 per cent. A generous allowance was always made to obviate the possibility of underestimating the sac volume in a normal pregnancy because of an error in measurement technique. The potential error in estimating gestation is 1 1 0 per cent (Robinson, 1975). A blighted ovum was therefore redefined as a pregnancy which had a
gestation sac volume of 2.5 ml or over at any single examination but in which no fetus could be identified by sonar, or, if the sac was less than 2 - 5 ml in volume, one which failed to increase in size by at least 75 per cent over a period of one week. Applying these stricter criteria to the 69 anembryonic pregnancies in this series it was found that in 37 a definitive diagnosis could be made at the first sonar examination. Of the 32 remaining patients, 17 aborted or were terminated during the following seven days and 15 were re-examined by sonar one week later. At this second examination a definitive diagnosis of blighted ovum was made in 11, the remaining 4 requiring a third examination. On no occasion was a firm diagnosis of a blighted ovum made in a pregnancy which was subsequently shown to be normal. Of the 69 patients, 34 aborted spontaneously and 35 had their pregnancies electively terminated. In no case did a macroscopic or microscopic examination of the aborted products of conception reveal the presence of a fetus. Missed abortions (69 patients). The sonar diagnosis of missed abortion is comparatively simple. If a fetus can be identified within the sac yet no fetal heart movements can be detected on a thorough examination, then the diagnosis can be made with complete confidence irrespective of any menstrual or clinical history. Thus 43 of the 69 cases of missed abortion were electively terminated on the grounds of the sonar findings and in all but nine a grossly macerated fetus was found. It is probable that in these latter nine cases the fetus had been broken up beyond recognition by the instrumentation at the time of the evacuation of the uterus. Of the 26 pregnancies which aborted spontaneously a macerated fetus was found in all but one. There were no patients in whom a diagnosis of a missed abortion was made who were subsequently shown to have a continuing pregnancy. While the fetus was invariably found to be in an advanced stage of maceration it was frequently noted that the trophoblast was macroscopically fresh either in whole or in part, thereby accounting for the common finding of a positive urine pregnancy test despite the fetus being unequivocally dead. That the trophoblast continues to function is further demonstrated in Figure 2
SONAR DIAGNOSIS OF PREGNANCY FAILURE Gestation sac volume (ml)
’7 150-
I
140-
130-
MISSED
120-
ABORTIONS
110100-
90-
8070-
605040-
3020-
100’ Menstrual age (weeks)
FIG2 Gestation sac volumes of the 14 patients with missed abortions who were examined on more than one occasion. In each case there was no doubt that the fetus was dead at the time of the examinations.
which shows the gestation sac volumes of the 14 patients who were examined on more than one occasion. In all but two of these pregnancies the gestation sac initially increased in volume after the death of the fetus, albeit at a slower than normal rate and in several instances spontaneous abortion was preceded by a fall in the sac volume after most of the trophoblast had presumably also died. There were 10 patients in this group in whom fetal heart movements had been clearly demonstrated by sonar at least once in the earlier part of pregnancy but in whom a subsequent sonar examination had shown the fetus to be dead. On reviewing the fetal heart rates in these pregnancies no consistent abnormality was found although there was a tendency for the rates to be faster than normal (Robinson and
853
Shaw-Dunn, 1973). In addition, the heart rate of one fetus was shown to be irregular, having a “pulsus trigeminus” pattern, but as this was the only arrhythmia found in the total series its significance is uncertain. Analysis of the fetal crown-rump lengths and gestation sac volumes showed no significant deviation from normality, and in particular three fetuses who died between successive weekly scans had apparently grown at a normal rate up to the time of death. If this latter finding holds true for missed abortions in general then the crown-rump length of a fetus can be used to estimate the gestational age at which it died (Robinson and Fleming, 1975). In this way the mean time of estimated fetal death was found to be 84 weeks with 54 of the 69 fetuses (78 per cent) dying between six and nine weeks of menstrual age. The maximum age of fetal death in this series was 14 weeks. In the group of 26 patients who aborted spontaneously the mean interval between the estimated time of fetal death and the actual abortion was 32 days, with a range of 10 to 73 days. Very few patients gave a history of bleeding within a period of one week before or after the estimated time of fetal death. Live abortions (24 patients). Of the 24 patients in this category, six aborted on or before the tenth week of menstrual age and 18 aborted at 14 weeks or later. This clear division of live abortions into early and late groups suggests different aetiological backgrounds, and for this reason they have been considered separately. All but three of the patients in the late live abortion group were examined by sonar on at least two occasions in the first trimester of pregnancy, and analysis of the findings revealed no abnormal trends or results in any of the three parameters studied. In the second trimester, the pregnancies were monitored by biparietal cephalometry and confirmation of fetal life using a Doppler apparatus, and again no significantly abnormal trends were noted even in the period immediately prior to abortion. Of the 18 patients 12 had clinical features which were suggestive of an incompetent cervix. In all cases the fetus was fresh at the time of abortion and its size was compatible with the given menstrual age or the maturity as determined by an earlier crownrump length measurement. Five of the six patients in the early live abortion
854 ROBINSON deviation of normality. This low “sac/fetus ratio” was found to be present in less than 5 per cent of those pregnancies which progressed normally. Hydatidiforrri mole (10 patients). An immediate diagnosis was made in all of the 10 patients in this group at the first examination, although in two an alternative diagnosis of a long-standing missed abortion could not be excluded. The sonar findings in these latter patients were atypical in that there were clear areas within the uterus, which, in retrospect, were probably attributable to blood clot, both patients having had considerable vaginal bleeding. In no patient, however, was there any doubt that the pregnancy
group were examined within two days of spontaneous abortion, the interval in the sixth being four days. At the time of abortion a fresh fetus with a crown-rump length very close to that determined by sonar was found in the products of conception in four cases. From the sonar findings in each of the six patients (Fig. 3), it can be seen that the fetal heart rates were all within normal limits with no tendency for the values to be either high or low. The remaining two parameters, however, show a consistently unusual pattern when considered together, in that, while the crown-rump lengths were all on or above the mean, the gestation sac volumes were all on or below the second standard
160 ‘Gettatton sac volumelmi)
I
1
lbeatrfmin) 180
140
170. 160150-
1301 120
140. 130-
110-
120-
100-
110-
I
90-
100-
800
6
7
8
9
10 11 12 Gestation (weeks)
13
14
15
70-
W90 Crown-Rump length lmrnl
5080
4070
3060
2050
1040
0’ 30
2c
ia C
6
7
8
9 10 11 12 Gertation(weeks1
13
14
FIG3 The fetal heart rates, crown-rump lengths and gestation sac volumes of the six patients who aborted “live” fetuses before 10 weeks of menstrual age. The patient represented by the “square” aborted four days after her examination, the remainder aborting within two days.
SONAR DIAGNOSIS OF PREGNANCY FAILURE
was not a continuing one, and all ten were subsequently terminated. Histology confirmed the diagnosis in all patients. Miscellaneous (4 patients). The four patients in this group were either lost to follow-up or aborted at home. In each case a provisional diagnosis of blighted ovum had been made. DISCUSSION This investigation has shown that given a knowledge of menstrual history it is possible to diagnose virtually all cases of blighted ovum, missed abortion and hydatidiform mole by a single sonar examination, but that if such sonar diagnoses are to be made with certainty then the menstrual history must be ignored and the normality or otherwise of the pregnancy assessed on the sonar evidence alone. This exclusion of the menstrual history when making a sonar assessment of an abnormal pregnancy is of great importance in practical terms if the possibility of a false positive diagnosis due to misleading or mistaken menstrual dates is to be avoided. On this basis and using the criteria defined in this study it has been shown that half of the cases of blighted ovum can then be diagnosed with confidence at the first examination, the remainder requiring a second and occasionally a third examination. While the number of firm diagnoses at the first examination is markedly decreased by this constraint, those pregnancies which have been diagnosed as blighted can then be managed, or terminated, without fear that an error may have been made. The diagnosis of a missed abortion however is not constrained by any knowledge or lack of knowledge of the menstrual history, for in these pregnancies a diagnosis is based on the simple premise that if a fetus can be seen yet no heart movements can be detected then the fetus is dead. The recognition of a hqdatidiform mole is similarly independent of the menstrual maturity. The diagnostic problem in these patients is a more academic one in that it can on occasion be difficult to differentiate a hydatidiform mole from a long-standing missed abortion and vice versa, especially when the placenta of the missed abortion has undergone hydropic degeneration. In either case, however, the pregnancy can confidently be labelled as noncontinuing.
855
The primary object of this investigation was to formulate criteria for the sonar diagnosis of abnormal pregnancies such that these diagnoses could be applied prospectively and with complete reliability in the active management of established early pregnancy failures. Before applying these criteria as a basis for intervention, however, clinical confidence in their accuracy had to be established, and in the earlier part of the series most pregnancies were allowed to take their natural course and abort spontaneously or were only terminated following several confirmatory examinations at weekly intervals. With increasing confidence, however, and an appreciation that no false positive results were being obtained it was felt that patients could be offered elective termination of their pregnancies whenever a firm diagnosis had been established. While there were incidental clinical grounds to support the sonar diagnosis in the majority of cases, termination of pregnancy was performed on sonar grounds alone in eight patients who had given no history of bleeding and whose urine pregnancy tests had been persistently positive. Any other worker wishing to provide a similar service for the diagnosis and management of early pregnancy failure must be prepared to establish the techniques in his own hands for an initial “non-intervention” period with good lines of communication with the obstetrician who is managing the patient. It may be argued that these pregnancies will all ultimately abort spontaneously and that early surgical intervention is simply needless interference. However, the great majority of patients with spontaneous abortions will ultimately come to theatre and the early diagnosis by sonar allows what may have been an emergency procedure to be carried out electively with increased safety for the patient. Further considerations are the psychological stress to the patient of awaiting a spontaneous abortion, and the cost of hospitalization if kept under specialist care. The diagnosis of blighted ova, missed abortion and hydatidiform moles can be made with accuracy but a much more difficult problem is the diagnosis or anticipation of fetal death or abortion in a living pregnancy. In the group of ten patients who were scanned before their pregnancies became missed abortions the sonar
856
ROBINSON
findings when the fetus was alive showed no significant departure from normal apart from the one fetus whose heart rate was irregular. Thus it is not yet possible to anticipate fetal death by sonar. These pregnancies, however, are to be distinguished from those in which the fetus is alive up to the time of spontaneous abortion. In the latter group sonar examination consistently showed the gestation sac volumes to be abnormally small in relation to the crownrump length of the fetus, a feature which was very uncommon in pregnancies which progressed normally. While this low “sac-fetus ratio” in the first ten weeks of pregnancy is not diagnostic of impending abortion it may serve as a warning that all may not be well. In contrast in those pregnancies which resulted in the abortion of a live fetus after the fourteenth week, the sonar findings were of no value in predicting outcome. The blighted ovum and missed abortion groups form the major portion of the abnormal pregnancies in this series. If the miscellaneous group and the patients with hydatidiform moles are excluded (most of them were referred from other hospitals) then the incidence of each of the above groups is seen to be 42.6 per cent, the equivalent figure for the early and late live abortions being 11 . I per cent and 3 . 7 per cent respectively. The relative incidence of each of the groups in this series is similar to that found by Fujikura et a1 (1966), who, in a pathological study of 327 aborted pregnancies found 5 . 5 per cent of their specimens to have totally empty intact sacs and a further 23.5 per cent ruptured but complete sacs with no evidence of a fetus or cord. Of their 72 (22 per cent) incomplete specimens it is probable that a significant proportion were anembryonic and if half of this group were added to those above then the incidence of blighted ovum in the two studies would be remarkably close. The remaining 50 per cent all had an accompanying fetus or there was evidence of a cord stump, and of these 31 (9.5 per cent) showed no evidence of maceration and might therefore be considered as “live” abortions. It is of note that there were no fetuses in this “live” group whose crown-rumps length was equivalent to an age of between 10 and 14 weeks, a finding which exactly parallels that in this investigation. All of the remaining fetuses were macerated to a variable degree and the
majority would therefore have been missed abortions. It is clear from these investigations that missed abortions are common and should not be confused with the more uncommon condition of “carneous mole”. This investigation has shown that early live abortions occur infrequently and that their natural history as illustrated by sonar is very different from that of missed abortions and blighted ova. A defect at the embryonal or fetal level such as a chromsomal abnormality (Carr, 1963 and 1965) would be the logical aetiology of blighted ova and missed abortions, while abnormality of the trophoblast may be the cause of early live abortions as the formation of amniotic fluid (largely a function of the trophoblast at this stage) has been shown to be markedly reduced. Conversely, the aetiology of late live abortions can usually be ascribed to cervical incompetence. In terms of the clinical management of a patient with a history of recurrent abortion a knowledge of the types of her previous abortions could be of considerable value. For example, there would be little point in exercising particular care in patients who had had missed abortions or blighted ova, while early live abortions might be due to hormonal deficiency and a recurrent abortion might be prevented by suitable supportive therapy.
ACKNOWLEDGEMENTS I am grateful to Professor Ian Donald for his advice and comments during this investigation and I thank the obstetric consultants of the Queen Mother’s Hospital whose patients were included in this study, and Mr J. Devlin and his department for preparing the illustrations. This work was performed under the provisions of a Medical Research Council Programme Grant in the University of Glasgow, with Professor Ian Donald as the Programme Leader. REFERENCES Brown, J . B., Evans, J. H., Beischer, N. A., Campbell, D. G., and Fortune, D. W. (1970): Journal of Obstetrics and Gynaecology of the British Commonwealth, 77, 690. Carr, D. H. (1963): Lancet, 2, 603. Carr, D. H. (1965): Obstetrics and Gynecology, 26, 308. Donald, I., Morley, P., and Barnett, E. (1972): Journal of Obstetrics and Gynaecology of the British Commonwealth, 79, 304.
SONAR DIAGNOSIS OF PREGNANCY FAILURE
Fujikura, T., Froehlich, L. A., and Driscoll, S. G . (1966): American Journal of Obstetrics and Gynecology, 95, 902. Hellman, L. M., Kobayashi, M., and Cromb, E. (1973): American Journal of Obstetrics and Gynecology, 115, 615. MacVicar, J., and Donald, I. (1963): Journal of Obstetrics and Gynaecology of the British Commonwealrh, 10, 387. Niven, P. A. R., Landon, J., and Chard, T. (1972): British Medical Journal, 3, 799.
857
Robinson, H. P. (1972): British Medical Journal, 4, 466. Robinson, H. P. (1973): British Medical Journal, 4, 28. Robinson, H. P., and Shaw-Dunn, J. (1973): Journal of Obstetrics and Gynaecology of the British Commonwealth, 80, 805. Robinson, H. P. (1975): British Journal of Obstetrics and Gynaecology, 82, 100. Robinson, H. P., and Fleming, J. E. E. (1975): British Journal of Obstetrics and Gynaecology, 82,102. Seppala, M., and Ruoslahti, E. (1972): British Medical Jortral, 4, 769.
