- Email: [email protected]

fluid with Ziehl-Neelsen staining. Cytologic and subsequent cultures of the fluid sputum and urine were negative. Results of fluorescence tests for antinuclear factor were also negative. On the basis of these investigations, a diagnosis of hydrothorax caused by ovarian hyperstimulation was made. The patient made an uneventful recovery with no further treatment and was discharged 4 days after admission. The patient returned to the Bourn-Hallam Centre two weeks after embryo replacement for review. She felt well and on examination had no evidence of recurrence of the effusion. A j3-subunit hCG assay was performed, and the level was >25 lUlL, indicative of pregnancy. She returned home for antenatal care but miscarried at 8 weeks' gestation. Comment This is the second reported case of ovarian hyperstimulation that presents solely as acute hydrothorax, and is the first case to occur after in vitro fertilization. The severe form of ovarian hyperstimulation syndrome is characterized by gross ovarian enlargement, ascites, pleural effusion, and often thromboembolic phenom-

August 1989 Am J Obstet Gynecol

ena that are potentially lethal conditions. In this case, although there was evidence of hemoconcentration, gross ovarian enlargement and ascites were not present either clinically or on ultrasonographic examination. Other investigations failed to reveal any other possible cause for the effusion. Severe ovarian hyperstimulation occurs more commonly in patients who conceive after therapy. Some workers have quoted an incidence as high as 6%. However, with present methods of monitoring gonadotropin therapy, the incidence is now much lower! The pathogenesis of hyperstimulation is thought to be a transudation of fluid through the capillary basement membranes of the grossly enlarged ovaries. The absence of ascites and gross ovarian enlargement in this case leaves the mechanism by which the hydrothroax developed debatable. REFERENCES 1. Jewelewicz R. Vande Wiele RL. Acute hydrothroax as the only symptom of ovarian hyperstimulation syndrome. AM J OBSTET GYNECOL 1975;121:1120. 2. Schenker JG. Weinstein D. Ovarian hyperstimulation syndrome: a current survey. Fertil Steril 1978;30:255-66.

Normal amniotic fluid volume changes throughout pregnancy Robert A. Brace, PhD, and Edward J. Wolf, MD La] oila, California The purpose of this study was to provide a quantitative characterization of the changes in amniotic fluid volume that occur throughout gestation. From 70S published amniotic volumes for normal pregnancies. we found that after log transformation. amniotic fluid volume had a uniform variability over 8 to 43 weeks' gestation. Thus the 9S% confidence interval covered the range of 112.S7 to 2.S7 times the mean volume at any given gestational age. Contrary to expected trends. mean amniotic fluid volume did not change significantly between 22 and 39 weeks and averaged 777 ml, with the 9S% confidence interval ranging from 302 to 1997 ml. The data are summarized in nomograms covering 8 to 43 weeks' gestation. (AM J OSSTET GYNECOL 1989;161 :382-8.)

Key words: Amniotic fluid volume, gestational age It is a well-recognized concept that maintenance of amniotic fluid volume within the normal range is imFrom the DivISIOn of Perinatal Medicine, Department ofReproductIVe Medicine, University of California, San Diego. Supported in part by NatIOnal Instztutes of Health grant HD23724 from the National Institute of Chzld Health and Human Development. Received for publicatIOn November 2, 1988; revISed January 1, 1989; accepted March 3,1989. Reprint requests: Robert A. Brace, PhD, Department of ReproductIVe Medicme T-002. University of Californza. San DIego, La Jolla. CA 92093.

382

portant for promoting fetal well-being. This is because aberrations in volume either above or below normal are associated with an increased incidence of fetal and neonatal morbidity and mortality. 1. 2 A problem with this concept is that "normal" has not been defined; that is, although there have been several studies in which amniotic fluid volumes have been measured at various stages of gestation,3-14 there has been neither an adequate characterization of the normal range of amniotic fluid volumes nor an adequate description of the variation in normal volumes with gestation. Thus the pur-

Gestational changes in amniotic fluid volume

Volume 161 Number 2

383

• ....

2500

•

.5 ~ 2000

3o >

S

• • ......................................................•............................!•...........•.....•.....~ .......... . • • •

1500

~

.J

• •

II.. U

l:! o

1000

H

Z %

500

GESTATIONAL AGE (weeks) Fig. 1. Amniotic fluid volumes in human pregnancies with normal fetuses (n = 705). Solul lme is polynomial regression line (Y = 525.6 -117.2X + 8.003X' -0.1237X', R = 0.718, P < 10- 6 ).

pose of the present study was to provide a quantitative characterization of the gestational changes in amniotic fluid volume that occur with normal pregnancies. Methods

Paired data on amniotic fluid volume and fetal gestational age were collected from 12 previously published studies.'·14 In these studies amniotic fluid volumes were measured either at the time of hysterotomy by direct collection," ,. 6. 9.11-14 or in continuing pregnancies by means of indicator dilution techniques!-B. 10. 13 For the present study, data were excluded if fetal anomalies, fetal death, spontaneous abortion, toxemia, or maternal or fetal disease were present. Thus data were included only if the fetus was considered normal. Fetuses with oligohydramnios or polyhydramnios were included if outcome was good. The data were collected either from published tables or figures. For the latter, photographic enlargement combined with linear measurement of X-Y coordinates was used to convert points in the figures into numeric values. Data analysis. Initially, the data are presented as the relationship between volume and age on a linear scale. For statistical analyses, amniotic fluid volumes were log transformed (base 10) because their variability was greater late in gestation than early in gestation and because the volumes at any given age were not normally distributed largely because of the presence of a few high volumes. The data were then analyzed with standard polynomial regression techniques. In addition,

mean values were computed for each 2-week period (i.e., 8 to 9.99, 10 to 11.99, etc., weeks). In the figures containing 2-week averages, volumes were plotted at the computed mean age rather than at the midpoint; for example, 8.2 was the mean for the 8.0 to 9.99 week window and 38.6 was the mean for 38 to 39.99 weeks. Statistical differences in volumes among age groups were tested by means of a one-way analysis of variance with Fisher's least significant difference for multiple comparisons." Variability was analyzed for the residuals calculated from the log of amniotic fluid volume minus the log as determined with the best-fit polynomial regression equation. To avoid biasing as a result of the different number of observations at each gestational age, the polynomial regression equation was determined from the means at 2-week intervals. Of 706 collected values, one volume of zero at 43 weeks was excluded from the analysis because it could not be log transformed and because there must have been at least a little fluid present. It should be noted that it was possible to statistically define confidence intervals only for log-transformed data because volumes were not normally distributed about the mean before log transformation. To convert these into easily interpreted values, antilogs must be taken. When this is done, the confidence interval is no longer equal to the mean ± a constant but rather equals the mean multiplied and divided by a constant. For example, if the mean and 95% confidence interval were 2 ± 0.3 for the log data, the mean volume would be 100 ml (antilog of 2), with a confidence interval of

384

Brace and Wolf

August 1989 Am J Obstet Gynecol

4000

-...

•

•

2000

,

•

1000

.! UJ

~

5>

o

H

~ tJ

400 200

·•

....:.. .

•

100

.............................................................................................

• •

•• •

40

•

H

I-

.',

o

H

Z % C

10 ........................................................................................................................... .

•

4

•

2~~~~

8

12

__

L-~~~

16

__L-~~~__L-~~~__L-~~

20

24

28

32

36

40

44

GESTATIONAL AGE (weeks) Fig. 2. Amniotic fluid volumes on a log scale in human pregnancies with normal fetuses (n = 705). Solid line is polynomial regression line (see text). 1.2

.8

'iii' ....

.... c:

~

01 0

C

0

UI oJ C ::J

0

H

UI III

......i"...... ·.-. "., M ••

.4

•• • • . ................................~ .............................. ~...........~ .........j . '. ~..... t 4 tJ ~. :~ ~•• •• I ..:. .,.~.~ .~~ ;. :1

¥ ...

. . !.. . •

-.4

--

.

.·1

fT' . ,

••• t

•

•

•:.. ' ", . " .... .. .• .,. .. •••••

••

............................................................!I! ..............................!!.................... .

••

a:

-.8

-1.2

.

.. Ltt~·"':·~·~S·tk~·;~t..· •i!"

••

• •• •

•

• •

• 8

•

•

• •

12

• 16

20

24

28

32

36

40

44

GESTATIONAL AGE (weeks) Fig. 3. Variability in amniotic fluid volumes as a function of gestational age. Data expressed as residuals calculated by subtracting value determined with polynomial regressional equation from individual points. Area between upper and lower dotted lines represents 95% confidence interval.

1/2 X 100 to 2 of 0.3 is 2.

X

100, or 50 to 200 ml, since the antilog

Results A total of 705 amniotic fluid volumes from normal fetuses were collected from 8 to 43.2 weeks' gestation. These are shown on a linear scale in Fig. 1 as a function

of gestational age, along with the line of best fit as determined by polynomial regression. From the regression equation, the peak was 931 ml and occurred at 33.8 weeks. Two major problems occur with the use of poly nomial regression to describe the changes in amniotic fluid volume. First, as seen in Fig. 1, the scatter about the

Gestational changes in amniotic fluid volume

Volume 161 Number 2

1000

400

385

.............................................................................................................................

••

•

ILl

:I: :J ...J

o > o

...

200

100 .......................................................................................................................... .

:J ...J IL. U

...

... ""

t-

40

o

Z :I:

10

............................................................................................................................ .

GESTATIONAL AGE (weeks) Fig. 4. Statistical analysis of gestational changes in amniotic fluid volume. Data expressed as mean ± SE for each 2-week interval. *p < 0.0001 compared with value at 39 weeks' gestation. 4000 2000

...

1000 7!11

.! ILl

:I: :J ...J

o >

... o

!IOI 2!51

400

!II

200

11

100

:J ...J IL.

...u b ...z ~

40 20

10

· · ........................................................................................................................ .

4

GESTAIONAL AGE (weeks) Fig. 5. Nomogram shows mean amniotic fluid volumes (dots) as a function of gestational age and percentiles. Volumes are plotted on a log scale. Fiftieth percentile line is polynomial regression calculated from biweekly means (log Y = -l.3916 + OA176X - 0.01622X' + 0.0003144X3 0.0000026X', R = 0.998, F = 811, P < 10- 6 ). Shaded area covers 95 % confidence interval.

mean was much greater late in gestation compared with early in gestation. Thus the underlying assumption of uniform variability has been violated. Second, as readily seen in Fig. 1 at 30 to 40 weeks, the scatter above the mean (i.e., the regression line) was much greater than the scatter below the mean. Thus the assumption of a normal distribution was violated.

In an attempt to overcome these problems, amniotic fluid volumes were log transformed and are shown on a log scale in Fig. 2. Because it is difficult to visually perceive variability about a curved line, Fig. 3 replots the data of Fig. 2 after subtracting the regression line. An analysis of variance of the data in Fig. 3 found no statistically significant changes in the residuals (i.e.,

386

Brace and Wolf

August 1989 Am J Obstet Gynecol

w

X :J

6 >

1500 991

a

H

:J ...J LI. U

H

...z ~

a

"' 500 . . .... . . . . . . . ... .

751

X

501

251

51

0~~~~~§E§S::::===-L-~~__~-L~~~~__L--=~11 8

12

16

20

24

28

GESTATIONAL AGE

32

36

40

44

(weeks)

Fig. 6. Nomogram shows amniotic fluid volumes as a function of gestational age on a linear scale. Dots are means for each 2-week interval. Percentiles are calculated from polynomial regression equation and SD of residuals. Shaded area covers 95% confidence interval.

variability) with gestational age. In addition, after log transformation, amniotic fluid volumes were normally distributed about the mean for any given gestational age. There are three major benefits of these results. First, polynomial regression now provides an adequate characterization of the changes in amniotic fluid volume with advancing gestation. When analyzed with this technique, each of the first four regression coefficients were statistically significant (P < 0.002). Thus the regression equation relating amniotic fluid volume (AFV) to gestational age (X) was log (AFV) = - 1.5595 + 0.4676 X -0.02014 X2 + 0.0004359 X' - 0.0000039 X', with R = 0.896, F = 713, and p < 10- 6 • This equation is represented by the solid line in Fig. 2. The second benefit is that standard statistical methods can be used to compare the changes in mean volumes. As determined with an analysis of variance, the changes in amniotic fluid volume (AFV) with gestational age (Fig. 4) were highly significant (F = 160, P < 10- 6 ). However, there were no statistically significant changes in volume between 22 and 39 weeks' gestation; volume averaged 777 ml, and individual means ranged from 630 to 817 ml (Fig. 4). The third benefit is that the variability of the data in Fig. 3 provides an overall estimate of the variability in amniotic fluid volume. Thus the standard deviation from the data in Fig. 3 can be used to characterize the normal range of amniotic fluid volumes throughout the range of gestational ages considered. From Fig. 3 the

SD is 0.2094 so that the 95% confidence interval (1.96 X SD) is 0.4105 (log units). This means that 95% of all women with a normal fetus have amniotic fluid volumes within the range of 1/2.57 to 2.57 times, or within 39% to 257% of the mean volume for any gestational age over 8 to 43 weeks' gestation (2.57 is the antilog of 0.4105). Nomograms showing the changes in amniotic fluid volume as a function of gestational age are shown in Fig. 5. As can be seen, these were calculated by adding or subtracting various multiples of the overall SD of the residuals to the mean volume calculated from the polynomial regression equation (i.e., the fiftieth percentile). The shaded area covers the 95% confidence interval (i.e., the 2.5 to 97.5 percentiles). These same data are replotted on a linear scale in Fig. 6 to facilitate comparisons. The mean changes in amniotic fluid volume calculated on a weekly basis (from the polynomial regression equation) are shown in Fig. 7. At 8 weeks, volume is increasing at 10 mil week. This increases to 25 mil week at 13 weeks and reaches a maximum of 60 mil week at 21 weeks' gestation. The weekly volume increment then decreases and reaches zero at 33 weeks' gestation, that is, volume is at maximum at this time. In contrast to this pattern, the weekly volume change when expressed relative to total volume decreases throughout weeks 8 to 43 of gestation (dashed line in Fig. 7). Thus mean amniotic fluid volume is increasing 45% per week at 8 weeks, 25% per week at 15 weeks, 10% per week at 24

Gestational changes in amniotic fluid volume

Volume 161 Number 2

387

25 ....................~- ................................................................................................... .. '" '.

'-

UJ

I.!)

Z

~

J: U

UJ

~

.J

o ...................................

::-;z:::::~::::,,~

IIEEKLY VOLUME CHANGE IS)

-25 ............................................................................................................................. .

o > >.J

Itl~

-50 ............................................................................................................................. .

x

-75BL-~~---~~~~~~-L~---~~~~L-3~6~~~40~~~44

12

16

20

24

2B

32

GESTATIONAL AGE (weeks) Fig. 7. Weekly changes in mean amniotic fluid volume as calculated from best-fit polynomial regression equation.

weeks, and 0% at 33 weeks. At 40 weeks, volume is decreasing an average of 8% per week. Comment

The major significance of this study is that we were able to statistically characterize for the first time normal amniotic fluid volumes over the last 80% of gestation. This also provides a statistical definition of abnormal amniotic fluid volumes so that limits can be set for oligohydramnios and polyhydramnios. For example, as seen in Figs. 5 and 6, at 30 weeks' gestation, the 95% confidence interval about the mean (817 ml) covers the range of 318 to 2100 m!. Thus with <318 ml of fluid oligohydramnios would be present and with more than 2100 ml polyhydramnios would be present. Each of these occur 2.5% of the time. Another observation is that after a progressive rise from 8 weeks' gestation, amniotic fluid volume reaches its statistical maximum at 22 weeks' gestation and remains at this level (630 to 817 ml) through 39 weeks' gestation, declining thereafter. Although Fuchs 16 had reported that mean amniotic fluid volume was relatively constant during the last trimester, it has been more widely accepted that amniotic fluid volume reached a peak at 34 weeks' gestation and declined thereafter. 10 This difference is largely attributable to the different methods of averaging (absolute values versus logs) in the studies; that is, a few cases of hydramnios had biased the mean value upward at 34 weeks' gestation. In addition, a statistical analysis of the volume changes

with time was not performed in the previous studies so that these two statements are not necessarily inconsistent with each other. This same contradiction arises in the present study in that the analysis of variance revealed no significant changes in volume between 22 and 39 weeks (Fig. 4). On the other hand, the mean volume as determined from the regression equation reached its peak at 33 weeks and declined thereafter, which is in agreement with Queenan et al. tO In this case, the polynomial regression equation most likely provides a better estimate of the trends in amniotic fluid volume. Thus with a much larger data set, a peak around 33 weeks' gestation would be more easily identified. A final new observation is that from 8 to 43 weeks' gestation, the variability in amniotic fluid volume about the mean is constant when expressed as logtransformed data. Thus 95% of the time amniotic fluid volume will be within the range of 1/2.57 to 2.57 times the mean volume at any given gestational age. At present we do not know why the variability is constant, but it probably relates to the volume regulatory mechanisms. It also should be noted that this variability is quite large. Factors, such as maternal age, gravidity and parity, and ethnicity and race, may have contributed to the variability, but we were unable to explore these factors because they were absent from the original reports. In summary, this study provides a statistical characterization of the changes in amniotic fluid volume that occur in normal human pregnancy. This also provides

Brace and Wolf

a definition of oligohydramnios and polyhydramnios over 8 to 43 weeks' gestation.

REFERENCES 1. Chamberlain PF, Manning FA, Morrison I, Marman CR, Lange IR. Ultrasound evaluation of amniotic fluid volume. I. The relationship of marginal and decreased amniotic fluid volumes to perinatal outcome. AM] OB5TET GYNECOL 1984; 150:245-9. 2. Chamberlain PF, Manning FA, Morrison I, Marman CR, Lange IR. Ultrasound evaluation of amniotic fluid volume. II. The relationship of increased amniotic fluid volumes to perinatal outcome. AM ] OB5TET GYNECOL 1984; 150:250-4. 3. Abramovich DR. The volume of amniotic fluid in early pregnancy.] Obstet Gynaecol Br Comm 1968;75:728-31. 4. Charles D,]acoby HE, Burgess F. Amniotic fluid volumes in the second half of pregnancy. AM] OB5TET GYNECOL' 1965;93:1042-7. 5. Gadd RL. The volume of the liquor amnii in normal and abnormal pregnancies. ] Obstet Gynaecol Br Commonw 1966;73:11-22. 6. Gillibrand PN. Changes in amniotic fluid volume with advancing pregnancy. ] Obstet Gynaecol Br Commonw 1969;76:527-9.

August 1989 Am J Obstet Gynecol

7. Haswell GL, Morris ]A. Amniotic fluid volume studies. Obstet Gynecol 1973;42:725-32. 8. Marsden D, Huntingford PJ. An appraisal of the Coomassie blue dilution technique for measuring the volume of liquor amnii in late pregnancy. ] Obstet Gynaecol Br Commonw 1965;72:65-8. 9. Nelson MM. Amniotic fluid volumes in early pregnancy. ] Obstet Gynaecol Br Commonw 1972;79:50-3. 10. Queenan ]T, Thomson W, Whitfield CR, Shah SI. Amniotic fluid volumes in normal pregnancies. AM J 085TET GYNECOL 1972;1l4:34-8. II. Rhodes P. The volume of liquor amnii in early pregnancy. ] Obstet Gynaecol Br Commonw 1966;73:23-6. 12. Sinha R, Carlton M. The volume and composition of amniotic fluid in early pregnancy. ] Obstet Gynaecol Br Commonw 1970;77:211-4. 13. Van Otterlo LC, Wladimiroff ]W, Wallen burg HC. Relationship between fetal urine production and amniotic fluid volume in normal pregnancy and pregnancy complicated by diabetes. ] Obstet Gynaecol Br Commonw 1977;84:205-9. 14. Wagner G, Fuchs F. The volume of amniotic fluid in the first half of human pregnancy.] Obstet Gynaecol Br Commonw 1962;69:131-6. 15. Dowdy S, Wearden S. Statistics for Research. New York: John Wiley & Sons, 1983. 16. Fuchs F. Volume of amniotic fluid at various stages of pregnancy. Clin Obstet Gynecol 1966;9:449-60.

Doppler velocimetry and placental disease Luis A. Bracero, MD, Debra Beneck, MD, Nancy Kirshenbaum, MD, Marianne Peiffer, RDMS, Patricia Stalter, CDMS, and Harold Schulman, MD Valhalla and Mineola, New York Quantitative placental examinations were performed on 47 women who had Doppler flow velocity studies of the umbilical artery during their pregnancy. The systolic-diastolic ratio of the umbilical artery was used as the measurement parameter to divide the study population into two groups. Group 1 consisted of women with normal systolic-diastolic ratios (systolic-diastolic < 3), and group 2 consisted of women with an elevated systolic-diastolic ratio (systolic-diastolic"" 3). The group with an increase in systolic-diastolic ratio had more perinatal complications as demonstrated by two stillbirths, a higher incidence of cesarean deliveries for fetal distress, and more admissions to the neonatal intensive care unit. Significant differences were found when gestational age at delivery, placental weight, birth weight, and the number of small muscular arteries in the placenta were compared. Since gestational age may have accounted for the difference in placental findings, patients were matched for gestational age. The placental weights were comparable, but there were fewer small muscular arteries in those patients with an increase in systolic-diastolic ratio (p < 0.001). In addition, when these findings were examined to determine the influence of diminished uterine flow velocity, none was found. (AM J OBSTET GVNECOL 1989;161 :388-93.)

Key words: Umbilical artery velocity waveforms, Doppler ultrasound, placenta From the Department of Obstetrics and Gynecology and the Department of Pathology, New York Medical College, Westchester County Medical Center, and the Department of Obstetrics and Gynecology, Wznthrop UniverSIty Hospital. Received for pubhcation April 1. 1988; revised February 22, 1989; accepted March 3, 1989. Repnnt requests: Luis A. Bracero. MD. Department of Obstetrics and Gynecology, Westchester County Medical Center, Valhalla. NY 10595.

388

Umbilical artery velocity waveform analysis has been proposed as a new placental function test. It has been reported that an umbilical artery systolic-diastolic ratio of <3 during the third trimester implies normal placental function, whereas a systolic-diastolic ratio 2: than 3 implies a strong probability of abnormal outcome. I These outcomes include small-for-gestational age fe-