(Radiology. 2000;215:453-457.)
© RSNA, 2000
Normal Fetal Pelvis: Important Factors for Morphometric Characterization with US1
Mark A. Kliewer, MD,
Barbara S. Hertzberg, MD,
Kelly S. Freed, MD,
Patricia J. McNally, RDMS and
David M. DeLong, PhD
1 From the Department of Radiology, Duke University Medical Center, Box 3808, Rm 2526, Blue Zone, South, Durham, NC 27710. Received February 15, 1999; revision requested April 5; revision received August 26; accepted August 30. Address correspondence to M.A.K.
 |
Abstract
|
|---|
PURPOSE: To prospectively evaluate iliac angle and iliac length in a large number of normal fetuses and to identify factors that may influence these measurements.
MATERIALS AND METHODS: At antenatal ultrasonography (US) in 356 fetuses, the iliac angle and iliac length were measured at two axial levels (superior and inferior). In mixed linear models, the statistical significance and magnitude of effect on the measurement of iliac angle and iliac length were estimated for gestational age, fetal sex, maternal diabetes status, axial level, and spine position relative to the transducer.
RESULTS: Statistically significant effects were found for gestational age, axial level, and spine orientation but not for fetal sex or maternal diabetes status. The iliac angle was found to decrease by 15.7° from the superior to inferior portion of the pelvis, decrease by approximately 0.37°/wk, and decrease by as much as 15.6° when the spine is directed to the side. Iliac length was found to increase by 0.8 mm/wk from 13 weeks to term, decrease by 1.2 mm from the superior to the inferior portion of the pelvis, and increase by as much as 1.29 mm when the spine is not directly subjacent to the transducer.
CONCLUSION: The axial level of measurement, gestational age, and spine orientation must be accounted for if these morphometric indexes are used to discriminate fetuses with and those without Down syndrome.
Index terms: Down syndrome, 856.1298, 856.873 • Fetus, abnormalities, 856.873 • Fetus, skeletal system, 856.873 • Fetus, US, 85.1298 • Pelvis, measurement, 44.1298, 856.873
|
Introduction
|
|---|
Morphometric analysis of the fetal pelvis has recently attracted attention as a potential method of adjusting risk of aneuploidy (1–4). This approach has its origins in much earlier radiology studies of infants and children in which dysmorphic pelvic features were used to help diagnose a wide range of genetic disorders and skeletal dysplasias (5–7). Findings in initial prenatal ultrasonographic (US) investigations suggest that measurements of the axial pelvic profile may be useful in the identification of fetuses at risk for trisomy 21, but conclusions from these studies are limited either by a retrospective study design (1,2), relatively small sample sizes (1–4), potential bias produced by foreknowledge of associated fetal anomalies (3,4), or failure to control for potentially significant covariants such as gestational age, axial level, and fetal sex that might influence the proposed measurements (1–4). The goals of this study were to prospectively establish biometric standards for the iliac angle and iliac length in a large number of normal fetuses and to test the importance of potentially confounding factors on these measurements.
|
MATERIALS AND METHODS
|
|---|
Between August 1997 and January 1998, the fetal pelvis was assessed prospectively in 420 consecutive US studies performed in 358 singleton pregnancies of women referred for obstetric US between 13 menstrual weeks gestation and term. Of these 358 fetuses, 296 were studied once, and 62 were studied twice. US studies were performed for a variety of indications such as estimation of fetal age, assessment of growth, evaluation of vaginal bleeding, and exclusion of fetal anomalies. Fetuses with anatomic or growth abnormalities and those with aneuploidy proved at amniocentesis were excluded from the study. All measurements were made prospectively at the time of the examination by using standard US equipment (models 128 and XP; Acuson, Mountain View, Calif) with 2.5-, 3.5-, or 5.0-MHz linear, sector, or curved linear electronically focused transducers.
For each study, pelvic measurements were obtained in two axial planes: a superior level in the upper half of the iliac wing and an inferior level in the lower half of the iliac wing (Fig 1). The transducer was positioned to optimize depiction of both sides of the pelvis in a true axial plane. At both the superior and inferior axial levels, measurements were made of the iliac angle and the anteroposterior lengths of the iliac wings (both right and left) (Fig 2). The iliac angle was defined as the angle formed by the convergence of lines drawn on the posterolateral aspect of the right and left wings of the ilium (1). This angle was measured by a sonologist (M.A.K.) with a hand-held goniometer for all studies. Linear measurements of the lengths of the iliac wings were made with electronic calipers.
View larger version (104K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 1a. US scans depict the axial pelvic profile at (a) superior and (b) inferior levels in a fetus at 16.1 weeks. Note the wider and more open configuration of the pelvis in a compared with that in b. Iliac wings (single arrows) are more divergent and oriented toward the coronal plane (double arrow) in a.
|
|
View larger version (111K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 1b. US scans depict the axial pelvic profile at (a) superior and (b) inferior levels in a fetus at 16.1 weeks. Note the wider and more open configuration of the pelvis in a compared with that in b. Iliac wings (single arrows) are more divergent and oriented toward the coronal plane (double arrow) in a.
|
|
View larger version (88K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 2a. US scans at a superior axial level in a fetus at 23 weeks gestation. (a) Iliac angle ( ) is measured between lines drawn tangential to the posterior margin of the iliac wings. (b) Iliac length is measured with electronic calipers positioned at the anterior and posterior margins of the bone.
|
|
View larger version (94K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 2b. US scans at a superior axial level in a fetus at 23 weeks gestation. (a) Iliac angle () is measured between lines drawn tangential to the posterior margin of the iliac wings. (b) Iliac length is measured with electronic calipers positioned at the anterior and posterior margins of the bone.
|
|
The best estimate of gestational age (BEGA) of the fetus, maternal diabetes status, and position of the fetal spine relative to the transducer were noted. The relative position of the fetal spine was recorded as if the spine pointed to a number on a clock face, with the transducer at the 12:00-o'clock position (Fig 3). For example, if the fetal spine was directly subjacent to the transducer (spine up), then the position was recorded as the 12:00-o'clock position; if the fetal spine was down, with the posterior elements directed away from the transducer, then the spine position was recorded as the 6:00-o'clock position. An attempt was made to ascertain the sex of all fetuses by means of visual inspection of the genitalia.
View larger version (106K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 3. US scan depicts the spine oriented at the 3-o'clock position at a superior axial level in a fetus at 19.4 weeks gestation. Spine position was recorded as if situated on a clock face, with the transducer at the top of the image designated as the 12-o'clock position.
|
|
Statistical analysis included calculation of means and SDs for the iliac angle and iliac length at six gestational age ranges (Table 1). Coefficients of variation were calculated by dividing the SD by the mean and then multiplying the result by 100. The correlations between the iliac angle and iliac lengths were characterized with Pearson product moment correlation coefficients.
For each of the morphometric indexes (iliac angle and right and left iliac lengths), a set of mixed linear models was fit for each of five predictive variables (gestational age, fetal sex, maternal diabetes status, axial level, and spine position) separately and for sets of these predictive variables. Model building progressed by initially creating univariate regression models that tested the effect of each predictive variable on each outcome variable (iliac angle and length) separately. Predictive variables found to have a statistically significant effect on a particular outcome measure were combined into a composite model and retested. There was, therefore, one composite model for iliac angle and one for iliac length. By means of a process of backward elimination, the least significant variables were removed one at a time from each composite model until a combination of predictive variables remained that were all significant at P values of .05 or less. The regression coefficients from the two final models provided estimates of the magnitude of the effect of each predictive variable on the morphometric indexes.
The mixed linear models allowed estimation of the magnitude of effect of an independent variable while controlling for all other variables and also for correlation between repeated measurements within a fetus. The intrinsic sampling variability of each morphometric index (the background noise of the measurement) was estimated. Differences in test results were considered statistically significant at P values of .05 or less.
|
RESULTS
|
|---|
Amniocentesis was performed for 23 of the 358 fetuses studied. There was a normal karyotype in 21 cases and aneuploidy in two (one case of trisomy 13 and one of Turner syndrome). The two aneuploid fetuses were removed from the study, which left a study group of 356 fetuses. For this study group, at least one axial image of the fetal pelvis could be obtained in 409 of 418 studies (97.8%). In these 409 studies, measurements were made at 805 axial levels: Both axial levels (superior and inferior) could be obtained in 396 studies, but only one level (superior or inferior) could be obtained in 13 studies. Fetal position precluded successful imaging of the pelvis in nine cases: Six of these were more than 20 weeks gestation, and three were less than 20 weeks gestation.
Table 1 summarizes iliac angles and lengths for six ranges of gestational age. The iliac angle tended to decrease throughout gestation, and the iliac length progressively increased. Stratification of these results on the basis of axial level further demonstrated that there is a difference of 15°–20° in iliac angle and of 0.8–6.0 mm in iliac length between the superior and inferior level depending on the BEGA. The mean iliac angle decreased for each successive BEGA category with the exception of the 29.0–32.9-week category. This fluctuation was likely a product of the variability of the angle throughout gestation, as evidenced by the relatively large SDs for the means in each BEGA category.
Calculation of coefficients of variation from Table 1 indicated that 1 SD in the measurement of these three variables was large relative to the means: One SD was between 23% and 30% of the mean values for the iliac angle and between 14% and 20% of those for the iliac length. The coefficients of variation were similar for the age groups, which indicates that the spread of the data was relatively constant throughout gestation.
Fetal sex was ascertained in 220 fetuses, and there were 100 female (46%) and 120 male (54%) fetuses. In the regression model, fetal sex did not have a statistically significant influence on iliac length (P = .99) but was close to significance for the iliac angle (P = .052). The iliac angle tended to be 3.8° less in male fetuses. Maternal diabetes mellitus was present in 24 cases (6.7%). Maternal diabetes status also did not exert a statistically significant effect on iliac length (P > .10) or iliac angle (P > .10).
The effects of BEGA, axial level, and spine orientation were statistically significant for the iliac angle and length (Table 2). After adjusting for the effects of BEGA and spine position, the effect of measuring at the inferior level rather than the superior level would be to decrease the iliac angle by 15.7° and decrease the iliac length by approximately 1.2 mm. With advancing gestational age, the iliac length increased by an estimated 0.80 mm/wk. Though statistically significant, the effect of BEGA on the iliac angle was less pronounced, tending to decrease the iliac angle by 0.37°/wk.
The effect of spine orientation relative to the transducer was estimated after the effects of BEGA and axial level were accounted for (Table 3). The measurements of iliac angle and length were grouped into quadrants according to whether the spine was directed toward the transducer (spine up, at the 11- to 1-o'clock positions), away from the transducer (spine down, at the 5- to 7-o'clock positions), or to either side of the image (either spine left, at the 2- to 4-o'clock positions or spine right, at the 8- to 10-o'clock positions). The iliac angle was 12.8°–15.6° less when the spine was directed to the side than when it was directed either up or down. There was no statistically significant difference between measurements with the spine up or down (P = .65) or with the spine right or left (P > .20). Iliac lengths increased by 0.79–0.93 mm when the spine was directed to the side and by 1.29 mm when the spine was directed away from the transducer (spine down). There was no statistically significant difference between measurements with the spine right or left (P = .56).
After adjustment for BEGA, axial level, and spine orientation, the right and left iliac lengths were found to be highly correlated, but iliac length did not correlate strongly with the iliac angle. The Pearson product moment correlation coefficient for the right and left iliac lengths was 0.74 (P = .001). In contrast, the iliac angle correlation to the left iliac length was 0.08 (P = .04) and to the right iliac length was 0.09 (P = .02).
The effect of BEGA, axial level, and spine orientation accounted for 32% of the observed variability in the iliac angles and 41% of the observed variability in the iliac lengths. The variability not accounted for by these effects results from unexplained differences between individual fetuses and from the random variation, or noise, in the iliac angles and lengths. With use of the residual variances from the model, 1 SD for iliac angle was 13° and for iliac length was 2.8 mm. These values are relatively large, considering that they represent 19% and 26% of the mean values for the iliac angle and iliac length, respectively.
|
DISCUSSION
|
|---|
The earliest studies of pelvic morphometry for clinical diagnosis of aneuploidy concerned structural irregularities at the hip on frontal radiographs (5,8). Complementing these observations, a few studies were performed to examine the pelvis in infants with Down syndrome from a craniocaudal perspective, and the iliac wings were more divergent than normal, as if rotated laterally toward the coronal plane (9–11). Recently, this perspective has been studied as a possible marker for aneuploidy at antenatal US (1–4).
Antenatal studies, however, have not yet fully accounted for the complexities of pelvic morphometry or the changes in size and shape of the fetal pelvis during gestation. Two initial investigations were performed to study the angular divergence of the iliac wings by means of an iliac angle, defined as the angle formed by the convergence of lines drawn along the posterolateral aspect of the right and left wings of the ilium (1,2). Findings in these studies suggested a mean difference in this measurement of 12°–15° between fetuses with and those without Down syndrome, but both studies were retrospective, and, therefore, the axial level at which the measurement was taken could not be controlled. Likewise, axial level was unspecified in other studies of the iliac angle (3) and iliac length (4). Further, in one study, results were averaged across the time of gestation without regard for the growth and structural development of the pelvis (3). To our knowledge, no study has been performed to systematically examine the effects of axial level, fetal sex, BEGA, maternal diabetes status, or spine orientation relative to the transducer.
This study demonstrates that measurements of iliac angles and lengths are substantially affected by the particular circumstances of measurement. Specifically, these measurements depend strongly on axial level, BEGA, and spine orientation relative to the transducer.
The iliac angle was most strongly influenced by axial level and spine orientation. The iliac angle was found to decrease by 15.7° from the superior to inferior portion of the pelvis and to decrease by 15.6° when the spine was directed to the side rather than toward the transducer. The iliac angle was relatively stable throughout gestation, tending to decrease by only 0.37°/wk from 13 weeks to term, and mean angles at the superior level decreased less than those at the inferior level (Table 1).
The iliac length was found to (a) decrease by 1.2 mm from the superior to the inferior portion of the pelvis, (b) increase by 0.8 mm/wk from 13 weeks to term, and (c) increase by 0.79–0.93 mm when the spine was directed to the side of the image (at the 3- or 9-o'clock positions) and by 1.29 mm when directed away from the transducer (at the 6-o'clock position). To illustrate, our calculations indicate that in a normal fetus the iliac length would be expected to be 10.3 mm at 18 weeks when the spine is oriented toward the transducer. Ten weeks later, the iliac length would be approximately 18.3 mm: 10.3 mm + (10 wk x 0.8 mm/wk) = 10.3 + 8.0 mm. Alternatively, if the fetal spine were directed to the 3-o'clock position, the iliac length at 18 weeks would be expected to increase 0.93 mm, resulting in a length of 11.23 mm.
The findings in this study are in concert with those in anatomic studies that the fetal pelvis, particularly the iliac bones, evolve in size and shape throughout gestation (12–18). Furthermore, the importance of accounting for the spine orientation relative to the transducer has been suggested in prior US studies that demonstrate the appearance of the fetal spine changes from varying perspectives (19–22) and that bone lengths measured at US can vary with different angulations of the bone (23).
Fetal sex did not appear to strongly influence the morphometric indexes under study, though there was a tendency for male fetuses to have a slightly smaller iliac angle (3.8° less). An effect for maternal diabetes was postulated in view of the well-documented and severe pelvic malformations sometimes seen in infants of diabetic mothers (24,25). Though this factor also did not appear to exert a measurable effect on these pelvic indexes, only a small number of diabetic mothers were included in the study, which limited its power for detection of small differences.
The results of this study have direct implications for the potential use of these measurements in the diagnosis of Down syndrome. After the effects of BEGA, axial level, and spine position are accounted for, the residual SD for iliac length is 2.8 mm and that for iliac angle is 13°. A mean shift of 1 SD corresponds to an area under the receiver operating characteristic curve of 0.78 (under the binormal model); therefore, for the measurements to be diagnostically useful, there would ideally be at least 1 SD of difference between the mean values for a pelvis in a Down fetus and that in a normal fetus. This suggests that the mean iliac length in Down fetuses should differ from that in the normal population by at least 2.8 mm, and the iliac angle in Down fetuses should differ from that in the normal population by at least 13°. This is analogous to detecting a signal (the true morphometric difference) over noise (the random variation in the measurement). In a small study of iliac length in Down syndrome, a length within 2 mm of the normal mean was seen in seven of 10 Down fetuses (4). Furthermore, the mean iliac angle in Down fetuses has been variously estimated as 13° and 15° greater than the normal mean in different studies (1,2). This magnitude of difference in the iliac angle attributable to aneuploidy is, however, similar to that attributable to axial level or spine orientation. Therefore, though the iliac angle seems to be the more promising index, it is imperative that axial level, BEGA, and spine orientation are taken into account if it is to be used as an indicator of aneuploidy.
In summary, findings in this study demonstrate the importance of axial level, BEGA, and spine orientation on the iliac angle and iliac length. Indeed, these results indicate the need to reexamine the assertions made in earlier retrospective studies that did not account for these factors. It is mandatory that all subsequent studies of iliac angle and iliac length in Down and non-Down fetuses be conducted under carefully prescribed conditions of measurement in order to ascertain the true magnitude of the morphometric disparities. Because the iliac angle was found to decrease less at the superior level throughout gestation and because both indexes seemed subjectively easier to measure at the superior axial level with the spine oriented toward the transducer (in the quadrant of the 11- to 1-o'clock positions), we propose the superior level with the spine up as the standard condition for measurement in subsequent studies. Until a standard approach is established, we suggest caution in the interpretation of these parameters for the diagnosis of aneuploidy.
|
Acknowledgments
|
|---|
The authors thank Susan Murray for preparation of the manuscript for this article.
|
Footnotes
|
|---|
Abbreviation: BEGA = best estimate of gestational age
Author contributions: Guarantor of integrity of entire study, M.A.K.; study concepts, M.A.K., B.S.H., D.M.D.; study design, P.J.M., M.A.K., K.S.F., D.M.D.; definition of intellectual content, M.A.K., B.S.H., K.S.F.; literature research, M.A.K., K.S.F.; clinical studies, P.J.M., K.S.F.; experimental studies, P.J.M., K.S.F.; data acquisition, P.J.M., K.S.F.; data analysis, K.S.F., D.M.D.; statistical analysis, D.M.D., M.A.K.; manuscript preparation and editing, M.A.K., B.S.H., K.S.F.; manuscript review, M.A.K., B.S.H., K.S.F., D.M.D.
|
References
|
|---|
-
Kliewer MA, Hertzberg BS, Freed KS, et al. Dysmorphologic features of the fetal pelvis in Down syndrome: prenatal sonographic depiction and diagnostic implications of the iliac angle. Radiology 1996; 201:681-684.[Abstract/Free Full Text]
-
Shipp TD, Bromley B, Lieberman E, Benacerraf BR. The iliac angle as a sonographic marker for Down syndrome in second-trimester fetuses. Obstet Gynecol 1997; 89:446-450.[Medline]
-
Bork MD, Egan JFX, Cusick W, Borgida AF, Campbell WA, Rodis JF. Iliac wing angle as a marker for trisomy 21 in the second trimester. Obstet Gynecol 1997; 89:734-737.[Medline]
-
Abuhamed AF, Kolm P, Mari G, Slotnick N, Evans AT, III. Ultrasonographic fetal iliac length measurement in the screening for Down syndrome. Am J Obstet Gynecol 1994; 171:1063-1067.[Medline]
-
Caffey J, Ross S. Pelvic bones in infantile mongoloidism-roentgenographic features. AJR Am J Roentgenol 1958; 80:458-467.
-
Taybi H, Kane P. Small acetabular and iliac angles and associated diseases. Radiol Clin North Am 1968; 6:215-221.[Medline]
-
James AE, Jr, Merz T, Janower ML, Dorst JP. Radiological features of the most common autosomal disorders: trisomy 21-22 (mongolism or Down's syndrome), trisomy 18, trisomy 13-15, and the cri du chat syndrome. Clin Radiol 1971; 22:417-433.[Medline]
-
Caffey J, Ross S. Mongolism (mongoloid deficiency) during early infancy: some newly recognized diagnostic changes in the pelvic bones. Pediatrics 1956; 17:642-649.[Abstract/Free Full Text]
-
Kaufmann HJ, Taillard WF. Pelvic abnormalities in mongols. BMJ 1961; 1:948-949.
-
Hall LAB. Increased curvature of the ilium: a new roentgenological sign of mongoloid pelvis. Dev Med Child Neurol 1968; 10:781-783.[Medline]
-
Mortensson W, Hall B. Abnormal pelvis in newborn infants with Down's syndrome. Acta Radiol Diag 1972; 12:847-855.
-
Delaere O, Kok V, Nyssen-Behets C, Dhem A. Ossification of the human fetal ilium. Acta Anat 1992; 143:330-334.[Medline]
-
O'Rahilly R, Müller F, Meyer DB. The human vertebral column at the end of the embryonic period proper. IV. The sacrococcygeal region. J Anat 1990; 168:95-111.
-
Laurenson RD. The primary ossification of the human ilium. Anat Rec 1964; 148:209-217.[Medline]
-
Noback CR, Robertson GG. Sequences of appearance of ossification centers in the human skeleton during the first five prenatal months. Am J Anat 1951; 89:1-28.[Medline]
-
Flecker H. Time of appearance and fusion of ossification centers as observed by roentgenographic methods. AJR Am J Roentgenol 1942; 47:97-159.
-
Schunke GB. The anatomy and development of the sacro-iliac joint in man. Anat Rec 1938; 72:313-330.
-
Bardeen CR. Studies of the development of the human skeleton. Am J Anat 1905; 4:265-302.
-
Budorick NE, Pretorius DH, Graft MR, Lou KV. Ossification of the fetal spine. Radiology 1991; 181:561-565.[Abstract/Free Full Text]
-
Dennis MA, Drose JA, Pretorius DH, Manco-Johnson ML. Normal fetal sacrum simulating spina bifida: "pseudodysraphism.". Radiology 1985; 155:751-754.[Abstract/Free Full Text]
-
Sauerbrei EE, Toi A. The fetal spine. In: Rumack CM, Wilson SR, Charboneau JW, eds. Diagnostic ultrasound. St Louis, Mo: Mosby, 1991; 895-910.
-
Kliewer MA, Hertzberg BS, George P, et al. Fetal bifid sacrum artifact: normal developmental anatomy simulating malformation. Radiology 1995; 195:673-676.[Abstract/Free Full Text]
-
Gamba JL, Bowie JD, Dodson WC, Hedlund LW. Accuracy of ultrasound in fetal femur length determination: ultrasound phantom study. Invest Radiol 1985; 20:316-323.[Medline]
-
Gabbe SG. Congenital malformation in infants of diabetic mothers. Obstet Gynecol Surv 1977; 32:125-132.[Medline]
-
Mills JI. Malformations in infants of diabetic mothers. Teratology 1982; 25:385-394.[Medline]
This article has been cited by other articles:
|
|
|
|
 
M. A. Kliewer, B. S. Hertzberg, K. S. Freed, P. J. McNally, and D. M. DeLong
The Anterior Iliac Separation: Alternative Index for Pelvic Morphometry in Fetuses with Down Syndrome
Am. J. Roentgenol.,
April 1, 2001;
176(4):
1003 - 1007.
[Abstract]
[Full Text]
[PDF]
|
|
|
TOP
Abstract
Introduction
