| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Genitourinary Imaging |
1 From the Institute of Diagnostic Radiology (T.M.K., S.C.A.M., G.E., K.T., B.M., R.A.K.H.) and Department of Obstetrics (A.R., R.H.), University Hospital, Zurich, Switzerland; and Department of Biostatistics, University of Zurich, Switzerland (B.S.). Received October 10, 2001; revision requested December 20; final revision received August 22, 2002; accepted August 27. Supported in part by a grant from the EMDO-Foundation, Switzerland. Address correspondence to R.A.K.H., Department of Radiology, Kantonsspital Baden, CH-5404 Baden, Switzerland (e-mail: rahel.kubik@ksb.ch).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
MATERIALS AND METHODS: MR pelvimetric data were retrospectively reviewed in 781 women (mean age, 28.9 years ± 5.2 [SD]) clinically referred, and the data were correlated to obstetric history to derive normative values. Five observers assessed results of multiple MR pelvimetric examinations in 10 female volunteers (mean age, 34.7 years ± 6.0; eight nullipara, two primipara) to provide data for measurement error analysis.
RESULTS: All values were higher in the spontaneous vaginal delivery subgroup (n = 100) and lower in the cesarean section or vacuum extraction subgroup (n = 130; intersubgroup difference, P < .001, Mann-Whitney U test). Pelvimetric parameters in the group undergoing spontaneous vaginal delivery were as follows: obstetric conjugate, 121.7 mm ± 8.6; interspinous distance, 112.3 mm ± 7.9; intertuberous distance, 120.6 mm ± 11.3; transverse diameter, 129.5 mm ± 8.7; and sagittal outlet, 115.8 mm ± 9.9. In the volunteer study, intraobserver, interobserver, and intraindividual reliabilities were high for the obstetric conjugate (0.94–0.96), interspinous distance (0.92–0.95), and transverse diameter (0.95–0.98) but low for intertuberous distance (0.64–0.87) and sagittal outlet (0.66–0.85).
CONCLUSION: Pelvimetric dimensions are smaller in women undergoing cesarean section or vacuum extraction than they are in those delivering vaginally. The pelvimetric parameters associated with the largest measurement errors are intertuberous distance and sagittal outlet.
© RSNA, 2003
Index terms: Pelvis, measurement, 44.92 • Pelvis, MR, 44.121411, 44.121412, 44.121416, 44.92 • Pregnancy, MR, 44.121411, 44.121412, 44.121416, 44.92
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
Whereas prenatal x-ray exposure has been associated with an increased risk of childhood cancer (10–12), findings of the numerous studies of MR imaging in human pregnancy have to date shown no experimental or clinical evidence of fetal harm (13,14). Its disadvantages are higher cost than those associated with conventional techniques, limited but increasing availability, and longer but shortening examination time. A substantial contraindication is claustrophobia; other contraindications (eg, pacemakers and metallic splinters) are comparatively rare in the obstetric population (15).
Conventional radiographic pelvimetry has largely provided the reference values available in the literature for comparison with the MR imaging data (16–20). To our knowledge, the largest MR imaging studies reported in the literature have included 53 women postpartum (3) and 52 women suspected of having fetal-pelvic disproportion (6). In the literature, to our knowledge, there are no reference values based on MR imaging measurements in a large study population, despite the fact that pelvimetric differences of just a few millimeters could have an important bearing on obstetric decision making, especially if we knew which parameters have the greatest clinical importance and which measurements are intrinsically the most reliable. Our purpose was to establish obstetric MR pelvimetric reference values in a large study population and stratify them according to delivery modality and to determine the intra- and interobserver error and intraindividual variability of MR pelvimetric assessment in volunteers.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
If a woman had given birth more than once in our institution, we chose the delivery that was temporally associated with MR pelvimetry. We excluded women who were referred by external institutions and did not deliver in our hospital, as we could not correlate their obstetric outcome with the MR imaging data.
These criteria yielded a study population of 781 women (mean age, 28.9 years ± 5.2 [SD]) available for further analysis. The study was performed according to our institutional review board guidelines. For this retrospective data analysis, no informed consent was needed.
All pelvimetric and obstetric data were obtained from the respective databases, supplemented as appropriate with data from the medical records. Data were collected by three coauthors (T.M.K., A.R., G.E.). Age, weight, height, and parity were recorded; multiple pregnancies were identified; and maternal status—pregnant or not pregnant—at pelvimetry was noted. Delivery date and modality—spontaneous delivery versus cesarean section or assisted delivery (eg, forceps or vacuum extraction)—were recorded, together with the reasons for intervention. A commercially available spreadsheet (Excel 2000; Microsoft) was used to correlate the radiologic and obstetric data.
Two subgroups were identified in the total study population: spontaneous delivery (subgroup 1: n = 100) and assisted delivery or cesarean section for fetal-pelvic disproportion (subgroup 2: n = 130). Fetal-pelvic disproportion was defined as arrest of labor for more than 2 hours despite adequate uterine contractions. Multiple pregnancies were excluded from the subgroup analysis. The remaining 551 cases included multiple pregnancies and pregnancies culminating in primary (ie, elective or due to factors such as twin pregnancy or breech presentation) or secondary (ie, after a trial of labor) cesarean section for reasons other than fetal-pelvic disproportion (eg, elective or emergency cesarean section). This subgroup of 551 was not evaluated separately, but it was included in the total population of 781 for demographic information and maternal height and weight analysis.
MR pelvimetry was performed with the patient in the supine position in a 1.5-T system (Signa Horizon EZ or CV/i; GE Medical Systems, Milwaukee, Wis) with use of the body coil. The protocol changed during the study in response to evidence in the literature (5,6,21) favoring gradient-echo over spin-echo sequences because of decreased examination time and energy deposition in tissue. Thus from May 1996 through February 1997, T1-weighted spin-echo sequences with the following parameters were used: 300–400/8 (repetition time msec/echo time msec), field of view of 32 cm, 256 x 192–256 matrix, two signals acquired, section thickness of 6–1.6 mm, section gap of 7–0 mm, total acquisition time of 7–9 minutes. Starting in March 1997, spin-echo sequences were replaced by T1-weighted fast spoiled-gradient-echo (FSPGR) sequences with the following parameters: 150/1.6, field of view of 32 cm, 256 x 192–256 matrix, two signals acquired, section thickness of 6–1.6 mm, and section gap of 7–0 mm. The cutaneous marker was positioned 2 cm above the superior edge of the symphysis to obviate additional localizing imaging.
Pelvimetric parameters were obtained by using midsagittal, transverse, and oblique sections of the pelvis. The midsagittal section included the obstetric conjugate, from the sacral promontory to the top of the symphysis pubis, and the sagittal outlet, from the end of the sacrum to the bottom of the inner cortex of the symphysis. The transverse section included the interspinous distance, or the narrowest distance between the ischial spines, and the intertuberous distance, or the widest distance between the ischial tuberosities. The oblique section, in a plane through the symphysis and the promontorium, included the transverse diameter, the largest transverse distance of the pelvis.
All measurements were made by a radiology suite technologist on a console (Sun Microsystems, Santa Clara, Calif) with the same software package (Advantage Network 3.1, GE Medical Systems). All technologists had been trained to make the measurements and were supervised by a radiologist.
Statistical Analysis: Retrospective Assessment
Continuous variables were presented as means and SDs. Pelvimetric measurements were compared between the two subgroups by using the Mann-Whitney U test. Further analysis focused on subgroup 1, spontaneous deliveries, as being most representative of a healthy population. The Spearman rank correlation was used to analyze the relationship to maternal height, weight, and parity. Measurements were also compared by using the Mann-Whitney U test between women who were pregnant and those postpartum and between the two MR imaging sequences, spin echo versus FSPGR. A P value less than .01 was used in all tests with Bonferroni correction for the five parameters. All analyses were performed with commercially avaliable software (StatView 5.0.1; SAS Institute, Cary, NC).
Volunteer Study
In addition to the retrospective data analyses, a prospective volunteer study was performed to determine intra- and interobserver error and intraindividual variability in MR pelvimetric assessment. MR pelvimetry was performed in 10 female nonpregnant (eight nullipara and two primipara—one after vaginal delivery, one after cesarian section) volunteers (mean age, 34.7 years ± 6.0; age range, 27–45 years) who provided their written consent after being informed about the MR imaging procedure and its contraindications and potential adverse effects. The protocol was approved by our institutional review board.
MR pelvimetry was performed as described for the patient population, but by using only FSPGR sequences and the parameters previously described, with a section thickness of 7 mm and a gap of 0 mm. Total acquisition time was 3–5 minutes. Images were analyzed by using the same hardware and software as described for the patient population, and the same pelvimetric measurements were obtained.
Each volunteer underwent imaging five times in succession in a single session, stretching their legs after each round of imaging before being repositioned on the MR imaging table. In single-assessment sessions, each of the five observers—one radiology suite technologist (K.T.), two residents experienced in MR pelvimetry (T.M.K., S.C.A.M.), and two newly trained physicians (G.E., A.R.)—then measured the five pelvimetric parameters from each imaging session, starting with the first one from volunteers 1–10 and ending with the last one from volunteers 1–10. This order was chosen to obviate the learning effect of assessing results from all five imaging sessions from the same volunteer directly one after the other. Each observer then repeated the assessment session at least 24 hours later. As a result, 500 pelvimetric measurements were available for analysis—five parameters measured twice from five imaging sessions in 10 volunteers.
Statistical Analysis: Volunteer Assessment
Mean values were used for all measurements. Inter- and intraobserver, as well as intraindividual, reliabilities were calculated by using random four-way analysis of variance. To improve the presentation, reliability values were subtracted from 1, representing the part of the variance influenced by the corresponding source of error, and tabulated with a commercially available spreadsheet (Excel 2000; Microsoft). Interobserver errors were demonstrated by displaying the measurements for each volunteer in a cell line chart. All analyses were performed with commercially available software (StatView 5.0.1; SAS Institute).
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
|
The remaining 513 women, 495 with primary cesarean sections and 18 with vacuum extractions, could not be placed into either subgroup for the following reasons: emergency indication for intervention (eg, preeclampsia), normal labor not attempted because the patient was suspected of having fetal-pelvic disproportion, breech presentation, patient request for elective cesarean section, or the impossibility of retrospectively identifying the indication for intervention.
All five sets of pelvimetric data (Fig 1) were higher in subgroup 1 than they were in subgroup 2 (P < .001) (Table 1). In subgroup 1, the only pelvimetric parameter to correlate with body weight was sagittal outlet (P = .001); obstetric conjugate (P < .001) and sagittal outlet (P = .001) correlated with height; no correlation was found between any parameter and parity. In the total study population (n = 743), all pelvimetric parameters correlated with weight and height (P < .001 in each case).
|
|
|
|
Volunteer Study
Pelvimetric results were as follows: obstetric conjugate, 119.3 mm ± 5.9 (range, 111–132 mm); interspinous distance, 113.0 mm ± 11.6 (range, 98–132 mm); intertuberous distance, 116.5 mm ± 8.4 (range, 103–128 mm); transverse diameter, 132.9 mm ± 9.7 (range, 113–144 mm); and sagittal outlet, 108.6 mm ± 5 (range, 97–115 mm). Interobserver reliability was as follows: obstetric conjugate, 0.94; interspinous distance, 0.92; intertuberous distance, 0.64; transverse diameter, 0.95; and sagittal outlet, 0.66. Intraobserver reliability was as follows: obstetric conjugate, 0.96; interspinous distance, 0.94; intertuberous distance, 0.83; transverse diameter, 0.97; and sagittal outlet, 0.83. Intraindividual reliability was as follows: obstetric conjugate, 0.95; interspinous distance, 0.95; intertuberous distance, 0.87; transverse diameter, 0.98; and sagittal outlet, 0.85. Reliabilities subtracted from 1 are plotted in Figure 2.
|
|
|
|
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
In previous attempts, small populations or unrepresentative inclusion criteria were used. In 1990, Pfammatter et al (3) published normative MR pelvimetric values derived from 53 women postpartum, while in 1994 Wentz et al (6) published data from 52 cases, most with women suspected of having fetal-pelvic disproportion (Table 2). Neither study has attained consensus recognition in the obstetric decision-making setting.
|
We had two options for our study. The first was a prospective study in a healthy population in which strictly defined inclusion criteria and an identical MR imaging protocol were to be used. Although methodologically ideal, this option, for logistic reasons, would not have yielded normative values in a large enough population over an acceptable time frame. We therefore opted for a retrospective analysis of clinical referrals for MR pelvimetry stratified for obstetric outcome, mindful of the fact that, because obstetricians were often liberal in their reasons for referral, our database, if sufficiently large, would include a sizable proportion of healthy women. We were able to conduct a separate statistical analysis in the uncomplicated spontaneous delivery subgroup (n = 100), which can be assumed to approximate the values in healthy women of childbearing age. It is in this subgroup, rather than in the total study population, that the burden of our study’s demonstration lies.
Because our study results show significant differences in all pelvic dimensions between the spontaneous delivery and fetal-pelvic disproportion subgroups, they document a commonsense assumption and also confirm and expand the findings of earlier investigators (Table 2). In 1994, Berger et al (25) found smaller pelvic dimensions in women with cephalopelvic disproportion (n = 13) than those in women delivering vaginally (n = 17), although only the difference in interspinous distance was significant. In 1997, Spörri et al (8) confirmed significantly lower interspinous distance and obstetric conjugate values in women with cephalopelvic disproportion (n = 28) versus those in a vaginal delivery control group (n = 10).
The hormonal environment of pregnancy has been postulated to increase ligament laxity primarily by means of increased serum estrogen levels in the third trimester and relaxin, a nonsteroidal polypeptide hormone, which circulates at particularly high levels in the first half of pregnancy (26–30). A clinically relevant increase in laxity has been reported in the third trimester in association with an increased risk of peripheral joint injury (30,31). However, physiologic laxity of symphyseal soft tissue is minor. Even during active expulsion of the fetal head through the pelvic inlet and past the ischial spine, ultrasonography (US) shows only minimal symphyseal distention (31,32).
Our study results can be used to confirm these findings because they show no pelvimetric differences between pregnant and nonpregnant women, nor any correlation between pelvimetric profile and parity. Our data can also be used to confirm the virtual absence of significant differences in pelvimetric data, except for transverse diameter, between spin-echo and FSPGR sequences already reported in smaller populations (6,21).
Our concern was not only to establish reference pelvimetric values but also to offer statistical qualification of the resultant data. If obstetricians are to put their faith in millimeter differences, they need to know which pelvic parameters are statistically the most reliable. Again, the data previously available could not be regarded as having a sufficiently solid statistical or population basis (6,33). Results from our volunteer study, incorporating five observers and 500 pelvimetric measurements, can be used to confirm that interobserver error is higher than intraobserver error, which in turn is higher than intraindividual variability, across all parameters except obstetric conjugate.
The greatest variation was in intertuberous distance and sagittal outlet. All observers remarked that intertuberous distance, the widest distance between the ischial tuberosities, was the most difficult parameter to define. A section cephalic to the image with the superior border of the inferior pubic ramus was selected in an attempt to measure the widest distance between these anatomic reference structures. However, since the ischial tuberosity is not a small structure but has a relatively wide outer contour, no precise measurement point could be chosen. The same difficulty applies in conventional radiography (34). Individual component analysis of the variance estimation was used to identify observer variance as the prime culprit. This could account for the intertuberous distance curve profiles (Fig 3d) in that each observer measured intertuberous distance according to individual interpretation of the criteria but did not change criteria between examinations.
The sagittal outlet measurement criteria, in contrast, are clearly defined, but the difficulty in identifying the junction between sacrum and coccyx is a potential source of error (5). Variance component analysis was used to identify observer-volunteer interaction as primarily responsible. Once an observer decided where the sacrococcygeal joint was in a given volunteer, the observer appeared to keep to that decision for all subsequent measurements in that volunteer. For example, one observer found higher sagittal outlet values in volunteers 1–4 than did all the other observers, but a lower sagittal outlet in volunteer 5 (Fig 3e).
Sagittal outlet measurements also showed the highest intraindividual variability, possibly because of the flexibility of the sacrococcygeal joint: The position of the coccyx relative to the sacrum may vary between MR imaging examinations and result in a shift of reference point. No such movement effects, on the other hand, account for the high intraindividual variability in intertuberous distance. A possible explanation is that different parts of the relatively widely contoured ischial tuberosity were mapped, depending on the volunteer’s position, and used as a reference point.
The clinical purpose of our study was to provide obstetricians with guidance in predicting fetal-pelvic disproportion, particularly in women at risk (eg, those with a history of prolonged labor and/or cesarean section). Long labor terminating in cesarean section is associated not only with maternal pain and discomfort but also with an increase in endomyometritis; amniotic infection; prolonged hospital stay; and higher risk for the newborn, as shown by a gradual decrease in fetal pH (1). Although opinion on the effect of pelvimetry on clinical decision making is divided (35,36), MR pelvimetry can only add to the obstetrician’s arsenal by providing information about a crucial factor in the mechanism of labor. Other factors are also clinically relevant, such as fetal biometric data, the position and deformability of the fetal head, and the strength of uterine contractions. When brought together with the sum of other information at the clinician’s disposal, MR pelvimetry has allowed better selection of the delivery route, with a significantly lower rate for emergency cesarean section (7).
Refinements of the MR imaging contribution to the management of labor include the fetal-pelvic index—a combination of pelvimetric and fetal biometric data—proposed in 1992 for predicting fetal-pelvic disproportion (37). A newer approach is to obtain a three-dimensional view of the pelvis and fetal head and calculate the capacities of the pelvic inlet and the middle of the pelvis from the pelvimetric data and relate them to fetal head volume; the results show good correlation between fetal-pelvic disproportion and failure to progress (8). MR imaging is perfectly capable of providing the same fetal biometric data (eg, biparietal and fronto-occipital diameter or shoulder measurements) as is US during the same imaging session (37–39).
In summary, our study offers what we believe to be the first reference MR pelvimetric values stratified according to obstetric outcome in a substantial population. We have confirmed that pelvimetric dimensions are significantly smaller in women undergoing cesarean section and assisted delivery than in those delivering vaginally, and we have shown that the pelvimetric parameters associated with the largest intra- and interobserver error and intraindividual variability are the intertuberous distance and sagittal outlet: Obstetric decision makers should therefore treat them with caution.
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
Author contributions: Guarantors of integrity of entire study, T.M.K., A.R., B.S., R.A.K.H.; study concepts and design, T.M.K., R.A.K.H., A.R.; literature research, T.M.K., A.R.; clinical studies, T.M.K., A.R., S.C.A.M., G.E., K.T., R.A.K.H.; data acquisition, T.M.K., A.R., S.C.A.M., G.E., K.T., R.A.K.H.; data analysis/interpretation, T.M.K., A.R., B.S., R.A.K.H.; statistical analysis, B.S., T.M.K., A.R.; manuscript preparation and definition of intellectual content, T.M.K., R.A.K.H.; manuscript editing, B.S., A.R., R.H., R.A.K.H.; manuscript revision/review and final version approval, all authors.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  V. L. Handa, H. K. Pannu, S. Siddique, R. Gutman, J. VanRooyen, and G. Cundiff Architectural Differences in the Bony Pelvis of Women With and Without Pelvic Floor Disorders Obstet. Gynecol., December 1, 2003; 102(6): 1283 - 1290. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
