Zaretsky, Michael V. MD1; Alexander, James M. MD1; McIntire, Donald D. PhD1; Hatab, Mustapha R. PhD2; Twickler, Diane M. MD2; Leveno, Kenneth J. MD2
Historically, pelvimetry was used to identify those women at greatest risk of cephalopelvic disproportion, for whom cesarean should be considered before undergoing labor. Baudelocque is said to have laid the early foundations of pelvimetry in 1775.1 Using external and, later, internal calipers as well as digital examination, pelvic dimensions were estimated. These early pelvimetric techniques were able to identify contracted pelvises, but they were limited by the imprecision of the measurements. Furthermore, the lack of safety of cesarean delivery in cases when contracted pelvis was identified resulted in very high mortality rates, thus limiting the usefulness of pelvimetry. The advent of the X-ray in 1895 promised an improvement over previous techniques, primarily because of the ability to accurately determine the transverse dimension of the pelvic inlet and the midpelvis as well as the ability to accurately measure the obstetric conjugative, measurements only indirectly obtained with digital examination or calipers.2 In addition, radiography was more practical to use than the calipers (especially the sometimes painful internal caliper) and produced more consistent results among examiners.1 This newfound ability to determine more exact pelvic dimensions led to widespread clinical use and multiple studies of its ability to predict abnormal pelvis anatomy that would preclude vaginal delivery.3–9 Although able to identify those women at greatest risk for cephalopelvic disproportion and cesarean delivery, X-ray pelvimetry did not do so with any precision. Its high cost, the exposure of the fetus to ionizing radiation, and the high false-positive rate currently limits its use to selected cases.
Newer imaging modalities, such as magnetic resonance imaging (MRI), have now been applied to pelvimetry in an effort to improve upon older techniques based on X-rays.3,10–14 Potential advantages of MRI include high quality imaging that does not require correction for image distortion, no exposure of the fetus to ionizing radiation, and calculation of the volume of the fetal head and the maternal pelvis.10 For this study, we sought to study whether MRI has the ability to identify those women at risk for dystocia and, thus, requiring cesarean delivery. We were particularly interested in the use of MRI to measure fetal head and maternal pelvic volumes and whether these novel measurements could improve the prediction of cephalopelvic disproportion leading to cesarean delivery for labor dystocia.
MATERIALS AND METHODS
At Parkland Memorial Hospital, women whose pregnancies reach 41 completed weeks are referred to an obstetrics complications clinic staffed by faculty and fellows in maternal–fetal medicine as well as research personnel. Women who do not labor spontaneously receive an induction of labor at 42 weeks. Information about each patient’s pregnancy, labor course, and neonatal outcomes are prospectively collected and entered into a computerized database. This study was approved by the Institutional Review Board at the University of Texas Southwestern Medical Center. Nulliparous women scheduled for induction were asked to participate in a study of MRI pelvimetry. Our goal was to determine whether MRI pelvimetry was able to identify those women who were determined to have dystocia requiring cesarean delivery.
Nulliparous women were seen in clinic, evaluated, and if otherwise uncomplicated, scheduled for induction at 42 completed weeks (42 0/7). Those women with an immediate indication for delivery were excluded from consideration for the study, as well as women with a prior cesarean, hypertension, insulin-dependent diabetes, known fetal anomalies, and stillbirths. In addition, women weighing more than 360 pounds were excluded from the study because of the weight limit of the MRI table. Candidates were approached and consent for the study was obtained by the primary investigator (M.V.Z.). Labor dystocia in the first stage of labor was diagnosed in women whose cervix dilated to at least 4 cm, who experienced no cervical change over 2–3 hours in the presence of at least 200 Montevideo units of uterine activity, and who underwent cesarean delivery. Dystocia in the second stage of labor was defined as failure of descent requiring cesarean delivery despite a second stage of at least 2 or 3 hours in women with a labor epidural.15 The practitioners managing labor were blinded to the MRI results.
After consent was obtained, the patients were escorted to the imaging suite where they underwent an MRI protocol consisting of 2 single-shot fast spin echo T2-weighted scans (TE = 60, 44 cm field of view [FOV], 512 × 256 matrix). A 1.5 Tesla GE Signa magnet (General Electric Medical Systems, Milwaukee, WI) and a torso coil were used on all but 3 patients whose large size necessitated the use of the body coil. The first sequence was a 7-mm acquisition without gap, aligned axial to the maternal uterus, and this incorporated the entire gravid uterus and pelvis. There were, on average, 40–50 images. The second sequence was a 4-mm acquisition, aligned axial to the maternal pelvis at an angle parallel to the obstetric conjugate. This sequence also averaged 40–50 images, which included the entire fetal head and the maternal pelvis.
From these two 90-second acquisitions, a single investigator (M.R.H.), an MRI physicist, performed all biometric and volumetric analyses. Using a 3-dimensional reconstruction workstation (GE Advantage Windows 4.1, General Electric Medical Systems), the investigator selected optimal orientations for obtaining 2-dimensional measurements of abdominal circumference and maternal pelvic interspinal diameter from the 7-mm acquisition. All other measurements, including fetal biparietal diameter (BPD) and head circumference, as well as maternal pelvimetry measurements of the obstetrical conjugate, transverse inlet, and sagittal midpelvis, were obtained from the 4-mm acquisition.
Fetal head volume (Fig. 1) and maternal pelvic volumes were calculated from the 4-mm acquisitions. Because MR pelvic volume measurements have not been previously described in the literature, we defined MR pelvic volume from historic pelvimetry descriptions of the inlet to the midpelvis. Because MR provides detail of the soft tissue components of the pelvis volumes, the bony and soft tissue pelvic volumes were attempted. Both volumes were arbitrarily defined as bordered superiorly by the inlet and inferiorly and laterally by either bony or muscular landmarks. The obstetric conjugate view was optimized in the sagittal plane by multiplanar reformatting on the postprocessing workstation; contiguous areas parallel to the obstetric conjugate were manually drawn, defined by the borders of the bony pelvis of each image until the midpelvic plane was reached (Fig. 2). This was repeated a second time, with the borders defined by the pelvic musculature rather than bone (Fig. 3).
The pelvic type was classified by using the technique described by Caldwell and Molloy.3 Single fetal and maternal pelvic measurements, as well as ratios of the two, were also compared. In addition, several previously described pelvimetric techniques and formulas were used and included the following: The method described by Mengert5 is based on the products of the transverse and sagittal diameters for both the inlet and the midpelvis, and pelvic contraction is defined as an area of less than 123 cm for the inlet or less than 106 cm for the midpelvis. The method described by Abitbol et al10 compares the smallest pelvic diameter (either the sagittal diameter of the inlet or the transverse diameter of the midpelvis) with the fetal biparietal diameter and indicates how much wider the smallest pelvic diameter is than the biparietal diameter. A positive cephalopelvic disproportion index is present if the pelvic diameter is less than 9 mm wider than the biparietal diameter.
The method described by Friedman and Taylor14 evaluates cephalopelvic disproportion by comparing fetal head volume with maternal pelvic inlet and midpelvis capacity. The pelvic capacities were calculated by using the formula: (π × d3)/6, where d equals the appropriate sagittal or transverse diameter of the respective plane, always taking the shorter of the two. Cephalopelvic disproportion is present if the fetal head volume is more than 50 cm3 larger than the smallest inlet capacity or the fetal head volume is more than 200 cm3 larger than the interspinal capacity or both. According to the method described by Spörri et al,11 cephalopelvic disproportion is present if the fetal head volume is greater than the smallest pelvic inlet or midpelvis capacity or both. The fetal-pelvic index described by Morgan et al12,13 compares the fetal head and abdominal circumferences with the respective inlet and midpelvis circumferences. The pelvic inlet and midpelvis circumferences were calculated according to the formula: (sagittal diameter + transverse diameter) × π/2. A positive fetal-pelvic index is present if the circumference of the fetal head or abdomen is larger than that of the maternal pelvic inlet or midpelvis, which indicates cephalopelvic disproportion or abdominal-pelvic disproportion.
Statistics used in the analyses included Student t test and Pearson χ2 for comparing results between the dystocia and nondystocia groups. The area under the curve (AUC) for the receiver operating characteristic curve was used as a measure of adequacy of prediction.16 The AUC is computed as the proportion of all possible pairs of subjects with predicted classification probabilities appropriately ordered according to the observed outcome. For example, if patient A is diagnosed with dystocia and patient B not diagnosed with dystocia, a correct classification would have the predicted probability of dystocia for patient A exceed that of patient B. A 95% confidence interval is presented for the AUC using components of the Mann-Whitney statistic for an estimate of the variance. Statistical analyses were calculated using SAS 8.0 (SAS Institute, Cary, NC), and statistical significance is assumed for significance levels less than .05. Statistical power was calculated by assuming an estimated prevalence of dystocia of 0.2 at the mean of the prediction variable and an odds ratio of 2.0 for a one standard deviation increase in the prediction variable. To achieve a statistical power of 0.8 for a 2-sided test of size 0.05, a sample size of 100 was required.
From July 7, 2003, to April 19, 2004, 120 total patients were gave their consent for the study. Of these, 13 women did not complete the MRI imaging. Of the remaining 107, 6 patients delivered at other hospitals and delivery information was not available. The remaining 101 patients were analyzed, and the results are reported here. Of the 101 patients, 63 women underwent vaginal delivery and 38 women underwent cesarean delivery. Of the cesarean deliveries, 22 were for dystocia and 16 were for fetal distress.
Table 1 shows demographic characteristics of the dystocia and on dystocia group. There were no significant demographic differences in the dystocia and no dystocia group.
Table 2 shows a comparison of selected fetal and pelvic measurements as well as MRI derived pelvic volume in cases with and without dystocia. The obstetric conjugate, mid pelvis anterior to posterior diameter and interspinous diameter were significantly less in the dystocia group. No other fetal or pelvis measurements were significantly associated with cesarean for dystocia.
Table 3 shows a comparison between fetal-to-pelvic measurement ratios, including MRI-calculated head and pelvis volumes in women with and those without dystocia. The most significant associations were seen in the ratios that included midpelvic measurements. Table 4 shows a comparison of previously described methods of estimating pelvic capacity in women with and in those without dystocia. In addition, MRI-derived fetal head volume is compared with MRI-derived pelvic volumes. Regardless of the technique used, women who developed dystocia had significantly different measurements from those of women without dystocia. The strongest association was seen when using Mengert’s midpelvis capacity.
Receiver operator characteristic curves were generated for those parameters from Tables 1, 2, and 3 that were significantly associated with dystocia (Fig. 4). The area under the curves ranged from 0.6 to 0.8. The most significant association with dystocia was seen in those techniques that included midpelvis assessment, with the greatest significance seen using Mengert’s midpelvis estimates. The P value for this association was < .001. Volume comparisons derived from MRI showed an AUC of 0.60 for the fetal head volume to bony pelvis and 0.64 for the fetal head volume to soft tissue pelvic volume.
As described above, the primary analysis focused on the ability of MRI to predict labor dystocia requiring operative delivery. A subanalysis was performed to examine the effect on the analysis of cesarean performed for fetal distress. In the primary analysis, these women were grouped in the no-dystocia group. In our subanalysis we excluded these women and compared cesarean delivery for dystocia with vaginal delivery. Our findings were unaffected. We then grouped all cesareans together, comparing vaginal delivery with cesarean delivery regardless of the indication. In this subanalysis we saw less of an association of MR pelvimetry with cesarean delivery, suggesting that MR has a greater ability to predict cesarean for dystocia than cesarean overall.
In this study of MRI pelvimetry, we found that MRI-derived measurements of the fetus and the maternal pelvis are significantly associated with labor dystocia. Several different individual measurements, as well as fetal-to-maternal pelvic ratios, and formulas were shown to be associated with labor dystocia requiring cesarean delivery, especially those measurements that took the midpelvis into account. Indeed, the most significant association with dystocia was seen with the anterior-posterior midpelvis diameter, the interspinous diameter, and the midpelvis capacity as determined by the technique described by Mengert.5 In addition to calculating the above dimensions, we were able to calculate pelvic volume based on both bony and soft tissue measures. The measured fetal head-to-pelvis volume ratio was comparable with, but did not improve upon, standard X-ray pelvimetry, even taking into consideration the soft tissue of the pelvis. Taken together, these results demonstrate an ability of MRI pelvimetry to identify women at risk for cesarean for dystocia, but none of the pelvic measurements or formulas studied showed an improvement over previously described pelvimetry techniques.
Rising cesarean rates, along with newer imaging modalities, have prompted renewed interest in pelvimetry. Sonography and magnetic resonance imaging have potential advantages over X-ray, which include an improved ability to assess fetal parameters without exposing the fetus to ionizing radiation. Morgan and Thurnau12,13 described the fetal pelvic index in 1986. This technique combined sonographically derived fetal head and abdominal measurements with conventional radiographic pelvimetry. They found the pelvic fetal index to be sensitive and specific, with an overall prediction value of 94–100%. Subsequent study of this technique, however, yielded a much lower sensitivity and specificity and an overall prediction value of only 65%, thus limiting its use.17 Using a similar, yet simpler approach, Abitbol et al10 studied women attempting a vaginal birth after cesarean (VBAC). They compared a sonographically derived BPD to the smallest pelvic diameter as determined by X-ray pelvimetry. This technique had an excellent specificity but poor sensitivity, resulting in an inability to predict those women who had a successful VBAC. It did, however, identify a small subset of women who had an extremely high chance of VBAC failure. Spörri and colleagues11 used MRI to evaluate the fetus and pelvis in 38 women at risk for dystocia. Using a variety of methodologies, they evaluated the ability of MRI-derived pelvimetry measurements to predict cephalopelvic disproportion. Although they were able to demonstrate correlation between MRI pelvimetry and dystocia, their results were not an improvement over previously reported techniques, and they could not recommend its routine use in clinical practice.
We embarked on this project with the hope that MRI and its unique ability to calculate fetal head and maternal pelvic volume would improve upon previously published pelvimetric techniques. Our results, however, are similar to those that have come before; pelvimetry identifies those women at risk for dysfunctional labor but cannot with accuracy predict those who will require cesarean delivery. This finding has been remarkably consistent across many studies, regardless of the pelvimetric technique and technology used and may be due to the inability of pelvimetry to take into account uterine function. Successful labor depends upon 3 interacting variables: the pelvis, fetus, and uterine activity. Our study and others before it have only assessed 2 of these variables: the fetus and pelvis. Including uterine function in the equation is problematic because it cannot be predicted before the onset of labor and is variable among pregnancies. Unless the variable of uterine activity and its effect on molding of the fetal head can be measured, it seems unlikely that further refinement of pelvimetric techniques will result in greater ability to predict failed labor, and those current recommendations limiting the routine use of pelvimetry in clinical practice are appropriate.
1. Baudelocque. L’art des accouchements. Nouvelle éd., 1789, t. l, 76–90.
2. Budin. Statistiques de la Maternité de Paris. L’Obstétrique, 1986, iii, 134. Photographie par les rayons × d’un bassin de Naegele. L’Obstetriqué, 1897, ii, 490.
3. Caldwell WE, Moloy HC. Anatomical variations in the female pelvis and their effect in labor with a suggested classification. Am J Obstet Gynecol 1933;26:479–505.
4. Ball RP. Roentgen pelvimetry and cephalometry. Surg Gynecol Obstet 1936;62:798–810.
5. Mengert WF. Estimation of pelvic capacity. JAMA 1948;138:169–74.
6. Steer CM. X-ray pelvimetry and the outcome of labor. Am J Obstet Gynecol 1958;76:118–31.
7. Hannah WJ. X-ray pelvimetry: a critical appraisal. Am J Obstet Gynecol 1965;91:333–41.
8. Parsons MT, Spellacy WN. Prospective randomized study of X-ray pelvimetry in the primigravida. Obstet Gynecol 1985;66:76–9.
9. Laube DW, Varner MW, Cruikshank DP. A prospective evaluation of x-ray pelvimetry. JAMA 1981;246:2187–8.
10. Abitbol MM, Taylor UB, Castillo I, Rochelson BL. The cephalopelvic disproportion index: combined fetal sonography and x-ray pelvimetry for early detection of cephalopelvic disproportion. J Reprod Med 1991;36:369–73.
11. Spörri S, Thoeny HC, Raio L, Lachat R, Vock P, Schneider H. MR imaging pelvimetry: a useful adjunct in the treatment of women at risk for dystocia. AJR Am J Roentgenol 2002;179:137–44.
12. Morgan MA, Thurnau GR, Fishburne JI Jr. The fetal pelvic index as an indicator of fetal pelvic disproportion: a preliminary report. Am J Obstet Gynecol 1986;155:608–13.
13. Morgan MA, Thurnau GR. Efficacy of the fetal-pelvic index in nulliparous women at high risk for fetal-pelvic disproportion. Am J Obstet Gynecol 1992;166:810–4.
14. Friedman EA, Taylor MB. A modified nomographic aid for x-ray cephalopelvimetry. Am J Obstet Gynecol 1969;105:1110–5.
15. Dystocia and augmentation of labor. ACOG Practice Bulletin No. 49. American College of Obstetricians and Gynecologists. Obstet Gynecol 2003;102:1445–54.
16. DeLong ER, DeLong DM, Clarke-Pearson DL. Comparing the areas under two or more correlated receiver operating characteristic curves: a nonparametric approach. Biometrics 1988;44:837–45.
17. Ferguson JE, Newberry YG, DeAngelis GA, Finnerty JJ, Agarwal S, Turkheimer E. The fetal-pelvic index has minimal utility in predicting fetal-pelvic disproportion. Am J Obstet Gynecol 1998;179:1186–92.
Figure. No caption available.
© 2005 The American College of Obstetricians and Gynecologists