First trimester uterine, placental and yolk sac haemodynamics in pre-eclampsia and preterm labour

K. Mäkikallio1, P. Jouppila and A. Tekay

Department of Obstetrics and Gynecology, University of Oulu, 90220 Oulu, Finland

1 To whom correspondence should be addressed. e-mail: kaarin.makikallio{at}oulu.fi


    Abstract
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
BACKGROUND: We hypothesized that impaired trophoblast invasion leads to umbilicoplacental blood flow disturbances that could be detected by Doppler ultrasonography during the first trimester of the pregnancy. METHODS: After successful fresh IVF or ICSI programme, 41 of 47 enrolled subjects were followed up every 1–2 weeks between weeks 6 and 11 of gestation. Ten patients who later developed pre-eclampsia and/or preterm labour formed the study group and the control group consisted of 31 uncomplicated IVF/ICSI pregnancies. Doppler parameters of uterine, spiral, intraplacental, chorionic, umbilical and yolk sac haemodynamics were assessed. RESULTS: At the week 8, the study group demonstrated higher (P < 0.05) maternal intraplacental resistance indices (RI) than the control group. A week later, yolk sac artery RI and umbilical artery mean velocity (Vmean) in the study group were lower (P < 0.05) compared to the control group. In late first trimester, increased (P < 0.01) velocities and RI were observed in chorionic arteries of the study group. During early pregnancy, no difference in uterine and spiral artery haemodynamics and in umbilical artery pulsatility index (PI) values was observed between the groups. CONCLUSIONS: Uterine and spiral artery RI and umbilical artery PI are unable to detect placental vascular disturbances during early pregnancy. Elevated intraplacental RI indicates increased maternal intraplacental impedance as early as week 8 of gestation. Decreased yolk sac artery RI and umbilical artery Vmean in the study group at week 9 of gestation were speculated to indicate hampered transition of blood supply from yolk sac to umbilical circulation, underlining the emphasized role of yolk sac function for the maintenance of pregnancy.

Key words: Doppler/early pregnancy/IVF/pre-eclampsia/preterm labour


    Introduction
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
The aetiology of pre-eclampsia is still unknown. In pre-eclampsia, as in the vast majority of pregnancies complicated by preterm labour, a vascular component plays an important role in the course of the disease (Hediger et al., 1995Go). Abnormal implantation and invasion of the trophoblastic tissue leading to structural and occlusive changes in the spiral arteries and underperfusion of placental circulation might be one plausible mechanism in the pathogenesis of pre-eclampsia. Placental histology has revealed similar findings in some cases of preterm labour (Pijnenborg et al., 1991Go; Salafia et al., 1995Go). Imbalance of vasoactive substances, increased cytokine concentrations, histopathological changes in placental vasculature and hereditary factors have been demonstrated in these conditions (Brosens et al., 1967Go; Walsh, 1985Go; Chesley and Cooper, 1986Go). Early recognition of these common perinatal problems could allow the use of preventive and therapeutic measures and thus benefit both the mother and the fetus.

Doppler ultrasonographic examinations of uterine and umbilical vasculature in the second and third trimesters have been used as a tool to predict the onset of pre-eclampsia and to describe placental function (Giles et al., 1985Go; Bower et al., 1993Go; Harrington et al., 1991Go, 1997a,b). The hypothesis behind the present study is that in pre-eclampsia and preterm labour without signs of chorioamnionitis, abnormal development of placental vasculature due to impaired trophoblast invasion leads to uterine and umbilicoplacental blood flow disturbances, which could be detected by Doppler ultrasonography during the first trimester of the pregnancy.


    Materials and methods
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
For this longitudinal and prospective study, 47 subjects were enrolled after successful fresh IVF or ICSI protocols in the University Hospital of Oulu. Five subjects were excluded due to spontaneous abortion between weeks 7 and 11 of gestation, and one because of tubal pregnancy. The study group consisted of 10 patients, who developed pre-eclampsia and/or preterm labour during their singleton pregnancies. The diagnostic criteria of pre-eclampsia followed the ACOG guidelines (American College of Obstetricians and Gynecologists, 1996Go). Preterm labour was defined as a delivery prior to 37 weeks of gestation without signs of clinical chorioamnionitis (fever, elevated CRP values, leukocytosis, fetal tachycardia). In the study group, three patients had severe pre-eclampsia and two had mild pre-eclampsia. Pre-eclampsia developed between weeks 29 and 33 of gestation, and in two cases Caesarean section was performed between weeks 33 and 35 of gestation due to severe maternal hypertension and headache. An additional five preterm labours in the study group occurred between weeks 25 and 35 of gestation (median week 33). The control group consisted of remaining 31 uncomplicated singleton IVF and ICSI pregnancies. Stimulation protocol and medication used for ovulation induction were similar in IVF and ICSI groups (Tomas et al., 1998Go). None of the patients had ovarian hyperstimulation syndrome. The perinatal data of the groups are presented in Table I. The local ethics committee approved the study protocol and all participants signed informed consent.


View this table:
[in this window]
[in a new window]
 
Table I. Perinatal data
 
A Sequoia System 512 (Acuson, USA) was used for the examinations. The operating frequencies of the transvaginal transducer (EV-8C4) using a curved linear array format with a footprint size of 29 mm were 5–8 MHz in B mode, 4–7 MHz in colour Doppler mode and 5 MHz in pulsed Doppler mode. The theoretically lowest measurable flow velocity with this equipment was 0.5 cm/s. The high pass filter was set at minimum. The acoustic output of the system was displayed using mechanical (MI) and thermal indices (TI). The maximum theoretical MI values were <1.0 in B mode, colour Doppler mode and pulsed Doppler mode. The maximum achievable soft tissue thermal index and bone thermal index (TIS/TIB) and related ultrasonic power values passing through a 1 cm window in auto-scanning mode were 1.9 and 78 mW in B mode, 1.8 and 60 mW in colour Doppler mode and 1.4 and 57 mW in pulsed Doppler mode. In each ultrasonographic examination, MI and TI values were kept <0.7 and <0.4 respectively. All the subjects were examined at weeks 6, 8, 9 and 11 of gestation. Gestational age was confirmed by the embryo transfer date and fetal crown–rump length (CRL) (Hadlock et al., 1992Go). The subjects were examined in the lithotomy position between 06:30 and 09:30 by a single observer (K.M.).

After colour Doppler identification of uterine, spiral, intraplacental, yolk sac, umbilical and chorionic arteries, blood velocity waveforms were obtained by pulsed Doppler ultrasonography. The size of the vessel under investigation determined the width of the pulsed Doppler sample volume gate (1–3 mm). The insonation angle was minimized <30° in every measurement and several vessels were assessed to determine the highest possible Doppler signals in small calibre vessels. Blood velocity waveforms of uterine arteries were recorded at the level of the cervicocorporeal junction of the uterus. The location of the placenta was determined at week 11 of gestation, and placental uterine artery parameters were used for comparisons between the groups. If the placenta was located in the midline position, the mean values of both arteries were used. Spiral artery blood velocity waveforms were recorded from the intensively coloured retrochorional area towards the insertion of the umbilical cord. The pulsatile blood velocity waveforms concordant with maternal heart rate and shown within the placenta were also investigated. For sake of the brevity, throughout the text these waveforms are referred to as intraplacental flow with maternal heart rate. Umbilical artery Doppler shift spectra were assessed near its placental insertion and blood velocity waveforms of fetal vessels from the chorionic surface of the placenta were obtained. Blood velocity waveforms of yolk sac artery were obtained from the area where the vitelline vessels entered the yolk sac. The maximum diameter of the yolk sac was measured. Fetal heart rate (FHR) was measured from umbilical artery blood velocity waveforms.

The examination was performed in 20 min at the most and the yolk sac was examined in 1–2 min. In the assessment of yolk sac flow parameters, the embryonic colour or pulsed Doppler ultrasonographic exposure was minimized. Measurements were made off-line using the ultrasound equipment’s own measurement package. Time-averaged maximum velocity (TAMXV), resistance index [RI = (peak systolic velocity – end diastolic velocity)/peak systolic velocity] of the examined arteries and pulsatility index [PI = (peak systolic velocity – end diastolic velocity)/TAMXV] of umbilical artery were measured from three consecutive cardiac cycles, and their mean values were used for further analysis. At weeks 9 and 11 of gestation, mean velocity (Vmean) of the umbilical artery blood flow was also calculated by the formula: Vmean= TVIxFHR, where time velocity integral (TVI) is calculated by planimetering the area under the Doppler spectrum. Intra-observer reproducibility of the Doppler measurements in uterine and spiral arteries was analysed in an additional 16 singleton pregnancies between weeks 8 and 11, each pregnancy being examined twice in a 30 min interval.

Statistical analysis between the groups was carried out by analysis of variance. Departure from a normal distribution was assessed by means of normal plots and the Kolmgorov–Smirnov test. If comparison was carried out between two groups and the data were normally distributed, Student’s t-test was used. If the data was not normally distributed the Mann–Whitney U-test was chosen. A level of P < 0.05 was considered statistically significant. All values are given as means (SD). Intraclass correlation coefficient (CC) and coefficient of variation (CV) were used to describe intra-observer reproducibility of the measurements.


    Results
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
The CRL values were within the normal reference range for gestational age calculated from the embryo transfer date and did not differ between the groups. During the study period, no difference in FHR was observed between the groups (Table II). Yolk sac size demonstrated no difference between the groups.


View this table:
[in this window]
[in a new window]
 
Table II. Changes in time-averaged maximum velocity (TAMXV) and resistance index (RI) values of uterine, spiral, intraplacental, chorionic and yolk sac arteries, and in umbilical artery pulsatility index (PI) values and mean velocity (Vmean) during early pregnancy
 
No significant differences between the study and the control groups were found in Doppler ultrasonographic parameters of uterine, spiral, intraplacental and chorionic blood velocity waveforms at the gestational week 6. At week 8 of gestation, intraplacental RI values in the study group were higher than those in the control group (P < 0.05) with no difference in TAMXV values (Figure 1, Table II). A week later, yolk sac artery RI values were significantly lower in the study group than in the control group (Figure 2) with no difference in yolk sac artery TAMXV values (Table II). At the same time, umbilical artery PI values did not differ between the groups. However, umbilical artery Vmean was significantly lower in the study group compared to the control group (Table II). At week 11 of gestation, TAMXV in fetal vessels at the placental surface (P < 0.001), as well as RI values (P < 0.01), were higher in the study group than in the control group (Figure 2, Table II). No differences in umbilical artery PI and Vmean values were observed at week 11 of gestation (Table II).



View larger version (11K):
[in this window]
[in a new window]
 
Figure 1. Resistance index (RI) values for intraplacental flow with maternal heart rate during early pregnancy. The values given are means. *P < 0.05 versus control group.

 


View larger version (16K):
[in this window]
[in a new window]
 
Figure 2. Resistance index values (RI) in chorionic and yolk sac circulation during early pregnancy. The values given are means. *P < 0.05 versus control group.

 
During the whole study period, uterine artery TAMXV and RI did not differ significantly between the groups. Neither did blood flow velocity waveform characteristics of spiral arteries show any differences between the groups. Longitudinal changes in uterine and spiral artery RI values are shown in Figure 3.



View larger version (14K):
[in this window]
[in a new window]
 
Figure 3. Resistance index (RI) values in uterine and spiral arteries during early pregnancy. The values given are means.

 
Intra-observer intra-class correlation coefficient of uterine artery TAMXV and RI varied between 0.87 and 0.90 and coefficient of variation between 12.1 and 18.8% respectively. Corresponding values in spiral artery varied between 0.84 and 0.96 and between 12.4 and 14.5% respectively.


    Discussion
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
The present study focuses on the question whether pre-eclampsia and preterm labour without chorioamnionitis is characterized by first trimester utero- and umbilicoplacental haemodynamic changes that could be detected by Doppler ultrasonography. Because of higher incidence of pre-eclampsia and preterm labour, IVF and ICSI pregnancies were selected for this longitudinal prospective study (Schenker and Ezra, 1994Go). The trophoblast migration to the intima of decidual and myometrial spiral arteries and replacement of musculo-elastic tissue of the vessel wall are essential for normal placental development (Brosens et al., 1967Go; Pijnenborg et al., 1983Go). In pre-eclampsia, these physiological vascular changes have been demonstrated to be restricted to the decidual segments of spiral arteries (Brosens et al., 1972Go; Khong et al., 1986Go). These fundamental disturbances in the development of the utero- and umbilicoplacental circulations occur already early during the first trimester of pregnancy. In cases of preterm labour caused by placental vascular disturbances, the histology of placental lesions is similar to those in pre-eclampsia, and the poor outcome of neonates has been speculated to be a consequence of this placental abnormality (Salafia et al., 1995Go; Germain et al., 1999Go). According to these anatomical findings, it has been postulated that failure in conversion of high-resistance spiral arteries to low-resistance uteroplacental arteries could be detected by non-invasive Doppler ultrasonography.

The results of this study demonstrate the chronological sequence of early haemodynamic abnormalities in utero- and umbilicoplacental circulations in pregnancies, which progress to pre-eclampsia and/or preterm labour. Spiral artery Doppler ultrasonographic parameters observed in the present study did not differ between the groups during the study period. At week 8 of gestation, intraplacental blood velocity waveforms with maternal heart rate demonstrated increased impedance in the study group. We speculate that these changes may indicate impaired vascular development in patients who later develop pre-eclampsia or preterm labour.

Early embryonic development occurs in a low oxygen environment and it has been postulated that the yolk sac has an essential role in transport of nutritive substances to the embryo (Gulbis et al., 1998Go; Burton et al., 1999Go). Previously, the abnormalities in yolk sac size have been connected to early pregnancy complications (Stampone et al., 1996Go). Our previous observations in normal pregnancies demonstrated that placenta replaces the yolk sac as a main source of nutritional supply after week 9 of gestation (Makikallio et al., 1999Go). In the present study, yolk sac artery RI values were significantly lower in complicated pregnancies at the week 9 of gestation. In addition, umbilical blood flow showed significantly lower mean velocity in the study group, while no difference in umbilical artery PI values were demonstrated between the study and the control groups. Mean velocity is directly proportional to volume blood flow, while PI describes the downstream impedance of the assessed vessel. Thus, according to these findings it seems that inadequate trophoblastic invasion hampers the transition of blood supply from yolk sac to umbilical arteries, and thus the importance of the yolk sac function during this time-period is underlined, explaining the decrease in impedance of yolk sac blood flow.

Increased impedance in fetal vessels at the placental surface, which may be indicative of impaired trophoblastic invasion, suggests that the developing fetus faces increased vascular resistance of the placenta already during the late first trimester. It may be possible to detect vessels at placental surface which will undergo occlusive changes later on during the pregnancy. At week 11 of gestation we did not find any difference in umbilical artery mean velocity or umbilical artery PI values. We speculate that with further involvement of deteriorated vessels on placental surface, raised resistance in umbilical artery haemodynamics becomes evident later during early second trimester as demonstrated by Harrington et al. (1997bGo).

The present study demonstrates changes in placental circulation that probably precede the changes in uterine artery blood velocity waveforms detected later during the second trimester of pregnancy. From a preventive and therapeutic point of view, early recognition of these pregnancy complications is essential. Previously, it has been demonstrated that routine uterine arterial Doppler ultrasonography at gestational weeks 18–24 in an unselected population predicts adverse obstetric outcome with sensitivity <44% and positive predictive value <27% (Mires et al., 1998Go). Harrington et al. (1997aGo,b) showed significant differences in uterine haemodynamics at 12–16 weeks of gestation between complicated and non-complicated pregnancies, demonstrating connection between early onset placental dysfunction and adverse obstetrical outcome. Doppler velocimetry of intraplacental, chorionic and yolk sac haemodynamics may be new tools to detect placental dysfunction during the ongoing trophoblast invasion when therapeutic measures would still be possible.

Intra-observer reproducibility of different Doppler ultrasonographic waveform parameters in this study was tested in uterine and spiral arteries. The uterine artery represented the most proximal artery in uterine circulation with high volume laminar blood flow, whereas turbulent blood flow pattern was typical of the distally located low-calibre spiral artery. Intra-observer reproducibility of spiral artery blood flow parameters was good. In uterine artery, indices describing impedance to blood flow were superior to velocity measurements in terms of both CC and CV. These findings confirm earlier reports that these parameters can be used reliably in the evaluation of uterine and uteroplacental circulation (Alcazar, 1997Go). The main problem in the measurement of absolute blood velocities is the angle between the vessel and the Doppler beam (Tessler et al., 1990Go). If the angle is kept <30°, the effect of the angle on the absolute velocities is minimal. In this study, special attention was paid to minimize the angle by achieving the highest Doppler shift signals, and in vessels with turbulent flow, several vessels were assessed to achieve the highest possible Doppler signals.

Concerns about the safety of first trimester ultrasonography have been raised and widely discussed in recent years. The time the embryo was exposed to pulsed Doppler ultrasonography during the yolk sac assessment was minimized and the lowest possible output was used. To further shorten the exposure time, off-line measurements were used. The on-screen display of thermal and mechanical indices enabled the observer to control continuously the ultrasound exposure. In animal models, no haematopoietic changes have been found after frequent ultrasound exposure of embryogenic structures during organogenesis (Tarantal and Hendrickx, 1990Go). Recently, ethics guidelines for research on the human fetus using Doppler ultrasonography in the first trimester have been published (Chervenak and McCullough, 1999Go).

In conclusion, uterine artery and spiral artery Doppler parameters are unable to detect placental vascular changes during the first trimester. The earliest signs suggesting impaired placental function can be detected in intraplacental blood velocity waveforms with maternal heart rate as indicated by increased impedance values. At week 9 of gestation, lower mean velocity in umbilical artery and decreased yolk sac RI values in complicated pregnancies indicate a hampered transition of blood supply from yolk sac to umbilical arteries. Finally, during late first trimester the impact of impaired placental function is demonstrated as increased impedance in fetal vessels at the placental surface, indicating the activation of fetal compensatory mechanisms in pregnancies progressing later to pre-eclampsia or preterm labour. Umbilical artery vascular impedance values do not reveal the differences in placental vascular development between normal and complicated pregnancies.


    References
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
Alcazar JL (1997) Intraobserver variability of pulsatility index measurements in three fetal vessels in the first trimester. J Clin Ultrasound 25,366–371.[CrossRef][ISI][Medline]

American College of Obstetricians and Gynecologists (1996) Hypertension in pregnancy. ACOG technical bulletin, no. 219. American College of Obstetricians and Gynecologists, Washington, DC.

Bower S, Schuchter K and Campbell S (1993) Doppler ultrasound screening as part of routine antenatal scanning: prediction of pre-eclampsia and intrauterine growth retardation. Br J Obstet Gynaecol 100,989–994.[ISI][Medline]

Brosens I, Robertson WB and Dixon HG (1967) The physiological response of the vessels of the placental bed to normal pregnancy. J Pathol Bacteriol 93,569–579.[ISI][Medline]

Brosens IA, Robertson WB and Dixon HG (1972) The role of the spiral arteries in the pathogenesis of preeclampsia. Obstet Gynecol Ann 1,177–191.[Medline]

Burton GJ, Jauniaux E and Watson AL (1999) Maternal arterial connections to the placental intervillous space during the first trimester of human pregnancy: the Boyd collection revisited. Am J Obstet Gynecol 181,718–724.[ISI][Medline]

Chervenak FA and McCullough LB (1999) Research on the fetus using Doppler ultrasound in the first trimester: guiding ethical considerations. Ultrasound Obstet Gynecol 14,161.

Chesley LC and Cooper DW (1986) Genetics of hypertension in pregnancy: possible single gene control of pre-eclampsia and eclampsia in the descendants of eclamptic women. Br J Obstet Gynaecol 93,898–908.[ISI][Medline]

Germain AM, Carvajal J, Sanchez M, Valenzuela GJ, Tsunekawa H and Chuaqui B (1999) Preterm labor: placental pathology and clinical correlation. Obstet Gynecol 94,284–289.[Abstract/Free Full Text]

Giles WB, Trudinger BJ and Baird PJ (1985) Fetal umbilical artery flow velocity waveforms and placental resistance: pathological correlation. Br J Obstet Gynaecol 92,31–38.[ISI][Medline]

Gulbis B, Jauniaux E, Cotton F and Stordeur P (1998) Protein and enzyme patterns in the fluid cavities of the first trimester gestational sac: relevance to the absorptive role of secondary yolk sac. Mol Hum Reprod 4,857–862.[Abstract]

Hadlock FP, Shah YP, Kanon DJ and Lindsey JV (1992) Fetal crown–rump length: reevaluation of relation to menstrual age (5–18 weeks) with high-resolution real-time US. Radiology 182,501–505.[Abstract]

Harrington KF, Campbell S, Bewley S and Bower S (1991) Doppler velocimetry studies of the uterine artery in the early prediction of pre-eclampsia and intra-uterine growth retardation. Eur J Obstet Gynecol Reprod Biol 42,S14–20.[ISI][Medline]

Harrington K, Carpenter RG, Goldfrad C and Campbell S (1997a) Transvaginal Doppler ultrasound of the uteroplacental circulation in the early prediction of pre-eclampsia and intrauterine growth retardation. Br J Obstet Gynaecol 104,674–681.[ISI][Medline]

Harrington K, Goldfrad C, Carpenter RG and Campbell S (1997b) Transvaginal uterine and umbilical artery Doppler examination of 12–16 weeks and the subsequent development of pre-eclampsia and intrauterine growth retardation. Ultrasound Obstet Gynecol 9,94–100.[CrossRef][ISI][Medline]

Hediger ML, Scholl TO, Schall JI, Miller LW and Fischer RL (1995) Fetal growth and the etiology of preterm delivery. Obstet Gynecol 85,175–182.[Abstract/Free Full Text]

Khong TY, De Wolf F, Robertson WB and Brosens I (1986) Inadequate maternal vascular response to placentation in pregnancies complicated by pre-eclampsia and by small-for-gestational age infants. Br J Obstet Gynecol 93,1049–1059.[ISI][Medline]

Makikallio K, Tekay A and Jouppila P (1999) Yolk sac and umbilicoplacental hemodynamics during early human embryonic development. Ultrasound Obstet Gynecol 14,175–179.[CrossRef][ISI][Medline]

Mires GJ, Williams FL, Leslie J and Howie PW (1998) Assessment of uterine arterial notching as a screening test for adverse pregnancy outcome. Am J Obstet Gynecol 179,1317–1323.[ISI][Medline]

Pijnenborg R, Bland JM, Robertson WB and Brosens I (1983) Uteroplacental arterial changes related to interstitial trophoblast migration in early human pregnancy. Placenta 4,397–413.[ISI][Medline]

Pijnenborg R, Anthony J, Davey DA, Rees A, Tiltman A, Vercruysse L and van Assche A (1991) Placental bed spiral arteries in the hypertensive disorders of pregnancy. Br J Obstet Gynecol 98,648–555.[ISI][Medline]

Salafia CM, Pezzullo JC, Lopez-Zeno JA, Simmens S, Minior VK and Vintzileos AM (1995) Placental pathologic features of preterm preeclampsia. Am J Obstet Gynecol 173,1097–1105.[ISI][Medline]

Schenker JG and Ezra Y (1994) Complications of assisted reproductive techniques. Fertil Steril 61,411–422.[ISI][Medline]

Stampone C, Nicotra M, Muttinelli C and Cosmi EV (1996) Transvaginal sonography of the yolk sac in normal and abnormal pregnancy. J Clin Ultrasound 24,3–9.[CrossRef][ISI][Medline]

Tarantal AF and Hendrickx AG (1990) The effects of exposure of the macague conceptus to ultrasound during organogenesis. Am J Primatol 237A.

Tessler FN, Kimme-Smith C, Sutherland ML, Schiller VL, Perrella RR and Grant EG (1990) Inter- and intra-observer variability of Doppler peak velocity measurements: an in-vitro study. Ultrasound Med Biol 16,653–657.[ISI][Medline]

Tomas C, Orava M, Tuomivaara L and Martikainen H (1998) Low pregnancy rate is achieved in patients treated with intracytoplasmic sperm injection due to previous low or failed fertilization in in-vitro fertilization. Hum Reprod 13,65–70.[Abstract]

Walsh SW (1985) Preeclampsia: an imbalance in placental prostacyclin and thromboxane production. Am J Obstet Gynecol 152,335–340.[ISI][Medline]

Submitted on August 15, 2003; accepted on November 14, 2003.