- Email: [email protected]
Rhythmic changes in the stomach movement of the human fetus with congenital duodenal obstruction during the third trimester of pregnancy
Early Human Development 54 (1999) 1–13
Rhythmic changes in the stomach movement of the human fetus with congenital duodenal obstruction during the third trimester of pregnancy Samia Mohamed Hussein a , Toshiyuki Yoshizato b , c b, d Shigehiro Fukushima , Takashi Koyanagi *, Kouhei Akazawa , Hitoo Nakano a a
Department of Gynecology and Obstetrics, Faculty of Medicine, Kyushu University, Fukuoka, Japan Maternity and Perinatal Care Unit, Kyushu University Hospital, c /o Department of Gynecology and Obstetrics, Faculty of Medicine, Kyushu University, Maidashi 3 -1 -1, Higashi-ku, Fukuoka 812 -8582, Japan c Applied Science of Audio and Visual Communication, Division of Audio and Visual Communication Studies, Kyushu Institute of Design, Fukuoka, Japan d Department of Medical Informatics, Kyushu University Hospital, Fukuoka, Japan
b
Received 6 February 1998; accepted 8 May 1998
Abstract In order to reveal whether or not rhythmic changes exist in fetal stomach movement (FSM), in utero FSM was assessed in three fetuses between 27 and 33 weeks’ gestation with congenital duodenal obstruction. A total of four observations, one each at 27, 29, 31 and 33 weeks’ gestation, was obtained. The longitudinal transection of each fetal stomach was continuously observed for 60 min using real-time ultrasound. The configuration and the area of the stomach were analyzed for each 15-s epoch. The complexity of the stomach configuration was quantified and defined as stomach complexity. For each case, the chronological changes of the stomach complexity were analyzed using the least median square of regression. The correlation between changes of the stomach complexity and the area of the stomach was analyzed using the cross-correlation method. (1) For gestational ages of 27, 29, 31 and 33 weeks, the 240 sequential measurements of the stomach complexity were significantly stratified into outlying and non-outlying points. The outlying points were 13.3% (32 / 240), 30.8% (74 / 240), 32.9% (79 / 240) and 36.3% (87 / 240) of the total observation points, respectively. (2) The percentages in which outlying points lasted 3 min (12 points) or more were 0% (0 / 240), 5.0% (12 / 240), *Corresponding author. Tel.: [email protected]
181-92-642-5394;
fax:
181-92-642-5414;
0378-3782 / 99 / – see front matter 1999 Elsevier Science Ireland Ltd. All rights reserved. PII: S0378-3782( 98 )00079-6
e-mail:
2
S.M. Hussein et al. / Early Human Development 54 (1999) 1 – 13
28.3% (68 / 240) and 30.4% (73 / 240) of the total observation points, respectively. (3) For each gestational age, no significant time series correlation was found between the stomach complexity and the area of the stomach. These findings suggest that: (1) two different conditions emerge in the FSM, at the latest at 27 weeks’ gestation, and begin manifesting from 29 weeks’ gestation onwards. (2) These chronological changes cluster into ‘active’ and ‘quiet’ phases from 31 weeks’ gestation onwards. (3) FSMs are not related to the changes in the stomach area throughout the observation periods. The underlying mechanism of this rhythmicity may represent the development of ultradian rhythm of the stomach movement, generated by the central nervous system. 1999 Elsevier Science Ireland Ltd. All rights reserved. Keywords: Congenital duodenal obstruction; Human fetus; Real-time ultrasound; Rhythmic changes; Stomach movement
1. Introduction Over the past two decades, an ever increasing number of studies have given rise to the organization of in utero behavior in the human fetus, in order to elucidate the manifestation and chronology of various motor nerve functions in relation to the development of the fetal central nervous system [13]. These include respiratory movement [25], extremity movement [7], heart rate changes [5,23], eye movement [2,3,12], penile tumescence [21] and sleep–waking pattern [11,18]. Previously, we studied the developmental characteristics of the gastric configuration in uncomplicated fetuses by using the three-dimensional reconstruction of ultrasound images. We revealed that the stomach maintained the same anatomical shape as seen in neonates and adults, i.e. with greater and lesser curvatures, from 27 weeks’ gestation onwards [17]. Fetuses with congenital duodenal obstruction had characteristics similar to those of uncomplicated fetuses with advance in gestation, even though their stomachs were extremely enlarged [27]. In studies using normal fetuses from 26 weeks’ gestation, Devan et al. reported the existence of temporal changes in the gastric volume, with peak-to-peak and trough-totrough duration having a periodicity of 35–55 min [6]. Thus, considering fetuses with congenital duodenal obstruction as a model, the aim of this study was to reveal whether or not rhythmic changes exist in the stomach movement in the third trimester of pregnancy and, if so, to clarify such changes in terms of a time-sequence profile and their relationship to area changes.
2. Materials and methods
2.1. Subjects Three fetuses between 27 and 33 weeks’ gestation with congenital duodenal obstruction were used in this study. The antenatal diagnoses were confirmed via
S.M. Hussein et al. / Early Human Development 54 (1999) 1 – 13
3
X-ray examination and operative findings in the postnatal periods. All the fetuses fulfilled the following criteria. (1) The gestational age calculated from the first day of the last menstrual period was subsequently confirmed by crown-rump length at 9 to 11 weeks’ gestation. (2) The biparital diameter, femur length and abdominal circumference were within the mean value 61.5 SD for the corresponding gestational age, according to the Japanese standards [22]. (3) Chromosomal analysis revealed normal karyotype. (4) No abnormalities were detected in utero except for dilated stomach and polyhydramnios. All the mothers were non-smokers with no complications or indications of drug administration throughout the observation periods. All were cared for in the Maternity and Perinatal Care Unit, Kyushu University Hospital, and consented to participate in this study.
2.2. Data acquisition Real-time ultrasound (Model SSA-270A, Toshiba, Tokyo) equipped with a 3.75MHz curvilinear transducer was used. With the mother in the semi-recumbent position, the longitudinal transection of fetal stomach was continuously observed for a 60-min period (Fig. 1). Each observation was made between 08:00 and 18:00 h, excluding mealtimes and 2 h following each meal. For each of the three fetuses, we encountered difficulties with scanning continuously for 60 min due to episodes of maternal intolerance and fetal gross movement, etc. This is why only four observations—one each at 27, 29, 31, and 33 weeks’ gestation—were obtained. The ultrasound images were recorded on videotape. While replaying the videotapes, the images were sampled each 15-s interval. For four observations, a total of 240 ultrasound images was used for subsequent analyses.
2.3. Data processing and parameter definition For each ultrasound image, the outline of the stomach was traced using a graphic digitizer (Model KD-4300, Graphtec, Tokyo) with a minimal resolution of 0.1 mm. The data thus obtained were directly transferred to an on-line microcomputer (NEC PC-9821 NB7, Tokyo) and processed. Dots were plotted automatically at a pitch of 1 / 20 of maximum longitudinal diameter in millimeters for each outline. These dots represented a polygon (Fig. 2A).
2.3.1. Stomach complexity In order to quantify the complexity of a stomach configuration, we used a newly devised parameter, defined as the stomach complexity. This parameter was calculated using a five-step process. First, the symmetrical axis of each stomach was determined using the division-based analysis of symmetry, which was devised by one of the authors and previously described in detail in Ref. [9]. In brief, the Delaunay triangulation (Fig. 2B) and the Voronoi diagram (Fig. 2C) were obtained for the dot pattern of the polygon. An edge of the Voronoi diagram is the perpendicular bisector
4
S.M. Hussein et al. / Early Human Development 54 (1999) 1 – 13
Fig. 1. Longitudinal transection of the fetal stomach obtained from 29 weeks’ gestation in the three different conditions of stomach activities: (I) resting, (II) active, (III) maximum activity. C, cardiac region; P, pyloric region; G, greater curvature; L, lesser curvature.
S.M. Hussein et al. / Early Human Development 54 (1999) 1 – 13
5
Fig. 2. Step-by-step process for obtaining the global symmetry of the stomach configuration using the division-based analysis of symmetry. (A) Dot pattern (approximation by a polygon) as a representation of the longitudinal tomography of the fetal stomach. (B) Delaunay triangulation showing a division of the dot pattern plane. (C) Voronoi diagram showing another division of the dot pattern plane. (D) Symmetric axis (broken line) of the figure (dot pattern) and connected paired dots (lines with arrows on both sides) symmetric to the axis.
of its dual edge of the Delaunay triangle. Therefore, geometrically, a Voronoi edge is locally the geometric symmetrical axis for its dual Delaunay edge. Thus, the longest path of the symmetrical axis along the spine of the polygon showed its global symmetry (Fig. 2D). Second, a normalized figure, N, was constructed by straightening the symmetrical axis of the original polygon (Fig. 3A). Third, we drew the convex hull, H, of the normalized figure (Fig. 3B). Fourth, we considered a line, l(x), which was perpendicular to the symmetrical axis of N and H and passed the point of x on the symmetrical axis. We defined the local width of N and H, Wn (x) and Wh (x), as the shortest distance between the point of x and the point at which l(x) intersected N and H, respectively. Local depression, D(x), at the point of x was determined as follows (Fig. 3C): D(x) 5 Wh (x) 2 Wn (x) Finally, the stomach complexity was calculated using the equation L
1E
1 Stomach complexity 5 ] D 2 (x)dx L 0
1/2
2
where L is the length of the symmetrical axis of the stomach. When the stomach is round or oval in shape, the stomach complexity indicates 0. The more complex the stomach configuration, the larger the value of the stomach complexity. When Wn (x) changes according to a sinusoidal function, the stomach complexity is determined only by its amplitude, independent of its cycle (see Appendix A).
6
S.M. Hussein et al. / Early Human Development 54 (1999) 1 – 13
Fig. 3. Quantitative description of the complexity of the stomach configuration. (A) Normalized figure, N, of the original dot pattern. (B) Convex hull, H, of the normalized figure (dot pattern). The abbreviations of l(x), Wn (x) and Wh (x) are seen in the text. (C) Local depression, D(x), between N and H. L is the length of the symmetrical axis of the stomach.
2.3.2. Area of the stomach The area of the stomach was obtained by calculating the total area of the dot pattern of the polygon.
2.4. Reproducibility of this experimental system In order to evaluate the reproducibility of the main frame study, another simulation experiment was performed on one fetus at 29 weeks’ gestation. Three ultrasound images at different stomach configurations as shown in Fig. 1 were chosen. The procedure was repeated independently 10 times from tracing the outline of the longitudinally transected image of the fetal stomach through to the calculation of the stomach complexity and the area of the stomach. The coefficients of variation were found to range from 4.9 to 7.0% for the stomach complexity and from 0.5 to 0.9% for the area of the stomach.
S.M. Hussein et al. / Early Human Development 54 (1999) 1 – 13
7
2.5. Statistical analysis 2.5.1. Chronological changes in the stomach complexity In order to describe the chronological changes in the stomach movement quantitatively, a scattergram was made for the stomach complexity versus elapsed time from the start of the observation. The scattergram was then analyzed using the least median of square regression, yielding a regression line and the standardized residual defined as the least median of square residual divided by the least median of square scale estimate sLMS for each measurement. In each case the outlying and non-outlying points among the measurements were defined as those with a standardized residual of .1.5 and ,1.5, respectively. Calculations were performed on a microcomputer system using the program by Rousseeuw [20].
2.5.2. Relationship between the chronological changes in the stomach complexity and the area of the stomach In order to reveal the relationship between the chronological changes in the stomach movement and the area of the stomach, the stomach complexity and the area of the stomach were analyzed using the cross-correlation method. Cross-correlation coefficients were calculated for 240 to 140 lags (210 to 110 min). This analysis is used to assess the significant relationship between two parameters in a time series. However, if one parameter contains autocorrelation or is not stationary about the mean, the cross-correlation function will not reflect the true relationship between the two parameters. Therefore, before calculating the crosscorrelation functions, we removed the autocorrelation component from each parameter by prewhitening or filtering the data, following the Box and Jenkins time series analysis [26]. The corresponding standard error (SE) of the correlation coefficient for a series of N samples at lag k can be estimated as (N 2 k)1 / 2 [4]. Significant correlation coefficients were determined when their absolute values exceeded the respective 2 SE. The calculations were carried out using the computer package program BMDP2T [15] on a SPARC station 20 (Sun microsystems, CA).
3. Results
3.1. Chronological changes in the stomach complexity For gestational ages of 27, 29, 31 and 33 weeks, the 240 sequential measurements of the stomach complexity were stratified, with statistical significance, into outlying and non-outlying points (Fig. 4). Alternating changes of outlying and non-outlying points were evident. The outlying points were 13.3% (32 / 240), 30.8% (74 / 240), 32.9% (79 / 240) and 36.3% (87 / 240) of the total observation points at 27, 29, 31 and 33 weeks’ gestation, respectively. The
8
S.M. Hussein et al. / Early Human Development 54 (1999) 1 – 13
Fig. 4. Chronological changes in the stomach complexity and the area of the stomach at 27, 29, 31 and 33 weeks’ gestation. Closed and open circles indicate the outlying points (standardized residual .1.5) and non-outlying points (standardized residual ,1.5), respectively. Arrows show the values of the stomach complexity at 29 weeks’ gestation in the three different activity conditions; I, II and III are as shown in Fig. 1.
percentages in which outlying points lasted 3 min (12 points) or more were 0% (0 / 240), 5.0% (12 / 240), 28.3% (68 / 240) and 30.4% (73 / 240) of the total observation points at 27, 29, 31 and 33 weeks’ gestation, respectively (Table 1). Table 1 Duration of outlying points of stomach complexity for each gestational age Gestational age (weeks)
Outlying points [duration (points)3incidence]
Percentage of duration of outlying points per total observation points
27 29
135, 235, 332, 431, 731 136, 232, 333, 431, 531, 732 931, 1131, 1231 331, 432, 2531, 4331 732, 1431, 1631, 2031, 2331
13.3% (32 / 240) 30.8% (74 / 240)
0% (0 / 240) 5.0% (12 / 240)
32.9% (79 / 240) 36.3% (87 / 240)
28.3% (68 / 240) 30.4% (73 / 240)
31 33
Percentage in which outlying points lasted 3 min (12 points) or more per total observationpoints
S.M. Hussein et al. / Early Human Development 54 (1999) 1 – 13
9
3.2. Relationship between the chronological changes in the stomach complexity and the area of the stomach For each gestational age, there was no significant relationship in time series between the stomach complexity and the area of the stomach (Fig. 5).
4. Discussion The gastrointestinal system has a dynamic function during intrauterine life [16] and the stomach has advantages in terms of its anatomical location, shape and mobility, observed in real-time in utero [17]. In order to assess the rhythmic changes in gastric movement, fetuses with congenital duodenal obstruction were used. This is because their extremely enlarged stomachs were easily observed via ultrasound. In addition, our previous study showed that their stomachs maintained almost the same developmental characteristics in utero as uncomplicated fetuses in terms of the three-dimensional configuration [27]. Aladjem et al. found a marked increase in the fetal motility between 29 and 31 weeks’ gestation; they called this the hyperkinetic period [1]. Accordingly, we selected this period of gestation for our study, and included the 33-week gestation period, in order to follow the chronological changes in the movement related to their area changes. There are at least two different kinds of movements in the fetal stomach: wall movement and axial movement. Here, we directly focused on the stomach wall movement by quantitatively analyzing changes in the stomach configuration. This was pre-processed to determine the symmetrical axis of each stomach configuration and straighten it; thus the stomach axial movement component was cancelled out. The least median of square (LMS) regression analysis was used for statistical discrimination between the two conditions in the stomach movement, which indicated two different underlying characteristics. The LMS regression analysis is a more robust version of the classical least square regression and can resist the affects of nearly 50% data contamination [20]. The chronological changes in the stomach wall movement suggest that two different conditions exist in the FSM. These emerge, at the latest, at 27 weeks’ gestation and begin manifesting from 29 weeks’ gestation onwards. In addition, these chronological changes cluster into ‘active’ and ‘quiet’ phases from 31 weeks’ gestation onwards. With respect to the fetal heart rate changes in the human fetus, it has already been noted that the rhythmic changes occur at around 28 weeks’ gestation, which manifest with advance in gestation; the quiet–activity cycle emerges at 34–35 weeks’ gestation onwards [5,23]. Since these rhythmic changes reflect the interaction between the two limbs of the autonomic nervous system controlling the heart, the autonomic control could also cause such rhythmicity in the fetal stomach movement. The stomach potentially fills and empties continuously and, therefore, gastric volume represents the net effect of both of these processes [6]. In utero, however, it is
10
S.M. Hussein et al. / Early Human Development 54 (1999) 1 – 13
Fig. 5. Cross-correlation coefficients of the chronological changes in the stomach complexity and the area of the stomach at 27, 29, 31 and 33 weeks’ gestation. The dotted lines indicate the 95% confidence limit for the null hypothesis.
S.M. Hussein et al. / Early Human Development 54 (1999) 1 – 13
11
impossible to measure the actual stomach volume in real time. Ultrasonography is the only method currently available for assessing gastric dimensions. Our previous study showed how one could estimate stomach volume using the area of a longitudinal transection, because the short-axially-transected stomach could be considered as a circle [17]. Thus, we calculated the values of area at every 15-s period and used these values as parameters to describe the changes in the fetal stomach volume. There was no statistical significance between the rhythmic changes in the stomach movement and the area. These findings suggest that the underlying mechanism of this rhythmicity is not a local mechanism. This is in line with a study which recorded the ultradian rhythm of the gastric contractile activity in adults during wakefulness. This study extended the idea of a feeding or oral cycle, derived from behavioral observations, to physiological correlates that must reflect neuronal or hormonal mechanisms, suggesting an underlying hypothalamic cycle [10]. Others suggested that both extrinsic nerve (especially the vagus) and hormonal factors (motilin in particular) were involved in regulation of gastric mobility [19]. Tomomasa et al. stated that antral smooth muscle motilin receptors appeared in the third trimester and that stimulation of these receptors caused antral contraction [24]. They also studied the difference in gastrointestinal motility between infants and adults and showed that, in infants, bands of rhythmic contractions were frequently recorded; some migrated caudally as in adults, while others did not and showed some difference in duration and amplitude. This indicated that the occurrence of rhythmic contraction and their migration in infants was regulated by different mechanisms [8]. On the other hand, Lavie et al. found neither of these findings in normal adults, stating that gastric rhythms were neither related to rapid eye movement / no-rapid eye movement cycles nor to any sleep stage [14]. Therefore, our study may provide further basis for studying stomach movement in order to establish such characters in normal cases and to show the relationship with other rhythmic changes in the human fetus.
Acknowledgements Supported by a Grant-in-Aid for Scientific Research (Nos. 07457390 and 07671801) from the Ministry of Education, Science and Culture, Japan. We thank Lynn Fujino for the support with manuscript preparation and Shin Nagata, M.D., Ph.D. and Shoji Satoh, M.D., Ph.D. for critical comments on this paper.
Appendix A Assume that Wn (x) changes according to a sinusoidal function (Fig. 6) and is given as Wn (x) 5 B 1 A cosh2p (n /L)xj
S.M. Hussein et al. / Early Human Development 54 (1999) 1 – 13
12
Fig. 6. The model of stomach configuration when its normalized figure is determined by a sinusoidal function.
where B is the baseline, A is the amplitude, L is the length of the symmetrical axis, and n is the number of waves of Wn (x).Thus, Wh (x) 5 B 1 A D(x) 5 Wh (x) 2 Wn (x) 5 A[1 2 cosh2p (n /L)xj] Therefore, L
1E
1/2
2
1 Stomach complexity 5 ] hA(1 2 cos 2p (n /L)x)j 2 dx L 0
5 (3 / 2)
1/2
A
Provided that the stomach complexity is determined only by the amplitude of Wn (x), A, and not by its cycle, n /L.
References [1] Aladjem S, Vuolo R, Pazos R, Lueck J. Antepartum fetal testing evaluation and redefinition of criteria for clinical interpretation. Semin Perinatol 1981;5:145–53. [2] Birnholz JC. The development of human fetal eye movement patterns. Science 1981;213:679–81. [3] Bots RSGM, Nijhuis JG, Martin Jr. CB, Prechtl HFR. Human fetal eye movement: detection in utero by ultrasonography. Early Hum Dev 1981;5:87–94. [4] Chatfield C. Estimation in the time domain. In: Analysis of time series, 3rd ed. New York: Chapman and Hall, 1984:60–71. [5] Dawes GS, Houghton CRS, Redman CWG, Visser GHA. Pattern of the normal human fetal heart rate. Br J Obstet Gynaecol 1982;89:276–84. [6] Devane SP, Soothill PW, Candy DCA. Temporal changes in gastric volume in human fetus in late pregnancy. Early Hum Dev 1993;33:109–16. [7] De Vries JIP, Visser GHA, Prechtl HFR. The emergence of fetal behavior. II. Quantitative aspects. Early Hum Dev 1985;12:99–120.
S.M. Hussein et al. / Early Human Development 54 (1999) 1 – 13
13
[8] Di Lorenzo C, Hyman PE. Gastrointestinal motility in neonatal and pediatric practice. Gastroenterol Clin North Am 1996;25:203–24. [9] Fukushima S. Division-based analysis of symmetry and its application. IEEE Trans Pat Anal Mach Intel 1997;19(2):144–8. [10] Hiatt JF, Kripke DF. Ultradian rhythms in waking gastric activity. Psychosom Med 1975;37(4):320– 5. [11] Horimoto N, Koyanagi T, Takashima T, Akazawa K, Nakano H. Changes in pupillary diameter in relation to eye-movement and no eye-movement periods in human fetus at term. Am J Obstet Gynecol 1992;167:1465–9. [12] Inoue M, Koyanagi T, Nakahara H, Hara K, Hori E, Nakano H. Functional development of human eye movement in utero assessed quantitatively with real-time ultrasound. Am J Obstet Gynecol 1986;155:170–4. [13] Koyanagi T, Nakano H. Functional development of the fetal central nervous system. In: Levene MI, Lilford RJ, editors. Fetal and neonatal neurology and neurosurgery, 2nd ed. London: Churchill Livingstone, 1995:31–44. [14] Lavie P, Kripke DF, Hiatt JF, Harrison J. Gastric rhythms during sleep. Behav Biol 1978;23:526–30. [15] Liu L. Box–Jenkins time series analysis. In: Dixon WJ et al., editors. BMDP statistical software manual. Berkley: University of California Press, 1992:435–88. [16] Mclain Jr. CR. Amniography studies of gastrointestinal motility of the human fetus. Am J Obstet Gyncol 1963;86:1079–87. [17] Nagata S, Koyanagi T, Fukushima S, Akazawa K, Nakano H. Change in the three-dimensional shape of the stomach in the developing human fetus. Early Hum Dev 1994;37:27–38. [18] Prechtl HFR. The behavioural states of the newborn infants (a review). Brain Res 1974;76:185–212. [19] Quigley EM. Gastric and small intestinal motility in health and disease. Gastroenterol Clin North Am 1996;25:113–45. [20] Rousseeuw PJ. Least median of square regression. J Am Stat Assoc 1984;79:871–80. [21] Shirozu H, Koyanagi T, Takashima T, Horimoto N, Akazawa K, Nakano H. Penile tumescence in the human fetus at term—A preliminary report. Early Hum Dev 1995;41:159–66. [22] Subcommittee of the Japan Society of Obstetrics and Gynecology. Standard curve of BPD and FL in Japanese fetuses assessed using ultrasound. Acta Obstet Gynecol Jpn 1993;45:391–4. [23] Timor-Tritsch IE, Dierker LJ, Hertz RH, Deagan NC, Rosen MG. Studies of antepartum behavioral state in the human fetus at term. Am J Obstet Gynecol 1979;132:524–8. [24] Tomomasa T, Itoh Z, Koizumi T, Kuroume T. Non migrating rhythmic activity in the stomach and duodenum of neonate. Biol Neonate 1985;48:1–9. [25] Trudinger BJ, Knight BC. Fetal age and pattern of human fetal breathing movements. Am J Obstet Gynecol 1980;137:724–8. [26] Vandaele W. Applied time series and Box–Jenkins Models. Ontario: Academic Press, 1983:257–321. [27] Yoshizato T, Koyanagi T, Nagata S, Takashima T, Fukushima S, Nakano H. Three-dimensional ultrasound image of the fetal stomach: congenital duodenal obstruction in utero. Early Hum Dev 1995;41:39–47.
Rhythmic changes in the stomach movement of the human fetus with congenital duodenal obstruction during the third trimester of pregnancy Samia Mohamed Hussein a , Toshiyuki Yoshizato b , c b, d Shigehiro Fukushima , Takashi Koyanagi *, Kouhei Akazawa , Hitoo Nakano a a
Department of Gynecology and Obstetrics, Faculty of Medicine, Kyushu University, Fukuoka, Japan Maternity and Perinatal Care Unit, Kyushu University Hospital, c /o Department of Gynecology and Obstetrics, Faculty of Medicine, Kyushu University, Maidashi 3 -1 -1, Higashi-ku, Fukuoka 812 -8582, Japan c Applied Science of Audio and Visual Communication, Division of Audio and Visual Communication Studies, Kyushu Institute of Design, Fukuoka, Japan d Department of Medical Informatics, Kyushu University Hospital, Fukuoka, Japan
b
Received 6 February 1998; accepted 8 May 1998
Abstract In order to reveal whether or not rhythmic changes exist in fetal stomach movement (FSM), in utero FSM was assessed in three fetuses between 27 and 33 weeks’ gestation with congenital duodenal obstruction. A total of four observations, one each at 27, 29, 31 and 33 weeks’ gestation, was obtained. The longitudinal transection of each fetal stomach was continuously observed for 60 min using real-time ultrasound. The configuration and the area of the stomach were analyzed for each 15-s epoch. The complexity of the stomach configuration was quantified and defined as stomach complexity. For each case, the chronological changes of the stomach complexity were analyzed using the least median square of regression. The correlation between changes of the stomach complexity and the area of the stomach was analyzed using the cross-correlation method. (1) For gestational ages of 27, 29, 31 and 33 weeks, the 240 sequential measurements of the stomach complexity were significantly stratified into outlying and non-outlying points. The outlying points were 13.3% (32 / 240), 30.8% (74 / 240), 32.9% (79 / 240) and 36.3% (87 / 240) of the total observation points, respectively. (2) The percentages in which outlying points lasted 3 min (12 points) or more were 0% (0 / 240), 5.0% (12 / 240), *Corresponding author. Tel.: [email protected]
181-92-642-5394;
fax:
181-92-642-5414;
0378-3782 / 99 / – see front matter 1999 Elsevier Science Ireland Ltd. All rights reserved. PII: S0378-3782( 98 )00079-6
e-mail:
2
S.M. Hussein et al. / Early Human Development 54 (1999) 1 – 13
28.3% (68 / 240) and 30.4% (73 / 240) of the total observation points, respectively. (3) For each gestational age, no significant time series correlation was found between the stomach complexity and the area of the stomach. These findings suggest that: (1) two different conditions emerge in the FSM, at the latest at 27 weeks’ gestation, and begin manifesting from 29 weeks’ gestation onwards. (2) These chronological changes cluster into ‘active’ and ‘quiet’ phases from 31 weeks’ gestation onwards. (3) FSMs are not related to the changes in the stomach area throughout the observation periods. The underlying mechanism of this rhythmicity may represent the development of ultradian rhythm of the stomach movement, generated by the central nervous system. 1999 Elsevier Science Ireland Ltd. All rights reserved. Keywords: Congenital duodenal obstruction; Human fetus; Real-time ultrasound; Rhythmic changes; Stomach movement
1. Introduction Over the past two decades, an ever increasing number of studies have given rise to the organization of in utero behavior in the human fetus, in order to elucidate the manifestation and chronology of various motor nerve functions in relation to the development of the fetal central nervous system [13]. These include respiratory movement [25], extremity movement [7], heart rate changes [5,23], eye movement [2,3,12], penile tumescence [21] and sleep–waking pattern [11,18]. Previously, we studied the developmental characteristics of the gastric configuration in uncomplicated fetuses by using the three-dimensional reconstruction of ultrasound images. We revealed that the stomach maintained the same anatomical shape as seen in neonates and adults, i.e. with greater and lesser curvatures, from 27 weeks’ gestation onwards [17]. Fetuses with congenital duodenal obstruction had characteristics similar to those of uncomplicated fetuses with advance in gestation, even though their stomachs were extremely enlarged [27]. In studies using normal fetuses from 26 weeks’ gestation, Devan et al. reported the existence of temporal changes in the gastric volume, with peak-to-peak and trough-totrough duration having a periodicity of 35–55 min [6]. Thus, considering fetuses with congenital duodenal obstruction as a model, the aim of this study was to reveal whether or not rhythmic changes exist in the stomach movement in the third trimester of pregnancy and, if so, to clarify such changes in terms of a time-sequence profile and their relationship to area changes.
2. Materials and methods
2.1. Subjects Three fetuses between 27 and 33 weeks’ gestation with congenital duodenal obstruction were used in this study. The antenatal diagnoses were confirmed via
S.M. Hussein et al. / Early Human Development 54 (1999) 1 – 13
3
X-ray examination and operative findings in the postnatal periods. All the fetuses fulfilled the following criteria. (1) The gestational age calculated from the first day of the last menstrual period was subsequently confirmed by crown-rump length at 9 to 11 weeks’ gestation. (2) The biparital diameter, femur length and abdominal circumference were within the mean value 61.5 SD for the corresponding gestational age, according to the Japanese standards [22]. (3) Chromosomal analysis revealed normal karyotype. (4) No abnormalities were detected in utero except for dilated stomach and polyhydramnios. All the mothers were non-smokers with no complications or indications of drug administration throughout the observation periods. All were cared for in the Maternity and Perinatal Care Unit, Kyushu University Hospital, and consented to participate in this study.
2.2. Data acquisition Real-time ultrasound (Model SSA-270A, Toshiba, Tokyo) equipped with a 3.75MHz curvilinear transducer was used. With the mother in the semi-recumbent position, the longitudinal transection of fetal stomach was continuously observed for a 60-min period (Fig. 1). Each observation was made between 08:00 and 18:00 h, excluding mealtimes and 2 h following each meal. For each of the three fetuses, we encountered difficulties with scanning continuously for 60 min due to episodes of maternal intolerance and fetal gross movement, etc. This is why only four observations—one each at 27, 29, 31, and 33 weeks’ gestation—were obtained. The ultrasound images were recorded on videotape. While replaying the videotapes, the images were sampled each 15-s interval. For four observations, a total of 240 ultrasound images was used for subsequent analyses.
2.3. Data processing and parameter definition For each ultrasound image, the outline of the stomach was traced using a graphic digitizer (Model KD-4300, Graphtec, Tokyo) with a minimal resolution of 0.1 mm. The data thus obtained were directly transferred to an on-line microcomputer (NEC PC-9821 NB7, Tokyo) and processed. Dots were plotted automatically at a pitch of 1 / 20 of maximum longitudinal diameter in millimeters for each outline. These dots represented a polygon (Fig. 2A).
2.3.1. Stomach complexity In order to quantify the complexity of a stomach configuration, we used a newly devised parameter, defined as the stomach complexity. This parameter was calculated using a five-step process. First, the symmetrical axis of each stomach was determined using the division-based analysis of symmetry, which was devised by one of the authors and previously described in detail in Ref. [9]. In brief, the Delaunay triangulation (Fig. 2B) and the Voronoi diagram (Fig. 2C) were obtained for the dot pattern of the polygon. An edge of the Voronoi diagram is the perpendicular bisector
4
S.M. Hussein et al. / Early Human Development 54 (1999) 1 – 13
Fig. 1. Longitudinal transection of the fetal stomach obtained from 29 weeks’ gestation in the three different conditions of stomach activities: (I) resting, (II) active, (III) maximum activity. C, cardiac region; P, pyloric region; G, greater curvature; L, lesser curvature.
S.M. Hussein et al. / Early Human Development 54 (1999) 1 – 13
5
Fig. 2. Step-by-step process for obtaining the global symmetry of the stomach configuration using the division-based analysis of symmetry. (A) Dot pattern (approximation by a polygon) as a representation of the longitudinal tomography of the fetal stomach. (B) Delaunay triangulation showing a division of the dot pattern plane. (C) Voronoi diagram showing another division of the dot pattern plane. (D) Symmetric axis (broken line) of the figure (dot pattern) and connected paired dots (lines with arrows on both sides) symmetric to the axis.
of its dual edge of the Delaunay triangle. Therefore, geometrically, a Voronoi edge is locally the geometric symmetrical axis for its dual Delaunay edge. Thus, the longest path of the symmetrical axis along the spine of the polygon showed its global symmetry (Fig. 2D). Second, a normalized figure, N, was constructed by straightening the symmetrical axis of the original polygon (Fig. 3A). Third, we drew the convex hull, H, of the normalized figure (Fig. 3B). Fourth, we considered a line, l(x), which was perpendicular to the symmetrical axis of N and H and passed the point of x on the symmetrical axis. We defined the local width of N and H, Wn (x) and Wh (x), as the shortest distance between the point of x and the point at which l(x) intersected N and H, respectively. Local depression, D(x), at the point of x was determined as follows (Fig. 3C): D(x) 5 Wh (x) 2 Wn (x) Finally, the stomach complexity was calculated using the equation L
1E
1 Stomach complexity 5 ] D 2 (x)dx L 0
1/2
2
where L is the length of the symmetrical axis of the stomach. When the stomach is round or oval in shape, the stomach complexity indicates 0. The more complex the stomach configuration, the larger the value of the stomach complexity. When Wn (x) changes according to a sinusoidal function, the stomach complexity is determined only by its amplitude, independent of its cycle (see Appendix A).
6
S.M. Hussein et al. / Early Human Development 54 (1999) 1 – 13
Fig. 3. Quantitative description of the complexity of the stomach configuration. (A) Normalized figure, N, of the original dot pattern. (B) Convex hull, H, of the normalized figure (dot pattern). The abbreviations of l(x), Wn (x) and Wh (x) are seen in the text. (C) Local depression, D(x), between N and H. L is the length of the symmetrical axis of the stomach.
2.3.2. Area of the stomach The area of the stomach was obtained by calculating the total area of the dot pattern of the polygon.
2.4. Reproducibility of this experimental system In order to evaluate the reproducibility of the main frame study, another simulation experiment was performed on one fetus at 29 weeks’ gestation. Three ultrasound images at different stomach configurations as shown in Fig. 1 were chosen. The procedure was repeated independently 10 times from tracing the outline of the longitudinally transected image of the fetal stomach through to the calculation of the stomach complexity and the area of the stomach. The coefficients of variation were found to range from 4.9 to 7.0% for the stomach complexity and from 0.5 to 0.9% for the area of the stomach.
S.M. Hussein et al. / Early Human Development 54 (1999) 1 – 13
7
2.5. Statistical analysis 2.5.1. Chronological changes in the stomach complexity In order to describe the chronological changes in the stomach movement quantitatively, a scattergram was made for the stomach complexity versus elapsed time from the start of the observation. The scattergram was then analyzed using the least median of square regression, yielding a regression line and the standardized residual defined as the least median of square residual divided by the least median of square scale estimate sLMS for each measurement. In each case the outlying and non-outlying points among the measurements were defined as those with a standardized residual of .1.5 and ,1.5, respectively. Calculations were performed on a microcomputer system using the program by Rousseeuw [20].
2.5.2. Relationship between the chronological changes in the stomach complexity and the area of the stomach In order to reveal the relationship between the chronological changes in the stomach movement and the area of the stomach, the stomach complexity and the area of the stomach were analyzed using the cross-correlation method. Cross-correlation coefficients were calculated for 240 to 140 lags (210 to 110 min). This analysis is used to assess the significant relationship between two parameters in a time series. However, if one parameter contains autocorrelation or is not stationary about the mean, the cross-correlation function will not reflect the true relationship between the two parameters. Therefore, before calculating the crosscorrelation functions, we removed the autocorrelation component from each parameter by prewhitening or filtering the data, following the Box and Jenkins time series analysis [26]. The corresponding standard error (SE) of the correlation coefficient for a series of N samples at lag k can be estimated as (N 2 k)1 / 2 [4]. Significant correlation coefficients were determined when their absolute values exceeded the respective 2 SE. The calculations were carried out using the computer package program BMDP2T [15] on a SPARC station 20 (Sun microsystems, CA).
3. Results
3.1. Chronological changes in the stomach complexity For gestational ages of 27, 29, 31 and 33 weeks, the 240 sequential measurements of the stomach complexity were stratified, with statistical significance, into outlying and non-outlying points (Fig. 4). Alternating changes of outlying and non-outlying points were evident. The outlying points were 13.3% (32 / 240), 30.8% (74 / 240), 32.9% (79 / 240) and 36.3% (87 / 240) of the total observation points at 27, 29, 31 and 33 weeks’ gestation, respectively. The
8
S.M. Hussein et al. / Early Human Development 54 (1999) 1 – 13
Fig. 4. Chronological changes in the stomach complexity and the area of the stomach at 27, 29, 31 and 33 weeks’ gestation. Closed and open circles indicate the outlying points (standardized residual .1.5) and non-outlying points (standardized residual ,1.5), respectively. Arrows show the values of the stomach complexity at 29 weeks’ gestation in the three different activity conditions; I, II and III are as shown in Fig. 1.
percentages in which outlying points lasted 3 min (12 points) or more were 0% (0 / 240), 5.0% (12 / 240), 28.3% (68 / 240) and 30.4% (73 / 240) of the total observation points at 27, 29, 31 and 33 weeks’ gestation, respectively (Table 1). Table 1 Duration of outlying points of stomach complexity for each gestational age Gestational age (weeks)
Outlying points [duration (points)3incidence]
Percentage of duration of outlying points per total observation points
27 29
135, 235, 332, 431, 731 136, 232, 333, 431, 531, 732 931, 1131, 1231 331, 432, 2531, 4331 732, 1431, 1631, 2031, 2331
13.3% (32 / 240) 30.8% (74 / 240)
0% (0 / 240) 5.0% (12 / 240)
32.9% (79 / 240) 36.3% (87 / 240)
28.3% (68 / 240) 30.4% (73 / 240)
31 33
Percentage in which outlying points lasted 3 min (12 points) or more per total observationpoints
S.M. Hussein et al. / Early Human Development 54 (1999) 1 – 13
9
3.2. Relationship between the chronological changes in the stomach complexity and the area of the stomach For each gestational age, there was no significant relationship in time series between the stomach complexity and the area of the stomach (Fig. 5).
4. Discussion The gastrointestinal system has a dynamic function during intrauterine life [16] and the stomach has advantages in terms of its anatomical location, shape and mobility, observed in real-time in utero [17]. In order to assess the rhythmic changes in gastric movement, fetuses with congenital duodenal obstruction were used. This is because their extremely enlarged stomachs were easily observed via ultrasound. In addition, our previous study showed that their stomachs maintained almost the same developmental characteristics in utero as uncomplicated fetuses in terms of the three-dimensional configuration [27]. Aladjem et al. found a marked increase in the fetal motility between 29 and 31 weeks’ gestation; they called this the hyperkinetic period [1]. Accordingly, we selected this period of gestation for our study, and included the 33-week gestation period, in order to follow the chronological changes in the movement related to their area changes. There are at least two different kinds of movements in the fetal stomach: wall movement and axial movement. Here, we directly focused on the stomach wall movement by quantitatively analyzing changes in the stomach configuration. This was pre-processed to determine the symmetrical axis of each stomach configuration and straighten it; thus the stomach axial movement component was cancelled out. The least median of square (LMS) regression analysis was used for statistical discrimination between the two conditions in the stomach movement, which indicated two different underlying characteristics. The LMS regression analysis is a more robust version of the classical least square regression and can resist the affects of nearly 50% data contamination [20]. The chronological changes in the stomach wall movement suggest that two different conditions exist in the FSM. These emerge, at the latest, at 27 weeks’ gestation and begin manifesting from 29 weeks’ gestation onwards. In addition, these chronological changes cluster into ‘active’ and ‘quiet’ phases from 31 weeks’ gestation onwards. With respect to the fetal heart rate changes in the human fetus, it has already been noted that the rhythmic changes occur at around 28 weeks’ gestation, which manifest with advance in gestation; the quiet–activity cycle emerges at 34–35 weeks’ gestation onwards [5,23]. Since these rhythmic changes reflect the interaction between the two limbs of the autonomic nervous system controlling the heart, the autonomic control could also cause such rhythmicity in the fetal stomach movement. The stomach potentially fills and empties continuously and, therefore, gastric volume represents the net effect of both of these processes [6]. In utero, however, it is
10
S.M. Hussein et al. / Early Human Development 54 (1999) 1 – 13
Fig. 5. Cross-correlation coefficients of the chronological changes in the stomach complexity and the area of the stomach at 27, 29, 31 and 33 weeks’ gestation. The dotted lines indicate the 95% confidence limit for the null hypothesis.
S.M. Hussein et al. / Early Human Development 54 (1999) 1 – 13
11
impossible to measure the actual stomach volume in real time. Ultrasonography is the only method currently available for assessing gastric dimensions. Our previous study showed how one could estimate stomach volume using the area of a longitudinal transection, because the short-axially-transected stomach could be considered as a circle [17]. Thus, we calculated the values of area at every 15-s period and used these values as parameters to describe the changes in the fetal stomach volume. There was no statistical significance between the rhythmic changes in the stomach movement and the area. These findings suggest that the underlying mechanism of this rhythmicity is not a local mechanism. This is in line with a study which recorded the ultradian rhythm of the gastric contractile activity in adults during wakefulness. This study extended the idea of a feeding or oral cycle, derived from behavioral observations, to physiological correlates that must reflect neuronal or hormonal mechanisms, suggesting an underlying hypothalamic cycle [10]. Others suggested that both extrinsic nerve (especially the vagus) and hormonal factors (motilin in particular) were involved in regulation of gastric mobility [19]. Tomomasa et al. stated that antral smooth muscle motilin receptors appeared in the third trimester and that stimulation of these receptors caused antral contraction [24]. They also studied the difference in gastrointestinal motility between infants and adults and showed that, in infants, bands of rhythmic contractions were frequently recorded; some migrated caudally as in adults, while others did not and showed some difference in duration and amplitude. This indicated that the occurrence of rhythmic contraction and their migration in infants was regulated by different mechanisms [8]. On the other hand, Lavie et al. found neither of these findings in normal adults, stating that gastric rhythms were neither related to rapid eye movement / no-rapid eye movement cycles nor to any sleep stage [14]. Therefore, our study may provide further basis for studying stomach movement in order to establish such characters in normal cases and to show the relationship with other rhythmic changes in the human fetus.
Acknowledgements Supported by a Grant-in-Aid for Scientific Research (Nos. 07457390 and 07671801) from the Ministry of Education, Science and Culture, Japan. We thank Lynn Fujino for the support with manuscript preparation and Shin Nagata, M.D., Ph.D. and Shoji Satoh, M.D., Ph.D. for critical comments on this paper.
Appendix A Assume that Wn (x) changes according to a sinusoidal function (Fig. 6) and is given as Wn (x) 5 B 1 A cosh2p (n /L)xj
S.M. Hussein et al. / Early Human Development 54 (1999) 1 – 13
12
Fig. 6. The model of stomach configuration when its normalized figure is determined by a sinusoidal function.
where B is the baseline, A is the amplitude, L is the length of the symmetrical axis, and n is the number of waves of Wn (x).Thus, Wh (x) 5 B 1 A D(x) 5 Wh (x) 2 Wn (x) 5 A[1 2 cosh2p (n /L)xj] Therefore, L
1E
1/2
2
1 Stomach complexity 5 ] hA(1 2 cos 2p (n /L)x)j 2 dx L 0
5 (3 / 2)
1/2
A
Provided that the stomach complexity is determined only by the amplitude of Wn (x), A, and not by its cycle, n /L.
References [1] Aladjem S, Vuolo R, Pazos R, Lueck J. Antepartum fetal testing evaluation and redefinition of criteria for clinical interpretation. Semin Perinatol 1981;5:145–53. [2] Birnholz JC. The development of human fetal eye movement patterns. Science 1981;213:679–81. [3] Bots RSGM, Nijhuis JG, Martin Jr. CB, Prechtl HFR. Human fetal eye movement: detection in utero by ultrasonography. Early Hum Dev 1981;5:87–94. [4] Chatfield C. Estimation in the time domain. In: Analysis of time series, 3rd ed. New York: Chapman and Hall, 1984:60–71. [5] Dawes GS, Houghton CRS, Redman CWG, Visser GHA. Pattern of the normal human fetal heart rate. Br J Obstet Gynaecol 1982;89:276–84. [6] Devane SP, Soothill PW, Candy DCA. Temporal changes in gastric volume in human fetus in late pregnancy. Early Hum Dev 1993;33:109–16. [7] De Vries JIP, Visser GHA, Prechtl HFR. The emergence of fetal behavior. II. Quantitative aspects. Early Hum Dev 1985;12:99–120.
S.M. Hussein et al. / Early Human Development 54 (1999) 1 – 13
13
[8] Di Lorenzo C, Hyman PE. Gastrointestinal motility in neonatal and pediatric practice. Gastroenterol Clin North Am 1996;25:203–24. [9] Fukushima S. Division-based analysis of symmetry and its application. IEEE Trans Pat Anal Mach Intel 1997;19(2):144–8. [10] Hiatt JF, Kripke DF. Ultradian rhythms in waking gastric activity. Psychosom Med 1975;37(4):320– 5. [11] Horimoto N, Koyanagi T, Takashima T, Akazawa K, Nakano H. Changes in pupillary diameter in relation to eye-movement and no eye-movement periods in human fetus at term. Am J Obstet Gynecol 1992;167:1465–9. [12] Inoue M, Koyanagi T, Nakahara H, Hara K, Hori E, Nakano H. Functional development of human eye movement in utero assessed quantitatively with real-time ultrasound. Am J Obstet Gynecol 1986;155:170–4. [13] Koyanagi T, Nakano H. Functional development of the fetal central nervous system. In: Levene MI, Lilford RJ, editors. Fetal and neonatal neurology and neurosurgery, 2nd ed. London: Churchill Livingstone, 1995:31–44. [14] Lavie P, Kripke DF, Hiatt JF, Harrison J. Gastric rhythms during sleep. Behav Biol 1978;23:526–30. [15] Liu L. Box–Jenkins time series analysis. In: Dixon WJ et al., editors. BMDP statistical software manual. Berkley: University of California Press, 1992:435–88. [16] Mclain Jr. CR. Amniography studies of gastrointestinal motility of the human fetus. Am J Obstet Gyncol 1963;86:1079–87. [17] Nagata S, Koyanagi T, Fukushima S, Akazawa K, Nakano H. Change in the three-dimensional shape of the stomach in the developing human fetus. Early Hum Dev 1994;37:27–38. [18] Prechtl HFR. The behavioural states of the newborn infants (a review). Brain Res 1974;76:185–212. [19] Quigley EM. Gastric and small intestinal motility in health and disease. Gastroenterol Clin North Am 1996;25:113–45. [20] Rousseeuw PJ. Least median of square regression. J Am Stat Assoc 1984;79:871–80. [21] Shirozu H, Koyanagi T, Takashima T, Horimoto N, Akazawa K, Nakano H. Penile tumescence in the human fetus at term—A preliminary report. Early Hum Dev 1995;41:159–66. [22] Subcommittee of the Japan Society of Obstetrics and Gynecology. Standard curve of BPD and FL in Japanese fetuses assessed using ultrasound. Acta Obstet Gynecol Jpn 1993;45:391–4. [23] Timor-Tritsch IE, Dierker LJ, Hertz RH, Deagan NC, Rosen MG. Studies of antepartum behavioral state in the human fetus at term. Am J Obstet Gynecol 1979;132:524–8. [24] Tomomasa T, Itoh Z, Koizumi T, Kuroume T. Non migrating rhythmic activity in the stomach and duodenum of neonate. Biol Neonate 1985;48:1–9. [25] Trudinger BJ, Knight BC. Fetal age and pattern of human fetal breathing movements. Am J Obstet Gynecol 1980;137:724–8. [26] Vandaele W. Applied time series and Box–Jenkins Models. Ontario: Academic Press, 1983:257–321. [27] Yoshizato T, Koyanagi T, Nagata S, Takashima T, Fukushima S, Nakano H. Three-dimensional ultrasound image of the fetal stomach: congenital duodenal obstruction in utero. Early Hum Dev 1995;41:39–47.