Microangiography and Vascular Permeability of the Subependymal Matrix in the Premature Infant

LE JOURNAL CANADIEN DES SCIENCES NEUROLOGIQUES

SACHIO TAKASHIMA AND KENZO TANAKA

SUMMARY: The microvascular ar Hemorrhage into the subepen MATERIALS AND METHODS chitecture of the subependymal matrix in dymal matrix, the periventricular /. Postmortem cerebral premature infants was studied with mic white matter or the choroid plexus is microangiography roangiography and benzidine stains. the source of intraventricular This reveal"d that the subependymal Postmortem cerebral microan hemorrhage in premature infants. giography was done on 30 neonates matrix is the end zone or the border zone Up to 32 weeks gestational age, the between cerebral arteries and the collec and 16 children without intracranial tion zone of the deep cerebral veins. most frequent sites of subependymal pathology and on 7 neonates with Focal hypoxic changes of this subepen hemorrhage are the striae terminalis subependymal hemorrhage ruptured dymal matrix may occur in hypoxemia and the head of the caudate nucleus. into the lateral ventricles. During the and ischemia because of the characteris The pathogenesis of subependymal autopsy, a colloidal solution of tic architecture. hemorrhage has been investigated barium sulfate and gelatin was hand The vascular permeability of these by many neonatologists and injected into the cerebral arteries vessels was studied in rabbits using three pathologists. Predisposing factors through the aortic arch for arteriog different molecular weights of FITC- such as the large subependymal mat raphy, or into cerebral veins through dextran. Vascular permeability was in rix (Riickensteiner et al,, 1929; creased in the subependymal matrix by the superior vena cava for veno Gruenwald, 1951; Saito et al., 1970) graphy. The pressure required for hypoxia and especially by hypoxia as thin vessel walls (Gruenwald, 1951), sociated with an increased venous pres some injections was measured and sure. These findings may be related to hemodynamic factors concerning found to be 120-150 mm Hg. The the pathogenesis of subependymal the internal cerebral venous system brains were removed and suspended hemorrhage in prematurity. (Larroche, 1964) and fibrinolytic ac in 10% formalin. The fixed brains tivity (Gilles et al., 1971; Takashima RESUME: L'architecture microvas- were cut coronally at 2 cm intervals and Tanaka, 1972) have been sug and these sections were radio culaire de la matrice sousependymaire gested. Various causal factors such des enfants prematures fut etudiee en graphed by soft X-ray technique. The combinant la microangiographie et les as hypoxemia (Grontoft, 1958), sections were then cut at 1 cm and colorations a la Benzidine. Ces etudes trauma or compression (Schwartz, 0.5 cm intervals and radiographs ont revile que la matrice 1961; Ross et al., 1965), venous made of each slice. These sections sousependymaire est situee au bout on a stasis (infarction; Towbin, 1968, in were cut again (500 to 1,000^) and la limite des arteres cerebrates pro- creased venous pressure; Cole et al., stained with benzidine. This stain fondes. line hypoxie focale de cette mat 1974), venous thrombosis (Lar rice sousependymaire peut ainsi etre gave a black color to the vessels roche, 1964), disseminated intravas containing blood and the microar trouvee dans I'hypoxemic et I'ischemie a cular coagulation (Gray et al., 1968), cause de cette architecture particuliere. chitecture of the veins and capil La permeabilite vasculaire de ces vais- hemorrhagic diathesis (Hemsath, laries could be differentiated from seaux fut etudiee chez le lapin en em- 1934; Applayard, 1970), hypernat- the arteries which were filled with ployant du FITC-Dextran de 3 differents remia (Simmons et al., 1974; Rober- barium solution. Venograms were poids moleculaires. La permeabilite ton et al., 1975) and increased arter made when barium sulfate was in vasculaire etait augmentee dans la mat ial pressure (Hambleton et al., 1976) jected into the superior vena cava. rice sousependymaire par i hypoxie et have been defined. However, the sourtout par I'hypoxie associee a une pathogenesis of subependymal 2. Cerebrovascular permeability in augmentation de la pression veineuse. hemorrhage remains unsolved Ces resultats ont peut-etre rapport avec fetal rabbits using fluorescein la pathogenese de I'hemorragie (Leech et al., 1974; Roberton, 1977). labelled dextrans as tracers sousependymaire de la prematurite. This is a report of the microan Pregnant rabbits weighing giographie findings in normal infants 3,300-4,000 g had Cesarean sections and premature infants with sub during which two-thirds of the From the Departments of Pediatrics and Pathol ependymal hemorrhage. Permeabil uterine arteries and veins were li- ogy, Kyushu University, Japan, 3-1-1 Maedashi, Fukuoka, Japan. ity of vessels in the subependymal gated for 40 min. before emptying matrix was also studied in fetal rab the uterus. This was done at the Reprint requests to Dr. Sachio Takashima, De partment of Pediatrics, Kyushu University, 3-1-1 bits with or without hypoxia and gestational ages of 24-26 (Group I) Maedashi, Fukuoka 812, Japan. increased venous pressure. and 28-30 days (Group II) (full term

Vol. 5, No. 1 FEBRUARY 1978 - 45 Downloaded from https://www.cambridge.org/core. IP address: 170.106.35.234, on 30 Sep 2021 at 21:37:58, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0317167100024744 THE CANADIAN JOURNAL OF NEUROLOGICAL SCIENCES

=31 days) to produce the hypoxic thalamus. In the neonates and chil fetuses. The 22 control cases were dren the distal portions of some long delivered before the uterine vessels perforating arteries bent at the bor were tied. Eight of 26 hypoxic fetal derline between the caudate nucleus rabbits were dead "in utero" (6 in or thalamus and subependymal mat group I and 2 in group II) and were rix, and branched into the subepen excluded. Arterial acid-base studies dymal matrix adjacent to the caudate were done on 3 of the living hypoxic nucleus and thalamus (Fig. 2). Small fetuses and the average values were branches of the ventriculofugal ar pH 7.01, PO2 17.5 and PCO2 67.3 teries, consisting of the lateral striate mmHg. As soon as the umbilical arteries, the choroidal arteries and vessels of the live fetuses were tied, some medullary arteries from the 0.4 ml of 5% solution of FITC- gyrus cinguli and gyri insulae spread dextran 3(MW = 3,000), 20 (MW into the subependymal matrix in the = 20,000) or 150 (MW = 150,000) deep white matter (Fig. 2, 3). There (Olsson et al., 1975; Schroder et al., fore, the subependymal matrix is the 1976) was injected into the sub terminal area for the thalamo-striate cutaneous cervical vein for 1 to 2 arteries and medullary arteries. The min. and the rabbit sacrificed 5 mi striae terminalis located between the nutes later. The whole brains were Figure 1—Arteriography of the occipital caudate nucleus and thalamus is the fixed in 10% formalin and then cut lobe in a premature infant (34 gesta border zone between the striate ar coronally. Sections of the frontal, tional weeks, hyaline membrane dis teries and the thalamic arteries (Fig. ease without intracranial pathology). 2). patietal and occipital lobes, cere The long medullary arteries extend to bellum and brain stem were mounted the subependymal matrix. The arteriography of the cases in paraffin, cut at 30 u, and then with subependymal hemorrhage viewed under a fluorescence mic (Fig. 1). With maturity the subepen which had ruptured into the lateral roscope. dymal matrix became narrower and ventricle showed an avascular area Six rabbits of 28 days gestation the vasculature appeared flattened. in the region of the striae terminalis with or without fetal hypoxia were The perforating arteries of the at the site of the subependymal injected via the cervical vein with striate and thalmic arteries branched hemorrhage (Fig. 3). In addition, 0.4 ml of 5% FITC-dextran solution from large arteries at the base and leaking, poorly filled and tortuous and then had 30 second intermittent extended to the basal ganglia and vessels were seen around the edge of clamping of the superior vena cava for 15 minutes together with artificial ventilatory support. The cerebral vascular permeability in those ani mals with intermittent clamping was compared to those without clamp ing. RESULTS 7. Postmortem Cerebral Microangiography Cerebral arteriography of the normal brains showed the long medullary arteries to be branching from the meningeal arteries and ex tending to the deep white matter and subependymal matrix. In all neon ates and children the distal portions of some long medullary arteries curved at the borderline between subependymal matrix and cerebral white matter and ran parallel to the ventricular wall. They then branched into subependymal matrix and the deep white matter. In the premature neonates some of the long medullary arteries extended into the Figure 2— Arteriography of the cerebrum in a one-year-old infant (acute broncho ventricular wall, and returned to pneumonia without intracranial pathology). The arterial border zone between the spread into subependymal matrix striate arteries and the thalamic arteries is apparent.

46 - FEBRUARY 1978 Subependymal Vasculature in Infants Downloaded from https://www.cambridge.org/core. IP address: 170.106.35.234, on 30 Sep 2021 at 21:37:58, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0317167100024744 LE JOURNAL CANADIEN DES SCIENCES NEUROLOGIQUES

Figure J—Arteriography of subependymal hemorrhage which Figure 4—Venography of the deep cerebral hemisphere in the has ruptured into the lateral ventricle in the premature infant premature infant without intracranial pathology. "V" indi (32 gestational weeks). Large arrow indicates subependymal cates lateral ventricle, ^terminal vein and|4thalamostriate hemorrhage, and small arrow indicates leaking lesion. veins.

the avascular area. In the deep white to form the internal cerebral veins matter, small leaking lesions were (Fig. 4, 5). The internal cerebral seen in 6 of 7 neonates with sub veins passed backward in the roof of ependymal hemorrhage (Fig. 3). the third ventricle. Cerebral venography of normal In microangiography of the brains revealed that the deep white neonatal normal brains, stained by matter was drained by a fan-shaped benzidine after barium solution was leash of short and long medullary injected into the cerebral arteries, veins, which flowed vertically into the capillary networks showed dif subependymal veins in the subepen ferent patterns in the caudate nuc dymal matrix. Some of them ran leus and thalamus, cerebral white parallel to the ventricular wall in the matter and subependymal matrix. subependymal matrix, and then The pattern of capillary network was drained retrograde into the subepen fine and almost circular in the caud Figure 5—Microvenography of periven dymal veins (Fig. 4, 5). ate nucleus and thalamus (Fig 6). In tricular white matter of the normal the cerebral white matter the net premature brain. The numerous The subependymal veins (terminal medullary veins flow into the sub veins) draining many small veins work was perpendicular to the ven ependymal veins in the subependymal from the subependymal matrix, tricular wall and larger and elliptical matrix which also receive drainage joined anteriorly with the septal, (Fig. 7). The capillary pattern in the from the small veins from the sub choroidal and thalamostriate veins, subependymal matrix was elliptical ependymal matrix. "V" indicates lat at the level of the Foramen of Monro and parallel to the ventricular wall eral ventricle.

Takashima and Tanaka FEBRUARY 1978 - 47 Downloaded from https://www.cambridge.org/core. IP address: 170.106.35.234, on 30 Sep 2021 at 21:37:58, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0317167100024744 THE CANADIAN JOURNAL OF NEUROLOGICAL SCIENCES

spaces of the control brains, but snowed slight permeability in the subependymal matrix of hypoxic brains (Fig. 9, Table 4). In the brain of a 28 gestational day rabbit exposed to intermittent clamping of the superior vena cava, there was marked permeability of FITC-dextran 20 into the subepen dymal matrix (Fig. 10). The permea bility in rabbits exposed to hypoxia Figure 9—Subependymal matrix and plus increased venous pressure was white matter of an hypoxic fetal rabbit higher than in rabbits with hypoxia or (26 gestational days) injected with increased venous pressure alone. FITC-dextran 150. Fluorescence mic Histological examination showed rograph. slight congestion and small hemor rhages in the subependymal matrix of the hypoxic cases (Fig. 8). There were no ventricular hemorrhages. The incidence of these hemorrhages

Figure 6—Microangiography of the caudate nucleus and subependymal matrix of the normal premature brain. Benzidine stains after arteriography. "V" indicates lateral ventricle. Figure 10—Subependymal matrix and white matter of a fetal rabbit (28 gesta tional days) with hypoxia and in (Fig. 6, 7). The size of the latter creased venous pressure. (FITC- network was intermediate, between dextran 20 injected). Fluorescence the caudate nucleus or thalamus and micrograph. the cerebral white matter. was higher in the hypoxic fetal 2. Fetal Rabbit vascular permeability brains (61.1%) than the control with fluorescein labelled dextran brains (13.6%) (Table 1). tracers The vascular permeability in the DISCUSSION subependymal matrix and the deep In the subependymal matrix, the white matter varied according to .the subependymal veins joined by septal molecular weight of the tracer. Figure 7—Microangiography of the deep choroidal, thalamostriate and post FITC-dextran 3 showed the greatest white matter and subependymal mat erior terminal veins flow towards the permeability in both the subepen rix of the normal premature brain. Foramen of Monro. The direction of dymal matrix and the deep white Benzidine stains after arteriography. flow changes at an acute angle at the matter. Permeability was higher in level of the Foramen of Monro to hypoxic brains than in the control form the two internal cerebral veins. brains. The deep white matter of Larroche (1964) felt that the risk younger brains (24 and 26 gestational of subependymal hemorrhage was days) showed higher permeability than related to hemodynamic distur that of more mature brains (28 and 30 bances (stasis and thrombosis) at the gestational days) (Fig. 9, Table 2). site of maximal flow near the Fora FITC-dextran 20 was barely rec men of Monro and at the head of the ognized in the perivascular space of caudate nucleus. the subependymal matrix and cere Hambleton et al. (1976) reported bral white matter in some of the that newborn brains up to 32 weeks younger brains. The permeability in gestational age have a rich capillary hypoxic brains was higher than in bed in the subependymal region. control brains (Table 3). Figure #—Subependymal hemorrhage in The important arteries are FITC-dextran 150 was not recog the hypoxic tetal rabbit following liga Heubner's artery, a branch of the nized at all in the perivascular tion of the uterine vessels. H&E X 72. anterior cerebral artery, lateral

48 - FEBRUARY 1978 Subependymal Vasculature in Infants Downloaded from https://www.cambridge.org/core. IP address: 170.106.35.234, on 30 Sep 2021 at 21:37:58, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0317167100024744 LE JOURNAL CANADIEN DES SCIENCES NEUROLOGIQUES

TABLE 1 may be related to the border zone Incidence of Subependymal Hemorrhage in theory of Lindenberg (1959). Hypoxic Rabbit Fetal Brains By microangiography, the sub ependymal matrix shows a charac TOTAL GESTATIONAL AGE (Days) teristic capillary network different I II from the cerebral white matter and 24-26 28-30 the basal ganglia and thalamus % % % (Takashima and Tanaka, 1972). The Hypoxic brains 11 61.1 2 28:6 9 81.8 subependymal veins run parallel to the ventricular wall, draining the 18 7 11 medullary veins and very small veins Control _3_ 13.1 1 10.0 1_ 16.7 in the subependymal matrix. Hamb- leton et al. (1976) found that the 22 10 12 capillary channels often open at right Number of animals with hemorrhage angles directly into the veins in the Number of animals examined subependymal matrix. TABLE 2 The subependymal matrix is prominent in premature infants and Vascular Permeability of FITC-Dextran 3 in has histological characteristics of Rabbit Fetal Brain (MW = 3,000) rich cellularity, poor connective tis Gestational Age (Days) sue and thin .vascular walls (Gruen- 24 26 28 30 wald, 1951). Electron microscopic Cont. Anoxia Cont. Anoxia Cont. Anoxia Cont. Anoxia examination of vessels in the sub - + + + + Cerebral Cortex + _ ependymal matrix indicates that the Cerebral White Matter ± ++ ± ++ ± + ± + pericapillary spaces are surrounded ++ + by primitive germinal cells. The foot Subependymal Matrix ± ± + ± ++ ± ++ processes of astrocytes develop Basal Ganglia _ + + + + around the capillaries in preterm — absent rabbits (Tanaka et al., 1977). Grun- + focal, slight net (1977) has demonstrated by EM + moderate deficiencies in the vascular integrity ++ marked at different ages. TABLE 3 Grontoft (1958) studied the blood Vascular Permeability of FITC-Dextran 20 in brain barrier (BBB) of the fetus and Rabbit Fetal Brain (MW= 20,000) newborn. A BBB impermeable to trypan blue is present in human Gestational Age (Days) 24 26 28 30 fetuses 5 to 30 cm long. It can be Cont. Anoxia Cont. Anoxia Cont. Anoxia Cont. Anoxia easily damaged by anoxia, but its resistance to anoxia increases with __ Cerebral Cortex + _ — + the development of the fetus. Cerebral White Matter _ + — + — + Cole et al. (1974) found the Subependymal Matrix — -r — + ± + amount of lower molecular weight Basal Ganglia + ± — — _ + protein in cerebrospinal fluid in creased in cases with subependymal - absent hemorrhage, and suggested this + focal, slight might be due to the increased venous +moderate or capillary pressure before ven tricular rupture. In our study of fetal rabbits the vascular permeability of striate artery and anterior choroidal as shown in Fig. 2 and 3. Small FITC-dextran 20 was increased in artery. Our work with cerebral mic branches of the ventriculofugal ar the cerebral white matter, especially roarteriography shows that the sub teries in the deep white matter also into the subependymal matrix of ependymal matrix is at the end site spread into the subependymal mat hypoxic brains. In the fetal brains of the thalamostriate arteries and rix, although these arteries are with hypoxia plus increased venous medullary arteries (Takashima and poorly developed in premature in pressure, the vascular permeability Tanaka, 1971). The striae terminalis fants (Takashima and Tanaka, in the subependymal matrix was — a common site of early hemor 1975). Therefore, it seems reasona even greater. rhage and subependymal pseudocyst ble to suggest that focal ischemia in In the pathogenesis of subepen — is a border zone between the the subependymal matrix, especially dymal hemorrhage in the premature, striate arteries and thalamic arteries, in the region of the striae terminalis, the histological immaturity and the

Takashima and Tanaka FEBRUARY 1978 - 49 Downloaded from https://www.cambridge.org/core. IP address: 170.106.35.234, on 30 Sep 2021 at 21:37:58, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0317167100024744 THE CANADIAN JOURNAL OF NEUROLOGICAL SCIENCES

TABLE 4 LINDENBERG, R. (1959). The pathology of the arterial borderzones of the brain. J. Vascular Permeability of FITC-Dextran 150 in Neuropath. Exper. Neurol. 18: 348-349. Rabbit Fetal Brain (MW= 150,000) OLSSON, Y., SVENSJO, E., ARFORS, K. Gestational Age (Days) E. and HULTSTROM, D. (1975). Fluores 24 26 28 30 cein labelled dextrans as tracers for vascu lar permeability studies in the nervous sys Cont. Anoxia Cont. Anoxia Cont. Anoxia Cont. Anoxia tem. Acta Neuropath. 33: 45-50. Cerebral Cortex — — — — — — ROBERTON, N. R. C. and HOWAT, P. Cerebral White Matter _ + + — _ + (1975). Hypernatraemia as a cause of in tracranial haemorrhage. Arch. Dis. Child. Subependymal Matrix _ + + + + ± 50: 938-942. Basal Ganglia - — — — - — ROBERTON, N. R. C. and HOWAT, P. (1977). Intraventricular hemorrhage and - absent alkali therapy (correspondence). Arch. Dis. ± focal, slight Childh. 52: 248-249. + moderate ROSS, J. J. and DIMMETTE, R. M. (1965). Subependymal cerebral hemorrhage in in vascular architecture of subepen COLE, V. A., DURBIN, G. M., OLAFF- fancy. Am. J. Dis. Child. 110: 531-542. dymal matrix are important predis SON, A., REYNOLDS, E. O. R., RIV RUCKENSTEINER, E. and ZOLLNER, F. ERS, R. P. A. and SMITH, J. F. (1974). (1929). Uber die Blutungen im Gebiete der posing factors. The subependymal Pathogenesis of intraventricular haemor Vena Terminalis bei Neugeborenen. Frank matrix is a border zone or end zone rhage in newborn infants. Arch. Dis. Child. furt, z. Path. 37: 568-578. and hence is subject to ischemia in 49: 722-728. SAITO, O. and MURATA, B. (1970). any hypoxemia or hypotensive GILLES, F. H., PRICE, R. A., KEVY, S. V. Neonatal intracranial hemorrhage; in event. This leads to a localized ven and BERENBERG, W. (1971). Fibrinolytic traventricular hemorrhage. Pediatrics ous stasis, increased vascular per activity in the ganglionic eminence of the (Japanese) 11: 734. meability and hemorrhage. Stasis premature human brain. Biol. Neonat. 18: SCHRODER, U., ARFORS, K. E. and 426-432. TANGEN, O. (1976). Stability of and increased vascular permeability GRAY, C. P., ACKERMAN, A. and fluorescein labelled dextrans in vivo and in may also originate from a distur FRASER, A. J. (1968). Intracranial vitro. Microvascular Research, II: 33-39. bance of the deep cerebral venous haemorrhage and clotting defects in low SCHWARTZ, P. (1961). Birth injuries of the system (as shown in the rabbits) so birth weight infants. Lancet, I: 545-548. newborn: morphology, pathogenesis, clini that a retrograde mechanism may GRONTOFT, O. (1958). Intracerebral and cal pathology and prevention. Hafner, New contribute to the rupture of the sub meningeal hemorrhages in perinatally de York. ependymal vessels. This study pro creased infants; intracerebral hemorrhages; SIMMONS, M. A., ADCOCK, E. W., pathologic-anatomical and obstetrical BARD, H. and BATTAGLIA, F. C. (1974). vides some evidence that the blood study. Acta Obstet. Gynec. Scand., 3: Hypernatremia and intracranial hemor supply of the subependymal matrix 308-334. rhage in neonates. New Engl. J. Med. 291: makes it vulnerable to alterations in GRUENWALD, P. (1951). Subependymal 6-10. blood flow. Ischemia and increased cerebral hemorrhage in premature infants, TAKASHIMA, S. and TANAKA, K. (1971). venous pressure in an animal model and its relation to various injuries Postmortem angiography of the neonatal produced increased vascular per influences at birth. Amer. J. Obstet. Gnec. cerebral vascular system. 2. Vascular ar 61: 1285-1292. chitecture of cerebral cortex and white mat meability. This may relate to the GRUNNET, M. L. (1977). Germinal plate ter in the neonates. Acta Neonat. Japonica production of subependymal cell hemorrhage in the premature infant. The 7: 222-228 (Japanese). plate hemorrhages. 53rd Annual Meeting of American Associa TAKASHIMA, S. and TANAKA, K. (1972). tion of Neuropathologists, Chicago. Microangiography and fibrinolytic activity HAMBLETON, G. and WIGGLESWORTH, in subependymal matrix of the premature J. S. (1976). Origin of intraventricular brain. Brain and Development 4: 222-227 haemorrhage in the preterm infant. Arch. (Japanese). ACKNOWLEDGEMENT Dis. Child. 51: 651-659. TAKASHIMA, S. and TANAKA, K. (1975). We are pleased to thank Drs. Dawna Arm HEMSATH, F. A. (1934). Ventricular cere Development of ventriculofugal arteries strong and L. Becker for their careful reading bral hemorrhage in the newborn infants; and periventricular leukomalacia in the of this manuscript and numerous constructive pathological and etiological study of 20 deep white matter. Acta Neonat. Jap. 11: suggestions. cases. Amer. J. Obstet. Gnec. 28: 343-354. 392-394 (Japanese). LARROCHE, J. C. (1964). Hemorrhagies TANAKA, K., TAKASHIMA, S. and cerebrales intraventriculaires chez le SUEISHI, K. (1977). Pathology of neonatal premature. I Anatomie et pathophysiologie. cerebral circulatory disturbances. Pediatric REFERENCES Biol. Neonat. 7: 26-56. clinics, 30: 715-724 (Japanese). APPLAYARD, W. J. and COTTON, D. G. LEECH, R. W. and KOHNEN, P. (1974). TOWBIN, A. (1968). Cerebral intraventricu (1970). Effect of asphyxia on thrombotest Subependymal and intraventricular hemor lar hemorrhage and subependymal matrix values in low birthweight infants. Arch. rhages in the newborn. Amer. J. Path. 77: infarction in the fetus and premature new Dis. Child. 45: 705-707. 465-475. born. Amer. J. Pathol. 52: 121-139.

50 - FEBRUARY 1978 Subependymal Vasculature in Infants Downloaded from https://www.cambridge.org/core. IP address: 170.106.35.234, on 30 Sep 2021 at 21:37:58, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0317167100024744