All experimental procedures were approved by the Monash University Animal Ethics Committee. Under general anaesthesia (1.5% halothane in O₂), aseptic surgery was performed on seven Border-Leicester X Merino ewes at 105 days gestation (term is ~147 days) as previously described [¹⁴]. A tube (ID 3.2 mm, OD 6.4 mm) was inserted into the fetal trachea and connected, via a Y-piece, to two large bore, saline-filled, ventilator circuit tubes (ID 9.5 mm, OD 14.3 mm). An additional saline-filled catheter (ID 3.2 mm, OD 6.4 mm) was inserted into the fetal upper trachea and connected to a saline-filled ventilation tube to create an exteriorised tracheal loop, allowing the normal flow of liquid into and out of the lung (Figure 1). A 4-mm ultrasonic flow transducer (Transonic Systems; Ithaca, NY) was placed around the left pulmonary artery (31), this probe directly measures blood flow to the left side of the lung (blood flowing from the right ventricle or shunting left-to-right across the ductus arteriosus). A small securing tie was then inserted into the wall of the vessel and a tapered polyvinyl catheter (BD Insyte Vialon, peripheral venous catheter, length: 48 mm, 0.03 mm ID, 0.041 mm OD, NJ USA) was inserted into the main pulmonary artery, by direct puncture. It was directed 2 cm down into the left pulmonary artery before it was secured into place using the securing tie. Polyvinyl catheters were placed in a fetal carotid artery, jugular vein and amniotic sac and exteriorised via the ewe's flank. Ewes and fetuses were allowed 5 days recovery following surgery. Fetal arterial blood PaO₂, PaCO₂, pH and SaO₂ were measured every second day to assess fetal wellbeing.

Carotid and pulmonary arterial and amniotic sac pressures (DTX, Viggo-Spectramed, California) and blood flow through the left pulmonary artery (LPBF) were recorded digitally at 1 kHz (Powerlab, ADI: Castle Hill, Australia) for 6 hours prior to starting in utero ventilation (IUV). Mean LPBF was calculated electronically from the instantaneous LPBF signal, the LPBF flow measures right ventricular output as well as the left to right shunting through the ductus arteriosus. Prior to IUV, the upper tracheal catheter was disconnected from the ventilator circuit and the liquid within the circuit passively drained by gravity into a sterile bag before connection to a neonatal ventilator (Draeger 8000+). Fetuses were ventilated in utero at 110 d gestation for 12 hrs (n = 7) as described previously (briefly PIP of 40 cmH₂O, PEEP of 4 cmH₂O, flow 10 L/min and a rate 60 inflations/min; FiO₂ 21%); each fetus acted as its own control. However, operated age-matched control fetuses were also used for comparison of blood-gases (110 d control fetuses) and fetal weight data (117 d control fetuses). Arterial blood gases were measured hourly. Following IUV, the ventilator circuit and fetal lung were refilled with lung liquid and reconnected to the upper tracheal catheter, to restore normal lung liquid flow [¹⁴]. Carotid arterial and amniotic sac pressures as well as PBF were recorded digitally for 6 hours after the cessation of IUV and ewes and fetuses were killed 7 days later (117 d GA) for post-mortem analysis.

Changes in the contour of the PBF waveform were measured by selecting waveforms from 10 consecutive cardiac cycles from each lamb at selected time points during experimentation[¹¹,¹⁶]: before IUV (pre-IUV) and then every 5 min for the first two hours of IUV and at 20 min intervals for the remaining 10 hours of IUV. Waveform parameters and calculations of pulsatility index (PI) have been described previously [¹¹].

PBF and PAP measurements represent an average value taken over a one minute period of recording with care being taken to avoid periods containing obvious movement artefacts caused by the ewe. Measurements of PBF, systemic arterial pressure (SAP) and PAP are expressed as mean ± SEM. Heart rate is expressed a percentage increase from the mean heart rate during the pre-IUV recording period. All values were then grouped and means and standard errors determined. Comparisons of PBF, HR, SAP, PAP and individual components of the PBF waveform were analysed using a two-way repeated measures ANOVA using the statistical package Sigma Stat (Version 3.1.1, Jandel Corporation, USA). The level of significance was p < 0.05 for all statistical analyses.

All fetuses had normal blood gas and acid-base status throughout the experimental period. Although all values were within normal ranges, the pH values in IUV fetuses were significantly higher than in age-matched controls (Figure 2). No time dependent differences were observed in blood gas and acid base status when compared to age-matched controls. No significant differences were observed between fetal body weights and lung/body weight ratios at post mortem compared to age-matched controls (117 d control; Table 1).

PBF in the left pulmonary artery tended to increase from pre-IUV control values of 6.9 ± 4.7 mL/min to 16.3 ± 6.1 mL/min in response to lung liquid drainage, although this was not statistically significant. The onset of IUV further, and significantly (p < 0.001), increased PBF from 16.3 ± 6.1 mL/min to a maximum value of 47.2 ± 19.1 mL/min at 10 minutes of IUV (Figure 3A). The PBF then decreased, despite continued mechanical ventilation, such that at 75 minutes after beginning IUV, PBF was not significantly different from pre-IUV values.

The change in PBF in response to IUV varied markedly between fetuses, particularly during the first 90 minutes of IUV, explaining the high variability in PBF displayed in Figure 3A. Based upon the PBF response to IUV, individual fetuses have been subdivided into two groups which differed markedly in their PBF response (Figure 3B). Group (i) called "responders" includes all fetuses with an end-diastolic flow >0 ml/min (indicating retrograde flow from the pulmonary arteries had ceased; see Figure 4) and a >50% increase in PBF (above control values) (Figure 3B; responders). Group (ii) called "non-responders", were fetuses who displayed minor or no alterations in PBF (Figure 3B, non-responders), including little change in end-diastolic flow (Figure 4). Of the seven fetuses studied, 4 were responders and 3 non-responders. GA, body weights and lung weights (Table 1) were not different between the groups.

In comparison to non-responding fetuses, responding fetuses markedly increased PBF in response to IUV, increasing from a pre-IUV value of 12.2 ± 6.6 mL/min to a maximum of 78.1 ± 23.1 mL/min at 10 minutes of IUV after which PBF gradually decreased to return to fetal levels at 2 hours of IUV (Figure 3B). Although all fetuses increased PBF in response to IUV in this group, the extent of the increase varied between fetuses during the early stages of IUV. In contrast, no change in PBF occurred in non-responding fetuses throughout the IUV period (Figure 3B). The changes in PBF were significantly different between responding and non-responding fetuses (p < 0.001) from the beginning until 110 mins of mechanical ventilation.

Numerous transient changes in PBF waveform characteristics were noted, although when both groups were combined, only the increases in mean diastolic flow and post-systolic minimum flow at 20 mins of IUV as well as the decrease in pulsatility index at 20 minutes and 2 hours were significant (Table 2). The inability to detect significant changes was largely caused by non-responding fetuses, as they had little or no variation in their PBF waveform (Figure 4; left hand graphs) in response to IUV. In contrast, fetuses that responded to IUV had markedly altered PBF waveforms (Figure 4; right hand graphs) that were similar to what we have shown in more mature fetuses [¹⁷].

As we have shown previously during preterm ventilation [¹⁷], IUV alter diastolic characteristics of the PBF waveform. Most changes were transient and paralleled changes in mean PBF (Table 3). In responding fetuses, mean diastolic flow, post-systolic minimum flow and end-diastolic flow were all significantly increased (Table 3; p < 0.05) from pre-IUV values at 20 minutes of IUV and returned to baseline levels by 2 hrs of ventilation. In 3 out of the 4 responding fetuses, PBF displayed no retrograde flow. PI significantly (p < 0.05), but transiently, decreased in responding fetuses (Table 3). Overall in all parameters measured, the responding fetuses were significantly different to the non-responding fetuses (p < 0.001).

Fetal heart rate tended to increase within 30 mins of IUV, although this increase was not significant either when the groups were considered together (Figure 5A) or following separation into responders and non responders (Figure 5B) due to the high level of variability.

Both the PAP and SAP were not significantly altered throughout IUV (Table 4). Following subdivision into groups based on their PBF response to IUV, neither pulmonary nor systemic arterial pressures were significantly altered by IUV (subgroup data not shown).

Previous studies have shown that reducing lung liquid volumes in near term fetal sheep, to a volume equivalent to FRC in newborn lambs, decreases PVR and increases PBF 2-4 fold [¹⁸,¹⁹]. Similarly, previous IUV studies conducted in near term fetal sheep have shown that IUV causes a sustained decrease in PVR and increase in PBF [⁷,²⁰,²¹]. In this study, lung liquid drainage prior to IUV caused only a small increase PBF, which failed to reach statistical significance, and mechanical ventilation caused a variable increase in PBF in only ~57% of fetuses. Clearly, the difference between the findings of this study and previous studies is the gestational age and lung maturity at the time of lung liquid drainage and IUV. This suggests that in very preterm infants, with a similar level of lung immaturity, the capacity of the lung to increase PBF in response to mechanical ventilation may be considerably less than previously acknowledged and non-existent in some.

The mechanisms responsible for the increase in PBF at birth in mature, near term, fetuses are thought to be multi-factorial and include effects of ventilation, increased oxygenation and the release of vasodilators [^22-24]. NO-induced vasodilation is thought to contribute to ~50% of the increase in PBF at birth [²²] which can be inhibited by blocking NO activity [²⁵]. The mature fetal lung can also vasodilate in response to vasoactive factors, such as prostaglandin D₂ [²⁶] and bradykinin [²⁷] both of which can decrease PVR and release vasoactive substances, such as prostaglandins in response to rhythmic distension [²⁸,²⁹]. However, the ability of the very immature pulmonary vascular bed to release NO and other vasodilators and respond to them is relatively unexplored and warrants further investigation.

Changes in transpulmonary pressures associated with lung aeration also contributes to the increase in PBF with ventilation onset [⁷,³⁰,³¹]. Before birth, the fetal lungs are liquid-filled providing a constant internal distending pressure that maintains the lungs in an expanded state [³²,³³]. Following lung aeration, the distending influence of lung liquid is lost and the presence of air creates an air-liquid interface [³²]. The resulting surface tension, even in the presence of surfactant, increases lung recoil and creates sub-atmospheric intra-pleural and peri-alveolar interstitial tissue pressures [³⁴,³⁵]. This increases both the capillary/alveolar wall and capillary/interstitial tissue wall transmural pressures leading to an increase in capillary recruitment [³⁶] and expansion of recruited capillaries [³¹]. On the other hand, increases in intra-luminal pressure caused by positive pressure ventilation, reduces the alveolar/capillary transmural pressure, capillary calibre and increases PVR. This is a well established relationship that has been demonstrated in adults [³⁷], near term fetuses [⁷], newborns [³⁸] and relatively mature preterm lambs [¹¹].

In adults, the large cross-sectional area of the pulmonary capillary bed allows the pulmonary circulation to accept the entire output of the right ventricle while maintaining pressures at ~1/8 of the systemic circulation. Much of the capillary bed forms (and the cross-sectional area of bed increases) during the alveolar stage of lung development, resulting in a gradual reduction in PVR during late gestation [³⁹]. At birth the large decrease in PVR can only occur if the cross-sectional area of the pulmonary vascular bed is sufficiently large. As very immature lungs will not have undergone this vital development at the time of preterm birth [⁴⁰] the ability of the immature lung to decrease PVR after birth must be diminished, irrespective of whether they have the capacity to dilate in response to vasodilatory stimuli. Indeed the decrease Pulsatility Index (PI) (Table 2 and 3) in fetal sheep following the initiation of mechanical ventilation in the current study indicates that PVR did decrease, although again the increased PI was not maintained for longer than 2 hours. The lack of a pulmonary vascular bed with a sufficiently high cross sectional area may explain why the increase in PBF was not sustained in our study and only occurred in 57% of fetuses.

The contribution of the low resistance placental circulation must also be considered when interpreting PBF changes in response to IUV, as the placental circulation was functional during our studies. We have recently shown that flow through the ductus arteriosus (DA) reverses with mechanical ventilation at birth, due to a reversal of the pressure gradient across the DA [¹⁶]. As a result, the majority of blood flow through the DA immediately after birth, flows from the systemic into the pulmonary circulation (referred to as left-to-right shunting) which contributes ~50% of PBF that gradually decreases with time [¹⁶]. The reversal of the pressure gradient between the systemic and pulmonary circulations [¹⁶,¹⁹] and the onset of left-to-right shunting through the DA, must be determined by both the decrease in PVR and the increase in downstream resistance in the peripheral vascular bed caused by umbilical cord occlusion. As the umbilical cord was not occluded in this study resistance in the systemic circulation would have remained low limiting the change in pressure gradient across the DA.

Heart rate in the responders, although tending to increase, was not significantly altered throughout the experiment. Given that fetal cardiac output is predominately determined by heart rate rather than stroke volume [⁴¹], it is possible that right ventricular output may have contributed to the initial increase in PBF. However, the decrease in PVR (as determined by the pulsatility index) likely plays a greater role in the present study as previously discussed. However, the increase in PBF is likely to have a combination of right and left ventricular output contributions. The responders had positive PBF throughout the cardiac cycle, evident by their respective PBF waveforms (Figure 4), indicating that 100% of right ventricular output is entering the pulmonary circulation. Left ventricular output may contribute to the increase in PBF as evident by the decrease in the PI (Table 4) which would facilitate left-to-right shunting through the DA. The gradual increase in pulsatility index, likely caused by factors which shall be discussed later, would reduce both left and right ventricular output contributions thus decreasing PBF to normal levels.

In responding fetuses, the contour of the PBF waveform changed in response to IUV to closely resemble the waveform seen in post-natal animals [¹⁷], with flow during diastole being particularly affected (Figure 4). Specifically, IUV significantly increased end- and peak-systolic flow as well as end- and mean-diastolic flow, resulting in decreased pulse amplitude in the early stages of IUV. The most significant change was the increase in diastolic flow about 20 mins after mechanical ventilation onset where it remained positive in 3 out of the 4 responding fetuses, indicating blood flowed towards the lung throughout the cardiac cycle at this time. The loss of retrograde flow during diastole is the greatest change to the PBF waveform during fetal to neonatal transition and is a sensitive indicator of a reduced PVR [¹¹]. Thus, the loss of retrograde flow in these very immature fetuses implies that PVR was reduced and that some left-to-right shunting through the DA may have occurred which contributed to the increase in PBF [¹⁶]. As diastolic PBF did not increase in non-responding fetuses, it is unlikely that PVR was significantly reduced or that significant levels of left-to-right shunting occurred in these fetuses in response to IUV.

The two fetal sub-groups had similar arterial pressures and blood gas status before and during IUV, and at autopsy had similar body and organ weights. However, in addition to the PBF changes, responding fetuses increased their heart rate in response to IUV. Although this may have contributed to the increase in PBF, it cannot account for the differences in PBF between subgroups. Indeed, one of the largest changes in PBF in response to IUV was increased diastolic flow, which is largely independent of heart rate and mostly determined by PVR [¹¹]. An increase in fetal heart rate, in the absence of a decrease in PVR, simply results in increased right-to-left flow through the DA. Previous IUV studies conducted in near term fetuses [⁹,¹²], also found that maximal increases in PBF occurred in only half of ventilated fetuses [⁷,⁴²]. Teitel and colleagues (1990) also separated fetuses into major and minor responders (determined by PBF changes) but found no differences between groups other than their response to IUV; they looked at gender, fetal weight, blood gas status, pulmonary vascular tone, ventricular function and alveolar ventilation [⁷]. We consider the primary difference between responding and non-responding fetuses is the effect of IUV on PVR.

It is also possible that fetal posture and the influence of the surrounding amniotic fluid may play a role in the differential PBF responses to IUV. Fetal body position cannot be held constant during IUV, either between animals or throughout experimentation in individual animals. Increased fetal flexion is known to increase abdominal pressure and elevate the fetal diaphragm [⁴³] which could limit the tidal volume and distribution of ventilation within the fetal lung. The net result would likely be a marked reduction in the ventilation-induced increase in PBF.

Despite the period of IUV lasting for 12 hours, the increase in PBF only lasted ~40 minutes in responding fetuses. Transient PBF changes in response to IUV have not been reported previously although this may be because ventilation only lasted for 30 min or less in those studies [^7-9,⁴²]. Attenuation of increases in PBF have been described in fetuses following occlusion of the DA [⁴⁴], exposure to high levels of inspired oxygen [⁴⁵] and to vasodilators [⁴⁶]. Abman and Accurso (1989) have suggested that the immature vascular bed may have a limited ability to release and maintain the necessary vasodilatory factors needed to sustain an increase in PBF. Alternatively, it is possible that PVR progressively increases, after an initial decrease, due to compression of the capillaries from increased interstitial tissue pressure caused by lung liquid retention in the tissue [³⁵] or to oedema caused by injury [⁴⁷]. It is also possible that an increase in vasoconstrictive properties within the vascular bed caused by neural, humoral, local or a myogenic responses contributes to the transient nature for the increase in PBF [⁴⁴]. Shear stress associated with mechanical ventilation is known to elicit myogenic responses [⁴⁸] which the pulmonary vasculature is particularly sensitive to during the perinatal period [⁴⁹].

In summary, our studies indicate that despite significant structural immaturity, the very immature lung can increase PBF in response to mechanical ventilation although changes are transient and highly variable. As previous studies, conducted ex utero, have not demonstrated the same variability this suggests either the degree of lung immaturity or the in utero environment is associated with the variability. Understanding the changes in pulmonary hemodynamics at birth in the extremely immature lung is necessary to improve the care and management of extremely preterm infants.

No statistical differences were present between fetuses that either responded or did not respond to IUV. All measurements were made at the time of post mortem.

Values that do not share a common letter are significantly different from each other (p< 0.05).

Values that do not share a common letter are significantly different from each other (p< 0.05). Overall in all parameters measured, the responding fetuses were significantly different to the non-responding fetuses (p<0.001).