© 2002 American Thoracic Society
Magnetic Resonance Imaging of Lung Water Content and Distribution in Term and Preterm InfantsDepartment of Paediatrics, Imperial College School of Medicine, Queen Mary's University Hospital, and Robert Steiner MRI Unit, Medical Research Council Clinical Sciences Centre, Hammersmith Hospital, London, United Kingdom; Hospital for Sick Children, University of Toronto, Toronto, Ontario, Canada Correspondence and requests for reprints should be addressed to Professor A. D. Edwards, Department of Paediatrics, Hammersmith Hospital, Du Cane Road, London W12 ONN, UK. E-mail: david.edwards{at}ic.ac.uk
An increase in lung liquid may contribute to respiratory disease in preterm infants. Uneven distribution of lung liquid may cause heterogeneity in the lung disease seen in these infants. We used magnetic resonance imaging to investigate lung water content and distribution in 16 preterm (24–31 weeks) and 9 term infants in the first week of life. Images of lung parenchyma were examined and relative proton density quantified to give an index of lung water. Lung water content and distribution were compared between preterm and term infants, and in preterm infants regional signal distribution between dependent and nondependent lung on T1 weighted images was also compared after turning between prone and supine positions. Relative proton density was higher in preterm than in term lung (p < 0.008) and greater in dependent than in nondependent regions, particularly in the preterm (p < 0.001). Repositioning preterm infants rapidly redistributed signal intensities, with more even distribution lying prone than supine (p < 0.001). Small, low-signal regions were seen in the lung parenchyma in preterm but not in term infants, which may indicate peribronchial fluid or overdistension of compliant lung units. We conclude that lung water content is higher in preterm than in term infants and is associated with gravity-related changes consistent with dependent atelectasis.
Key Words: preterm • neonate • respiratory distress syndrome • magnetic resonance imaging • lung water
The causes of persistent lung dysfunction in preterm infants are poorly understood. Surfactant deficiency makes a significant contribution, and neonatal mortality has been reduced by the routine administration of exogenous surfactant soon after delivery (1). However, further doses of surfactant are usually much less effective, and respiratory morbidity remains significant in the surfactant-treated population (2). Increased lung water content may be an important component of neonatal respiratory disease. During fetal life, fluid is secreted into the lungs, whereas at term, secretion stops and reabsorption occurs rapidly during the latter part of labor (3); this process may be immature in the preterm, leading to higher residual lung fluid content (4). The inflammatory processes that can be detected in the airways of many preterm infants soon after delivery and persistent patency of the arterial duct may also increase lung water (5). The distribution of lung liquid may also be important, as an increased lung water burden may cause more dependent areas of lung to collapse, resulting in uneven ventilation of the lung. An effective method for examining lung water in preterm infants has not previously been available. In adults with acute respiratory distress syndrome, computerized tomography (CT) scanning has been used to look at lung water distribution and has made a significant contribution to the understanding of this disease (6, 7). Unfortunately, this approach is unsuitable as a research tool for preterm infants in view of the large doses of radiation used. Magnetic resonance (MR) imaging allows detailed three-dimensional imaging without the use of ionizing radiation. By altering the way in which the image is obtained it is possible to expose the magnetic properties of T1 recovery (spin-lattice) and T2 relaxation (spin-spin) within different tissues. Proton density–weighted images aim to minimize these magnetic effects. By knowing the signal produced on both proton density and T1-weighted images, it is possible to calculate the relative proton density within tissue. The lung parenchyma contains insignificant concentrations of fat and other hydrogen-bound complexes such that proton density measured by MR is principally related to water content. Proton MR imaging of the adult lung has been challenging due to limited signal caused by low proton density of the lung, blood flow artifacts, and physiological motion from cardiac pulsation and respiration. However, improvements in MR techniques allowing more rapid imaging and ECG/respiratory gating mean that MR is now used in the assessment of a variety of lung conditions in adults (8). Animal and in vitro experiments have also shown that it is possible to quantify lung liquid content (9, 10), and lung water content has been examined by MR in the normal adult lung (11) and in some animal models of pulmonary edema (12–14) The preterm infant has not been imaged due to both the practical and technical difficulties of obtaining such images. However, we have established within our Neonatal Intensive Care Unit a 1.0T neonatal MR system (Marconi Medical Systems/Oxford Magnet Technology), which has full intensive care facilities (14). This has made it possible to obtain MR images of very premature infants safely and without physiologic disturbance (15). To study preterm pulmonary dysfunction, we have obtained multislice transverse images of lung in preterm and term infants, measured regional MR signal, calculated relative proton density, and investigated the effects of gravity on these variables by changing the infants' recumbent position. The study addressed the questions: is lung water content increased in preterm infants compared with normal term infants, and is any increase distributed nonuniformly with associated gravity-related atelectasis? We also used the images to look for changes within lung parenchyma consistent with edema, atelectasis, or tissue injury.
Patients The Hammersmith Hospitals Research Ethics Committee gave approval for MR studies of preterm infants. Informed parental consent was obtained for each patient. Table 1 summarizes the main clinical details of the infants studied. At the time of scan, infants receiving mechanical ventilation required a mean airway pressure of 8–11 cm H2O, fraction of inspired oxygen (FIO2) 0.25–0.5, and infants receiving nasal continuous positive airway pressure required pressures of 6–8 cm H2O, FIO2 0.21–0.55. Self-ventilating preterm infants had received a median 22 hours (range, 2 hours to 3 days) ventilatory support before scanning but had no oxygen requirement at the time of scan. All infants were at or below their birth weight and none were clinically edematous. All term controls recruited from the postnatal wards had had no respiratory symptoms or oxygen requirement since birth.
Image Acquisition T1- and proton density–weighted images were obtained with the infants lying supine or prone according to their position before scanning. Preterm infants were then turned and further T1-weighted images acquired within 15 minutes of turning and repeated 10 to 20 minutes later. Further information on scanning acquisition, is given in Table 2 .
Image Analysis MR images were reviewed by a radiologist who was unaware of the patients' clinical condition. Focal or regional differences in the appearance of the lung parenchyma and pleura were described. Chest radiographs were not obtained as part of the study protocol because the additional radiation exposure was ethically unacceptable. However, those obtained for clinical purposes within 24 hours of the MR imaging study were compared with the MR findings. For quantitation, circular regions of interest (6–10 pixels in diameter) were selected in comparable dorsal and ventral lung from a single anatomical slice at the insertion of the pulmonary veins into the left atrium. Signal intensity on T1- and PD-weighted images were combined to give a value for relative proton density (as an index of water content). Additional information for the method used to calculate proton density is given in the online data supplement. Relative proton density was compared in term and preterm infants, and regional signal intensity and relative proton density distribution between dependent and nondependent lung was compared. T1 signal intensity distribution was compared before and after turning preterm infants between supine and prone positions.
Statistics
Image Acquisition No complications occurred during or immediately after the MR studies. A full set of data was not acquired on all infants because it was sometimes clinically inappropriate to turn the infant or prolong the examination, or because the infants could not be settled. Of the 16 preterm infants, 12 were initially imaged supine and 4 were imaged prone. A total of 11 were imaged both supine and prone, 4 were imaged supine only, and 1 was imaged prone only. Of those imaged both prone and supine, five underwent a third examination 10 to 20 minutes later. Data for calculations of relative proton density were obtained in 13 preterm infants, all lying supine. All control infants were studied lying supine, and data for calculations of relative proton density were obtained in seven of these control infants.
MR Image Analysis
Signal intensity was increased in dependent relative to nondependent lung tissue on initial images in both preterm (p < 0.0001) and term (p < 0.0007) infants. This increase was more apparent in basal than in apical regions; it was also more apparent in preterm than in control subjects. In 11 preterm infants, increased signal intensity in dependent lung tissue was associated with an irregular pleural margin that lacked clear definition between lung parenchyma and pleura; all term infants had thin, clearly defined pleural margins (Figure 1). Calculated relative proton density was greater in dependent than nondependent regions in both preterm (p < 0.0002) and term (p < 0.003) infants, but the gradient from the nondependent to dependent regions was greater in preterm infants because proton density was particularly high in dependent preterm lung (p < 0.0002). These data are shown in Figure 2. Preterm infants scanned before and after turning from supine to prone positions demonstrated a significant change in signal distribution after repositioning (p < 0.0001). The signal intensity gradient seen was reversed after repositioning in six infants and reduced in five (Figures 3 and 4) . Signal distribution was more even with lower dorsal (p < 0.008) and greater ventral (p < 0.03) signal intensity when subjects were in the prone position than in the supine position. There was no further change in the small number of infants who underwent repeated scanning 10–20 minutes later.
Seven preterm infants had regions of focal segmental consolidation that were unrelated to posture and unaffected by changes in position (Figure 5) . In 11 preterm infants, but no term infants, multiple circular regions of decreased signal intensity were present: some were nearer the hilar regions and surrounded by a high signal halo (Figure 5); others were located mainly in nondependent lung tissue, particularly at the apex, and usually multiple (Figure 6) .
Thirteen infants had chest radiographs performed within 24 hours of the MR scan. A gravity-dependent gradient in MR signal intensity was seen in 12 infants but could not be detected on chest radiograph, although 10 of these infants had a diffuse ground glass appearance on chest X-ray. Focal segmental consolidation was seen in seven infants on MR and in six infants on chest radiographs. Focal low-signal regions were seen on MR in eight infants, predominantly in the upper lobes, and comparable lesions were identified on the chest radiograph of one infant.
This study shows that MR image–visible proton density is higher and more unevenly distributed throughout the lungs of preterm infants compared with term infants. In the absence of significant concentrations of fat or other hydrogen-bound complexes in the lung, high proton density provides strong evidence of higher lung water content in the the lungs of preterm infants. Liquid is secreted into the lung during fetal life, and its reabsorption during normal parturition through the opening of amiloride-sensitive epithelial sodium channels (ENaC) is essential for air breathing (3, 16, 17). This process is deficient early in gestation in animals, and probably in humans (18, 19). Lung water content is thus thought to be high in preterm infants. Lung water may also be increased due to inflammatory edema. Intrauterine infection is a common antecedent of preterm birth and is associated with lung disease (20). Proinflammatory cytokines have been detected in the amniotic fluid (21), umbilical cord blood (22), and airways and bronchoalveolar lavage fluid (23) of preterm infants (5). Oxygen toxicity and mechanical ventilation, particularly alveolar over-distension, may also induce inflammation (24). Other factors that might increase lung fluid include the effects of a patent arterial duct and the administration of excessive intravenous fluids. Lung water was distributed in a gradient of increasing density from nondependent to dependent lung. The gradient was small in term control infants and consistent with data from adult studies (25), which suggests a slight gravity-induced gradient in blood volume (26) and decreased alveolar size in dependent zones (27). However, in preterm infants, the gradient in lung fluid density was very much greater. Two aspects of the present study support the conclusion that the increased lung water burden was routinely associated with airway collapse and atelectasis. First, MR images showed loss of definition between lung parenchyma and pleura in dependent regions, consistent with localized collapse. Second, turning infants from the supine to the prone position led to a rapid reduction in the density of previously dependent lung regions. The short time course is best ascribed to reopening of closed (previously dependent) alveoli, although movement of intraalveolar edema may also contribute to the changes. Routine dependent atelectasis may be explained by high pulmonary fluid content, which magnifies the hydrostatic compressive force induced by gravity. This amplifies the effects of the pleural pressure gradient and ratio of closing volume to functional residual capacity, both of which are already increased by smaller body size (28). Analogous increases in density in dependent lung regions have been demonstrated in acute respiratory distress syndrome using computed X-ray tomography (6). In acute respiratory distress syndrome, turning from the supine to the prone position redistributed dependent densities with an increase in oxygenation (6, 29). The prone position may have intrinsic advantages: the lungs no longer support the weight of the heart or abdominal contents, the pleural pressure gradient is more evenly spread, and lung recruitment is improved (30). Preterm infants show improved oxygenation and increased lung compliance when placed in the prone position (31, 32). The observation that nonuniform lung density with dependent atelectasis is very common in preterm infants has significant clinical implications. Nondependent regions are compliant and at risk of overdistension resulting in inflammation and injury, while repetitive opening and closing of atelectatic lung units can also initiate inflammatory processes and damage alveolar architecture (33). Many infants in the present study had small regions of low signal intensity within the lung parenchyma, which may represent overdistension of compliant air spaces or peribronchial edema. The importance of alveolar overdistension and ventilator-induced lung injury has recently been highlighted in the National Institutes of Health Acute Respiratory Distress Syndrome Network trial, where low tidal volumes with positive end-expiratory pressure resulted in 25% fewer deaths (34). The infants studied had mild respiratory disease with modest ventilatory requirements, which suggests that even those without severe respiratory distress may be vulnerable to injury. This may be a partial explanation for the observation that chronic lung disease frequently develops in infants who have mild respiratory symptoms initially (35). Thus far we have only applied multi-slice MR techniques and have worked with relative rather than absolute signal values. There is scope for developing MR methods tailored to the examination of the neonatal lung, reducing artifacts, determining T1 and T2 relaxation times, lung volume, and total water content in absolute units. Ventilation–perfusion scanning and assessment of alveolar size and permeability may also be possible (36, 37). In conclusion, MR imaging demonstrated that despite surfactant treatment, the premature infant routinely faces an excessive and nonuniform lung water burden with marked dependent atelectasis. This gravity-related heterogeneity may make the lung vulnerable to damage due to over- and underdistension of alveoli.
The authors are grateful to Dr. L. Al-Nakib, Dr. A. Herlihy, Mrs. J. M. Allsop, and the staff of the Neonatal Intensive Care Unit, Hammersmith Hospital, London, United Kingdom, for invaluable assistance. They also acknowledge and thank infants in the study and their parents for their essential involvement in this project. Support was provided by SpARKS, The Medical Research Council, Marconi Medical Systems, and the Garfield Weston Foundation.
This article has an online data supplement, which is accessible from this issue's table of contents online at www.atsjournals.org Received in original form April 25, 2001; accepted in final form May 2, 2002
This article has been cited by other articles:
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
TOP
ABSTRACT
INTRODUCTION
