THE ANATOMICAL RECORD 252:92–101 (1998) Computerized Morphometry of the Pulmonary Vasculature Over a Range of Intravascular Pressures JILL A. OHAR,* K.S. WALLER, TANYA J. WILLIAMS, DOUGLAS A. LUKE, AND DAPHNE E. DEMELLO Departments of Internal Medicine, Pathology, and Public Health, St. Louis University School of Medicine, St. Louis, Missouri 63110–0250 ABSTRACT Recent availability of computerized image analysis has fostered hope that barium injection and landmarking of pulmonary arteries would be unnecessary for morphometric assessment when using this technique. We reasoned that if barium injection altered morphometric variables, it would do so in a linear fashion correlating with incremental increases in injection pressure of the barium. The two goals of the present study were to determine whether barium injection into arteries affected morphometric measurements and to determine whether incremental increases in injection pressure correlated with alterations in morphometric measurements in a linear fashion. Computerized image analysis was used to measure the internal elastic lamina (IEL) and external elastic lamina (EEL). Medial area (MA), luminal area (LA), percentage of medial thickness, IELÎMA, and idealized LA were calculated. Barium injection did not alter morphometric variables in a linear fashion correlating with incremental increases in injection pressure of the barium except the percentage of arteries that filled with barium. Maximum recruitment for pre-acinar arteries occurred at 40 mmHg pressure and 60 mmHg distending pressure for intra-acinar arteries. Incremental increases in injection pressure did not affect IEL, EEL, or calculated morphometric variables. However, IEL, medial thickness, and MA were all smaller in injected vessels than in uninjected vessels. IELÎMA and the ratio of measured vs. idealized LA were both increased in injected lungs. We suspect that vascular injection selects for evaluation, a population of smaller, thin-walled vessels, which in the uninjected lungs are collapsed and hence excluded from analysis. Anat. Rec. 252:92–101, 1998. r 1998 Wiley-Liss, Inc. Key words: computer-assisted diagnosis; pulmonary artery; pulmonary circulation; pulmonary hypertension; vascular resistance; vascular smooth muscle Computerized interactive morphometric analysis entails measurement of the internal elastic lamina (IEL) circumference and the outer limit of the media, often demarcated in larger vessels, by the external elastic lamina (EEL). The computer then, by using integrals, calculates the actual area occupied by the media and lumen. Computer-assisted morphometric analysis is based on the premise that medial cross-sectional area and IEL circumference of the vessel wall are independent of intravascular distending pressure, thus raising the hope that the practice of uniform vascular distension would no longer be necessary (Cook and Yates, 1972). Traditional morphometric analysis requires preparation of the lungs r 1998 WILEY-LISS, INC. under uniform conditions to ensure accurate comparisons. Lungs for morphometric evaluation are traditionally fixed after intravascular injection of a barium sulfate/gelatin suspension at a standard supraphysiologic pressure (73 mmHg). This method has been used to eliminate indi- Grant sponsor: American Heart Association, Missouri affiliate. *Correspondence to: Jill A. Ohar, M.D., Division of Pulmonology, St. Louis University Health Sciences Center, 3635 Vista Avenue, 7FDT, St. Louis, MO 63110–0250. E-mail: email@example.com Received 14 November 1997; Accepted 10 April 1998 COMPUTERIZED PULMONARY MORPHOMETRY vidual variability and permit measurement under a uniform convention of true structural change caused by experimental conditions or disease. The viscosity of the barium–gelatin mixture permits filling of the arterial tree but not capillaries; thus, it aids in the histologic distinction of intra-acinar arteries from veins (Elliott and Reid, 1965). Furthermore, barium sulfate–gelatin solidifies rapidly and therefore maintains uniform distention of pulmonary arteries at the pressure with which the barium sulfate gel was injected. Barium sulfate angiograms can be used to evaluate the overall growth pattern of the vasculature and alterations in flow distribution, as in thromboembolic disease and Swyer-James syndrome. The growth pattern or flow distribution of the pulmonary vasculature may also be assessed in a section of lung by the number of bariumfilled arteries relative to the number of alveoli visible in a high power field. Traditionally, a ratio is calculated of the number of alveoli per the number of barium-filled arteries. This value increases when small arteries become sclerosed, as in pulmonary hypertension, or fail to develop normally, as in congenital heart disease (Rabinovitch et al., 1978; Rabinovitch and Reid, 1980). This ratio is also altered in experimental models of pulmonary hypertension induced by monocrotaline or hyperoxia (Jones et al., 1985). Theoretically, raising intravascular distending pressure increases pulmonary vascular recruitment and could influence the findings in this assessment. However, barium sulfate injection of lung tissue necessitates an intact pulmonary arterial tree to permit vascular injection and thus precludes the use of biopsy specimens, thereby limiting its usefulness in humans. Implicit in computer-assisted morphometric studies of the pulmonary vasculature is the assumption that the IEL and EEL do not change length as a consequence of pathologic remodeling or when fixed at supraphysiologic pressures. Although several studies using this technique have been published (Ono and Voelkel, 1991, 1992; Ono et al., 1992; Sheedy et al., 1996; Zhao et al., 1996), the assumption that the IEL and EEL do not change has not been proven by a controlled comparison of computerassisted vs. traditional methods. Cook and Yates (1972) referred to unpublished data to state that at physiological pressures no lengthening of the IEL occurs when an artery is distended, presumably when compared with undistended arteries; they did not comment on the effect of the supraphysiologic pressures characteristic of pulmonary hypertension. Further, they discussed systemic rather than pulmonary arteries and did not refer to disease states. In contrast, Fernie and Lamb (1985) concluded that, at supraphysiologic intravascular distending pressure, the IEL of fixed pulmonary arteries vs. undistended arteries does stretch by a factor of 1.5. They deduced this by comparing the slope of the linear regression of the IEL circumference related to the square root of the medial area (MA) in pulmonary arteries injected at a pressure of 75 mmHg; the arteries were grouped by size rather than by landmarking to the accompanying respiratory structure. The lungs used in these studies were obtained at autopsy from patients who died of obstructive lung disease, heart failure, and other unrelated conditions (Fernie and Lamb, 1985; Fernie et al., 1985). Published research has not supported the assumption that the IEL and EEL are unaffected by pathologic remod- 93 eling. In the normal rat, Davies et al. (1985) showed that the IEL is longer than EEL in pulmonary arteries more than 48 µm in external diameter, whereas the EEL is longer than the IEL in those less than 48 µm. In the hypoxic rat, this relationship is changed, thus demonstrating unreliability in comparing an elastic lamina in disease with normal lamina. Furthermore, in an animal model of pulmonary hypertension produced by platelet activating factor (PAF) (Ohar et al., 1990, 1991), we showed that at all levels the IEL and EEL circumferences are reduced under the experimental conditions. This finding and the reduction in external diameter (ED) of hilar pulmonary arteries that occurs in chronic hypoxia (Davies et al., 1985) because of contracture of the arterial wall points to a plasticity of the vascular wall that is quickly apparent in disease states. Also implicit in the applications to date of computerassisted morphometric analysis in the pulmonary vasculature is the assumption that landmarking is not necessary. However, by definition, a morphometric study is one that quantifies structural elements. In the pulmonary vasculature, structural characteristics of the arterial wall such as muscularization are dictated by the level of the artery within the branching system, and pathologic processes extend the level to which muscularization is seen. The level of branching of the pulmonary arteries is by convention ‘‘landmarked’’ by identifying the accompanying respiratory structure (i.e., bronchiole, terminal bronchiole, respiratory bronchiole, alveolar duct, and alveolar wall). Landmarking of pulmonary arteries by the accompanying airway and categorization of vessel walls as muscular, partly muscular, or nonmuscular is an accepted method for quantification of structural characteristics that permits comparison of normal with diseased vessels or with vessels of different species at the same level in the branching system. Therefore, studies that do not landmark risk comparison of structurally different populations of pulmonary arteries in diseased vs. normal specimens or among species from different species. To evaluate the validity of computerized morphometry in the analysis of pulmonary vascular structure, we studied the effect of a range of intravascular distending pressures on IEL circumference, medial cross-sectional area, and the ratio of the number of alveoli per number of barium-filled arteries per unit area. The two goals of the study were to determine (1) whether the presence of barium sulfate injected under pressure into the pulmonary arteries affected morphometric measurements at all and (2) whether incremental increases in injection pressure affected incremental alterations in morphometric variables. MATERIALS AND METHODS Animal Preparation Female New Zealand white rabbits weighing 2–3 kg were killed with a pentobarbital injection. The chest was immediately opened, the pulmonary artery cannulated, and the lungs flushed at an intravascular distending pressure of 20 mmHg with phosphate buffered saline at room temperature. The left atrium was opened widely to prevent back-pressure before flushing commenced, and flushing continued until the left atrial effluent was clear. The lungs were submerged in normal saline solution and stored frozen at 0°C until they were injected with barium 94 OHAR ET AL. Fig. 1. Barium sulfate angiogram. Barium was injected at a pressure of 40 mmHg for 15 min. The background blush in the periphery of the lung is consistent with adequate filling. sulfate and fixed with formalin. Freezing eliminated intravascular tone and also allowed injection of the lungs in batches to reduce variability of injection conditions. Frozen lungs were compared with lungs injected without the freezing step, and no difference in vascular filling and structural integrity was noted. Barium Sulfate Injection and Fixation Lungs were thawed and warmed to 37°C by submersion in warmed normal saline solution. The main pulmonary artery was cannulated (2-mm outer diameter tubing), and a solution consisting of barium sulfate (70% weight/ volume) suspended in gelatin (7% weight/volume solution in water warmed in a water bath to 60°C) was injected into the pulmonary artery at a constant pressure until vessels underlying the visceral pleura were visible and filled in a typical snowflake pattern (5–15 min). Distending pressures used were 20, 30, 40, 60, 70, and 80 mmHg. The snowflake pattern was visible in all lungs, but a longer injection time was required at lower pressures. For a seventh, ‘‘zero-pressure group,’’ lungs were not injected with barium. The trachea of each lung was then cannulated and filled with formalin by gravity. A head pressure of formalin (20 cm water) was maintained while the lungs were submerged in formalin for 3 days. The lungs were then radiographed to visualize filling of the intra-acinar Fig. 2. Fully and partly muscularized arteries are visible at all airway levels in the rabbit. The size range, quantified by external diameter, of muscularized arteries differed widely across airway levels. B, bronchiole; TB, terminal bronchiole; RB, respiratory bronchiole; AD, alveolar duct; AW, alveolar wall. vasculature by the presence of a uniform background ‘‘haze’’ and thus suitability for morphometric studies. In each group, lungs from four rabbits were selected based on the radiographs (Fig. 1). In the lower pressure group, although four angiograms could be included, many more COMPUTERIZED PULMONARY MORPHOMETRY 95 Fig. 3. The percentage of arteries filled with barium progressively increases with increasing intravascular distending pressure up to 60 mmHg for intra-acinar arteries and up to 40 mmHg for pre-acinar arteries. No further recruiting occurs at higher pressures. injections were rejected as unsatisfactory vs. those in the higher pressure groups. Percentage of Arteries Filled by Barium and the Arterial:Alveolar Ratio From the angiograms it was apparent that the different distending pressures did not alter filling of the larger pre-acinar arteries. To assess recruitment, three features were analyzed in each high power field (HPF: (1) alveolar density as a measure of lung distension, (2) the number of barium-filled vessels per unit area, and (3) the total number of vessels less than 200 µm and greater than 10 µm in diameter per unit area. The calculated ratio of feature 2 to feature 3 is a measure of arterial recruitment. Twenty HPFs from the left lower lobe were analyzed for each of four animals in each intravascular pressure group. Using the number of alveoli within an HPF as an index of lung distension (presuming normal alveolar development in these adult rabbits), the number of veins included in the total vessel count should remain constant. Sectioning and Staining Five blocks of lung tissue approximately 1.5 cm 3 1.5 cm were obtained from each animal by slicing along the lingula and upper and lower lobes parallel to the hilum: 1.0 cm from the hilum for the right upper lobe, 1.4 cm for the right middle lobe, 1.8 cm for the right lower lobe, 1.4 cm for the left upper lobe, and 1.8 cm for the left lower lobe. Lung blocks were embedded in paraffin, sectioned (5 µm thick), and stained by the Verhoeff–van Gieson method. One stained slide was examined from each block, yielding a total of five stained slides that were studied from each animal. In the rabbit, fully muscularized arteries are present down to the respiratory bronchiolar level, with a range of 1,000–94 µm, and include a complete IEL and EEL. Distal to this, partly muscular arteries (150–31 µm) are seen as far as the alveolar wall and only an IEL is present. Nonmuscularized arteries can be as large as 150 µm in diameter (Fig. 2). Morphometric Analysis A computer-assisted image analysis system (American Innovision, San Diego, CA) was used to measure several variables in cross-sectioned arteries accompanying bronchioli, terminal bronchioli, respiratory bronchioli, and alveolar ducts and within alveolar walls. Twenty arteries per airway level per animal were evaluated, resulting in evaluation of 100 arteries per animal. Morphometric measurements were made directly from the glass slide. Magni- 96 OHAR ET AL. Fig. 4. Internal elastic lamina (IEL) circumference measurements are unaffected by intravascular distending pressures. B, bronchiole; TB, terminal bronchiole; RB, respiratory bronchiole; AD, alveolar duct; AW, alveolar wall. IEL circumference (inset) is significantly less in injected intra-acinar (AW and AD) arteries (striped bars) than in uninjected arteries (solid bars). Barium sulfate injection likely facilitates selection of smaller arteries for evaluation by the morphometrist. These smaller arteries are often collapsed in uninjected specimens and thus excluded from the evaluation. fied images were relayed from a Nikon Labophot (Model 197873, Japan) light microscope through a Javelin Chromochip II (Model JE3462HR, Javelin Electronics, Japan) video camera to a standard videoscreen. For vessel morphometry, only barium-filled arteries were measured. Only arteries that were cut in good cross section (defined as the lengths of perpendicular diameters differing by ,10%) and measuring 50–1,000 µm in diameter were analyzed. The measured parameters included the circumference of the IEL and EEL when present. The IEL and EEL of each arterial profile were traced with the computer cursor (mouse) on the screen to record the circumference. From these tracings, a computer program using integrals calculated the two areas, the area between the EEL and IEL (medial area), and the area enclosed by the IEL (luminal area). The intima was so thinned by the intravascular distension that the endothelium sat directly on the IEL, and intimal thickness (area between IEL and endothelium) could not be measured. Calculated Morphometric Variables Calculated variables included percentage of medial thickness, IELÎMA, and idealized lumen area (LAideal). Percent of medial thickness was calculated according to the following formula: medial thickness 5 [(M1 1 M2)/2] diameter 3 100%. The thickness (M1 and M2) was taken as it was cut at two positions along the shortest ED. The IEL divided by the square root of the medial area was taken to indicate the relation of medial thickness to arterial size (Fernie et al., 1985; Fernie and Lamb, 1985). LAideal is an idealized circle that the IEL would circumscribe if it were ‘‘smoothed out.’’ LAideal was calculated from the IEL circumference according to the following formula: LAideal 5 p 3 2p 4 . IEL 2 COMPUTERIZED PULMONARY MORPHOMETRY 97 Fig. 5. External diameter is unaffected by intravascular distending pressure. B, bronchiole; TB, terminal bronchiole; RB, respiratory bronchiole; AD, alveolar duct; AW, alveolar wall. At the alveolar wall level (AW), external diameter is significantly smaller in barium-injected arteries (striped bars) than in uninjected arteries (solid bars in inset). In idealized, fully distended arteries, LAmeasured/LAideal is equal to 1 (Davies et al., 1985). In constricted arteries, the ratio is ,1. The more constricted an artery, the lower the ratio value. 0.01 was considered to be significant. Data are expressed as the mean 6 the standard deviation. Statistical Analysis RESULTS Percentage of Arteries Filled by Barium and the Arterial:Alveolar Ratio Two separate statistical analyses were conducted. First, to determine whether morphometric variables measured in uninjected lungs were comparable to morphometric variables measured in injected lungs, the means of variables from all injected groups (20–80 mmHg) were compared with the means of variables measured in uninjected specimens (0-mmHg pressure) by Student’s t-test. Second, to determine the effect of different distending pressures on the variables measured from the arteries in each injection pressure group, the means of the variables measured in each pressure group (20–80 mmHg) were compared by analysis of variance followed by a test for linear trend. P , At the pre-acinar level, the percentage of barium-filled arteries, a measure of vascular recruitment, tended to increase with increasing intravascular distending pressure in a linear fashion. The trend apparent at the pre-acinar level was not statistically significant. At the intra-acinar level however, the percentage of arteries filled with barium increased significantly with increasing intravascular distending pressure (Fig. 3). Full recruitment or complete filling of pre-acinar arteries occurred at an intravascular distending pressure of 40 mmHg, whereas 60 mmHg was required for full recruitment of intra-acinar arteries. As would be expected from these data, the alveolar: 98 OHAR ET AL. Fig. 6. Medial area is unaffected by increasing intravascular distending pressure. B, bronchiole; TB, terminal bronchiole; RB, respiratory bronchiole; AD, alveolar duct; AW, alveolar wall. In barium-injected arteries (striped bars), the medial area is significantly smaller than in uninjected arteries (solid bars) at all airway levels (inset). arterial ratio tended to be greater in lung tissues injected at 20–40 mmHg than in lung tissues injected at 60–80 mmHg (data not shown). Morphometric Analysis The IEL circumference tended to be smaller in injected arteries than in uninjected arteries. This difference in IEL circumference was significant at the alveolar wall and alveolar duct levels (Fig. 4). When injected arteries were evaluated for effect of intravascular distending pressure on IEL circumference, no linear relationship could be identified (Fig. 4). The ED was significantly smaller in the injected than in the uninjected arteries only at the alveolar wall level (Fig. 5). There was no significant difference in ED between injected and uninjected arteries at all other airway levels, and ED was unaffected by distending pressure at all airway levels (Fig. 5). The MA was significantly smaller at every airway level in injected arteries than in uninjected arteries (Fig. 6). A linear trend between MA and distending pressure was not identified. Fig. 7. Luminal area measurements were unaffected by increasing intravascular distending pressure or by barium injection (inset). There was no significant difference in lumen area between injected and uninjected arteries at any airway level (Fig. 7). Further, among injected arteries, no linear trend was found between distending pressure and measured lumen area at any airway level (Fig. 7). Calculated Morphometric Variables At every airway level the medial thickness was significantly smaller in the injected than in the uninjected arteries (Fig. 8). When injected arteries were evaluated for effect of intravascular distending pressure on medial thickness, no linear relationship could be identified (Fig. 8). The IELÎMA was significantly smaller in uninjected than in injected arteries at every airway level (Fig. 9). Among injected arteries, there was no significant linear trend between IELÎMA and intravascular distending pressure identified at any airway level (Fig. 9). The ratio of measured vs. idealized lumen area was significantly smaller in uninjected than in injected arteries at all airway levels (Fig. 10). However, no significant linear trend at any airway level could be established between intravascular distending pressure and lumen area ratio (Fig. 10). COMPUTERIZED PULMONARY MORPHOMETRY 99 Fig. 8. Medial thickness is unaffected by increasing intravascular distending pressure. B, bronchiole; TB, terminal bronchiole; RB, respiratory bronchiole; AD, alveolar duct; AW, alveolar wall. Medial thickness (inset) is significantly smaller in barium-injected arteries (striped bars) than in uninjected arteries (solid bars) at all airway levels. DISCUSSION One goal of the present study was to determine whether the presence of barium injected into arteries affected vascular morphometric measurements. We reasoned that if barium injection did affect measurement of morphometric variables, it would do so in a linear fashion correlating with incremental increases in injection pressure. Therefore, the second goal of the present study was to determine whether incremental increases in barium injection pressure correlated linearly with alterations in vascular morphometric measurements, indicating a cause-and-effect relationship. We found that comparison of injected (injecting pressure 20–80 mmHg) with uninjected arteries showed significant differences: in injected vessels, IEL circumference, medial thickness, and medial area were all smaller than those in uninjected vessels. IELÎMA and the ratio of measured vs. idealized lumen area were both increased in injected lungs. This result is contrary to our expectation that increasing intravascular distending pressure would progressively stretch the IEL and thin the media, thus producing an incremental increase in IEL circumference and a decrease in medial thickness and cross-sectional medial area. Whereas barium-injected arteries did have a reduced medial thickness and medial area as compared with uninjected arteries, the IEL circumference was also decreased. One explanation for this phenomenon is that vascular injection selects for evaluation, a population of smaller, thin-walled vessels, which in the uninjected lungs are collapsed and hence excluded from analysis. The only measurement that was linearly affected by increasing intravascular distending pressure was the increased percentage of arteries that filled with barium. For pre-acinar arteries, maximum recruitment occurred at 40-mmHg pressure; for intra-acinar arteries, this was achieved at 60-mmHg distending pressure. In injected vessels, incremental increases in the barium injection pressure did not affect IEL circumference, external diameter, medial thickness, medial area, or lumenal area. Today there is widespread acceptance of computerassisted image analysis techniques for morphometric evalu- 100 OHAR ET AL. Fig. 9. Top: Comparison of the slope of the linear regression of the internal elastic lamina circumference related to the square root of the medial area (MA) (IELÎMA) is unaffected by increasing intravascular distending pressure. B, bronchiole; TB, terminal bronchiole; RB, respiratory bronchiole; AD, alveolar duct; AW, alveolar wall. Bottom: IELÎMA is significantly smaller in uninjected arteries (solid bars) than in injected arteries (striped bars) at all airway levels. ation of pulmonary arteries. The evolution of these techniques was predicated on inflexibility of the circumference of the IEL and the medial area despite increasing intravascular distending pressures. Our study validates this concept as did the work of Cook and Yates (1972). However, Fernie and Lamb (1985) claimed that the IEL does stretch based on the different slopes of the regression lines derived from IELÎMA in a lung injected with barium sulfate at a pressure of 73 mmHg compared with the uninjected contralateral lung. However, their study included patients with known pulmonary pathology, and the vessels analyzed were not landmarked according to accompanying respiratory structure. In many pathologic conditions, vascular remodeling occurs, resulting in extension of the muscular coat to a more distal level in the vascular tree, medial hypertrophy, and lengthening or shortening of the elastic laminae (Jones et al., 1985; Ohar et al., 1991). Therefore, in the study by Fernie and Lamb (1985), it is not clear that vessels at the same level in the vascular tree were compared. Furthermore, in their study, the medial areas of the measured vessels were not reported. Fig. 10. Top: The ratio of measured to idealized lumen area is unaffected by increasing intravascular distending pressure. B, bronchiole; TB, terminal bronchiole; RB, respiratory bronchiole; AD, alveolar duct; AW, alveolar wall. Bottom: The ratio of measured to idealized lumen area is unaffected by barium injection (striped bars, injected arteries; solid bars, uninjected arteries). Computer-assisted image analysis for evaluation of pulmonary arteries represents a technologic advance that allows rapid measurement of large numbers of morphometric variables that can be easily downloaded for statistical analysis. It does not circumvent the need for landmarking of pulmonary arteries to accompanying respiratory structure, and it does not facilitate direct comparison of bariuminjected vessels to uninjected vessels because barium injection facilitates evaluation of a population of thinwalled vessels that is generally eliminated from analysis in uninjected specimens because of vessel collapse. ACKNOWLEDGMENT We are indebted to Lynne M. Reid, M.D., for her patience, assistance, and critical review of the manuscript. LITERATURE CITED Cook TA, Yates PO. A critical survey and techniques for arterial mensuration. J. Pathol. 1972;108:119–127. COMPUTERIZED PULMONARY MORPHOMETRY Davies P, Maddalo F, Reid L. Effects of chronic hypoxia on structure and reactivity of rat lung microvessels. J. Appl. Physiol. 1985;58:795– 801. Elliott FM, Reid L. Some new facts about the pulmonary artery and its branching pattern. Clin. Radiol. 1965;16:193–198. Fernie JM, Lamb D. A new method for quantitating the medial component of pulmonary arteries: The measurements. Arch. Pathol. Lab. Med. 1985;109:156–162. Fernie JM, McLean A, Lamb D. New method for quantitating the medial component of pulmonary arteries: Factors affecting the measurements. Arch. Pathol. Lab. Med. 1985;109:843–848. Jones R, Langleben D, Reid LM. Patterns of remodeling of the pulmonary circulation in acute and subacute lung injury. In: FI Said, ed., The Pulmonary Circulation and Acute Lung Injury. Mount Kisco, NY: Futura Publishing Co., New York, 1985. Ohar JA, Pylle JA, Waller KS, Hyers TM, Webster RO, Lagunoff D. A rabbit model of pulmonary hypertension induced by the synthetic platelet-activating factor acetylglyceryl ether phosphorylcholine (AGEPC). Am. Rev. Resp. Dis. 1990;141:104–110. Ohar JA, Waller KS, deMello D, Lagunoff D. Administration of chronic intravenous platelet activating factor induces pulmonary arterial 101 contracture and hypertension in rabbits. Lab. Invest. 1991;65:451– 458. Ono S, Voelkel NF. PAF antagonists inhibit monocrotaline-induced lung injury and pulmonary hypertension. J. Appl. Physiol. 1991;71: 2483–2492. Ono S, Voelkel NF. PAF receptor blockade inhibits lung vascular changes in the rat monocrotaline model. Lung 1992;170:31–40. Ono S, Westcott JY, Voelkel NF. PAF antagonists inhibit pulmonary vascular remodeling induced by hypobaric hypoxia. J. Appl. Physiol. 1992;73:1084–1092. Rabinovitch M, Reid LM. Quantitative structural analysis of the pulmonary vascular bed in congenital heart defects. Pediatr. Cardiovasc. Dis. 1980;11:149–169. Rabinovitch M, Haworth SG, Cataneda AR, Nadas AS, Reid LM. Lung biopsy in congenital heart disease: A morphometric approach to pulmonary vascular diseases. Circulation 1978;58:1107–1122. Sheedy W, Thompson JS, Morice AH. A comparison of pathophysiological changes during hypobaric and normobaric hypoxia in rats. Respiration 1996;63:217–222. Zhao L, al-Tubuly R, Sebkhi A, Owji AA, Nunez DJ, Wilkins MR. Angiotensin II receptor expression and inhibition in the chronically hypoxic rat lung. Br. J. Pharmacol. 1996;119:1217–1222.