Abnormal pulmonary perfusion heterogeneity in patients with Fontan circulation and pulmonary arterial hypertension

Abstract

Many factors affect the distribution of pulmonary perfusion, which may be disrupted by cardiopulmonary disease, but this is not well studied, particularly in rare conditions. An example is following the Fontan procedure, where pulmonary perfusion is passive, and heterogeneity may be increased because of the underlying pathophysiology leading to Fontan palliation, remodelling, or increased gravitational gradients from low flow. Another is pulmonary arterial hypertension (PAH), where gravitational gradients may be reduced secondary to high pressures, but remodelling may increase perfusion heterogeneity. We evaluated regional pulmonary perfusion in Fontan patients (n = 5), healthy young controls (Fontan control, n = 5), patients with PAH (n = 6) and healthy older controls (PAH control) using proton magnetic resonance imaging. Regional perfusion was measured using arterial spin labelling. Heterogeneity was assessed by the relative dispersion (SD/mean) and gravitational gradients. Mean perfusion was similar (Fontan = 2.50 ± 1.02 ml min⁻¹ ml⁻¹; Fontan control = 3.09 ± 0.58, PAH = 3.63 ± 1.95; PAH control = 3.98 ± 0.91, P = 0.26), and the slopes of gravitational gradients were not different (Fontan = −0.23 ± 0.09 ml min⁻¹ ml⁻¹ cm⁻¹; Fontan control = −0.29 ± 0.23, PAH = −0.27 ± 0.09, PAH control = −0.25 ± 0.18, P = 0.91) between groups. Perfusion relative dispersion was greater in both Fontan and PAH than controls (Fontan = 1.46 ± 0.18; Fontan control = 0.99 ± 0.21, P = 0.005; PAH = 1.22 ± 0.27, PAH control = 0.91 ± 0.12, P = 0.02) but similar between patient groups (P = 0.13). These findings persisted after removing contributions from large blood vessels and gravitational gradients (all P < 0.05). We conclude that patients with Fontan physiology and PAH have increased pulmonary perfusion heterogeneity that is not explained by differences in mean perfusion, gravitational gradients, or large vessel anatomy. This probably reflects the effects of remodelling in PAH and possibly in Fontan physiology.

Introduction

Many factors affect the heterogeneity of pulmonary perfusion, including hydrostatic effects due to gravity, vascular branching structure (Hlastala & Glenny, 1999; Glenny & Robertson, 2010) and active regulatory mechanisms (see Glenny & Robertson, 2011 for review). Perfusion heterogeneity may also be disrupted by cardiopulmonary disease, but this is not well studied, particularly in rare conditions. Such is the case with the Fontan procedure (Akintoye et al. 2019), a palliative cardiac surgery to treat patients with only a single usable ventricle (Fontan & Baudet, 1971). Following Fontan palliation, the vena cavae are directly anastomosed to the pulmonary arteries and there is no subpulmonary ventricle; pulmonary perfusion is non-pulsatile, passive and dependent on the pressure gradient between the systemic venous and pulmonary postcapillary vessels. The remaining single ventricle, which may be either the right or left ventricle, pumps to the systemic circulation. After Fontan palliation, many children survive into adulthood, although ultimately ventricular failure or other complications may lead to the need for cardiac transplant (Akintoye et al. 2019).

In patients with Fontan physiology, hypoplastic pulmonary vessels may result from the underlying pathophysiology that comprised the primary cardiac abnormality (Gewillig & Goldberg, 2014) leading to Fontan palliation, or may arise as a result of decrease in the overall lung perfusion and failure of normal vascular development. Overall low pulmonary vascular pressures may have Zone model effects (West et al. 1964a) on the gravitational distribution of perfusion. Any of these factors may increase pulmonary perfusion heterogeneity in these individuals, but this is unknown.

Another rare condition affecting the pulmonary circulation is pulmonary arterial hypertension (PAH). PAH is a vasculopathy characterized by progressive remodelling of the pulmonary circulation, increased precapillary resistance, elevated pulmonary arterial pressure, and ultimately, right heart failure and death (Thenappan et al. 2018). In PAH, conflicting factors may affect heterogeneity: increased pressure may reduce gravitational gradients and heterogeneity due to Zone model effects, whereas remodelling of the vascular structure may increase heterogeneity, but, again, this is not well described.

ASL-FAIRER (ASL, arterial spin labelling; FAIRER, flow-sensitive alternating inversion recovery with an extra radio frequency pulse) (Mai & Berr, 1999; Mai et al. 2001) is a magnetic resonance imaging technique that uses magnetic field gradients and radiofrequency pulses to label endogenous protons allowing non-invasive high-resolution perfusion measurements without exposure to ionizing radiation. Using ASL-FAIRER we measured the distribution of pulmonary perfusion in two groups of patients with rare cardiopulmonary disease: those who had undergone Fontan procedure and those with a known vasculopathy, PAH, and in age- and sex-matched controls, one control group for each patient population. We hypothesized that both PAH patients and those with Fontan circulation would have an increase in perfusion heterogeneity consistent with underlying remodelling of the pulmonary circulation. We also hypothesized, consistent with Zone model predictions (West et al. 1964a), that PAH patients would have a decreased gravitational gradient in pulmonary perfusion, whereas those with Fontan circulation would have an increased gradient compared to controls. Some data have been previously reported in abstract form (Hopkins et al. 2016).

Methods

This study conformed to the Declaration of Helsinki standards and was approved by the University of California, San Diego’s Human Subjects Research Protection Program (no. 121259) and Rady Children’s Hospital (no. 140960). This study was not registered in a database. A total of 22 subjects were recruited: five patients with Fontan circulation (three male, two female), five age- and sex-matched normal controls for Fontan patients (three male, two female), six patients with PAH (all female) and six age- and sex-matched controls for PAH patients. Subjects were recruited from patient registries and by advertisement and gave written, informed consent. Subjects underwent screening for exclusion criteria including an inability to lie flat in the scanner, need for supplemental oxygen or contraindications to magnetic resonance imaging. Normal control subjects had no history of significant illness, cardiopulmonary disease, smoking or other inhalant use, or use of any regular medication. All patients were maintained on their regularly prescribed medications which included PAH targeted therapies in the case of PAH patients. Spirometry was performed (Easy One spirometry, Zurich, Switzerland) using ATS standards and the Third National Health and Nutrition Examination Survey references values (Hankinson et al. 1999).

Protocol overview

Subjects were placed supine in a 1.5T Signa HDx TwinSpeed MRI system (General Electric Medical Systems, Milwaukee, WI, USA). An eight-channel torso coil was placed around the subject’s chest to allow for the acquisition of pulmonary blood flow images. MRI-compatible ECG electrodes (InVivo ECG Quadtrodes) were placed on the left chest and a reference silicone phantom was placed on the right chest prior to the placement of the top elements of the torso coil. The reference phantom allows for absolute quantification of perfusion and proton density (see below).

All imaging results are for a single 15 mm sagittal slice in the right lung acquired within an 8–10 s breath-hold. The slice location was positioned in the mid-clavicular line to capture the maximum anterior-posterior diameter of the lung and to avoid motion artefacts from the aorta and heart in the left hemithorax. Subjects were trained to breath-hold at the end of a normal expiration (at functional residual capacity, FRC) beginning just before the images were acquired and ending when the imaging sequence finished.

Proton density imaging

Regional proton density (e.g. lung water content) was measured using a fast gradient echo sequence described in detail elsewhere (Theilmann et al. 2009). The multi echo gradient echo sequence acquires multiple single echo acquisition alternating between two echo times in a single brief breath-hold. Six images (even images: 2, 4, 6, 8, 10, 12) were acquired with an echo time of 1.1 ms and six images (odd images: 1, 3, 5, 7, 9, 11) were acquired at an echo time of 1.8 ms. The even images were averaged to improve signal-to-noise ratio and generate a short echo time image. The process was repeated for the odd images to generate the long echo time image. Proton density was determined by fitting multi-echo data to a single exponential and back extrapolating the signal to an echo time of zero (e.g. TE = 0). Since the phantom does not exhibit the same image decay constants to that in the lung, a correction factor was empirically determined for the static sequence parameters permitting the mean phantom signal to be used as a reference for absolute calibration of water content (e.g. proton density) (Theilmann et al. 2009; Holverda et al. 2011). Proton density imaging sequence parameters were TR = 10 ms, flip angle = 10 deg, slice thickness = 15 mm, field of view = 40 cm, receiver bandwidth = 125 kHz, and a full acquisition matrix of 64 × 64 (reconstructed by scanner to 256 × 256). Proton density images were collected with both the torso and with the body coil built into the scanner to allow for subject specific correction of perfusion images for heterogeneity in signal intensity based on proximity to the coil elements (described below in Image Processing).

Measuring pulmonary blood flow using ASL

Arterial spin labelling (ASL) MRI is a non-invasive technique where arterial blood water is magnetically labelled (‘tag’ image) with a combination of radiofrequency pulses and spatial magnetic field gradients. After a period of time, the inflowing magnetically labelled arterial blood water alters the total tissue magnetization, and consequently the image intensity after acquisition. The acquisition is repeated without labelling the arterial blood (‘control’ image). Regional perfusion is proportional to the difference in magnetization between the ‘control’ and the ‘tag’ image.

Regional pulmonary perfusion was assessed using a flow-sensitive alternating inversion recovery with an extra radiofrequency pulse (FAIRER) (Bolar et al. 2006; Hopkins et al. 2007), an ASL technique, followed by a Half-Fourier Acquisition Single-shot Turbo spin-Echo (HASTE) acquisition (Mai et al. 1999; Bolar et al. 2006) using the torso coil. With ASL FAIRER, the magnetization of the arterial blood protons is prepared with the same timing but is spatially different for the ‘tag’ and ‘control’ images. The ‘tag’ image is acquired after a global inversion pulse that inverts arterial blood both inside and outside the lung slice being imaged whereas the ‘control’ image is acquired after a slice-selective inversion pulse which inverts the arterial blood inside the lung slice only. The inversion preparation (global or slice-selective) is applied during diastole with the acquisition occurring after a delay chosen to be approximately 80% of one R-R interval with blood flowing into the slice during this delay. The difference between ‘control’ and ‘tag’ signals within each voxel is proportional to the amount of blood delivered during the delay, or inversion time (TI) interval, weighted by a decay factor due to the relaxation of the blood magnetization during that interval (Henderson et al. 2009). FAIRER-ASL imaging sequence parameters were as follows: TI = 600–800 ms (based on subject’s heart rate), TE = 21.3 ms, field of view = 40 cm, slice thickness = 15 mm. The collected image matrix size was 256 × 128 (reconstructed by scanner to 256 × 256).

Image processing

Quantification of regional lung density (in g ml⁻¹).

Proton density data were acquired using the body coil to avoid signal intensity inhomogeneity often present with multi-coil acquisitions. After back-extrapolating the signal to an echo time of zero on a voxel-by-voxel basis, regional lung proton (water) density was obtained by referencing the signal in the phantom which had been previously characterized. This proton density, which reflects protons in both tissue and blood, is subsequently referred to in this article as density (in units of g H₂O per ml lung) (Theilmann et al. 2009). The details of this technique for quantifying regional lung density has been published and validated, showing excellent reliability (R² = 0.95, P < 0.0001) in repeated studies and excellent validity < 5% error (Holverda et al. 2011).

Quantification of regional blood delivered (in ml min⁻¹ ml⁻¹).

Subtraction of the ‘tag’ from the ‘control’ image produces an image in which the signal from a voxel is proportional to the amount of pulmonary blood delivered averaged over the previous heart cycle. Quantification of regional blood flow (in units of ml min⁻¹ (ml lung)⁻¹) is obtained by referencing the pulmonary blood flow signal intensity to the mean signal in the silicone phantom after correcting for magnetization relaxation effects in the phantom (T₁ = 620 ms, T₂ = 160 ms) and human pulmonary arterial blood (Spees et al. 2001) that occurs during the inversion time (TI) (Henderson et al. 2009). This technique has been validated in a lung phantom model (Hopkins & Prisk, 2010) with excellent results.

A torso coil with eight channels with a higher sensitivity profile than the body coil was used for the ASL acquisition to maximize signal-to-noise ratio. Due to the eight channel multi-coil acquisition, the intensity inhomogeneity in data acquired with the torso coil was corrected as follows: A coil sensitivity map was calculated from the density data obtained with torso and the body coil. Data were averaged from the short echo time images from both coils. These data were spatially smoothed in frequency space, equivalent to convolving with a 2D Gaussian in image space with standard deviation of 2.1 cm and full width at half-maximum of 5 cm. The resultant smoothed images were divided (torso/body) to define each subject’s coil sensitivity map. A quantitative map of perfusion, independent of the spatial variations in coil sensitivity, was calculated after dividing the calibrated ASL image by the coil sensitivity map on a voxel-by-voxel basis.

Quantification of regional perfusion (in ml min⁻¹ ml⁻¹) by removal of large vessels.

The arterial spin labelling technique measures the amount of blood delivered into the imaging slice from outside the imaging slice, comprising both that in large conduit vessels and also that representing perfusion in the capillary vessels. To distinguish these two components, we applied a threshold of 35% of maximum blood delivered (in ml min⁻¹ ml⁻¹) calculated from the mean value of the top 1% of voxels in the image. This resulted in two categories of signal in the images: (1) signal equal to or exceeding this intensity was designated as conduit vessels, and (2) signal below this thresh-hold was considered to represent capillary perfusion. The 35% blood flow thresh-hold was chosen based on previously published modelling studies of our technique (Burrowes et al. 2012). For this study, regional perfusion data were analysed with and without conduit vessels by applying a binary mask. Voxels containing larger conduit blood vessels were not excluded from density images since these voxels contribute to the regional lung density.

Data analysis

The relative dispersion (standard deviation/mean) was calculated and used as an index of perfusion heterogeneity (Glenny, 1998). To evaluate the effects of gravity on the perfusion distribution, data were binned in 1 cm sheets starting from the most gravitationally dependent lung, and mean values at each location compared between groups. The slope and strength of the association (R²) was obtained. This approach explicitly excludes variability within an isogravitational plane but includes the effect of gravitationally induced tissue deformation (Hopkins et al. 2007). To assess heterogeneity independently of gravitational effects, relative dispersion was also calculated from the summed weighted average of each 1 cm isogravitational sheet (Hall et al. 2014). Since flow in both large, conduit, vessels and perfusion in the capillaries contribute signal to the ASL-FAIRER images that may increase heterogeneity in the image, the relative dispersion was calculated with and without conduit vessels.

ANOVA was used to compare differences between subject groups; in the case of a significant omnibus F, post hoc testing used Students t test (Statview 4.1, SAS Institute Inc., Cary, NC, USA) to identify where these differences occurred. Significance was accepted at P < 0.05, two-tailed.

Results

Subject descriptive data and spirometry

All of the control subjects fulfilled the entry criteria and were on no medications, and without a history of inhalant use, smoking or cardiopulmonary disease. The Fontan cohort consisted of one subject with right ventricular morphology and four with left ventricular morphology and the average time from Fontan completion to imaging was 17 ± 6 years. Medications in this cohort included acetylsalicylic acid in all subjects, angiotensin converting enzyme (ACE) inhibitors in two subjects, and ACE inhibitors plus a pulmonary vasodilator in one subject. The PAH cohort included two patients with idiopathic PAH (World Symposium on Pulmonary Hypertension,WSPH, group 1.1), one patient with PAH associated with drug use (WSPH group 1.3) and three patients with PAH associated with connective tissue disease (WSPH group 1.4). The average time from PAH diagnosis to imaging was 6.1 ± 4.0 years. All of the PAH patients were maintained on oral pulmonary vasodilator therapy.

Spirometry was normal for all controls (Table 1). There was no significant difference between the patient groups and their respective controls for the demographic variables. However, the Fontan patients had a significantly lower forced vital capacity (FVC) percentage of predicted (P = 0.02), and forced expiratory volume in one second (FEV₁) percentage of predicted (P = 0.03) than their controls. Similarly, the PAH patients had a lower FVC percentage predicted (P = 0.03), and FEV₁ percentage predicted (P = 0.04). Heart rate and oxygen saturation data are given in Table 1. Both the Fontan and PAH patient groups had a significantly lower oxygen saturation than their control subjects (P = 0.003 and P = 0.03, respectively) consistent with the clinical presentation of their disease.

Table 1.

Subject demographic and pulmonary function data (means ± SD)

	Fontan	Fontan control	PAH	PAH control	P Fontan vs. Fontan control	P PAH vs. PAH control
N	5	5	6	6	—	—
Sex (M/F)	3/2	3/2	0/6	0/6	—	—
Age (years)	20 ± 5	21 ± 2	52 ± 10	48 ± 13	0.78	0.57
Height (cm)	159 ± 15	164 ± 7	161 ± 10	163 ± 10	0.56	0.78
Weight (kg)	54.1 ± 15.2	65.5 ± 8.7	62.9 ± 13.9	74.1 ± 7.5	0.18	0.11
FVC %	80 ± 18	108 ± 8	84 ± 24	106 ± 6	0.02	0.03
FEV1 %	81 ± 15	100 ± 4	74 ± 19	98 ± 8	0.02	0.04
FEV1/FVC %	105 ± 5	100 ± 12	89 ± 9	93 ± 6	0.45	0.45
PAP (mmHg)#	11 ± 2	—	51 ± 4	—	—	—
PAW (mmHg)#	7 ± 1	—	7 ± 3	—	—	—
PVR (Wood Units)#	0.8 ± 0.2	—	12.8	—	—	—
		—	5.3	—	—	—
Heartrate	58 ± 10	59 ± 8	72 ± 11	67 ± 9	0.79	0.42
SpO2	94 ± 1	98 ± 0	95 ± 4	98 ± 1	0.003	0.03

Data are means ± standard deviations. Abbreviations: FEV₁, forced expiratory volume in 1 s; FVC, forced vital capacity; HR, heart rate; PAP, pulmonary arterial pressure, PAW, pulmonary arterial wedge pressure; PVR, pulmonary vascular resistance; S_po2, oxygen saturation measured by pulse oximetry.

^#Data from clinical record.

Mean image density and perfusion data

Representative perfusion images are given in Fig. 1. There were no significant differences between groups for mean density (P = 0.66), mean perfusion (P = 0.26) or mean perfusion exclusive of conduit vessel signal (P = 0.19) in the images. Perfusion varied with height from dependent lung (P < 0.0001), consistent with Zone model effects (West et al. 1964a; Hopkins et al. 2007) but this was not different between groups (Fig. 2, P = 0.15). There was also no significant difference between groups for slope or degree of association of the regression of perfusion vs. height from dependent lung (Table 2).

Figure 1. Representative sagittal ASL perfusion images from one subject from each of the subject groups.

Note the overall small lung size in the Fontan patient compared to control, and the increased size of the large vessels seen as the red tubular structures in the central image in the patient with PAH. Colour scale is perfusion in ml min⁻¹ ml⁻¹.

Figure 2. Perfusion vs. height from the most gravitationally dependent lung (mean and standard deviation).

For visualization purposes, standard deviation is shown for Fontan and PAH patients only. Closed circles, Fontan patients (n = 5); open circles, Fontan controls (n = 5); closed squares, PAH (n = 6); open squares, PAH controls (n = 6).

Table 2.

Mean imaging data (means ± SD)

	Fontan (n = 5)	Fontan control (n = 5)	PAH (n = 6)	PAH control (n = 6)	Omnibus P
Number of voxels SD	5178 ± 1051	5554 ± 1000	5316 ± 1193	4849 ± 1302	0.47
Proton density (g ml−1) SD	0.25 ± 0.04	0.24 ± 0.02	0.27 ± 0.05	0.27 ± 0.03	0.66
Perfusion (ml min−1 ml−1) SD	2.50 ± 1.02	3.09 ± 0.58	3.63 ± 1.95	3.98 ± 0.91	0.26
Perfusion (ml min−1 ml−1)* SD	1.60 ± 0.63	2.50 ± 0.77	2.61 ± 1.50	2.92 ± 0.63	0.19
Slope (ml min−1 ml−1 cm−1)	−0.23 ± 0.09	−0.29 ± 0.23	−0.27 ± 0.09	−0.25 ± 0.18	0.91
R2	0.65 ± 0.28	0.66 ± 0.22	0.64 ± 0.17	0.63 ± 0.27	0.31

^*Data exclusive of signal in large, conduit vessels. Slope is the rate of decrease of perfusion with increasing height from the most dependent lung as determined by linear regression and R² is the strength of the association. There were no significant differences between subject groups for mean data or the distribution of perfusion as a function of gravity and therefore post hoc testing was not done.

Indices of heterogeneity

There were highly significant differences between subject groups for measures of perfusion heterogeneity (Table 3): Overall, relative dispersion was different between subject groups (P = 0.002). Both populations of patients were different from their respective controls (Fontan vs. Fontan control, P = 0.005; PAH vs. PAH control, P = 0.02), but Fontan patients did not differ significantly from PAH (P = 0.13). The group differences persisted even when signal from large conduit vessels were excluded (P = 0.0008, Table 3) and when only heterogeneity in the isogravitational plane was considered (P = 0.004), but again, Fontan patients did not differ significantly from PAH patients (Table 3, Fig. 3).

Table 3.

Measures of heterogeneity (means ± SD)

	Fontan (n = 5)	Fontan control (n = 5)	PAH (n = 6)	PAH control (n = 6)	P Omnibus	P Fontan vs. Fontan control	P PAH vs. PAH control	P Fontan vs. PAH
Relative dispersion	1.46 ± 0.18	0.99 ± 0.21	1.22 ± 0.27	0.91 ± 0.12	0.002	0.005	0.02	0.13
Relative dispersion Isogravitational	1.28 ± 0.12	0.95 ± 0.12	1.16 ± 0.26	0.88 ± 0.13	0.004	0.007	0.01	0.36
Relative dispersion*	1.30 ± 0.34	0.73 ± 0.13	0.91 ± 0.24	0.66 ± 0.08	0.0008	0.01	0.03	0.06

^*Data exclusive of signal in large, conduit vessels. P values represent comparisons when the overall main effect for group was significant.

Figure 3. Perfusion heterogeneity in patients with Fontan circulation and PAH patients compared to controls.

Top, heterogeneity of perfusion as measured by the relative dispersion of perfusion including gravitational effects and large vessels. There was a significant main effect of group (P = 0.002) and both patient groups differed from their respective controls, but the Fontan patients did not differ significantly from PAH patients. Bottom, isogravitational relative dispersion. The differences in heterogeneity between patient groups and their respective control groups persisted when the influence of gravity was eliminated (Omnibus P = 0.004), but the Fontan patients did not differ significantly from PAH. Horizontal bar and error bars are means ± SD for each group.

Discussion

The results of this study show that the spatial heterogeneity of perfusion was significantly increased in patients with Fontan circulation and in PAH patients compared to control subjects. The extent of heterogeneity was similar between the two patient groups and the differences from controls were not due to the effects of gravity or large vessel anatomy. Surprisingly, there were no differences in the gravitational gradient in perfusion for either patient cohort for reasons that are unclear. Possibly, the extent of remodelling was such that structural effects predominated, or alternately the small distance in the direction of the gravitational vector in supine posture may have made detection of gravitational gradients difficult.

Perfusion heterogeneity in patients with Fontan physiology

Overview.

Approximately 16,000 patients underwent Fontan palliation between 2001 and 2014 (Akintoye et al. 2019). For those patients surviving the initial perioperative period, the majority reach adulthood, and over 60% of patients survive 20 years after the operation. This decreases to <50% survival by 30 years post-procedure (Pundi et al. 2015), with heart failure as the biggest predictor of mortality (Alsaied et al. 2017). There are several variations in the Fontan circuit (see Gewillig, 2005 for review), but the absence of active pumping of blood into the pulmonary circulation is ubiquitous.

Pulmonary arterial pressure is typically not elevated in Fontan physiology, but because of the lack of active pumping of blood into the pulmonary circulation, a low pulmonary vascular resistance is essential to avoid complications. However, pulmonary vascular resistance may be increased (Khambadkone et al. 2003; Mitchell et al. 2004) for reasons that are unclear: low overall cardiac output, and low perfusion pressures may affect capillary recruitment; lack of pulsatile flow may also affect capillary dilatation and recruitment (Presson Jr et al. 2002) and lack of pulsatile flow may cause dysregulation of shear-stress-mediated release of vasoactive molecules, resulting in endothelial dysfunction (Busse & Fleming, 1998). Since long-term failure of Fontan procedure occurs when the single ventricle can no longer meet the circulatory demands, optimal perfusion of the lung at low pulmonary vascular resistance is critical to both maintain cardiac output (Gewillig & Goldberg, 2014) and prevent systemic venous congestion. For this reason, the use of pulmonary vasodilators is not uncommon (Ovaert et al. 2009; Goldberg et al. 2011; Rhodes et al. 2013).

Although ^99mTc lung perfusion scintigraphy and phase-velocity MRI have been used to evaluate the right to left shunt and patterns of drainage from the inferior vena cava into the right or left main pulmonary artery (Pruckmayer et al. 1999; Fratz et al. 2002), little is known about the spatial patterns of pulmonary perfusion in Fontan physiology. In the present study all of our Fontan patients had adequate ventricular function, and pulmonary vascular resistance measured by right heart catheterization was low. Nonetheless, these patients had significantly greater pulmonary perfusion heterogeneity than age-matched control subjects and the extent of the perfusion heterogeneity was comparable to patients with known pulmonary arterial hypertension (PAH).

Zone model effects on heterogeneity.

In the normal human lung, perfusion heterogeneity arises because of hydrostatic effects (i.e. Zone model effects: West et al. 1964a; Hughes et al. 1968), because of vascular branching structure affecting local resistances (Hlastala & Glenny, 1999; Glenny & Robertson, 2010) and also from local regulatory mechanisms such as hypoxic pulmonary vasoconstriction (see Glenny & Robertson, 2011 for review). After correction for tissue distribution, perfusion heterogeneity persists, as does a small gravitational component, even in supine posture (Hopkins et al. 2007) where the gravitation vector is operating over a smaller distance. Nonetheless, the spatial distribution of perfusion is relatively uniform in normal subjects, consistent with adequate ventilation-perfusion matching and gas exchange (Glenny & Robertson, 2011), although both perfusion heterogeneity (Levin et al. 2007) and ventilation-perfusion mismatch (Cardus et al. 1997) increase somewhat with increasing age. All of these mechanisms may be altered in disease.

There are several potential reasons why perfusion heterogeneity could be increased in patients with Fontan physiology. Pulmonary arterial pressures were relatively low in our Fontan patients and this could affect bulk perfusion in the direction of the gravitational vector by Zone model effects, if the hydrostatic pressure was sufficiently low in gravitationally non-dependent regions. This is unlikely because the measured pulmonary arterial pressure in these subjects averaged 11 ±n 2 mmHg (~15 cmH₂O). Consequently, in the supine posture, Zone 1 conditions are not expected and most of the lung is expected to be under Zone 3 conditions, with perhaps some contribution from Zone 2. In Zone 3, flow is dictated by the difference between pulmonary arterial and venous pressures. Although driving pressure is similar over the Zone 3 lung, the absolute pressure is greater in the most dependent regions. In Zone 2, blood flow is dependent on the difference between alveolar and pulmonary arterial pressure. Therefore, non-uniform dilatation and recruitment may occur if blood vessels in the dependent lung are more distended and/or more are recruited than in the non-dependent lung. However, the differences between Fontan patients and controls persist even when perfusion heterogeneity was evaluated only across isogravitational planes, and not in the direction of the gravitational vector. This is also supported by both the lack of differences in mean perfusion and the distribution of perfusion as a function of height between groups. However, since we studied our subjects in supine posture we cannot rule out a greater influence of Zone model effects in upright posture.

Large vessel effects on heterogeneity.

Another potential mechanism for increased heterogeneity in perfusion images is the presence of large ‘conduit’ blood vessels. In ASL images, some blood vessels that supply or drain lung regions outside the image plane are seen in the images of perfusion and contribute signal. The signal from these structures is not perfusion, i.e. the delivery of blood to the capillary bed, if the blood is destined for another part of the lung not contained in the image. Conduit vessels contribute high signal to the images and have the potential to alter metrics of heterogeneity, especially in the presence of disease affecting the pulmonary vasculature that might alter the size and shape of these structures. However, the difference in heterogeneity was significant between groups when this signal was eliminated, and so this cannot be the explanation for the difference we observed.

Perfusion heterogeneity in pulmonary arterial hypertension

Overview.

Like patients with Fontan physiology, the quantitative evaluation of the distribution perfusion is limited in PAH. PAH arises because of vascular remodelling of the distal pulmonary arterioles resulting in increased precapillary resistance (Thenappan et al. 2018). The histopathology has been well described and consists of hypertrophy of the media of small pulmonary vessels and intimal fibrosis, arranged in concentric, laminar structures with an onionskin appearance (Wagenvoort & Wagenvoort, 1970). In addition, fibrinoid vascular necrosis and arteritis and regions of dilated thin-walled vessels are reported. Plexiform lesions, consisting of closely packed endothelial cells in a mucopolysaccharide matrix are also described (Wagenvoort & Wagenvoort, 1970; Fishman, 2000). These alterations in the vascular structure all have the potential to increase perfusion heterogeneity.

Although imaging forms an important part of patient evaluation and many studies have reported right heart structure and function (Francois & Schiebler, 2016; Freed et al. 2016) as well as flow velocities in the main pulmonary vessels, the spatial distribution of pulmonary blood flow in PAH has not been well described. In PAH alterations in perfusion, chiefly patchy defects, are reported (Hachulla et al. 2016; Giordano et al. 2017), and the number of such defects correlates with disease severity (Ameli-Renani et al. 2014). Increased perfusion heterogeneity has been reported with positron emission tomography imaging in patients with exercise-induced PAH compared to controls (Kohli et al. 2019). In our study we also found a significant increase in perfusion heterogeneity in PAH patients compared to controls.

Zone model and large vessel effects on heterogeneity.

The high pulmonary arterial pressure that is the hallmark of PAH is not expected to increase perfusion heterogeneity per se. Rather, increased pressure is expected to reduce any underlying heterogeneity arising due to Zone model effects and uneven dilatation and recruitment. Since these mechanisms are expected to reduce perfusion heterogeneity, it is unlikely that increased pulmonary arterial pressure is the cause of the increased perfusion heterogeneity. Consistent with this, single-photon emission computed tomography imaging shows a loss of the normal redistribution of perfusion following a change in posture (Lau et al. 2014) in PAH. Similarly, and as for the Fontan patients, since the increased heterogeneity persists when signal for larger vessels is eliminated, the increased heterogeneity cannot be explained on the basis of large vessel anatomical changes. Instead, the most likely explanation in these patients is remodelling due to the underlying disease process: lung regions distal to remodelled arterioles will have reduced perfusion whereas regions that have not undergone remodelling will have increased perfusion such that overall heterogeneity is increased.

The similarity of the extent of heterogeneity between PAH patients and those with Fontan physiology suggests that underlying remodelling is also a possible explanation for our findings in Fontan patients, where remodelling may arise from a lack of pulsatile pulmonary flow. In animal models, non-pulsatile flow is associated with endothelial dysfunction and apoptosis of vascular smooth muscle cells, and in Fontan patients there are increases in pulmonary arterial intimal layer thickness on post-mortem histological examination (Ridderbos et al. 2015). These changes may be partly reversible, however, as pulmonary vascular resistance has been shown to normalize in individuals with Fontan physiology after heart transplant restores normal cardiac anatomy and pulsatile pulmonary flow (Kaza et al. 2015). Alternately, lack of pulsatile flow may increase perfusion heterogeneity by purely mechanical effects: pulsatile flow has been shown to recruit capillaries to a greater extent than steady flow at a given perfusion pressure (Presson Jr et al. 2002).

Gravitational gradients in perfusion in Fontan physiology and PAH

The Zone model of pulmonary perfusion (Permutt et al. 1962; West et al. 1964b) describes the distribution of blood flow based on the relationship between alveolar, pulmonary arterial and pulmonary venous pressures. Both pulmonary arterial and venous pressures vary hydrostatically as a result of gravity, and thus are higher in the gravitationally dependent (i.e. the base of the upright lung) than the non-dependent lung. In supine posture, particularly at FRC, the vertical distance between the heart and non-dependent lung is small, and most of the lung is expected to be under Zone 3 conditions. Regions of Zone 2 may be present and Zone 4 conditions (see below) have been described (Levin et al. 2007; Prisk et al. 2007; Arai et al. 2009; Hopkins et al. 2010)).

We hypothesized that the gravitational gradients in perfusion would be increased in those with Fontan physiology and reduced in those with PAH, consistent with Zone model predictions. The finding that the gradients in perfusion were not different in either patient group was surprising, since previous studies have shown a redistribution of pulmonary blood flow to the non-dependent lung in normal humans and animals when pulmonary arterial pressure is increased by alveolar hypoxia (Fowler & Read, 1963; Dawson, 1969; Wagner Jr et al. 1979). We are not aware of any previous studies in those with Fontan physiology, but there are limited data in patients with PAH. Consistent with the present work, Ley et al. (2007) found similar dorsal to ventral blood flow distributions between PAH and controls measured in supine posture with contrast enhanced MRI. However, previous work using contrast enhanced electron-beam computed tomography showed a loss of gravitational gradient in perfusion in supine posture at full inspiration in PAH (Jones et al. 2004), and data obtained using SPECT-CT shows a loss of redistribution of perfusion with the transition from supine to upright posture in PAH patients compared to controls (Lau et al. 2014). The loss of redistribution was associated with markers of disease severity related to right heart function and functional class.

The reasons for the differences between some of these prior studies and the present work may relate to several factors. First, in the face of extensive vascular remodelling, patchy increased precapillary resistance may disrupt the normal pressure relationships in the lung such that structure becomes the predominant determinant of flow (Glenny & Robertson, 2010). Second, the constraints of the MRI scan environment dictated that our subjects were studied supine and at FRC and thus the vertical distance in the direction of the gravitational vector was small (~10-15 cm) compared to these prior patient studies, and any differences may have been subtle. Thirdly, we were only able to study a single posture and we could not evaluate redistribution with changes in posture. Finally, like previous studies, we had a small number of subjects and we cannot rule out the possibility that the results are unique to this population of subjects.

In addition to the similarity of the overall gravitational gradients, perfusion in the most dependent lung was also similar between groups. In the original work describing Zone 4, Hughes attributed the progressive reduction in the most dependent lung to increased resistance in extra-alveolar vessels (Hughes et al. 1968), since this region of the lung did not obey Zone model predictions. The calibre of these vessels was described as being determined by a balance between expansion of the lung parenchyma by radial traction (Permutt, 1965) acting to increase vessel calibre and interstitial pressure and inherent vascular tone decreasing (Hughes et al. 1967) vessel calibre, with the latter two effects predominating. This was based measurement of regional perfusion at total lung capacity, functional residual capacity and residual volume and the technology available to Hughes and colleagues did not permit either absolute quantification of perfusion or correction for gravitationally based tissue deformation. More recently our group showed in normal subjects that perfusion in the most gravitationally dependent lung did not change significantly between residual volume and total lung capacity when these factors were taken into account. This suggests that increasing lung volume may increase resistance to flow by elongating pulmonary blood vessels axially counterbalancing the effects of radial traction, consistent with isolated lung studies. (Sun et al. 1987).

Another explanation for Zone 4 effects is that dependent airway closure and lower ventilation-perfusion ratio in dependent regions may reduce perfusion in these regions (Prefaut & Engel, 1981; Petersson et al. 2006) if hypoxic pulmonary vasoconstriction is activated. If this is the case, one might expect greater Zone 4 effects, because of a lower overall arterial P_O2 in the patients with PAH or Fontan physiology. The data in the present study do not support this mechanism of Zone 4. Consistent with this, we have shown that hyperoxic gas mixtures have little effect on the distribution of pulmonary blood flow measured at FRC (Arai et al. 2009), in normal subjects. Additionally, modelling work implicates vascular structure and longer vascular pathways as important determinants of the apparent Zone 4 effect (Burrowes et al. 2005a,b).

Clinical implications

In the present study, the Fontan and PAH patient’s spirometry, while low or low-normal in most cases, was not markedly abnormal. Global measures of lung function (i.e. spirometry, mean pulmonary arterial pressure measurements) are dominated by information from lung that is relatively spared by the disease process. This means that there is considerable loss of function before symptoms develop and functional status declines. Thus, evaluation of regional measures of function such as perfusion heterogeneity may provide increased sensitivity to underlying abnormalities. ASL-FAIRER is highly reproducible (Levin et al. 2007), completely non-invasive, and high resolution repeated measurements can be made in human subjects without contrast injection or ionizing radiation. For this reason, it is suitable for evaluating the distribution of pulmonary perfusion in a variety of patient populations and may potentially provide information useful for monitoring both the patient populations studied here, and possibly others with pulmonary vascular disease. For example, these techniques may be useful in the early detection of pulmonary vascular disease in patients with unexplained exertional dyspnoea. Alternately it is possible that measurement of perfusion heterogeneity could provide information about the acute response to vasodilators such as inhaled nitric oxide. However, the clinical significance of the findings of increased heterogeneity is uncertain. Particularly in the PAH cohort there were marked differences in the extent of perfusion heterogeneity for reasons that are unclear. While the two subjects with the highest relative dispersion had the lowest oxygen saturation on pulse oximetry, they were otherwise unremarkable.

Study limitations

The study is limited by the small patient populations studied. Although the differences between patients and controls were highly significant, they may not be reflective of a larger patient population. We studied only a single slice of the right lung. While the anatomical location was similar between groups, this is a limitation of the present study. In addition, the MRI scanner used necessitated that we study our subjects supine, where the effects on gravity on the distribution of pulmonary perfusion are expected to be lessened compared to upright posture. As mentioned previously, it is possible that differences in the gravitational distribution of perfusion between groups may have been more apparent in upright posture. The finding of greater perfusion heterogeneity is non-specific, and changes may arise from several potential mechanisms and are unlikely to be diagnostic of any particular disease. For example, our group has previously shown changes in perfusion heterogeneity during (Hall et al. 2014) and after (Burnham et al. 2009) exercise, with normal ageing (Levin et al. 2007) and with exposure to hypoxia in individuals with a history of high altitude pulmonary oedema (Hopkins et al. 2005). In Fontan physiology the lung is not perfused by active pumping and for this reason the change in intrathoracic pressure during breathing may affect perfusion in these individuals to a greater extent than in our other study groups. The images were obtained during an ~8–10 s breath-hold at FRC and the subject was instructed to maintain an open glottis, but we cannot rule out that suspension of respiration may have affected the results.

Thus, although changes in local vascular structure due to remodelling are a possible explanation for our findings, it is not proof of this mechanism.

Conclusions

Both patients with Fontan physiology and PAH have an increase in the spatial heterogeneity of perfusion measured with proton MRI compared to controls. These differences were not a result of gravitational influences, Zone model effects or large vessel structural abnormalities. The magnitude of the increase in Fontan patients was similar to those with PAH and may indicate underlying vascular remodelling in these individuals.

Key points.

• The distribution of pulmonary perfusion is affected by gravity, vascular branching structure and active regulatory mechanisms, which may be disrupted by cardiopulmonary disease, but this is not well studied, particularly in rare conditions.

• We evaluated pulmonary perfusion in patients who had undergone Fontan procedure, patients with pulmonary arterial hypertension (PAH) and two groups of controls using a proton magnetic resonance imaging technique, arterial spin labelling to measure perfusion. Heterogeneity was assessed by the relative dispersion (SD/mean) and gravitational gradients.

• Gravitational gradients were similar between all groups, but heterogeneity was significantly increased in both patient groups compared to controls and persisted after removing contributions from large blood vessels and gravitational gradients.

• Patients with Fontan physiology and patients with PAH have increased pulmonary perfusion heterogeneity that is not explainable by differences in mean perfusion, gravitational gradients, or large vessel anatomy. This probably reflects vascular remodelling in PAH and possibly in Fontan physiology.

Translational perspective.

Under normal conditions very low hydrostatic pressure results in less flow at the apex of the upright lung than at the base, but this may be affected by very high or low pressures in disease. Active regulatory mechanisms aid in the matching of perfusion to ventilation but may also disrupt the uniformity of perfusion. Finally, perfusion heterogeneity may also be affected by vascular remodeling. By evaluating the distribution of perfusion, functional imaging may offer increased ability to detect abnormalities in the pulmonary circulation that may be useful for diagnosis or monitoring.

We evaluated pulmonary perfusion in patients who had undergone Fontan procedure for congenital heart disease, where the vena cavae are directly anastomosed to the pulmonary arteries and perfusion is passive, using a proton magnetic resonance imaging technique, arterial spin labelling (ASL). We also assessed patients with pulmonary arterial hypertension, PAH, and two groups of healthy controls. We hypothesized that both PAH patients and those with Fontan circulation would have an increase in perfusion heterogeneity consistent with underlying remodeling of the pulmonary circulation and also have altered gravitational gradient compared to controls.

Surprisingly, there were no significant differences in gravitational gradients in perfusion. Perfusion heterogeneity was similar both patient groups and greater than controls suggesting that underlying remodeling may explain our findings in Fontan patients. The clinical significance of these findings are unclear. However,ASL is noninvasive and without ionizing radiation and thus may be useful in monitoring or the early detection of pulmonary vascular disease.

Biography

Susan R. Hopkins is Professor of Medicine and Radiology and Director of the Pulmonary Imaging Laboratory at UC-San Diego. Dr Hopkins obtained her Medical degree from Memorial University of Newfoundland and completed a fellowship in Sports Medicine and a PhD (Pulmonary and Exercise Physiology) at the University of British Columbia, before joining John West and Peter Wagner at the UC-San Diego as a postdoctoral fellow. Her research focuses on the lung response to stress such as hypoxia, exercise and pulmonary disease. In addition to classic gas exchange techniques such as the multiple inert gas elimination technique (MIGET), she uses novel functional MRI techniques to quantify ventilation, perfusion and ventilation–perfusion ratio distributions to offer insights into mechanisms of disease.

Data availability statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.