Pulmonary Hypertension in Idiopathic Interstitial Pneumonia Is Associated with Small Vessel Pruning

To the Editor:

Chest computed tomography (CT) can quantify both vascular and parenchymal lung disease; however, its role in the simultaneous assessment of pulmonary hypertension (PH) associated with interstitial pneumonia is undefined (1, 2). We sought to quantitatively compare the small pulmonary vasculature of patients with pulmonary fibrosis (PF) with PH (PF-PH) to those without PH (PF-NPH), while accounting for regional parenchymal heterogeneity.

Methods

We identified patients with idiopathic interstitial pneumonia (IIP) (determined retrospectively by multidisciplinary review, including explant pathology) who underwent right heart catheterization before lung transplantation at the UCLA Medical Center from 2013 to 2018 (IRB #11-003042). The subset with noncontrast, thin-slice (1 mm) CT imaging (n = 12) was identified, after meeting criteria for either severe PF-PH (mean pulmonary arterial pressure [mPAP] ⩾ 35 mm Hg and LVEDP ⩽ 15 mm Hg; or mPAP ⩾ 25 mm Hg and cardiac index < 2.5 L/min/m²) (3) or PF-NPH (mPAP ⩽ 20 mm Hg demonstrated on two right heart catheterizations, done at lung transplantation evaluation and day of lung transplantation).

Lung regions were automatically segmented, and pulmonary vasculature was reconstructed with the Chest Imaging Platform (cip.bwh.harvard.edu). Lung regions were divided into equal-volume zones (upper, middle, lower). Quantitative interstitial lung disease (QILD; includes fibrotic reticulation, honeycombing, and ground-glass patterns) and lung fibrosis (QLF; includes fibrotic reticulation pattern only) scores were measured as a percentage of total lung/region volume (1). The volume of small arteries or veins <5 mm² in cross-sectional area was normalized by total arterial (aBV5/total blood volume [TBV]) or venous volume (vBV5/TBV), respectively, as a measure of small artery/vein pruning (2). Arterial/venous BV5/TBV was measured for the whole lung and regions with the least QILD and most QLF. Pre-acinar arterial dilation reflected by small arterial vessel volume 5–20 mm² in cross-sectional area normalized by total arterial volume (aBV5–20/TBV) was also calculated. At the pulmonary artery bifurcation, main pulmonary artery and ascending aorta diameters were manually measured. Two-sided P values were calculated using exact Wilcoxon rank-sum and Fisher’s tests and Spearman correlations. Statistical analyses were performed in SAS 9.4 (SAS Institute). Some of these findings were presented at the 2023 American Thoracic Society meeting (4).

Results

The cohort consisted of patients with IIP with PF-PH (n = 7) or PF-NPH (n = 5). Table 1 lists baseline characteristics, hemodynamics, and CT measures. The PF-PH subgroup had elevated median mPAP of 40 mm Hg (interquartile range [IQR], 29.0–49.3 mm Hg) and pulmonary vascular resistance (PVR) of 5.2 Wood units (IQR, 5.0–6.8 Wood units). Subjects with PF-PH had increased whole lung QILD (Figure 1) and QLF scores. Minimal QILD predominated in upper-lung regions (n = 11; right [n = 3]; left [n = 8]), whereas maximal QLF was in lower-lung regions (n = 11; right [n = 6]; left [n = 5]). One patient had an inverse disease distribution, with minimal QILD in the left lower region and maximal QLF in the right upper lung.

Table 1.

Characteristics, Hemodynamics, and CT–derived Vascular and Parenchymal Measures between Patients with PF with and without PH

	PF-PH (n = 7)	PF-NPH (n = 5)	P Value
Characteristic
Age, yr	60.7 (56.6 to 64.7)	66.3 (63.3 to 68.8)	0.268
Sex, M, n (%)	7 (100)	4 (80.0)	0.417
FVC% predicted	42.0 (37.0 to 56.0)	44.6 (44.0 to 46.0)	0.639
DlCO corrected for Hgb, % predicted	25.5 (19.7 to 39.0)	35.9 (33.1 to 44.0)	0.126
Median time between RHC and CT, d	88.0 (4 to 145)	−4.0 (−21 to 119)	0.755
Median time from RHC to LT, d	66.0 (49 to 334)	115.0 (87 to 127)	0.639
Hemodynamics
mPAP, mm Hg	40.0 (29.0 to 49.3)	18.0 (17.3 to 18.3)	0.003
PVR, Wood units	5.2 (5.0 to 6.8)	1.6 (0.4 to 2.0)	0.003
CI, L/min/m2	1.9 (1.7 to 2.9)	2.9 (2.5 to 3.4)	0.202
CT vascular measures in the whole lung
Arterial BV5/TBV in whole lung, %	0.29 (0.23 to 0.32)	0.40 (0.35 to 0.42)	0.005
Venous BV5/TBV in whole lung, %	0.23 (0.16 to 0.25)	0.30 (0.27 to 0.39)	0.149
Combined BV5/TBV in whole lung, %	0.26 (0.19 to 0.27)	0.33 (0.30 to 0.40)	0.073
Arterial BV5–20/TBV in whole lung, %	0.38 (0.37 to 0.44)	0.33 (0.32 to 0.35)	0.018
PA diameter, mm	37.0 (32.8 to 38.9)	30.6 (29.9 to 33.5)	0.106
PA/A ratio	1.09 (0.88 to 1.10)	0.93 (0.93 to 0.99)	0.268
CT vascular measure in the lung region with minimal QILD
Arterial BV5/TBV in lung region with minimal QILD, %	0.36 (0.23 to 0.47)	0.54 (0.52 to 0.56)	0.003
Venous BV5/TBV in lung region with minimal QILD, %	0.30 (0.20 to 0.39)	0.51 (0.49 to 0.58)	0.048
Combined BV5/TBV in lung region with minimal QILD, %	0.33 (0.22 to 0.40)	0.52 (0.52 to 0.59)	0.018
CT vascular measure in the lung region with maximal QLF
Arterial BV5/TBV in lung region with maximal QLF, %	0.26 (0.24 to 0.38)	0.38 (0.33 to 0.39)	0.530
Venous BV5/TBV in lung region with maximal QLF, %	0.19 (0.11 to 0.25)	0.22 (0.19 to 0.27)	0.530
Combined BV5/TBV in lung region in maximal QLF, %	0.20 (0.14 to 0.31)	0.27 (0.20 to 0.33)	0.432
CT measures of parenchymal disease burden
Whole lung QILD, %	58.9 (54.6 to 66.6)	40.8 (34.5 to 42.3)	0.035
QILD in minimal lung region, %	34.5 (19.3 to 46.4)	19.0 (15.5 to 23.4)	0.073
Whole lung QLF, %	36.1 (23.3 to 44.9)	23.0 (19.3 to 24.6)	0.028
QLF in maximal lung region, %	80.6 (70.2 to 90.4)	55.8 (55.0 to 57.2)	0.010
Total lung volume (TLC) on CT, L	3.6 (3.2 to 4.3)	3.1 (2.6 to 3.3)	0.268

Definition of abbreviations: arterial BV5–20 = volume of pulmonary arteries with cross-sectional area between 5 mm² and 20 mm²; BV5 = volume of pulmonary arteries/veins with cross-sectional area < 5 mm²; CI = cardiac output measured by thermodilution, indexed to body surface area; CT = computed tomography; Hgb = hemoglobin; LT = lung transplantation; mPAP = mean pulmonary artery pressure; PA = pulmonary artery; PA/A = PA divided by aortic diameter ratio; PVR = pulmonary vascular resistance; QILD = quantitative interstitial lung disease; QLF = quantitative lung fibrosis; RHC = right heart catheterization; TBV = total blood volume of all pulmonary arteries or veins; TLC = total lung capacity.

Data are presented as median (interquartile range) unless otherwise noted. One patient’s Dl_CO measurement was missing.

Figure 1.

Examples of vascular reconstructions in a patient with pulmonary fibrosis (PF) (A) without pulmonary hypertension (PF-NPH) and (B) with PH (PF-PH). Note the loss of small vessels (orange; cross-sectional area of <5 mm²; BV5) in PF-PH, with redistribution of blood to larger (green) arteries (cross-sectional area between 5 mm² and 20 mm²). Representative axial computed tomography images for the corresponding patients with PF-NPH and PF-PH show arterial (blue) and venous (red) vessel segments superimposed on the parenchyma. In PF-PH versus PF-NPH, comparison of (C) small artery vessel volume reflected by whole lung arterial BV5/TBV and (D) quantitative interstitial lung disease burden. TBV = total blood volume.

Figure 1 displays a sample pulmonary vascular reconstruction. Whole lung aBV5/TBV in PF-PH was reduced compared with PF-NPH (0.29 [IQR, 0.23–0.32] vs. 0.40 [IQR, 0.35–0.42]; P = 0.005; Table 1); however, no differences were seen with vBV5/TBV. Whole lung aBV5–20/TBV was increased in PF-PH (0.38 [IQR, 0.37–0.44]) compared with PF-NPH (0.33 [IQR, 0.32–0.35]; P = 0.018). In the minimal QILD region, there were reductions in both aBV5/TBV and vBV5/TBV in PF-PH, compared with PF-NPH. Conversely, arterial/venous BV5/TBV between groups did not differ in the lung region with maximal QLF.

Across the PF cohort, whole lung aBV5/TBV was inversely associated with mPAP (ρ, −0.66; 95% confidence interval [CI], −0.89 to −0.12; P = 0.019) and PVR (ρ, −0.89; 95% CI, −0.97 to −0.62; P < 0.001), with similar findings in the minimal QILD region. vBV5/TBV in the minimal QILD region (not whole lung) also correlated with PVR (ρ, −0.61; P = 0.036), but not with mPAP.

Among CT measures of ILD/PF severity, whole lung aBV5/TBV negatively correlated with whole lung QILD (ρ, −0.68; 95% CI, −0.90 to −0.15; P = 0.014) and QLF (ρ, −0.70; 95% CI, −0.90 to −0.17; P = 0.012) scores, with similar findings in the minimal QILD region. vBV5/TBV in the whole lung and minimal QILD regions were not associated with QILD/QLF scores. aBV5/TBV and vBV5/TBV did not correlate with FVC or Dl_CO.

Discussion

In an advanced IIP cohort, the presence of severe PH (compared with no PH) on quantitative CT imaging was associated with: 1) increased whole lung small arterial pruning with evidence of compensatory pre-acinar arterial dilation; 2) increased small vein pruning, albeit only in the minimal ILD/PF region; 3) no differences in small artery or vein pruning in the lung region with maximum ILD/PF; and 4) increased whole lung ILD/PF burden.

CT imaging can quantify small vessel loss in PH (2). A report of the ILD/PF subgroup from the Assessing the Spectrum of Pulmonary hypertension Identified at a REferral centre Registry demonstrated CT small vessel pruning in severe PH compared with mild-to-moderate PH (5). We have extended these observations by incorporating quantitative assessments of ILD/PF and distinguishing the contribution of small arteries and veins to the overall burden of small vessel loss. Finally, in an effort to include an indisputable group 3 PH phenotype, only moderate and severe ILD/PF were included in this study, keeping in mind that mild/modest ILD has been routinely included in prior group 1 pulmonary arterial hypertension studies (6, 7).

The quantitative CT assessment of small vasculature is inherently compromised in areas of ILD/PF. As such, the region least affected by ILD/PF may be ideal to quantify the small pulmonary vasculature. When analyses were confined to this minimal ILD/PF (QILD) area, both arterial and venous BV5/TBV were significantly reduced in PF-PH, whereas similar assessments in the maximum ILD/PF region revealed no differences. Interestingly, prior evaluation of pulmonary vascular histologic changes in a similar population also found small vein remodeling to be most notable in the region least affected by ILD/PF (8). In addition, small vessel pruning (artery and vein) in the lung region with minimal QILD directly correlated with PVR at or above the 5 Wood unit threshold known to be associated with worse prognosis (9).

FVC—a routine measure of ILD/PF severity—did not correlate with pruning or the presence of PH, findings supported by prior work (10). Despite similar CT whole lung volumes, PF-PH was associated with significantly greater QILD/QLF compared with PF-NPH. As such, this study is the first to indicate a possible association between the burden of ILD/PF and PH. It should be noted, however, that there is considerable overlap between the CT parenchymal findings of ILD/PF (without PH) and group 1 pulmonary arterial hypertension, particularly ground-glass and reticulations, which may confound this finding (11).

Our study has limitations. Our cohort was small, albeit well characterized, owing to the inclusion of the opposite extremes of the PH hemodynamic phenotype and the requirement for retrospective multidisciplinary discussion, including explant pathology review. As we decided a priori to focus this study on the extremes of the precapillary PH hemodynamic spectrum in an advanced PF population, the findings may not necessarily be extrapolated to the mild-to-moderate phenotypes of PH and ILD/PF, which we plan to characterize in follow-up work.

Conclusions

In the context of IIP, severe precapillary PH is associated with small vessel pruning involving arteries and veins and is best assessed in the lung region with minimal ILD/PF. Quantitative assessments of ILD/PF suggest there is a relationship between the extent of ILD/PF and PH. Automatic quantitative processing of CT imaging for simultaneous pulmonary vasculature and parenchyma evaluation may distinguish patients with IIP with and without PH. These initial observations require confirmation in larger cohorts.