Effect of Local Tidal Lung Strain on Inflammation in Normal and Lipopolysaccharide-Exposed Sheep

Abstract

Objective

Regional tidal lung strain may trigger local inflammation during mechanical ventilation, particularly when additional inflammatory stimuli are present. However, it is unclear whether inflammation develops proportionally to tidal strain or only above a threshold. We aimed to: (1) assess the relationship between regional tidal strain and local inflammation in vivo during the early stages of lung injury in lungs with regional aeration heterogeneity comparable to that of humans; and (2) determine how this strain-inflammation relationship is affected by endotoxemia.

Design

Interventional animal study.

Setting

Experimental laboratory and positron emission tomography (PET) facility.

Subjects

Eighteen 2–4-month-old sheep.

Interventions

Three groups of sheep (n=6) were mechanically ventilated to the same plateau pressure (30–32 cmH₂O) with High-Strain (V_T=18.2±6.5 ml/kg, PEEP=0), High-Strain plus intravenous lipopolysaccharide (LPS) (V_T=18.4±4.2 ml/kg, PEEP=0), or Low-Strain plus LPS (V_T=8.1±0.2 ml/kg, PEEP=17±3 cmH₂O). At baseline, we acquired respiratory-gated PET scans of inhaled ¹³NN to measure tidal strain from end-expiratory and end-inspiratory images in six regions of interest (ROIs). After 3 hours of mechanical ventilation, dynamic [¹⁸F]fluoro-2-deoxy-D-glucose (¹⁸F-FDG) scans were acquired to quantify metabolic activation, indicating local neutrophilic inflammation, in the same ROIs.

Measurements and Main Results

Baseline regional tidal strain had a significant effect on ¹⁸F-FDG net uptake rate K_i in High-Strain LPS (p=0.036) and on phosphorylation rate k₃ in High-Strain (p=0.027) and High-Strain LPS (p=0.004). LPS exposure increased the k₃-tidal strain slope 3-fold (p=0.009), without significant lung edema. The Low-Strain LPS group showed lower baseline regional tidal strain (0.33±0.17) than High-Strain (1.21±0.62; p<0.001) or High-Strain LPS (1.26±0.44; p<0.001), and lower k₃ (p<0.001) and K_i (p<0.05) than High-Strain LPS.

Conclusions

Local inflammation develops proportionally to regional tidal strain during early lung injury. The regional inflammatory effect of strain is greatly amplified by intravenous LPS. Tidal strain enhances local ¹⁸F-FDG uptake primarily by increasing the rate of intracellular ¹⁸F-FDG phosphorylation.

INTRODUCTION

Ventilator-induced lung injury (VILI) is thought to result from excessive deformation of the lungs during mechanical ventilation (1, 2). This deformation can be described by volumetric strain, defined as the change in lung volume relative to an initial lung volume (2). Protti et al. recently reported a critical threshold of whole-lung strain, corresponding to tidal volumes (V_T) >20 ml/kg, above which ventilator-induced lung edema ultimately develops in normal pigs (3). However, VILI has been shown to develop at significantly lower V_T in patients (4–7) and animals (8, 9). This discrepancy may be explained by at least two factors. First, even presumably safe tidal volumes may produce excessive local strains as a result of strain heterogeneity in supine patients (10) and large animals (11–15). Second, lung sensitivity to strain may increase in the presence of additional inflammatory stimuli, such as lipopolysaccharide (LPS) (16, 17).

Lung inflammation is a key early process in VILI that may precede and contribute to the development of edema. Bellani et al. indicated that pulmonary [¹⁸F]fluoro-2-deoxy-D-glucose (¹⁸F-FDG) uptake, a well-established marker of neutrophilic inflammation in lung injury (18–22), is linearly related to tidal gas volume changes in normally aerated regions of acute lung injury (ALI) patients mechanical ventilated for 9±7 days (23). However, because they studied the later stages of injury and did not measure tidal strain (24), the role of regional tidal strain in the early development of inflammation remains unclear.

We recently developed a respiratory-gated Positron Emission Tomography (PET) technique to measure regional tidal strain during mechanical ventilation (13), and advanced methods to quantify regional pulmonary inflammation from ¹⁸F-FDG kinetics (25, 26). In the present study, we leverage these PET techniques to investigate the following hypotheses in mechanically ventilated, heterogeneously aerated sheep lungs: (a) regional lung inflammation, assessed from ¹⁸F-FDG kinetics, develops in proportion to regional tidal lung strain; (b) the effect of tidal strain on regional inflammation is synergistically increased by systemic LPS exposure; and (c) reduction of regional tidal strain is a major determinant of decreased pulmonary inflammation during protective ventilation.

MATERIALS AND METHODS

Animal Preparation

Study protocols were approved by the Subcommittee on Research Animal Care of the Massachusetts General Hospital, and handling of the animals was in accord with National Institutes of Health guidelines. Sheep (21.5±4.9 kg) were fasted overnight and premedicated with intra-muscular ketamine (4 mg/kg) and midazolam (1–2 mg/kg). Following induction of anesthesia with intravenous propofol (2–4 mg/kg), an endotracheal tube was inserted, and a femoral artery and jugular vein were cannulated. General anesthesia was maintained with continuous infusion of propofol and fentanyl titrated to heart rate and blood pressure. Pancuronium was used for muscle paralysis at induction (0.1 mg/kg) and repeated every 90 minutes (0.02–0.04 mg/kg).

Experimental Procedures

We studied three groups of mechanically ventilated sheep (n=6 animals/group). The first group (High-Strain) aimed to produce high regional tidal strain while avoiding excessive plateau pressure (P_PLAT) by using zero positive end-expiratory pressure (PEEP) and continuously adjusting V_T to maintain P_PLAT within 30–32 cmH₂O (baseline V_T=18.2±6.5 mL/kg). The second group (High-Strain LPS) combined the same protocol (baseline V_T=18.4±4.2 mL/kg, PEEP=0) with continuous intravenous LPS infusion (10 ng/kg/min, Escherichia coli O55:B5, List Biologic Laboratories Inc., Campbell, CA) for the duration of ventilation (3.1±0.8 h), beginning after baseline image acquisition. The third group (Low-Strain LPS) received the same LPS dose and protective ventilation (27), with V_T=8 mL/kg and PEEP continuously adjusted to maintain P_PLAT=30–32 cmH₂O (baseline PEEP=17.1±3.4 cmH₂O). This protocol aimed to reduce the tidal (i.e. dynamic) component of lung strain, while maintaining peak strain similar to the other groups.

In each animal, a recruitment maneuver (40 seconds at airway pressure = 35 cmH₂O) was performed at baseline to standardize lung volume history. Additional ventilatory settings were: inspired O₂ fraction (F_IO₂) initially at 0.3 and adjusted to achieve an arterial O₂ saturation >0.88, inspiratory-to-expiratory time ratio I:E=1:2, and respiratory rate RR=18 breaths/minute or higher to maintain the arterial carbon dioxide pressure (P_aCO₂) between 32 and 45 mmHg. A variable dead space was added to the breathing circuit if P_aCO₂ was <32 mmHg with RR=18 breaths/minute.

PET Imaging

Sheep were positioned supine in the PET camera (Scanditronix PC4096, GE Healthcare, Milwaukee, WI), with the most caudal slice of the 9.7 cm field-of-view adjacent to the diaphragmatic dome. The following scans were acquired after 10 minutes (baseline) or 3.1±0.8 hours of mechanical ventilation (end):

1. Transmission scans (baseline, end): obtained during 10 minutes of continuous breathing to correct emission scans for tissue attenuation and measure lung density. These scans were processed to construct images of average fractional gas content at baseline (F_{GAS, BL}) and end (F_{GAS, END}) (13).

2. Respiratory-gated ¹³NN emission scans (baseline, end): acquired to measure regional tidal strain from regional aeration at end-expiration and end-inspiration (13). Briefly, ¹³NN gas was added to a rebreathing system and equilibrated with the lungs for 6–10 minutes (gas activity~300 kBq/mL). Respiratory-gated imaging was used to track the dynamic changes in pulmonary ¹³NN concentration during continuous mechanical ventilation. Over a 5-minute acquisition period, images of regional ¹³NN activity were collected into six bins of equal duration, representing six distinct phases of the respiratory cycle (i.e., two bins during inspiration and four during expiration).

3. ¹⁸F-FDG emission scans (end): acquired to assess local metabolic activation (18). ¹⁸F-FDG (~200 MBq) was infused over 60 s through a jugular catheter. Simultaneously, a dynamic PET scan was started, consisting of 37 sequential frames (9×10 s, 4×15 s, 1×30 s, 7×60 s, 15×120 s, 1×300 s).

PET images were reconstructed with voxel size of 2×2×6.5 mm using a convolution backprojection algorithm. Images were decay corrected to time zero and filtered in-plane with a circular moving average filter of diameter 12 mm and along the z-axis with a 2-point moving average filter. Each frame yielded a 128×128×14 matrix with effective volumetric resolution of 1.66 cm³.

Definition of Lung Fields for Analysis

Volumetric masks of lung fields were delineated at both time points (baseline, end) for end-expiration (EE), end-inspiration (EI), and mean lung volume by: (a) including all voxels with F_GAS>0.5; (b) adding perfused but poorly-aerated lung regions viewed in the end-apnea frames of the ¹³NN infusion scan; and (c) manually excluding the trachea, two main bronchi, and major blood vessels. Six regions-of-interest (ROIs) were defined for each animal by dividing the lung field in half along the cephalo-caudal axis, and further dividing each half along the ventral-dorsal axis into three sub-regions of equal height (Figure 1A).

Figure 1.

(A) Configuration of regions-of-interest (ROIs) for sVol computation, (B) distributions of baseline sVol (sVol_BL) along the gravitational axis, and (C) changes in sVol (ΔsVol) from baseline to the study end (sVol_END). Use of 6 ROIs in each sheep (A) resulted in a median ROI volume of 89 mL, with distinct ROI positions along the gravitational axis. Mixed-model regression lines in (B) show a significant gravitational dependence of sVol_BL in the combined High-Strain (gray triangles) and High-Strain LPS (black circles) groups, but not in the Low-Strain LPS group (white circles). Values of ΔsVol were mostly small (C), with averages (mean±SD) close to zero. No systematic change in sVol was observed in any of the groups.

Image Analysis

End-expiratory and end-inspiratory images of equilibrated ¹³NN were normalized by the ¹³NN specific activity to obtain images of fractional gas content at end-expiration (F_EE) and end-inspiration (F_EI) (13). In each ROI, tidal specific volume change (sVol, defined as the change in gas volume during inspiration divided by end-expiratory gas volume) was computed from average F_EE and F_EI, according to Fuld et al. (11):

$$\mathit{sVol}=\frac{F_{EI}-F_{EE}}{F_{EE}(1-F_{EI})}$$

In corresponding ROIs in the ¹⁸F-FDG images, the three-compartment Sokoloff model (28) was fit to ¹⁸F-FDG kinetics using iterative optimization. The ¹⁸F-FDG concentration within a blood pool ROI drawn over the right heart was used as an input function for the model after calibration with manual blood samples (29). For each ROI, we obtained three parameters describing ¹⁸F-FDG kinetics: the fractional distribution volume of ¹⁸F-FDG in tissue (F_e), the rate of ¹⁸F-FDG phosphorylation (k₃), and the net uptake rate (K_i), where K_i=F_e·k₃.

Statistical Analysis

Data are presented as mean ± standard deviation unless otherwise noted, and significance was set at p<0.05. Global variables were compared between groups using one-way ANOVA with Tukey-Kramer post-hoc tests for normally distributed data, or Kruskal-Wallis with Bonferroni-corrected Mann-Whitney tests otherwise. All regional measurements, including sVol, F_EE, F_EI, K_i, F_e, and k₃, were compared between groups using linear mixed-effects models with group (categorical) as a fixed effect and random intercepts for each animal to account for repeated measurements (i.e., six ROIs per animal). To test for effects of ROI position on sVol, we also modeled sVol with fixed effects for ROI ventral-dorsal and cephalo-caudal positions in each group, and random intercepts and position effects in each animal. To determine effects of sVol on ¹⁸F-FDG parameters, we modeled each ¹⁸F-FDG parameter (K_i, F_e, and k₃) as an interaction between sVol and group (i.e. allowing different effects of sVol in each group) and included random intercepts and sVol effects for each animal. A heterogeneous covariance structure was used in the K_i, F_e, and k₃ models, to allow for different variances among groups. All analyses were performed using Matlab (R2013b Statistics Toolbox, The Mathworks, Natick, MA).

RESULTS

Global Cardiorespiratory Variables

Consistent with the study design, High-Strain groups received higher V_T and lower levels of PEEP and respiratory rates than the Low-Strain LPS group (Table 1). P_aO₂/F_IO₂ decreased by the end of the study only in the High-Strain LPS group.

Table 1.

Weight and Cardiorespiratory Variables

	High-Strain (n=6)		High-Strain LPS (n=6)		Low-Strain LPS (n=6)
	Baseline	End	Baseline	End	Baseline	End
Weight, kg	20 (18–21)		23 (21–24)		20 (19–21)
VT, mL·kg−1	18.2 (17.0–23.5)b	20.0 (13.9–24.3)b	18.8 (15.9–21.5)b	13.8 (11.5–18.7)b,e	8.1 (8.0–8.3)	8.0 (7.9–8.1)
PEEP, cmH2O	0 ± 0b	0 ± 0 b	0 ± 0 b	0 ± 0b	17 ± 3	17 ± 2
RR, bpm	19 ± 1b	19 ± 3	19 ± 2b	21 ± 3	25 ± 2	26 ± 2
FIO2	0.3 (0.3–0.3)	0.35 (0.3–0.4)	0.3 (0.3–0.35)	0.35 (0.3–0.6)	0.3 (0.3–0.4)	0.3 (0.3–0.4)
CO, L·min−1	5.9 ± 1.3 b	6.3 ± 1.2b	4.4 ± 1.1	4.4 ± 0.7 b,c	3.3 ± 0.7	3.2 ± 0.5
MAP, mmHg	74 ± 14	81c 13	92 ± 12	78 ± 13 f	87 ± 13	81 ± 13
MPAP, mmHg	14 ± 8	15 ± 9 b	16 ± 9	21 ± 5 a	22 ± 6	32 ± 6 f
PaO2/FiO2, torr	243 ± 66	228 ± 88	255 ± 74	162 ± 67d	351 ± 117	261 ± 112
PaCO2, torr	34 (30±41)	36 (32±51)	33 (30±40)	43 (41±46)	39 (36±42)	39 (33±43)

V_T=tidal volume; PEEP=positive end-expiratory pressure; RR=respiratory rate; F_IO₂=inspired O₂ fraction; CO=cardiac output; MAP=mean arterial pressure; MPAP=mean pulmonary arterial pressure; P_aO₂=arterial O₂ pressure; P_aCO₂=arterial CO₂ pressure. Data are shown as mean ± standard deviation if normally distributed, or median (25^th–75^th percentile) otherwise. Groups were compared using ANOVA with Tukey-Kramer post-hoc tests for normally distributed data, or Kruskall-Wallis with Bonferroni-corrected post-hoc tests otherwise, Time points were compared using paired t-tests.

^ap<0.05 vs. Low-Strain LPS group.

^bp<0.01 vs. Low-Strain LPS group.

^cp<0.05 vs. High-Strain group.

^dp<0.10 vs baseline.

^ep<0.05 vs baseline.

^fp<0.01 vs baseline.

Topographic Heterogeneity of Tidal Strain

At baseline, the High-Strain groups showed large, heterogeneous changes in aeration between end-expiration and end-inspiration (Figure 2, Supplemental Digital Content 1). In contrast, animals in the Low-Strain LPS group showed smaller and more homogeneously distributed tidal changes in aeration. There was a significant gravitational dependence of sVol, with highest values in dependent regions, in both the High-Strain (p<0.001) and High-Strain LPS groups (p=0.004), but not in the Low-Strain LPS group (Figure 1B). No dependence of sVol on ROI cephalo-caudal position was detected in any group.

Figure 2.

Images of transverse planes from a representative animal in each group. Fractional gas content at end-expiration (F_EE) and end-inspiration (F_EI) as well as ¹⁸F-FDG net uptake rate (K_i) showed notable heterogeneity along the gravitational axis in both High-Strain groups. With High-Strain ventilation, K_i increased with LPS exposure, particularly in the dependent lung. Concentrated ¹⁸F-FDG uptake was not observed with protective ventilation (Low-Strain), where F_EE and F_EI were more homogeneous and LPS appeared to produce a small global increase in K_i.

The magnitude of regional sVol was different among the groups at baseline, as expected from the applied ventilator settings. Both High-Strain groups presented significantly higher sVol than the Low-Strain LPS group (Table 2, Figure 1B). Of note, in the High-Strain groups, a considerable fraction of ROIs presented sVol>1.5 (28% of all non-LPS ROIs, 17% for LPS), with some regions even showing sVol>2 (8% for both), despite global sVol values (computed from whole-lung average F_EE and F_EI) being close to 1 (Table 2). Measurements of sVol at the end of the experiments reflected similar distributions to those at baseline (Table 2, Figure 1C). No systematic changes in sVol over time were observed in any of the groups. Strong correlations were found between sVol at the end and baseline sVol for measurements from all groups (sVol_end=0.76·sVol_baseline+0.15, r=0.76, p<0.001), indicating stable topographic distributions of sVol over time.

Table 2.

Regional Aeration and Regional and Global Tidal Strain

Regional Variables:	High-Strain		High-Strain LPS		Low-Strain LPS
	Baseline	End	Baseline	End	Baseline	End
FEE	0.45 ± 0.14 a	0.41 ± 0.14 a	0.50 ± 0.16 a	0.42 ± 0.17 a	0.69 ± 0.08	0.60 ± 0.12
ΔFEE	−0.04 ± 0.08 b		−0.08 ± 0.10 b		−0.09 ± 0.10 b
FEI	0.62 ± 0.12 a	0.59 ± 0.14	0.67 ± 0.16	0.57 ± 0.21	0.74 ± 0.08	0.65 ± 0.12
ΔFEI	−0.04 ± 0.05 b		−0.11 ± 0.14 b		−0.10 ± 0.10 b
sVol	1.21 ± 0.62a	1.22 ± 0.51a	1.26 ± 0.44 a	1.05 ± 0.58a	0.33 ± 0.17	0.29 ± 0.18
ΔsVol	0.02 ± 0.49		−0.21 ± 0.51		−0.05 ± 0.11
Global sVol	1.17 ± 0.49a	1.08 ± 0.23 a	1.06 ± 0.20 a	0.87 ± 0.31 a	0.31 ± 0.09	0.27 ± 0.13

F=fractional gas content; EE=end-expiration; EI=end-inspiration; Δ=change in parameter from baseline to end; sVol=tidal specific volume change (i.e., tidal strain); global sVol=sVol measured from whole-lung average F_EE and F_EI. Regional variables were compared between groups with a linear mixed-effects model including group as a fixed effect and random intercepts for each animal. Global sVol was compared between groups using ANOVA with Tukey-Kramer corrected post-hoc tests.

^ap<0.01 vs. Low-Strain LPS group.

^bp<0.01 for the hypothesis that the variable equals 0.

Relation of Regional Metabolic Activation with Tidal Strain

Regional ¹⁸F-FDG net uptake rate K_i, volume of distribution F_e, and rate of phosphorylation k₃ showed significant gravitational dependence (Supplemental Digital Content 2), with highest values in dependent regions (Figure 2). Applying the mixed-effects model to measurements in ROIs, K_i showed a significant linear association with sVol in the High-Strain LPS group (p=0.036), but not in the High-Strain or Low-Strain LPS groups (Figure 3A). The components of K_i (K_i=F_e·k₃) were differently associated with regional strain. Whereas the ¹⁸F-FDG volume of distribution F_e was not associated with sVol in any of the groups (Figure 3B), the ¹⁸F-FDG rate of phosphorylation k₃ was associated with sVol in the High-Strain (p=0.027) and High-Strain LPS groups (p=0.004) (Figure 3C) but not in the Low-Strain LPS group (Figure 3D). The High-Strain LPS group showed higher values of k₃ (p<0.001) than the Low-Strain LPS group. These findings imply that the increase in K_i with increasing sVol following LPS exposure occurred predominantly through a higher k₃, indicating enhanced intra-cellular phosphorylation of ¹⁸F-FDG in regions with higher tidal strain.

Figure 3.

(A) ¹⁸F-FDG net uptake rate K_i, (B) volume of distribution F_e, and (C,D) phosphorylation rate k₃ vs. regional sVol for all ROIs of all animals in the High-Strain group (gray triangles), High-Strain LPS group (black circles), and Low-Strain LPS group (white circles). Regression lines in (A,C) show significant effects of sVol in the High-Strain (gray line) and High-Strain LPS (black line) groups, determined with the mixed-effects models of K_i and k₃. No significant effects of sVol were found in the Low-Strain LPS group. (C) Comparing the High-Strain groups, the slope of k₃ vs. sVol was 3.3 times higher (p=0.009) and the intercept was 60% higher (p=0.108) with LPS exposure. (D) Comparing the LPS groups, the Low-Strain protocol led to lower values of sVol (p<0.001) and k₃ (p<0.001), which fell along the same k₃–sVol relationship defined by the High-Strain LPS group. When the mixed-effects model was fit to all LPS animals together (dotted line), the regression line was not significantly different from that of High-Strain LPS alone (solid line), implying a similar dependence of k₃ on sVol in all LPS animals.

Effect of LPS on the Relationship between Tidal Strain and Metabolic Activation

By exposing the lungs to LPS during High-Strain ventilation, the magnitude of the effect of sVol on k₃ increased by a factor of 3.3 (2.14×10⁻² vs. 0.65×10⁻² min⁻¹, p=0.009, Figure 3C). Thus, the lung metabolic response to tidal strain was enhanced by more than a factor of 3 in the presence of LPS. This increase in the k₃ vs. sVol slope with LPS exposure was clearly evident by examining individual animal regressions [0]in the mixed-effects model (Supplemental Digital Content 3). In addition to these differences in slope, intercepts of the mixed-effects model of k₃ tended to be higher in LPS groups (High-Strain =3.90×10⁻² min⁻¹, Low-Strain =3.86×10⁻² min⁻¹) than in the High-Strain group (2.45×10⁻² min⁻¹, p=0.108 vs. High-Stain LPS, p=0.022 vs. Low-Strain LPS, Figure 3C). These results suggest that LPS has both an independent effect on metabolic activation, as well as an important interaction with tidal strain that further increases metabolic activation.

Data from the Low-Strain LPS group were highly concentrated in the k₃–sVol plot (Figure 3D), highlighting the homogeneity of those variables in this group. Importantly, those points fell along the mixed-model regression line defined by the High-Strain LPS group. To show this quantitatively, the data were grouped according to LPS exposure and reanalyzed with the mixed-effects model (i.e., Low-Strain and High-Strain LPS groups combined). In this model, parameters of the combined LPS group were not significantly different from those of the High-Strain LPS alone (Figure 3D), in terms of slope (1.96×10⁻² [all LPS] vs. 2.14×10⁻² [High-Strain LPS only] min⁻¹, p=0.79) or intercept (3.58×10⁻² vs. 3.90×10⁻² min⁻¹, p=0.60). Given that data from all LPS-exposed animals followed the same global trend (dotted line, Figure 3D), the reduction in metabolic activation with protective (i.e., Low-Strain) ventilation presumably occurred through the reduction of sVol, shifting lung regions downward along that regression line.

The maximum regional k₃ was highest in the High-Strain LPS group (Figure 4). Among LPS-exposed animals, Low-Strain ventilation reduced the maximum k₃ by 43% (p<0.05). Comparing High-Strain groups, there was a trend toward higher minimum k₃ with LPS (p<0.10), supporting the presence of a small independent effect of LPS on metabolic activation even in regions of lowest tidal strain.

Figure 4.

Minimum and maximum regional k₃ in each group. Comparing High-Strain groups, LPS appeared to have a small effect on minimum k₃, and significantly increased maximum k₃. Comparing LPS groups, the Low-Strain LPS protocol greatly reduced the maximum k₃ in relation to High-Strain LPS, though minimum values of k₃ were similar. ^†p<0.10, *p<0.05, **p<0.01.

Additional Variables Influencing Regional Inflammation

In addition to tidal strain, we also studied the associations of k₃ with other variables potentially influencing inflammation: regional tidal recruitment, perfusion, average lung inflation level (F_GAS), hyper-inflation, consolidated atelectasis, and intra-regional heterogeneity of tidal strain (Supplemental Digital Content 1). Using the same mixed-effects model used to study the association of sVol with k₃, we found no associations of perfusion or hyper-inflation with k₃. Average F_GAS was inversely associated with k₃ in all groups. In the High-Strain LPS group but not in the High-Strain group, k₃ was associated with tidal recruitment of voxels with F_GAS<0.1 (p=0.019) and 0.1<F_GAS<0.3 (p<0.001; Figure 5), as well as with intra-regional heterogeneity of specific ventilation (p=0.002), measured from ¹³NN washout (30). Among all studied independent variables, sVol was the most strongly associated with regional k₃ in lungs with and without LPS exposure (Supplemental Digital Content 1).

Figure 5.

Relation between regional ¹⁸F-FDG phosphorylation rate k₃ and estimates of tidal recruitment of distinct aeration compartments, based on decrease in the fractions of voxels in those compartments during inspiration. Significance of the effect of each predictor variable, derived with the mixed-effects models, is shown for each group.

Histological Injury

Regional tissue injury was mild in all groups (Supplemental Digital Content 4). LPS exposure significantly increased tissue neutrophil counts in both ventral and dorsal regions, with similar neutrophil counts between LPS groups. No interstitial or alveolar edema was observed in any group.

DISCUSSION

Using sheep models of early acute lung injury, we found that: (a) mechanical ventilation in the supine position with zero PEEP and high tidal volumes results in topographically heterogenenous local tidal strain, with highest values in dependent regions; (b) in both normal and LPS-exposed lungs ventilated with high tidal volumes, ¹⁸F-FDG phosphorylation rate k₃ is significantly associated with regional tidal strain; (c) exposure to LPS substantially increases the slope of the k₃ vs. sVol regression, indicating local synergistic interaction between tidal strain and LPS; and (d) use of low tidal volumes and high PEEP decreases the magnitude of regional tidal strain and produces a corresponding reduction in k₃, supporting a direct effect of tidal strain on regional metabolic activation.

We found significant topographic heterogeneity of tidal strain in the High-Strain groups. Of note, a substantial fraction of lung regions presented sVol>1.5, despite global values being close to 1, reinforcing that high levels of regional strain may be present even when global strain is within a presumably safe range (11–15, 31). Although we used relatively large V_T to achieve those high strains, ARDS lungs may exhibit similar levels of strain at much lower V_T. In fact, Chiumello et al. found global strains in the range of 1–1.5 in patients ventilated with V_T between 6–12 ml/kg and PEEP = 5 cmH₂O (32). Considering the substantial heterogeneity of strain and compliance observed in ARDS models (14, 33), it is plausible that local strains as high as 2 could exist in patients with global strains >1.

Tidal strain appeared to have a direct effect on the regional ¹⁸F-FDG net uptake rate K_i in lungs with LPS exposure. We and others have established that K_i is a sensitive marker of neutrophilic inflammation in humans (34, 35) and animal models of lung injury (19–21). Since K_i is the product of k₃ and F_e, we further examined the dependence of these K_i components on tidal strain. We found that k₃ had the strongest associations with sVol, both with and without LPS exposure, while F_e showed no associations with sVol. In contrast to K_i and F_e, which are expected to be linearly related to local tissue density, k₃ should not depend directly on tissue density, as it represents the average rate of intracellular ¹⁸F-FDG phosphorylation within the tissue compartment of a region-of-interest (28). Indeed, given that K_i=F_e•k₃, if F_e is linearly related to tissue density (as supported by our data in Supplemental Digital Content 2), k₃ must be independent from it in order to confer the linear dependence of K_i on tissue density. Thus, k₃ is expected to depend only on the specific metabolic state of the cells present in the tissue, providing an index of cellular metabolic activity independent of tissue density. In the High-Strain group, increased k₃ in response to tidal stretching may have resulted from normal physiologic responses to stretch such as deformation-induced lipid trafficking (36) or cytoskeletal reorganization (37). Given the high neutrophil counts observed after LPS exposure, the increased metabolic activation with strain in LPS animals more likely reflects the triggering of inflammatory processes, with increased number and activation of neutrophils as well as contributions from other cells (38). Interestingly, while neutrophil counts were similar between the LPS groups, metabolic activation was significantly greater with High-Strain than with Low-Strain ventilation, suggesting that tidal strain primarily affected the activation of sequestered neutrophils, rather than causing recruitment of additional neutrophils.

Intravenous LPS remarkably amplified the effect of tidal strain on metabolic activation, increasing the slope of the k₃–sVol regression by more than a factor of 3. Thus, the increase in metabolic activation caused by a moderate LPS dose was larger in regions of higher tidal strain, implying the presence of local synergy between tidal strain and LPS. While synergy between stretch and LPS in producing cytokine release has been documented in isolated epithelial cells (39), macrophages (40), and small animal models (16, 17), our findings demonstrate the relevance of regional interactions between strain and other inflammatory stimuli in large, heterogeneous lungs. Such interactions could play an important role in generating the marked topographic heterogeneity of inflammation observed in ARDS patients (41). Additionally, increased sensitivity of the lungs to tidal strain when additional inflammatory stimuli are present may help to reconcile the findings of Protti et al., where injury did not occur until V_T>20 ml/kg (3), with evidence of VILI in patients ventilated with much lower V_T (5, 42–44).

The absence of histological lung edema in all groups, despite significant neutrophilic inflammation in the LPS groups, suggests that the inflammatory response to tidal strain may either precede the formation of edema, or occur at lower values of strain, below the threshold for edema if such exists. Accordingly, our findings suggest that metabolic changes may be more sensitive than measurements of edema for detection of early processes in the development of lung injury. Importantly, the observed association between regional lung strain and inflammation occurred over a range of strain values lower than those whole-lung strains previously observed to cause edema (3). Thus, even if strains applied in routine mechanical ventilation are not large enough to ultimately produce edema, such strains could still trigger early inflammation, potentially with clinically relevant consequences.

Mechanical and physiologic factors other than tidal strain may have provoked inflammation in our studies. In the High-Strain LPS group, we found associations of k₃ with tidal recruitment and ventilation heterogeneity (Supplemental Digital Content 1), though these were not as strong as the association with tidal strain. Nonetheless, it is possible that low-volume injury mechanisms may have contributed to the genesis of inflammation in this group. Indeed, high interfacial stresses that occur during tidal recruitment can injure epithelial cell plasma membranes (45, 46). In normal lungs (e.g., High-Strain group), healthy surfactant function may prevent or delay injury by these mechanisms. However, LPS-dependent surfactant dysfunction can produce higher surface tensions (47), with consequent increase in the risk of cell injury. Such effects of LPS could be partly responsible for the observed association between inflammation and tidal recruitment in the High-Strain LPS animals. However, our data point to tidal strain as a more important determinant of early regional inflammation in the studied normal and LPS-exposed lungs.

Our finding that average lung inflation (i.e., F_GAS) was negatively associated with metabolic activation (Supplemental Digital Content 1) suggests that inflammation was not provoked by high static (i.e., continuously applied) strain, but by the dynamic component of strain induced by tidal inflation. This result is in line with recent findings of dynamic strain as a more important determinant of whole-lung edema and inflammation than static strain (3). We expand on those findings by showing that this predominant sensitivity of the lungs to dynamic strain applies to the early regional development of inflammation, before any histologic measures of tissue injury or edema are detectable.

Methodological limitations to our study include: (a) Our measurements of sVol could be affected by tidal recruitment. PET estimates of recruitment were associated with k₃, yet less so than tidal strain and only in the High-Strain LPS group (Supplemental Digital Content 1). Higher resolution CT techniques may be required to further study the respective roles of recruitment and tidal strain. (b) We computed strain using end-expiratory gas volume in the denominator as done previously (48), though others have used functional residual capacity (49). While the definition of lung “resting volume” remains arbitrary (1) and will affect the computation of strain (2), our choice was consistent with the goal of quantifying the dynamic component of strain in the studied conditions. (c) Registration errors between end-expiration and end-inspiration could produce error in sVol measurements. We mitigated this by studying large ROIs (13) and computing sVol from average regional F_EE and F_EI, which are relatively insensitive to registration errors (11, 13); (d) Regional metabolic activation was used as a surrogate for inflammation. Although we did not measure tissue markers of inflammation (e.g. cytokines), ¹⁸F-FDG uptake in itself is an established marker of neutrophilic inflammation in acutely injured lungs (18–22, 50).

CONCLUSIONS

During mechanical ventilation of supine lungs similar in size to human lungs, high regional lung strains may be present even when global lung strain is within acceptable limits. Regional lung metabolic activation, reflecting the development of local inflammation, shows a positive linear relation with tidal strain, and this dependence is substantially amplified by moderate LPS exposure even when lung edema is not apparent. Such localized metabolic activation is prevented by reducing and homogenizing regional tidal strain with high PEEP and low V_T.