Assessment of Lung Inflammation With ¹⁸F-FDG PET During Acute Lung Injury

Abstract

Objective

The purpose of this review is to describe the current experimental and clinical data regarding the fundamentals and applications of ¹⁸F-FDG PET during acute lung injury (ALI) and acute respiratory distress syndrome (ARDS).

Conclusion

Lung inflammation is a key feature of ALI. During ALI, FDG PET can be used to monitor lung neutrophils, which are essential cells in the pathophysiologic mechanisms of ALI. Pulmonary FDG kinetics are altered during experimental and human ALI and are associated with regional lung dysfunction, histologic abnormalities, and prognosis. FDG PET may be a valuable noninvasive method for gaining comprehensive understanding of the mechanisms of ALI/ARDS and for evaluating therapeutic interventions.

PET after IV administration of ¹⁸F-FDG has been used to study the global and regional metabolism of solid organs such as the brain and the heart [1–3] and has become a mainstay in oncology for diagnosis and treatment monitoring [4, 5]. Given that tissue FDG uptake is determined essentially by metabolic activity, it has also been used to study inflammation in nontumoral conditions [6, 7].

Acute lung injury (ALI) and acute respiratory distress syndrome (ARDS) are frequently encountered in critically ill patients and result in significant morbidity and mortality [8, 9]. Pulmonary inflammation is a key feature of both conditions and bears great relevance in their progress and resolution [10–12]. Accordingly, there is considerable interest in the application of FDG PET to ALI and ARDS because noninvasive and accurate quantification of pulmonary inflammation throughout the lung could be obtained, improving understanding of the disease mechanisms and assessment of therapeutic strategies. FDG PET also is an important advance over existing methods of studying lung inflammation. Open lung biopsy, the current reference standard, is invasive and restricted to the sampled region, and bronchoalveolar lavage, the less-invasive alternative, also is topographically limited [13]. In this review, we describe the current experimental and clinical data on the fundamentals and applications of FDG PET in ALI and ARDS.

ALI and ARDS

ALI is a clinical syndrome characterized by acute onset, infiltrates on chest radiographs, and hypoxemia (PaO₂/FiO₂ ratio < 300 mm Hg) without an excessive increase in pulmonary capillary pressure [9, 14]. ARDS is the most severe form of ALI (PaO₂/FiO₂ < 200 mm Hg). Both conditions can be the consequence of pulmonary or extrapulmonary insults [15, 16]. Increased alveolar capillary membrane permeability is a central feature of the acute phase of ALI and ARDS [17]. It is associated with diffuse alveolar damage, which includes alveolar flooding, infiltration by neutrophils and macrophages, and formation of hyaline membranes [12, 18]. The inflammatory process can evolve to a subacute phase characterized by fibroproliferation [10, 19]. Depending on the balance between fibroproliferation and alveolar repair, some patients recover without sequelae, but others experience pulmonary fibrosis and chronic respiratory failure [15, 20].

Mechanisms of Lung FDG Uptake

FDG Uptake by Cells

FDG is a glucose analogue used as a marker of glucose metabolism. It is transported into cells via the same GLUT family of transporter proteins as glucose (primarily GLUT-1). There, in the presence of hexokinase, it is phosphorylated to ¹⁸ F-FDG-6-phosphate, which unlike glucose-6-phosphate cannot be further metabolized via the Krebs cycle [21]. In tissues with low dephosphorylase activity (such as lung, brain, and heart) [22–26], dephosphorylation has been found to be negligible for at least 1 hour after FDG injection. FDG therefore is considered trapped intracellularly [26].

FDG is a biochemically active antimetabolite of glucose at concentrations several orders of magnitude greater than those encountered during in vivo clinical studies. This property seems negligible in clinical practice [27]. Indeed, cerebral glucose utilization measured with 6-¹⁴C-glucose strongly correlates with ¹⁴C-deoxyglucose uptake (r² = 0.97) in various physiologic situations [28]. Understanding of the factors determining FDG uptake is crucial for adequate signal interpretation and introduction of FDG PET to clinical use.

Neutrophils

Neutrophils become highly activated in response to inflammatory stimuli and are key modulators of the magnitude of injury during ALI and ARDS [29, 30]. They rely on anaerobic glycolysis for energy production and consume 20–30 times more glucose when activated than at rest [31]. The current concept, derived from experimental studies, is that the FDG imaging signal during ALI is predominantly determined by the combination of the absolute number of neutrophils in the field of view and their state of activation [32–35].

In rabbit models of pneumococcal pneumonia and bleomycin pneumonitis, autoradiography of histologic sections of lung after IV administration of ³H-deoxyglucose showed that deoxyglucose uptake was localized to neutrophils [34]. Regional instillation of complement component C5a and amorphous silica into rabbit lungs confirmed the specific localization of the signal to neutrophils [36]. The investigators also documented that neutrophil numbers in the lung were proportional to the FDG signal intensity after C5a instillation and pneumococcal pneumonia [35]. Lung FDG uptake has been associated with neutrophilic infiltration in a rat model of pancreatitis-associated lung injury [33]. During ventilator-induced lung injury in sheep, lung neutrophil counts were correlated with regional FDG uptake [35]. The same group found that this correlation persisted at the regional level in a sheep model of mild endotoxemia and heterogeneously distributed lung inflammation [37].

Neutrophil activation is characterized metabolically by an increase in glucose utilization per cell and, thus, increased FDG uptake. There are limited data on the biochemical determinants of FDG uptake by neutrophils. In an in vitro study [38], change in neutrophil shape, an early event after neutrophil activation [30], correlated best with metabolic activity. Surprisingly, neutrophil ³H-deoxyglucose uptake correlated with neither respiratory burst nor degranulation, two major processes in the neutrophil response to a bacterial insult [38].

Chen and Schuster [32] highlighted the relevance of the concept of neutrophil activation in determining FDG uptake. They found greater ³H-deoxyglucose uptake after IV administration of endotoxin than after administration of oleic acid for an equivalent number of bronchoalveolar lavage–harvested neutrophils in dogs [32]. Increased activation also was implied in a rat model of pancreatitisassociated lung injury in which changes in lung FDG uptake were larger than measurements of neutrophil accumulation or sequestration, such as myeloperoxidase activity and chemotactic peptide uptake [33].

Differences in activation also have been suggested in patients with bronchiectasis and pneumonia [39]. During pneumonia (more than 3 days after the onset of symptoms) high FDG uptake was associated with negligible lung influx of radiolabeled neutrophils, indicating high postmigratory activity of the accumulated inflammatory cells. In contrast, during bronchiectasis, FDG uptake was lower for substantial neutrophil influx, implying low metabolic activity of those cells [39].

Other Inflammatory Cells

Neutrophil depletion experiments yielded evidence of the contribution of other cells to the FDG signal [35, 40]. Indeed, ALI and ARDS occur in neutropenic patients [41] and involve several types of inflammatory cells [42].

Monocytes and macrophages

Monocytes and macrophages are the first cells activated in various models of ALI [43, 44]. Moreover, in endotoxin models of ALI, the increase in macrophage counts is similar to [45, 46] or higher than [47] that of neutrophils. In other models of ALI, macrophage depletion has been found to protect the lung from dysfunction and injury [48–54]. Although macrophages take up FDG in atherosclerotic plaques [55–57] and in tumors [16], a ³H-deoxyglucose microautoradiographic study did not show any alveolar macrophage signal in acutely inflamed lung tissue [34]. This smaller contribution of macrophages to the FDG signal in acute lung states may be due to the inability of these cells to increase glucose uptake to the same extent as neutrophils do [58], despite the documented threefold increase in ³H-deoxyglucose uptake by peritoneal macrophages after infusion of lipopolysaccharide into mice [59].

Lymphocytes

Lymphocyte FDG uptake increases in vitro after activation [60]. However, FDG uptake during lung graft rejection, a predominantly T lymphocyte–mediated process, has been found similar to that in lung transplant recipients without rejection [61], suggesting a discrete contribution of lymphocytes to the FDG signal.

Lung parenchymal cells

The exact contribution of endothelial and epithelial cells to lung parenchymal FDG uptake during ALI and ARDS is unknown. That parenchymal cells contribute to the FDG signal is suggested by experimental data such as those obtained in a mouse model of ALI caused by endotoxin, in which lung FDG uptake decreased approximately one third with neutrophil depletion [40], implying that two thirds of uptake was due to other cells. In a sheep model of ventilator-induced lung injury, neutrophil depletion was associated with 70% reduction in FDG uptake [35]. Results of such studies not only support the concept that other cells contribute to the pulmonary FDG signal [35] but also suggest that differences in lung parenchymal FDG uptake may occur and be associated with species differences and disease states [35, 40].

Endothelial cells are highly metabolically active [62, 63] and take up FDG after nitric oxide stimulation [64]. Regarding epithelial cells, the sodium–glucose cotransporter is a known key mechanism of lung metabolism and homeostasis at the apical membrane of alveolar epithelial cells, enhanced activity occurring when alveolar glucose concentration increases [65]. During ALI, plasma glucose enters the alveoli, a mechanism that suggests these cells also can contribute to FDG uptake.

Methods of Quantifying FDG Uptake in the Lung

The simplest methods of quantifying pulmonary FDG are based on static indices such as the standardized uptake value (SUV) [66]. The interest in more accurate quantification of FDG uptake by lung parenchyma, not merely FDG activity in lung fields, led to the use of modeling methods. Models used to quantify FDG kinetics have been developed and validated mostly to study solid organs such as brain [2, 3], heart [1], and liver [67]. It has been recognized more recently, however, that specific models for studying pulmonary FDG kinetics, particularly in conditions of ALI, are required [68]. These models are needed because compared with solid organs, the lung has greater air content, lower basal glucose consumption [69], a larger perfusion to tissue ratio, and a larger parenchymal edema and flooding to tissue ratio during organ injury [70].

Static Indices

Simplified quantitative measures that reflect raw FDG uptake have been developed in oncology because they are compatible with static whole-body FDG PET studies. Calculation of SUV is the most widely used method. The SUV represents FDG uptake within a region of interest (ROI) measured over a certain interval after FDG administration (typically > 45 minutes [71]) and normalized to the dose injected (FDG_dose) and to a body distribution factor such as body weight (BW). SUV is computed as follows:

$$SUV={AC}_{ROI}/({FDG}_{\mathit{\text{dose}}}/BW)$$

where AC_ROI is the average activity concentration in the specified ROI. SUV quantification is affected by many factors reviewed elsewhere [71]. Among those, the main concern is that SUV encompasses blood and tissue activity within the studied ROI. As a result, intersubject and intrasubject variability in tissue blood volume contributes to the uncertainty of the SUV estimate. That SUV correlated poorly with dynamic indices in models of oleic acid– and endotoxin-induced lung injury in dogs [72] suggested that it was not suitable for quantifying FDG uptake during ALI.

In contrast, the tissue-to-plasma activity ratio, another static index, has been linearly correlated with the in vitro ³H-deoxyglucose uptake in that ALI model in dogs [72]. This index also was strongly correlated with dynamic indices [32], suggesting its usefulness in the quantification of FDG uptake. The tissue-to-plasma activity ratio is calculated by dividing the tissue activity data during the late imaging frames by the radioactivity in the plasma determined from the mean of two blood samples obtained at the end of that frame period [72]. However, the tissue-to-plasma activity ratio was not useful for discriminating patients with cystic fibrosis from healthy subjects [73], raising doubts on its clinical applicability.

Dual-time point scanning [74] is another static method useful for the diagnosis of malignancy [75] and assessment of idiopathic interstitial pneumonia [76], but to our knowledge it has not been studied in ALI.

Dynamic Indices

Dynamic indices are derived from the kinetics of FDG, typically over a period of 60–75 minutes starting from the IV injection of the radiotracer [77–80]. Availability of these kinetic data together with plasma samples allows the fitting of compartmental models. These models are used to obtain estimates of variables related to FDG transport and uptake in tissue not only at the global organ level but also in specific ROIs. As in other applications [1–3], blood samples are drawn during imaging to determine the input function to the lung in the form of a plasma time–activity curve [81]. The following three compartmental models with an increasing number of compartments have been used to study pulmonary FDG kinetics during ALI.

Two-compartment (Patlak) model

Patlak et al. [2] proposed a model initially conceived to estimate cerebral glucose utilization. The model is composed of a central compartment in rapid equilibrium with blood plasma and a peripheral compartment in which the radiotracer is irreversibly trapped (Fig. 1). Patlak graphic analysis is based on an xy-plot on which the y-axis represents FDG activity in an ROI normalized to plasma activity and the x-axis is the integral of plasma activity in time normalized to plasma activity [2]. The net uptake rate (Patlak K_i[K_iP]) of the tissue under study is the slope of the straight portion of the curve, and the initial FDG distribution volume as a fraction of tissue volume (V_e) is the y-intercept of the linear regression (Fig. 2).

Fig. 1.

Tracer kinetic models of ¹⁸F-FDG: generalized two-compartment (Patlak), three-compartment (Sokoloff) model, and lung-specific four-compartment model, which includes two equilibrating compartments. Arrows indicate directions of mass transport. C_p(t) = activity concentration of FDG in plasma; C_e(t) and C_ei(t) = FDG concentrations in extravascular compartment serving as substrate pool for hexokinase in Sokoloff (C_e) and in four-compartment (C_ei) models, C_ee(t) = FDG concentration in extravascular or noncellular compartment, C_m(t) = concentration of phosphorylated FDG. Rate constants k₁ and k₂ account for forward and backward transport of FDG between blood and tissue; k₃ is rate of FDG phosphorylation; k₅ and k₆ account for forward and backward transport of FDG between substrate (intracellular) and nonsubstrate (extracellular) compartments.

Fig. 2.

Patlak plots for lung subjected to aggressive ventilation strategy (peak inspiratory pressure, 50 cm H₂O; end-expiratory pressure, −10 cm H₂O) (solid triangles) and lung protected from aggressive mechanical ventilation (open triangles). Plots represent ¹⁸F-FDG activity in region of interest (C_ROI) normalized to plasma activity (C_p) versus integral of plasma activity normalized to plasma activity. FDG uptake rate, calculated as slope of linear portion of plot, is higher in lung subjected to aggressive mechanical ventilation. (Reprinted with permission from [35])

The main advantages of this model are simple application and interpretation. Jones et al. [39] proposed to express K_iP normalized to V_e to take into account the distribution volume of FDG in lung tissue. There is no reference standard and thus no agreement on this normalization, which significantly modifies the values of K_iP in some cases [39, 82] but not others [79, 83]. Moreover, such normalization corrects for the amount of tissue in the ROI, but it is unknown whether it takes into account the lung edema/tissue ratio. This uncertainty is a major limitation of application of the Patlak model in ALI.

Three-compartment (Sokoloff) model

The three-compartment (Sokoloff) model was originally conceived to measure cerebral glucose uptake in rats [3] and subsequently in humans [84]. It has a blood and two tissue compartments, precursor and metabolic (Fig. 1). Transfer rates characterize the transport between compartments: k₁ quantifies facilitated FDG transport from blood into the tissue (precursor compartment) per unit of lung volume; k₂ quantifies tracer transport from the precursor compartment back into the blood; and k₃ characterizes FDG phosphorylation to ¹⁸ F-FDG-6-phosphate (metabolic compartment), which is assumed to be proportional to hexokinase activity [3]. From these compartments, a global measure of the transfer rate of FDG from plasma to tissue (K_iS) and the distribution volume of the precursor compartment (F_e) as a fraction of lung volume can be computed:

$$\begin{array}{c}{K}_{iS}=k_{\mathit{1}}\times k_{\mathit{3}}/(k_{\mathit{2}}+k_{\mathit{3}})\\ F_e=k_{\mathit{1}}/(k_{\mathit{2}}+k_{\mathit{3}})\end{array}$$

Thus

$$K_{iS}=F_e\times k_{\mathit{3}}$$

The Sokoloff model includes the assumptions that after phosphorylation the radiotracer is irreversibly trapped in the tissue and that all extravascular unmetabolized FDG in the region of interest is present in a single compartment, immediately available for hexokinase-mediated phosphorylation (precursor compartment). The latter assumption may be inaccurate for acutely injured lungs, in which large pools of edematous tissue can be functionally separated from cells trapping FDG. Therefore, the following compartment model specifically designed to reflect lung FDG kinetics during ALI has been developed [68].

Four-compartment model

The lung-specific four-compartment model was designed for the assessment of FDG kinetics in acutely inflamed lungs. It is aimed at situations in which pulmonary edema can increase the distribution volume of FDG. This model includes an extravascular and noncellular compartment in addition to blood and parenchyma and allows discrimination between the distribution volume of FDG that is a precursor for phosphorylation (F_ei, intracellular) and the volume that is not (F_ee, extracellular and noncellular). The rate constants k₁, k₂, and k₃ are the same as in the Sokoloff model. The rate constants k₅ and k₆ describe the changes in ROI activity concentration due to forward and backward FDG transfer between the precursor and the extravascular and noncellular compartments (Fig. 1). Thus in this model:

$$K_{iF}=F_{ei}\times k_{\mathit{3}}$$

Compared with the Sokoloff model, in which the Akaike information criterion (an index for quantifying quality of fit of a mathematic model, accounting for the number of parameters [85]) is used, the four-compartment model has yielded best fits in most injured lungs during ALI experiments [68]. The pulmonary FDG net uptake rate and distribution volume in the precursor pool for phosphorylation are correlated between the Sokoloff and the four-compartment models [68]. However, those variables were overestimated in the Sokoloff model (de Prost N et al., presented at the 2009 meeting of the American Society of Anesthesiologists). For all these reasons, the four-compartment lung-specific model seems best suited to describing FDG kinetics during ALI.

Input function for compartmental modeling of FDG in the lung

Assessment of lung FDG kinetics implies knowledge of the time–activity curve in pulmonary artery plasma as an input function. Acquisition of this input function typically involves sequential blood sampling, a process that is invasive and prone to measurement artifacts, exposes the clinical staff to radiation and blood, and adds costly laboratory procedures [86]. The lung input function can be estimated from a two-parameter model of the blood-pool ROI that accounts for partial-volume and spillover artifacts [81]. The method requires only two blood samples for calibration of the raw PET-derived blood activity data and yields estimates of FDG kinetics parameters highly correlated with blood samples [37, 68].

Examples of FDG kinetics

Regional FDG activity is influenced by the number of metabolically active cells, the activity of these cells, and the diffusion of radiotracer into edematous tissue [70]. Use of the described models allows identification of some of these factors. For instance, de Prost et al. (de Prost et al., 2009 ASA meeting) found in a sheep model of alveolar lavage that an increase in lung water can generate an additional volume of distribution to FDG. This volume is not a precursor for phosphorylation and can artifactually increase lung FDG uptake independently of lung inflammation. Figure 3 illustrates this concept and shows the reduction of K_i and of the estimates of the radiotracer distribution volume with the four-compartment model compared with the Sokoloff and Patlak models in a sheep model known to produce discrete inflammation and large alveolar flooding (saline lavage). In contrast, in states with larger contributions of inflammation, the increase in cell metabolism is the main determinant of K_i, and the values of K_i produced by the four-compartment model are therefore close to those produced with the Sokoloff and Patlak models (Fig. 4).

Fig. 3.

Modeling of kinetics of ¹⁸F-FDG in dependent region of sheep lung subjected to surfactant depletion with saline lavage in three kinetics models. C_p = activity concentration of FDG in plasma; C_e and C_ei = region of interest (ROI) concentrations of extravascular FDG serving as substrate pool for hexokinase in Sokoloff and four-compartment models, respectively; C_ee = ROI concentration of FDG in extravascular and noncellular compartment; C_m = ROI concentration of phosphorylated FDG.

A, Patlak model. K_iP = 0.0071/min, y-intercept = 0.46 (Patlak equation, y = 0.0071x + 0.46). C_ROI = ¹⁸F-FDG activity in region of interest.

B, Sokoloff model. K_iS = 0.0051/min, k₃ = 0.0159/min, and F_e = 0.32 (K_iS = k₃ × F_e).

C, Four-compartment model. K_iF = 0.0046/min, k₃ = 0.0260/min, F_ei = 0.1786, and F_ee = 0.1769 (K_iF = k₃ × F_ei). K_iF < K_iS, and K_iP, and F_ei < F_e and Patlak y-intercept. This last result shows that use of four-compartment model leads to smaller size of compartment that is direct precursor for phosphorylation than in Patlak and Sokoloff models. This reduction in estimates of precursor volume accompanies reduction in estimates of net FDG uptake rate K_i.

Fig. 4.

Modeling of kinetics of ¹⁸F-FDG in dependent region of normal sheep lung after administration of 2-hour infusion of endotoxin (10 ng/kg/min) in three kinetics models. Ki_F, K_iS, and K_iP are closer than in Figure 3. Although F_ei estimate is lower than F_e and y-intercept of Patlak plot, effect on Ki is negligible, likely because in this situation k₃ is main determinant of K_i (k₃ in this example is three times as high as in Fig. 3). C_p = activity concentration of FDG in plasma; C_e and C_ei = region of interest (ROI) concentrations of extravascular FDG serving as substrate pool for hexokinase in Sokoloff and four-compartment models, respectively; C_ee = ROI concentration of FDG in extravascular and noncellular compartment; C_m = ROI concentration of phosphorylated FDG.

A, Patlak model. K_iP = 0.0125/min, y-intercept = 0.46 (Patlak equation, y = 0.0125x + 0.46). C_ROI = ¹⁸F-FDG activity in region of interest.

B, Sokoloff model. K_iS = 0.0136/min, k₃ = 0.0550/min, and F_e = 0.25 (K_iS = k₃ × F_e).

C, Four-compartment model. K_iF = 0.0126/min, k₃ = 0.0690/min, F_ei = 0.1834, and F_ee = 0.1374 (K_iF = k₃ × F_ei).

Use of FDG PET in Inflammatory Lung Diseases

Pulmonary Infectious Diseases

FDG PET facilitates early visualization of lung infection in various experimental and clinical situations. Using a rabbit model of Streptococcus pneumoniae pneumonia [34], Jones et al. found that lung FDG uptake was pronounced 15 hours after instillation of bacteria but had decreased to almost baseline values 48 hours after instillation [39]. In healthy subjects receiving a moderate local lung instillation of endotoxin, FDG PET can depict inflammation, suggesting its sensitivity in quantifying the early phase of an acute inflammatory response [79]. FDG PET also has been used to measure the effect of antiinflammatory interventions, suggesting its applicability in the testing of novel drugs [83]. The clinical value of lung FDG uptake has been exemplified in cystic fibrosis patients with no sign of acute infectious exacerbation. The magnitude of uptake was associated with the speed of decline in pulmonary function [73].

Subacute and Chronic Lung Inflammatory Diseases

In a rabbit model of pulmonary fibrosis [36], the FDG signal intensity increased and stayed high for as long as 4 weeks after administration of fibrogenic agents; the signal was predominantly derived from neutrophils. A 2009 study [87] showed that in patients with pulmonary fibrosis, the highest FDG PET SUVs coregistered with high-resolution CT patterns, such as honeycombing, which is classically understood as representative of irreversible parenchymal scarring. The authors hypothesized that this finding reflected high fibroblastic activity, suggesting an opportunity for pharmacologic manipulation.

Applications of FDG PET in the assessment of airway inflammation have resulted in higher FDG uptake in persons with chronic COPD than in those with asthma [82]. Although both groups had high sputum neutrophil counts, this count correlated with FDG uptake only in COPD patients, implying a difference in neutrophil status. This finding suggests that FDG PET provides a potentially important marker of persistent lung inflammation.

Experimental Acute Lung Injury

Smoke inhalation

Schroeder et al. [80] found a significant and early (2 hours) increase in regional FDG uptake in sheep lungs acutely exposed to smoke inhalation. Regional FDG uptake (Fig. 5) appeared to be predictive of regional gas exchange impairment in this model, suggesting a function–inflammation relation. The findings also suggested a direct association between the degree of ventilation– perfusion heterogeneity before injury and the postinjury magnitude of FDG uptake.

Fig. 5.

Sheep exposed to unilateral smoke inhalation. Parametric PET images of ventilation (10⁻³ mL/s) (left), Patlak net uptake rate (K_iP, min⁻¹) (center), and K_iP corrected for tissue fraction (K_ip/F_tissue, min⁻¹) (right) in two sheep (one in each row) exposed to unilateral smoke inhalation. Smoke-exposed left lung (right side of image) exhibits higher K_ip than right (nonexposed) lung, even after correction for lung density. Areas of high ¹⁸F-FDG uptake may correspond not only to alveolar units (bottom row) but also central airways (top row), as attested by concomitant decrease in ventilation of injured lung. (Reprinted with permission from [80])

Ventilator-induced lung injury

Musch et al. [35] found that increased regional FDG uptake exclusively due to mechanical forces can be detected in a sheep model of ventilator-induced lung injury after only 90 minutes of mechanical ventilation. When positive end-expiratory pressure was associated with overdistention, increased FDG uptake was observed and associated with mild neutrophil infiltration despite the absence of impairment of gas exchange and respiratory system compliance, suggesting that the FDG signal preceded regional dysfunction (Fig. 6). The combination of overdistention with lung collapse led to a marked increase in FDG uptake, marked regional gas exchange and mechanical dysfunction, and pronounced neutrophilic infiltration.

Fig. 6.

Three sheep exposed to unilateral ventilator-induced lung injury. Sheep were prone during imaging. Right side of each PET image corresponds to left lung. Parametric images of Patlak net uptake rate (K_ip) (three left images) and lung tissue sections stained with toluidine blue (right). In all cases, left lung was subjected to aggressive ventilation strategy involving both alveolar overexpansion and collapse (peak inspiratory pressure, 50 cm H₂O; end-expiratory pressure, −10 cm H₂O) (top) or exclusively alveolar overexpansion (peak inspiratory pressure, 50 cm H₂O; end-expiratory pressure, +10 cm H₂O) (bottom). Right lung was protected from aggressive mechanical ventilation. Arrows indicate lung corresponding to tissue sections displayed on same row. Lung subjected to alveolar overexpansion and collapse (top) shows higher ¹⁸F-FDG uptake than does control lung in addition to heterogeneous distribution of aeration with poorly and well-aerated areas, intraalveolar proteinaceous material, and marked cellular infiltration in histologic section. At fourfold higher magnification (inset), intraalveolar cells seem to be mainly neutrophils. Lung subjected exclusively to overexpansion (bottom) has uniform aeration and some neutrophils. (Reprinted with permission from [35])

Endotoxemia

Using the Sokoloff model, Costa et al. [37] found that FDG PET can be used to detect early regional cellular activation during mild endotoxemia associated with a moderately aggressive mode of ventilation (Fig. 7). Moreover, heterogeneous lung aeration due to mechanical ventilation of the supine lungs with large tidal expansion produced spatially heterogeneous FDG uptake associated with heterogeneous pulmonary neutrophilic infiltration.

Fig. 7.

Sheep exposed and sheep not exposed to two-hit model of acute lung injury. A and B, Parametric images of Patlak net uptake rate (K_iP) in two supine sheep exposed (A) or not exposed (B) to two-hit model of acute lung injury with mild doses of IV endotoxin (10 ng/kg/min for 2 hours) and moderately aggressive mechanical ventilation. K_iP is higher in sheep exposed to two-hit injury (color scale is same for A and B). In addition, distribution of K_iP is heterogeneous, being larger in dependent lung regions.

Clinical Acute Lung Injury

Traumatic lung injury

Addressing the predictive value of FDG PET, Rodrigues et al. [88] conducted a study with eight patients with pulmonary contusion, none initially meeting the criteria for ARDS. At FDG PET 24–72 hours after admission, three of the four patients in whom ARDS developed had diffuse FDG uptake throughout the lungs (Fig. 8A), but all four of the patients in whom ARDS did not develop had significant FDG uptake only in areas of focal lung opacity on CT images (Fig. 8B). The SUV of the normally aerated lung was higher in patients with ARDS than in those in whom ARDS did not develop, suggesting that FDG PET provides valuable predictive information during the early stages of ALI.

Fig. 8.

Patients admitted with thoracic trauma and lung contusion who did not meet criteria for acute respiratory distress syndrome (ARDS) at image acquisition. Coronal ¹⁸F-FDG PET (left), transaxial CT image (center), and corresponding slice through same region of transaxial PET scan (right). (Courtesy of Morton K, University of Utah School of Medicine, Salt Lake City, UT)

A, Patient in whom ARDS developed within days after acquisition of these images. FDG PET images show high and diffuse FDG uptake in all portions of lungs, both focally nonaerated and poorly aerated and normally aerated on CT scan.

B, Patient in whom ARDS did not develop. FDG PET images show moderate FDG uptake in nonaerated and poorly aerated regions and low FDG uptake in normally aerated regions.

ARDS

Bellani et al. [77] studied the distribution of FDG uptake, as measured with K_iP, in the lungs of patients with established ARDS. Those investigators found that the relation between FDG uptake and lung density followed two different patterns. In seven patients, K_iP was highest in regions with the highest density, and in three patients, K_iP was higher in normally or poorly aerated regions than in nonaerated regions. This finding suggests that aerated regions (so-called baby lung) can be considerably infiltrated with inflammatory cells and emphasizes the great potential of FDG PET for noninvasive exploration of the relations between structure, function, and inflammation in ALI and ARDS.

Conclusion

In the acutely inflamed lung, FDG PET facilitates noninvasive assessment of regional lung neutrophil infiltration and activation. This ability has led to considerable improvement in the experimental and clinical pathophysiologic investigation of ALI and ARDS. To our knowledge, lung-specific compartmental models that account for the presence of an extracellular compartment have never been used in humans. Results of animal studies suggest use of these models may improve quantification of the FDG signal in ALI and ARDS. Further research is needed to explore the biologic correlates of FDG uptake. Better understanding of such mechanisms is required before FDG PET can be used as a decision-making tool in the treatment of patients with ALI and ARDS.