Healthy lung tissue predominantly relies on glucose utilization to sustain function, although its energy and metabolic needs are relatively modest compared to other organs such as the heart and liver (Fisher et al., 1974). Approximately 50% of the glucose utilized by the lung tissue converts to lactate (Fisher, 1984). The lungs typically maintain the glycolytic intermediary balance (lactate-to-pyruvate ratio) in the blood by utilizing the excess blood lactate, which results in a negligible difference in transpulmonary lactate concentration (Johnson, 2011). However, this function of the lung is compromised in many lung pathologies – especially in ARDS/ALI (Iscra et al., 2002). Figure 4 demonstrates lung lactate production measured by the difference in the lactate concentration across the lungs (arteriovenous difference in lactate) in 122 patients with a variety of lung disorders (De Backer et al., 1997). Lungs of patients with ALI (N = 43) produce significantly more lactate than other pathologies. What is more is that lactate production in ALI patients was strongly correlated with lung injury score as shown in Figure 4B. The injury score is representative of the severity of opacities observed in chest radiographs and computed tomography images, as well as the severity of overall respiratory failure measured by P_aO₂/F_iO₂ ratio and loss of pulmonary compliance (Murray et al., 1988).

Although increased glycolysis and lactate production by the lung tissue may reflect the presence of hypoxia due to elevated anaerobic metabolism, several studies suggest that the elevated lactate production by the lungs in ARDS/ALI is primarily associated with increased lung inflammation and neutrophil activity, and can occur even in the absence of tissue hypoxia. This suggests that increased lactate concentration and lactate-to-pyruvate ratio in the lung tissue may be a surrogate for lung injury and inflammation (De Backer et al., 1997; Kellum et al., 1997; Iscra et al., 2002).

The radiolabeled analog is first synthesized at a cyclotron facility by bombarding a radioligands with accelerated protons to produce unstable radioactive isotopes e.g., ¹⁸F and ¹¹C that are then used in a biosynthesizer unit to produce radio tracers. The tracer is then injected intravenously into the patient. The nucleolus of the radiolabeled atom then undergoes β radioactive decay, in which a positron is released and travels for a short distance in the tissue (< 1 mm) before colliding with an electron to produce two γ-ray photons that travel in opposite directions (Pauwels et al., 2000), which are detected using a ring-shaped array of sensor around the patient. By resolving the time difference between the arrival of photons at sensors in the opposite direction, using the time-of-flight algorithm, 2D or 3D images can be tomographically generated (Gambhir et al., 2001). Data acquisition is often performed over a period of 30–60 min, providing temporal information as well. The raw data obtained from the scanner can be converted to markers of metabolic activity using timed-blood sampling to measure overall radioactivity combined with kinetic approaches that fit temporal changes of data to multi-compartmental models to decompose relative contributions of signal intensity from solid organs, blood and extracellular matrix (Chen et al., 2017).

PET scanners do not obtain anatomical information and thus are often combined with CT scanners (PET/CT scanners) to overlay the functional information on the anatomical images.

The most commonly used tracer for PET imaging is [¹⁸F]-fluorodeoxyglucose (¹⁸F-FDG). ¹⁸F-FDG is glucose analog and is similarly transported into the cell by glucose transporter 1 (GLUT-1) and subsequently phosphorylated. ¹⁸F-FDG cannot progress through the Krebs cycle and thus remains trapped in cells. Therefore, it can specifically be used to assess glucose uptake as a surrogate for overall glycolytic activity.

While ¹⁸F-FDG-PET is routinely used in neuro-radiology (Gambhir et al., 2001) and oncology (Pastorino, 2010; Heusch et al., 2014; Miles et al., 2018), it is not clinically used for the management of ARDS/ALI. However, several studies have demonstrated its capability as a powerful molecular imaging tool to delineate regions in the injured lungs with elevated glycolysis. As activated neutrophils are largely responsible for the uptake of glucose, elevated glycolysis can be used as a surrogate for inflammation. This has been validated in other studies that used autoradiography of the lung tissue in animals after administration of ¹⁸F-FDG showing that radioactivity was localized to neutrophils in the lung tissue.

Neutrophil recruitment and activation are heightened in ARDS, leading to elevated glycolysis, which can be regionally measured using ¹⁸F-FDG-PET. The ability of this molecular imaging technique to regionally highlight alterations in metabolic activity has made ¹⁸F-FDG-PET an invaluable research tool to non-invasively and quantitatively assess the severity lung inflammation. ¹⁸F-FDG-PET has been employed in both animal models of lung injury (de Prost et al., 2014) and in preclinical studies (Chen et al., 2006b; Bellani et al., 2012), and has provided valuable insight about the progression of lung injury (de Prost et al., 2014) as well as the importance of inflammation in patient outcome (Bellani et al., 2009, 2012).

In a preclinical study by de Prost et al. the authors assessed the impact of ventilation strategy on distribution and progression of lung inflammation using ¹⁸F FDG-PET (de Prost et al., 2013) in an endotoxemia model of lung injury in sheep. They found that primary inflammatory injury progressed rapidly in sheep ventilated with a non-protective strategy, while the inflammation was contained in animals ventilated with protective ventilation. Furthermore, the study demonstrated that metabolic activity was significantly higher in the dependent regions of the lungs.

In patients, ¹⁸F FDG-PET has been shown to be capable of localizing areas with higher neutrophilic activity the lung tissue as well. Chen et al. (2006b) showed that glucose uptake was significantly elevated in lungs of human subjects 24 h after instillation of 1–4 ng/kg endotoxin in airways using bronchoscopy. In a different study, Rodrigues et al. demonstrated the use of ¹⁸F-FDG-PET at early stages of ARDS (1–3 days after admission) to predict outcome (Figure 5); patients with significantly higher glycolysis in the lung tissue hard poorer outcome that others (Rodrigues et al., 2008). Bellani et al. (2009, 2011) showed the use of ¹⁸F FDG-PET to investigate the regional distribution of inflammatory activity in lungs of mechanically ventilated ARDS patients. Their work suggests that active inflammation is localized to non-dependent regions of the lungs at early stages. However, the inflammation spreads throughout the entire lung after days of ventilation. This data is supported by older work using localized biopsy that demonstrated similar destitution of inflammation (Papazian et al., 1998).

¹⁸F-FDG-PET have been used to localize inflammatory activity in other lung pathologies as well. Chen et al. (2006a) showed elevated FDG uptake in lungs of patients suffering from cystic fibrosis (CF) compared to healthy subjects (Chen et al., 2006a). The uptake rate correlated strongly with the number of neutrophils present in the bronchoalveolar lavage. What was more interesting was that the glucose uptake rate was especially higher in patients with rapidly declining pulmonary function. In a different study, the authors assessed FDG uptake in lungs of patients with chronic obstructive pulmonary disease (COPD) patients with and without chronic bronchitis (Scherer and Chen, 2016). They showed increased average uptake of FDG in lungs with patients with chronic bronchitis. The CT scans of the same patients also showed more heterogeneously distributed emphysema.

Although these studies suggest a strong link between neutrophilic inflammation and increased FDG uptake, other inflammatory cells such as macrophages and eosinophils are also capable of accumulating FDG (Scherer and Chen, 2016). Nevertheless, FDG uptake can represent a measure of overall inflammatory response in the lungs. Given that inflammation and inflammatory burden is associated with decline in lung function, disease severity and lung tissue destruction in many lung pathologies, ¹⁸F FDG-PET be used to classify patients with various disease severities to predict and assess response to treatments in these patients (Scherer and Chen, 2016).

There are a number of other less commonly used PET tracers that can provide more specific information about the inflammatory process. For instance, ⁶⁸Ga-citrate has been shown to bind to the lactoferrin within the neutrophil, therefore localizing specifically neutrophilic inflammation (Scherer and Chen, 2016). 18F-nitric oxide synthase (18F-NOS) can assess expression of nitric oxide synthase in lung epithelium, which has been shown to be elevated in patients with progressive asthma, COPD, ARDS, and emphysema (Huang et al., 2015). Lastly, Summers et al. (2014) administered a bolus of ¹¹¹Indium-tropolonate-labeled neutrophils to healthy subjects and ARDS patients and showed increased retention of neutrophils in and their delayed clearance in the ARDS patients.

The first principal limitations for clinical use of PET imaging is the cost of preparing the radiolabeled compound, which is done at a cyclotron facility followed by an on-site chemical synthesis apparatus to produce the final compound. Such facilities are expensive to maintain and thus are available only at a few universities and hospitals. Therefore, radio tracers that have a long-half life, such as ¹⁸F-FDG (109.8 min) are often produced remotely at a cyclotron facility and transported to near-by locations. Since samples are radioactive, they need to be delivered via specially licensed road transport, or, for longer distances, via dedicated small commercial jet services, thereby making the scans costly. Another limitation is the long scan time (10–50 min) (Cereda et al., 2019) that is can be difficult for critically ill patients or patients with lung injury. Additionally, PET tracers expose patients to ionizing radiation, which limits use of this technique to monitor patient’s response to interventions through repeated measurements.

Another potential disadvantage of ¹⁸F-FDG-PET is that while it enables examining abnormalities in the uptake of the glucose analog fluorodeoxyglucose, it is unable to reveal changes in downstream metabolism as it cannot progress through the Kerbs cycle. Such information may be crucial to the evolution of inflammation and injury (Fisher and Dodia, 1984; De Backer et al., 1997). Lastly, given the long half-life of ¹⁸F-FDG-PET, the radioactivity of the probe from a single injection can last for hours, which limits the possibility of repeated scans as frequently as needed. This is especially important in small animal research, where lung injury progression occurs on a time scale that is significantly shorter than in human patients.

¹³C MRSI enables study of metabolic flux in the tissue due to its unique ability to distinguish metabolites through their distinct resonance frequencies (chemical shifts). Due to low natural abundance of ¹³C nuclei, the MRI scan is performed after administration of an exogenous non-radioactive ¹³C-labeled compound. Subsequently, spectroscopic imaging methods can be used to highlight changes in cellularity and metabolic pathways. Although this method has been shown to be insightful for tumor and neuroimaging it is limited as it requires a long scan time due to low intrinsic nuclear spin of the ¹³C nuclei (Kurhanewicz et al., 2011, 2019). To overcome these challenges the signal can be increase through nuclear hyperpolarization.

Hyperpolarization is a process to temporarily enhance the sensitivity of the MRI signal by over 10,000-fold over conventional MRI (Kurhanewicz et al., 2011). Hyperpolarization of the ¹³C nuclei is typically achieved through a process called dynamic nuclear polarization (DNP), which transfers spin alignment from the sparse unpaired electrons of an electron paramagnetic agent (EPA) to the adjacent ¹³C labeled nuclei using resonant microwave irradiation at high magnetic fields (∼3.3T) at ∼1 K temperature (Ardenkjaer-Larsen et al., 2003). Once the sample reaches the desired polarization level (usually within 1–3 h), it is dissolved rapidly using a hot isotonic buffer to yield a highly polarized neutralized solution, which is then administered intravenously to the subject.

Upon injection, spectroscopic imaging must be carried out quickly and efficiently for two reasons; first, given the short life-time (T₁ relaxation time constant) of the probes (10–120 s depending on the probe), the data must be acquired quickly. Second, in conventional MRI 2D or 3D images are acquired, whereas HP ¹³C MRSI requires data acquisition in the spectral dimension as well, which adds further complexity to the criteria for pulse sequence development. What is more is that any RF excitation causes additional irreversible signal loss. Several imaging and spectroscopic pulse sequences have been developed to address these challenges by limiting the number of excitations and exploiting the long T₂, and T₂^∗ relaxation times of ¹³C species in many organs (Yen et al., 2009), as well as ¹³C species’ large range of the chemical shift (Kurhanewicz et al., 2011).

The most widely used hyperpolarized DNP probe is [1-¹³C] pyruvate, a small and highly soluble molecule with high polarizability (up to 60% polarization reported) and a long T₁ relaxation time constant (40–60s). Because pyruvate is at a central branching point in several key metabolic pathways in cancer and inflammatory diseases, it be used to study a wide variety of metabolic perturbations in tissues and presents unique opportunities to characterize metabolic flux in various metabolic, Therefore [1-¹³C] pyruvate is perhaps the most attractive HP ¹³C imaging probe to date (Figure 6).

HP [1-¹³C] pyruvate has been used extensively to study metabolic alterations in heart (Lau et al., 2010), liver (Lee et al., 2013), kidneys (Laustsen et al., 2015), and tumors (Day et al., 2007) in intact animals and has more recently been used in a number of human studies (Nelson et al., 2013; Cunningham et al., 2016; Park et al., 2018). Nevertheless, given that the lungs receive the full blood supply during each circulation, and therefore, play a major role in maintaining body’s homeostasis and whole body’s metabolic status, HP ¹³C MRSI can be used as valuable tool for evaluating lung metabolism and pathology. In the context of lung injury, HP [1-¹³C] pyruvate MRSI can be used to regionally measure lactate-to-pyruvate ratio as a surrogate for lactate labeling, which is significantly elevated in inflamed lungs (Figure 7; Pourfathi et al., 2018; Pourfathi, 2019).

In an ex-vivo perfused lung study with an experimental model of bleomycin-induced lung inflammation in rats, Shaghaghi et al. (2014) showed a significantly increased rate of hyperpolarized lactate labeling after injection of hyperpolarized [1-¹³C] pyruvate in inflamed lungs. The lactate-to-pyruvate ratio declined but remained higher than healthy lungs even after fibrotic remodeling of the tissue at later stages of the injury. The study also showed a strong correlation between the lactate-to-pyruvate ratio and neutrophil scores derived from histological assessment of the lung samples. Interestingly there was no correlation between the macrophage count and the lactate signal suggesting that the primary source of the lactate is from neutrophils.

This technique has also been used for in-vivo imaging in small animals; Thind et al. (2012) showed elevated HP lactate-to-pyruvate ratio thus lactate labeling in irradiated lungs after radiation induced lung injury (RILI) compared to healthy animals. The authors showed a strong correlation between the lactate-to-pyruvate ratio and inflammatory markers measured from bronchoalveolar lavage. Pourfathi et al., showed that in an experimental model of aspiration pneumonia, HP lactate-to-pyruvate ratio was significantly elevated in mildly injured rats that received (intratracheally) a low volume of hydrochloric acid (HCl). Interestingly, the authors report a decline in the lactate-to-pyruvate ratio in more severely injured rats that received a larger volume of HCl. The authors showed that the although the lactate signal appeared to increase in the posterior regions of the injured lungs, the skewed measurement was in fact caused by a dramatic increase in the blood volume in the injured lungs.

Another study with HP [1-¹³C] pyruvate MRSI by Pourfathi et al. showed that this technique can be used to mechanisms of injury progression by secondary ventilator-induced lung injury; the authors assessed the impact of recruiting atelectasis on the trajectory of lung injury and inflammation in ventilated rats with primary aspiration pneumonitis, and reported that positive end-expiratory pressure (PEEP) and recruitment contains regional pulmonary lactate production and inflammation. The study supported a direct relationship between pulmonary inflammation and increased HP lactate-to-pyruvate ratio, consistent with the link between increased glycolysis caused by recruitment and activation of neutrophils as part of an innate inflammatory cascade, that was proposed by previous FDG-PET studies (Jones et al., 1994), and showed a strong correlation between the lactate-to-pyruvate and inflammatory markers of neutrophilic activity (MPO) and adherence (ICAM-1).

Finally, Siddiqui et al. showed that HP [1-¹³C] pyruvate MRI has the potential to be used as predictor for lung rejection. In a direct comparison between this technique and microCT in a lung allograft rejection rat model, the authors observed elevated lactate-to-pyruvate ratio prior to observing features in the microCT that are indicative of rejection. The authors also showed a strong relationship between the presence of markers of adaptive immunity CD4+ and CD8+, and the elevated lactate-to-pyruvate ratio in the transplanted lung (Siddiqui et al., 2019).

The preliminary results of these studies demonstrates the potential of HP [1-¹³C] pyruvate MRSI to detect elevated pulmonary lactate-to-pyruvate ratio. This imaging marker can serve as a surrogate to regionally assess increased glycolysis and subsequent lactate production by injured lungs as a result of inflammation (Pourfathi et al., 2017). While lactate mapping can be a marker of neutrophilic infiltration, it may be used to assess other biological changes in the lung related to lung injury that promote lactate production as well, such as fibrosis (Kottmann et al., 2012). The ability to interrogate alterations in critical downstream metabolic pathways such as glycolysis, may enable non-invasive and. frequent assessment of patients’ response to therapies early after treatment. This opens up opportunities to assess the response of various therapeutic apaches to select optimal strategy to attenuate lung injury and its consequences in the early stages (Shankar-Hari and McAuley, 2017).

The most critical limitation of HP ¹³C MRSI is the very short lifetime of the hyperpolarization that limits the available “window-of-opportunity” to acquire data. Another limitation is the need for a clinical hyperpolarizer that is currently available at around 30 sites across the world. Unlike PET tracers, HP ¹³C agents cannot be produced remotely and delivered to the site-of-interest given that the lifetime of the hyperpolarization is significantly shorter than that of the half-life of PET tracers. Therefore, clinical dissemination of this HP ¹³C MRSI technology requires a polarizer at every site.

Other technical limitations of this technology for lung imaging in clinical studies are the field inhomogeneity in the lung tissue causing rapid spin dephasing at air-tissue interfaces and lung’s overall low tissue density. This difficulty is further exacerbated in the case of metabolic imaging by lung’s modest overall energy needs. These challenges limit both signal-to-noise ratio (SNR) and the suitability of rapid pulse sequences that are useful for imaging other organs. While many studies suggest that these challenges can be addressed (Pourfathi et al., 2017, 2018) the clinical utility of ¹³C HP MRSI in critically ill patients raises safety concerns primarily because of the need to use MRI–compatible monitoring and ventilation equipment.

Lastly, quantification of the absolute concentration of metabolites using HP ¹³C MRSI is non-trivial, as the absolute signal level is subject to variability due to polarization level and physiological conditions (Pourfathi et al., 2017). Although the use of hyperpolarized lactate-to-pyruvate ratio as a surrogate for endogenous lactate concentration (De Backer et al., 1997) and glycolysis (Siddiqui et al., 2016) can mitigate this variability, it can still be subject to bias caused by excess fluid in the extracellular space resulting from capillary bed leakage (Musch et al., 2007); the presence of such excess fluid can increase pyruvate concentration or limit its uptake, thereby reducing the lactate-to-pyruvate ratio and the sensitivity of the method (Pourfathi et al., 2017). PET imaging studies have addressed similar challenges by using compartment models to pinpoint the local source of signal (Chen et al., 2017). In the case of MRI, this bias may be corrected for by using rapid MRI pulse sequences that allow the acquisition of multiple images to characterize metabolic flux in various tissues (Siddiqui et al., 2016).