The most studied metabolic phenomenon in cancer is its tendency to increase its’ rate of glycolysis in adjunct to oxidative phosphorylation, despite the presence of sufficient levels of oxygen in the TME. This feature of cancer metabolism is called the Warburg effect, named after the German scientist who first described this (76). Although glycolysis is less efficient in producing ATP, it does generate increased levels of additional metabolites for the biosynthesis of ribose, glycosylation precursors, amino acids, and lipids (77, 78). Inefficient ATP production would only be problematic in scarcity of nutrients, which is in general not the case in cancer. Therefore ‘aerobic glycolysis’ means a survival advantage for cancer cells in terms of increased anabolism and avoidance of oxidizing precious carbon-carbon bonds (79).

To meet their greatly enhanced demand for glucose under conditions of ‘aerobic glycolysis’, cancer cells have upregulated levels of the key glucose transporter 1 (Glut-1) on cell membranes (10) and associated hexokinases. Hexokinases are enzymes that phosphorylate six-carbon sugars, primarily glucose, by transferring a phosphate group from ATP to its’ substrate. As phosphorylation charges the hexoses, it is trapped intracellularly and available for further metabolic processes, resulting in a stable down-slope gradient that drives glucose transport into the cell. For this reason, hexokinase activity is the rate limiting step for most metabolic pathways involving glucose. The isoform hexokinase II (or hexokinase B) is the dominant isoform in many cell types (80), including most cancers, and located at the outer mitochondrial membrane to have direct access to ATP (81). Upregulation of aerobic glycolysis results in an increase of pyruvate, which is further metabolized into lactate (77). Intracellular lactate is transported out of the cell, along with protons, via monocarboxylate transporters (e.g., MCT-1 and MCT-4) into the TME. In addition to lactate shuttles, intracellular acidification is prevented by carbonic anhydrase 9 (CAIX), a transmembrane metalloenzyme that facilitates secretion of acids produced under oxidative stress. Indeed, it has long been noticed that tumors often have an acidic environment (82).

Metabolic reprogramming in cancer is partly due to oncogenic activation of signal transduction pathways and transcription factors, HIF-1α is a master regulator of glycolysis and the pentose phosphate pathway (20, 83, 84). In lung cancer, oncogenes and pathways divert intracellular glucose flux towards increased usage of glucose into the hexosamine biosynthesis, required for protein glycosylation and pentose phosphate pathway [reviewed in (4)]. Well-studied signaling pathways, including PI3K/Akt/mTOR and RAS/RAF/MEK/MAPK, with high prevalence in lung cancer, associate with increased glycolysis as well as metabolic plasticity, by initiating compensatory mechanisms and facilitating alternative metabolic sources, e.g., amino acids, nucleotides or fatty acid biosynthesis and macropinocytosis. At a transcriptional level, the transcription factor nuclear factor erythoid-2-related factor (NFE2L2/Nrf2) is identified as one of the main regulators of metabolic reprogramming in lung cancer, and its activity is associated with poor survival (85).

Epigenetic mechanisms also contribute to the regulation of gene expression involved in cancer metabolism (86). Disruption of the epigenome is present in cancer cells, including DNA methylation, histone proteins and histone modification enzymes, as well as proteins that regulate the function of metabolic enzymes (87). Reciprocally, activity of histone and DNA modifying enzymes regulates the expression of metabolism-associated genes, leading to a complex interplay between metabolism and epigenetic during cancer progression (88). Understanding the relation between metabolism, signaling pathways and epigenetics may open new avenues for anti-cancer immune therapy (89), which will be discussed later.

Upon activation, naïve T-cells also undergo metabolic adaptation to meet the increased bioenergetic demands associated with proliferation and effector function (29, 90–93). In contrast to static cancer cells, which can thus invest in creating a favorable metabolic niche, effector T-cells migrate through the body and are merely passengers who need to adapt to changing environmental conditions, from well-supplied lymph nodes and spleen to rather oxygen and nutrient deprived cancer lesions (94). In general, nutrient competition between cells strongly influences cell fate (36, 95) and function (90). More recently, this interplay between cancer cells and immune cells has been reviewed (51, 96, 97). Aerobic glycolysis is not required for activation or proliferation during early stages of T-cell activation (98), however, it is essential for optimal T-cell effector function in the TME (99, 100). In vitro models previously demonstrated that cancer cells outcompete T-cells for glucose, directly restricting cytokine mediated anti-cancer immunity (101). Also in vivo, tumor infiltrating CD8+ T-cells face restricted glucose availability, which consequently hampers increased rate of glycolysis by restricted mTOR activity and thus reduced IFN-γ production (98, 102).

In addition to direct competition for glucose, limiting the magnitude of aerobic glycolysis in T-cells, high lactate excretion by cancer cells further suppresses T-cell effector functions (103–105), directly correlated it to reduced survival rates in e.g., head and neck cancer (106). The acidic TME inhibits both T-cell trafficking and cytotoxicity (102, 103, 107) and sheds new light on the role of lactate as immune metabolic mediator (14). The enzyme lactate dehydrogenase A (LDHA) which converts pyruvate into lactate, not only plays a central role in cancer cell aerobic glycolytic capacity but exerts similar function in T-cell function through PI3K signaling (31, 108).

The costimulatory molecule CD28 on T-cells, ligating to CD80 during antigen-specific activation, induces this PI3K signaling (30), resulting in increased expression of Glut-1. By facilitating glycolysis increase, CD28 signaling prepares T-cells to anticipate on changing metabolic demands associated with sustained effector functions. This necessary metabolic switch is furthermore under the control of inhibitory members of the CD28 superfamily (mainly PD-1 expression), with the intend to delicately control T-cell activation (97, 109–111). PD-1 on T-cells is mostly studied as an exhaustion marker, induced by chronic antigen exposure and endurable stages of activation. Its increased expression on T-cells indicates a critical stage of T-cell development, at the verge of going in retraction and clearance (109, 112). The expression of PD-L1, by cancer cells and myeloid derived suppressor cells in the TME not only suppresses cytotoxic effector function of T-cells, but it also entangles the metabolic reprogramming of T-cells via ligation of PD-1. PD-1 ligation suppresses the ability of T-cells to perform glycolysis and glutaminolysis, thus pushing T-cells further towards retraction. Therefore, one of the effects of therapeutic monoclonal antibodies targeting CTLA-4 (interacting with CD28) and PD-L1 (interacting with PD-1), is allowing T-cells to maintain their increased glycolytic and glutaminolytic capacity to execute anti-cancer effector functions in the TME (113, 114).

The most widely applied tracer to image the upregulation of glycolysis is 2’-deoxy-2’-[¹⁸F]fluoro-D-glucose ([¹⁸F]FDG). [¹⁸F]FDG is extensively used for the detection of primary tumors, metastases and recurrences, and monitoring responses to anti-cancer treatments (115–117). [¹⁸F]FDG uptake by glycolytic cancer cells is directly related to upregulated levels Glut-1 transporters (118) and hexokinases activity (119). Consequently, levels of [¹⁸F]FDG uptake also correlate with increased levels of derivates of the glycolytic pathway; pyruvate and lactate (120).

Glycolysis, even when increased, contributes relatively little to cellular ATP content; its majority is provided by oxidative phosphorylation in mitochondria, which requires oxygen in the electron transport chain (142). Although oxygen availability is only limiting mitochondrial electron transport chain at very low levels (<0.07% oxygen) (143, 144), its lowering levels are sensed carefully. Through HIF1a activation, hypoxia promotes glycolysis in addition to increasing oxidative phosphorylation (145, 146). Imbalances in oxygen levels occur in a range of physiological conditions, e.g., wound healing (147), and disease conditions, e.g. chronic obstructive pulmonary disorders (148). In cancer however, the chaotic tissue vascularization results in chronic diffusion-limited hypoxia as well as acute perfusion limited hypoxia (15). Enduring long-term hypoxia results in additional oxidative stress caused by increased levels of reactive oxygen species (ROS) produced from mitochondrial complex III in cancer cells (149, 150) Excessive levels of intracellular ROS can cause oxidative damage to intracellular lipids, protein and DNA, which might reciprocally drive diversification of cancer phenotypes (151) but at a given point culminates in cell cycle arrest and apoptosis (152).

Intracellular oxygen homeostasis is regulated by the hypoxia-inducible factor (HIF), a heterodimer that is composed of two subunits HIF-1α and HIF-1β (153). HIF-1α transcriptional activation is triggered by short-term hypoxia of 2–24 h with oxygen tensions <0.1% oxygen, while the isoform HIF-2α activation occurs under milder hypoxic conditions (<5% oxygen). Under normoxic conditions, HIF-1α is degraded under control of the von Hippel-Lindau (VHL) protein. Under hypoxic conditions, HIF-1α is stabilized and binds to HIF-1β before translocating to the nucleus to bind the hypoxia response elements (HRE) that targets genes involved in intracellular acid-base balances, such as carbonic anhydrase IX (CAIX) (154). Furthermore, it induces transcription of genes involved in glycolysis (including Glut-1, hexokinases (155, 156)), angiogenesis and proliferation (157). While HIF initiates increase in glycolysis, the glycolytic products pyruvate and lactate in their turn induce HIF-1α accumulation, indicative of a sustained feed-forward mechanism driving tumor metabolism towards glycolysis (158, 159).

Another result of HIF-1α upregulation is increased expression of CAIX as is described above (160) and MCT-4 (161). The interaction of CAIX with MCT-1 and MCT-4 is linked to acidification of the TME (162) and associated with poorer prognosis (163) and immune suppression (164). However, whether CAIX expression can serve as a surrogate for tumor hypoxia is debatable (165) and clinical studies on CAIX expression in lung cancer are scarce.

As VEGF is the main mediator of angiogenesis in many types of cancer to cater to chronic hypoxic conditions and as VEGF is under control of HIF signaling, it is also aberrantly expressed in lung cancer (166), in particular in adenocarcinoma (167). The level of VEGF expression is correlated with micro-vessel density and development of hypoxia and is involved in the so-called secondary vascular growth phase (168–170). It is suggested that, although VEGF expression stimulates angiogenesis, the disorganized and immature features of newly formed blood vessels in fact sustain the presence of intra-tumoral regions of hypoxia (171). Consequently, VEGF expression is in most studies correlated to a worse survival in NSCLC (172).

In parallel to oxidative stress, oncogenic mutations in lung cancer can also induce HIF activation, e.g., phosphatase and tensin homolog (PTEN), PI3K/Akt/mTOR pathway (152, 173), or epigenetic alterations (174, 175). As a consequence of increased HIF-1α signaling, PD-L1 expression on lung cancer cells increases (176–179).

On a general note, cancer cells show a greater metabolic plasticity than effector T-cells and have evolved to manipulate the host TME to their benefit, which enables them to utilize a variety of alternative metabolic pathways and substrates also under hypoxic conditions. Consequently, these alternative metabolic pathways often come with side-products, such as ROS, which require an additional set of processes to compensate for collateral damage. Although T-cells have differential metabolic preferences throughout their lifespan, they display limited plasticity or compensating pathways to deal with the ‘metabolic waste’ from cancer cells, resulting in ‘exhausted’ states in the TME (93, 101). For example, high levels of ROS in the TME are toxic for T-cells (180–182). Central to these effector function insufficiencies is mitochondrial function, which shows hyperpolarization, fragmentation, and increased ROS production in the TME (35, 59, 114, 130, 183). Furthermore, intratumoral hypoxia also limits T-cell migration away from the blood vessels into the tumor micro-environment, creating hypoxic immune privileged niches within the tumor (184). Thus, in addition to the limited availability of glucose itself, hypoxia further restricts T-cells’ capacity to perform aerobic glycolysis (paragraph 2.3), hampers T-cell infiltration and hypoxia-related waste products directly affects T-cell viability.

As mentioned above, induction of glycolysis is essential for T-cell effector functions and this induction is under control of mitochondrial ROS signaling and HIF-1α under normoxic conditions (6, 94) In particular Th17, Th1 CD4+ T-cells and CD8+ T-cells rely on increased glycolysis, whereas regulatory T-cells show less glycolysis dependency (93). For example, upon activation of CD3/CD28 on CD8+ T-cells, the expression of HIF-1α increases via PI3K/AKT/mTOR pathways to allow for increased glycolysis and effector functions such as IFN-γ and TNF-α secretion (185, 186). Under hypoxic conditions, HIF-1α induces downregulation of IFN-γ production by Th1 cells (187). These observations suggest a complex dual role for HIF-1α signaling in T-cells, which is environment and stimulus dependent. While glycolysis is required for T-cell effector functions, glutaminolysis and the pentose phosphatase pathway are necessary for biosynthesis. T-cell receptor triggering increases amino-acid transporters, along with upregulation of glucose metabolism, and therefore contributes to T-cell activation. However, given its different role glutaminolysis does not compensate for the dependency on glycolysis under hypoxic conditions. In fact, depletion of amino acids in the TME such as L-arginine by myeloid derived suppressor cells, inhibits T-cell proliferation (101). Perhaps the most important alternative metabolic pathways for T-cells in the TME to meet their metabolic demands is fatty acid oxidation (188). T-cell effector function is partially preserved by upregulating PPAR-a signaling to metabolize fatty acids under hypoxic and hypoglycemic conditions (39). Promotion of fatty acid metabolism could synergize with PD-1 blockade to control tumor growth, as shown in a preclinical melanoma model.

Several preclinical and human studies have identified roles of hypoxia in blunting treatment efficacy, as a longstanding notion across cancer types (106, 189, 190), and in particular in radiotherapy (191). Radiosensitivity starts to decrease at oxygen tensions below 2% oxygen, most directly by decreased availability of molecules for radiolysis to produce ROS by ionizing radiation. The hypoxia found in cancer also leads to downregulation of the type I IFN pathway, while this pathway is necessary for an adequate immune response.

In addition to directly reducing the therapeutic potential of ionizing radiation, the downregulated type I IFN pathway due to hypoxia impairs immune activation upon immunogenic cell death, a phenomenon that is observed for radio- as wells as chemotherapy (192). Furthermore, regulatory T-cells and memory CD8+ T-cells largely depend on oxidative phosphorylation, which is also restricted under hypoxia (101), and at least partly explains the arduous task of immune activation in hypoxic tumor regions. Lastly, the disturbed vascularization in tumors is known to hamper the intra-tumoral delivery of therapeutic agents, resulting in sub-therapeutic intra-tumoral concentrations (193, 194). To this end, anti-angiogenic treatments have been introduced in adjunct to radiotherapy (191, 195) targeted- or chemotherapy (196, 197). The overall results over combination treatments targeting VEGF in NSCLC so far have been disappointing (198).

More recently, other processes involving tumor vasculature associated endothelial cells have been identified, which act in addition to the typical vessel sprouting induced by hypoxia-driven or mutation-driven PI3K/Akt signaling. These processes include vessel co-option and vascular mimicry and may partly explain previous ambiguous results of combination treatments in NSCLC. These alternative angiogenic process also illustrate the complex network between NSCLC, supporting stromal cells, such as endothelial cells and pericytes, and mobile immune cell populations. It is generally accepted that angiogenesis factors drive an immune suppressive microenvironment (16, 184). In preclinical models VEGF inhibition resulted in enhanced T-cell infiltration and improved anti-cancer immune responses (199) and help the induction of tertiary lymphoid structures. These studies sparked the interest in combining anti-angiogenic treatment with immune checkpoint inhibitors (200).

Most of the current clinical hypoxia PET tracers are ¹⁸F-fluorinated nitroimidazole compounds, which target the altered redox status in cancer cells and its uptake is increased in hypoxic cells. The mechanism of fluorinated nitroimidazoles is based on an oxygen-reversible single-electron reduction of the nitro group, resulting in the formation of oxygen radicals which covalently bind to macromolecules in hypoxic cells (201), resulting in intracellular trapping of the tracer. In clinical setting, ¹⁸F-fluoromisonidazole ([¹⁸F]FMISO) is the most widely used tracer for hypoxia (202, 203). However, [¹⁸F]FMISO has slow clearance and low tumor uptake (204), which led to the development of second generation 2-nitroimidazole tracers, [¹⁸F]fluoroazomycinarabinofuranoside ([¹⁸F]FAZA), [¹⁸F]FETNIM, [¹⁸F]EF3, [¹⁸F]EF5 (205). Even a third generation 2-nitroimidazole hypoxia tracer ([¹⁸F]HX4, Figure 4 ) has been developed and clinically tested, showing more favourable pharmacokinetic and clearance properties than other ¹⁸F-fluorinated nitroimidazole compounds (206, 207). [¹⁸F]HX4 is for these reasons favored over previous hypoxia tracers for response monitoring to (chemo-)radiation therapy (208–210).

Besides nitroimidazole analogues, other compounds that target the redox status in cancer cells are diacetyl-bis(N (4)-methylthiosemicarbazone (ATSM), radiolabeled with different copper radioisotopes, or ionic copper (II) (211). [⁶⁴Cu]ATSM has several advantages over other nitroimidazole derivative hypoxia markers, including rapid tumor uptake and faster clearance from normoxic tissues (212). Several studies in lung cancer have shown that radiolabeled ATSM targets different regions within a tumor as compared to [¹⁸F]FDG (213, 214), and enable prediction of response to radiotherapy (215). Similar findings have been observed in patients with locally advanced HNSCC, in which [⁶²Cu]ATSM was evaluated as a predictor of response (216), with results paralleling [¹⁸F]HX4.

As hypoxia upregulates expression of CAIX on cancer cells, multiple radiotracers have been identified and tested pre-clinically for the imaging of CAIX, such as the anti-CAIX monoclonal antibody (mAb) G250, girentuximab (cG250), girentuximab antibody fragment Fab’ and F(ab’)2, and more recently affibody molecules. In a recent comparative preclinical study, the affibody ZCAIX:2, antibody fragment girentuximab-F(ab’)2, and a complete antibody-based tracer were evaluated for imaging upregulation of CAIX in head and neck cancer xenograft models (217). Radiolabeled girentuximab, girentuximab Fab’ and F(ab’)2 fragments are also evaluated in human colorectal cancer xenografts (218). According to these studies the complete girentuximab IgG tracer showed the most promising results in both human tumor xenografts. In the clinical setting, the chimeric mAb girentuximab is mostly tested for targeting of CAIX in clear cell renal cell carcinoma (ccRCC) (219), but no clinical studies have been performed on primary lung cancer.

Alternatively to molecular imaging tracers with a particular target in a hypoxia related pathway, multi-modal imaging that combines tissue characteristics, using dynamic contrast enhanced CT, and glucose metabolism, using routine [¹⁸F]FDG PET, was shown to accurately predict the presence of intra-tumoral regions with hypoxia (as defined by [¹⁸F]HX4 accumulation (208)).

As the agreement among different hypoxia-related tracers for PET imaging, or agreement with regional [¹⁸F]FDG uptake in NSCLC is modest (220–224), it remains critical to obtain tissue validation or solid clinical endpoints (225) when incorporating hypoxia tracers in NSCLC studies. However, the most extensively tested hypoxia-related imaging tracer for response prediction in the clinical setting is [¹⁸F]FMISO. In early stage NSCLC, the combined pattern of high [¹⁸F]FDG and high [¹⁸F]FMISO uptake was associated with an increased risk of recurrence after stereotactic radiotherapy (226). Such metabolic profile based in molecular imaging could help in guiding intensity-modulated treatment, as demonstrated in locally advanced NSCLC to avoid deleterious effects on organs-at-risk (227–229).

No studies have yet been performed regarding hypoxia imaging and immunotherapy in NSCLC, but in HNSCC [¹⁸F]FMISO imaging was used pre-clinically in combined anti-PD-1 and anti-CTLA-4 treatment to monitor changes in the TME during treatment. Preliminary data shows the potential to predict response to checkpoint blockade with anti-PD-1 and anti-CTLA-4 therapy [Reeves et al. J Nucl Med 2020, volume 61, supplement 1; 407, meeting report]. In HNSCC patients, increased lymphocyte infiltration is seemingly determined by a hypoxia-dependent response to chemoradiation (230), and persistent hypoxia during definitive chemoradiation treatment correlated with persistent PD-L1 expression and reduced outcomes (230), illustrating the potential of hypoxia related imaging to probe the tumor microenvironment. Several clinical trials are ongoing with [¹⁸F]FMISO as read-out in radiotherapy trials. One phase IB/II trial is ongoing to examine the feasibility and safety of the combination of two immune checkpoint inhibitor therapies (nivolumab and ipilimumab) in the neoadjuvant setting in resectable HNSCC. In this study, hypoxia measured by [¹⁸F]FMISO PET imaging is investigated as determinant for the effect of immune checkpoint inhibitors on the intratumoral T cell capacity (NCT03003637).

In addition to glucose, most tumor types also display increased uptake of amino acids, such as glutamine, to meet their high demands in biosynthesis and macromolecular synthesis (79, 231). Glutaminolysis is the intracellular conversion of glutamine to glutamate by glutaminase (GLS). This process is facilitated by the upregulation of the alanine-serine-cysteine transporter 2 (ASCT2, also known as SLC1A5) receptors in different cancer types (231), including lung cancer (232). In particular under low-oxygen conditions, glutamine becomes a carbon source for proliferating cancer cells to perform lipogenesis via reductive carboxylation (142), taking over up to 80% of de novo lipogenesis in A549 lung carcinoma cells (233). Via several other routes, glutaminolysis provides a back-up for metabolic pathways that are usually sustained by glucose metabolism; by providing a source of NADPH (234) and the glycolytic intermediate PEP when gluconeogenesis can no longer be performed (235). Thus, increased glutaminolysis in most cancer types illustrates their metabolic plasticity and provides an alternative source to glycolysis for intracellular bioenergetics. This dual reliance of lung cancer is further illustrated by the upregulation of glutaminolysis once glycolysis is suppressed (236).

Another important role for glutamate in cancer cells is its conversion into glutathione, a critical intracellular redox buffer, which is necessary to counteract the oxidative stress inflicted by aerobic glycolysis (237).

Similar to glycolysis, increased glutaminolysis is driven by increased signaling in the PI3K and/or Akt, which results in increased signaling of mTOR. Lung cancer frequently harbors mutations in the receptor tyrosine kinases or further downstream (238), and some of the metabolic heterogeneity observed in lung cancer cell lines can be attributed to mutations in KRAS or Trp53, apart from their histological subtype being adenocarcinoma (239) or squamous cell carcinoma (240).

Engagement of the T-cell receptor and the co-stimulatory molecule CD28 triggers pathways under the control of transcription factors HIF-1α and mTOR, which not only increase glycolysis, but also upregulate the expression of amino acid transporters (7). Thus activated and proliferating T-cells also display increased glycolysis and glutaminolysis (94, 100, 241), which associates with the increased expression of SCL1A5 for glutamine (242), similar to cancer cells. In vitro experiments demonstrated that glutamine deprivation indeed reduces T-cell proliferation, suppresses differentiation towards Th1 phenotypes but stimulates regulatory FoxP3+ phenotypes (243). In addition to its role as intracellular antioxidant similar to cancer cells, glutathione in T-cells also supports mTOR and NFAT activation, thus driving glycolysis and glutaminolysis (92) and promoting inflammatory responses.

Blocking glutaminolysis in lung cancer cell lines results in upregulation of PD-L1 expression via NF-κB activity and dampened T-cell activation, but when glutaminolysis is inhibited together with PD-L1 blockade, the balance tips towards T-cell mediated cancer cell death (37).

Besides glucose, and glutamine, T cells also consume tryptophan. Deprivation of tryptophan can impair the function of these T cells (244). Pathologic conditions as hypoxia induce the presence of IDO on tumors, resulting in a significantly increased tryptophan metabolism by the kynurenine pathway (245). This increase of the metabolic product kynurenine is toxic for T-cells and leads to immunosuppression (246). Tryptophan 2,3-dioxygenase (TDO) is like IDO as it also catalyzes tryptophan into kynurenine (247, 248). Since IDO and TDO both convert tryptophan into kynurenine, they are both important targets to image this tryptophan metabolism.