What is hypoxia?

Hypoxia generally refers to sub-physiologic tissue oxygen levels (<5–10 mmHg). Tumour hypoxia, a hallmark of malignancy, is a common and important feature of the tumour microenvironment. It is the consequence of an oxygen delivery versus consumption mismatch that occurs when cell proliferation outstrips neoangiogenesis during tumour growth. This results in very low oxygen levels (<5 mmHg) in tumours versus 40–60 mmHg in healthy tissues [1]. Hypoxia can generally be classified as (1) perfusion-related (acute) hypoxia due to insufficient blood flow, (2) diffusion-related (chronic) hypoxia caused by an increase in diffusion distances with tumour expansion, and (3) anaemic hypoxia caused by a decrease in oxygen transport capacity [2]. The latter two are considered relatively stable, whereas the degree of acute hypoxia may change in a short time. Cancer cells respond differently to decreased oxygen tension by eliciting cell death or cell survival, which partially depends on the time of exposure to hypoxia.

The origin of chronic hypoxia in human tumours was postulated by Thomlinson and Gray in 1955 [3]. Chronic hypoxia, also referred to as diffusion-limited hypoxia (DLH), is caused by consumption of oxygen by cells close to vessels, leaving inadequate oxygen for the cells further away from the vessels (>100 µm of capillary blood vessels), as demonstrated by means of phosphorescence lifetime imaging of R3230AC tumours in dorsal flap window chambers [4]. Chronic hypoxic changes are exacerbated in larger tumours and contribute to long-term cellular changes such as high frequency of DNA breaks, accumulation of DNA replication errors, potentially leading to genetic instability and mutagenesis [5, 6].

Brown and colleagues [7] were the first to present a second form of hypoxia: acute hypoxia. Acute hypoxia is an abrupt and brief exposure to short-term hypoxia (between a few minutes and up to 72 h) which occurs consequent to fluctuations in tumour perfusion accompanying functionally and structurally defective vascular network in tumour (overdilated, hyperpermeable, tortuous, and disrupted), and associated with high-interstitial pressure of the extracellular matrix [8]. This leads to periods of better or worse oxygenation [9, 10] promulgating the lexicon—cycling hypoxia [11]. Temporal occlusion of blood vessels caused by blood clots or tumour emboli can also cause acute hypoxia [12]. Acute hypoxia can lead to generation of high levels of reactive oxygen species (ROS) that can damage cells [13]. Cellular adaptations to these conditions have been enumerated and include decreasing oxidative metabolism and activating autophagy [14, 15]. Increased radio-resistance of cancer cells [13, 16], induction of spontaneous metastasis [10, 17, 18], and genomic instability due to delayed DNA damage response and rapid p53-dependent apoptosis [19, 20] can also result from hypoxia, leading to an aggressive tumour phenotype.

Hypoxia represents a unique tumour vulnerability to be exploited in the context of newly emerging personalised medicine strategies. Undoubtedly, both chronic and acute tumour hypoxia directly affect clinical responses to therapy by influencing tumour growth, ability to metastasize, and resistance to cell death.

Methods

A comprehensive PubMed literature search was performed, identifying articles relating to types of hypoxia, biological significance of hypoxia, measurement of hypoxia, MRI methods, and PET imaging of hypoxia in preclinical and clinical settings, up to December 2016. Search terms that were used to identify such articles were “hypoxia imaging,” “MRI,” “FMISO”, “FAZA”, “FETNIM”, “EF5”, “HX4”, “RP-170”, “Cu-ATSM”, and “PET” or “positron emission tomography.” Original publications in English were selected for inclusion in this review.

Biology and clinical significance of hypoxia

Tumour hypoxia is frequently seen in solid tumours, and tumour cells survive by activating different signalling pathways leading to a plethora of temporally or spatially heterogeneous changes in tumours (Table 1), elicited at different thresholds of oxygen tension [21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52]. In fact, during malignant growth, hypoxic regions are associated with increased genetic instability and more aggressive phenotype which correlate with tumour metastasis risk. Likewise, hypoxia causes unequivocal resistance to cancer treatments, such as reduced drug penetration, intrinsic chemoresistance (by mechanisms including loss of sensitivity to p53-mediated apoptosis or diminution of cell proliferation by metabolic stress), and resistance to ionizing radiation (reduced ability of oxygen to fix DNA lesions).

Table 1 Signalling pathways activated by tumour hypoxia promoting cell survival
Full size table

A number of biological consequences of low oxygen levels have been elegantly described by Höckel and Vaupel [53]. At pO2 levels less than 10–15 mmHg, cells become radioresistant and gene expression of hypoxia-regulated genes under control of hypoxia-inducible factor (HIF1) increases. Decreased adenosine triphosphate (ATP) synthesis is seen at pO2 levels less than 10 mmHg and together with decreased protein synthesis leads to lower oxygen consumption by cells. Finally, pO2 levels less than 1 mmHg reduce oxidative phosphorylation and conversely enhance glycolysis to maintain adequate ATP levels [54].

Role of HIF1alpha

Pathological hypoxia is a common microenvironment factor in tumours that facilitates cell survival and propagation of the tumour. The cross-talk between tumour and its microenvironment is essential for tumour survival [55]. Hypoxia-inducible changes not only affect tumour cells but also the tumour microenvironment [56]. Hypoxia-inducible factor 1 and 2 (HIF1 and HIF2, respectively) are oxygen-sensitive, heterodimeric transcription factors that act as key mediators of the cellular adaptation to low oxygen. HIF1 regulates glycolysis and pyruvate metabolism, and HIF-2 controls fatty acid metabolism. HIF1 is a heterodimeric protein consisting of HIF1alpha (oxygen regulated) and HIF1beta (constitutively expressed) dimers. Hypoxia stabilises HIF1alpha which stimulates expression of a variety of genes controlling metabolic pathways, pH regulation, angiogenesis, metastatic potential, DNA replication, protein synthesis, and treatment resistance, which (1) enhances cell survival via growth factor signalling and inhibition of pro-apoptotic pathways, (2) contribute to tumour neovascularization via VEGF, VEGF receptors, COX-2, iNOS, (3) regulate cell detachment via down regulation of adhesion molecules such as cadherins, and (4) induce cell migration and invasion through matrix degrading enzymes [57,57,59] (Table 1, Fig. 1).

Fig. 1
figure 1

Regulation of HIF1alpha in normoxic and hypoxic conditions and biological consequences of hypoxia

Full size image

Resistance to chemotherapy and radiotherapy mediated by HIF1 signalling

Drug resistance could potentially occur at the cellular level or secondary to changes in the tumour microenvironment. Tumours have convoluted vasculature which results in proliferating well-nourished cells closer to the functional blood vessels and regions of hypoxic cells located away from the functional blood vessels. Irregular blood flow and large distances between functional blood vessels in solid tumours lead to poor drug distribution, resulting in therapeutic resistance [60].

It is often difficult to discriminate between the effects of hypoxia per se and HIF1, and the literature on the effects of the hypoxic microenvironment and HIF1 on drug efflux and multidrug resistant phenotype [61], for instance, is controversial. Zhao and co-workers recently reported that HIF1alpha suppresses MDR1/P-glycoprotein in gastric cancer by inhibiting miR-27a expression in gastric cancer [62], and in colon cancer cells, inhibition of HIF1 leads to downregulation of p-glycoprotein and reversal of multidrug resistant phenotype [63]. In contrast, pronounced hypoxia has minor effect on p-glycoprotein expression and activity [64], while acidosis, a feature of the hpoxic micro-environement, increases p-glycoprotein activity [65]. In addition, regarding drug efficacy, the hypoxic environment can modify the efficacy of drugs that require molecular oxygen as part of their mechanism of action, e.g., bleomycins [66, 67], or are activated by reductases under hypoxia, e.g., evofosfamide, tarloxotinib, tirapazamine, and SN30000 [68, 69], and is a barrier to drug delivery generally independent of HIF1 [70]. Hypoxia and HIF1 also confer treatment resistance of cancer cells by inducing cell cycle arrest (quiescence) [71], making drugs that target cycling cells ineffective), and by supporting the highly tumourigenic stem cell niche [72]. In glioblastoma, HIF1alpha+ quiescent stem-like are found to locate within the peri-necrotic region and confer higher tumourigenic potential [73]. During severe or prolonged hypoxia, most of the cells undergo programmed cell death. However, some of the tumour cells adjust to environmental stress and survive by avoiding necrosis, inhibition of apoptosis [48, 74, 75], and decreasing senescence of cells [76], mediated by HIFalpha, resulting in an aggressive phenotype and resistance to treatment.

During fractionated radiotherapy, HIF1alpha protects the tumour microvasculature from radiation-induced endothelial apoptosis, via induction of vascular endothelial growth factor (VEGF) and other pro-angiogenic factors and facilitates tumour cell survival by increasing the antioxidant capacity of tumours to counteract radation-induced oxidative stress [21]. Irradiation also induces changes in the tumour microenvironment such as vascular, stromal, and immunological changes which may promote radioresistance and tumour recurrence [77]. These effects eventually lead to the resistance of tumour cells to chemotherapy and radiation.

Measurement of hypoxia

Knowledge of the hypoxia state enables prediction of treatment outcome and selection of patients for hypoxia modifying treatment. The relative prevalence of diffusion limited hypoxia (DLH), cyclic and perfusional hypoxia in human tumours or animal models is not known, and it is predicted that different hypoxia modes require different diagnostic and therapeutic approaches. Several noninvasive techniques (direct or indirect measurements) are available to obtain an absolute or relative value of the oxygenation status of tumours. The various strategies available are described in Table 2 [78,79,80,81,82,83,84,85,86,87,88,89,90,91,92,93,94,95,96,97,98,99,100,101,102,103]. Each of the techniques described interrogates different aspects of the hypoxic microenvironment, and they provide information on hypoxia at different locations: Oxygen electrodes and OxyLite sampling predominantly measure interstitial hypoxia. PET, SPECT, and extrinsic markers report on intracellular hypoxia and PET/SPECT images quantify data on a macroscopic scale in tumour regions. Blood oxygen level-dependent (BOLD)-MRI and Oxy-R fraction allow assessment of blood oxygenation, while indirect methods that report on hypoxia-induced molecular events (e.g., GLUT1, CA9 expression) rather than hypoxia itself have also been utilised as markers of tumour oxygenation.

Table 2 Methods of hypoxia assessment
Full size table

MRI methods

These include MR-based gradient-recalled echo techniques, electron paramagnetic resonance, and MR spectroscopy. MRI methods for interrogating tumour oxygenation are attractive since MRI scanners are widely available and they avoid the complication of short-lived radioactivity.

Blood oxygen level dependent (BOLD)-MRI

The most facile contrast mechanism, which depends on blood oxygenation—blood oxygen level dependent (BOLD) MRI—avoids the need for reporter molecules by imaging differences between diamagnetic oxy-haemoglobin and paramagnetic deoxy-haemoglobin. The presence of deoxy-haemoglobin in a blood vessel causes susceptibility differences between vessel and its surrounding tissue resulting in a decrease in T2* leading to darkening in tissues containing the vessel in a T2*-weighted imaging protocol. A limitation is that it is also sensitive to changes in Hb concentration (due to alterations in vascular volume and flow as well as interconversion of oxy- and deoxy-haemoglobin). Therefore, this technique provides qualitative assessment of changes in oxygenation rather than quantitative measurements. The technique is widely used for functional brain mapping [104, 105], where it is thought to primarily reflect changes in flow.

Baudelet and Gallez have rigorously investigated correlations between pO2 estimated using fibre optic probes and BOLD signal changes and have found general correlations, but a given BOLD response may reflect vastly different changes in pO2 [102]. BOLD MRI has the advantage of both high spatial and temporal resolution and it can be repeated as needed; however, it can be susceptible to subtle motion artefacts [106]. Rijpkema et al. used BOLD to evaluate patients during the ARCON trial for head and neck cancer and found significant changes in T2*-weighted MRI contrast accompanying hyperoxic gas breathing [107]. No accompanying changes were observed by traditional T1-weighted gadolinium dynamic contrast-enhanced MRI. Preliminary analysis of 11 women being treated with chemotherapy for locally advanced breast cancer showed a significantly different BOLD response to breathing oxygen before the course of chemotherapy for tumours of women with good therapeutic outcome versus those with poor response. Indeed, three women with complete pathologic response showed a signal change greater than 7%, whereas those with poor outcome showed less than 3% [108]. It is arguable whether the differential response reflects perfusion or oxygenation, but traditional dynamic contrast-enhanced MRI failed to provide similar discrimination.

The biologic sequelae of hypoxia are also amenable to imaging. Prolonged hypoxia can lead to increased lactate in tissues and 1H MRI can be used to image lactate [109, 110]. Furthermore, alteration of the redox state of nonprotein thiols, such as glutathione, adenine nucleotide redox state, NADH or NADPH in hypoxic cells can lead to accumulation of radiopharmaceuticals in hypoxia. All of these tests measure downstream consequences of hypoxia and often do not instantly return to normal values after an adequate O2 supply has been established. For more information, the reader is referred to a recent review that addressed the role of functional MRI (fMRI) methods to assess tumour oxygenation for predicting outcome [111].

PET imaging of hypoxia

Positron emission tomography (PET) has inherent advantages for studying hypoxia, as it can employ radiotracer probes that directly report on cellular oxygen levels, and not via hypoxia-mediated changes in phenotype, thereby permitting the non-invasive and three-dimensional assessment of intratumour oxygen levels in a more direct manner [112]. In contrast to histologic characterisation, PET can monitor whole tumours although at low spatial resolution [113]. PET has very high sensitivity and specificity compared to MR imaging and it enables the identification of regional hypoxia in vivo in preclinical and clinical settings [103].

PET tracers for hypoxia imaging and their mechanisms of action

The criteria for development of radiotracer probes includes improving relative tumour uptake by using isotopes with longer half-lives and ensuring rapid clearance of the parent compound from systemic circulation and normoxic tissue (hydrophilic compounds), while being sufficiently lipophilic to enter cells and allow uniform tissue distribution. The charcteristics of an ideal hypoxic tracer include: retention in low partial oxygen pressure (pO2) regions (hypoxia specific), pharmacokinetic profile and tissue distribution independent of confounding factors such as blood flow/tissue perfusion or pH, high stability, suitable tissue kinetics to enable imaging in a specified time frame, ease of synthesis, favourable dosimetry profile, reproducibility and effectiveness in multiple tumour types.

Radionuclide detection of hypoxia in tumours was first reported in 1981 with [14C]misonidazole autoradiography [114]. Subsequently, two main tracer classes have been developed to specifically study regional tumour hypoxia with PET: [18F]labelled nitroimidazoles and Cu-labelled diacetyl-bis(N4-methylthiosemicarbazone) analogues [112]. Multiple PET tracers suitable for the detection of hypoxia have been developed, validated and shown to exhibit different characteristics; some of these are discussed in Table 3 [115,116,117,118,119,120,121,122,123,124,125,126,127]. The first [18F]labelled drug to be clinically tested was fluoromisonidazole (FMISO) [128] and it remains the most extensively tested agent [129, 130].

Table 3 Salient characteristics of hypoxic radiotracers
Full size table

[18F]Nitroimidazoles

2-Nitroimidazole compounds were originally developed as hypoxic cell radiosensitisers and were introduced as hypoxia markers in the 1970s (Fig. 2) [115]. Nitroimidazoles enter cells by passive diffusion and subsequently undergo reduction forming reactive intermediate species. Hypoxic conditions cause further reduction of the nitro-anion radical, which is irreversibly trapped in the cell when the oxygen tension is less than 10 mmHg [129]. The reduction of nitroimidazoles requires the presence of ubiquitously expressed tissue reductases, which enables these compounds to accumulate within viable hypoxic cells, but not apoptotic or necrotic cells [130,130,132]. Under normoxic conditions, nitro-anion radical is re-oxidised into the parent compound, which can diffuse out of the cell. The mechanism of of [18F]MISO intracellular trapping is shown in Fig. 3 [133]. Therefore, hypoxic tissues can be delineated as an area of high tracer uptake after allowing a sufficient period of time for the nonspecific tracer to be excreted from the cells [134, 135].

Fig. 2
figure 2

Structures of clinically used [18F]-labelled nitroimidazole compounds

Full size image
Fig. 3
figure 3

Schematic representation of FMISO uptake in hypoxic conditions

Full size image

FMISO uptake was closely correlated with pimonidazole immunohistochemistry and has been found to reflect hypoxia in head-and-neck cancer [136,136,137,138,139,140,141,142,143,144,146], glioma [147,147,148,149,150,152], colorectal cancer [153], breast cancer [154], lung cancer [155, 156], and renal cell carcinoma [157, 158].

In view of FMISO’s slow plasma clearance, FMISO imaging usually requires an interval of longer than 2 h (ideally 4 h) after administration to obtain good contrast [159] with a hypoxia threshold in general defined as SUVmax of 1.5 or tumour:muscle ratio of 1.4 [103]. Although its biodistribution properties do not result in high-contrast images, the 2-h image unambiguously reflects regional pO2 in the range where it is clinically significant. However, due to perceived concerns regarding FMISO stability in vivo [160], metabolite formation, slow clearance properties [129], and failure to achieve image intensities in humans comparable to what had been achieved in animal models, alternative hypoxia PET tracers with different clearance and hydrophilicity characteristics have been developed in an attempt to overcome these limitations. These include fluoroazomycin arabinoside (FAZA), fluoroerythronitroimidazole (FETNIM), fluoroetanidazole (Fig. 4), and fluorinated etanidazole derivatives (EF3, EF5), HX4 [161,161,163].

Fig. 4
figure 4

Published in Br J Cancer 2004 (Barthel et al.) (Color figure online)

Hypoxia imaging with radiolabelled 2-nitroimidazole. a Chemical structure of [18F]fluoroetanidazole. The nitro moiety is necessary for hypoxia selective retention. b Cellular uptake of [18F]fluoroetanidazole in RIF-1 cell line culture grown under normoxia or hypoxia (nitrogen gas). The amount of radioactivity bound to cells was counted. c Imaging of [18F]fluoroetanidazole by PET showing tracer localisation in HT1080 (subclone 1-3C) xenograft. A 0.5-mm transverse slice of the 30–60 min image acquired in a small animal PET scanner is shown. Arrow, tumour. Courtesy of EOA

Full size image

[18F]Fluoroazomycin-arabinofuranoside (FAZA)

[18F]Fluoroazomycin-arabinofuranoside (FAZA) is more hydrophilic than FMISO. Consequently, it has faster clearance kinetics, resulting in improved tumour-to-reference tissue ratios, and thus hypoxia-to-normoxia contrast. Head-to-head comparisons between FAZA, [124I]IAZA, and FMISO in preclinical animal studies imaged at 3 h after injection demonstrated faster vascular clearance of FAZA, resulting in an increased tumour-to-blood ratio (5.19) relative to that of [18F]fluoromisonidazole (3.98). More recently, clinical studies have successfully evaluated the feasibility of FAZA for imaging hypoxia in gliomas [118], lymphomas [118], lung [164, 165], head-and-neck [118, 166,166,168], cervical [169], and rectal tumours [170], and the results have been shown to compare favourably with equivalent FMISO data, especially as improved hypoxic–normoxic contrast was obtained at earlier time points. High FAZA tumour-to-reference tissue values have been associated with reduced disease-free survival and have shown prognostic potential in the detection of hypoxia in head-and-neck patients [167]. Due to the higher tumour-to-reference tissue ratios in comparison with FMISO, FAZA is gaining popularity for PET imaging of tumour hypoxia. Despite the fact that FAZA is not widely available at present, increasing research demand may persuade more sites to produce it.

Next-generation tracers

[18F]2-(2-nitro-1H-imidazol-1-yl)-N-(2,2,3,3,3-Pentafluoropropyl)-acetamide (EF5)

The nitroimidazole EF5 has been extensively used for ex vivo immunohistochemical detection of bioreduced adducts, which indicate regions of tumour hypoxia. However, [18F]2-(2-nitro-1H-imidazol-1-yl)-N-(2,2,3,3,3-pentafluoropropyl)-acetamide (EF5), first investigated as a hypoxia PET tracer in 2001 [171], has only recently appeared in the clinical setting. In contrast to many of the second-generation hypoxia tracers, EF5 is highly lipophilic, resulting in greater cell membrane permeability and slower blood clearance [119], thus improving rates of tumour uptake and homogeneity of tracer distribution. The main drawback of EF5 is the complex labelling chemistry in comparison to the simple nucleophilic displacement reactions used for the mono-fluorinated 2-nitroimidazoles [171].

[18F]3-fluoro-2-(4-((2-nitro-1H-imidazol-1-yl)Methyl)-1H-1,2,3-triazol-1-yl)propan-1-ol (HX4)

[18F]3-fluoro-2-(4-((2-nitro-1H-imidazol-1-yl)methyl)-1H-1,2,3-triazol-1-yl)propan-1-ol (HX4), a next-generation 2-nitroimidazole tracer contains a 1,2,3-anti-triazole moiety (as a synthetic convenience) rendering it more hydrophilic than FMISO, specifically designed to maximize pharmacokinetic and clearance properties. Initial studies in humans demonstrate rapid renal clearance and urinary excretion of HX4, with a favourable dosimetry profile similar to that of FMISO [120].

Preclinical studies validated that the tracer uptake was indeed oxygen-dependent though tumour-to-background ratios appeared similar to those reported for FMISO in studies using the same tumour model [172]; thus, it remains to be seen if HX4 provides any significant advantage over FMISO in a clinical setting. A phase I study of 6 patients (4 non-small-cell lung carcinoma, 1 thymus carcinoma, and 1 colon carcinoma) has shown a median tumour-to-muscle ratio of 1.40 at 120 min after injection, although no attempt was made to determine the optimal imaging time points [120]. In head-and-neck tumours, HX4 produced tumour-to-reference tissue values similar to FMISO at relatively early time points post injection, indicating the potential advantage of shorter acquisition times [173]. A more recent study in non-small cell lung cancer (NSCLC) patients [174] suggested that later scan times (2–4 h p.i.) can further enhance the hypoxic-to-normoxic signal.

[18F]Fluoroerythronitromidazole (FETNIM)

The hydrophilic nature of [18F]Fluoroerythronitromidazole (FETNIM) accounts for its rapid renal clearance and low liver uptake, compared with FMISO. This also could explain the positive correlation between tumour blood flow and initial tumour FETNIM uptake [121]. Recent clinical studies in head-and-neck [121, 175, 176], lung [177, 178], cervical cancer [179], and oesophageal cancer [180] showed that high tissue uptake of FETNIM was indicative of reduced progression-free and overall survival. However, as with HX4, it is not clear whether the use of this tracer presents any advantages over FMISO imaging protocols. Clinical studies with FETNIM have been mainly carried out at the University of Turku, Finland.

[18F]1-(2-1-(1H-methyl)ethoxy)-methyl-2-nitroimidazole (RP-170)

1-(2-1-(1H-methyl)ethoxy)-methyl-2-nitroimidazole was developed as a 2-nitroimidazole-based hypoxic radiosensitiser, which has also been labelled with fluorine-18 ([18F]RP-170). The hypoxic selectivity of RP-170 was demonstrated in glioma patients on the basis of significant correlations between uptake, oxygen tension measurements, and HIF1alpha immunostaining [123]. Studies in brain [123, 181] and lung [182] tumours indicated higher SUV (calculated at 1 h post injection), for hypoxic than normal tissues. The shorter interval before scanning, combined with improved hypoxic contrast compared with FMISO, could make it attractive for clinical imaging.

Copper (Cu)-diacetyl-bis (N4-methylthiosemicarbazone) (Cu-ATSM)

An alternative class of agents for the study of hypoxia with PET that has been intensively investigated in both preclinical and clinical studies is the complex of Cu with diacetyl-bis(N4-methylthiosemicarbazone) (ATSM) ligands, among which ATSM is the prototype (Fig. 5). The potential of these agents for hypoxia imaging was first reported by Fujibayashi et al. [124]. Copper (Cu)-diacetyl-bis (N4-methylthiosemicarbazone) (Cu-ATSM) is a hypoxic marker that is selectively retained in hypoxic tissues. Cu-ATSM rapidly diffuses into the cells due to its high membrane permeability and low redox potential, secondary to its lipophilicity and low molecular weight. After cellular entry, Cu(II)-ATSM is reduced to an unstable Cu(I)-ATSM species, which further dissociates into the metal complex Cu(I), and ATSM, thus irreversibly trapping the Cu(I) within the cellular copper metabolic processes (Fig. 6) [183]. In normoxic conditions, the [Cu(I)-ATSM] can be re-oxidised to its parent compound, allowing efflux from the cell [184]. One of the advantages of Cu-ATSM is that it can reveal molecular contrast within 10–15 min post injection mainly due to its rapid tracer kinetics [125, 185].

Fig. 5
figure 5

Structure of [64Cu]Cu-ATSM

Full size image
Fig. 6
figure 6

Schematic representation of proposed mechanism of [64Cu]Cu-ATSM

Full size image

However, it has been observed that high uptake in tumours may only partly be a direct consequence of hypoxia [185]. Nevertheless, extremely high-contrast images of Cu-ATSM have been obtained in a variety of tumour sites [186]. The lack of correlation between Cu-ATSM distribution and immunohistochemistry hypoxia markers casted some doubt on the hypoxia selectivity of Cu-ATSM [187]. The suggested reason for the low correlation between Cu-ATSM uptake and hypoxic distribution, in some tumours, was the differing redox status of the tumour types. This has been further seen in pre-clinical experiments, where it was demonstrated that in the cell lines tested, [64Cu]Cu-ATSM and [64Cu]Cu-acetate had almost identical uptake in vivo over 2–16 h, post injection. However, up to 1 h post injection, [64Cu]Cu-acetate had a superior tumour-to-muscle ratio [188]. Several factors could explain the phenomenon; indeed, some tumours might have a lower than-average redox potential with high concentrations of electron donors causing reduction and trapping of Cu-ATSM in both hypoxic and normoxic areas. This observation does not discount the fact that [64Cu]Cu-ATSM may still be clinically relevant as a tracer for hypoxia, perhaps HIF1 status, as suggested by some investigators [189]. The timing of image acquisition is crucial, as the initial phase of tracer uptake can be perfusion and hypoxia-driven, whereas at later time points uptake is probably more indicative of tumour hypoxia.

Validation of MRI and PET hypoxia imaging

As discussed thus far, both MRI and PET play an important role in hypoxia imaging. However, there are few reports that compare these two imaging modalities. Preclinically, a clear correlation between [18F]FAZA PET image intensities and tumour oxygenation was demonstrated by Tran et al. [190]. [18F]FAZA accurately showed improved uptake when rats with subcutaneous rhabdomyosarcomas were treated with air, in contrast to carbogen. This correlated well with an invasive OxyLite probe, although the probe demonstrated a relatively high heterogeneity in the oxygen value measured depending on the specific point within the tumour. Functional MRI ([19F]MRI, however, did not show any discernible difference in T1 spin–lattice relaxation time. In a more recent study, Valable et al. validated tissue saturation studied by MRI against FMISO PET wth high sensitivity and specificity in a rat glioma model [191].

In the clinical setting, Swanson et al. performed a detailed spatial analysis of the hypoxic tumour burden visible on the FMISO PET relative to the imaging changes associated with tumour neovasculature, necrosis, invasion, and edema seen on gadolinium-enhanced T1-weighted MRI (T1Gd) in 24 patients with glioblastoma [152]. Hypoxic Volume (HV), defined within the tumour as sections that had a tumour-to-blood ratio of higher than 1.2, showed a consistent correlation with the MRI-defined regions within the tumour, supporting the idea that there is a definite link between the PET and MRI images of hypoxia. Furthermore, it was found that HV, and the respective surface areas of HV and T1Gd abnormality were the most significant predictors of survival. Simoncic et al., showed a strong correlation between FMISO PET and dynamic contrast enhanced MRI (DCE-MRI) kinetic parameters in 6 head and neck cancer patients [192].

These studies suggest that both MRI and PET could complement each other and provide a future direction in selecting the best modality to image hypoxia.

Clinical applications

There is evidence from numerous clinical studies across a range of tumour types to support the existence and importance of the “hypoxia driver” phenotype both in pre-clinical [193,193,194,195,196,197,198,199,200,201,202,203,204,205,206,207,208,209,210,211,212,213,214,215,216,217,219] (Table 4) and clinical studies [220,220,221,222,223,224,225,226,227,228,229,230,231,232,233,234,235,236,237,238,239,240,241,242,243,244,245,247] (Table 5).

Table 4 Preclinical studies of hypoxia imaging
Full size table
Table 5 Clinical hypoxia PET imaging studies in various tumours
Full size table

Identification of tumour hypoxia and prediction of prognosis/response to treatment

The clinical significance of hypoxia PET imaging is to identify individuals with poor prognosis and those likely to benefit from hypoxia-targeted therapy. Several studies have shown that hypoxia PET imaging predicts outcome. High FMISO retention has been associated with higher risk of loco-regional failure and shorter progression-free survival in head-and-neck [142, 248, 249, 252,252,254] and renal cancer [158]. Furthermore, a meta-review of the clinical data of over 300 patients concluded that FMISO is a predictor of poor treatment response and prognosis [131]. Similarly, FETNIM uptake in lung [176], head-and-neck [175], and oesophageal cancer [180], were also associated with poor outcomes. Studies conducted with FAZA in squamous cell carcinomas of the head and neck [167] and Cu-ATSM in patients with cervical [125, 261, 263], lung [125, 261], and rectal cancer [262] have shown that lower tumour-to-muscle ratio (TMR) is indicative of better prognosis.

These findings have been discussed in a recent meta-analysis of PET hypoxia studies which have demonstrated a common tendency towards predicting outcome in tumours showing higher tracer accumulation [162]. Decreased FMISO uptake with treatment has been widely reported in brain [152], head-and-neck [250, 255], lung [258, 260], and renal tumours [158]; although this was not seen in some tumours [142, 156]. Decrease in semi-quantitative imaging parameters such as tumour-to-muscle ratios (TMRs) signifying response to chemotherapy have also been demonstrated with Cu-ATSM in lung [125, 261] and head-and-neck tumours [257], and FAZA in lung cancer [165].

Radiotherapy planning

It is well known that tumours demonstrate temporal changes and/or heterogeneity in the spatial distribution of hypoxia. Identification of these areas with PET hypoxia scans enables image guidance and hence, radiation dose escalation to radioresistant sub-volumes [162, 248, 259]. Boosting the dose to intra-tumoural areas of biological resistance (dose painting) is being pursued as a strategy to overcome radioresistance and improve outcomes [264]. This is made possible due to the advances in imaging and radiation treatment planning. The feasibility of this strategy has been investigated in cancers of the head and neck, lung, and brain with Cu-ATSM [256], FMISO [251], and FAZA [166], mostly on anthrophomorphic phantoms [166, 249, 251], and further studies are required for translation into clinical benefit.

Discussion and concluding remarks

Hypoxia research has a long history; however, accurate and reproducible measurement of clinically relevant hypoxia with high sensitivity continues to evade the scientific community. Although radionuclide measurements of hypoxia started in the early 1980s we are yet to have a widely accepted method. In addition to studies with oxygen electrodes, imaging utilising exognous probes including FMISO-PET, FAZA-PET, HX4, and immunohistochemistry with pimonidazole have been the mainstay of hypoxia assessment in clinical studies; the EU-funded METOXIA consortium for example utilises HX4 for assessment of hypoxia [265]. In the post-genome era, genetic methods are also making an important entrance—with a 26-gene signature in validation for assessing hypoxia [266]. There have been some successes in the use of hypoxia measurements as part of clinical trials: a number of studies confirmed that hypoxia predicts locoregional failure to radiotherapy [81, 267, 268] and chemoradiotherapy with hypoxia-modulated cytotoxin tirapazamine [249]. Conversely, there have been several challenges, in particular poor sensitivity of the methods requiring long assessment periods (as pertains to nitroimidazole PET methods), invasive nature (as pertains to pimonidazole immunohistochemistry), an inability to relate measured output to oxygen tension (as pertains to MRI methods despite high spatial resolution) and wide heterogeneity within and between the tumours of the same patients and temporally with treatment (all methods).

Notably, despite design of newer nitroimidazoles with substantially different physicochemical properties—high hydrophilicity—concomitant improvements in signal-to-noise ratio and, thus, reduction in imaging times, have not been achieved, indicating that the ideal chemical design has not yet been realised. There is also paucity of studies examining heterogeneity of hypoxia using parametric imaging to detect the influence of hypoxia sub-volumes and whether this additional detail will have prognostic or predictive value. With the advent of PET-MRI scanners, it will be feasible to multiplex imaging modalities to provide addition information such as perfusion to increase accuracy of hypoxia measurements or provide complementary information with higher predictive value. Whatever selected method will require assessment of precision of measurement which is non-trivial with such a spatio-temporally evanescent phenomenon as hypoxia.

Thus, it is accepted that hypoxia could have significant prognostic and predictive value in the clinic; however, the best method for hypoxia assessment has in our opinion not been realised.