Imaging of hypoxia in brain tumours



The validation path of hypoxia PET imaging: a focus on brain tumours

Natale Quartuccio, Marie-Claude Asselin*

Division of Informatics, Imaging and Data Sciences, Wolfson Molecular Imaging Centre, University of Manchester, UK

Abstract

Gliomas are brain tumours arising from the glia, the supportive tissue of the central nervous system (CNS) and constitute the commonest primary malignant brain tumours. Gliomas are graded from grade I to IV according to their appearance under microscope. One of the most significant adverse features of high-grade gliomas (grade III-IV) is hypoxia, a biological phenomenon that develops when the oxygen concentration becomes insufficient to guarantee the normal functions of a tissue. Since tissue hypoxia influences negatively patient outcome and targeting hypoxia has potential therapeutic implications, there is currently great interest in imaging techniques measuring hypoxia. The aim of this review is to provide up to date evidence on the radiotracers available for measuring hypoxia in brain tumours by means of Positron Emission Tomography (PET), the most extensively investigated imaging approach to quantify hypoxia. The review is based on preclinical and clinical papers and describes the validation status of the different available radiotracers. To date, [F-18]fluoromisonidazole ([18F]FMISO) has been the most widely used radiotracer for imaging hypoxia in patients with brain tumours, but experience with other radiotracershas expandedin the last two decades.

Validation of hypoxia radiotracers is still on-going and essential before these radiopharmaceuticals can become widely used in the clinical setting. Availability of a non-invasive imaging method capable of reliably measuring different levels of oxygen inbrain tumours would be of outstanding clinical importance to select patients that may benefit from tailored treatment strategies targeting hypoxia.

Keywords: Hypoxia, PET, brain tumour, glioma, preclinical, patients, FMISO, FAZA

*Address correspondence to this author at the Wolfson Molecular Imaging Centre, University of Manchester, 27 Palatine Road, Manchester, M20 3LJ, UK; Tel/Fax:+44 (0)161 275 0068; E-mail: marie-claude.asselin@manchester.ac.uk

1. INTRODUCTION

1.1 Importance of hypoxia and therapeutic implications of measuring tumour hypoxia in brain tumours

Hypoxia is a state characterized by inadequate oxygen concentration to allow the normal execution of biological functions (1). Hypoxic state in tumours, together with abnormal vasculature, is a particularly relevant feature of gliomas, a group of brain tumours which arise from the glia, the supportive cells of the central nervous system (CNS) (2).Gliomas are classified by the WHO in four grades according to their microscopic features. High-grade gliomas are the most aggressive ones and include anaplastic astrocytoma (grade III) and grade IV glioma, formerly known as glioblastoma multiforme (GBM)(3).Noteworthy the median pO2 in normal brain tissue is above 35 mmHg, whereas brain tumours have a median pO2 around 13 mmHg and contain also regions with severe hypoxia (pO2≤ 2.5 mmHg)(4, 5).

Hypoxia compromises the effectiveness of radiotherapy, diminishes the efficacy of standard chemotherapy plans and carries a worse prognosis (1, 6-8). Hypoxia imaging has been shown to have multiple potential applications for glioma patients, such as grading, prediction of prognosis, evaluation of treatment response and stratification of patients with higher chance of benefit of hypoxia-modifying treatments[pic](9). Strategies to fight hypoxia have classically included the improvement of tumour oxygenation (e.g. carbogen administration), administration of agents increasing the sensitivity of tumour cells to irradiation (radiosensitizers) or drugs that become activated only under hypoxia (hypoxia-activated cytotoxic prodrugs)(10, 11). Trials employing hypoxia-modifying treatments have demonstrated variable modification of disease-free survival and loco-regional tumour control in different tumour types(11); however these studies were characterized by an aprioristic approach and lacked an accurate assessment of hypoxia, complicating the evaluation of the real benefit of hypoxia-modifying treatments (12-14). The use of hypoxia imaging to guide hypoxia-tailored radiotherapy plans, as indicated by simulation and feasibility studies including patient with head-and-neck cancers, appears theoretically feasible and could result in higher tumour control (15-20). There is lack of similar pilot studies in patients with brain tumours, however a computation model, taking into account a patient with a grade IV glioma, suggested that hypoxia PET imaging could predict areas of high cellular density with hypoxia-related resistance (16). In gliomas an additional problem for targeting hypoxia derives from the tumour location because the hypoxia-targeting drugs should be ideally constituted of small lipophilic molecules to cross the blood-brain barrier (BBB). PET imaging is an attractive technique to measure hypoxia since it is relatively non-invasive and allows full tumour coverage. The ideal radiotracers for imaging hypoxia should have a mechanism of tracer retention exclusively lead by hypoxia, being minimally influenced by additional factors such as blood flow or pH. Furthermore, low non-hypoxic metabolism is desirable. In the specific application for brain tumours, a tracer should have a convenient octanol:water partition coefficient (P - a parameter representing the relative lipophilic or hydrophilic nature of a compound) to cross the BBB and wash out from normoxic cells, finally steadily localizing into tissue with clinically relevant hypoxia (21). The BBB is a permeable but highly selective barrier that isolates brain extracellular fluid in the central nervous system (CNS) from circulating blood and is comprised of endothelial cells, separated by tight junctions, astrocytes end-feet and pericytes(22). Lipophilic radiotracers are able to cross quickly cell membranes but have the drawback of slow pharmacokinetics; indeed the slow clearance of the unbound tracer from normoxic tissue lead to low hypoxic-to-background contrast. For fluorinated lipophilic tracers, the relative short half-life limits the time allowed to permit the optimal clearance from normoxic tissue. On the contrary, hydrophilic radiotracers have a lower first pass extraction coefficient resulting in a lower penetration through the BBB; however hydrophilic radiotracers have faster clearance from normoxic cells, leading theoretically to a better hypoxic-to-normoxic contrast at an earlier time (23). To further complicate the situation, the BBB in high grade gliomas is dysrupted, therefore the validation of a tracer would optimally require a cross-validation with a free diffusible tracer to estimate the blood flow and to ascertain that the tracer does not simply reflect just a BBB leakage. Due to the fact that the BBB limits the passage of hydrophilic molecules, hydrophilic tracers are more susceptible to reflect the disruption of the BBB. Therefore the question whether a hydrophilic radiotracer is far better than a lipophilic tracer for the imaging of hypoxia in brain tumours is not simple to be answered. Another important distinction is between fluorinated and non-fluorinated radiotracers. In general fluorinated radiotracers demonstrate low tumour-to-background ratio (TBR). On the contrary, non-fluorinated radiotracers (e.g. labelled with radioisotopes of I-124, Ga-68 or Cu-64), although leading to better image contrast, have been less investigated and the reliability of some of them has been questioned(11, 24).Hypoxia PET images are conventionally decay corrected and each voxel is absolutely calibrated in Bq/ml tissue. The voxel values are then normalised to blood radioactivity concentration typically obtained from blood samples taken during the PET scan, leading to tissue-to-blood ratio(T/B) at a voxel level (25). Generally, values of T/B>1.2 or above (e.g. 1.4) have been used to identify hypoxic regions in brain tumours by means of [F-18]fluoromisonidazole[18F]FMISO)(26, 27). Delineation of a reference region within the imaging the field of view has also been used as an alternative to blood sampling, for brain tumours; due to the particular environment, the use of a cerebellar region has been suggested for the calculation of the TBR(28). Some authors found calculation of the tumour hypoxic volume (HV) (29) or of the fractional hypoxic volume (FHV) useful as an indicator of the extent and severity of tumour hypoxia (29, 30). The estimation of the HV, which is the volume of the tumour that is hypoxic, requires setting a T/B threshold while the FHV requires in addition the measurement of the total tumour volume (2, 11), most often from anatomical images.

Hypoxia PET tracers are reviewed in turn, for each the status of the validation presented, followed by preclinical studies brain tumour models and leading to clinical studies in glioma patients. PET tracers used to image hypoxia in patients with brain tumours arecovered in greatest details first and compared in table 1. Results from preclinical PET studies in brain tumour models with hypoxia tracers are summarised in table 2. Hypoxia PET tracers evaluated in tumours other than glioma are covered next and finally other PET tracer in development for measuring tumour hypoxia are covered last.

|Radiotracer |Year introduced |Main characteristics |Drawbacks |References in clinical |

| | | | |setting (brain tumours) |

|[18F]FMISO |1987 |Selective binds to hypoxic viable cells; widely validated in different |Slow clearance kinetics and low TBR. |(27-29, 31-40) |

| | |types of tumours. Potential use as prognostic indicator. logP=0.4 | | |

|[18F]EF5 |2001 |Promising tracer with possible correlation with outcome. Limited | Low tumour-to-normal brain ratio Discordant results |(41-43) |

| | |formation of metabolites. High lipophilicity logP=0.6 |regarding biological validation of the tracer with the | |

| | | |non-radiolabelled counterpart EF5. | |

|[18F]FAZA |2002 |Faster clearance and higherTBR than [18F]FMISO. Less lipophilic than |Less validated than [18F]-FMISO. Poor penetration of |(44) |

| | |[18F]FMISOEF5.logP=0.04 |intact blood–brain barrier. | |

| | | |Lower absolute tumour uptake in comparison with FMISO in | |

| | | |preclinical studies with different tumour xenograft | |

| | | |models. | |

|[18F]FRP170 |2002 |Correlation of high SUVmax with HIF-1 level. A cut-off of 1.7 for TBR |Limited data in literature. Lack of comparison studies |(45-47) |

| | |distinguishes between hypoxia and normal tissues. Shorter interval |with 18F-FMISO. | |

| | |between injection and acquisition than [18F]-FMISO. logP=0.094 | | |

|[60,61,62,64Cu]A|1997 |Freely diffusible, rapid visualization of hypoxia (good hypoxic-normoxic|Limited availability of Cu isotopes. Uptake related to the|(48, 49) |

|TSM | |uptake ratio). |activity of intracellular reductive agents. High radiation| |

| | |The radioisotopes of copper used to radiolabel ATSM range from |exposure for the patients (64Cu). Coupling with blood flow| |

| | |short-lived (62Cu = 9.7 min and 60Cu = 23.7 min) to long-lived (61Cu = |when imaging with short-lived radioisotopes. | |

| | |3.33 h and 64Cu = 12.7 h). logP=1.85 | | |

TABLE 1: Comparison of the PET tracers used to image hypoxia in patients with brain tumours

logP= logarithm of octanol/water partition coefficient. Log P represents a measure of lipophilicity; higher values indicate higher lipophilicity.

Methods

A comprehensive literature search of the different scientific databases (PubMed, Scopus, Embase-Ovid, EBSCO and Web of Science) was performed looking for relevant original articles regarding the use of PET or PET/CT in the evaluation of patients withglioma. The literature databases were investigated using a string based on the following keywords: PET AND hypoxia AND brain AND (tumour OR tumour OR glioma OR glioblastoma). Both prospective and retrospective studies were included. The search was updated until the end of November 2016. No start date was used. To expand the search, the references of the retrieved articles were also considered. Abstracts and articles in languages other than English were excluded.

2.1 [18F]FMISO

2.1.1 Pharmacokinetics and validation status

At the present time, [18F]FMISOis the most widely used PET tracer to quantify tumour hypoxia (9, 50). [18F]FMISO belongs to the 2-nitroimidazole-based class of compounds. Nitroimidazolespassively diffuses into cells at with different kinetics depending to their lipophilicity(51). Under hypoxic conditions, with pO2 ≤10 mmHg, the compound undergoes electron reductions and forms reactive radicals. These bind covalently to intracellular macromolecules, so that the tracer is trapped inside hypoxic cells. Conversely, in normoxic cells, the tracer is reoxygenated and reconverted to the original form, passing out through the cell walls (52).

The logP value for FMISO is close to 1 (approximately 0.4), therefore, at equilibrium, the normoxic tissue-to blood ratio would approximate the unity, whereas higher values should be found for the hypoxic-to-normoxic tissue ratio (21).The main drawbacks of [18F]FMISOare the slow uptake in the target tissue and the slow clearance of the unbound tracer from non-hypoxic tissue resulting in high background in PET images (53). Most of the injected [18F]FMISO activity has been demonstrated to homogeneously distributesin organs of human healthy volunteers and radiation exposure is similar to other fluorinated tracers (54).Classically, early images with [18F]FMISO acquired up to 5 minutes after injection allow estimation of delivery, whereas delayed images, taken about 3 hours after injection, following washout of non-bound tracer, provide estimation of hypoxia[pic](9). Some other authors have proved that performing static acquisition at 4 hours p.i. in HNSCC patients provide a superior image contrast than [18F]FMISO 2h p.i.(55). However, Peeters et al. reported that T/Bof [18F]FMISO was still increasing at 6 hours (9 ± 0.8) in rhabdomyosarcoma R1-beraing rats (n=16) undergoing serial dynamic scan up to 6 hours post injection[pic](56). Moreover, in vitro experiments in different cell lines suggest that anoxic/oxic binding ratios for FMISO can be influenced by other intrinsic factors including glucose level and cell growth state (57).Pharmacokinetics of [18F]FMISOhas been compared systematically to other tracers in different preclinical studies including also brain tumour models (9L) (table 2) (56, 58-61).

[18F]FMISO has been assessed and validated in vitro, in preclinical studies and in patients in a wide range of malignancies including brain tumours showing high uptake in hypoxic tumour tissue and validated against the “gold standard” immunohistochemical hypoxia marker pimonidazole (62-65). Reproducibility of [18F]FMISO PET-derived hypoxia parameters in brain hasnever been tested in brain tumours but has been demonstrated in other preclinical tumour models, in patients with lung cancer and, with discordant results, and in patients with head-and-neck cancer(56, 66-68).Noteworthy, in the above mentioned study of Peeters and co-workers, [18F]FMISOresulted the tracer with the best reproducibility and the most sensitive one to reoxygenation (tested in rhabdomyosarcoma R1-bearing ratsunder exposure to carbogen/nicotinamide) compared to [F-18]Fluoroazomycinarabinofuranoside([18F]FAZA) and Flortanidazole ([18F]HX4). Conversely, [18F]FAZA and [18F]HX4 resulted to be more sensitive to acute hypoxia (exposure to 7% oxygen breathing)(69).The main findings of the preclinical PET hypoxia studies in brain tumour models are summarized in table 2.

2.1.2. Preclinical evidence in brain tumour models

Many preclinical studies have investigated [18F]FMISO in animal studies with different brain tumour models(61, 64, 65, 70-76).

2.1.2.1 Biological validation

Many studies have biologically validated [18F]FMISO using pimonidazole staining (65, 70, 77). In a preclinical studyaiming to investigate the biologic characteristic of [18F]FMISOdistribution in tumours and involving a C6 glioma xenograft rat model (n=6) were evaluated by [14C]FDGautoradiography and[18F]FMISOPET autoradiography. A significant higher percentage of positively stained areas of Glucose transporter-1 and hexokinase-II (whose genes are overexpressed under hypoxic conditions) was found in regions with high [18F]FMISO uptake (determined by autoradiography) compared to regions with low [18F]FMISO uptake. No significant differences in [14C]FDG uptake and Ki-67 index were found between [18F]FMISO positive and negative regions, suggesting a not strictly direct correlation between hypoxia status, proliferation and glucose metabolism in the C6 brain tumour model (65).

2.1.2.2 Evaluation of hypoxia in different tumour models

Evaluation of hypoxia severity by means of [18F]FMISO PET scan has also been demonstrated to be associatedwith the effect of antiangiogenic treatment (decrease of vascular permeability and increase of CBV) in a preclinical study assessing the hypoxic volume (HV) in 9L (n=6) and C6 (n=6) rat models undergoing a small animal PET scan (dynamic=0-3 h)(74). In this study the 9L model did not present [18F]FMISO uptake as opposed to the C6 model, which is less vascularised and more hypoxic. Difference in[18F]FMISO uptake have also been described for the U87 (n=3) and U251 (n=3) orthotopic rat models in a preclinical study evaluating [18F]FMISO uptake by a 20-min static PET scan acquired2 hours post injection.[18F]FMISO uptake was demonstrated in the U251 but not in the U87 model, whose hypoxia level does not probably fall below the threshold triggering FMISOtrapping(70). The feasibility of detecting hypoxia in the U251 model has also been further confirmed recently in another study including seven U251 orthotopic rat modelsalong with C6 (n=7) and 9L (n=7)models undergoing a 20-min dynamic scan two hours following [18F]FMISO injection. In this study [18F]FMISO PET detected hypoxiaonly in the U251-MG and C6 models demonstrated and was cross-validated with pimonidazole staining and multiparametric quantitative blood-oxygen-level dependent (BOLD) MRI (77). (75)

2.1.2.3 Treatment response

A recent study has investigated the utility of [18F]FMISOin evaluating the fluctuations of the hypoxic status in C6 tumour cells and the corresponding C6 rat models (n=32) during treatment with the radiosensitizer irisquinone (IQ). The rats were treated either with IQ, radiotherapy or IQ+radiotherapy scanned with [18F]FDG PET and [18F]FMISO (2 hours p.i.) in two consecutive days before and after treatment. The rats receiving radiotherapy or IQ+ radiotherapy demonstrated decreased post-treatment [18F]FMISO uptake, whereas rats treated with QT alone and a control group showed increased [18F]FMISO and [18F]FDG uptake (75). Evaluation of effects of the hypoxic drug THL-302 on tumour growth has instead investigated in another study including C6 (n=8) and 9L (n=6) orthotopic rat models undergoing dynamic PET scanning before and after treatment. In this study C6 and 9L showed significant decrease in tumour growth rate compared to a control group (78).

2.1.3 Clinical evidence in brain tumours

Valk et al. were the first to demonstrate that [18F]FMISO PET is a feasible method for detecting hypoxia in high-grade gliomas in 3 patients (2GBM and 1 anaplastic astrocytoma) by means of static PET scan (range:120-180 min p.i.) reporting the tumour-to-plasma (T:P) ratio (0.71-1.49 at 2h)(79). Interestingly, a dual tracer PET study, utilizing [15O]H2O and dynamic (0-90 min) and static (at 150-170min p.i.) [18F]FMISO PET documented no impact of BBB at late time-pointson [18F]FMISO uptakein 6 GBM patients (80). The distribution of [18F]FMISO uptake (determined by calculation of the uptake rate K1) was highest in the tumour margins but low in the central core of the tumour, where also [15O]H2O was low. Pixel-by-pixel for each patient analysis for[18F]FMISO and [15O]H2O images showed variable good correlation (r=0.42-0.86) at early time (0-5 minutes) after injection but not at 60-90 minutes or later time-points (150–170 min p.i.), suggesting that distribution of [18F]FMISO may be spatially independent of perfusion at a regional level at late time (81). Dynamic PET acquisition and arterial blood sampling however may be difficultly translated to clinical routine. Therefore, as alternative, another group used an image-derived (ID) blood surrogate reference (a 2-cm diameter region in the left and right cerebellum) and selected an ID T/B ≥1.2 to define tumour hypoxia in 64 glioma patients(28). Despite the suboptimal time window of the 20-min static PET scanning (started 90-140 min p.i.), the ID blood activity demonstrated high correlationto measured activity (R2=0.84) in venous blood samples.

Distribution of [18F]FMISOhas also been compared with distribution of [18F]FDG in patient with GBM at a voxel level with heterogeneous and discordant correlations between hypoxia and glucose metabolism (26, 82, 83). The biological link between [18F]FMISO distribution and tumour aggressiveness has instead been explored in a study of Kawai and colleagues (34) (n=10 patients with GBM) undergoing serial static PET scans at 15min and late time points (130-160min. The authors identified a strong correlation between hypoxia parameters (HV and T/B≥1.2) with the metabolic volume defined on [11C]methionine PET, identified by using a [11C]methionine PET index of ≥1.3. Although the authors did not investigate correlation with immunohistochemistry markers, the study supports the hypothesis that [11C]methionine PET and FMISO PET may play a complementary role with the first able to identify the infiltrative pattern of GBM and the second one useful to detect the interlinked hypoxic burden and related phenomena (e.g. neovascularisation) and eventually influence the surgical and post-operative treatments. Also of interest is the complex relationship existing between hypoxia and angiogenesis in GBM where PET imaging may provide complementary information to MRI. This wasshownin a study of Swanson et al. (32). The authors documented a strong correlation between HV, determined bystatic [18F]FMISO PET (120-140 min p.i.), and altered vasculature, documented on Gd-enhanced T1-weighted MRI sequences, with the HV partially overlapping the outer edge of the enhancing region (32).

2.1.3.1 Grading

An important role of [18F]FMISO PET may be the grading of gliomas. Hirata et al. first documented findings supporting theutility of delayed static 18F]FMISO scan (at 240 min) in 23 patients differentiating GBM from lower grade gliomas based on the SUVmax and HV (tumour voxels with SUV >1.3 timesthe mean SUV in a cerebellar ROI)(35).Also Yamamoto reported no hypoxia (no pixel with T/B max ≥ 1.2) in grade I (n=1) and II (n=6), as opposed to grade III (n=7), and IV (n=16) gliomas performing static PET scan 2 hours p.i.(31). In another study with 32 newly diagnosed and 16 recurrent high grade gliomas, higher T/B max and HV were also found in patients with newly diagnosed grade IV glioma compared to patients with newly diagnosed grade III glioma (p ................
................

In order to avoid copyright disputes, this page is only a partial summary.

Google Online Preview   Download