Positron emission tomography to assess hypoxia and perfusion in lung cancer

Abstract

In lung cancer, tumor hypoxia is a characteristic feature, which is associated with a poor prognosis and resistance to both radiation therapy and chemotherapy. As the development of tumor hypoxia is associated with decreased perfusion, perfusion measurements provide more insight into the relation between hypoxia and perfusion in malignant tumors. Positron emission tomography (PET) is a highly sensitive nuclear imaging technique that is suited for non-invasive in vivo monitoring of dynamic processes including hypoxia and its associated parameter perfusion. The PET technique enables quantitative assessment of hypoxia and perfusion in tumors. To this end, consecutive PET scans can be performed in one scan session. Using different hypoxia tracers, PET imaging may provide insight into the prognostic significance of hypoxia and perfusion in lung cancer. In addition, PET studies may play an important role in various stages of personalized medicine, as these may help to select patients for specific treatments including radiation therapy, hypoxia modifying therapies, and antiangiogenic strategies. In addition, specific PET tracers can be applied for monitoring therapy. The present review provides an overview of the clinical applications of PET to measure hypoxia and perfusion in lung cancer. Available PET tracers and their characteristics as well as the applications of combined hypoxia and perfusion PET imaging are discussed.

Key Words: Molecular imaging, Positron emission tomography, Hypoxia, Perfusion, Quantification, Lung cancer

Core tip: This review provides an overview of the current applications of positron emission tomography for hypoxia and perfusion imaging in lung cancer. Available PET tracers are discussed and the benefits of combined hypoxia and perfusion PET imaging are clarified. Hypoxia imaging could aid in selecting patients for hypoxia-specific treatment strategies. To achieve this, consensus about the optimal imaging protocol and quantification method is essential. Large clinical trials are needed to confirm the value of hypoxia imaging for improving patient care.

INTRODUCTION

Worldwide, lung cancer is the most common cause of cancer related death among men and women[1]. Every year, approximately 1.2 million new cases of lung cancer are diagnosed globally and 1.1 million patients die of this disease[2]. Non-small cell lung cancer (NSCLC) and small cell lung cancer (SCLC) are the main histological types and represent approximately 85% and 15% of the lung cancer cases, respectively[3,4]. The prognosis of both NSCLC and SCLC is poor and depends on the stage of the disease[5,6]. For example, the 5-year overall survival is approximately 1% and 2% for stage IV NSCLC and extensive stage SCLC, respectively. Treatment of lung cancer depends on histological type, stage and performance status. The available treatment options include surgery, radiation therapy and chemotherapy, or a combination of these modalities. Systemic therapy of lung cancer consists mainly of a platinum-based doublet, such as cisplatin or carboplatin, in combination with a third generation cytotoxic drug such as gemcitabine, pemetrexed, paclitaxel or docetaxel[7,8]. In addition, targeted agents, including gefitinib, erlotinib, bevacizumab and crizotinib, have been introduced for the treatment of advanced NSCLC[9-16]. For the last decades, several tumor characteristics have been under investigation in order to further understand the biology of lung cancer and enhance the efficacy of the several treatment modalities.

In lung cancer, tumor hypoxia is a characteristic feature[17], which is associated with a poor prognosis[18-20] and resistance to both radiation therapy[21] and chemotherapy[22]. Hypoxia is a reduced O₂ tension in tissue and is defined between normoxia (pO₂ levels of 40-60 mmHg) and anoxia (0 mmHg)[23]. In clinical practice, no consensus has been achieved for hypoxic thresholds in tumors, but tumors with pO₂ values below 10 mmHg are usually considered hypoxic[23]. Tumor hypoxia is the result of an imbalance between oxygen supply and consumption and can be caused by the following mechanisms[23]: (1) the structurally and functionally abnormal tumor vasculature leads to a perfusion-limited delivery of oxygen[24], thereby inducing “acute” hypoxia; (2) tumor proliferation increases the distance between tumor cells and blood vessels that provide nutrients and oxygen to tumor cells. Consequently, the distances to blood vessels can become larger than the diffusion distance of oxygen (> 70 μm), locally causing diffusion-limited hypoxia (referred to as “chronic” hypoxia); (3) tumor hypoxia is also associated with a systemic decrease in oxygen supply, i.e., anemia, which can be caused by tumor-related factors as well as anticancer therapy.

To promote cell survival in hypoxic conditions hypoxia inducible factor-1 (HIF-1) is upregulated, which in turn activates a number of processes including growth factor signaling, angiogenesis, proliferation, glycolysis, tissue invasion, and finally metastasis[25]. As a result, markers of the HIF signaling cascade such as HIF-1α, glucose transporter-1, and vascular endothelial growth factor (VEGF), have been investigated as surrogate markers for tumor hypoxia in lung cancer[18,19,26,27]. Alternatively, immunohistochemical staining using injectable exogenous bioreductive markers like pimonidazole and 2-(2-nitro-1[H]-imidazol-1-yl)-N-(2,2,3,3,3-pentafluoropropyl)-acetamide (EF5) can be applied[28]. However, immunohistochemistry requires tissue samples and represents an indirect measurement of tumor hypoxia. Alternatively, pO₂ levels in tumors can be directly assessed using Eppendorf polarographic electrodes. This an invasive technique that can be applied in tumors that are easily accessible[29]. In lung cancer, this technique is not feasible[17], as these tumors are usually deeply seated within in de body. Positron emission tomography (PET) may be useful, as PET enables direct assessment of tumor hypoxia in patients non-invasively[30].

As the development of tumor hypoxia is associated with decreased perfusion, perfusion PET imaging may provide more insight into the relation between hypoxia and perfusion in malignant tumors. PET scans may not only reveal the prognostic significance of hypoxia and perfusion in lung cancer, but may also help to select patients for specific treatments including radiation therapy, hypoxia modifying therapies, and antiangiogenic drugs[31,32]. This review provides an overview of the clinical applications of PET to measure hypoxia and perfusion in lung cancer.

PET PRINCIPLES

PET enables non-invasive 3D imaging of dynamic processes in vivo. To this end, molecules of interest are radiolabeled with positron emitting radionuclides. For PET imaging, commonly used radionuclides are oxygen-15 (¹⁵O), carbon-11 (¹¹C) and fluorine-18 (¹⁸F). These radionuclides are isotopes of elements that are often naturally present in organic molecules as well as in chemically produced molecules, e.g., anticancer drugs. After replacing one of the molecules’ atoms by its radioactive isotope, the molecular structure is unchanged, leaving chemical properties unaffected. After intravenous injection of a PET tracer, the radiolabeled molecules can be located within the body by detecting the emitted photons. Since only a small amount of radiotracer is required for PET imaging, it is assumed that the radiotracer does not affect the dynamic process under study.

PET is based on the detection of positron emission. During radioactive decay, the radionuclide, e.g., ¹⁸F, emits a positron which, after traveling a short distance (few mm) in tissue, annihilates with a nearby electron to emit two 511 keV photons in opposite directions. These two “annihilation” photons are registered by the PET scanner using a coincidence detection circuitry, providing 3D information of the tracer distribution with high sensitivity and resolution. To achieve quantitative accuracy, imaging data needs to be corrected for attenuation: when emitted from tissues deeper in the body, photons are more likely to be absorbed than from superficial structures. As a result, 3D images would falsely show low tracer concentrations in deeper structures compared to superficial structures. In PET, the attenuation perceived by the annihilation photon pairs, traveling in opposite directions over a line through the body, is mathematically equivalent to the attenuation perceived by one photon transmitted through the body over that same line. Therefore, accurate attenuation correction can be achieved using a transmission source, e.g., computed tomography (CT). In addition, PET/CT systems can correct for false detections due to random coincidence detection or scattered annihilation photons. As a result, PET provides radioactivity measurements with high quantitative accuracy[33].

Quantification of tracer uptake, however, remains challenging. First, the measured radioactivity concentration in tissue depends on the tracer concentration in blood over time, which, in turn, depends on the injected dose and distribution volume. The standardized uptake value (SUV) takes this variability into account, as the radioactivity concentration in tissue is normalized by the ratio of the injected dose to patient weight. Second, the PET signal does not necessarily reflect specific uptake, e.g., trapping of the tracer by the process of interest. A tracer could also be free in tissue, trapped by a different process or reside in blood vessels within the region of interest, e.g., tumor (Figure 1). Pharmacokinetic modeling can be applied to distinguish between the various kinetic processes and separates the total signal into these components[34].

Figure 1 Graphical representation of the different components that determine the total positron emission tomography signal. Examples 1 and 2 can represent either different patients, different lesions in one patient or different scans of one patient, for example before and after therapy. In both examples the contributions of specific uptake (the uptake of interest) are equal, but the total signal is different due to differences in contribution of other (non-relevant) signals. Measured standardized uptake value (SUV) values are reflected by the dashed lines. As SUV does not only reflect the specific signal, its use should be validated before it is used in a clinical setting, i.e., it is required to assess if contributions from non-specific signals affect SUV values in a non-predictable way. For the purpose of illustration, the Y-axis represents SUV values on an arbitrarily chosen scale.

In addition to spatial information, temporal information of the tracers’ distribution is used in pharmacokinetic modeling. To obtain information on the changes in tracer activity concentrations over time (time activity curves or TAC), sequential PET images are acquired over the same body area. In addition, accurate temporal data on tracer concentration in plasma is obtained from arterial blood sampling and dedicated lab analysis. Mathematical models (“compartment models”) are then used to extract measures of the relevant components of the tracers’ kinetics, such as specific uptake or binding. As absolute quantification by kinetic modeling can be challenging and cumbersome in the clinic, alternatives have been introduced to measure tracer uptake. Before clinical implementation, these “simplified parameters” (such as SUV) should be validated and correlated with parameters from pharmacokinetic modeling.

To date, 2-deoxy-2-[¹⁸F]fluoro-d-glucose ([¹⁸F]FDG) is the most commonly used PET tracer. As [¹⁸F]FDG is a glucose analogue, it accumulates in malignant tumors with high glucose consumption. As a result, [¹⁸F]FDG PET is extensively used for diagnosis, staging and response monitoring of cancer. Currently, [¹⁸F]FDG PET is routinely performed for initial staging[35] and pre-operative staging[36,37] of patients with NSCLC. As tumor hypoxia is associated with increased glycolysis, it is conceivable that hypoxia is associated with increased [¹⁸F]FDG uptake. However, results on [¹⁸F]FDG to assess tumor hypoxia have been conflicting[38], indicating that [¹⁸F]FDG is not specific enough to identify hypoxia. Therefore, other PET tracers have been developed to measure hypoxia and perfusion in tumors more specifically. In the following paragraphs, these PET tracers will be discussed.

TUMOR HYPOXIA IMAGING

Clinical relevance

Tumor hypoxia is associated with resistance to both radiation therapy[21] and chemotherapy[22]. Radiation therapy requires oxygen to induce DNA damage and hypoxic cancer cells are three times less sensitive to radiation therapy than normoxic cancer cells[39,40]. In addition, the resistance to anticancer drugs is attributed to the lack of O₂ available for drug activation, the increased genetic instability, the antiproliferative effects of hypoxia, and the increased gene transcription induced by HIF-1[41,42]. Currently, drugs that selectively target tumor hypoxia and its increased gene transcription are still under study and have entered the first clinical trials[43-45]. Since tumor hypoxia may affect clinical outcome, hypoxia imaging may be useful to determine prognosis and tumor response in lung cancer patients. Furthermore, hypoxia assessment may help to optimize treatment strategies in individual patients.

In particular, the efficacy of radiation therapy may be increased by several interventions. First, the systemic oxygenation level can be increased by hyperbaric chamber treatment[46], carbogen breathing[47] and improved oxygen transport by hemoglobin. For the latter, blood transfusions and erythropoietin injections are available[48]. Oxygen transport can be further improved by agents that improve perfusion and affect vascular permeability[49]. Second, the apparent oxygenation level in tumors can be increased using radiosensitizers, which are usually based on a nitroimidazole-group and specifically target hypoxic tumor cells (pO₂ < 10 mmHg). Once incorporated in hypoxic tumor cells, radiosensitizers mimic oxygen, thereby increasing the efficacy of radiation therapy[50]. Third, the radiation therapy plan can be adjusted to increase the dose administered to hypoxic tumor tissue. This can be achieved by dose boosting to the whole tumor, dose painting, or dose painting by numbers[51]. For dose boosting, an increased dose is administered to hypoxic areas, thereby increasing the radiation dose to normal tissue and, potentially, its associated side effects. For dose painting, the dose to a specific area (e.g., hypoxic area) is increased, whereas the radiation to the remaining part of the tumor can be either maintained or decreased. In the latter case the total dose level can be maintained. Dose painting can be further refined when it is directly based on the voxel-by-voxel values of a PET image (referred to as “dose painting by numbers”). For successful implementation of the previous mentioned radiation therapy strategies, hypoxia imaging may help to identify hypoxic tumors, prevent unnecessary side effects in patients with normoxic tumors, and reveal heterogeneous distribution of hypoxia within tumors. Figure 2 summarizes the potential applications of hypoxia imaging for radiation therapy.

Figure 2 Radiation therapy treatment strategies for tumor hypoxia.

Characteristics of a hypoxia PET tracer

The ideal hypoxia tracer would freely and rapidly diffuse to tissue, including remote areas. For optimal contrast of the PET image, accumulation of the tracer should be high in hypoxic cells, whereas no binding should occur in normoxic cells. To achieve the best image quality, an optimal balance between tracer half-life, accumulation rates and clearance rates is required: the tracers’ half-life should be long enough to obtain a high signal-to-noise ratio whilst allowing the tracer enough time to diffuse and bind to hypoxic cells and clear from normoxic tissues and blood. Accumulation and clearance rates are influenced by the tracers’ octanol/water partition coefficient. More lipophilic compounds may more readily pass through the cell membrane. On the other hand, more hydrophilic compounds may more easily diffuse across tissues and show faster clearance from blood and normoxic tumors through the urinary pathway[30,52]. Besides these hypoxia specific characteristics, the tracer should be metabolically inert, since the formation of radiolabeled metabolites results in a decreased amount of the original tracer available for hypoxia specific uptake, poor image contrast and inaccurate tracer quantification.

For clinical implementation, hypoxia tracers require fast kinetics, allowing for rapid accumulation in hypoxic tissues, thereby limiting the time between tracer injection and imaging. In addition, simplified and reproducible methods (e.g., SUV) are needed to quantify tracer uptake.

Hypoxia tracers for PET

Over the last decades, several PET tracers have been developed to measure tumor hypoxia. To identify all relevant hypoxia tracers in lung cancer, a literature search was conducted in PubMed to identify studies published before 1 January 2014. To this end, PET specific search terms (PET, positron emission tomography) were combined with hypoxia specific search terms (hypoxia, anoxia), and/or lung cancer specific search terms (lung cancer, lung neoplasms, non-small cell lung cancer, small cell lung cancer), and/or kinetic modeling specific search terms (kinetic modeling, modeling), and/or radiation therapy specific search terms (radiation therapy, radiation). For these search terms, the corresponding Mesh terms were included. Thereafter, the obtained English abstracts were evaluated for relevance. Based on the obtained publications, a specific search strategy was subsequently performed for each identified hypoxia PET tracer. Additional publications were identified by cross-referencing. Brain studies were excluded since the blood-brain barrier may affect tracer kinetics. Figure 3 and Table 1 give an overview of the identified hypoxia tracers that have been evaluated in oncology. The tracer names and abbreviations are displayed in Table 2. These hypoxia tracers can be subdivided in nitroimidazole-based and thiosemicarbazone-based tracers. In the following paragraphs, these tracers and their potential applications in lung cancer patients will be discussed.

Table 1 Evaluated hypoxia tracers in oncology.

Tracer1	Half-life	Validation studies2,3					Evaluated in clinical oncology3
		BC	Probe	Ex-M	En-M	FMISO	Lung cancer	Other cancer types
[*Cu]ATSM	[60Cu]: 23.7 min [61Cu]: 3.3 h [62Cu]: 9.7 min [64Cu]: 12.7 h	Lewis et al[133], Yuan et al[134], Kersemans et al[135], Matsumoto et al[136]	Ballegeer et al[137], Bowen et al[138], Lewis et al[133], Myerson et al[139], O'Donoghue et al[72]	Ballegeer et al[137], Hansen et al[140], Matsumoto et al[136], McCall et al[66], O'Donoghue et al[72], Oh et al[74], Yuan et al[134]	Grigsby et al[141], Tateishi et al[142], Valtorta et al[143], Weeks et al[144]	Dence et al[145], Kersemans et al[135], Lewis et al[146], Matsumoto et al[136], O'Donoghue et al[72]	Dehdashti et al[86], Lohith et al[67], Wong et al[80], Zhang et al[81]	Chao et al[147], Dehdashti et al[148], Dehdashti et al[149], Dietz et al[150], Grassi et al[151], Grigsby et al[141], Kositwattanarerk et al[152], Laforest et al[153], Lewis et al[154], Minagawa et al[155], Nyflot et al[156]
[18F]EF1	110 min	NA	NA	Evans et al[157]	NA	NA	NA	NA
[18F]EF3		Mahy et al[158], Mahy et al[159]	NA	Mahy et al[159]	NA	Mahy et al[158], Dubois et al[160]	NA	Mahy et al[161]
[18F]EF5		NA	NA	Yapp et al[162], Ziemer et al[163]	NA	NA	NA	Komar et al[164]
[18F]FAZA [18F]FETA		Reischl et al[165], Piert et al[166], Picchio et al[167], Maier et al[168], Tran et al[169] Barthel et al[187]	Busk et al[170], Mortensen et al[171], Piert et al[166], Tran et al[169] Barthel et al[187]	Busk et al[170], Busk et al[172], Busk et al[173], Busk et al[174], Busk et al[175], Graves et al[176], Maier et al[168], Solomon et al[177] NA	Belloli et al[178], Picchio et al[167], Troost et al[179], Valtorta et al[143] NA	Sorger et al[57], Piert et al[166], Reischl et al[165] Rasey et al[188]	Postema et al[180], Bollineni et al[181], Trinkaus et al[58], Verwer et al[59] NA	Grosu et al[182], Souvatzoglou et al[183], Schuetz et al[184], Shi et al[106], Mortensen et al[185], Havelund et al[186] NA
[18F]FETNIM		Grönroos et al[189]	Yang et al[60]	NA	Hu et al[85]	Grönroos et al[189], Tolvanen et al[190], Yang et al[60]	Hu et al[85], Li et al[84]	Lehtiö et al[99], Lehtiö et al[191], Lehtiö et al[192], Vercellino et al[193], Yue et al[194]
[18F]FMISO		Bentzen et al[195], Rasey et al[196], Bentzen et al[197], Troost et al[198]	Bentzen et al[125], Gagel et al[126], Bruehlmeier et al[199], Lawrentschuk et al[200], O'Donoghue et al[72], Piert et al[166], Sørensen et al[201], Gagel et al[202], Carlin et al[203], Chang et al[204], Mortensen et al[127]	Hatano et al[205], Huang et al[206], Huang et al[207], Oehler et al[208], Cho et al[209], Troost et al[210], Matsumoto et al[136], Troost et al[198], Dubois et al[211]	Dubois et al[211], Cherk et al[212], Riesterer et al[213], Lehmann et al[214], Chen et al[215], Campanile et al[216], Cheng et al[217], Sato et al[218], Norikane et al[219]		Cherk et al[212], Eschmann et al[77], Gagel et al[82], Koh et al[87], Rasey et al[220], Vera et al[83]	Koh et al[221], Liu et al[222], Yeh et al[223], Rischin et al[224], Bentzen et al[125], Rajendran et al[225], Gagel et al[126], Rajendran et al[226], Hicks et al[227], Lawrentschuk et al[200], Loi et al[228], Thorwarth et al[229], Rajendran et al[230], Rischin et al[231], Thorwarth et al[232], Zimny et al[124], Eschmann et al[233], Gagel et al[202], Thorwarth et al[234], Thorwarth et al[235], Lee et al[236], Lin et al[237], Nehmeh et al[238], Roels et al[239], Dirix et al[240], Lee et al[241], Mortensen et al[127], Wang et al[56], Abolmaali et al[242], Bowen et al[138], Eary et al[243], Kikuchi et al[244], Hugonnet et al[131], Yamane et al[245], Mammar et al[246], Zips et al[247], Cheng et al[217], Henriques de Figueiredo et al[248], Okamoto et al[249], Sato et al[218], Segard et al[250], Tachibana et al[251], Norikane et al[219]
[18F]FRP170		NA	NA	NA	NA	NA	Kaneta et al[252]	NA
[18F]FPIMO		NA	NA	Busk et al[174]	NA	NA	NA	NA
[18F]HX4		Dubois et al[61]	NA	Dubois et al[61]	Chen et al[215]	Chen et al[215]	van Loon et al[253], Zegers et al[78]	Chen et al[215]
[124I]IAZA	4.2 d	Reischl et al[165]	NA	NA	NA	Reischl et al[165]	NA	NA
[124I]IAZG		NA	Zanzonico et al[254]	NA	NA	Riedl et al[255], Zanzonico et al[254]	NA	NA

¹Refer to Table 2 for full tracer names;

²Preclinical and clinical studies comparing the uptake of the hypoxia tracer under study with other hypoxia markers;

³Excluding brain studies. BC: Induced hypoxia by breathing conditions; Probe: Polarograpic electrode; Ex-M: Exogenous hypoxia marker (pimonidazole, EF3 or EF5); En-M: Endogenous hypoxia marker (HIF-1, CA IX); FMISO: [18F]FMISO PET; NA: Not available.

Table 2 Hypoxia positron emission tomography tracer abbreviations.

Abbreviation	Full name or chemical name
[*Cu]ATSM	[*Cu]-diacetyl-bis(N4-methylthiosemicarbazone)
[18F]EF1	2-(2-Nitroimidazol-1H-yl)-N-(3-[18F]fluoropropyl)acetamide
[18F]EF3	2-(2-Nitroimidazol-1H-yl)-N-(3,3,3-[18F]trifluoropropyl)acetamide
[18F]EF5	2-(2-nitro-1H-imidazol-1-yl)-N-(2,2,3,3,3-[18F]-pentafluoropropyl)-acetamide
[18F]FAZA	[18F]fluoroazomycin arabinoside
[18F]FETA	[18F]fluoroetanidazole
[18F]FETNIM	[18F]fluoroerythronitroimidazole
[18F]FMISO	[18F]fluoromisonidazole
[18F]FRP170	1-(2-[18F]fluoro-1-[hydroxymethyl]ethoxy)methyl-2-nitroimidazole
[18F]FPIMO	[18F]pimonidazole
[18F]HX4	3-[18F]fluoro-2-(4-((2-nitro-1H-imidazol-1-yl)methyl)-1H-1,2,3,-triazol-1-yl)-propan-1-ol
[124I]IAZA	[124I]iodazomycin arabinoside
[124I]IAZG	[124I]iodazomycin galactoside

Figure 3 Timeline for development and evaluation of hypoxia specific tracers that have been evaluated by preclinical or clinical positron emission tomography. Development and in-vitro analysis (blank), preclinical positron emission tomography (PET) evaluation (grey), and clinical PET evaluation (black). The number of published clinical studies in oncology is indicated (excluding brain studies). See Table 2 for full names.

Nitroimidazole-based tracers: Originally, nitroimidazoles have been developed as radiosensitizers. Already in 1984, Chapman[53] have proposed nitroimidazoles for hypoxia imaging. Upon entering the cell, nitroimidazole undergoes electron reduction, thereby becoming a radical. In normoxic cells, this reaction is reversed by O₂. In hypoxic cells, the radical can react with an intracellular macromolecule instead and remains trapped. As the latter process occurs at pO₂ < 10 mmHg, an oxygenation level associated with increased radiation therapy resistance, nitroimidazoles are able to detect clinically relevant hypoxia[54].

Among the developed hypoxia tracers for PET (see Figure 3), [¹⁸F]FMISO has been investigated most extensively. Although [¹⁸F]FMISO showed rapid metabolism in mice studies, it appeared to be a robust hypoxia tracer in humans, with parent fractions up to 96% at 90 min after injection[55]. Since [¹⁸F]FMISO is rather lipophilic with a partition coefficient (log P) of 0.4, clearance from blood and normoxic tissues is slow. Therefore, the required time intervals between injection and imaging are long, at least 3 h[56]. Efforts have been made to develop hypoxia tracers with more favorable characteristics. Being the most evaluated and validated hypoxia tracer to date, the performance of new hypoxia tracers is often compared with [¹⁸F]FMISO (see Table 1). Among these tracers, [¹⁸F]FAZA has been introduced in the clinic. [¹⁸F]FAZA (log P = 0.04) is more hydrophilic than [¹⁸F]FMISO and shows faster clearance from blood and normoxic tissues[57]. This allows for a shorter time interval between injection and imaging[58]. In addition, [¹⁸F]FAZA has a high parent fraction during imaging, accounting for a parent fraction of 90% at 70 min after injection[59]. Other more hydrophilic nitroimidazole tracers include [¹⁸F]FETNIM and [¹⁸F]HX4, which have a partition coefficient (log P) of 0.17[60] and -0.69[61], respectively. An example of a more lipophilic tracer is [¹⁸F]EF5, which is the ¹⁸F-labelled version of exogenous hypoxia marker EF5, with a partition coefficient (log P) of 0.6.

Thiosemicarbazone-based tracers: Thiosemicarbazone-based tracers represent another subgroup of hypoxia tracers for PET. Thiosemicarbazones possess a strong antitumor activity, particularly when coupled with a metal ion like copper (Cu)[62]. [Cu]ATSM is a therapeutic agent which, by replacing the Cu atom with a suitable radioactive Cu isotope, can be used for hypoxia PET imaging[63]. In nuclear medicine, Cu is of particular interest for its favorable radiochemical properties. First, Cu is relatively easy to incorporate in molecules and has multiple radioactive isotopes suitable for PET imaging. Second, with half lives ranging from 24 min to 13h for ⁶⁰Cu and ⁶⁴Cu, respectively, Cu has several potential applications. The short-lived radionuclides can be used for sequential measurements, whereas radionuclides with longer half lives do not require a cyclotron on-site and are more suitable for the clinical setting. Remarkably, ⁶⁴Cu can also be applied as radiation therapy agent, since it also emits a β^- particle (40% yield)[64,65]. In oncology, [Cu]ATSM has been evaluated both preclinically and clinically. This tracer shows favorable kinetics with rapid uptake in hypoxic tissue and fast clearance from normoxic tissues, enabling imaging within 30 min after injection[66,67]. However, the exact uptake mechanism of [Cu]ATSM is still under debate[63,68,69] and several preclinical studies have shown that [Cu]ATSM uptake depends on tumor type and other characteristics than hypoxia alone[70-76].

Clinical evaluation of hypoxia PET tracers in lung cancer

Hypoxia PET imaging is in development and most clinical studies have been focused on notoriously hypoxic cancer types such as cervical cancer and head and neck cancer. Nevertheless, several clinical PET studies have evaluated hypoxia imaging in lung cancer (Table 3). In the following paragraphs, data acquisition, quantification and clinical observations of these hypoxia tracers will be discussed.

Table 3 Hypoxia tracer studies in lung cancer.

Tracer1	Year	Authors	N2	Stage	Time3	Duration4	Measure5	Therapy6
[60Cu]ATSM	2003	Dehdashti et al[86]	18	I-IV	30 min	30 min	T/M	Radiation
								Chemoradiation, chemotherapy
[62Cu]ATSM	2000	Takahashi et al[79]	6	NA	10 min	10 min	T/B
	2008	Wong et al[80]	2	NA	15 min	5 min	SUV
	2009	Lohith et al[67]	13	I-IV	10 min	10 min	SUVmean
	2013	Zhang et al[81]	5	I-IV	15 min	5 min	SUVhypoxia/perfusion7
[18F]FAZA	2009	Postema et al[180]	13	NA	2-3 h	3-4 min	T/Bg
	2013	Trinkaus et al[58]	11	III	4 h	30 min	T/Bg	Chemoradiation
	2013	Bollineni et al[181]	11	III-IV	2 h	NA	T/Bg
	2013	Verwer et al[59]	9	NA	0 h	70 min8	VT9
[18F]FETNIM	2010	Li et al[84]	26	III	2 h	20 min	T/Bmax	Radiation
								Chemotherapy
	2013	Hu et al[85]	25	II	2 h	NA	T/Me	Chemotherapy
[18F]FMISO	1995	Koh et al[87]	14	III	2 h	40 min	T/B	Radiation
	1996	Rasey et al[220]	21	III-IV	2 h	40 min	T/B
	2005	Eschmann et al[77]	8	III-IV	4 h	NA	T/Me	Radiation
	2006	Cherk et al[212]	21	I-II	2 h	NA	SUVmax
	2006	Gagel et al[82]	8	III-IV	3 h	30 min	SUV, T/M	Chemoradiation
	2011	Vera et al[83]	7	III	3 h	NA	SUVmax	Chemoradiation Chemotherapy
[18F]FRP170	2007	Kaneta et al[252]	3	NA	0 h	60 min8	SUV TAC
[18F]HX4	2010	van Loon et al[253]	4	IV	2 h	NA	T/B
	2013	Zegers et al[78]	15	II-IV	4 h	30 min	T/B

¹Refer to Table 2 for full tracer names;

²Number of lung cancer patients (evaluable scans);

³Start time after injection of the positron emission tomography (PET) frame that was used for quantification;

⁴Duration of the PET frame that was used for quantification;

⁵(Semi)quantitative measure used for evaluation;

⁶Evaluated therapy;

⁷Hypoxia marker uptake normalized to perfusion marker uptake: mean (SUV[Cu]ATSM/SUV[Cu]PTSM);

⁸Dynamic PET data used for quantification;

⁹Volume of distribution derived from full pharmacokinetic modeling. NA: Data not available; SUV: Standardized uptake value; T/M: Tumor-to muscle ratio; T/Me: Tumor-to-mediastinum ratio; T/B: Tumor-to-blood ratio; T/Bg: Tumor-to-background ratio; VT: Volume of distribution; TAC: Time activity curve.

Data acquisition and analysis: Nitroimidazole based tracers require relatively long time intervals for accumulation in hypoxic cells and clearance from normoxic cells. In concordance, most studies used images > 2 h after injection for hypoxia assessment. The length of the time interval between injection and imaging may affect the tracers’ distribution pattern in tumors. For example, it has been shown that the distribution of [¹⁸F]FMISO at 2 h is significantly different from the distribution at 4 h, whereas only the 4 h data are predictive of tumor recurrence[77]. In contrast, the distribution of [¹⁸F]HX4 was similar at 2 h and 4 h[78]. Compared to nitroimidazole based tracers, [Cu]ATSM shows fast kinetics and images were acquired after time intervals as short as 10 min after injection[67,79-81].

Quantification of hypoxia: To identify hypoxia in tumor tissue, several simplified parameters have been used, including tumor-to-blood ratio, tumor-to-background ratio, tumor-to-muscle ratio, tumor-to-mediastinum ratio, and SUV. In addition, several studies have used dynamic PET scans to investigate the tracers’ kinetics in more detail, for example by using pharmacokinetic modeling for quantification[59]. Furthermore, consecutive imaging using multiple tracers has been performed to facilitate the identification and quantification of hypoxia. For example, consecutive PET scans have been performed with hypoxia tracer [Cu]ATSM and perfusion tracer copper-pyruvaldehyde-bis(N4-methylthiosemicarbazone ([Cu]PTSM). Here, the ratio of [Cu]ATSM SUV to [Cu]PTSM SUV has been used as a measure of hypoxia[81].

To date, it is not known which measure and threshold accurately reflects pO₂ levels in tumors. As repeated measurements with a polarographic electrode are not feasible in lung cancer, a more pragmatic approach is required. The clinical relevance of a threshold can be determined by clinical parameters like tumor response, progression-free survival and overall survival.

Determination of clinically relevant hypoxia: Among the clinical studies on hypoxia PET tracers in lung cancer, most studies have only evaluated tumor hypoxia prior to treatment. For [¹⁸F]FMISO, a pretreatment threshold of > 2 for tumor-to-mediastinum ratio was associated with poor outcome after radiation therapy. However, the shape of the time activity curve appeared to be a better predictor of response[77]. In contrast to these results, other authors did not find a predictive value for [¹⁸F]FMISO after chemoradiation[82,83]. In other patients treated with chemoradiation, a threshold of tumor-to-blood ratio > 1.9 for [¹⁸F]FETNIM[84,85] and > 3.0 for [Cu]ATSM[86] was associated with poor overall survival and tumor response, respectively. In addition, a number of studies have evaluated the changes in hypoxia tracer uptake during therapy. While hypoxic cells are considered to be more resistant to radiation therapy, most studies in lung cancer reported a decrease in hypoxia tracer uptake after radiation therapy[58,82,87].

PET FOR TUMOR PERFUSION MEASUREMENTS

Tumor angiogenesis

Blood flow is not only required for the delivery of PET tracers and anticancer drugs to tumors, but also for the transport of nutrients, e.g., glucose, and oxygen. Under hypoxic conditions in tumors, the HIF protein is usually upregulated. Activated HIF translocates to the nucleus of tumor cells and results in transcription of a large repertoire of genes including VEGF[88,89]. VEGF is a potent protein and plays a key role in tumor angiogenesis, which is the formation of new blood vessels. This tumor angiogenesis is essential for tumor growth, metastatic spread and survival of tumor cells. As a result, VEGF signaling has become an important therapeutic target for the treatment of malignant tumors. To date, several antiangiogenic drugs have been developed including monoclonal antibodies that bind circulating VEGF (e.g., bevacizumab[90]) and tyrosine kinase inhibitors that target the intracellular domain of the VEGF receptors (e.g., sunitinib and sorafenib[91]). Among the currently available antiangiogenic drugs, bevacizumab has been registered for the treatment of patients with NSCLC. In combination with paclitaxel and carboplatin, bevacizumab has been approved for first-line treatment of non-squamous NSCLC[15]. As tumor vascularization is an important factor in the biology of malignant tumors, and antiangiogenic strategies have been introduced for the treatment of lung cancer, imaging techniques are increasingly used for perfusion measurements in lung cancer.

Imaging of tumor perfusion

PET is a sensitive technique to quantify tumor perfusion[92]. To this end, perfusion tracers like rubidium-82 (⁸²Rb[93]), radioactive ammonia ([¹³N]NH₃[94]), radioactive water ([¹⁵O]H₂O[95-101]) can be administered. Other PET tracers such as [⁶⁸Ga]transferrin[102] and [¹¹C]methylalbumin[103] are available to assess vascular permeability. Currently, experience with perfusion PET tracers is rather limited in oncology, except for [¹⁵O]H₂O. In particular, previous PET studies have shown that quantification of tumor perfusion using [¹⁵O]H₂O is feasible in patients with lung cancer [97,104,105].

[¹⁵O]H₂O PET

As [¹⁵O]H₂O is a freely diffusible tracer with near 100% extraction over a wide perfusion range (0-6 mL/min per mL), its kinetics directly reflect tumor perfusion. As a result, [¹⁵O]H₂O is an ideal tracer for quantitative perfusion imaging. The short half-life of ¹⁵O, which is 2.03 min, enables sequential PET scans using both [¹⁵O]H₂O and another tracer, e.g., [¹⁸F]FDG[97] or a hypoxia tracer[106]. However, it requires the presence of a nearby cyclotron. Because [¹⁵O]H₂O is metabolically inert and is not retained in cells, quantification using SUV, which is a parameter for quantification of irreversible uptake, is not possible. Instead, pharmacokinetic modeling, using short (< 10 min) dynamic PET scans, is required to quantify tumor perfusion.

Monitoring tumor perfusion during treatment

Currently, [¹⁵O]H₂O PET scans are increasingly used to assess response of the tumor vasculature to antiangiogenic therapy[107-110]. As [¹⁵O]H₂O PET has shown high reproducibility in lung cancer[105], it can be applied for response monitoring during treatment. de Langen et al[111] have investigated changes in tumor perfusion in 44 NSCLC patients who were treated with bevacizumab and erlotinib. Three weeks after the start of treatment, a mean decrease of 11% in tumor perfusion was measured using [¹⁵O]H₂O PET[111]. A significant reduction in tumor perfusion was measured in patients with a partial response according to the response evaluation criteria in solid tumors (RECIST[112]). More importantly, patients with > 20% reduction in tumor perfusion had an improved progression-free survival as compared to other patients (12.5 mo vs 2.9 mo). The latter findings indicate that [¹⁵O]H₂O PET may have predictive value in lung cancer patients who are treated with antiangiogenic drugs. For early prediction of tumor response, early perfusion measurements may be useful, as the effects of antiangiogenic can be very rapid[113].

Tumor perfusion and drug delivery

As the short half-life of ¹⁵O enables sequential PET scans using both [¹⁵O]H₂O and an additional tracer, [¹⁵O]H₂O PET is a useful tool to investigate drug delivery of radiolabeled anticancer agents by correlating uptake of radiolabeled drugs with [¹⁵O]H₂O perfusion data[114,115]. Apparently, it has been shown that tumor perfusion is an important determinant of drug tumor exposure, as indicated by several PET studies on [¹⁸F]5-fluorouracil(FU)[116-118], [¹¹C]DACA[119], and [¹¹C]docetaxel[120,121]. Consequently, tumor perfusion may be predictive of tumor response to the above mentioned anticancer drugs. These findings advocate further studies investigating the predictive value of tumor perfusion for tumor response to chemotherapy. As tumor perfusion is the key factor for the uptake of several anticancer drugs in tumors[122], antiangiogenic drugs may affect drug exposure in tumors. To investigate this concept, a PET study has been performed in NSCLC patients using both [¹⁵O]H₂O and the radiolabeled taxane [¹¹C]docetaxel[113]. In that study, bevacizumab reduced both perfusion and net influx rate of [¹¹C]docetaxel within 5 h. These rapid effects persisted after 4 d and were not associated with significant changes in tumor heterogeneity. The mentioned studies indicate that [¹⁵O]H₂O PET may reveal the role of perfusion in drug delivery and antiangiogenic therapy in malignant tumors[123].

IMAGING HYPOXIA AND PERFUSION

It is conceivable that the development of tumor hypoxia is associated with a decrease in tumor perfusion. This may complicate PET imaging, as tracer delivery will be reduced in these areas. Although the uptake of the ideal hypoxia tracer is not directly related to perfusion, lack of perfusion will limit tracer delivery.

Diffusion-limited hypoxia is present in tumor cells located away from capillaries, i.e., further than the diffusion distance of oxygen. As perfusion is relatively low in these areas, tracer delivery may be limited and this may, in turn, affect uptake of hypoxia tracers. In addition, low perfused areas can become necrotic. The PET signal will be decreased in areas containing necrosis even though these areas may also contain highly hypoxic cells. In Figure 4, these hypothetical considerations are summarized. The figure also illustrates the limitations of using a predefined threshold to delineate hypoxic areas on a PET image, as areas likely to contain the most severely hypoxic cells will be missed. The mentioned considerations may explain the conflicting results between the uptake of hypoxia tracers and the direct assessment of tissue oxygenation using polarographic electrodes[124-127]. [¹⁵O]H₂O PET may help to understand these conflicting results and may identify the remote, low perfused areas. An example of images obtained from consecutive perfusion and hypoxia PET imaging is displayed in Figure 5.

Figure 4 Representation of hypothetical considerations on the link between perfusion and the hypoxia signal as measured by imaging positron emission tomography. The continuous curve represents hypothetical level of hypoxia in tissue for increasing levels of perfusion (i.e., closer to the capillaries). The dotted line represents the positron emission tomography signal obtained from hypoxia imaging using an optimal imaging protocol. The horizontal line represents a threshold used for delineation of hypoxic areas.

Figure 5 Example of consecutive perfusion and hypoxia positron emission tomography in a patient with non-small cell lung cancer. A: Low dose computed tomography; B: Perfusion image (averaged image acquired over time interval 30-120 s after injection of 370 MBq [¹⁵O]H₂O); C: Hypoxia image (averaged image over time interval 40-70 min after injection of 185 MBq [¹⁸F]FAZA).

Acute hypoxia is directly caused by a (temporary) lack of tumor perfusion. Since acute hypoxia is presumed to be transient or even cycling[128], hypoxia tracer uptake may not accurately reflect this type of hypoxia. [¹⁵O]H₂O PET may help to study the effect of acute hypoxia and its relation with hypoxia tracer uptake.

Besides the previous considerations for combining [¹⁵O]H₂O perfusion PET imaging with hypoxia tracer PET imaging, the combination may provide further insight into the effects of treatment. Jain has previously proposed that antiangiogenic therapy may normalize the abnormal tumor vasculature, thereby decreasing tumor hypoxia and improving drug delivery of cytotoxic agents[129,130]. This is underscored by the fact that a decrease in [¹⁸F]FMISO uptake has been measured in renal cell cancer after treatment with sunitinib[131]. On the other hand, an increase in [¹⁸F]FMISO uptake after sorafenib[132] and a rapid decrease in tumor perfusion after bevacizumab have been reported as well[113]. The latter findings suggest that antiangiogenic therapy may decrease tumor perfusion and subsequently the delivery of hypoxia tracers to tumors. To further clarify these findings, future PET studies need to combine hypoxia tracers with [¹⁵O]H₂O at different time points after drug administration.

FUTURE PERSPECTIVES

In the present review, the currently available tracers for PET imaging of hypoxia and perfusion in lung cancer patients were discussed. Considering the currently available studies, PET seems feasible to assess hypoxia and perfusion in lung cancer. In contrast to traditional probe measurements, PET hypoxia imaging is non-invasive and provides information on the heterogeneous distribution of hypoxia in tumors. In addition, whole body PET scans using a hypoxia tracer can reveal hypoxic areas not only in primary tumors, but in metastases as well. To date, several tracers have been developed to measure tumor hypoxia, whereas tumor perfusion has been mostly quantified using [¹⁵O]H₂O. While acquisition and quantification of [¹⁵O]H₂O data is rather straightforward, several challenges remain for PET hypoxia tracers.

Although several hypoxia tracers have been developed and evaluated in the clinical setting, no consensus has yet been reached on the most feasible tracer, the optimal timing of acquisition, and the most accurate quantification method. In lung cancer, the studies on hypoxia PET tracers are preliminary and include a limited number of patients. Ideally, the clinical impact of hypoxia imaging would be evaluated in large clinical trials validating hypoxia tracers for prediction of tumor response and survival. In addition, clinical trials are needed to reveal the clinical value of hypoxia tracers for advanced radiation therapy strategies such as dose painting. In NSCLC, several trials are currently recruiting patients for [¹⁸F]FMISO based (NCT01576796) and [¹⁸F]FDG based (NCT01024829) dose boosting.

In patients with lung cancer, quantification of tracer uptake can be challenging due to tumor movement during respiration. As PET acquisition usually takes 10min to 1 h, patient motion during PET imaging is unavoidable and the acquired image of the lung tumor will be blurred, which complicates accurate delineation of hypoxic areas. As a result, these images are less suitable for dose painting techniques, especially for dose painting by numbers. For PET imaging, respiratory gated imaging (4D imaging) is currently under study. In respiratory gated imaging, patient motion is continuously monitored during acquisition. As a result, PET data can either be corrected for the registered motion or PET data from a specific interval of the respiratory cycle can be used for reconstruction. As similar techniques are also under study for radiation therapy, dose painting strategies may be further improved by combining 4D PET hypoxia imaging with 4D radiation therapy.

Since the introduction of antiangiogenic drugs, perfusion measurements have been increasingly applied in the clinic. [¹⁵O]H₂O PET provides quantification of tumor perfusion and may be useful for response monitoring during antiangiogenic therapy. Further studies are needed to evaluate the predictive value of tumor perfusion for tumor response to anti-cancer drugs. In addition, tumor perfusion may not only affect the delivery of drugs to tumors, but also the delivery of PET tracers such as hypoxia tracers.

In conclusion, PET using both [¹⁵O]H₂O and a hypoxia tracer is a promising method to further understand the development of hypoxia in lung cancer. As previously mentioned, these PET scans are promising for response monitoring of radiation therapy and antiangiogenic drugs. In addition, hypoxia tracers may be useful to select patients for treatment with radiosensitizers (e.g., nimorazole, NCT01733823) and realize a more precise radiation plan including dose boosting and dose painting. As the available PET studies on hypoxia and perfusion are rather preliminary in patients with lung cancer, further studies are needed for validation and clinical implementation in this patient population.

ACKNOWLEDGMENTS

This review was written within the framework of CTMM, the Center for Translational Molecular Medicine.