¹⁸F-FMISO is a commonly used hypoxia tracer in preclinical and clinical studies. Due to its lipophilicity, this molecule easily crosses the cell membranes and is then trapped if intracellular hypoxia remains below a threshold of 10 mmHg. The cellular clearance of ¹⁸F-FMISO is quite low in normoxic tissues, thereby hampering the contrast between normoxic tissues and moderate hypoxic tumor tissues. As a result, a TBR ratio of 1.2 is usually used to delineate regions of hypoxia after a minimum delay of 2 h (20, 21). The best signal-to-noise ratio has been observed 4 h after tracer injection (22). In a study of Mortensen et al. (23), it was not possible to correlate ¹⁸F-FMISO with Eppendorf electrodes in the clinical setting in head and neck tumors (23).

In preclinical and clinical studies, the level of hypoxia highlighted by ¹⁸F-FMISO has been correlated with the response to therapy and outcome (24). Non-hypoxic volume estimated using ¹⁸F-FMISO uptake showed significantly better local control after single-dose irradiation than hypoxic tumors in FaDu hSCC xenografts (25). In a recent review on PET imaging, Fleming and colleagues listed all the applications of ¹⁸F-FMISO in clinical trials (21). This tracer has been successfully used to image hypoxia in gliomas, head and neck, and breast and renal tumors. However, the use of ¹⁸F-FMISO in sarcomas, pancreatic cancers, or rectal cancers was compromised because of the non-specific accumulation of ¹⁸F-FMISO in normoxic surrounding tissues or because of insufficient tracer uptake. Four head and neck tumor studies were able to correlate one ¹⁸F-FMISO-related tumor parameter (T:Bmax, SUVmax, or T:Mmax) with disease-free survival or locoregional failure (21), whereas one study was not able to establish any correlation (26). To date, different TBRmax thresholds for stratification have been reported; therefore, standardized methods still need to be determined in multicenter studies (27). Tumor mapping of hypoxia with ¹⁸F-FMISO could be useful for planning intensity modulated radiation therapy (IMRT) on patients with head and neck cancers, since when the hypoxic regions are well delineated, it is possible to boost the dose delivered to those areas. ¹⁸F-FMISO maps were used for this purpose on two patients in a study from Lee and colleagues in 2008. With the knowledge of hypoxic areas that they gained, these authors were able to escalate the dose to 84 Gy for 10 patients. Moreover, they raised the delivered irradiation doses up to 100 and 105 Gy for two patients in hypoxic areas (28). A single-center trial combining multimodal hypoxia imaging, including ¹⁸F-FMISO, and IMRT in patients with inoperable stage III non-small cell lung carcinoma (NSCLC) tumors was started in 2012 (29). Recent data also suggest that selective dose painting to hypoxic tumor subvolumes requires adaptation during treatment (30). ¹⁸F-FMISO has also been used to monitor reoxygenation of the tumors during the course of radiotherapy: in 10 patients, a decrease in the uptake of ¹⁸F-FMISO was observed in eight tumors after the delivery of 20 Gy (31). The reoxygenation process has also been observed in patients with glioblastoma treated by fractionated radiotherapy and concomitant temozolomide administration (32). The results showed a significant decrease in tumor hypoxia attributed to the radiotherapy effect. Finally, it has been suggested that the hypoxic areas of the tumors are correlated with neovascularization and with the tumor metabolism rate in glioblastoma multiform (33). This conclusion comes from a preliminary study involving 10 patients who underwent magnetic resonance imaging (MRI) to evaluate tumor perfusion after the injection of a gadolinium-based contrast agent and several PET imaging protocols: the first of these was ¹⁸F-FMISO as a reporter of hypoxia and the second was l-methyl-11C-methionine (¹¹C-MET), an amino acid whose uptake reflects tumor activity and which is currently used for glioma detection and grading (34).

¹⁸F-fluoroazomycin-arabinofuranoside (¹⁸F-FAZA) is a more hydrophilic nitroimidazole that displays faster clearance from blood and normal tissues than ¹⁸F-FMISO. As a result, imaging tumor hypoxia with this radiotracer improves the signal-to-noise ratio. In a preclinical study on rhabdomyosarcoma, a correlation between ¹⁸F-FAZA uptake and actual values of pO₂ measured by EPR has been established, reflecting quantitative aspects of the method (35). Moreover, ¹⁸F-FAZA seems to be predictive of the response to radiotherapy: less hypoxic rhabdomyosarcoma tumors (defined by a lower uptake of ¹⁸F-FAZA) demonstrated better local tumor control 90 days after radiotherapy than more hypoxic tumors (36). Similarly, a significant correlation between ¹⁸F-FAZA T/B ratio and tumor growth delay was found in 9L glioma (37). With regard to clinical applications, ¹⁸F-FAZA imaging has been successfully performed in gliomas, lymphomas, lung, head and neck, and cervical and rectal tumors (21). The results of the DAHANCA 24 trial on head and neck cancers have proven that ¹⁸F-FAZA uptake is a good prognostic factor of tumor response to radiotherapeutic treatment (38). Finally, ¹⁸F-FAZA-PET images have been successfully exploited to delineate radiotherapy planning for head and neck squamous cell carcinoma, with 86 Gy being the dose to deliver in hypoxic areas. The treatment protocol included three phases and was based on ¹⁸F-FAZA-PET images acquired before irradiation and after the 7th and 17th fractions (39). ¹⁸F-flortanidazole (¹⁸F-HX4) is a hydrophilic nitroimidazole which quickly clears from normoxic tissues, allowing imaging 90 min after tracer administration; its low accumulation in the brain, heart, and gastrointestinal tract enables these body parts to be imaged (20). In a comparative study looking at several markers of hypoxia in an in vivo model (head and neck carcinoma cells SQ20b), ¹⁸F-FAZA, ¹⁸F-HX4, and ¹⁸F-FMISO uptakes were correlated with hypoxia, despite a relatively low accumulation of ¹⁸F-FAZA in muscles and tumors (40). In a second comparative study, ¹⁸F-HX4 and ¹⁸F-FAZA were found to be sensitive to an increase of hypoxia, induced by the breathing of a gas mixture containing 7% O₂, when the tumor-to-blood ratio was used. However, when only the tumor-to-muscle was used, only ¹⁸F-FAZA revealed a significant decrease in tumor oxygenation (41).

¹⁸F-labeled fluoroerythronitroimidazole (FETNIM) was suggested as another marker of tumor hypoxia for use with PET in 1995 (42). Initial data suggested that ¹⁸F-FETNIM shows low peripheral metabolism, little defluorination, and possible metabolic trapping in hypoxic tumor tissue (43). ¹⁸F-FETNIM distribution has been positively correlated with ¹⁸F-FAZA in murine mammary tumors under normoxic and hyperoxic conditions (44). It has been tested in head and neck, lung, cervical, and esophageal clinical cancer studies, with significant correlations between patient outcome and either high fractional hypoxic volumes, F-FETNIM uptake, or baseline SUVmax (21). Cross-validation studies with other quantitative oxymetric markers are lacking. However, F-FETNIM has been compared to F-FAZA in the preclinical and clinical settings, with positive correlations (21, 44, 45).

Copper (II) diacetyl-bis (N4-methylthiosemicarbazone) (Cu-ATSM) can be used as a radiotracer with ^60–64Cu with variable half-times (46). This agent displays high lipophilicity and rapid clearance from normoxic tissues, thereby enabling imaging 30 min after its administration (47). In the absence of oxygen, the Cu(II) is irreversibly reduced to Cu(I) in viable mitochondria and therefore becomes trapped in hypoxic cells. In the study by Carlin and colleagues, the ⁶⁴Cu-ATSM molecule displayed a better uptake in tumor than ¹⁸F-FMISO, ¹⁸F-FAZA, and ¹⁸F-HX4. However, its distribution within the tumor was not similar to the other tracers: the accumulation of ⁶⁴Cu-ATSM was greater at the tumor periphery and the uptake was lower in the tumor center where perfusion was also reduced (40). The low accumulation of ⁶⁰Cu-ATSM in the urinary tract makes it an ideal candidate for imaging pelvic organs. For example, the uptake of this radiotracer has been inversely correlated with the patient outcome (in terms of progression-free survival) for 38 patients with cervical cancer (48). Similar observations have been performed in head and neck, and rectal and lung tumors, as reviewed in (16) and in more recent studies in NSCLC and head and neck tumors (49, 50). Few preclinical studies have attempted to link Cu-ATSM uptake and outcome; one study of canine tumors was not able to establish any correlation (51). Planning of dose painting can also be achieved using Cu-ATSM, which detects the hypoxic regions in preclinical and clinical models (52–54). However, Cu-ATSM uptake does not only reflect hypoxia: in a study of six tumor cell lines, the maximum uptake was cell line dependent and was linked to the redox status of tumor cells. The retention of Cu was higher in cells with an abnormally reduced status (55). Moreover, the in vitro results demonstrated that hypoxia selectivity was optimal 30–60 min after the administration of Cu-ATSM, but this is a limiting factor for in vivo applications, since the distribution of the tracer during the first hour after its administration is limited by a reduced tumor blood flow. The latter imaging is suggested to be rather linked to the active transport of Cu alone that has been dissociated from the Cu-ATSM complex, although these observations are again cell line dependent (55). The fact that copper metabolism may also play a role in the uptake mechanism of ⁶⁴Cu-ATSM was confirmed in a more recent publication showing similar contributions between ⁶⁴Cu-ATSM and ⁶⁴Cu-acetate (56). Further studies have demonstrated that Cu-ATSM uptake is not correlated with an increase in hypoxia (57, 58) or that Cu-ATSM uptake is not co-localized with hypoxia marked with immunohistochemistry (40, 59, 60). Only one recent study has concluded in favor of a positive correlation between tracer accumulation and hypoxia but not in both tumor models under study (61). It seems that Cu-ATSM is not a specific marker of tumor hypoxia, but it has been successfully correlated with the NADH and NADPH levels: Cu-ATSM uptake is rather observed in tumors with abnormally reduced status, which may or may not be linked to hypoxia (62, 63). Finally, another study by Vavere and Lewis investigated the link between Cu-ATSM uptake and the fatty acid synthesis pathway, which consumes NADPH, and correlated the level of fatty acid synthase with the Cu-ATSM uptake (64). Consequently, Cu-ATSM images cannot be interpreted in terms of oxygenation only and, although Cu-ATSM is predictive of radiotherapy outcome, it is unclear whether this is linked to tumor hypoxia.

¹⁹F-MRI is a non-invasive method able to map tumor hypoxia quantitatively, after the injection of a perfluorocarbon emulsion. Calibration curves relating the longitudinal relaxation rate to pO₂ can be acquired for a given temperature and a given perfluorocarbon (65) and used to map tumor oxygenation quantitatively. A major advantage of this calibration is the independent property of the absolute ¹⁹F signal intensity, linked to perfluorocarbon uptake. The fluorocarbon (PFC) relaxometry using echo planar imaging for dynamic oxygen mapping method developed by Mason and colleagues has been successfully used to monitor positive and negative changes in tumor oxygenation (65–68) as well as to map the heterogeneity of response to hyperoxic challenges within each tumor: it appears that well-oxygenated areas at baseline will display an increase in oxygenation earlier than hypoxic areas (65). Similarly, ¹⁹F-MRI mapping was able to monitor the effect of a radiosensitizer, S-nitrosocaptopril, which induces a significant increase in tumor pO₂ from 20 to 60 min after its administration (69). Due to an acquisition time reduced to 1.5 min, Jordan and colleagues were able to monitor spontaneous oxygenation fluctuations in the range of 5–30 mmHg in transplantable mouse liver tumors and to identify hypoxia cycles in this model (70, 71). PFC can remain in the tumor and enables repeated measurements. The injection of perfluoro-15-Crown-Ether in Shionogi tumors, a murine mammary carcinoma which has acquired a dependence on androgens, has shown that ¹⁹F-MRI is able to distinguish three hormone-dependent oxygenation statuses (72). Tumor tissue heterogeneity can be assessed by the diffusion-based multispectral technique in order to distinguish tumor necrosis from viable tumor tissue and to detect subcutaneous adipose tissue. In a recent study, Shi and colleagues monitored the tumor response to hyperoxic and hypoxic challenges by considering the tumor as a whole or by considering each tissue type separately. The pO₂ increased significantly when the authors considered the tumor as a whole, and this response was enhanced further when they focused on the viable tumor tissue (73). However, in a study comparing ¹⁸F-PET imaging and ¹⁹F-MRI, it was shown that fluorine mapping with MRI was less sensitive to small pO₂ changes (from 3 to 5 mmHg) in some tumors (35). Moreover, before performing ¹⁹F-MRI, the toxicity of the chosen PFC needs to be taken into account since it has been observed, for example, that the early toxicities (thrombosis and tissue necrosis) observed with HFB could be avoided by using 15C5 (74). Despite this, approval is being awaited from the FDA for the investigation in a clinical trial of PFCs as a biomarker of tumor response to radiotherapy.

Dynamic contrast-enhanced MRI is a method widely used in preclinical and clinical research to assess information on tumor hemodynamics. A bolus of gadolinium-based contrast agent is injected, and its distribution within the tissue of interest is analyzed through signal enhancement, thereby providing information on perfusion and permeability. This technique is regularly combined with oximetric methods such as PET imaging, blood oxygen level-dependent (BOLD)-MRI, or EPR oximetry to assess tumor hemodynamic parameters and their impact on therapy (75–77). Two parameters are regularly assessed by DCE-MRI, using the Tofts model: k^trans, representing the volume transfer constant between blood plasma and extravascular extracellular space, and v_p, defining the blood plasma volume per unit volume of tissue. Low-perfused tumor areas suffer from an insufficient supply of oxygen, thereby leading to hypoxia. It has therefore been suggested that DCE-MRI could be used as an indirect method to detect hypoxic areas in tumors. In a recent preclinical study by Øvrebø and colleagues, k^trans was found to be predictive of tumor response to radiotherapy: low k^trans was associated with an increased radioresistance in hypoxic melanoma xenografts, suggesting that DCE-MRI is a biomarker of tumor radioresistance in hypoxic tumors (78). Further studies on two cervical cancer xenografts have confirmed those results, indicating that the radiotherapy outcome can be correlated with k^trans values measured before the treatment taking each tumor model separately (79). Furthermore, skewness in the distribution of the k^trans parameter was also correlated with poorer patient outcome, highlighting the heterogeneity of perfusion within these tumors (80). Conversely, a study on fibrosarcoma was not able to show any correlation between DCE-related parameters (k^trans, v_p, K_ep, % of perfused voxels) and the outcome of radiation therapy (81). With respect to validation of the technique with other oximetric methods, attempts have been made in preliminary clinical studies to assess correlations between the level of hypoxia and permeability, with mixed results, in head and neck cancer and gliomas using pimonidazole staining (an immunohistological staining aimed at detecting tumor hypoxia) and in cervical cancer using polarographic electrodes (82–84). Søvik and colleagues have also demonstrated that DCE-MRI could be used to monitor the changes in tumor oxygenation during the course of radiotherapy in order to adapt IMRT to changes in hypoxia distribution within the tumor after several doses of irradiation (85). The Brix model can also be used for the analysis of DCE-MRI images. Perfusion or permeability is then assessed by the parameter A_Brix, also known to measure the extravascular extracellular space (86). In order to compare the Brix and Tofts models and their ability to predict the outcome of patients, Andersen and colleagues tested both models on patients with cervical cancers. They concluded that low values of k^trans and A_Brix can be associated with poor outcome (87). A recent study also demonstrated that low A_Brix could be correlated with an upregulation of genes involved in the response to hypoxia (88).

Nevertheless, numerous precautions have to be taken when interpreting DCE-MRI images in terms of oxygenation because perfusion is not the only feature influencing pO₂. Moreover, despite the establishment of relations between k^trans and oxygen tensions or immunohistochemical measurements in some studies, estimates of perfusion remain indirect estimates of hypoxia and, in some circumstances, do not relate to hypoxic status (89). It is also important to mention that the contrast agent distribution can also be altered by perfusion and extracellular volume, leading to misestimation of oxygenation in necrotic areas (85).

Blood oxygen level-dependent MRI, or fMRI, uses endogenous contrast and is sensitive to the effective transversal relaxation rate of protons (R2*=1/T2*). This measurement is sensitive to the ratio of oxyhemoglobin and deoxyhemoglobin, the latter being a paramagnetic agent that shortens T2*.

In preclinical studies, BOLD-MRI has proven its ability to monitor changes in oxygenation levels during hyperoxic challenges (90–92). Our group has compared the changes in BOLD signal following the administration of carbogen or isosorbide dinitrate and has observed that the magnitude of the changes is stronger during the carbogen challenge (affecting the hemoglobin saturation) than after the administration of the NO donor (93). The oxygenation concentration is not the only parameter that affects R2*: changes in tumor blood flow, blood volume, blood pH, or metabolic status can also influence the R2* measurements (94, 95). Changes in R2* should therefore be carefully considered when they are treated as an indicator of changes in tumor oxygenation. Moreover, no correlation has been established between R2* measurements and absolute values of pO₂ (96). In comparisons with pimonidazole staining, both correlation and inverse correlation have been observed in prostate and mammary tumors, respectively (97, 98). BOLD-MRI is therefore used to monitor tumor oxygenation changes rather than to map tumor hypoxia quantitatively. Spontaneous fluctuations have been successfully monitored by BOLD-MRI, which has led to the identification of several cycles of hypoxia, with periods ranging from 3 min to 1 h (99). Moreover, the tumor regions in which tumor oxygenation fluctuates have been related to areas with functional vasculature. The prognostic value of R2* was investigated in a preclinical study. The authors subjected rats with GH3-prolactinomas and mice with RIF-1 fibrosarcomas to carbogen breathing before radiation therapy. The GH3 prolactinomas displayed large R2* and a large response to the hyperoxic challenge (ΔR2*), and this was predictive of a transient reduction in the tumor size after irradiation. However, the inhibition of tumor growth exhibited with the RIF-1 fibrosarcoma was smaller, and this was related to a low R2* at baseline and to a poor response to the hyperoxic challenge (100). Recently, our group observed that R2* was predictive of radiation therapy outcome in rat 9L-glioma tumors but not in rhabdomyosarcoma tumors (101). In a study by Kim and colleagues, the results from a small sample of cancer cervical patients suggested that the R2* values are predictive of the radiotherapeutic response (102).

It is important to remember that the BOLD signal is related to the amount of deoxyhemoglobin and therefore linked to the blood pO₂ and blood saturation of oxygen (SO₂). There is therefore a real interest in quantifying the BOLD signal (103). A technique known as multiparametric quantitative BOLD has been recently developed to achieve a quantitative mapping of tissue oxygenation. The method is based on the acquisition of several images, with standard sequences, aimed at measuring the blood volume fraction, field inhomogeneities (by mapping B0), and the tissue T₂ before and after the administration of a contrast agent. These three values are then integrated in a model describing the T2* signal in order to calculate average oxygen saturation in each voxel (95). This method has been successfully applied to map tumor hypoxia in the brain in stroke or gliomas (104). This approach allows a quantitative measurement of the blood oxygen saturation and represents a major improvement in the use of BOLD imaging to map tumor hypoxia and to monitor tumor oxygenation changes.

Oxygen is a paramagnetic agent that shortens the longitudinal relaxation time (T₁) of surrounding protons. Consequently, T₁ mapping appears as a possibility for mapping tumor hypoxia. This method does not require any contrast agent and is widely available in medical imaging centers. Moreover, a correlation can be established between pO₂ values and relaxation rates, as was done, for example, between relaxation rates and arterial blood oxygen pressure in a pig (105), revealing the quantitative aspect of such measurements. However, since T₁ relaxation is also influenced by temperature, tissue of interest, blood flow, and basal blood oxygen saturation, a calibration between T₁ values and pO₂ cannot be established so easily. Nevertheless, assessing T₁ measurements has been a useful method for monitoring changes in tissue oxygenation.

In a recent study by Muir and colleagues reporting the use of a hyperbaric chamber for rodents, T₁ values measured in the brain were found to be significantly reduced (with an increase in related R₁ values) when successive switches were made from normobaric air to hyperbaric air and then to hyperbaric oxygen (106). On the basis of a comparison between the changes in T₁ values in liver, kidney, and muscle of healthy rats obtained during transitions from air to pure oxygen, to carbogen (10% CO₂ and 90% O₂), or to a mixture of ambient air with 10% CO₂, it was concluded that these measurements were sensitive to oxygen dissolved in tissues when there was no concomitant change in blood flow (107). However, in the same study, the authors highlighted the lack of sensitivity of T₁ measurements to a decrease in oxygenation during a switch from pure oxygen to carbogen or from air to a mixture of air and 10% CO₂. The explanation may lie in the vasodilatation induced by CO₂, which offsets the decrease in tissue oxygenation, resulting in unexpected positive R₁ changes (107). However, this issue is controversial: in a preclinical study aimed at evaluating brain oxygenation during a hyperoxic challenge, the R₁ (R₁ = 1/T₁) values were similarly increased in the cerebral cortex and in the pituitary gland during both carbogen and pure oxygen breathing, despite a decrease in brain perfusion induced by pure oxygen breathing (indeed, in the absence of CO₂, vasoconstriction occurs) (108).

T₁ measurements have been assessed together with T2* for monitoring tumor oxygenation in murine prolactinoma models and prostate tumor xenografts undergoing a hyperoxic challenge. The changes in relaxation rates were found to be related to the basal oxygenation status: the most hypoxic tumors exhibited significantly reduced R₁ values and significantly higher R2* values (R2*=1/T2*) (109).

T₁ measurements have recently been used in conjunction with perfusion MRI to quantify the hypoxic fraction in multiple models with differing hypoxic and vascular phenotypes (110).

In a preliminary human study on healthy volunteers, this technique was successfully used to monitor the increase in oxygenation in normal myocardium, spleen, and arterial blood. However, there was no significant change observed in liver, skeletal muscle, or subcutaneous fat (111). These discrepancies in the results were attributed to a lack of sensitivity, to the uncontrolled motion of organs, and to differences in blood flow, blood volume, and regional oxygen consumption. Two years later, a study by Noseworthy and colleagues also failed to monitor the oxygenation in skeletal muscles subjected to a hyperoxic challenge involving pure oxygen with T₁ measurement despite a significant change in T₂ (112). Nevertheless, further work demonstrated the effect of oxygen in shortening the longitudinal relaxation time in healthy volunteers’ muscles, spleen, renal cortex, subcutaneous fat, placenta, and liver (113–116). Furthermore, in another study by O’Connor and colleagues, the authors observed significant differences of response induced by either pure oxygen or carbogen depending on the tissue of interest: carbogen induced a lower T₁-shortening effect than pure oxygen in the spleen, whereas the opposite phenomenon was observed in the liver. Ten patients with abdominal tumors were then subjected to pure oxygen breathing. A significant increase in R₁ values was observed in eight of these patients with ovarian, cervical, or gastrointestinal malignancies (117). The effect of hyperoxia has also been investigated in the brain using a dynamic T₁-weighted sequence called tissue oxygen level dependent (TOLD) (118). This method allows reduced acquisition times and is more suited for a dynamic assessment of changes induced by a hyperoxic challenge. Haddock and colleagues (119) monitored the oxygenation of brain tissue during two protocols: the first was composed of a twice-repeated switch from ambient air to pure oxygen with two 2-min intervals, while the second began with a normoxic phase followed by a switch to pure oxygen breathing during 7 min followed by a final breathing of air. Due to the T₁ effect, they could observe changes in signal intensity whose magnitude was related to the changes in brain oxygenation monitored with BOLD-MRI. Furthermore, TOLD measurements have been demonstrated to be sensitive to dynamic hyperoxic challenges (120) and seem to be predictive of the radiotherapy outcome (92). Prostate tumors that are well reoxygenated during pure oxygen breathing before radiotherapy display a significantly larger regrowth delay than low responders to hyperoxia (92). Combined BOLD and TOLD contrasts were also recently assessed in Dunning rat prostate tumors and were shown to be useful prognostic factors for predicting tumor response to hypofractionation (121). However, these results remain preliminary and need to be further investigated. The study by Burrell and colleagues highlights the complementary character of concomitant T₁ and T2* measurements: although the mean response was an increase in R₁ values and a decrease in R2* values in both models, they observed that the magnitude of R₁ and R2* changes was dependent on the basal oxygenation status. The explanation may be linked to hemoglobin saturation: in well-oxygenated tumors, the increase in oxygen supply will raise the amount of dissolved oxygen rather than increase the already well-saturated hemoglobin. This results in higher amplitude changes of positive ΔR₁ rather than of negative ΔR2* (109). It is also important to remember that the T2* measurement is influenced by the hemoglobin and deoxyhemoglobin ratio and is therefore sensitive to vascular oxygenation, whereas the T₁ measurement is rather sensitive to oxygen dissolved in tissues. As there is a real interest in combining the measurement of R₁ and R2*, Ding and colleagues have recently proposed a new sequence enabling simultaneous acquisition of T2* and T₁ measurement. With this new method, they monitored the dynamic changes in oxygenation of abdominal organs (spleen, medulla, and renal cortex) (118).

From the foregoing, we can conclude that there is real value in measuring the ¹H relaxation time to obtain information on tissue oxygenation non-invasively. By comparing the sensitivity of ¹H-MRI oximetry with ¹⁹F-MRI using perfluorocarbons, Tadamura and colleagues observed that: “T1 shortening effect with oxygenation observed using ¹⁹F MR system with PFCs is much greater than that observed with ¹H MRI because of large oxygen solubility of PFC compared with that of aqueous media. Therefore, the PFC method is more sensitive to tissue oxygenation state.” (111). In all the techniques described above, the signal was mainly influenced by protons of the water molecules. As oxygen solubility is higher in lipids than in water, focusing on the relaxation of protons belonging to lipid molecules would improve the sensitivity of the techniques described above. Consequently, a new technique called mapping of oxygen by imaging lipids relaxation enhancement (MOBILE) aimed at selectively measuring the T₁ of the lipid component has been proposed (122). This method has demonstrated its ability to distinguish the oxygenation levels in tumor tissue homogenates submitted to different oxygen concentrations. Moreover, the MOBILE technique can be used to map local pO₂ in several tissues such as liver, muscles, brain (infarcted or not), and mammary tumors. Both positive and negative changes in tumor oxygenation can be monitored with the MOBILE technique. Its quantitative aspect has also been demonstrated on mammary tumor models presenting high lipid content (123). However, the method was not applicable to tumors with low lipid content. An alternative method based on the deconvolution of global T₁ in fat and water components was recently developed. However, a quantitative aspect could not be demonstrated in rhabdomyosarcomas and glioma models, and lipids T₁ and global T₁ turned out to be less predictive of the outcome of radiation therapy than R2* (101). Considering the clinical application of MOBILE, the method has demonstrated its ability to identify hypoxia in stroke areas (124). The method is currently being investigated in human gliomas, with a pilot study showing that global R₁ and lipid R₁ values are significantly lower in tumors than in the “normal appearing white matter” of patients or the healthy brains of volunteers and that lipid R₁ measurements enable discrimination between tumor areas and peritumoral edema (125).

Using the exogenous source of contrast hexamethyldisiloxane (HMDSO), R. P. Mason’s group identified a source of oxygen-sensitive contrast, called “proton imaging of siloxanes to map tissue oxygenation levels,” which was validated using hyperoxic breathing challenge in Dunning prostate R3327 MAT-Lu tumor-implanted rats but was not assessed as a predictive marker of the outcome of radiation therapy (126).