Monitoring Oxygenation Levels Deep in the Tumor Core: Noninvasive Imaging of Hypoxia, Now in Real-Time 3D Free

It is common knowledge that tumor vasculature is aberrant and dysfunctional (1), leading to poor delivery of oxygen and nutrients, and insufficient drug delivery to the target tumor tissue during treatment. Understanding tumor blood flow and especially monitoring oxygenation status is valuable, not just for better understanding of cancer physiology and systems biology, but also for obtaining information on the status and behavior of tumor node(s) in the clinical setting. It also has potential application in personalized medicine. Tumors can be highly heterogeneous; even within a single node, significant heterogeneity may be observed (2). Revealing hypoxia heterogeneity may help to understand tumor biology and may eventually aid in the selection of appropriate treatment strategies for particular patients.

For practical use, especially with the goal of personalized and precision medicine, obtaining tumor tissue oxygenation information is not trivial. Traditional optical methods, such as pulse oximetry, are superficial and provide low spatial resolution. Preclinical research aimed at understanding the role of tissue oxygenation in tumor biology is important for understanding of the critical pathways that determine tumor and stromal cell fate, proliferation, metabolism, and mutation rates. Research methods to obtain real-time information about the tumor in vivo are either invasive (3), or limited by complicated, expensive, and time-consuming tests, such as PET (4) and MRI, which require strong magnetic fields or ionizing radiation. Imaging hypoxia takes significant time, and imaging techniques used in these studies are not always capable of live monitoring. The optoacoustic (also known as photoacoustic) approach presented by Ron and colleagues (5) in this issue of Cancer Research and others, for example, Tomaszewski and colleagues (6), holds significant potential to address this challenge and helps transition tumor tissue oxygenation imaging from research toward clinical practice.

As a physical phenomenon, optoacoustics is not new. The photoacoustic effect was first discovered by Alexander Graham Bell in 1881 (7). Modern incarnation of this technique in biomedicine is based on rapid heating during short periods of time (nanoseconds, using tissue irradiation with laser light pulses). The resulting heat-induced expansion generates ultrasound waves. Tissue heating is dependent on the ability of the interrogated tissue to absorb light of a certain wavelength. Because of the fact that pulses are ultra-short, overall light fluence, expressed as the amount of delivered energy necessary to perform transcutaneous studies, is within the safety limits of laser use in healthcare, as established by the American National Standards Institute (8). As the level of blood oxygenation defines the optical energy absorbance spectrum of hemoglobin, the measurement of spectral dependence of the photoacoustic signal provides the ability to assess the level of oxygen in the tissue. Purely optical methods are only capable of monitoring events that take place close to the surface, because light is effectively scattered by the biological tissues. For example, intravital light microscopy offers excellent spatial resolution, but it is not able to monitor events deeper than a fraction of 1 mm below the surface; cremaster muscle, mesenteric vasculature, skin flap, or mouse ear studies in animal models are common, mostly requiring skin removal to minimize light scattering. Noninvasive imaging of the events inside tumors at over a millimeter depth is not feasible with this technique, furthermore, the hypoxic core only forms in tumors over several mm in size. Photoacoustics resolves the problem of penetration depth, as well as the problem of spatial resolution. It allows much deeper penetration—heat-generated ultrasound wave is not attenuated significantly even after it travels through several centimeters of tissue, so, the distance required for light to travel is half of what would be necessary for purely optical techniques. As the photoacoustic signal can be successfully generated by diffused photons deep inside the tissue, ultrasound detection by the transducer array can monitor signals generated far from the surface. Spatial resolution of the method benefits from poor scattering of ultrasound waves in soft tissues and may be down to approximately 0.1 mm depending on the receiving frequency optimum of the transducers used, for example, 10 MHz in this case (5).

The study performed by Ron and colleagues (5) has taken advantage of the ability of their system to perform volumetric ultrasound detection with a spherical array of transducers, with imaging depth completely covering the tumor, approximately 1 cm for this particular study. The additional advantage of this study is in its reliance on the intrinsic oxygen-sensitive optical contrast agent available in the bloodstream, hemoglobin, so there is no need to administer external synthetic contrast dyes. Upon oxygen binding, the infrared absorbance spectrum of hemoglobin changes significantly at some of the interrogated wavelengths between 730 and 850 nm, which allows to compute the oxygen saturation of hemoglobin, sO₂, and assign oxygen saturation to each voxel in the image dataset. A combination of the multi-spectral laser light excitation and volumetric ultrasound detection of photoacoustic signal gives this technique the name volumetric multispectral optoacoustic tomography (vMSOT).

Ron and colleagues (5) demonstrated, using vMSOT, that the levels of hypoxia (presented as sO₂ in the tumor mass) can be monitored within three major tissue segments, which they designated as normoxic rim, normoxic core, and hypoxic core. Immunohistologic staining for carbonic anhydrase IX, a tissue marker of hypoxia, seemed to match hypoxia observed in the sO₂ image frame. Interestingly, hypoxia levels looked quite significant in some core areas of the tumor that have strong expression of endothelial marker CD31 (in turn, looking similar to carbonic anhydrase IX histology), confirming the dysfunctional status of tumor vasculature in those segments. Even more interesting is the ability of the method to perform the study in real time. Thus, the study presented sO₂ volumetric kinetic data, obtained as a response to oxygen challenge, which showed rapid (within a minute or two) increase of oxygen levels in the normoxic rim. The sO₂ level increased moderately in response to oxygenation challenge in tumor core tissue designated as normoxic; this increase rate was slower than in the normoxic rim. Remarkably, not an increase, but a reduction of sO₂ level in most of the areas of the tumor assigned as hypoxic core was observed.

An added benefit of the dynamic imaging potential of vMSOT is the ability to monitor the cycling of hypoxia levels in each voxel, and the ability of the method to present the frequencies of opening/closing events, observed up to several times per hour. Cyclic variation of hypoxia levels in tumors has already been suggested as an important parameter that influences angiogenesis, resistance to anticancer treatments, intratumoral inflammation, and tumor metastasis (9). The noninvasive deep penetration optical tool, vMSOT, may help simplify investigation of these interdependencies.

Attention needs to be brought to the limitations of vMSOT as an imaging technique (5) that may complicate practical translation. The main issue is the moderate penetration depth of this technique, to about an inch currently; it will have difficulty in imaging bone and lung, due to inherent physics constraints of sound and light propagation. Another limitation is the narrow field of view, which hampers the use of this apparatus for scouting assessment and tumor node detection. Special attention has to be given to motion compensation (hence some of the volumetric imaging frames in this study had to be rejected from processing), which may be complicated further by the moderate frame rate of the laser, and time it takes to switch between the multiple optical wavelengths of the laser used for excitation. So, vMSOT equipment would clearly benefit from frame rate increase, both for laser firing frequency and wavelength change time. Achieving portability of the equipment would be the next logical step toward the attractiveness of the approach for general use. MSOT technology might benefit from a combination with conventional ultrasound, administered by the same acoustic array, so that scouting anatomy information would be easily available from traditional ultrasound, Doppler would assist in locating large vessels, and ultrasound contrast-specific pulse sequences would aid monitoring tissue perfusion down to the level of microvasculature.

Potential practical usefulness of the results presented in Ron and colleagues' article is clear. The ability to monitor in real time the levels of sO₂ and its heterogeneity within tumor nodes, as well as observing cyclic sO₂ temporal changes, enable feedback for immediate decision-making during image-guided interventions and ability to check whether hypoxia level changes in the target tumor nodes in response to therapeutic interventions. It should be able to assess tumor response quickly, well before tumor volume change. This may be especially helpful for rapid evaluation of the efficacy of chemotherapy and antivascular therapy, including chemoembolization and anti-VEGF therapeutics. Clinical testing of vMSOT has recently been reported in the cardiovascular imaging setting (10). Successful imaging of carotid artery bifurcation in human volunteers has been demonstrated, with imaging depth reaching up to 30 mm. Therefore, photoacoustics in general, and vMSOT in particular, may aid both cardiovascular medicine and oncology. In the future, combination of photoacoustics with other imaging tools will provide direct volumetric comparison between the status of hypoxia, actual blood flow rate, and metabolism rate in the interrogated tissues. This will help uncover physiologic mechanisms of tumor development and progression that are not yet known.

A.L. Klibanov reports receiving a commercial research grant from SoundPipe Therapeutics NIH Grant subcontract and has ownership interest (including stock, patents, etc.) in Targeson Inc. (now dissolved). No potential conflicts of interest were disclosed by the other author.

A.L. Klibanov was supported by institutional grant from NIH (NIH R01 EB023055) and current subcontract via NIH R44 HL139241 from SoundPipe Therapeutics. S. Hu was supported by institutional grant from NIH (NIH R01 NS099261).