New transparent liquid bandage can detect tissue oxygenation as well as the gold standard of an oximeter


In the first human clinical trial, researchers at Massachusetts General Hospital (MGH) and surgeons at Beth Israel Deaconess Medical Center (BIDMC) have validated the practicality and accuracy of an oxygen-sensing liquid bandage that measures the concentration of oxygen in transplanted tissue.

The trial, published in Science Advances, compared the performance of a novel, paint-on bandage made with phosphorescent materials to a wired tissue oximeter (ViOptix device) – the current standard for monitoring tissue oxygenation – in women undergoing breast reconstruction surgery after cancer.

“Our trial showed that the transparent liquid bandage detected tissue oxygenation as well as the gold standard of an oximeter, which uses old technology, is uncomfortable for the patient, obstructs visual inspection of the tissue, and can give false readings based on lighting conditions and the patient’s movements,” says Conor L. Evans, Ph.D., the paper’s senior author and a principal investigator at MGH’s Wellman Center for Photomedicine.

“The standalone bandage is a major advancement from a wired oximeter that restricts a patient’s movements and is complicated to use.”

The research team took on the challenge of building a better tissue oxygenation sensor following a request from the Department of Defense seeking to reduce failure rates of tissue transplant surgeries and skin grafts in injured soldiers. The technology underlying the bandage was developed through the support of the Military Medical Photonics Program.

The trial tests the bandage in breast reconstruction, a common type of free-flap transplant surgery in which plastic surgeons harvest skin, fat, arteries and blood vessels from the patient’s abdomen and microsurgically reattach the tissue and vessels to the chest.

“Up to 5% of free-flap surgeries can fail, typically within 48 hours after surgery, if blood flow to the transplanted tissue is interrupted or inadequate, which is a devastating outcome,” says Samuel J. Lin, MD, MBA, plastic and reconstructive surgeon at BIDMC and senior author. By monitoring how much oxygen gets to the transplanted tissue, surgeons can quickly detect a vascular problem and intervene to save the transplant.

Five women undergoing breast reconstruction were enrolled in the trial from March to September 2017. The liquid bandage was painted on in a 1 cm by 1 cm area on seven transplanted flaps (two women had both breasts reconstructed). A wired oximeter was also placed on each flap, and tissue oxygenation was monitored for 48 hours after surgery.

The bandage measures the amount of oxygen getting to the tissue itself, while the ViOptix reads the amount of oxygen saturation in the blood with near-infrared spectroscopy—a less direct measurement of crucial blood flow to the transplant.

In this study, a clinician-researcher took photos of the bandage with a digital camera with custom filters following surgery. The flash from the camera excited the phosphorescent material in the bandage, which then glowed red to green based on the amount of oxygen present in the tissue.

Liquid bandage detects tissue oxygenation without the drawbacks of wired oximeters
Comparison of the two tissue oximetry modalities used in this study. Left: The current standard of care ViOptix, which relies on the near-IR absorbance of oxy- and deoxyhemoglobin to detect the tissue oxygen saturation %stO2 under the wired lead. Right: A phosphorescence-based approach to tissue oximetry, which uses the phosphorescence of metalloporphyrin painted onto the surface of the skin to map tissue oxygenation. Photo credit: Juan Pedro Cascales, MGH. Credit: Science Advances  18 Dec 2020: Vol. 6, no. 51. DOI: 10.1126/sciadv.abd1061

Evans and colleagues have since developed a battery-powered sensor head for the bandage that eliminates the need for the camera and makes the bandage self-contained. The prototype study was published in Biomedical Optics Express.

In all seven flaps, the tissue oxygenation rate of change measured by the bandage correlated with the oximeter, and all seven flaps were successful. The researchers are currently designing a clinical trial to study how well the bandage detects a flap that is failing from lack of oxygen.

“The ability to have a wireless oxygen monitoring device for blood flow is potentially a gamechanger,” says Lin.

Clinical applications for an oxygen-sensing bandage could include monitoring wound healing, tissue transplants for trauma, skin grafts for burns, limbs affected by peripheral artery disease and chronic ischemia (reduced blood flow).

“The technology might also detect important tissue changes in patients with heart disease and other chronic medical conditions, providing an early warning signal that disease is progressing,” says Lin. “And there are likely other clinical uses we haven’t yet considered.”

The overall goal of this review is to facilitate clinically effective use of measurements of molecular oxygen (O2) in tissues with the explicit intent of improving clinical care, that is, improving the accuracy and effectiveness of diagnoses, treatments, and prognoses for individual patients.

This review focusses especially on improving personalized medicine and outcomes of care, by carefully considering the basis and validity of clinical measurements of O2 in tissues and how those measurements can be used to advance diagnosis and therapy. While measurements of O2 in tissues have been recognized as an important factor in the clinical evaluation and treatment of many diseases, especially cancer (Busk, Overgaard, & Horsman, 2020), and pathologies involving ischemia (such as in peripheral vascular disease and wound healing), insufficient attention often has been paid to the meaning of the values that have been obtained. (Note: this review is derived, in part, from a series of recent papers on this topic; Flood et al., 2020; Swartz, Flood, et al., 2020; Swartz, Vaupel, et al., 2020.)

Instead, all too often, when a measurement technique has indicated that the level of O2 in a given tissue is “X,” that is, is some specific quantitative number for the O2 in the tissue, researchers, and clinicians alike assume that “X” is a reliable, accurate representation of the “true” oxygenation status of the tissue.

This approach ignores the complexity and dynamics of O2 in living biological systems. The reality is that any O2 measurement has been taken at only one point in time of a distribution in the subvolume that was interrogated by the method, while the O2 is in fact varying with time and across space in the tissue and is unlikely to be uniform in the volume that is being interrogated.

In this review we focus on the biological/clinical meaningfulness of O2 measurements made in living organisms, while recognizing that tissue O2 is in constant flux. We emphasize that, to obtain maximum clinical utility of the measurements, it is necessary to consider the goal of the measurements and the limitations of the data that are obtained. We particularly focus on the clinical value of making repeated measurements of O2, especially in association with strategies/events that potentially change O2 levels.


Physical concepts and terminology for reporting on oxygen levels in tissues

The level of molecular oxygen, that is, O2, is usually reported as partial pressure of oxygen (pO2) or concentration of oxygen ([O2] or cO2). These terms have physically rigorous meanings that can usefully be extended to describe gases (such as O2) that are dissolved in liquids or solids, including tissues. Partial pressure is the pressure exerted by oxygen in a mixture of gases, while concentration is the content of oxygen in the gas mixture or solid. Partial pressure is commonly expressed in mmHg, and these units are sometimes referred to as torr or kPa (SI unit used in the EU), while concentration of O2 is commonly expressed in mL of O2 per 100 ml, for example, in blood.

However, the solubility of oxygen varies greatly in different media (Bennett, Swartz, Brown, & Koenig, 1987; Jordan et al., 2013) and this affects the relationship between pO2 and [O2]. The transport of O2 across lipid membranes is known to depend on both diffusion and solubility in the bilayer, and to be affected by changes in the physical state and by the lipid composition, especially the content of cholesterol and unsaturated fatty acids.

For example, because O2 partitions preferentially into lipophilic media, such as membranes, the solubility of O2 in membranes is about four times greater than in aqueous solutions (Möller et al., 2016). This difference has significant consequences for physical and chemical interactions involving molecular oxygen in biological systems because these interactions depend on the number of oxygen molecules that are present and their rate of diffusion.

Does it matter clinically to know whether the technique is reporting [O2] or pO2? While these measures are not identical, there is a known relationship between them. According to the ideal gas law, pO2 is directly proportional to concentration, assuming the volume and temperature are constant, that is,

PV = nRT

where P = pressure (pO2); V = volume; N = amount of substance [O2]; R = ideal gas constant; T = temperature.

It is less straightforward in biological systems. If the solubility of oxygen in each component of tissue is known (and this is not always fairly readily derived experimentally) and pO2 can be measured, then it is relatively straightforward to calculate [O2]. Conversely, if the component in which [O2] is measured and the solubility of O2 in that compartment is sufficiently known, then it should be feasible to determine pO2. However, it often is not feasible to measure these parameters readily.

Because of the biological complexities in assessing O2 in tissues, each reported measure of O2 level in a tissue can be better considered as an average value. Average is used here in its more colloquial usage rather than as a statistically defined term, because different techniques output their measurement of O2 using differing methods pertinent to that technique. Each technique gathers information from a particular volume of tissue (irrespective of whether that volume is well characterized), which we refer to hereafter as the “interrogated volume.”

The sampling of data within the interrogated volume is then used to produce a measure based on a sort of average O2 within that volume. Characterizing that “average measure” is made difficult both by the imprecision of knowing the exact volume queried, but also because the detection of O2 in that volume may be affected by factors such as its distance from the detector or, for optical techniques, different rates of scattering (that also may not be well characterized). Hence, we conclude that it is important to bear in mind that the measurements of O2 in vivo are fundamentally based on a sort of “average” within the interrogated volume.

Because of all of these challenges in obtaining precise measurements of relevant parameters necessary to assess whether the data obtained by any technique is truly measuring either pO2 or [O2] and because of the imprecisions of knowing the volume being assessed and the tissues within that volume, it is more realistic to acknowledge the complexity of these issues for in vivo measurements in tissues by using a less precise term for measures of molecular oxygen in tissues such as “O2 levels,” which is the convention we follow in this review. We also argue that, while it is important to recognize the biological imprecisions in these measures, there are still many clinically viable uses of this information (such as assessing change in O2 levels).

Note too, within this paper, while focusing on the uncertainties due to sampling issues of each technique and variations due to biological factors, we are not taking into consideration further uncertainties due to inevitable instrumental noise, variations in the placement of the detector, etc.

Expected levels of O2 in tissues

Table 1 presents some illustrative data on the O2 levels in various tissues, both in normal states and as altered by some diseases. (These measures are presented here as reported in the literature.) The first column presents the median pO2 obtained, using the Eppendorf electrode (or comparable polarographic techniques) to measure O2 levels in patients; also presented are two other indications of O2 levels: the hypoxic fraction (the percentage of measurements in a given type of tissue that is below a defined “hypoxic level,” in this case 2.5 mmHg) and the range of pO2 values found experimentally.

These data illustrate both the variation in median O2 levels between types of tissues and how they may vary with physiology or disease. For example, (intertissue) in general the median O2 levels are lower in skeletal muscle and heart compared to the spleen; (intratissue) the median O2 in skeletal muscle at rest is higher than with exercise while, in contrast, there is almost no variation between the normal spleen and with Hodgkin’s disease.

The data in Column 4 illustrate the wide range that any given measurement can have in the “same” type of tissue, that is, almost all tissues range from ~0 to ~100 mmHg, even when their median value is quite different (see again spleen vs. bone). However, these very high values in the upper range may include experimental artifacts due to the measurement being taken within or very close to an arteriole, for example, a vessel feeding the microcirculatory bed.

Table 1

Oxygenation status of organs/tissues

Organ/tissueMedian pO2 (mmHg)HF 2.5 (%)pO2 range (mmHg)References
Cortex45–501–21–97#Günther, Aumüller, Kunke, Vaupel, and Thews (1974)
Outer medulla382–52–96Same
Inner medulla119–110–32Same
Liver25–301–21–96Kallinowski and Buhr (1995a)
Pancreas5721–95Koong et al. (2000)
Normal681–22–96Vaupel, Wendling, Thomé, and Fischer (1977)
In hypersplenism6922–97Wendling, Vaupel, Fischer, and Brünner (1977)
Hodgkin’s disease6722–96Same
Subepicardial18–2611–96Winbury, Howe, and Weiss (1971)
Subendocardial10–17n.a.1–94Moss (1968) (for both)
Oral5211–96Kallinowski and Buhr (1995a)
Large bowel55n.a.1–95Same
Normal65010–96Vaupel, Schlenger, Knoop, and Höckel (1991), Vaupel and Harrison (2004)
Fibrocystic disease6705–98Same
Prostate2641–96Vaupel and Kelleher (2013)
Uterine cervix4181–97Höckel, Schlenger, Knoop, and Vaupel (1991)
Cortical3230–96Spencer et al. (2014)
Hematopoietic marrow2210–95Same
Adipose marrow2620–95Same
Skeletal muscle
Resting27–320–20–96Landgraf and Ehrly (1980)
Exercise105–100–96Jung, Kessler, Pindur, Sternitzky, and Franke (1999)
Hypovolemic shock4400–40Harrison and Vaupel (2014)
PAOD6–7~300–90Landgraf, Schulte‐Huermann, Vallbracht, and Ehrly (1994)
Thermoneutral conditions25–35n.a.40–70Carreau et al. (2011)
Critical limb ischemia5–8180–96Harrison and Vaupel (2014)
Limbs, venous disease15n.a.40–65Clyne, Ramsden, Chant, and Webster (1985)
Gray matter2811–96Vaupel (1994)
White matter10–15n.a.n.a.Same
Retina~20n.a.0–70Hogeboom van Buggenum, van der Heijde, Tangelder, and Reichert‐Thoen (1996), Linsenmeier and Zhang (2017)
White adipose tissue
Nonobese56n.a.40–74Pasarica et al. (2009), Hodson (2014)
Obese47n.a.29–63Lempesis, van Meijel, Manolopoulos, and Goossens (2020)
Abbreviations: #: arterial; HF2.5: hypoxic fraction ≡ fraction of pO2 values ≤2.5 mmHg; n.a.: information not available; PAOD: peripheral arterial occlusive disease.

Table 2 presents the same types of information for a more detailed analysis of changes in an important pathology, cancer, where O2 levels have been an especially important focus for informing clinical treatment and prognosis. To give the reader a sense of how well supported the numbers are, the data in Table 2 have been ordered by the number of patients included in each row.

Table 2

Pretherapeutic oxygenation status of human tumors

Tumor type (ordered by no. of patients)No. of patientsMedian pO2 (mmHg)HF 2.5 (%)pO2 range (mmHg)References
Cervix cancer7309–10280–88Vaupel et al. (2007)
Head and neck cancer59210210–90Vaupel (2009)
Prostate cancer4387260–95Vaupel (2011)
Soft tissue sarcoma28314130–96(data synopses)
Breast cancer21210300–95These 3 ref. apply to all
Glioblastoma10413260–50Above the line
Vulvar cancer5411250–92Vaupel, Thews, Mayer, Höckel, and Höckel (2002), Vaupel, Mayer, and Höckel (2006), Stone et al. (2005)
Rectal cancer2925n.a.0–92Kallinowski and Buhr (1995a), Mattern, Kallinowski, Herfarth, and Volm (1996)
Lung cancer2616130–95Falk, Ward, and Bleehen (1992), Le et al. (2006)
Malignant melanoma (metastatic)181250–96Lartigau et al. (1997)
Non‐Hodgkin’s lymphoma818360–92Powell et al. (1999)
Pancreas cancer82590–91Koong et al. (2000), Graffman, Bjork, Ederoth, and Ihse (2001)
Brain metastases510260–87Rampling, Cruickshank, Lewis, Fitzsimmons, and Workman (1994)
Liver metastases46n.a.0–90Kallinowski and Buhr (1995a, 1995b)
Renal cell carcinoma310n.a.0–90Lawrentschuk et al. (2005)
Gall bladder cancer14n.a.0–10Graffman et al. (2001)
Bile duct cancer18n.a.0–15Graffman et al. (2001)
Abbreviations: HF2.5: hypoxic fraction ≡ fraction of pO2 values ≤2.5 mmHg; n.a.: information not available.

Of interest here, all seven cancer types with at least 50 patients studied have a fairly consistent and fairly hypoxic median O2 level (~10 mmHg). Similarly, all but soft tissue sarcoma have a similar hypoxic fraction, ~25%; (sarcoma appears to be about half that).

Glioma appears to be an outlier on the range; glioma had no patient whose O2 level was above 50 while all others (as was true for the tissues in Table 1) have at least one measurement in the upper 80s or 90s.

These occasional high readings are not surprising since it is plausible that, randomly, some readings will have been obtained in or very near to arterioles/feeding microvessels. In contrast to the first seven cancers, the 10 types of cancers with fewer than 30 patients appear to be more varied in their O2 levels, but this is possibly due to being based on few patients.

Nevertheless, even though (as noted elsewhere in this review), the data presented are not unconditionally “absolute values” of O2 levels (as they are sometimes referred to; e.g., Koch, 2002; Macnab, Gagnon, Gagnon, Minchinton, & Fry, 2003). Nevertheless (as argued in this review), they can provide very useful data as long as their limitations are recognized by researchers and clinicians.

Impact of pathology on heterogeneity of O2 levels

The presence of pathology often significantly increases the amount and extent of oxygen heterogeneity both spatially and temporally (Vaupel & Harrison, 2004; Vaupel & Mayer, 2016). The presence of pathology often impacts the structure/morphology of the vessels. In tumors there often is a significant amount of neoangiogenesis which results in much less ordered and less functional blood vessels (Busk et al., 2020).

The resulting vessels are much less efficient in delivering blood and also tend to be much more prone to leak. Leakage from these vessels can cause increases in the interstitial pressure, which can reduce the effectiveness of the microcirculation due to reduction of the perfusion pressure within the tumor capillaries (Fukumura, Duda, Munn, & Jain, 2010).

Pathological changes also can result in altered consumption of O2. In malignant tumors O2 consumption is likely to decrease due to poor oxygen delivery and/or because of a switch to glycolysis due to metabolic reprogramming (i.e., the Warburg effect) as a consequence of HIF‐1α overexpression, upregulation of oncogenes, downregulation of suppressor genes, and activation of certain signaling pathways (Vaupel & Multhoff, 2020; Vaupel, Schmidberger, & Mayer, 2019).

Pathology can also impact the integrity of the blood vessels. For example, tumor growth may physically impinge on the integrity of the blood vessels, and the metabolic abnormalities in diabetes can impact the structure of blood vessels (causing either microangiopathy and/or macroangiopathy). The results of these processes can produce very significant local variations in the availability of fully functional vascular structures, resulting in locally hypoxic regions.

Pathology also can impact temporal changes of oxygen and the response to treatment. The presence of pathology, especially cancer (e.g., acute and cycling hypoxia in cancers) and peripheral vascular disease, can result in significantly greater variability in O2 levels (Braun, Lanzen, & Dewhirst, 1999). These include short‐term changes, especially associated with the structural abnormalities of the microcirculation resulting in increased local variability in flow, and long‐term changes that develop over time, such as those due to disease progression and responses to therapy (Baudelet et al., 2004; Kimura et al., 1996; Konerding, Fait, & Gaumann, 2001; Matsumoto et al., 2006).

There also is a potential for pathologies to interrelate with each other. For example, anemic hypoxia can develop in tumors due to the underlying systemic anemia of the patient (Vaupel & Mayer, 2014).

In addition to these underlying effects of pathology on tissue oxygen levels, any applied therapeutic interventions are very likely to induce changes. For example, cell killing due to radiation or chemotherapy will alter oxygen consumption patterns. These same therapies will also affect the O2 supplying vasculature via both antiangiogenic effects and—possibly—normalization of vessel structure (Jain, 2005). The effects of therapies will generally vary both spatially and temporally, reiterating the complexity of meaningfully characterizing tissue oxygen levels.


Although many techniques are often thought to measure actual O2 in tissues, only a few actually have the potential to make O2 measurements directly in the tissues of interest (Springett & Swartz, 2007; Tatum et al., 2006). Techniques that can potentially assess O2 directly in tissues include: EPR (Epel et al., 2019; Swartz et al., 2014; Swartz et al., 2014), the Eppendorf electrode (Vaupel, Höckel, & Mayer, 2007), some optical methods based on direct measurements of target molecules in tissues, for example, phosphorescence quenching of optical sensors placed directly in tissues or as part of a physical probe such as the “OxyLite” (Wen et al., 2008), and NMR relaxation techniques (Colliez et al., 2014).

Two other types of measurements assess O2 in the vascular system. Blood gases do this directly, while optical methods that measure both hemoglobin saturation and total hemoglobin (especially near infrared spectroscopy [NIRS]; Scheeren, Schober, & Schwarte, 2012) provide a plausible link to the pO2 in the blood.

However, the techniques most often used clinically to characterize tissue oxygenation do not in fact measure O2 directly; instead, they measure “indirect” parameters that can be plausibly linked to actual O2 levels but only under appropriate/defined circumstances. This latter group of techniques includes positron emission tomography (PET) imaging of glucose derivatives (Neveu et al., 2015), PET imaging of drugs that localize in hypoxic tissues (Tran et al., 2012), laser Doppler flow, measures of metabolites that may be affected by O2 levels, for example, lactate and redox intermediates, and several magnetic resonance imaging/nuclear magnetic resonance (MRI/NMR), blood oxygenation level dependent (BOLD) imaging (Baudelet & Gallez, 2002), and MRI (Egeland et al., 2006). Note: If their basis is understood and the data considered accordingly, these can all provide clinically and physiologically useful information even though they do not provide direct information on the amount of O2 in the tissues.

Direct measures of O2 in targeted tissues that potentially can be used in human subjects
These are techniques that, while they have the capability of providing direct quantitative measurements of O2 in homogeneous media, cannot provide such data in tissues in vivo because the volumes that they sense are larger than the volumes of homogeneity of O2 in actively metabolizing tissues. Consequently, all in vivo measurements of O2 are inherently averages of the actual oxygen content in that volume. Even neglecting the need to include measures of heterogeneity inside cells, based on the usual volume of cells and assuming that differences are sought for aggregates of ≤3 cells, for a measurement of heterogeneity sensed within a 10 mm diameter volume, the spatial resolution needed to appropriately characterize O2 levels in this volume becomes 8 million voxels. The measuring techniques may not even provide a well‐defined averaged pO2 value within the volume that they sense. For example, sensors for the signal that is being measured.

In the next sections we review the characteristics of each technique that can directly measure O2 levels in tissues. We also briefly remark on the volumes they measure and how the measures obtained can be useful clinically (see also Ortez‐Prado, Dunn, Vasconez, Castillo, & Visco, 2019).

EPR oximetry
Using appropriate particulate paramagnetic materials, EPR oximetry can provide direct measurements of O2, that is, the EPR signal is directly proportional to the amount of O2 (Epel, Bowman, Mailer, & Halpern, 2014; Swartz, Vaupel, et al., 2020). Because each multisite sensor senses a volume that is much larger than capillary networks, these techniques provide a volume averaged sampling of all compartments within the tissues. The time resolution of the techniques can be milliseconds or shorter.

The measured parameter of an EPR spectrum that indicates the amount of O2 present is the line width of the observed resonance peak. There usually is a fixed relationship between the line width and the amount of O2, with the relationship being specific for each type of paramagnetic material, for example, carbon, charcoal, or phthalocyanine particulates. Using particulate oximetric materials, measurements can be continuous over any span of time and can be repeated indefinitely (see example in Figure 2).

The method requires that the sensing material be injected or implanted in one or more regions of interest, but thereafter all measurements can be made entirely noninvasively. Importantly the measurements can be carried out in a clinical setting and can fit into the workflow needed for patient care.

The initial clinical EPR measurements of oxygen in tissues have used India Ink as the oxygen sensor (Swartz et al., 2014). The carbon particles are the components that respond to oxygen (Lan, Beghein, Charlier, & Gallez, 2004). After injection of 30–50 µl of the suspension through a small needle, the carbon particles disperse nonuniformly through the local region as small extracellular aggregates. T

hey are often engulfed by macrophages. The resulting EPR spectra in the region probed by the resonator (i.e., the surface coil used for signal detection) are a composite of the oxygen‐dependent line widths from each of the particles. In reality, because of the relatively broad lines from the India Ink particles, the range of “oxygen levels” that are likely to be present in the tissue, and the limited number of particles in each subregion, it is a challenge to resolve directly even the major groups of similar line widths. Therefore, using the observed line width to provide a quantitative measure of oxygen would seem to have modest utility in itself.

The other method of clinical EPR oximetry is based on the use of micro‐crystalline probes (e.g., LiPc, LiNc‐BuO), encapsulated in biocompatible polymers (Swartz et al., 2014). Clinical measurements currently are being performed using the “OxyChip” which consists of oxygen sensitive microcrystals of lithium octa‐n‐butoxynaphthalocyanine (LiNc‐BuO) embedded in polydimethylsiloxane (PDMS; Hou, Khan, Gohain, Kuppusamy, & Kuppusamy, 2018; Hou et al., 2016; Jarvis et al., 2016).

The dimensions currently used in humans are cylinders that are 5 mm long with a diameter of 0.6 mm. The EPR signal from the sensor (OxyChip) reflects the pO2 within the PDMS, which itself reflects an average of the pO2 in contact with the external surface of the cylinder.

The dimensions of the OxyChip are much greater than those of a microcirculatory unit and therefore reflect many different such units. The microcirculatory networks sampled by the OxyChip are likely to include regions with quite different O2 levels. The “values” of oxygen that are obtained therefore do not, in themselves, provide any information on the heterogeneity of the sampled capillary networks.

In contrast to EPR spectroscopy based on particulates with stable free radicals, EPR oximetry imaging can provide spatially resolved measurements using soluble paramagnetic materials with stable free radicals, especially nitroxides and trityls (Epel et al., 2014; Epel et al., 2019). An advantage of pulse techniques using spin lattice relaxation rate images of dissolved O2 spin probes is the virtual absence of confounding variation of the spin from the probe affecting the measurement of the effect of O2 on relaxation rates.

This is of particular importance for low oxygenation levels where the spin probe concentration and its effect on the relaxation rate approaches that of O2 (Epel et al., 2019). Nonetheless, the volumes of even the smallest resolvable voxel in the images will still be too large to avoid averaging of the O2 levels sensed within each voxel.

The time resolution is usually at least several minutes. This technique cannot provide repeated measurements without readministering the paramagnetic material each time. At this time, this technique is available only for preclinical use, because there are no FDA‐approved soluble paramagnetic materials.

The Eppendorf electrode
The Eppendorf electrode has been used clinically to provide direct tissue pO2 measurements along electrode tracks. To create a track, a series of points is obtained by progressing the 200 µm microelectrode through the tissue in a sequence of “Pilgrim steps,” that is, the probe is advanced a prescribed distance and then withdrawn a fraction of that distance to minimize pressure effects. The volume sampled by each point is estimated to be 100–500 cells around the tip of the probe (Vaupel et al., 2007).

Therefore, the measurement usually reflects the average of a range of pO2 values, especially in the presence of pathophysiology. The histogram of values of a tumor was thought to be a representative sample of three to seven tracks through the volume, although this still was only a sampling of the true heterogeneity of the tissue O2. Nevertheless, in tumors, very useful clinical correlations have been found with the number of points below a threshold value, for example, median pO2 or hypoxic fractions (Vaupel & Mayer, 2007).

Note that this does not require individual, true absolute pO2 values. Instead, the separation between severely hypoxic and less hypoxic values relies on averaged oxygen tensions. When used with multiple point measurements (>70), this is a good but not infallible technique to determine the presence of hypoxic regions. Regardless of its potential advantages, unfortunately this technique is no longer available clinically. This technique also had some practical limitations, including difficulties to repeat the measurement in a given tissue subvolume because of the local trauma produced from the preceding measurements.

Currently the only version of the oxygen electrode suitable for use in humans is a technique requiring temporary implantation of an oxygen electrode that extends through the skin and skull and is used to monitor severe brain trauma. Such a device, while very useful for its intended purpose, is not suitable for measurements in tumors (Stewart et al., 2008).

Optical methods
The OxyLite technology is based on quenching by O2 of fluorescence with a sensor whose diameter is 230–750 µm at the end of a fiber‐optic fiber (User Manual, 2020). This cross‐section corresponds to the diameter of a tissue subvolume of 10–75 mammalian cells. And, therefore, the sensor will provide an average of the distribution of pO2 values throughout this volume. The temporal resolution can be quite rapid. It could be used in a manner similar to the Eppendorf electrode to obtain a series of similarly averaged measurements at different locations.

Direct injection of phosphorescent agents can provide one way to directly sample tissue O2, where the signal comes from the phosphorescence lifetime changes that result from excited triplet state quenching of O2 (Wilson, Harrison, & Vinogradov, 2012). High resolution mapping with very fast time resolution can be provided, although care must be taken in choosing the probe that localizes in the compartments of interest, as some provide intracellular information, some provide purely extracellular information, and some are simply perivascular in nature (Esipova et al., 2011).

A key part of advancing these molecular probes has been to ensure that they sample the oxygen in the environment with a buffer around them such that the oxygen diffuses into the sensor and any measurement of oxygen by the triplet state quenching does not alter the local oxygen level, making the signal potentially nonlinear. Many unprotected or bare nanoparticle probes can have a signal dependence which is not ideal for linearity with oxygen or repeated measurements, or they could have uncertain localization.

Dendrimer particles of the Oxyphor complex have been used as biocompatible large particles with pegylation to the exterior to ensure a known biodistribution (Esipova et al., 2019). Their localization is largely extracellular as well, but of course the localization can be tailored by specifically altering the surface chemistry, and they have been used to create probes for many unique biological environments such as the gut, bone marrow, or brain.

One of the more promising methods for oxygen measurement in vivo has been through delayed phosphorescence from protoporphyrin IX (Mik et al., 2006), which is present in all tissues to some extent, produced in the mitochondria, and can be proactively enhanced by administration of aminolevulinic acid. This molecule has triplet state quenching by oxygen as well, and the reverse intersystem crossing that can occur in the absence of oxygen allows for a delayed fluorescence signal which is uniquely sensitive to the local oxygen environment. This work was pioneered in cardiac tissues by Mik (2013) and shown to be a logical way to sample tissue oxygen (Scholz, Cao, Gunn, Brůža, & Pogue, 2020).

An important aspect of all these optical methods is that the depth of measurement is dominated by the light input/output geometry, which is typically limited to just a millimeter or two of depth for diffuse illumination and detection. Microscopy studies are widespread as they lend themselves well to thin tissues and take advantage of coherence or optical confinement tools. Thicker tissues are less studied, as this regime is more dominated by the tissue scattering nature than the features of the light signals.

The development of x‐ray induced Cherenkov‐excited luminescence methods (which concomitantly use a specially developed porphyrin Oxyphor PtG4, developed by David Wilson and Sergei Vinogradov, University of Pennsylvania) have been shown to be useful for imaging through a few centimeters of tissue.

This takes advantage of the deeper penetration of x‐rays. It samples all emissions from the tissue, reconstructing the detectable light. The reconstruction then infers the site of the origin of the light where the radiation is absorbed and therefore which emits Cherenkov to the excited sensor molecule. This measurement therefore provides information on oxygen at that site (Pogue et al., 2018).

This technique can provide excellent spatial resolution but more development of the technique is needed in order to take advantage of the benefits of spatial resolution. More generally, the benefits of optical methods come largely from their ability to tailor the imaging system to the problem. Currently, however, the logistics have not been solved for routine human use, nor have the probes been made in an FDA‐approved format for use in humans.

Photoacoustic tomography (PAT) is another optical technique that is becoming increasingly used to define perfusion and O2 saturation in preclinical cancer and investigational studies of breast tumors, with ongoing commercial development for broader clinical use in a number of ventures. The method relies on optical absorption of a pulsed laser source by chromophores in the tissue, including hemoglobin, to produce ultrasound waves which are detected at the surface.

The major benefits of PAT in lesion imaging are high resolution for neovasculature imaging through ~1–2 cm of tissue, and the ability to image features of blood oxygen saturation with spectroscopic PAT. Quantitation of tissue oxygenation using PAT remains difficult because an indirect indicator of oxygenation is measured (i.e., blood oxygen saturation) and unknown tissue properties skew the light penetration in tissue, making accurate spectroscopic measurements challenging at nonsuperficial depths (Cox, Laufer, Beard, & Arridge, 2012).

NMR relaxation methods
Based on the principle that molecular O2 impacts the relaxation time of nuclei (Bennett et al., 1987; Swartz et al., 1985), several different approaches have been developed to try to use the power of NMR to measure O2 in tissues (Matsumoto et al., 2006). The measurement of O2 by NMR using isotopes with spin (O17 for NMR; Zhu & Chen, 2011) which is measured when the O2 is incorporated into water (and hence may better be considered as a measure of metabolism (Gallez, Neveu, Danhier, & Jordan, 2017; Neveu et al., 2015) and O15 for PET (Hattori et al., 2004) is possible, but these are extraordinarily expensive and not suitable for routine clinical use.

The most developed method has been the use of the relaxation time of fluorine nuclei in fluorinated hydrocarbons injected directly into tissues (Liu et al., 2011; Zhao, Jiang, & Mason, 2004). The measurements are based on the relaxation time of the F atoms in the emulsion, but the volume of the fluorine containing hydrocarbon is much larger than several capillary networks. Consequently, the data obtained are an average of a range of partial pressures. To date this approach has not advanced to clinical use.

There are a number of other NMR‐related approaches that have been suggested using the impact of O2 on the relaxation of protons (Colliez et al., 2014; O’Connor et al., 2009, 2016). Potentially these might provide spatially resolved data that could be quite useful. However, it is very unlikely that any could have the resolution to resolve the spatial heterogeneity of O2 in tissues. Thus, these measurements of O2 will be similar to the other techniques discussed, that is, their data will be based on assessing changes in “averaged” O2 levels.

Measures of O2 in the vascular system
Measures of O2 in the vascular system are based on the known relationship between the O2 saturation of hemoglobin (Hb) and the ambient pO2 as shown in the sigmoidal oxygen dissociation curve. However, the vascular system is, by its nature, not uniform with respect to its saturation. Arterial and venous systems have very different saturations, with the capillaries (which is where most O2 is delivered to the tissue) being in between these extremes. Therefore, when measurements such as pulse oximetry or NIRS report HbO2 saturation (Sakata, Grinberg, Grinberg, Springett, & Swartz, 2005), it is important to understand what part of the vascular system is being measured.

The problem with applying measurements of oxygen made in blood to measurements in tissues is that, while the amount of O2 in the vascular system is an important parameter impacting O2 levels in the perfused tissues, it is not possible to go directly from measuring O2 in blood to knowing the O2 in tissues. That is because O2 in the tissue is determined by many factors in addition to the potential supply of O2 from the red blood cells.

Perfusion and diffusion must occur to get into the tissues. In some tissues, such as skeletal muscle and myocardium, the situation is even more complicated because of binding of oxygen by myoglobin. For example, some consumption of oxygen by cells will occur between the vascular system and the tissues of interest, and the number of cells and level of oxygen consumption by the cells can vary greatly. If tumors are present, the situation is further impacted by the chaotic nature of the neovasculature that arises in the tumors, with poorer perfusion and abnormal diffusion out of the vessels (Collins, Rudenski, Gibson, Howard, & O’Driscoll, 2015).

The other type of measurement of O2 in the vascular system is the measurement of blood gases (Davis, Walsh, Sittig, & Restrepo, 2013). These are usually done in larger vessels. These measurements have the same limitations as NIRS in regard to being able to provide information on the O2 levels in the extravascular tissue compartment.

Even though they are not measuring the actual increase of oxygen in tissues, measurements in the vascular system may be helpful in assessing response to techniques to increase O2 availability, because of ease of use and potential depth of sampling (Kim & Liu, 2008; Sunar et al., 2006).

While these tools are volume averaging over centimeters of tissue and the signal is dominated by areas of higher perfusion, the data on changes can provide clinically valuable information indicating that the oxygenation technique could be effective in raising levels of oxygen in the tissues of interest.

A related methodology, which is widely available, BOLD MRI (Thulborn, Waterton, Matthews, & Radda, 1982), is sometimes mistakenly thought to provide a direct measurement of O2 in the vascular system. But because it provides only one parameter, the amount of DeoxyHb, it cannot be used to assess HbO2 saturation or pO2 in the circulatory system.

In principle, changes in the amount of the BOLD signal could be used to provide an indication of changing O2 levels. However, this type of use is limited by the fact that, if the blood volume within the sampled voxels changes, the amount of DeoxyHb will change even if the saturation remains the same and, therefore, the results may be confounded (Baudelet & Gallez, 2002). The potential time resolution is very fast.

Other indirect measures of tissue oxygen
There are a number of methods used clinically that attempt to provide indications of O2 levels in tissues, which are based on plausible but indirect relationships to the actual O2 levels. While these therefore intrinsically do not directly measure O2 in tissues, because of their widespread availability and, perhaps, some naivety among clinical users, they are sometimes considered to provide direct measurements of O2.

This confusion has important implications for their usefulness in clinical applications because, while these techniques may provide valid and quantitative measures of parameters associated with their biological or chemical modes of interaction, their quantitative relationships to O2 levels may be poorly defined and potentially misinterpreted.

The use of molecules such as the nitroimidazoles and Cu‐ATSM that selectively localize in hypoxic tissues (O2 content <1%) has been widely employed, most often using PET labels to indicate their location (Busk et al., 2020; Chapman, Franko & Sharplin, 1981; Gutfilen, Souza, & Valentini, 2018; Kelada et al., 2017; Tran et al., 2012).

The principle is that with critically low levels of O2 these can be reduced to reactive intermediates which, if not reoxidized by O2, can then bind to cellular components (Sealy, Swartz, & Olive, 1978). While these probes are widely used and often have been clinically useful, they clearly cannot provide a direct measure of the O2 content of the tissue. They can, nonetheless, provide a qualitative indication of regions with moderate to good perfusion that were hypoxic at the time that the tracer was delivered. When the nature of the data is understood, these techniques can be quite useful.

Another widely used PET clinical imaging technique is based on the observation that many tumors have a high rate of anaerobic and aerobic glycolysis and therefore a high uptake of glucose, which in turn can be followed by using the imaging agent 18F‐FDG (18F‐fluorodeoxyglucose; Lopci et al., 2014).

The amount of uptake is based on a number of different factors including perfusion rate, extent of hypoxia, expression of the Warburg effect (i.e., aerobic glycolysis), number of cells, uptake of glucose analogs and rate of glycolysis (Vaupel & Multhoff, 2020; Vaupel et al., 2019). Clinical use of this technique is widespread and considered clinically useful for a number of contexts, especially in identifying regions where metabolically active tumor cells are located.

Some clinicians even consider areas of high uptake as potentially indicating tumor hypoxia although the Warburg effect is mainly responsible for the high glucose uptake rates (Vaupel & Multhoff, 2020). A major complication is the high rate of false positives, due to many other causes of uptake of the agent (Britton & Robinson, 2016; Metser & Even‐Sapir, 2007). There can be very significant discrepancies between the pattern of localization of FDG and the histological evidence of hypoxia (Christian et al., 2009).

Endogenous hypoxia markers and other surrogates
Endogenous hypoxia markers (e.g., HIF‐1α, GLUT‐1, CA IX) are often used to judge the oxygenation status of malignant tumors. Responses to low oxygen levels in normal and cancer cells are mainly initiated by HIF‐1α, a key regulator for genes responsible for mammalian oxygen homeostasis.

These responses (among others) include an increased erythropoiesis, angiogenesis, and glycolysis. The latter is mainly initiated by an enhanced cellular glucose uptake through GLUT‐1 transporters (Semenza, 1999, 2000). However, the expression of these markers is also driven by a series of other parameters as described in Section 3.2. Therefore, these proteins per se are not useful at all to judge the oxygenation/hypoxia status of benign and malignant tumors. The respective problems have been discussed in detail by Mayer, Höckel, and Vaupel (2008).

There is also a group of physiological measurements that are sometimes indirectly linked to O2. Measurements of blood flow, for example, MRI perfusion (Essig et al., 2013) and laser Doppler flow (Briers, 2001) are often used clinically to obtain a (surrogate) parameter that is potentially linked to the supply of O2.

Because of the many factors that affect how much O2 is delivered and the impact of utilization on the amount of O2 is available, these techniques cannot by themselves provide reliable insights into the oxygenation status of the tissues. Measurements of metabolites that may be affected by O2 levels, for example, lactate and redox intermediates, are frequently used. However, their levels are the result of complex interactions that depend on many factors in addition to the amount of available O2, and therefore cannot reliably indicate O2 levels in tissues. Nonetheless, these physiological measurements, especially if repeated and related to appropriate other parameters, may in some instances provide useful clinical information.

reference link :

More information: Haley Marks et al. A paintable phosphorescent bandage for postoperative tissue oxygen assessment in DIEP flap reconstruction. Science Advances  18 Dec 2020: Vol. 6, no. 51. DOI: 10.1126/sciadv.abd1061


Please enter your comment!
Please enter your name here

Questo sito usa Akismet per ridurre lo spam. Scopri come i tuoi dati vengono elaborati.