Advertisement

The First-in-Man “Si Se Puede” Study for the use of micro-oxygen sensors (MOXYs) to determine dynamic relative oxygen indices in the feet of patients with limb-threatening ischemia during endovascular therapy

      Objective

      Patients with limb-threatening ischemia exhibit uneven patterns of perfusion in the foot, which makes it challenging to determine adequate topographic perfusion by angiography alone. This study assessed the feasibility of reporting dynamic relative oxygen indices and tissue oxygen concentration from multiple locations on the foot during endovascular therapy using a novel micro-oxygen sensor (MOXY; PROFUSA, Inc, South San Francisco, Calif) approach.

      Methods

      A prospective, 28-day, single-arm, observational study was performed in 10 patients who underwent endovascular therapy for limb-threatening ischemia. At least 24 hours before therapy, four microsensors were injected in each patient (one in the arm, three in the treated foot). The optical signal from the microsensors corresponded to tissue oxygen concentration. A custom detector on the surface of the skin was used to continuously and noninvasively measure the signals from the microsensors. The ability to locate and read the signal from each injected microsensor was characterized. Oxygen data from the microsensors were collected throughout the revascularization procedure. The timing of therapy deployment was recorded during the procedure to assess its relationship with the microsensor oxygen data. Oxygen data collection and clinical evaluation were performed immediately postoperatively as well as postoperatively on days 7, 14, 21, and 28.

      Results

      The study enrolled 10 patients (50% male) with ischemia (30% Rutherford class 4, 70% Rutherford class 5). Patients were a mean age of 70.7 years (range, 46-90 years), and all were Hispanic of varying origin. Microsensors were successfully read 206 of 212 times (97.2%) in all patients during the course of the study. Microsensors were compatible with intraoperative use in the interventional suite and postoperatively in an office setting. In nine of 10 revascularization procedures, at least one of the three MOXYs showed an immediate change in the dynamic relative oxygen index, correlating to deployed therapy. Moreover, there was a statistically significant increase in the concentration of oxygen in the foot in preoperative levels compared with postoperative levels. No adverse events occurred related to the microsensor materials.

      Conclusions

      This MOXY approach appears to be safe when implanted in patients with limb-threatening ischemia undergoing endovascular recanalization and is effective in reporting local tissue oxygen concentrations over a course of 28 days. Further testing is needed to determine its potential effect on clinical decision making, both acutely on-table and chronically as a surveillance modality, which ultimately can lead to improved healing and limb salvage.
      The estimated global prevalence of peripheral artery disease increased 23.5% from 2000 to 2013, rising to an estimated 202 million cases.
      • Fowkes G.
      • Rudan D.
      • Rudan I.
      • Aboyans V.
      • Denenberg J.
      • McDermott M.
      • et al.
      Comparison of global estimates of prevalence and risk factors for peripheral artery disease in 2000 and 2010: a systematic review and analysis.
      There has been a parallel increase in percutaneous lower extremity interventions. In the United States alone, the percentage of percutaneous interventions increased by 296.8% comparing 1997 to 2008, although the number of amputations has only decreased by 4.6%.
      • Jim J.
      • Owens P.L.
      • Sanchez L.A.
      • Rubin B.G.
      Population-based analysis of inpatient vascular procedures and predicting future workload and implications for training.
      The accelerated rate of endovascular therapy adoption and development of new percutaneous technologies have not been met with consensus-based algorithms that allow physicians to engage in evidence-based clinical practice. Most therapeutic guidelines used to date are based solely on arterial lesion anatomy
      • Norgren L.
      • Hiatt W.
      • Dormandy J.
      Inter-society consensus for the management of peripheral arterial disease (TASC II).
      and fail to include many other important factors such as patient risk factors and frailty, renal function, affected angiosome, extent of the wound, or presence of infection. New classification schemes aim to address some of these concerns and are in the process of being refined and validated.
      • Mills Sr., J.L.
      • Conte M.S.
      • Armstrong D.G.
      • Pomposelli F.B.
      • Schanzer A.
      • Sidawy A.N.
      • et al.
      The Society for Vascular Surgery Lower Extremity Threatened Limb Classification System: risk stratification based on wound, ischemia, and foot infection (WIfI).
      As more experience is gained with advanced endovascular techniques, it is clear that most lesions can be traversed and treated.
      • Montero-Baker M.
      • Schmidt A.
      • Bräunlich S.
      • Ulrich M.
      • Thieme M.
      • Biamino G.
      • et al.
      Retrograde approach for complex popliteal and tibioperoneal occlusions.
      The question remains: What is the therapeutic goal? Some authors have suggested the rather simplistic approach of advocating that the end goal is achieving in-line flow to the foot.
      • Norgren L.
      • Hiatt W.
      • Dormandy J.
      Inter-society consensus for the management of peripheral arterial disease (TASC II).
      A more contemporary approach supports the use of angiosome-guided therapy,
      • Bosanquet D.C.
      • Glasbey J.C.
      • Williams I.M.
      • Twine C.P.
      Systematic review and meta-analysis of direct versus indirect angiosomal revascularisation of infrapopliteal arteries.
      which at least attempts to follow a more physiologic and topographically logical perspective to the act of revascularization. Unfortunately, apart from a bidimensional angiogram, there exists no validated on-table tool to evaluate local perfusion improvement as a means to assess the end-point of interventions.
      The “Si Se Puede” study aims to characterize the safety and feasibility of using a microsensor-based tissue oxygen monitor called MOXY (PROFUSA, Inc, South San Francisco, Calif) to determine the local tissue oxygen concentration in different locations of the foot of patients with severe limb ischemia during and after endovascular revascularization. This report represents the first-in-man evaluation of a potential tool that can provide on-table feedback regarding local optimization of perfusion as well as a means to evaluate long-term perfusion status.

      Methods

      A prospective, single-cohort, phase 1, first-in-man, 28-day, observational safety and feasibility study was reviewed and approved by the CEATE Foundation (Center of Academic Excellence in Endovascular Therapy), a local ethics committee of the San Juan de Dios Hospital (San Jose, Costa Rica). Each patient provided informed consent before any study-related procedure. This was an investigator-initiated study. No funding was provided by PROFUSA, Inc other than the donation of the investigational devices used during the study.

       Study population

      Study candidates who met the following inclusion criteria were recruited: age 18 to 90 years with a history of symptomatic severe lower limb ischemia (Rutherford clinical categories 4-6), planned endovascular revascularization procedure ≤1 week of screening, able to take acetylsalicylic acid and thienopyridine, and a life-expectancy of at least 1 year. Excluded were candidates who were pregnant, premenopausal, had a history of sensitivity to light, keloid formation, dermatitis, known allergy to materials used in the study, or in the investigator's opinion were not suitable for study participation.

       Study protocol

      After study screening and enrollment, patient demographic and clinical characteristics were collected. At least 1 day before the patient's endovascular procedure, the investigator selected one location on the arm and three locations on the foot of the index lower limb to be treated for microsensor injections. The arm sensor was placed in a region of presumed normal perfusion to serve as a reference. The foot sensors were placed in three locations of interest with consideration of their proximity to an ulcer or wound (if present, ∼5-10 mm away from the ischemic border), angiosome anatomy, and their likelihood of presenting a change in oxygenation as a result of the planned revascularization. Sensors were placed in the subcutaneous tissue, ∼2 to 4 mm below the skin surface, and were injected using sterile technique with a custom-made 18-gauge injector. Sensor injection sites were marked and photographed to facilitate their subsequent localization.
      Before the revascularization procedure, each of the four injected microsensors was located using a detector. The portion of the detector in contact with the patient was wrapped in a sterile cover, and once the sensor was located, the detector was affixed to the skin using adhesive tape or elastic wrap, or both, to enable continuous sensor data collection throughout the procedure. After all sensors were located and detectors secured, the nursing staff began sterile preparation for the surgery.
      The investigators performed all endovascular revascularization procedures according to the standard of care defined by the hospital and were blinded to all data acquisition from the implanted sensors. During the procedure, local tissue oxygen concentration data were continuously recorded once every 5 to 15 seconds from all accessible microsensors. The time of each therapeutic step, specifically balloon angioplasty (balloon inflation and deflation) and vascular stenting, was recorded. Serial pulse oximetry measurements and intraoperative angiograms were recorded. After the investigator announced the revascularization procedure was complete, oxygen data collection continued for at least 5 minutes before all detectors were removed from the patient. The manner in which the sensors work is discussed below.
      After revascularization, data from microsensors were collected weekly until the patient's participation in the study was complete (defined as 28 days after sensor injection). At these weekly measurement sessions, investigators evaluated the skin area of the sensor injection sites. Tissue oxygen concentration was measured from each injected sensor by using a reader to locate the sensor, affixing the reader onto the skin with adhesive tape, and continuously recording sensor data for at least 5 minutes. Upon study completion, the sensors remained in situ.
      The investigators and the clinical care team were blinded to all data collected from the microsensors throughout the entire study to avoid any potential treatment bias. The study protocol therefore had no effect on the conduct or the outcome of any subject's revascularization therapy or clinical management.

       Tissue oxygen monitor

      The tissue oxygen monitor consists of an injectable, oxygen-sensitive microsensor, an injection device, and a detector. The microsensor is composed of a biocompatible hydrogel called poly (2-hydroxyethyl methacrylate; pHEMA) and a near infrared (NIR) oxygen-sensitive palladium-benzoporphyrin molecule (Pd-MABP). The microsensor senses oxygen in the body based on the principle of phosphorescence quenching of metalloporphyrins, a well-established technique with excellent sensitivity and specificity to physiologic oxygen.
      • Rumsey W.L.
      • Vanderkooi J.M.
      • Wilson D.F.
      Imaging of phosphorescence: a novel method for measuring oxygen distribution in perfused tissue.
      • Lo L.W.
      • Koch C.J.
      • Wilson D.F.
      Calibration of oxygen-dependent quenching of the phosphorescence of pd-meso-tetra (4-carboxyphenyl) porphine: a phosphor with general application for measuring oxygen concentration in biological systems.
      • Vinogradov S.A.
      • Grosul P.
      • Rozhkov V.
      • Dunphy I.
      • Shuman L.
      • Dugan B.W.
      • et al.
      Oxygen distributions in tissue measured by phosphorescence quenching.
      • Wilson D.F.
      • Vinogradov S.A.
      • Grosul P.
      • Sund N.
      • Vacarezza M.N.
      • Bennett J.
      Imaging oxygen pressure in the rodent retina by phosphorescence lifetime.
      The pHEMA hydrogel is biocompatible, has good oxygen permeability, excellent mechanical properties, and a long history of use in medical devices.
      • Montheard J.P.
      • Chatzopoulos M.
      • Chappard D.
      2-hydroxyethyl methacrylate (HEMA): chemical properties and applications in biomedical fields.
      The Pd-MABP molecules are covalently attached to the pHEMA hydrogel, ensuring that the sensing chemistry is retained in the hydrogel structure.
      The miniature sensors (0.5 × 0.5 × 5 mm) were designed to remain in the body permanently. They were soft and tissue-like to minimize stress at the material-tissue interface caused by motion and pressure, which can damage or stimulate adjacent immune cells and prolong the inflammatory phase.
      • Helton K.L.
      • Ratner B.D.
      • Wisniewski N.A.
      Biomechanics of the sensor-tissue interface-effects of motion, pressure, and design on sensor performance and the foreign body response-part I: theoretical framework.
      • Klueh U.
      • Liu Z.
      • Feldman B.
      • Henning T.P.
      • Cho B.
      • Ouyang T.
      • et al.
      Metabolic biofouling of glucose sensors in vivo: role of tissue microhemorrhages.
      To place a microsensor into the tissue, a custom-made 18-gauge injection device was used to advance the microsensor to the desired depth and hold the microsensor in place while the needle was retracted, leaving the microsensor in the tissue of interest.
      A detector was used to measure tissue oxygen concentration from a microsensor. Inside the detector there was a light-emitting diode, a temperature sensor, and photodetector circuitry. Each detector was connected to a central controller with a laptop computer for configuring the detector and viewing data. A maximum of four detectors were used at a time in this study. When a detector was placed on the skin over an injected microsensor, a visual indicator provided confirmation of proper sensor-detector alignment. The light-emitting diode of the detector sent pulses of light into the skin to excite the microsensor, which in turn emitted a light signal back through the skin and to the detector.
      The detector processed this signal, resulting in a parameter called phosphorescent lifetime decay (τ). Phosphorescent lifetime decay of Pd-MABP correlates directly to the oxygen concentration of its surrounding environment and thus is a robust parameter unaffected by variations in the optical properties of the tissue.
      • Lakowicz J.R.
      Introduction to fluorescence. Principles of fluorescence spectroscopy.
      Different methods were used to analyze and present oxygen levels: τ from a sensor was normalized to the τ from a reference sensor to provide a relative oxygen value; calibration algorithms could also be used to convert τ to the oxygen concentration in molarity (μM).
      The current detector reports τ as frequently as every 5 seconds and could record from up to 16 sensors in parallel, allowing continuous, multisite, tissue oxygen monitoring. Long-term monitoring is also possible, as has been shown in an animal model with >13 months of use.

       Data presentation and analysis

      The safety and the technical performance of the tissue oxygen monitor were evaluated in this study. To characterize safety, the number and nature of adverse events (AEs) were captured. AEs were further classified into research material-related, study procedure-related, or unrelated. To evaluate technical performance, the positive sensor detection rate, defined as the number of sensors located and measured successfully divided by the number of sensors injected across patients and across study days, was characterized. To report local relative tissue oxygen level, a dynamic relative oxygen index (DROID) was defined as the ratio of the phosphorescent lifetime decay of the arm and the foot sensor. Perioperative DROIDs were presented as time series data annotated with therapy deployment times to qualitatively assess the relationship between deployed therapy and local tissue oxygen level (Fig 1; Supplementary Fig, online only). The perioperative DROIDs were reviewed by patient and by sensor. Point measurements of the DROID postoperatively were calculated using the average of the latter half of the data collected during a session to allow time for measurements to reach steady state. Descriptive statistics of the postoperative DROIDs are reported across patients and across sensors. Data were analyzed using R 3.0.3 software (The R Foundation for Statistical Computing, http://www.r-project.org/foundation/). General additive model fits were performed using the package mgcv 1.8-3.
      Figure thumbnail gr1
      Fig 1A case study of dynamic relative oxygen indices (DROIDS) vs time during endovascular therapy shows the relationship of delivered therapy on local tissue oxygen in the foot. On the far right side of the image is a topographic descriptor of where the sensors were placed in the patient's foot. The (online only) demonstrates quick changes in tissue oxygen levels after delivery of therapy in patients with popliteal-tibial disease.
      Phosphorescent lifetime decay and sensor temperature were converted into estimated oxygen concentration based on lifetime data collected in vitro. These data were fit to the general additive model:
      ln(02+0.001)=c0+cl×tp(lifetime,temperature)+e


      where ln is the natural log, c0 is a constant intercept, c1 is a constant coefficient, e is a normally distributed error with a mean of 0, and tp is a tensor product of smooth bases.
      • Wood S.N.
      Low-rank scale-invariant tensor product smooths for generalized additive mixed models.
      The resulting oxygen concentrations for arm and foot sensors on the day of surgery are reported. Sensor data are reported in oxygen concentration units (μM) for accuracy because conversion to the more familiar oxygen tension units (mm Hg) required knowledge of the exact oxygen solubility coefficient for plasma.
      • Sinaasappel M.
      • Ince C.
      Calibration of Pd-porphyrin phosphorescence for oxygen concentration measurements in vivo.
      Rather than introduce error in the values reported to conform to transcutaneous oximetry measurement units, we present our data in concentration and plan to validate these values in future studies.
      Serial correlation patterns in sensor signal lifetime data were identified using the Box-Jenkins approach.
      • Box G.E.
      • Jenkins G.M.
      • Reinsel G.C.
      “Model identification”. Time series analysis: forecasting and control.
      Patterns were explored by graphing autocorrelation, partial autocorrelation, and extended autocorrelation functions of raw data and residuals from fitted models. Independence of model residuals was checked using the Ljung-Box test.
      • Ljung G.M.
      • Box G.E.
      On a measure of lack of fit in time series models.
      Stationarity before and after differencing was checked using the augmented Dickey-Fuller unit root test.
      • Said S.E.
      • Dickey D.A.
      Testing for unit roots in autoregressive-moving average models of unknown order.
      We modeled estimated oxygen concentrations from each sensor as a function of three periods: a baseline period immediately preceding the endovascular intervention (period 1), the period of endovascular intervention (period 2), and the period immediately after the endovascular intervention (period 3). The model matrix for the fit was coded in such a way that the coefficients for period 3 represent (and can be used to test for the significance of) the differences in oxygen levels (concentrations) and slopes (rate of change in oxygen concentration as a function of time) between period 1 and period 3. Generalized least squares
      • Kariya T.
      • Kurata H.
      “Generalized least squares estimators”. Generalized least squares.
      was used to fit the data to represent means and slopes in a similar manner as ordinary least squares regression, while explicitly accounting for the autocorrelation (ie, nonindependence) among sequential measurements common in continuous biological monitoring data.
      • Breton M.
      • Kovatchev B.
      Analysis, modeling, and simulation of the accuracy of continuous glucose sensors.
      Inclusion criteria for the data included availability of equally spaced data from the beginning of period 1 to the end of period 3 with no more than one missing data point in a row anywhere in the interval (no large gaps). In addition, we required a minimum of 50 data points for each of periods 1 and 3 to improve model reliability. As a result, all sensor data were excluded from patient 1 (too few data points in period 3), patient 2 (due to a large gap in period 1), and patient 7 (<50 data points available in period 3). Data from the left foot plantar 1 sensor of patient 5 were excluded (due to a gap starting at the end of period 2 extending into period 3).
      The present study was a first-in-human experience designed to assess the feasibility of a microsensor based approach to characterize local tissue oxygen levels to aid in the revascularization and care of patients with limb-threatening ischemia. The positive sensor detection rate was powered sufficiently to evaluate if the microsensors could be located and function consistently >90% of the time using a binomial test. This 90% threshold was chosen based on prior experience and failure rate when operating other tissue oxygen-monitoring equipment. Other time series analyses on DROID were done in a preliminary and exploratory fashion; thus, they were not sufficiently powered to draw statistical conclusions at this point.

      Results

      Patient demographic information is summarized in Table I. A total of 40 microsensors were injected in 10 patients: 10 sensors were placed in the upper arms as reference, 30 were placed in the feet in the anterior tibial (22 [73%], posterior tibial (7 [23%]), and peroneal (1 [4%]) angiosomes.
      Table IDemographics of patients in the study
      ParameterPatients (N = 10)
      Mean ± SD or No. (%)
      Age, year70.7 ± 11.5
      Male gender5 (50)
      Body mass index, kg/m225.2 ± 3.4
      Hispanic or Latino ethnicity10 (100)
      Risk factors
       Hypertension8 (80)
       Cardiovascular disease3 (30)
       Dyslipidemia4 (40)
       Type 2 diabetes6 (60)
       Renal disease1 (10)
       Tobacco use3 (30)
      Rutherford lower extremity ischemia category
       Category 4, ischemic rest pain3 (30)
       Category 5, minor tissue loss7 (70)
      Lesion classification (TASC D)
       Aortoiliac1 (10)
       Femoropopliteal4 (40)
      Tibial
       Short2 (20)
       Long3 (30)
      ABI
       Baseline
      Paired data (preoperative and postoperative ABI) were available for 4 of the 10 patients.
      0.42 ± 0.12
       Postoperative
      Paired data (preoperative and postoperative ABI) were available for 4 of the 10 patients.
      0.73 ± 0.23
      ABI, Ankle-brachial index; SD, standard deviation; TASC, TransAtlantic Inter-Society Consensus.
      a Paired data (preoperative and postoperative ABI) were available for 4 of the 10 patients.
      Of 21 AEs reported in the study, 0 were research material-related, 13 were study procedure-related, and 8 were unrelated, including 3 serious AEs. The 13 procedure-related AEs were classified as mild bruising at the injection site (32.5%) and occurred in six patients. The three serious AEs were associated with complications of the endovascular procedure: closure device failure with need for operative intervention in two patients and one intraprocedural iliac arterial perforation treated with a polytetrafluoroethylene-covered stent.
      During the course of the 28-day study, 212 measurement attempts were made from the injected microsensors: 97.2% of sensors were successfully located and measured, exceeding the predetermined per-protocol threshold of 90% (P = .0004). In five of six failed attempts, the microsensors were located and measured successfully at a later time. One of the failed attempts occurred at the patient's last day of participation, and thus, the sensor's functionality could not be reassessed.
      Perioperative DROIDs were calculated using data from 10 patients. During these 10 endovascular revascularization procedures, 96 timestamps of therapy deployment were recorded. Fig 1 illustrates two of the 28 perioperative DROID time series collected. Changes in DROID appeared to align temporally with some deployed therapies: a decrease in DROID after balloon inflation (dark-gray sensor, t = 28, 39 minutes), an increase after balloon deflation (dark-gray sensor, t = 32, 44 minutes), and an increase after stent deployment (dark-gray and light-gray sensors, t = 51 minutes) were observed. Differential changes in DROID were also noted across microsensors in this individual: the dark-gray sensor appeared to be responsive to deployed therapies, whereas the light-gray sensor did not indicate a change in DROID until ∼t = 48 minutes later on in the procedure.
      Preoperative and postoperative DROIDs were calculated from data obtained in 38 sessions with participants over a course of 4 weeks, yielding 138 data points. Measurements were categorized into preoperative (n = 28), immediately postoperative (n = 28), 1 week (n = 15), 2 weeks (n = 21), 3 weeks (n = 23), and 4 weeks (n = 23) after the revascularization procedure. Descriptive statistics of this data set are presented as a box plot in Fig 2.
      Figure thumbnail gr2
      Fig 2Box and whisker plot shows the dynamic relative oxygen index (DROID). DROID is calculated as the micro-oxygen sensor (MOXY) value from the arm/MOXY value from the foot. The horizontal lines in the middle of each box represent the median value of the sensors at each time of measurement included in the study protocol. The top and bottom borders of the box mark the 75th and 25th percentiles, respectively, the whiskers mark the lowest and highest datum within 1.5 of the interquartile range of the lower and upper quartile, respectively (Tukey boxplot), and the circles indicate outliers. Stdev, Standard deviation.
      In vitro calibration data were used to convert the phosphorescent lifetime decay and temperature values collected from this study into oxygen concentration in μM in an exploratory analysis. Converted data in μM were fitted into a regression model to specifically examine (1) the value changes in oxygen concentration and (2) the slope changes in oxygen concentration before and after the revascularization procedure. Fig 3 illustrates the oxygen concentration value measured in the beginning of the revascularization procedure, before any interventions had been performed (pre) and at the end of the procedure, with the patient still on the table (post). These results were further categorized into arm reference sensors and foot sensors. The median oxygen concentration of the reference arm sensors decreased nominally from 58.1 μM to 55.2 μM (n = 6), and the median oxygen concentration of the foot sensors increased from 10.7 μM to 28.7 μM (n = 16). The increase in median oxygen concentration observed in the foot sensors was statistically significant using a one-tailed Wilcoxon test (P = .0042). Moreover, we observed that the oxygen concentration might not have reached steady state upon procedural completion because the sensor signals trended upwards, indicating continued reoxygenation of the tissue after revascularization. This observation was verified in some cases in which increasing oxygen concentrations were captured in the recovery room. Table II summarizes the arm and foot sensor data in the postintervention period.
      Figure thumbnail gr3
      Fig 3Comparison of arm and foot sensors' oxygen concentration (μM) before (pre) and after (post) revascularization. The horizontal lines in the middle of each box represent the median value, the top and bottom borders of the box mark the 75th and 25th percentiles, respectively, the whiskers mark the lowest and highest datum within 1.5 of the interquartile range of the lower and upper quartile, respectively (Tukey boxplot), and the circles indicate outliers. Stdev, Standard deviation.
      Table IISummary of oxygen sensor slope data for arm and foot sensors in the immediate postintervention period
      Positive slopes that are significant compared with preintervention slopes (P < .05) are bolded. Most foot sensors (12 of 16) demonstrated a significant increasing oxygen concentration immediately after the intervention. Arm sensors exhibited negligible or decreasing oxygen slopes. Refer to the Results for details on patient exclusions.
      Patient IDSensor locationPostintervention slope, μM/minP value
      3Arm−2.964<.001
      Foot 13.192<.001
      Foot 27.860<.001
      Foot 32.244<.001
      5Arm−0.336.669
      Foot 10.024<.001
      Foot 22.292<.001
      6Arm−1.572<.001
      Foot 10.216.078
      Foot 2−0.548<.001
      Foot 30.304.060
      8Arm−0.432.001
      Foot 11.968<.001
      Foot 20.504<.001
      9Arm0.056.888
      Foot 10.056.779
      Foot 20.012<.001
      Foot 30.324.002
      10Arm0.036.465
      Foot 10.600<.001
      Foot 20.120.033
      Foot 31.320<.001
      a Positive slopes that are significant compared with preintervention slopes (P < .05) are bolded. Most foot sensors (12 of 16) demonstrated a significant increasing oxygen concentration immediately after the intervention. Arm sensors exhibited negligible or decreasing oxygen slopes. Refer to the Results for details on patient exclusions.
      A positive slope indicates oxygen levels were increasing. The regression model showed that 94% of the foot sensors (n = 16) demonstrated a positive slope change, in which 80% of them were considered significant compared with slopes in the preintervention period (P < .05). Arm sensors exhibited negligible or decreasing oxygen slopes.

      Discussion

      A major tenet of limb salvage therapy is that one must recognize perfusion deficiencies and be able to grade the severity and monitor their improvement after revascularization. In fact, one of the essential components of the Society for Vascular Surgery Wound, Ischemia, and foot Infection (WIfI) Threatened Limb Classification System is to assess perfusion in all patients with threatened limbs. In modern tertiary vascular practice, where many patients have diabetes and medial calcinosis, with open foot wounds or previous amputations, traditional means of measuring perfusion are problematic and often not applicable or reliable.
      The MOXY sensor provides a new method of measuring tissue oxygen concentration without perturbing the tissue after the initial injection. In this first-in-man study, the evaluation of the sensors in the perioperative period demonstrated their safety for use in patients with limb-threatening ischemia. Most device-related AEs were limited to small ecchymotic areas, most likely secondary to the 18-gauge needle inserted to administer the sensor. A simpler and less invasive method of delivery to avoid patient discomfort (although reported to be minimal during the implantation period for this cohort) is a straightforward product development task.
      The implantation of the devices was simple and effective, and the sensors showed a very high detection rate of 97.2% during a total of six patient visits (preoperative, postoperative, and at weeks 1, 2, 3, and 4). Although this is a small cohort with a relatively short 28-day follow-up, the ease of sensor localization and robust sensor signal are promising and provide insight into the potential clinical applicability in the acute and chronic phases of care. There exists a straightforward path to detector miniaturization and a form factor conducive to measuring oxygen in the operating room and, eventually, home use, which is critical to technology adoption and a clear potential advantage over other methods, such as fluorescence angiography, for sensing biological oxygen.
      • Mills J.
      Intraoperative fluorescence vascular angiography: a case report.
      Compared with other oxygen-sensing devices in the clinical setting, MOXY offers distinct advantages. First, it directly measures tissue oxygen concentration within the tissue, which differs from completely noninvasive (no injectable sensor) technologies, such as pulse oximetry and near-infrared spectroscopy (NIRS), that measure oxygen saturation in the vasculature (Fig 4). Pulse oximetry measures the oxygen saturation in the arterioles and requires a pulse and is therefore less useful for monitoring tissue oxygen saturation in ischemic tissue. NIRS measurements do not require a pulse but provide estimates of the oxygen available for diffusion into the tissue (StO2). Although studies have been published on the use of NIRS in peripheral arterial disease and ischemic tissues,
      • Comerota A.J.
      • Throm R.C.
      • Kelly P.
      • Jaff M.
      Tissue (muscle) oxygen saturation (StO2): a new measure of symptomatic lower-extremity arterial disease.
      • Cheatle T.R.
      • Potter L.A.
      • Cope M.
      • Delpy D.T.
      Near-infrared spectroscopy in peripheral vascular disease.
      • Keller A.
      Noninvasive tissue oximetry for flap monitoring: an initial study.
      it does not provide a direct measurement of tissue concentration and is susceptible to interference from patient motion and slight changes in pressure.
      Figure thumbnail gr4
      Fig 4Graphic comparison of current modalities for clinical and nonclinical oxygen (O2) evaluation in tissue. Micro-oxygen sensor (MOXY) sensor (as depicted in the tissue by the light gray bar) measures the tissue oxygen concentration (Pto2) directly from the tissue. HB, Hemoglobin; HbO2, oxyhemoglobin; NIR, near-infrared.
      The most clinically validated technique in the setting of PAD care is transcutaneous oxygen pressure (TcpO2), a measurement of the oxygen that diffuses from the tissue to a surface electrode.
      • Caselli A.
      • Latini V.
      • Lapenna A.
      • Di Carlo S.
      Transcutaneous oxygen tension monitoring after successful revascularization in diabetic patients with ischaemic foot ulcers.
      The electrode warms the surface of the skin to 44°C to disrupt the outermost skin layer (stratum corneum) so that oxygen can exit the tissue. TcpO2 has been demonstrated to predict the ability of diseased tissue to heal (eg, TcpO2 readings <40 mm Hg correlate with impaired wound healing). Criticisms of the technique include the time and complexities involved with obtaining reliable measurements, interference from edema and inflammation, and the requirement for elevated temperatures at the measurement site causing an artificial measurement environment.
      • Braun J.D.
      • Rajguru P.
      • Armstrong D.G.
      • Mills J.L.
      Indocyanine green angiographic criteria using ingress and ingress rate to detect SVS lower extremity threatened limb classification (WIfI) grade 3 ischemia.
      • Arsenault K.A.
      • Al-Otaibi A.
      • Devereaux P.J.
      • Thorlund K.
      • Tittley J.G.
      • Whitlock R.P.
      The use of transcutaneous oximetry to predict healing complications of lower limb amputations: a systematic review and meta-analysis.
      The microsensor technology was proven to be reliable in providing real-time information during the perioperative period. Although the blinded condition of the operators limited a systematic evaluation of sensor response to each interventional step, there was a clear trend upwards in tissue oxygen in the immediate postoperative period. The interpatient baseline oxygen tissue concentration is potentially affected by hypoxemic morbidities as chronic obstructive pulmonary disease or congestive heart failure, as well as intraoperative interventions such as oxygen supplementation.
      The DROID relationship was created in an effort to normalize any potential confounding systemic oxygen effects (Fig 2). Although not statistically significant because of the underpowered nature of a small sample, there was a clear trend of improvement, which was fairly acute in nature: the largest change was realized within the postoperative period and stabilized by 1 week. Moreover, the level was fairly stable out to 4 weeks. This observation suggests the potential use of the microsensor as an on-table tool to optimize results. Unpublished data in healthy individuals demonstrated MOXY's ability to sense changes in oxygen levels within seconds, which, if also true for this group of patients, could empower the operator to tailor an intervention to optimal perfusion-goal performance.

      Conclusions

      The “Si Se Puede” first-in-man study has shown that the MOXY sensor appears to be a safe and effective tool to measure tissue oxygen concentrations in real-time in patients with limb-threatening ischemia during the perioperative period of planned revascularization procedures. The study results show that use of these sensors merits further testing to determine their potential effect during on-table clinical decision making and long-term follow-up, with the ultimate goal of improved wound healing and limb salvage rates.

      Author contributions

      Conception and design: NW, SG, KH, MMB, LMA
      Analysis and interpretation: KA, KH, MMB
      Data collection: KA, MMB, LMA, MC, NW
      Writing the article: KA, KH, JM, MMB
      Critical revision of the article: KA, KH, JM, MMB, NW
      Final approval of the article: KA, NW, SG, KH, MMB, JM, MC, LMA
      Statistical analysis: KA, KH
      Obtained funding: KH, NW
      Overall responsibility: MMB
      The authors wish to acknowledge Dr Mitch Kostich for his contributions to data and statistical analysis.

      Appendix.

      Figure thumbnail fx1
      Supplementary Fig (online only)A, A.second case study (patient 3) of dynamic relative oxygen indices (DROIDSs) vs time during endovascular therapy in a patient with popliteal-tibial disease. The image at the right is a topographic descriptor of sensor locations in the patient's foot. Traces demonstrate increase in tissue oxygen levels resulting from therapy delivery. B, A third case study (patient 9) of DROIDS during endovascular therapy in a patient with proximal iliac disease. In this example, the magnitude and responsiveness of sensors is not as pronounced as in A and in . The sensors exhibit changes, but whether they are significant in response to deployed therapies is less clear.

      References

        • Fowkes G.
        • Rudan D.
        • Rudan I.
        • Aboyans V.
        • Denenberg J.
        • McDermott M.
        • et al.
        Comparison of global estimates of prevalence and risk factors for peripheral artery disease in 2000 and 2010: a systematic review and analysis.
        Lancet. 2013; 382: 1329-1340
        • Jim J.
        • Owens P.L.
        • Sanchez L.A.
        • Rubin B.G.
        Population-based analysis of inpatient vascular procedures and predicting future workload and implications for training.
        J Vasc Surg. 2012; 55 (discussion: 1399-400): 1394-1399
        • Norgren L.
        • Hiatt W.
        • Dormandy J.
        Inter-society consensus for the management of peripheral arterial disease (TASC II).
        J Vasc Surg. 2007; 45: S5-S67
        • Mills Sr., J.L.
        • Conte M.S.
        • Armstrong D.G.
        • Pomposelli F.B.
        • Schanzer A.
        • Sidawy A.N.
        • et al.
        The Society for Vascular Surgery Lower Extremity Threatened Limb Classification System: risk stratification based on wound, ischemia, and foot infection (WIfI).
        J Vasc Surg. 2014; 59: 220-222
        • Montero-Baker M.
        • Schmidt A.
        • Bräunlich S.
        • Ulrich M.
        • Thieme M.
        • Biamino G.
        • et al.
        Retrograde approach for complex popliteal and tibioperoneal occlusions.
        J Endovasc Ther. 2008; 15: 594-604
        • Bosanquet D.C.
        • Glasbey J.C.
        • Williams I.M.
        • Twine C.P.
        Systematic review and meta-analysis of direct versus indirect angiosomal revascularisation of infrapopliteal arteries.
        Eur J Vasc Endovasc Surg. 2014; 48: 88-97
        • Rumsey W.L.
        • Vanderkooi J.M.
        • Wilson D.F.
        Imaging of phosphorescence: a novel method for measuring oxygen distribution in perfused tissue.
        Science. 1988; 241: 1649-1651
        • Lo L.W.
        • Koch C.J.
        • Wilson D.F.
        Calibration of oxygen-dependent quenching of the phosphorescence of pd-meso-tetra (4-carboxyphenyl) porphine: a phosphor with general application for measuring oxygen concentration in biological systems.
        Anal Biochem. 1996; 236: 153-160
        • Vinogradov S.A.
        • Grosul P.
        • Rozhkov V.
        • Dunphy I.
        • Shuman L.
        • Dugan B.W.
        • et al.
        Oxygen distributions in tissue measured by phosphorescence quenching.
        Adv Exp Med Biol. 2003; 510: 181-185
        • Wilson D.F.
        • Vinogradov S.A.
        • Grosul P.
        • Sund N.
        • Vacarezza M.N.
        • Bennett J.
        Imaging oxygen pressure in the rodent retina by phosphorescence lifetime.
        Adv Exp Med Biol. 2006; 578: 119-124
        • Montheard J.P.
        • Chatzopoulos M.
        • Chappard D.
        2-hydroxyethyl methacrylate (HEMA): chemical properties and applications in biomedical fields.
        J Macromol Sci Rev. 1992; 32: 1-34
        • Helton K.L.
        • Ratner B.D.
        • Wisniewski N.A.
        Biomechanics of the sensor-tissue interface-effects of motion, pressure, and design on sensor performance and the foreign body response-part I: theoretical framework.
        J Diabetes Sci Technol. 2011; 5: 632-646
        • Klueh U.
        • Liu Z.
        • Feldman B.
        • Henning T.P.
        • Cho B.
        • Ouyang T.
        • et al.
        Metabolic biofouling of glucose sensors in vivo: role of tissue microhemorrhages.
        J Diabetes Sci Technol. 2011; 5: 583-595
        • Lakowicz J.R.
        Introduction to fluorescence. Principles of fluorescence spectroscopy.
        Springer, Berlin1999: 1-23
        • Wood S.N.
        Low-rank scale-invariant tensor product smooths for generalized additive mixed models.
        Biometrics. 2006; 62: 1025-1036
        • Sinaasappel M.
        • Ince C.
        Calibration of Pd-porphyrin phosphorescence for oxygen concentration measurements in vivo.
        J Appl Physiol (1985). 1996; 81: 2297-2303
        • Box G.E.
        • Jenkins G.M.
        • Reinsel G.C.
        “Model identification”. Time series analysis: forecasting and control.
        4th edition. John Wiley and Sons, Inc, Hoboken, NJ2008: 195-230
        • Ljung G.M.
        • Box G.E.
        On a measure of lack of fit in time series models.
        Biometrika. 1978; 65: 297-303
        • Said S.E.
        • Dickey D.A.
        Testing for unit roots in autoregressive-moving average models of unknown order.
        Biometrika. 1984; 71: 599-607
        • Kariya T.
        • Kurata H.
        “Generalized least squares estimators”. Generalized least squares.
        John Wiley & Sons, Chichester, UK2004: 25-66
        • Breton M.
        • Kovatchev B.
        Analysis, modeling, and simulation of the accuracy of continuous glucose sensors.
        J Diabetes Sci Technol. 2008; 4: 4-14
        • Mills J.
        Intraoperative fluorescence vascular angiography: a case report.
        J Diabetes Sci Technol. 2012; 6: 204-208
        • Comerota A.J.
        • Throm R.C.
        • Kelly P.
        • Jaff M.
        Tissue (muscle) oxygen saturation (StO2): a new measure of symptomatic lower-extremity arterial disease.
        J Vasc Surg. 2003; 38: 724-729
        • Cheatle T.R.
        • Potter L.A.
        • Cope M.
        • Delpy D.T.
        Near-infrared spectroscopy in peripheral vascular disease.
        Br J Surg. 1991; 78: 405-408
        • Keller A.
        Noninvasive tissue oximetry for flap monitoring: an initial study.
        J Reconstr Microsurg. 2007; 23: 189-197
        • Caselli A.
        • Latini V.
        • Lapenna A.
        • Di Carlo S.
        Transcutaneous oxygen tension monitoring after successful revascularization in diabetic patients with ischaemic foot ulcers.
        Diabet Med. 2005; 22: 460-465
        • Braun J.D.
        • Rajguru P.
        • Armstrong D.G.
        • Mills J.L.
        Indocyanine green angiographic criteria using ingress and ingress rate to detect SVS lower extremity threatened limb classification (WIfI) grade 3 ischemia.
        J Vasc Surg. 2014; 60: 538
        • Arsenault K.A.
        • Al-Otaibi A.
        • Devereaux P.J.
        • Thorlund K.
        • Tittley J.G.
        • Whitlock R.P.
        The use of transcutaneous oximetry to predict healing complications of lower limb amputations: a systematic review and meta-analysis.
        Eur J Vasc Endovasc Surg. 2012; 43: 329-336