Noninvasive monitoring of hemodynamic stress using quantitative near-infrared frequency-domain photon migration spectroscopy

. Hemorrhagic hypovolemia and inotropic agent administration were used to manipulate cardiac output (CO) and oxygen delivery in rabbits to investigate the correlation between noninvasive frequency domain photon migration (FDPM) spectroscopy and invasive hemodynamic monitoring parameters. Frequency-domain photon migration provides quantitative measurements of light absorption and reduced scattering ( m a and m s 8 , respectively) in tissue. Wavelength dependent m a values were used to calculate invivo tissue concentration of deoxyhemoglobin [Hb], oxyhemoglobin @ HbO 2 # , total hemoglobin [TotHb], and water @ H 2 O # as well as mixed arterial-venous oxygen saturation ( S t O 2 ) in tissue. FDPM-derived physiologic properties were correlated with invasive measurements of CO and mean pulmonary artery pressure (mPAP), FDPM-derived [TotHb] and S t O 2 correlated signiﬁcantly with hemorrhaged volume (HV), mPAP, and CO. Correlation coefﬁcients for [TotHb] vs HV, mPAP, and CO were −0.77, 0.86, and 0.70, respectively. Correlation coefﬁcients of S t O 2 vs HV, mPAP, and CO were −0.71, 0.55, and 0.61, respectively. Dobutamine induced changes resulted in correlation coefﬁcients between FDPM-derived and invasively measured physiologic parameters that are comparable to those induced by hypovolemia. FDPM spectroscopy is sensitive to changes in mPAP and CO of as little as 15%. These results suggest that FDPM spectroscopy may be used in clinical settings to noninvasively monitor central hemodynamic parameters and to directly assess oxygenation of tissues. © 2002SocietyofPhoto-Optical InstrumentationEngineers.


Introduction
Adequate perfusion and oxygenation are critical for tissue viability. Insufficient tissue oxygenation is the underlying cause of tissue damage for an extensive range of medical emergencies, e.g., hypovolemic shock, peripheral vascular disease, and stroke. Tissue oxygenation depends on oxygen delivery (DO 2 ) as well as oxygen consumption (VO 2 ). DO 2 is a product of cardiac output ͑CO͒ and arterial oxygen content and is an accurate indicator of the overall oxygenation status of the body. 1,2 Regional oxygen delivery depends on DO 2 and the local vascular resistance, while VO 2 is the difference between arterial delivery and venous return of oxygen ͕i.e., VO 2 ϭCOx(͓aO 2 ͔ -͓vO 2 ͔)͖. Since arterial ͓aO 2 ͔ and venous ͓vO 2 ͔ oxygen contents are generally proportional to hemo-globin oxygen saturation, DO 2 and VO 2 usually correlate to arterial (S a O 2 ) and venous (S v O 2 ) oxygen saturation and cardiac output. Hemodynamic monitoring of DO 2 and VO 2 is a widely accepted principle for managing various medical emergencies, particularly circulatory dysfunction and shock. 1,[3][4][5] Noninvasive arterial blood pressure and heart rate are considered to be poor indicators of early shock. 4,6 Pulmonary artery thermodilution and Swan-Ganz catheterization to determine CO and pulmonary artery wedge pressure ͑PWP͒ are common invasive clinical procedures to measure DO 2 and VO 2 that may cause serious complications. While DO 2 and VO 2 are accurate indicators of the overall oxygenation status of the body, they are not necessarily representative of regional tissue oxygenation. The fractional distribution of CO to various organs is cumbersome if not im-possible to determine, especially in vasculature with multiple arterial supplies and instances of regional heterogeneity. A noninvasive modality capable of quantifying local tissue oxygenation that is also sensitive to factors contributing to DO 2 and VO 2 is of clinical importance for monitoring tissue hypoxia secondary only to shock.
Several noninvasive optical techniques are currently under investigation, including laser Doppler imaging 7 and nearinfrared ͑NIR͒ reflectance. [8][9][10][11] The methodology and merits of NIR spectroscopy have been intensely examined by numerous researchers. [12][13][14][15][16] Substantial progress has been made during the past decade in our ability to quantify tissue optical and physiologic properties, largely due to the emergence of time and frequency-domain photon migration techniques. 11,17 Frequency-domain photon migration ͑FDPM͒ uses high frequency ͑MHz-GHz͒ intensity modulated near-infrared light to noninvasively quantify light absorption ( a ) and reduced scattering ( s Ј) parameters in tissue. a and s Ј are sensitive to biologically important light-absorbing and -scattering molecules and structures. Intrinsic NIR absorbers in tissue are generally assumed to be hemoglobin ͓deoxy-͑Hb͒ and oxy-(HbO 2 )͔, water (H 2 O), fat, cytochromes, and melanin. 18,19 Quantitative measurements of wavelength-dependent light absorption can be used to determine the in vivo concentrations of these species. Additional parameters such as tissue blood volume fraction, hemoglobin oxygen saturation (S t O 2 ), and oxygen utilization can be deduced. Note that, in this article, the term tissue oxygenation (S t O 2 ) refers to the oxygenation of arteriovenous hemoglobin found within vessels that have diameters ranging from capillaries to medium-size veins/ arteries. Although S t O 2 and oxygen partial pressure ͑tension͒ are proportional, they are distinct quantities. Light scatterers are cellular ͑e.g., organelles, proteins͒ and extracellular ͑e.g., collagen, elastin͒ tissue components. 20,21 Measurements of wavelength-dependent scattering properties offer insight into the composition, density, and organization of tissue structures. Consequently, changes in wavelength-dependent scattering properties may accompany processes such as ischemia, necrosis, wound healing, and malignancy.
The sensitivity of optical properties to the tissue's functional and structural components may provide a means for noninvasive assessment of pathophysiologic processes, such as ischemia and hypoxia. Indeed, previous studies have shown significant changes in the tissue's optical and physiologic properties in response to local ischemia. 17,22,23 In this pilot study, substantial hemodynamic changes were induced in a rabbit model to investigate how changes in DO 2 affect tissue oxygenation of the limb. The aim of this study was to assess ͑1͒ the feasibility of using FDPM spectroscopy to quantify local tissue oxygenation and ͑2͒ the sensitivity of FDPMderived parameters to changes in tissue oxygenation due to perturbation in DO 2 .

Materials and Methods
Experimental setups based on hypovolemia and dobutamine treatment were selected on the basis of clinical relevance. Hypovolemia was induced in the rabbit in order to decrease CO, and thereby decrease DO 2 . Dobutamine was administered to increase CO and DO 2 . FDPM data were collected concurrently with systemic hemodynamic parameters, specifi-cally, CO, mean pulmonary artery pressure ͑mPAP͒, arterial saturation (S a O 2 ), and mean systemic arterial pressure ͑mSAP͒. FDPM-derived physiologic properties were correlated to hemodynamic parameters.

Animal
Pathogen-free white New Zealand male rabbits ͑Myrtle Rabbitry Inc., Thompson Station, TN͒, weighing 4.0 ͓Ϯ0.4 standard deviation ͑SD͔͒ kg, were used. Animals were housed in a pathogen-free animal facility and were given a commercial basal diet and water ad libitum. The study was approved by the Institutional Laboratory Animal Care and Use Committee, University of California, Irvine ͑ARC protocol No. 97-1556͒.

Procedures
Hypovolemic hemorrhagic shock was induced in six animals ͑the hypovolemic group͒. Dobutamine was administered to another six animals ͑the dobutamine group͒. The animals were anesthetized by an intramuscular injection of a ketamine-xylazine mixture at an induction dose of 50 mg/kg ketamine ͑Ketaject, Phoenix Pharmaceutical Inc., St. Joseph, MI͒ and 5 mg/kg xylazine ͑Anased, Lloyd Laboratories, Shenandoah, IA͒. Anesthesia was maintained as needed by intravenous administration of the ketamine-xylazine mixture at doses of 2.5-5.0 mg/kg ketamine and 0.5-1.0 mg/kg xylazine. Immediately following anesthesia induction, animals were intubated using a size 3.0 endotracheal tube and mechanically ventilated ͑dual phase control respirator, model 32A4BEPM-5R, Harvard Apparatus, Chicago, IL͒ at a respiration rate of 32/min and a tidal volume of 50 cc at 100% inspiratory oxygen. The animals were shaved at the thorax and thighs. The chest wall was opened to expose the thoracic cavity, allowing direct access to the heart and aorta. The animals were euthanized by exsanguination immediately after the study in accordance with federal guidelines.

Physiology Measurements
A calibrated flow transducer ͑T106 small animal flow meter, Transonic System, Inc., Ithaca, NY͒ was placed around the ascending aorta to measure CO. A time trace of the cardiac flow was recorded for ϳ10 s, and an average of the time of the flow profile was calculated to yield the mean amount of CO. The heart rate ͑HR͒ was also determined from the time profile of cardiac flow. To measure the pulmonary arterial pressure, a calibrated pressure transducer ͑TSD104A transducer and MP100 WSW System, Biopac Systems, Inc., Santa Barbara, CA͒ was connected to a fluid-filled line terminated by an 18 gauge catheter. The catheter was inserted directly into the right ventricular chamber and advanced into the pulmonary artery under continuous monitoring for the duration of the measurement. A time trace of the pulmonary arterial pressure was recorded for 10 s. The systolic and diastolic pressures corresponded to the peak and trough of the pressure profile; the mPAP was calculated digitally. Systemic arterial pressures were similarly measured, with the transducerconnected catheter inserted into the ear artery. The mSAP was recorded.
Blood was drawn from the artery of the ear for arterial blood gas chemistry. All fluid-filled lines connected to the transducers were intermittently flushed with small volumes ͑Ͻ1/2 cc͒ of heparin to prevent blood clots. A pulse oximeter probe ͑Biox 3700 pulse oximeter, Ohmeda, Hurst, TX͒ was placed on the forelimb to measure the S a O 2 and HR. Baseline values for all hemodynamic and noninvasive measurements were determined prior to induction of hypovolemia or administration of dobutamine.

Hypovolemia
An 18-gauge needle was inserted into the right atrium and 5 cc of hemorrhaged volume ͑HV͒ was slowly removed. Three minutes later, hemodynamic parameters were measured again in conjunction with FDPM spectroscopy. The procedure was repeated until a total HV of 60 cc ͑corresponding to an average of 42% of the total blood volume͒ was removed, except for in one rabbit where a total volume of 120 cc was removed.

Dobutamine Administration
Dobutamine doses of 30, 40, and 50 g/kg were sequentially administered into the ear vein at a average rates of 40, 53, and 67 g/min, respectively. One minute after each dose, the hemodynamic parameters were measured simultaneously with the FDPM measurements. Two trials of administration of dobutamine were carried out on each rabbit. The values obtained from the two trials were averaged.

FDPM
Theory and instrumentation of the FDPM approach to quantify tissue optical properties are described in detail elsewhere. 24,25 Briefly, FDPM launches sinusoidally intensity modulated NIR light into highly scattering tissues. Light that is sinusoidally intensity modulated at angular frequency propagates in tissues as highly damped diffuse photon-density waves ͑PDWs͒ with well-defined wave characteristics, specifically, the phase velocity and decay rate. 26 PDW propagation depends on the bulk tissue optical properties and can be described by a diffusion approximation for light transport in turbid media, e.g., tissues.
Analytically derived solutions to the diffusion equation for the semi-infinite boundary condition yield nonlinear PDW model functions. The model functions are fit to measurements of the frequency-dependent phase shift and amplitude attenuation of diffusely reflected PDWs in order to explicitly calculate tissue a and s Ј . Wavelength-dependent absorption is used to determine the tissue concentration of deoxyhemoglobin ͓Hb͔, oxyhemoglobin ͓HbO 2 ͔, ͓TotHb͔ (ϭ͓Hb͔ Extinction coefficients for Hb, HbO 2 , and H 2 O were obtained from published values. 18 Least-squares fits were used to solve Eq. ͑1͒, with the constraint that the concentration values must be greater than zero. 27 For this study, a portable broad-bandwidth FDPM instrument was used to quantify tissue optical properties at multiple NIR wavelengths. FDPM measurements were performed using a probe with a source-detector separation of 9.45 mm. Both the source and detecting fibers were gently placed on the quadricep muscles to collect the FDPM data set, which consists of phase and amplitude data for the wavelengths at 674, 798, 811, 849, and 956 nm. The quadricep muscles were chosen because of their large tissue mass and thus best fulfill the semi-infinite boundary condition. Each data set consists of three to four repetitive measurements. Figure 1 is a schematic of the setup used to perform FDPM and hemodynamic measurements.

Data Analysis
Baseline data from the hypovolemic ͑six rabbits͒ and dobutamine ͑six rabbits͒ group were combined, i.e., hemodynamic-and FDPM-derived data on the 12 rabbits were pooled. Mean and standard deviation ͑ϮSD͒ were calculated from the pooled baseline values for CO, mPAP, mSAP, and FDPM-derived optical and physiologic parameters.
In order to compare data obtained from different animals, the percentages of change from baseline values were computed for CO, mPAP, mSAP, and FDPM-derived physiologic concentrations. The percentages of change for CO, mPAP, mSAP, ͓Hb͔, ͓HbO 2 ͔, ͓TotHb͔, ͓S t O 2 ͔, and ͓H 2 O͔ from the hypovolemic group were combined. The percentages of change in the FDPM-derived physiologic parameters were correlated to HV, ⌬CO, and ⌬mPAP. To quantify the extent of association between these variables, the correlation coefficient and the 95% confidence interval were computed. Data from the dobutamine-treated group were analyzed in the same manner.

Hemodynamic Measurements
Values of CO, mPAP, and mSAP were measured for each incremental HV or dobutamine dose. Figures 2͑a͒ and 2͑b͒ show typical time traces of the cardiac flow and pulmonary arterial pressure, respectively. HR and CO were determined from the time trace of the cardiac flow, and mPAP was determined from time trace of pulmonary arterial pressure. The mean systemic arterial pressure was recorded. Baseline values for CO, mPAP, and mSAP from the hypovolemic-and dobutamine-treated groups were combined and are summarized in Table 1.

FDPM
An example of the FDPM phase and amplitude data is shown Figure 3͑a͒ for 798 nm. Model functions were fit to the phase ͑ᮀ, left y axis͒ and amplitude ͑ࡗ, right y axis͒ of the PDW to determine the tissues' optical properties. Model functions ͑lines͒ fit the phase and amplitude data well and yielded a and s Ј values. Representative a ͑ᮀ, left y axis͒ and s Ј ͑ࡗ, right y axis͒ of the five wavelengths are shown in Figure 3͑b͒. Pham et al. Also shown is a line that represents a least-squares fit to the wavelength-dependent a values, resulting in ͓Hb͔ ϭ25.8 M, ͓HbO 2 ͔ϭ49.6 M, ͓TotHb͔ϭ75.4 M, S t O 2 ϭ65.8%, and ͓H 2 O͔ϭ31.6 M for the specific data set shown in Figure 3. Table 2 summarizes the baseline optical properties of the five wavelengths used, and Table 3 summarizes the baseline FDPM-derived physiologic properties. Figure 4 shows a representative example of the progression of physiologic parameters as a function of the increasing severity  (2), five NIR diode lasers (3) at 674, 798, 811, 849, and 956 nm, optical fiber for the light source (4) and signal detection (5) mounted 9.45 mm apart on a probe (6), and an avalanche photodiode (7). The probe was placed gently on the shaved area of the hind limb. CO was measured using a calibrated flow transducer (8) placed around the aorta. PAP and SAP were measured with pressure transducers placed, respectively, in the right ventricle and advanced into the pulmonary artery (9) and in the artery of the ears (10). S a O 2 was initially determined from and arterial blood gas analysis and subsequently monitored with a pulse oximeter placed on the forelimb. The heart rate is also determined from the cardiac flow time profile. (b) Time trace of the pulmonary artery pressure. A needle, which was connected to the transducer, was inserted into the right atrium in order to measure PAP. Systolic and diastolic PAP correspond to the peak and trough, respectively. An average of the systolic and diastolic PAP gives the mean PAP. of hypovolemia. In Figure 4͑a͒, CO ͑ࡗ, left y axis͒ and mPAP ͑ᮀ, right y axis͒ started to decrease when as little as 10 mL ͑ϳ7% total blood volume͒ of blood was removed, while mSAP ͑, right y axis͒ remained relatively constant until the HV reached ϳ30 mL ͑ϳ21% of the total blood volume͒.  Figure 4͑c͒ is a plot of TotHb ͑ᮀ, left y axis͒ and mixed S t O 2 saturation ͑ࡗ, right y axis͒ versus HV. As was observed with CO and mPAP values, ͓HbO 2 ͔, ͓TotHb͔, and S t O 2 significantly decrease when as little as 10-15 mL of blood is removed. Interestingly, ͓Hb͔ and ͓H 2 O͔ remained relatively constant even when the hemorrhage volume reached 60 mL ͑ϳ42% of the total blood volume͒. For each animal in the hypovolemic group, we calculated the percentages of change from the baseline values for the hemodynamic-and FDPM-derived physiologic parameters in response to blood loss. The percentage of change data on all animals in the hypovolemic group were combined. To ascertain that acute hemorrhaging decreased mPAP and CO, we determined the correlation coefficients for ⌬mPAP vs HV and ⌬CO vs HV. Correlation coefficients of Ϫ0.76 and Ϫ0.82 were obtained for ⌬mPAP vs HV and for ⌬CO vs HV, respectively. The mean PAP and CO decreased by as much as 50% and 80%, respectively, in response to blood loss of 60 cc. Correlations between ⌬͓TotHb͔ vs HV, ⌬͓TotHb͔ vs ⌬mPAP, and ⌬͓TotHb͔ vs ⌬CO are shown for the hemorrhaged group in Figures 5͑a͒-5͑c͒, respectively. ⌬͓TotHb͔ is seen to be correlated with HV, ⌬mPAP, and ⌬CO, resulting in correlation coefficients of Ϫ0.77, 0.68, and 0.70, respectively. Furthermore, the ͓TotHb͔ measured decreased approximately 10% in response to 60 mL of blood loss, while mPAP decreased 50%, and CO decreased 80%. Figures 6͑a͒-6͑c͒ illustrate the relationship of changes in tissue hemoglobin oxygen saturation (⌬S t O 2 ) and HV, ⌬mPAP, and ⌬CO, respectively. As shown Figure 6, ⌬S t O 2 varies with each, resulting in correlation coefficients of Ϫ0.71, 0.55, and 0.61, respectively. As indicated by the smaller magnitude of the correlation coefficients, the association among HV, ⌬mPAP, and ⌬CO vs ⌬S t O 2 is less dramatic than these parameters versus ͓TotHb͔. Table 4  is obtained for the wavelength-dependent a data in order to calculate [Hb], ͓HbO 2 ͔, and ͓H 2 O͔. The solid line through the a data represents the least-squares fits.    correlation coefficients and 95% confidence interval for the associations of ⌬͓TotHb͔ and ⌬S t O 2 vs HV, ⌬mPAP, and ⌬CO.

Dobutamine
Dobutamine increased the CO and mPAP in a dose dependent manner as shown in Figure 7͑a͒. Cardiac output increased by ϳ16% and ϳ32% in response to dobutamine doses of 30 and 40-50 g/kg, respectively. Likewise, dobutamine doses of 30 and 40-50 g/kg increased mPAP ϳ25% and ϳ35%, respectively. Figure 7͑b͒ shows the mean changes in ͓TotHb͔ and S t O 2 in response to administration of dobutamine. The ͓TotHb͔ measured increased ϳ2% and ϳ2.5% in response to dobutamine doses of 30 and 40-50 g/kg, respectively, while S t O 2 increased ϳ0.5% for doses between 30 and 50 g/kg.
The percentages of change in ͓TotHb͔ and S t O 2 were compared to the dobutamine-mediated increase in CO and mPAP. Figures 8͑a͒ and 8͑b͒ show ⌬͓TotHb͔ vs ⌬CO and ⌬S t O 2 vs ⌬CO, respectively. A correlation coefficient of 0.70 was observed for ⌬͓TotHb͔ vs ⌬CO. As shown in Figure 8 ͓TotHb͔ increased approximately 4% in response to a 50% increase in CO. This rate is slightly less than but comparable to the percentages of change in ͓TotHb͔ in response to a hypovolemiamediated decrease in CO, where ͓TotHb͔ decreased ϳ6% in response to a decrease of 50% in CO. The correlation between ⌬S t O 2 vs ⌬CO yielded a considerably smaller value of 0.42. Mixed S t O 2 increased by less than 1% in response to a dobutamine-mediated increase of 50% in CO. This is considerably less than the changes in S t O 2 due to the hypovolemiamediated decrease in CO, where S t O 2 decreased ϳ4% in response to a 50% decrease in CO. The correlation coefficients for ⌬͓TotHb͔ vs ⌬mPAP or for ⌬S t O 2 vs ⌬mPAP were 0.30 and 0.17, respectively. Table 5 summarizes the correlation coefficients and 95% confidence intervals for statistically significant associations in the dobutamine study.

Discussion and Conclusion
Acute hypovolemia is the leading cause of shock. 6,28 Early identification of impending tissue hypoxia accompanied by rapid intervention constitutes the most effective way in which to prevent organ failure. 29,30 Presently, the recommended strategy for identifying hypovolemic shock is to assess the clinical signs and symptoms, central blood volume, cardiac output, arterial blood chemistry, and systemic blood pressure. 3,31 Pulmonary wedge pressure ͑PWP͒ has been shown to be a sensitive and accurate indicator of central blood volume; 32 low PWP means that the preload volume is insufficient to maintain adequate cardiac output. However, measurements of PWP necessitate invasive monitoring devices that include placing a balloon catheter through the right ventricle into the pulmonary artery. Some of the potential complications of PWP measurements include infection, the rupture of veins, embolisms, and pneumothorax. 33,34 Pulmonary artery thermodilution catheterization is the most commonly accepted method for measuring CO in a clinical setting. [35][36][37] Like PWP measurements, the procedure is invasive. Doppler ultrasound is a noninvasive alternative for measuring cardiac output. 38 However, Doppler ultrasound is less accurate than invasive procedures, it cannot be accurately obtained in many cases, it requires a highly experienced operator, and it cannot be used for continuous monitoring. 39 Near real-time magnetic resonance imaging is potentially an ideal modality by which to quantify cardiac output as well as to assess anatomical features of the heart. Presently, the use of magnetic resonance imaging for cardiac output measurements is expensive, and is not practical for unstable patients or for frequent serial assessments. Systemic blood pressure and arterial hemoglobin oxygen saturation (S a O 2 ) are noninvasive measurements, but are not sensitive indicators of tissue perfusion and oxygenation. Reflex mechanisms, such as systemic vasoconstriction and increased cardiac output can maintain nearly constant blood pressure when faced with severe hypovolemia. Thus, blood pressure decompensation often indicates that shock has progressed or advanced to irreversible stages of organ damage. 3 Because S a O 2 represents the oxygenation state of arterial blood, it is an indicator of the oxygen exchange processes of the lung, not of the oxygenation status of local tissues nor a measure of the delivery of oxygen to tissues. Although the product of S a O 2 and CO is a good indicator of the delivery of whole body oxygen, it is only an indirect measure of local tissue oxygenation because regional levels depend on the O 2 distribution as well as utilization. In fact, many of the current physiologic measures ͑CO, PWP, S a O 2 , etc.͒ primarily indicate the perfusion and oxygenation status of the whole body. These whole body hemodynamic parameters may poorly represent delivery of oxygen and utilization of certain organs. 40,41 Percutaneous oxygen electrodes measure tissue oxygenation in the local milieu of the sensor. 42 However, electrode-based measurements often reflect the oxygenation status of superficial tissues, they are locally invasive, and they are unreliable in vivo for a variety of reasons. Consequently, a noninvasive method that assesses the local tissue oxygen status and, at the same time, correlates to standard hemodynamic parameters is of great clinical interest. 4,29,43 FDPM-derived physiologic properties ͓TotHb͔ and S t O 2 are direct indicators of the status of local tissue oxygenation. ͓TotHb͔ represents hemoglobin in the arteries, capillaries, and veins of tissues at the measurement site. Likewise, S t O 2 quan-  tifies the mixed arterial-venous oxygen saturation of local tissues. ͓TotHb͔ and S t O 2 values depend not only on the delivery of oxygen but also on utilization of local tissue oxygen. ͓TotHb͔ decreases in proportion to a reduction in delivery (DO 2 ) of regional oxygen. Vasoconstriction and decreased hematocrit are likely to accompany the lowering of ͓TotHb͔. Mixed tissue saturation, i.e., S t O 2 , decreases when the utilization of oxygen exceeds delivery. In fact, FDPM-derived ͓TotHb͔ and S t O 2 have experimentally been shown to be sensitive to changes in tissue perfusion and oxygenation. 17,25,44 Moreover, ͓TotHb͔ and S t O 2 may also be sensitive to perturbations in CO, PWP, and S a O 2 , since these parameters clearly affect local tissue oxygenation. The measurements of CO and mPAP during administration of dobutamine were consistent with the known pharmacological effects of dobutamine on the circulatory system. Dobutamine increases cardiac output and elevates pulmonary artery pressure. 45,46 Our results show that dobutamine increased CO and mPAP in a dose-dependent manner ͑Figure 7͒. Analysis of the dobutamine data indicates that FDPM-derived ͓TotHb͔ and S t O 2 are statistically correlated with dobutaminemediated changes in CO. The dobutamine-mediated changes in CO are comparable to the sensitivity of ͓TotHb͔ to hypovolemia-mediated changes in CO. Interestingly, the sensitivity of ⌬S t O 2 to dobutamine-mediated ⌬CO was considerably lower than its sensitivity to hypovolemia-induced ⌬CO. Specifically, S t O 2 increased 1% per 50% increase in dobutamine-mediated ⌬CO, while S t O 2 decreased 4% in response to a 50% decrease in hypovolemia-mediated ⌬CO. A plausible explanation may be that hypovolemia causes intense stress not found in the administration of dobutamine. First, hypovolemia caused systemic vasoconstriction and cardiac output may be preferentially shunted away from the region of the skin to the vital organs, thereby exaggerating the decrease of S t O 2 in the periphery. Second, tissues may enter hypercatabolic states in response to ischemia, leading to maximal extraction of oxygen from arterial and venous blood as well as causing acidosis. Both factors contribute to the heightened S t O 2 sensitivity to CO perturbation observed during hypovolemic shock. In the case of the administration of dobutamine the delivery of oxygen exceeds the demand. The excess oxygen contributes to an increase in S t O 2 , but the tissues' oxygen saturation clearly does not change to the degree observed in hypovolemia.
FDPM technology is capable of separately measuring a and s Ј . So far, little is known about the effect͑s͒ of perfusion changes on tissue s Ј . In this study, tissue s Ј did not show statistically significant changes in response to hemodynamic Correlation coefficient (r) and 95% confidence intervals for the data are shown in (a) and (b). All data from animals in the dobutaminetreated group are combined. manipulations. This observation is consistent with the fact that FDPM measurements were performed during the acute phase of tissue ischemia. Because s Ј depends mainly on the tissues' structural milieu, 20,21 it may be sensitive to long-term pathophysiologic processes that alter tissue architecture, including cellular swelling, reperfusion injury, tissue necrosis, and wound healing. These processes do not occur during acute hypovolemia, which may explain why s Ј was unperturbed following acute hemorrhaging or administration of dobutamine.
In summary, the results from this pilot study using a rabbit model indicate that the quantitative, FDPM-derived ͓TotHb͔ and S t O 2 values statistically correlate with blood loss as well as to hypovolemia-and dobutamine-mediated perturbations in CO and mPAP. Significant is the fact that ͓TotHb͔ and S t O 2 are responsive to ͑1͒ blood loss of as little as ϳ7% and ͑2͒ changes in mPAP and CO of as little as 15%. These results suggest that ͓TotHb͔ and S t O 2 may be used to noninvasively identify impending hypovolemic shock and to monitor the efficacy of therapeutic interventions, such as restoration of the blood volume and administration of dobutamine. Although we have specifically addressed hypovolemic shock in this study, noninvasive FDPM spectroscopy may also be useful for other clinical manifestations associated with tissue hypoxia, such as peripheral vascular disease 47 and septic or cardiogenic shock. FDPM technology warrants further investigation in surgical and critical care settings where quantitative, dynamic monitoring of local tissue oxygenation would be of vital clinical importance.