INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Estimating the extent of pulmonary edema and tissue destruction impairing blood oxygenation is crucial (1) for monitoring clinical status in acute lung injury (ALI), including acute respiratory distress syndrome (ARDS) (2) and reimplantation response after lung transplantation (3). Although several techniques have been developed to evaluate these pathological changes in ALI, none has proved entirely satisfactory in clinical situations. Chest roentgenography is a convenient and widely used method for the diagnosis of pulmonary edema, but is not sensitive enough to detect subtle but critical increases in lung water content, and differentiating atelectasis from edema is difficult by this method (2, 4, 5). Arterial blood gas analysis is a well established method of generally evaluating pulmonary oxygenation function, but is incapable of estimating regional or unilateral changes of the lung. Computed tomography can estimate the pathological change of injured lung (6), and nuclear magnetic resonance imaging is a reliable and noninvasive method of assessing lung tissue water content (7, 8). However, these methods are not generally suitable for the intensive care unit setting. The double-indicator dilution method is convenient to use for bedside monitoring (1, 9), but is influenced by cardiac function and the distribution of ventilation and perfusion (5). Thus, a sensitive, clinically convenient, noninvasive method is currently desired to assess pulmonary edema and tissue distraction in ALI (2, 10).

We have investigated the potential of in vivo near-infrared (NIR) spectroscopy to determine lung tissue water content as a parameter of pulmonary edema. NIR spectroscopy was introduced into the medical field by Jöbsis (11), and has been applied to the noninvasive monitoring of regional blood volume changes, hemoglobin (Hb) saturation, and the redox state of cytochrome oxidase (Cyt.aa₃) in intact solid organs, such as brain, muscles, and liver (12). We have recently demonstrated the feasibility and validity of this technique for evaluation of tissue oxygenation in living lungs (16). Because water has a unique absorption band that is different from that of any biological pigments in the NIR region (17), we predicted that changes in lung water content could be quantified by NIR spectroscopy, and it could serve as a useful parameter for monitoring the pathological status of ALI, in addition to pulmonary oxygenation. In the present studies, we demonstrated the ability of NIR spectroscopy to assess changes in water content in an in vitro model simulating the spectrophotometric characteristics of the lung. We next estimated the sensitivity and feasibility of the NIR spectroscopy for detecting pulmonary water content and Hb oxygenation in living lungs in an oleic acid (OA)-induced ALI model in rats.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

NIR Spectroscopic Detection of Changes in Water Content in In Vitro Model

To investigate directly the ability of NIR spectroscopy for detecting changes in water volume in biological materials containing air, we constructed a model (phantom), which simulates spectrophotometric characteristics of the lung, by modifying a previously reported in vitro system (Figure 1A) (16).

View larger version (27K): [in this window] [in a new window]   Figure 1.   In vitro model simulating the spectrophotometric characteristics of the lung. (A) Schematic illustration of a transverse view of the in vitro model. The in vitro model consists of a glass test tube (outer diameter 10 mm, inner diameter 8 mm) containing a red blood cell (RBC) suspension and 18 sealed capillary tubes (hematocrit tubes). The hematocrit tubes were filled with either air or water in varying ratios (number of tubes containing air: number of tubes containing water = 18:0, 12:6, 6:12, and 0:18), and were sealed. They were then closely packed into the glass test tube in random fashion. Human blood with sodium citrate was suspended in phosphate-buffered saline at two different Hb concentrations (0.6 mM and 1.0 mM). The gaps between the hematocrit tubes were filled with the RBC suspensions with and without milk (scattering material) containing fat and biological proteins, as previously reported by other and our groups (16, 18). (B and C) Relationship between H2O measured by NIR spectroscopy and number of hematocrit tubes containing water in two Hb concentrations, 0.6 mM (open circles) and 1.0 mM (open squares). Linear correlation between the changes in H2O and number of hematocrit tubes containing water without milk and with milk (B: without milk, r = 0.982, p < 0.0001; C: with milk, r = 0.918, p < 0.0005). The data obtained by NIR spectroscopy are shown in arbitrary units (AU).

A multichannel photodetector (MCPD-2000; Otsuka Electrical Co., Osaka, Japan) with quartz optical fibers and a 300-W halogen lamp (Petite Ace 25; Sanyo Denki Co., Tokyo, Japan), as a light source, were used to observe variations in water volume in this system. The tips of the optical fibers were fixed on the test tube with a special attachment. The reflected light from the in vitro system was scanned within the 500 to 1,100 nm ranges, and the sampling time of each scan was 200 ms. The spectra in a series of 20 scans were averaged with a personal computer (PC-9821 Xs; NEC, Tokyo, Japan). Absorption spectra were transformed as described previously (16, 18, 19) to correct their flattening shape caused by light scattering. The differences between the corrected spectra of the model without water-containing hematocrit tubes and that of models containing different numbers of water-containing tubes were calculated. Multicomponent analysis of differences in the spectra was performed in a wavelength range of 700 nm to 1,000 nm by using an equation based on the Beer-Lambert law:

(1)

where OD(), L(), and e_1-3() are optical density, mean path length, and extinction coefficients, respectively, of each component at a wavelength of ; oxy-Hb is oxyhemoglobin and deoxy-Hb is deoxyhemoglobin. Least-square curve-fitting was used to calculate the existential rate of the three components (20).

OA-induced ALI in Rats

All procedures involving rats were performed according to the guidelines of the National Institutes of Health Guide for the Care and Use of Laboratory Animals.

Male Wistar rats weighing 300 to 400 g (Charles River Japan, Yokohama, Japan) were prepared with intramuscular injection of atropine (0.4 mg/kg) and intraperitoneal anesthesia with sodium pentobarbital (50 mg/kg) and ketamine (40 mg/kg). After endotracheal intubation with a 16-gauge polyvinyl tube (Terumo Co., Tokyo, Japan), the rats were mechanically ventilated (SN-480-7; Shimano Co., Tokyo, Japan). The ventilator settings were: tidal volume, 10 ml/kg; respiratory frequency, 80 breaths/min; and positive end-expiratory pressure, 3 cm H₂O. The left carotid artery was cannulated with a 3-F polyethylene tube (Atom Co., Tokyo, Japan) for blood pressure monitoring and arterial blood sampling. The trachea was exposed and encircled with 5-0 silk braid in order to fix the intubation tube; the trachea was ligated at different times.

The rats were divided into the three experimental groups according to differences in ventilation settings. In the initial ligation group (n = 13), tracheal ligation was performed immediately before injection of OA (Nacalai Tesque Inc., Kyoto, Japan) (0.1 ml/kg) into the penile vein following the NIR spectroscopic measurements. In the late ligation group (n = 10), to produce a severer grade of lung injury, tracheas were ligated 20 min after OA injection. Time-matched sham rats (sham group: n = 4), in which saline (0.1 ml/kg) was administered via the penile vein following the same procedure as in the initial ligation group, served as a control.

In Vivo NIR Spectroscopy

For in vivo NIR spectroscopy, the same instruments as those used in in vitro study were used. After performing thoracotomy in the left fifth intercostal space, the tips of the fiber bundles for NIR spectroscopy (the distance between the bundles was 3 mm) were fixed at a position approximately 3 mm above the lung, and the NIR spectrum under control conditions was obtained before OA or saline injection. NIR spectroscopic measurements were performed at 5-min intervals for 60 min after OA or saline injection as described previously (16). Briefly, the differences between the spectra obtained from the lungs before and after OA or saline injection were subjected to multicomponent analysis, i.e., the difference in the gross optical densities between them was calculated at 2-nm wavelength intervals to obtain the "subtracted spectrum." Then multicomponent analysis of the "subtracted spectrum" was performed by use of least-square curve-fitting on the basis of singular value decomposition to quantify the changes in each component (20). The five components were fitted with the following equation:

(2)

The obtained data are presented as relative changes in the optical density of each component at the NIR range (700 to 1000 nm).

Gravimetric Measurements of Lung Water Content

The left lung was excised and weighed immediately after completion of NIR spectroscopic measurement, and dried for 48 h at 60° C to make gravimetric measurements of lung water content (LW_Gr: [wet lung weight  dry lung weight]/dry lung weight; g H₂O/g dry lung).

Blood Gas Analysis

Whole blood was sampled from the left carotid artery immediately after thoracotomy and at the end of the NIR spectroscopic measurements, and Sa_O2 was measured with a blood gas analyzer (ABL 510; Radiometer Trading Co., Copenhagen, Denmark).

Histological Examination

The right lung was excised after completion of NIR spectroscopic measurement, and was fixed in 10% buffered formalin, embedded in paraffin, sectioned, and stained with hematoxylin-eosin.

Statistical Analysis

All data are shown as mean ± SEM. The paired t test was used for comparisons of Sa_O2 between pre- and post-OA administration. Statistical analysis between experimental groups was performed by one-way analysis of variance, and Scheffé's F tests were applied to compare the individual groups. Regressions between variables were assessed by linear regression analysis. Statistical analysis was carried out with StatView 4.5 package software (Abacus Concepts Inc., Berkeley, CA) for Macintosh. Significance was accepted at p values < 0.05.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Ability of NIR Spectroscopy to Detect Changes in Water Volume (In Vitro Study)

Figure 1B shows the relationship between the changes in water volume measured by NIR spectroscopy and the number of hematocrit tubes containing water. Under both conditions with and without milk (enhancing photoscattering), there was significant correlation despite the differences in Hb concentration (without milk: r = 0.982, p < 0.0001, and with milk: r = 0.918, p < 0.0005), indicating the ability of NIR spectroscopy to quantify changes in water contents in biological materials.

Histology of OA-induced ALI in Rats

In the sham group, the gross appearance of the lungs did not change throughout the examination, and in the initial ligation group, the lungs maintained the same degree of expansion as in the sham group. In contrast, in the late ligation group, the lungs gradually deflated for the first 20 min after OA administration, but by ligating the trachea the lungs were expanded and kept inflated during the last 40 min, the same as in the initial ligation group or the sham group.

A summary of the histological findings is shown in Table 1. In the initial ligation group, mild to moderate intra-alveolar edema and mild perivascular cuffing were observed (Figure 2A), and focal intra-alveolar hemorrhages and septal necrosis were also noted. In the late ligation group, all findings progressed more extensively than in the initial ligation group (Figure 2B), and these findings were similar to the histological findings of ARDS. No clear findings except capillary congestion were seen in the sham group.

View larger version (156K): [in this window] [in a new window]   Figure 2.   Photomicrographs of tissue section from the initial ligation group (A) (hematoxylin-eosin stain; original magnification: ×200) and late ligation group (B) hematoxylin-eosin stain; original magnification: ×200). (A) Moderate capillary congestion, intra-alveolar edema, and focal septal necrosis are seen. (B) Marked intra-alveolar edema, multifocal septal necrosis, and capillary congestion are seen.

In Vivo NIR Spectroscopic Measurement

Figure 3A shows representative absorption spectra of the lungs in each group, and Figure 3B shows standard absorption spectra used for multicomponent analysis of those lung spectra. In the OA-administered groups, the absorption at 760 nm (peak for deoxy-Hb) and 975 nm (peak for water) was elevated when compared with that in the sham group.

View larger version (17K): [in this window] [in a new window]   Figure 3.   (A) Representative NIR spectra, 60 min after OA administration. Elevation of spectra around 760 nm (deoxy-Hb) and 975 nm (water) were observed in the OA-administered groups. (B) Absorption spectra used as standard spectra in the multicomponent analysis of NIR spectroscopy. The spectrum of water has a strong and characteristic peak around 975 nm and that of deoxyhemoglobin also has a peak around 760 nm.

Time-course changes in lung water content measured by NIR spectroscopy (LW_NIR) are shown in Figure 4A. After OA administration in the initial ligation and late ligation group, LW_NIR increased similarly by 20 min (at the time of tracheal ligation in the late ligation group). In the initial ligation group, LW_NIR increased gradually and reached a plateau. In contrast, LW_NIR in the late ligation group increased linearly to almost twice that in the initial ligation group until 60 min after OA administration. In the sham group, LW_NIR was constantly maintained at the initial level throughout the observation period.

View larger version (14K): [in this window] [in a new window]   Figure 4.   Time-course change of LWNIR (A) and correlation between LWNIR and LWGr (B). (A) Closed squares: late ligation group; closed circles: initial ligation group; closed triangles: sham group. Results are expressed as mean ± SEM. (B) Relationship between LWGr and LWNIR. closed circles: late ligation group; closed squares: initial ligation group; open squares: initial ligation (20 min) group; closed triangles: sham group. Initial ligation (20 min) group: five rats in the initial ligation group in which measurements were made for only 20 min to evaluate progression of the edema.

Table 2 shows the relative changes in each parameter (LW_NIR, oxy-Hb, deoxy-Hb, oxidized-Cyt.aa₃, and reduced-Cyt.aa₃) of NIR spectroscopy at the end of observation (60 min) in each group. LW_NIR increased significantly in the OA-injured lung, i.e., the initial ligation group and the late ligation group, compared with the sham group (p < 0.005, and p < 0.0001), and LW_NIR in the late ligation group was significantly higher than in the initial ligation group (p < 0.01). Oxy-Hb was increased in all groups after injection of OA or saline, and oxy-Hb in the late ligation group was considerably increased. Deoxy-Hb in the late ligation group was also significantly increased compared with the sham group (p < 0.05).

Oxidized-Cyt.aa₃ showed no significant differences in any of the groups. However, reduced-Cyt.aa₃ in the late ligation group increased significantly compared with the other groups (versus sham group p < 0.0001 and versus initial ligation group p < 0.05), and in the initial ligation group it also increased compared with the sham group (p < 0.05).

Correlation between NIR Spectroscopic Data and Blood Gas and Gravimetric Data

The results of blood gas analysis and gravimetric measurement of lung water content are summarized in Table 3. In the OA-injected group, LW_Gr 60 min after OA administration was significantly higher than in the sham group (p < 0.05), and in the late ligation group LW_Gr was higher than in the initial ligation group (p < 0.05). There were no significant differences in Sa_O2 before injection (Sa_O2-pre) between the groups. In the OA-injected group, Sa_O2 60 min after injection (Sa_O2-post) was significantly lower than Sa_O2-pre in every group (initial ligation group: p < 0.05; and late ligation group: p < 0.001). Sa_O2-post in the late ligation group was significantly lower than in the sham and the initial ligation groups (p < 0.05). Sa_O2 tended to decrease in the OA-injected groups but there were no significant differences between the experimental groups.

The relationship between LW_Gr and LW_NIR is shown in Figure 4B. A statistically significant positive linear correlation was found between them (n = 27, r = 0.877, p < 0.0001). In addition, there was a statistically significant linear correlation between the deoxy-Hb and Sa_O2 (n = 27, r = 0.798, p < 0.0001) (not shown).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

In the 700 to 1,300 nm range of NIR light, a significant amount of radiation can be effectively transmitted through biological materials over long distances. In vivo NIR spectroscopy is based on the fact that the absorption intensity of NIR light depends on the level of Hb oxygenation, the redox state of Cyt.aa₃, and the biological pigment content (11). Thus changes in each characteristic absorption spectrogram noninvasively provide useful information about quantitative changes in the component (21, 22). Water also has special absorption bands that differ from those of any biological pigments in the NIR range (17). It has a strong well-defined absorption feature at 975 nm and has weaker overtones at 820 nm and 760 nm (Figure 3B). In the medical field, NIR spectroscopy is currently being applied to evaluation of oxygen availability and oxygen consumption relationships within the living tissue such as brain, muscle, and liver (12, 21), and to measure tissue water content in the skin and the bovine lens (23, 24). These applications encouraged us to examine the potential of NIR spectroscopy for quantifying tissue water content in the living lung for diagnosis of pulmonary edema, because the lung is susceptible to edema as a result of acute injury and the water content is an important indicator, along with the tissue oxygenation state, for monitoring its pathological status (1). We previously demonstrated the applicability of in vivo NIR spectroscopy to the living lung as a means of assessing the tissue oxygenation state (Hb oxygenation, Hb volume, and redox state of Cyt.aa₃) of the lung (16). However, quantification of lung water content by NIR spectroscopy was not evaluated, because the NaCN administration and hypoxia loading in our prior model of impaired oxygenation did not allow evaluation of fluxes in lung water content.

The present in vitro study showed that NIR spectroscopy could precisely detect the changes in water volume displacing air space in the hematocrit tubes. The model used in the in vitro study resembles the lung in terms of containing scattering materials (cells and fat), chromophore (Hb), and air which possibly affect photon scattering. The water volume data measured by NIR spectroscopy were unaffected by either changes in Hb concentration or by enhanced scattering (which resulted from adding milk). These results show that NIR spectroscopy can be used to quantify changes in lung water content despite varying regional Hb concentrations and individual variations in scattering materials in lung tissues.

To further evaluate the potential of in vivo NIR spectroscopy to quantify lung water content, we applied NIR spectroscopy to OA-injured lung, widely used in experimental models of ARDS, because of its similar morphologic and cellular changes (25, 26). Other investigators have reported that the degree of pulmonary edema and hypoxia induced by OA is affected by ventilation conditions (27). In our preliminary experiments, we found that delayed timing of tracheal ligation led to inadequate inhalation because of decreased compliance and facilitated the OA-induced lung injury. By varying the timing of tracheal ligation, we were able to induce different grades of ALI (mild/moderate, and severe) in OA-injected rats. In this model, LW_NIR correlated significantly with LW_Gr. The histological findings revealed that LW_NIR reflected the degree of pulmonary edema in each group. Even in histologically mild ALI (in the initial ligation group), which showed only subtle changes in Sa_O2, NIR spectroscopy was able to detect the increased lung water content, indicating that this method is sensitive enough to diagnose pulmonary edema in the early phase of ALI and to estimate therapeutic interventions. Furthermore, NIR spectroscopy allowed accurate monitoring of sequential changes in lung water.

Our method enables simultaneous estimation of lung water content and tissue oxygenation, by multicomponent analysis with the curve-fitting method using the spectra of purified standards. OA administration increases the intrapulmonary fraction (28), which is associated with the increase of deoxygenated Hb in the lung tissues. In the present study, NIR spectroscopy detected the increases in deoxy-Hb in OA-injured lung, and the elevations were correlated with decreases in Sa_O2. In the late ligation group, oxy-Hb also increased (i.e., total Hb increased), which might be caused by the vascular congestion and the reduced oxygen consumption at the cellular level in severely injured lung tissues.

Noninvasive estimation of tissue viability by measuring Cyt.aa₃ is one of the features of NIR spectroscopy (15, 21, 29), although there is still some controversy about its accuracy and reliability in estimating Cyt.aa₃ (30). Our previous report demonstrated the usefulness of NIR spectroscopic assessment of Cyt.aa₃ level in rat lungs exposed to NaCN (16). Cyt.aa₃ levels measured by NIR spectroscopy in the present study were markedly decreased in the late ligation group, in which multifocal septal necrosis was observed histologically, indicating consistency between the Cyt.aa₃ data and the histological picture. Thus, our NIR spectroscopy can simultaneously provide three important indicatorstissue water content, Hb oxygenation, and the redox state of Cyt.aa₃for the precise diagnosis of ALI. Considering the portable, nondestructive, and continuous availability of our NIR spectroscopic system, this method may be a useful tool in clinical pulmonary care.

In the in vivo experiment, we performed a thoracotomy for NIR spectroscopic measurement to obtain a pure spectrum directly from lung without interference of spectra from adjacent tissues (i.e., thoracic wall or heart). For the same reason, we fixed the tips of the optical fiber bundles as close to the lung surface as possible without direct contact that might affect local concentration of spectrophotometric parameters. Although this measurement is nondestructive to lung tissue, a noninvasive system without thoracotomy is ultimately desired to bring this technique closer to clinical application. Previous reports have demonstrated that NIR light penetrates through the living tissue up to 6 cm in human forearm (31), and that NIR spectroscopic extracranial measurement is able to estimate cerebral tissue oxygenation in infants (12). Considering such good transparency of biological tissues to NIR light, noninvasive NIR measurement through the thoracic wall should be theoretically possible. Technological improvement such as a more powerful light source and more effective probe are needed to facilitate noninvasive clinical NIR spectroscopic monitoring of the lung. Furthermore, simultaneous multipoint scanning, which allows subtracting the spectrum of superficial skin or muscles from the measured gross spectrum (32), should be employed to obtain the pure spectrum from the living lung through the thoracic wall. These possibilities are currently under investigation.

In regard to "field of view" of NIR spectroscopic measurement, it has been established that the "mean light pathlength," labeled as "field of view" in vivo spectroscopy, is four- to sixfold over the distance between the transmitting and receiving optical bundles when those are vertically applied to the tested tissues and the intensity of light source is adequate (33). Thus, the spectrophotometric view could be adjusted by varying that distance and the intensity of light source. We preliminarily confirmed the validity of this theory by use of our in vitro model, i.e., our NIR spectroscopy could detect the absorption spectrum of indocyanine green in a capillary tube placed at 10 mm in depth when the distance between the transmitting and receiving optical bundles was 2.5 mm. In our in vivo NIR spectroscopic system, the distance between the transmitting and receiving optical bundles was 3 mm, thus the mean light pathlength, spectrophotometric view, was theoretically 12 mm, which was long enough to represent a lobe of rat lung. If NIR spectroscopy was clinically applied, however, multipoint measurement of NIR spectroscopy should be employed to obtain a representative absorption spectrum of the injured lung rather than that of a limited area of lung in the nondependent area.

Another concern related to the clinical use of NIR spectroscopy is the possible dynamic changes in cellar density of the lung tissue (i.e., in case of atelectasis or hyperinflation of lung), which might affect the data of the NIR spectroscopy because such geometrical change of the tissue affects the light scattering and absorption (34). However, the changes in cell density of the lung tissue might affect not only water absorption but also other components including oxy-, deoxy-Hb, and oxidized-, reduced-Cyt.aa₃, thus these parameters might show similar trend of their alteration. Therefore, by comparing each component, we could judge the influence of geometrical change of lung tissue.

In conclusion, our NIR spectroscopic measurements allow simultaneous evaluation of both the lung water content of the lung and Hb oxygenation in OA-injured lung and may be useful in assessing ALI, such as ARDS and ischemia-reperfusion injury.