Blood rheology (also termed hemorheology) is an integrated branch of physics and medicine that deals with the blood flow and deformation behavior of blood components (1, 2). Changes of blood rheology are thought to be crucial in the progression of many severe diseases, such as stroke, heart attack, anemia, diabetes, and sickle cell disease (1, 3–5). These changes may also dramatically complicate therapeutic interventions (e.g., infusion of heparin or warfarin). The grand challenge of hemorheology in live organism is the complexity of multiple dynamic relationships. The most important rheological property of blood is its resistance to flow, or viscosity. The major flow-affecting determinants of whole-blood viscosity, and thus hemorheology, are 1) hematocrit (Ht), the fraction of the blood volume occupied by red blood cells (RBCs); 2) RBC deformability, the ability of RBCs to undergo deformation in flow, depending on the shape, membrane deformability, and other properties (e.g., chemical structure or clustering capability) of hemoglobin (Hb); and 3) RBC aggregation, the formation of reversible (physiological) rouleaux or irreversible (pathological) clumps of RBCs. Viscosity is also determined by shear rates (velocity difference) and flux of RBCs (number of RBCs per second)(1–4). Another rheological determinant, plasma viscosity, usually has little influence on whole-blood viscosity. Indeed, when RBCs are added to plasma, blood viscosity becomes increasingly sensitive to Ht, with an almost exponential relationship between the Ht value and blood viscosity (1). In particular, at moderate-to-high shear rates, a 1% change in Ht (from 45% to 46%) increases blood viscosity by 4%.
Hemorheology has been intensively studied for decades by many methods ex vivo, including rotational, viscosimetry, micropipette aspiration, sedimentation rate, micropore filtration, and flow cytometry (1–5). However, these methods are limited by one or more of the following factors: 1) single-parameter measurement (e.g., RBC aggregation or Ht); 2) invasiveness, which may unpredictably alter rheological properties and prevent long-term monitoring in the native biological environment; 3) time-consuming preparation procedures (several hours if not an entire day); 4) discontinuous sampling with limited, discrete time points; and 5) the small blood volume extracted (typically a few milliliters). Furthermore, ex vivo/in vitro tests cannot replicate the transient flow-affecting character of rheological changes, which only become fully apparent in the natural blood circulation in vivo under the influence of endothelial, nervous, and humoral stimuli.
Technological progress in the development of methods for evaluating blood rheology in vivo promises to extend its clinical significance (2). Among numerous in vivo diagnostic techniques (e.g., ultrasound, magnetic resonance imaging, positron emission tomography, and various optical techniques), photothermal (PT), and photoacoustic (PA) methods have exhibited a high level of sensitivity in detecting of individual cells (6–10) and monitoring oxygenation and total Hb content in humans (11–13). Fluorescence techniques can also detect and image individual cells and blood flow (14–18). However the use of fluorescence labeling in vivo raises potential problems for translating this technology to humans because of1) the cytotoxicity of available fluorescent tags,2) photobleaching or blinking of tags,3) undesired immune responses to tags, and4) the strong influence of scattering light and an autofluorescence background, which allow the assessment of only superficial microvessels with a slow flow velocity. Our contribution in this field includes the development of PA and PT flow cytometry (PAFC/PTFC) for in vivo real-time detection of individual circulating cells either with intrinsic absorbing markers (e.g., Hb in RBCs or melanin in melanoma cells) or synthetic gold and other nanoparticles as PT and PA contrast agents (7–10, 19–26). PT spectroscopy has demonstrated capabilities to identify several Hb chemical subtypes (e.g., sulpha-, met-, CO- or HbS [in sickle cells]) (27). Our group also developed Raman and integrated PA-PT-Raman spectroscopy and imaging for in vivo identification with chemical specificity of individual cells and nanoparticles (28, 29). However, the application of PT and PA techniques to the study of blood rheology has not yet been reported. Here, we analyze the capabilities of a combined platform (Fig. 1), for integrating high-speed PT, PA, and optical imaging for in vivo real-time monitoring of multiple rheological parameters at the single-blood-cell level.
Two integrated setups were built on the technical platform of an upright and invert microscopes (Olympus BX51 and Olympus IX81, Olympus America) and two tunable optical parametric oscillators (OPO) (Lotis Minsk, Belarus and Opolette HR 355 LD, OPOTEK, Carlsbad, CA) producing pump laser pulses with the following parameters: wavelength, 420–2,300 nm and 410–2500 nm; pulse width, 8 ns and 5 ns; a repetition rate, 10 Hz and 100 Hz; pulse fluence, 0.01–100 J/cm2 and 0.001–20 J/cm2, respectively, as described elsewhere (21, 30, 31, 32–36 ). PA signals were detected using two ultrasound transducers: (1) unfocused, model 6528101, 3.5 MHz, 5.5 mm in diameter (Imasonic); and (2) focused, model V316-SM, 20 MHz, focal length, 12.5 mm (Panametrics-NDT). The signals after amplifier (model 5662: bandwidth 50 kHz–5 MHz, gain 54 dB; and model 5678: 40 MHz, and gain 60 dB; both from Panametrics-NDT) were recorded with a PC and a Tektronix TDS 3032B oscilloscope, and analyzed with standard and customized software. The transducer was gently attached to the samples and warm water was topically applied for good acoustic coupling.
In one channel PT thermal lens schematic (30), pump laser-induced temperature-dependent variations of the refractive index around absorbing zones caused defocusing of a continuous-wave collinear probe beam from He-Ne laser with wavelength of 633 nm and power of 1.4 mW (model 117A, Spectra-Physics). The subsequent change in the beam's intensity at its center (referred to as PT signal) is detected after passing through a pinhole by a photodiode with built-in preamplifier (PDA36A, 40 dB amplification, ThorLabs). PT signals were detected either from whole cells using relatively large laser beam with diameter of 10–30 μm or from localized intracellular zones using focused laser beams with diameter of 300–700 nm.
PT imaging (PTI) was performed in two setups using different modes. In the first one using the upright microscope, laser-induced the refractive index variations were visualized with a multiplex thermal-lens schematic and a CCD camera (AE-260E, Apogee) using a pump laser pulse from OPO and second, probe pulse from Raman Shifter (wavelength, 639 nm; pulse width, 12 ns; pulse energy, 2 nJ; pulse delay, 0–10 μs) (33). Refractive index variations were measured by comparison of the sample probe images before and right after the excitation with pump laser pulse (31, 33) with each pixel of the CCD camera acting as an individual photodiode with pinhole. The pump and probe beam diameters were adjustable (range 10–60 μm and 5–50 μm, respectively) by changing condensers and their axial moving. The lateral resolution was 700 nm with a 20× microobjective and 450 nm with a 60× water-immersion objective. The cells in the standard microscopic slide were navigated by using an automatic scanning microscopic stage (Conix Research) and in-house written Visual Basic™ software.
In the second setup using the inverted microscope, imaging was provided by scanning cells with a two-dimensional (2D; X-Y) translation stage (H117 ProScan II, Prior Scientific, Rockland, MA) having a positioning accuracy of 50 nm (34, 35). The intensity of each pixel of PT image represents the average of PT signals from several OPO pulses. The lateral resolution of the PTI level at 0.5–6 μm was determined by the focal spot size of the pump laser beam. This setup was equipped with a high-speed (200 MHz) analog-to-digital converter board (National Instruments Corp., PCI-5152, 12-bit card, 128 MB of memory), specialized software (LabVIEW; National Instruments), and a Dell Precision 690 workstation with a quadcore processor, 4 GB of RAM, and Windows Vista 64-bit operating system.
High-resolution (∼300 nm) transmission digital microscopy (TDM) module with a cooled, color CCD camera (DP72, Olympus), a high-speed (up to 40,000 frames per second [fps]) CMOS camera (MV-D1024-160-CL8; Photonfocus AG, Lachen, Switzerland), and high sensitive CCD camera (Cascade: 512; Photometrics, Roper Scientific, Inc.) were used for the navigation of laser beams and cell imaging.
Preparation of Samples In Vitro
Blood samples were collected from donor mice and rats (protocols were approved by the University of Arkansas for Medical Sciences Institutional Animal Care and Use Committee). RBCs were prepared by their isolation through initial centrifugation (200×g for 6 minutes) of whole blood after the removal of plasma and leukocyte layers. Then, RBCs were washed via centrifugation at 1,000g for 10 minutes at room temperature. The pellet was then resuspended in PBS.
Diamide (Sigma), Chlorpromazine and distilled water were added to the RBC solution to form rigid RBCs, stomatocytes and spherocytes, respectively. RBC aggregation was produced by Dextran500 (MW = 500; Sigma). The Hb powder (Sigma) was dissolved in plasma to obtain physiological concentration of total Hb (110–160 g/L).
Each sample (∼8 μL) was placed in the individual well (S-24737, Molecular Probes) and assessed with PT/PA imaging and TDM. The well depth of 120 μm allowed keeping the original shape of RBCs and their aggregates.
In vivo experiments involved a nude mice (nu/nu), the genetically modified mice expressing human sickle hemoglobin (STOCK Hbatm1Paz Hbbtm1Tow Tg(HBA-HBBs)41Paz/J mice) and rats (Sprague-Dawley rats), in accordance with protocols approved by the University of Arkansas for Medical Sciences Institutional Animal Care and Use Committee. We selected rat mesentery as an almost ideal minimally invasive animal model because it has a thin tissue layer with the clearly distinguished single-blood vessels with individual fast moving cells (37). After anesthesia (ketamine/xylazine, 50/10 mg/kg, i.m.), the animal was laparotomized and a mesentery was exteriorized on a customized, heated (38°C) microscope stage. It was then suffused with warmed Ringer's solution (38°C, pH 7.4) containing 1% bovine serum.
Experiments with the mice were performed on a mouse ear, with well-distinguished blood vessels at 50–100 μm depth and 50–150 μm diameter. The anesthetized animal was placed on the heated microscope stage with a topical application of water on the ear for acoustic coupling of the ultrasound transducer and the tissue.
Various alterations of hemorheological status in vivo were produced by intravenous injections of different drugs with well-established effects: saline solution (400 μL; 0.9% solution of NaCl) to decrease Ht; Diamide (20 mg/kg) to produce rigid RBCs; distillate water to form swollen spherocytes); and Dextran (2.4 g/kg of blood) for RBC aggregation.
A minimum of three animals were used for each experiment unless otherwise noted. Results are expressed as a mean ± standard error. Spearman correlations for which P <0.05 were considered statistically significant. MATLAB 7.0.1 (MathWorks, Natick, MA), and LabVIEW (National Instruments) were used for the statistical calculations. Data were summarized as a mean, standard deviation (SD), median, interquartile range, and full range. Comparisons of PA and PT data were performed via scatter plot in conjunction with the Spearman correlation analysis.
Phenomenological Models of In Vivo Blood Rheology Using PT and PA Methods
Each physical process responsible for PT and PA effects (excitation → nonradiative relaxation → heating → thermal relaxation → acoustic wave) can be sensitive to rheological parameters. Because these effects are also spectrally, spatially, and energy selective, the changes of the laser wavelength, pulse energy, and beam diameter can be used to obtain more information of blood properties. For example, spectrally selective optical excitation may provide information on RBC chemical composition (27). The quenching of radiative relaxation (leading to increased PT/PA signal amplitude or a shortened rise time) (26) may depend on the RBC/Hb functional state and properties of surrounding proteins. The processes of thermal relaxation depend on RBC size and shape (31). Nonlinear effects at high laser energy levels, such as multiphoton absorption (7), may amplify spectral imaging contrast. Laser-induced nanobubble and microbubble formation (10, 30) around localized overheated zones with high Hb concentrations (38) can be a marker of abnormal Hb clustering within individual RBCs. The PT/PA effects may also be sensitive to the changes in blood rheology induced by environmental (e.g., nicotine, alcohol) or therapeutic (e.g., radiation, drugs) impacts. On the basis of our previous and current findings, we summarize below the methodological details for in vivo PT and PA measurement of blood rheological parameters.
High-speed imaging RBCs
Because of high endogenous absorption of Hb and low absorption of plasma proteins, label-free, high-speed (up to 10,000 fps), high-resolution (250–300 nm) transmission (39) and PT imaging (38, 40, 41) of thin tissue, such as ear and mesentery in animal models, enables time-resolved determination of the shape of single RBCs in a capillary with single-file, flow even in a microvessel (39). This can provide information on RBC flux, shear stress, and Ht. Resolution in the z direction (along the light path) can be increased by confocal PT microscopy (42). Pathological cell shapes (e.g., sickle cells) can be identified by high-resolution PT imaging. In addition, multiwavelength PAFC can provide measurements of circulating blood volume (43).
PT and PA measurements of individual RBC velocity and blood velocity profiles
Measurement of RBC velocity with high spatial resolution can be obtained by either (1) video-rate analysis (37); (2) label-free, two-beam (pump-probe) PT flow velocimetry, in which a pump laser induces a heated zone in blood flow while the movement of this zone with flow is detected through refraction or defocusing of the probe beam (time-of-flight technique) (19); or (3) monitoring of PA signal width determined by the transit time of RBCs through the laser beam (9, 25). In latter case, PA signal width is acquired by the averaging of many PA signals from the same cells with high-pulse-repetition-rate laser. The transit time of cells through the laser beam depends on cell velocity and beam and cell sizes. Fast scanning of a vessel cross-section with a strongly focused laser beam may be used to estimate velocity profiles. Because the behavior of cells in flow (e.g., velocity) depends on their morphological type, PT/PA measurements of velocity may allow cell discrimination (e.g., fast moving RBCs in axial flow is distinguished from rolling leukocytes near the vessel wall).
PT monitoring of RBC size and shape
Under the influence of a short laser pulse, the temperature elevation of RBC ΔT(t), referred to as PT signals, can be described in the first approximation as follows (30, 31):
where ΔTmax is the maximum temperature elevation in the irradiated cells, τRT is the rise time, and τT is the thermal relaxation time. The typical range of τRT is 0.1–10 ns, which depends on the nonradiative relaxation time, the time to the photodetector's response, the laser pulse width (tp), and spatial averaging of thermal effects from individually heated Hb molecules within the whole RBC (30). If the laser pulse width, tp, is shorter than the cell thermal relaxation time (thermal confinement, tp < τT), for RBCs with two basic geometries — (a) a sphere with radius R, and (b) a planar discoid with radius RD and thickness d(d ≪ RD) — the thermal relaxation time τT may be estimated, respectively, as follows (30)
where kT is the heat diffusion coefficient. For example, for a discoid RBC with d = 2 μm and a spherical RBC with R = 3 μm, estimations of τT (kT = 1.44 × 10−3 cm2/s for water parameters) are approximately 4.4 μs and 14.4 μs, respectively. Thus, PT signals demonstrate the standard fast-rising peak associated with rapid RBC heating and a slower trailing edge (referred to as the signal tail) related to the cooling of RBCs, with their shape and size dependent on time (Fig. 1, oscilloscope on bottom). As a result, the transformation of normal RBCs with a standard discoid form to a pathological sphere-like shape can be controlled directly in flow without imaging through the monitoring of the PT signal duration. This approach can also be used to monitor the volume of individual RBCs when they swell, shrink, or aggregate (see below).
PA and PT monitoring of hematocrit (Ht)
The acquisition of many PT/PA signals during irradiation of relatively large vessels with high-pulse-repetition-rate lasers creates a constant PT/PA background from RBCs (Fig. 1A, right, bottom insert). The average level of this background is proportional to the number of RBCs in the irradiated volume, thus allowing the estimation of the Ht after corresponding calibration taking into account blood oxygenation and hemoglobin concentration (11). The fluctuation of the background is associated with changes of RBC number in the irradiated volume. These fluctuations increase as vascular diameter decreases. Ultimately, in a capillary with single-file cell flow, PT/PA signals appear above the noise when individual RBCs sequentially pass through the irradiated volume (Fig. 1B). In arterioles and venules in which several RBCs appear simultaneously in the irradiated volume, the background level increases, but its fluctuation decreases. In large vessels with many RBCs (hundreds and thousands) in the irradiated volume, the background continues to rise and fluctuations are minimized and are ultimately determined by laser energy instability. The dilution of blood with a physiological solution leads to a decrease in the background level that can be used to calculate Ht (44).
Monitoring of RBC aggregation and changes of RBC volume
When RBC aggregates pass through the irradiated volume, they should lead to PA signal fluctuations or transient strong PA peaks from many small aggregates or the single large aggregate, respectively. RBC aggregates can be detected because they have a higher absorption than the normal RBC background. For example, five RBC aggregates have a volume of ∼450 fL (corpuscular volume of a single RBC is 90 fL (45) with a total Hb amount of 150 pg (∼30 pg of Hb per one RBC), while the same volume of whole blood with a Ht of 45% contains just two RBCs with 60 pg of Hb. Thus, the expected ratio of PT to PA signals from even small RBC aggregates to blood background is 2.5, which is quite enough for the detection of aggregates. The increase of local absorption in two- and three-dimensional aggregates lead to a significant increase of PT/PA signal fluctuations as a marker of abnormal RBC aggregation, compared with linear RBC rouleaux.
The change of a single RBC's volume leads to a change of the PT/PA signal amplitude, which is strongly dependent on RBC size (as (1/R3) (46). Therefore, even a small geometric RBC change can lead to an enhanced signal change. In particular, hypo-osmolarity (<300 m Osmol/kg) causes of RBC swelling, while hyperosmolarity (>300 m Osmol/kg) leads to RBC shrinking. These RBC-size variations can be detected through the cell size-dependent behavior of PA signal amplitude (larger size, smaller amplitude) at the same concentration of Hb in RBCs.
Sickle RBCs can be identified among normal RBCs on the basis of their specific shape (sickle vs. biconcave), as well as spatial distribution, chemical structure and the optical properties of Hb (HbS vs. HbA). In particular, the identification of sickle RBCs can be based on differences in PT/PA signal amplitudes, even without monitoring of cell shape.
It is well recognized that blood viscosity is determined mainly by Ht, RBC aggregation, and shear-induced deformation of RBCs, and much less by plasma viscosity (1). In turn, RBC deformability depends on their intrinsic factors, including cell geometry, membrane flexibly, and internal viscosity. The capability of PT/PA methods to measure both RBC and plasma parameters is the key point to estimate blood viscosity.
White blood cell (WBC) rheology
WBCs, whose normal concentration in blood is relatively low, may increase blood viscosity only in rare cases of hyperleukocytic leukemia (1). However, we believe that WBCs' role in blood rheology needs to be reevaluated. While WBCs have a slightly larger diameter than RBCs (8 μm vs. 5 μm), they are spheres, and, therefore, they have twice the increased volume compared to RBCs. Furthermore flexible RBCs represent simple membrane capsules enclosing Hb, whereas WBCs are much more rigid cells, owing to their rigid nuclei, granular cytoplasm, and active contractile elements (1, 52). The capacity of PT and PA techniques to detect WBCs in circulation (47) can shed light on this important, still unstudied question. Detection can be based on significant differences in the absorption spectra of WBCs and RBCs owing to the presence of cytochromes and Hb as intrinsic cell-specific markers, respectively (22), as well as to the strong dependence of PT/PA signal amplitude on cell size.
Even these simple phenomenological models show the high potential of PT and PA methods to assess rheological parameters. The diagnostic principles are based on the high sensitivity of time-resolved PT/PA parameters to dynamic changes of Hb properties (e.g., clustering, concentration, spatial distribution, chemical composition) that are specifically associated with alterations of the main hemorheological determinants. For example, fluctuations of PA signal amplitude can be related to pathological aggregation of RBCs, while decreasing PA signal amplitude can be associated with a low Ht. The rheological parameters (e.g., aggregation or viscosity) in the progression of some diseases can change significantly quickly (a few hours, if not minutes) (1). Thus, real-time in vivo multiparameter monitoring of hermorheological parameters may be important for patient survival.
PT tests of Hemorheological Parameters Ex Vivo
The phenomenological models were first verified in ex vivo studies. We explored whether a new technical platform was capable of differentiating single RBCs, RBC aggregates, and Hb in solution (Fig. 2A). High-resolution TDM imaging was used to control cell and laser-beam position. When both pump and probe beams at a diameter of 10–20 μm covered an entire normal or abnormal single RBC or small RBC aggregates (10–15 μm), PT signals from single RBCs and RBC aggregates (Fig. 2A, right, top and middle) exhibited standard signal shapes for all samples: a fast-rising peak and a slower tail. However, the PT signals had a different amplitude and tail duration, depending on individual RBC sizes or number of RBCs in aggregates. TDM images revealed a relatively homogenous size and shape distribution for normal RBCs that correlated with data in the literature (36, 48). PT signal amplitude from individual RBCs in PBS also exhibited a low (3–5%) level of heterogeneity at wavelengths from 530 to 580 nm corresponding to strong absorption bands of Hb in the visible-spectral range. The duration of the tail was 4–6 μs, which correlated with our estimation for discoid RBCs (see above). The irradiation of whole blood with a physiological Ht value (45%) in a 120-μm-thick well (i.e., ∼50 RBCs with 1,500 pg of Hb in the irradiated volume) led to a dramatic increase in the PT signal amplitude; however, the tail duration was similar to those for single RBCs (Fig. 2 A, right, top), suggesting that cooling rates were mostly determined by the cooling of individual RBCs with an average distance between them larger than heat diffusion length at a typical Ht of 20–45%. Nevertheless, we cannot exclude the dynamic fluctuation of cooling time in blood flow due to the interaction of thermal fields from individual RBCs at a higher Ht or because of frequent RBC collisions that may create random temporal RBC complexes (rouleaux) having an increased PT amplitude and a longer cooling time.
For comparison, the PT signal amplitude from 1,500 pg of a relatively homogeneous Hb solution in animal plasma was lower than that from 50 RBCs with a similar total Hb concentration (see above), while the PT signal tail was much longer. This can be explained by the larger signal amplitudes from localized zones with high Hb concentrations in RBCs or the influence of nonlinear amplification of PT signals from overheated localized intracellular zones with Hb clusters.
These conclusions are in agreement with our experimental results related to increases in the PT signal amplitudes and tail durations when RBCs form aggregates (Fig. 2A, right, middle). The amplitude of PT signals correlated with the size of an aggregate controlled by TDM (i.e., larger aggregates, higher amplitude and longer tails). Large aggregates (>10 RBCs) produced nonlinear signals at a relatively high laser energy level due to bubble formation around overheated local zones (31, 49). As expected, the signals from RBC aggregates exceeded those from background blood two to fivefold.
We then focused on studying RBC deformability, which can be accompanied by changes in cell shape, size, and membrane rigidity, as well as Hb distribution within a cell. The high-resolution TDM imaging was used to verify the size and shape of RBCs. Distillate water, chlorpromazine, or Diamide induced a well documented transformation of biconcave RBCs: “swollen” spherocytes with a 40–50% increase in diameter (Fig. 2A, bottom), stomatocytes with shape like a uniconcave cup (Fig. 2B, middle), and rigid RBCs (Fig. 2B, right), respectively. All of these low-deformable types of RBCs have increased volume when compared with normal cells. In turn, the concentration of Hb molecules per a unit of RBC volume decreased when accompanied by a decrease in PT signal amplitudes (46). For example, the mean amplitude of the PT signal from a “non-swollen” spherocyte (Fig. 2B, left) and a “swollen” spherocyte (Fig. 2A, left, bottom), whose volume increased approximately two to fourfold, decreased PT signals from RBCs by 50 and 75%, respectively (e.g., Fig. 2A, right, bottom row) compared to normal discoid RBCs (Fig. 2A, right, top row).
Next, we tested PT signals from sickle RBCs containing human HbS. Under TDM control, we found that single sickle RBCs provided PT signals with lower amplitudes (2.5–4-fold) than PT signals from normal RBCs (Figs. 2A and 2C). Furthermore, round and biconcave RBCs with HbS exhibited low-amplitude signals, compared with round and biconcave RBCs with HbA (Fig. 2C, bottom row). The shape of PT signals from RBCs with HbS was also characterized by a longer tail.
To obtain high-resolution images of RBCs, we applied PT confocal scanning microscopy (42) (Fig. 3). High-amplitude PT signals (coded by red in Fig. 3) at 530 nm were associated with Hb because the PT spectrum corresponded in the first approximation to the absorption spectrum of Hb (Fig. 3D). Because Hb fills the entire RBC volume, the PT signals from Hb replicated the shape of the RBCs. In particular, normal biconcave RBCs exhibited low amounts of Hb in their central thin parts with increased concentration in the peripheral thick ring (Fig. 3C, left). The high sensitivity of PT imaging was also able to distinguish the internal spatial heterogeneity of Hb distribution in RBCs. Thus, these data confirm our previous finding (38, 41) of the spatial heterogeneity of PT signals from RBCs as indicators of an uneven spatial distribution of Hb molecules and, probably, Hb clusters (Figs. 3B and 3C). Using the method of time-resolved PT microscopy beyond the diffraction limit based on size-dependent PT signal shape (31), we roughly estimated that the sizes of localized zones with presumably high local Hb concentrations lay in the broad range from 50 nm to 300 nm. These highly localized phenomena were not distinguished on high-resolution optical images of the same cells due to low absorption sensitivity (Fig. 3A, left).
High-sensitivity PT imaging allowed distinguishing sickle RBCs on the basis of both their specific shape and intracellular PT patterns (Fig. 3F). Indeed, sickle RBCs had more profound spatial heterogeneity (Fig. 3F middle and right) than normal RBCs. This may be associated with increased clustering of HbS compared to HbA.
PT Imaging of RBCs in Blood Flow
We applied PT image cytometry/microscopy (19, 38, 40, 43), using an advanced PT multiplex thermal-lens imaging mode (33) to visualize individual RBCs in the microcirculation of thin (∼250 μm) nude-mouse ear (Fig. 4) with the use of a two-beam (pump-probe) technique (30, 31). The experiments were performed on capillaries with diameters of 5–7 μm and on relatively small microvessels (arterioles, venules) with diameters of 8–12 μm and flow velocities of 0.5–2 mm/s, as well as on relatively large microvessels with diameters of 20–50 μm and flow velocities of 3–8 mm/s (Figs. 4B–4D, respectively).
Both pump and probe laser beams had a circular geometry with diameters comparable to vessel diameters (Fig. 4A, right). We obtained high-resolution (300–500 nm) images of individual RBCs in capillaries and microvessels. It is well known that RBCs are highly deformable and are able to move through narrow capillaries, transforming from their basic biconcave-disk shape to a parachute-like shape (36, 39). PT imaging allowed the monitoring of this dynamic shape deformation with high image contrast (Fig. 4B, top).
In both capillaries with single-file flow and small venules accommodating the simultaneous passage of several RBCs, we imaged the parachute-like shape of moving RBCs (Fig. 4C, top). The RBC deformation index, representing the ratio of length to width was found to be around 1.5–2. In large microvessels with fast multi-file flow (e.g., 40–50-μm arterioles), we observed shapes that were less parachute-like and more elongated (stick-like) (Fig. 4D, top). Despite the overlapping of images from individual RBCs, PT image flow cytometry enabled not only distinguishing single RBCs in vivo but also imaging of local structures associated with the spatial distribution of Hb in RBCs. These structures could hardly be seen on TDM because of their low absorption sensitivity. We also again observed tiny, localized absorbing heterogeneities in PT images of RBCs, which are consistent with our previous findings (38) and with in vitro data above. We hypothesize that these structures could be associated with nanoscale Hb clusters with increased local absorption. If we further exclude any possible artifacts and confirm this hypothesis, PT imaging should be powerful tool for in vivo monitoring of intracellular Hb distribution at the single-RBC level under normal and pathological conditions, including the presence of HbS in sickle RBCs.
Label-Free PA Monitoring of RBC Aggregation In Vivo
In vivo real-time PAFC monitoring of mesenteric arteries (41 ± 3 μm in diameter) and veins (45 ± 4 μm) in healthy rats demonstrated relatively small fluctuations of PA signal amplitude levels that are likely related to formation of physiologically reversible RBC rouleaux. Specifically, at a laser pulse energy instability of 1–2 %, these fluctuations ranged from 1 to 5% in the arteries and increased to 7–10% in the veins and to 10–12% in bifurcation zones, where hydrodynamic conditions intensify rouleaux formation.
Intravenous introduction of Dextran500 produced well known pathological irreversible RBC aggregates that led to the appearance of strong fluctuations of PA signal amplitude (Fig. 5A, bottom) that were significantly higher than fluctuations from intact blood before injection (Fig. 5A, top). These fluctuations started 1–2 minutes after the Dextran500 injection. For example, in an ear blood vessel with a mean diameter of 40 ± 4 μm, the fluctuations started 1.8 min after the Dextran500 injection into the tail vein achieves 30–60% compared with average blood background. These data suggest the potential for PAFC to monitor RBC aggregation by analyzing the fluctuations of PA background (and probably signal frequency spectra) from blood in various vessels.6
Real-Time Dynamic Monitoring of Hematocrit and RBC Deformability
The PTFC) was conducted in microvessels of rat mesentery with one-file flow and a mean diameter of 6.5 ± 2 μm. The appearance of a single RBC in the small irradiated volume was accompanied by PT signals of different amplitudes and tails (Fig. 5B, top). These differences were attributed with random changes of RBC shape, distances between individual RBCs, and RBC velocity. The number of PT signals per one minute was used to measure real-time changes of Ht. High-speed optical imaging (1,000–5,000 fps) with the resolution at the one-cell level provided independent measurements of RBC number, dynamic shape, and linear velocity that allowed to verify PT data (Fig. 5B, middle and bottom). For example, during PTFC monitoring of vein with the diameter (D) of 8-μm and RBC velocity (Vrbc) of 1 mm/sec, we detected ∼55 signals per one minute associated with RBC number. This means that RBCs occupied 4,950 fL (4.95 × 10−7 mL) of bulk blood volume passed through the investigated vessel per minute. The bulk volumetric flow rate (Qbulk) of 29.44 × 10−7 mL/min was calculated from Qbulk= Vbulk×A, where as Vbulk = Vrbc/1.6 and A = πD2/4 (50). Thus, for the particular experimental data we found Ht = 16.8%. In addition, based on a Newtonian definition (8Vbulk/D), the wall shear rate was estimated as ∼500 sec−1. Real-time PTFC during one hour demonstrated that the Ht value was varied from 8 to 17%. This estimation was correlated with optical image data, in particular, with maximal and minimal Ht values (Fig. 5B).
Real-Time Dynamic Optical Imaging of RBC Deformability
In another set of experiments, we obtained the images of blood flow in microvessels after RBC deformability was altered with chemicals. High-resolution, high-speed video monitoring showed the presence of low-deformable RBCs after an injection of Diamide or Chlorpromazine (Figs. 5C and 5D, bottom) compared to control measurement before injection (Fig. 5C and 5D, top). This effect was profound in a specific localized area in curved vessels (Fig. 5C), where maximum centrifugal forces were acting on cells, and in bifurcation zones (Fig. 5D) with higher RBC acceleration.
PAFC of Blood Circulation in Mice with Human Sickle Cell Disease
PAFC monitoring of normal mice and genetically modified mice with sickle cell disease revealed a slightly lower (20–30%) PA background from ear microvessels and more profound PA signal amplitude fluctuations in mice with sickle cells. These data are consistent with in vitro results above demonstrating lower PT signal amplitudes from sickle cells. Photodamage thresholds associated with bubble formation were higher for sickle cells, suggesting that laser-induced local temperature in these cells was lower than that in normal RBCs. Indeed, at a high energy level, the threshold for nonlinear PA signal fluctuations was two to fourfold lower than in mice with sickle cells. These data demonstrate that both PT and PA signals from sickle cells have decreased amplitudes that can be used, after additional verification, as a marker of sickle cell disease.
In vivo Noninvasive PA Monitoring of Complex Changes in Blood Rheology
In this application, noninvasive PAFC was tested with the use of mouse ear blood vessels having a mean diameter of 50 ± 7 μm. Dilution of blood by intravenous injection of 400 μL of physiological solution led to a 20% decrease in PA signals at 580 nm, which corresponded to a ∼7% decrease in the systemic Ht. The complex alterations of systemic blood rheology were created by subsequent intravenous injections first of distillate water, which produced both spherocytosis and reduced Ht, and, then of Dextran500, which induced RBC aggregation. PAFC monitoring of mouse ear vessels at 580 nm revealed a 42% decrease in the PA signal amplitude after the first injection due to blood dilution and formation of “swollen” spherocytes, leading to a decrease in both Ht and average intracellular Hb concentration in large, round RBCs, and hence in PA signals. The second injection of Dextran500 with a lower blood background signal resulted in high-amplitude fluctuations of PA signals that reflected the presence of RBC aggregates in the blood circulation.
Here, we describe a platform integrating PT and PA techniques with conventional optical imaging for real-time, studies of blood rheology. This platform, verified in an animal model, represents a further development of our previous findings in which we applied PT/PA techniques to the in vivo investigation of lymph rheology, permitting the assessment of the “lymphocrit,” cell deformability in lymph flow, and lymph viscosity (22, 37). The diagnostic principles are based on the sensitivity of PT/PA signal parameters (e.g., amplitude, temporal profile, and dynamic behavior) to the local concentration and spatial distribution of Hb in RBCs and RBCs in flow, that is related to dynamic changes in the hemorheological properties of blood flow.
This technology can provide dynamic multiparameter monitoring and imaging of RBC aggregation, alterations of RBC deformability, and Ht. We also demonstrated in vivo high-speed (up to 10,000 fps) and high-resolution (up to 300 nm) imaging for the continuous monitoring of transient RBC deformability in fast blood flow. The relationship between cellular functions and mechanical properties suggests that cell deformability can be used as a marker of early or latent stages of different pathological processes. We demonstrated that the integration of PAFC, PTFC and PT imaging provides in vivo the real-time rheological status of circulating blood in microvessels through dynamic measurements of multiple parameters, including RBC aggregation, Ht, wall shear rate, dynamic shape of moving RBCs, RBC deformability, and intracellular distribution of Hb. We also found for the first time that this technology could, noninvasively and without labeling, assess relatively large vessels and detect changes in two important rheological parameters— Ht and RBC aggregation— together. Finally the capability of the PT/PA technical platform was tested ex vivo and in vivo in a mouse model of human sickle cell disease. We believe that in vivo PA blood testing and rapid (510 min) PT image cytometry (38, 40) of RBCs ex vivo using a small drop of blood can add valuable information about diagnosis of sickle disease.
Although we demonstrated proof of this concept using the visible-spectral range, where the absorption of Hb is maximal, our previous finding suggests that individual RBCs can produce detectable PA signals in near-infrared (NIR) range, in which light achieves maximum penetration into biotissues (20,21). The further development of PT/PA imaging by increasing its speed and resolution could provide accurate measurement of blood vessel diameter for calculation of the Ht.
In addition to measuring the main determinants of hemorheology that we demonstrated here, the capabilities of PTFC and PAFC can be significantly extended to measure many more rheological parameters such as blood viscosity as a complex parameter that depends on Ht, RBC deformability, RBC aggregation, and concentrations of plasma proteins with high molecular weights. PT/PA monitoring of plasma proteins can be performed in far NIR range, where they provide readable absorption in the absence of a Hb background (e.g., in capillaries).
Previously, we and others demonstrated that PT and PA techniques can measure blood flow velocity (9, 19, 25, 51), which could be the basis for calculating the velocity profile and shear rate of the vascular wall. Such spatially resolved measurements of wall shear rates have the potential to answer long-standing questions about ways that blood shear forces exerted on endothelial cells contribute to physiological and pathological processes, such as atherogenesis and angiogenesis (44). For example, it is well known that endothelial cells' sensing of the shear stress gradient promotes activation of genes responsible to measure, vasoregulation, and proliferation. Previously, there were no methods to measure the shear stress gradient in vivo. Because we found that PT and PA signals from a Hb solution and RBCs under the same conditions are significantly different, PAFC and PTFC can likely be used to diagnose pathological intravascular hemolysis in vivo.
A novel diagnostic platform using safe laser parameters and a label-free, noninvasive approach can be quickly translated to use in humans. Indeed, compared with other optical modalities, noninvasive PA methods offer higher resolution, sensitivity, and penetration depth (up to 3 cm), and the minimally invasive delivery of laser radiation through tiny fibers (9) could allow the assessment of potentially any site in the human body. The clinical prototype could be in the form of a portable fiber-based device (e.g., see prototype in the supplementary information for Ref. 9) that is placed over different vessels ranging from capillaries in the nail or eye to large vessels in the hand or neck area. Patient management may be improved though ultrasensitive point-of-care monitoring of blood rheology in vivo, instead of existing invasive time-consuming in vitro testing (see Introduction).
The testing that we have described could be useful for the early, sensitive diagnosis of a broad spectrum of diseases accompanied by alterations of multiple rheological parameters in the acute stage of critically ill patients, in cardiovascular dysfunction, cancer, sickle cell disease, infections, and intoxications. Potential applications include use in emergency departments and intensive care units for patients with shock, trauma, anemia, renal insufficiency, neurological deficit, or congestive heart failure.
This testing would also enable physicians to individualize therapy and assess the therapeutic efficacy of existing and novel treatments. The diagnostic approach may integrate new PT/PA rheological testing with techniques that we recently developed including 1) PT/PA Raman flow cytometry (28, 29), which provides additional diagnostic parameters of blood chemistry through differences in PA Raman vibrational contrast from blood components; 2) identification of spectral signatures of the internal chemical structures of different forms of Hb (e.g., sulpha-, met-, CO-, or HbS [in sickle cells]) in the NIR region in the background of reduced and oxygenated Hb (27); 3) high-speed spectral imaging for measurement of oxygenation and Hb at the single-RBC level compared to bulk integrated parameters; 4) PA detection of circulating clots; and 5) monitoring of changes in the volume of circulating blood. Hypothetically, many standard blood parameters that are currently measured by conventional in vitro blood testing can be measured in real-time in vivo more quickly (a few minutes vs. several hours) and with greater sensitivity.
The authors like to thank Evgeny Shashkov, Dmitry Nedosekin, Mustafa Sarimollaoglu, and Scott Fergusson for their help in experiments; the Office of Grants and Scientific Publications at UAMS for editorial assistance with the preparation of this manuscript.