1.

Introduction

Photoimmunotherapy (PIT) is a cell-specific cancer therapy with minimal side effects, which is based on an armed antibody conjugate that induces rapid cellular necrosis after exposure to near-infrared (NIR) light. The conjugate consists of a hydrophilic photosensitizer phthalocyanine dye, IR700, which is covalently bound to a humanized monoclonal antibody. When exposed to NIR light, the conjugate induces highly selective and rapid cancer cell death both in vitro and in vivo.^1,2 It has been hypothesized that the selective destruction of the perivascular layers of cancer cells by PIT leads to expansion of the tumor vessel diameter and results in a 24-fold increase in the accumulation of various nanomaterials in the tumor bed. This effect is most marked for nanosized molecules including nanoscale anti-cancer drugs and thus reflects a relative increase in the permeability of vessels for larger molecules.³ This effect has been termed superenhanced permeability and retention (SUPR) to differentiate it from the more modest-enhanced permeability and retention that is commonly seen in tumors. Histology and fluorescence imaging with fluorescently labeled macromolecules confirms that high concentrations of nonspecific nanomaterials can permeate into the tumor bed after PIT. However, little is known about the specific effects of PIT on vessel size and blood flow changes at a microscopic level. In this study, we used intravital optical coherence tomography (OCT) to monitor the dynamics of the tumor vasculature during PIT.

OCT is an emerging biomedical imaging tool which provides depth-resolved images similar to ultrasound images.⁴ However, OCT sends light waves instead of sound waves into tissues and thus has much better spatial resolution than ultrasound ($\sim 10\mu \mathrm{m}$). OCT can provide high resolution three-dimensional images of tumor microarchitecture, including the vessels and lymphatics.⁵ OCT has been widely used in cancer research to study both cellular and vascular responses to cancer therapy.⁶ For instance, Doppler OCT (DOCT) has been used to monitor vascular changes caused by photodynamic therapy.⁷ However, it remains challenging to obtain in situ information without resorting to window chamber-implanted tumors due to motion artifacts. In this study, we overcome this challenge with real-time motion gating using speckle variance (SV) imaging. Furthermore, we develop a vessel lumen recognition algorithm to automatically quantify the size of the lumen and measure the blood velocity. These methods can be used to monitor acute hemodynamic changes after PIT.

2.

Material and Methods

2.2.

OCT Imaging System

One unique feature of OCT compared with conventional optical imaging modalities is its capability to resolve the information from different depths by using an interferometer. The interference fringe from different depths reveals the scattering intensity from the specific depth and has a characteristic oscillation frequency. By resolving signals with specific frequencies, a depth-resolved OCT image can be obtained. Figure 1 shows the schematic of the swept-source OCT (SS-OCT) setup. The SS-OCT system^8–10 utilizes a wavelength-swept laser as its light source. The spectrum bandwidth of the laser is 100 nm centered at 1325 nm (Thorlabs, Newton, New Jersey, SL1325-P16), which gives a $10\text{-}\mu \mathrm{m}$ axial resolution in tissue. The wavelength-swept frequency is 16 kHz with a 12-mW output power; therefore, for 1000 axial line images, the frame rate is equivalent to $16\text{frames}/\mathrm{s}$ ($8\text{frames}/\mathrm{s}$ if only one direction is used). A Mach-Zehnder interferometer receives 3% of the laser output power and uses it to generate a clock signal with a uniformly spaced optical frequency to trigger the sampling of the OCT signal in the analog-to-digital converter. The circulator sends the photons from port 1 to port 2 and the reflected light from port 2 to port 3. The sample and reference arms of a Michelson interferometer receive equal portions of the remaining 97% of the laser power. The galvanometer scanning mirrors deflect the sample arm light into the sample through an objective lens, which provides a $15\text{-}\mu \mathrm{m}$ lateral resolution. The backscattering light from the tissue is recombined with the reflected light from the reference mirror, and the interference fringe is detected by the dual-balanced detector. The signal will be acquired by the data acquisition card and then be sent to the computer for image construction and display.

Fig. 1

Schematic of the swept-source optical coherence tomography (OCT) system. SS laser: swept source laser. FC: fiber coupler. PC: polarization controller. BD: balanced detector. C: collimator. O: objective lens. GSM: galvo scanning mirrors. M: mirror.

2.3.

Phantom Studies

Figures 2(a) and 2(b) show the results of the phantom study. Two imaging protocols, DOCT¹¹ and SV,¹² are tested with a vessel phantom. The vessel phantom is a capillary tube with a 0.4-mm-inner diameter. Intralipid solution (1% volume concentration) was pushed through the tube at various speeds controlled by an injection pump (Fisher Scientific, Rockingham Country, New Hampshire, Single Syringe Pump). Figure 2(a) shows that SV can accurately map the lumen of vessel, but DOCT reaches a detection limit at low velocities. SV is capable of mapping out the vessel lumen even when the bulk flow speed is zero. On the other hand, Fig. 2(b) shows that DOCT linearly changes with the flow velocity, but SV does not vary with the flow velocity. Therefore, we utilize DOCT to measure the flow velocity and SV to measure vessel diameter. The standard deviation is calculated from 10 different images under a specific flow velocity.

Fig. 2

(a) Cross-sectional Doppler OCT (DOCT) and speckle variance (SV) images of a vessel phantom (0.4-mm-inner diameter) at various flow velocity. (b) Total intensity of DOCT and SV versus flow velocity. (c) Real-time motion detection using SV images ($6\text{frames}/\mathrm{s}$). Frames 259 and 266 show the SV images when the tissue is at rest. Frames 263 and 264 show the SV images when the tissue is moving. By locating the local minimum (indicated by the red triangles), we can isolate images with no motion. (d to g) Automatic region of interest (ROI) selection procedure. (d) The original image with blood vessel and background noise. (e) Binary conversion of (d). (f) An image with all the connected pixels labeled with a specific color. Noise is dramatically reduced when choosing the ROI with the largest area. (g) The final image without the background noise.

3.

Results

Figures 3(a)–3(d) show the diameter and the blood velocity of a PIT-treated tumor blood vessel. OCT provides the structure information. DOCT provides the blood velocity information and SV provides the size of the open lumen. The first 60 s are obtained as a baseline before PIT. At 60 s, NIR irradiation ($100\mathrm{J}/{\mathrm{cm}}^2$) is turned on and the blood velocity quickly drops below the detection limit. The open space in the vessel lumen also decreases with time, but remains open after 600 s of light irradiation. After 600 s, the NIR light is turned off and post-treatment data is obtained for another 60 s. The vessel size and the velocity remain at a relatively constant level after irradiation. At the beginning of irradiation, the DOCT image [Fig. 3(a)] demonstrates that blood velocity is high and the SV image shows a large lumen. After 83 s of irradiation, the DOCT image [Fig. 3(b)] shows that the velocity has significantly decreased, but the size of the lumen remains almost unchanged. The blood velocity drops below the detection limit after 372 s of irradiation, but the lumen remains open [Fig. 3(c)].

Fig. 3

Vascular changes in a photoimmunotherapy (PIT)-treated and a control tumor blood vessel. (a to c) Representative OCT/DOCT/SV images at different time points (indicated by the orange dashed lines) during the PIT treatment. (d) Time plot of the blood velocity (red line) and the vessel width (blue line). In the first 60 s (indicated by the black dashed line), the laser light is off to obtain baseline information. After 60 s, near-infrared (NIR) laser light is turned on for 10 min (60 to 660 s). The dosage is $100\mathrm{J}/{\mathrm{cm}}^2$. In the last 60 s (660 to 720 s), the light is turned off to obtain post-treatment information (indicated by the second black dashed line on the right). (e to g) Representative OCT/DOCT/SV images at different time points (indicated by the orange dashed lines) of a control tumor blood vessel (with NIR light, but no PIT drug). (h) Time plot of the blood velocity (red line) and the blood vessel width (blue line) of the control vessel. The dosage is $100\mathrm{J}/{\mathrm{cm}}^2$. (i) Ratio of vessel size and blood velocity in PIT-treated and control tumors (control is NIR irradiation but no PIT drug). The vessel size of PIT-treated vessels ($n=4$, from 3 mice) is slightly smaller than its size in the first minute, but it is not statistically significant ($P>0.05$) compared to the control ($n=5$, 3 mice). On the other hand, the blood velocity of PIT-treated blood vessels drops significantly in the end of treatment compared to the beginning and it is statistically different from the control ($P<0.05$).

Control experiments were performed on mice that did not receive the antibody-photosensitizer conjugate but underwent light exposure. Figures 3(e)–3(h) show that blood velocity and lumen size during the 600 s of NIR irradiation fluctuate moderately, which is probably due to physiologic variation and motion, but there is no sustained blood velocity drop in the controls. Indeed, the images show that the blood velocity and the lumen size remain essentially the same throughout the measurement period.

To quantify the effect of PIT, the ratio of the blood velocity and lumen size at the beginning and end of PIT (comparing the first and last minute during irradiation) was determined [Fig. 3(i)]. The PIT-treated group consisted of 4 tumor vessels from 3 different animals. Control experiments (with irradiation but no drug) were performed on 5 tumor vessels from 3 different animals. The lumen size of PIT-treated tumor vessels decreased with time (the ratio is smaller than 1), but it is not statistically different from the control vessels (T test, $P>0.05$). On the other hand, the drop in blood velocity in PIT-treated blood vessels was statistically different compared with the control group ($P<0.05$). We conclude that while there are small nonsignificant changes in vessel lumen diameter, there is a measurable acute decrease (more than one order of magnitude) in blood velocity in the PIT-treated vessels.

Figures 4(a) and 4(b) demonstrate the histology of the peripheral tumor blood vessels. With H&E staining, the red blood cells do not display a nucleus and can be found in the well-preserved blood vessels (indicated by the red arrows). The vessels in both the control and PIT-treated tumors have well-defined boundaries and are easily identifiable indicating that the vessels are not destroyed by PIT. This data is consistent with the in vivo result, which suggests blood vessels are not damaged. On the other hand, PECAM-1 staining [Figs. 4(c) and 4(d)], which labels the endothelial cells of blood vessels (brown color), demonstrates that the PIT-treated capillaries in the deep tumor bed are markedly dilated compared with the control vessels.

Fig. 4

(a) Histology (H&E staining) of control peripheral tumor vessel. Intact blood vessels are indicated by the red arrows. (b) Histology (H&E staining) of PIT-treated peripheral tumor vessel. Intact blood vessels are indicated by the red arrows. (c) Histology (PECAM-1 staining) of control central tumor vessel. The inset shows a zoom-in image. (d) Histology (PECAM-1 staining) of PIT-treated central tumor vessel. The inset shows a zoom-in image. The scale bar represents $100\mu \mathrm{m}$.

4.

Discussion

There are several advantages of using OCT in this setting. It allows us to monitor tumor vascular responses without the injection of a contrast agent or the implementation of an artificial tumor window chamber. Motion gating with SV successfully overcomes the challenge of bulk motion artifact and permits in situ imaging without interfering with the tumor. This method requires a fast imaging speed to obtain images not degraded by bulk motion from a breath cycle. Since the normal mouse respiratory rate is around 3 breaths per second, according to the Nyquist sampling theory, imaging speeds $>6\text{frames}/\mathrm{s}$ are necessary to record the entire cycle. A high imaging frame rate limits the number of A-lines per image (more A-lines per image, slower frame rate) and thus limits the field-of-view (FOV). We used 1000 A-lines per image within a $715\text{-}\mu \mathrm{m}$ FOV to achieve a reasonable DOCT sensitivity. The FOV could be increased by two orders to 71.5 mm while maintaining the same frame rate by using a laser source with a two orders faster wavelength sweep rate.¹³ In addition, high speed lasers can potentially enable four-dimensional tissue images. In addition to FOV, the spatial resolution could be further improved. Micro-OCT¹⁴ with a resolution of $1\mu \mathrm{m}$ in three dimensions will enable the detection of small vessel dilation in vivo.

A potential alternative will be the use of fluorescent proteins, which are excellent endogenous fluorescence emitters to be used for depicting various biological processes and cells both in vitro and in vivo. By labeling circulating cells with fluorescent proteins, blood flow in blood vessels can be clearly monitored.^15,16 Moreover, by inserting a needle-type objective into tissue, the tumor-host interaction and their responses to the drugs can be observed in situ.¹⁷ These techniques are powerful tools for the development of anti-tumor drugs, but the use of fluorescence proteins requires virus-mediated in vivo gene transfection, which is unlikely to be permitted in humans, at least in the near future. For the practical medical application, OCT does not require any contrast agent and has been FDA approved to image retina blood flow in ophthalmic patients.¹⁸ Moreover, for either fluorescence imaging or OCT imaging, bulk motion can blur and degrade the image. In this study, we develop an algorithm that can reduce the motion artifact and permit us to image a moving tissue (see Sec. 2.4).

Unlike most other ablative therapies, both in vivo SV imaging [Figs. 3(a)–3(c)] and histology images [Fig. 4(b)] verify that the peripheral tumor blood vessels remain intact during PIT treatment. Furthermore, SV confirms that the vessels are filled with free moving blood cells and are not thrombosed. An obstructed vessel would result in the absence of free moving scatters and the vessel would lose its variance contrast relative to static tissue, none of which was observed. The vessels are open and filled with free moving blood cells. However, the central and peripheral vessels respond very differently. Due to the limitation of spatial resolution and penetration depth, only big ($>100\mu \mathrm{m}$) peripheral tumor vessels are observed in vivo with OCT. These big peripheral vessels do not dilate [Fig. 3(i)], while the deep capillaries are markedly dilated [Fig. 4(d)]. This heterogeneous response may be due to the difference in vessel structure. The thick peripheral vessels typically have full structures including muscular bundles in the wall, which make them less likely to change during PIT.

Figures 3(a)–3(d) demonstrate that the blood velocity in peripheral tumor blood vessels drops significantly during PIT treatment, while the lumens of the vessels remain open. Blood velocity is maintained in control animals who received the conjugate but no light irradiation [Fig. 3(d), 0 to 60 s] and the controls who received light irradiation but no conjugate [Fig. 3(h)]. This blood velocity drop may be due to the increase of blood pool volume as a consequence of dilated central vessels. A much larger central blood pool volume as well as possible decreased intravascular pressure in the peripheral vessels could slow the blood to the periphery that could induce enhanced nanodrug delivery and reasonably explain the hemodynamic basis of the PIT-induced SUPR effect.

In conclusion, PIT treatment acutely reduces blood speed while the blood vessel is undamaged or thrombosed. Slow blood flow speed implies a long drug circulation time. On the other hand, dilated central vessels may lead to slow periphery flow and an increase in drug pool volume in tumor vessels.

In our next project, we will use multiphoton microscopy to observe the diffusion of the antibody-IR700 conjugate and estimate cancer cell viability in situ. This study will help elucidate how PIT destroys the perivascular barrier and enhances the delivery of anticancer drugs.