Optical coherence tomography (OCT), a three-dimensional imaging technology, was used to monitor and characterize the growth kinetics of multicellular tumor spheroids. Precise volumetric quantification of tumor spheroids using a voxel counting approach, and label-free dead tissue detection in the spheroids based on intrinsic optical attenuation contrast, were demonstrated.
Tumor spheroids have been developed as a three-dimensional (3D) cell culture model in cancer research and anti-cancer drug discovery. However, currently, high-throughput imaging modalities utilizing bright field or fluorescence detection, are unable to resolve the overall 3D structure of the tumor spheroid due to limited light penetration, diffusion of fluorescent dyes and depth-resolvability. Recently, our lab demonstrated the use of optical coherence tomography (OCT), a label-free and non-destructive 3D imaging modality, to perform longitudinal characterization of multicellular tumor spheroids in a 96-well plate. OCT was capable of obtaining 3D morphological and physiological information of tumor spheroids growing up to about 600 µm in height. In this article, we demonstrate a high-throughput OCT (HT-OCT) imaging system that scans the whole multi-well plate and obtains 3D OCT data of tumor spheroids automatically. We describe the details of the HT-OCT system and construction guidelines in the protocol. From the 3D OCT data, one can visualize the overall structure of the spheroid with 3D rendered and orthogonal slices, characterize the longitudinal growth curve of the tumor spheroid based on the morphological information of size and volume, and monitor the growth of the dead-cell regions in the tumor spheroid based on optical intrinsic attenuation contrast. We show that HT-OCT can be used as a high-throughput imaging modality for drug screening as well as characterizing biofabricated samples.
Cancer is the second leading cause of death in the world1. Developing drugs targeting cancer is of crucial importance for patients. However, it is estimated that more than 90% of new anti-cancer drugs fail in the development phase because of a lack of efficacy and unexpected toxicity in clinical trials2. Part of the reason can be attributed to the use of simple two-dimensional (2D) cell culture models for compound screening, which provide results with limited predictive values of compound efficacy and toxicity for the following stages of drug discovery2,3,4. Recently, three-dimensional (3D) tumor spheroid models have been developed to provide clinically relevant physiological and pharmacological data for anti-cancer drug discovery3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25. Since these spheroids can mimic tissue-specific properties of tumors in vivo, such as nutrient and oxygen gradient, hypoxic core as well as drug resistance19, the use of these models can potentially shorten drug discovery timelines, reduce costs of investment, and bring new medicines to patients more effectively. One critical approach to evaluating compound efficacy in 3D tumor spheroid development is to monitor the spheroid growth and recurrence under treatments9,26. To do this, quantitative characterizations of the tumor morphology, involving its diameter and volume, with high-resolution imaging modalities, are imperative.
Conventional imaging modalities, such as bright-field, phase contrast7,9,22,24, and fluorescence microscopy8,9,16,18,22 can provide a measurement of the spheroid’s diameter but cannot resolve the overall structure of the spheroid in 3D space. Many factors contribute to these limitations, including penetration of the probing light in the spheroid; diffusion of the fluorescent dyes into the spheroid; emitting fluorescent signals from excited fluorescent dyes inside or on the opposite surface of the spheroid due to strong absorption and scattering; and depth-resolvability of these imaging modalities. This often leads to an inaccurate volume measurement. Development of the necrotic core in spheroids mimics necrosis in in vivo tumors6,10,15,19,25. This pathological feature is unlikely reproduced in 2D cell cultures19,25,27,28. With a spheroid size larger than 500 µm in diameter, a three-layer concentric structure, including an outer layer of proliferating cells, a middle layer of quiescent cells, and a necrotic core, can be observed in the spheroid6,10,15,19,25, due to lack of oxygen and nutrients. Live and dead cell fluorescence imaging is the standard approach to label the boundary of the necrotic core. However, again, penetrations of both these fluorescent dyes and visible light hinder the potential to probe into the necrotic core to monitor its development in its actual shape.
An alternative 3D imaging modality, optical coherence tomography (OCT) is introduced to characterize the tumor spheroids. OCT is a biomedical imaging technique that is capable of acquiring label-free, non-destructive 3D data from up to 1-2 mm depths in biological tissues29,30,31,32,33,34. OCT employs low-coherence interferometry to detect back-scattered signals from different depths of the sample and provides reconstructed depth-resolved images at micron-level spatial resolutions in both lateral and vertical directions. OCT has been widely adopted in ophthalmology35,36,37 and angiography38,39. Previous studies have used OCT to observe the morphology of in vitro tumor spheroids in basement membrane matrix (e.g., Matrigel) and evaluate their responses to photodynamic therapy40,41. Recently, our group established a high-throughput OCT imaging platform to systematically monitor and quantify the growth kinetics of 3D tumor spheroids in multi-well plates42. Precise volumetric quantification of 3D tumor spheroids using a voxel counting approach and label-free necrotic tissue detection in the spheroids based on intrinsic optical attenuation contrast were demonstrated. This paper describes the details of how the OCT imaging platform was constructed and employed to obtain high-resolution 3D images of tumor spheroids. The step-by-step quantitative analyses of the growth kinetics of 3D tumor spheroids, including accurate measurements of spheroid diameter and volumes, is described. Also, the method of the non-destructive detection of necrotic tissue regions using OCT, based on the intrinsic optical attenuation contrast is presented.
1. Preparation of Cells
2. High-throughput OCT Imaging Platform
NOTE: See referenced work29,30,31,32,33,34 for a thorough review of principles and applications of OCT. See Figure 1 and Huang et al.42 for details of the custom OCT imaging system used in this study.
3. OCT Scanning and Processing of Tumor Spheroids
4. Morphological Quantification of 3D Tumor Spheroids
NOTE: A custom written code in MATLAB processes this quantification. Click the Run button to initiate the process. See Figure 2B for the flowchart of the steps of morphological quantification of spheroids.
5. Dead-Cell Region Detection of 3D Tumor Spheroids
NOTE: In a homogeneous medium, OCT back-scattered intensity detected as a function of depth (I(z)) can be described by the Beer-Lambert Law49: , where z represents the depth, μ is the optical attenuation coefficient, and I0 is the incident intensity to the sample. Hence the derived optical attenuation coefficient can be expressed as: . Since OCT images are often plotted on a logarithmic scale, the slope of the OCT intensity profile can be retrieved to derive the optical attenuation coefficient. See Figure 2C for a flowchart of the generation of the optical attenuation maps.
6. Histology and Immunohistochemistry
NOTE: Histology and immunohistochemistry (IHC) stained images of tumor spheroids are obtained to correlate with the corresponding OCT results.
High Throughput Optical Coherence Tomography Imaging of Spheroids in a 96-well Plate
Figure 3 exhibits the result of HT-OCT scanning of a 96-well plate with HCT 116 tumor spheroids on Day 3. The sequential scan of the whole plate starts from the bottom-right well (H12). Figure 3B shows the flow chart of the software implementation of the HT-OCT system. After one spheroid data were collected and processed, the plate would move to next well, wait for ~2 s to allow the spheroid to rest, and collect the next spheroid data. Each OCT data consist of 400 x 400 x 1024 voxels, which corresponded to an actual volume of 1.0 x 0.84 x 2.3 mm3. Figure 3C shows a collage of en-face OCT images of HCT 116 spheroids generated from the processed data. The result is comparable with images from other 2D high-throughput imaging system22. Given the 3D imaging capability of the OCT, we could also generate the collage of 2D cross-sectional spheroid images from 96 wells (Figure 3D) to monitor spheroid heights and visualize spheroid inhomogeneity in the vertical direction. A collage of 3D-rendered spheroid images is also feasible from any predefined angle (Figure 3E) to visualize the overall 3D shape and evaluate the roundness of the spheroid. Note that the overall OCT imaging and process time for the whole 96-well plate would be ~21 min and ~25 min when the line-scan camera is running at a speed of 92 kHz and 47 kHz, respectively. See Video 1 for an example.
Longitudinal Morphological and Physiological Monitoring of the Tumor Spheroid
After we obtained the OCT structural images of tumor spheroids from the plate for multiple time points, we could further analyze these data by quantifying the morphological and physiological information of the tumor spheroids. Figure 4 shows the different approaches to characterize tumor spheroids and obtain longitudinal morphological and physiological information from them.
Figure 4B shows different ways to visualize the tumor spheroid. With the aid of either commercial or free software, we could load the 3D data into the software and create a "volume" of the tumor spheroid (3D rendering), which shows the overall structure of tumor spheroid in 3D space. With proper thresholding, we could generate a surface plot of tumor spheroid (Figure 4B), which could be used to segment the spheroid and measure the volume. We could also generate the orthogonal slides (ortho slides) from different cross-section planes in different orientations (Figure 4B, XZ, YZ, and XY) and measure the diameter and height of the tumor spheroid from these ortho slides.
Gathering the OCT data of the same spheroid from multiple time point, we could quantify the morphological information and generate the growth curve of the spheroid to show its longitudinal changes. Figure 4C shows representative data of an HCT 116 tumor spheroid being monitored for 21 days. From the segmented data and ortho slides, we measured the diameter, height and voxel-based volume of the spheroid for all the time point, which were listed in the table. We also calculated the diameter-based volume for a comparison. The growth curves in size and volume were plotted, respectively. From the growth curves, we could see that this HCT 116 tumor spheroid followed a linear growth pattern in volume before day 11. Before this time point, the spheroid kept growing and maintained a relatively uniform shape. However, after day 11, the spheroid became disrupted, flattened and fully collapsed on day 21. The growth curve of voxel-based volumes clearly shows the trend, with a gradually decreased volumes after day 11.
Based on the OCT data, we can also obtain the physiological information of the distribution of dead-cells within the tumor spheroids by analyzing the pixel-by-pixel optical attenuation from 2D cross-sectional images. Following the methods illustrated in Figure 2 and Protocol 5, we could quantitatively determine the dead-cell regions and monitor the growth of these regions as a function of time. Figure 4D shows a representative result of longitudinal tracking of the increase of dead-cell areas in the tumor spheroid. The areas highlighted in red, which had high optical attenuation, show the labeled necrotic areas. From the 3D rendered optical attenuation map during the 14 day development, we could see the red sector expanding, indicating the increase of the necrotic regions. As the percentage of the necrotic areas increased, the tumor spheroid could not maintain its perfect shape. Therefore, they would tend to disrupt and collapse, which were seen in the longitudinal monitoring of tumor morphology in Figure 4C.
The proposed nondestructive dead tissue region detection technique was verified by comparing OCT optical attenuation map of HCT 116 tumor spheroid with corresponding images obtained by histology and IHC. Figure 4D presents such a comparison with a Day 14 HCT 116 spheroid. A good match between the OCT attenuation map and corresponding H&E and TUNEL slices were found, which was indicated by analyzing the features within the regions in H&E and TUNEL slices marked by dash lines derived from the contour of OCT high attenuation regions. In H&E slices, the dead tissue regions were indicated by less dense and aggregated structure located within the dashed line region. In TUNEL slices, a good match was observed between high attenuation region and TUNEL-labeled apoptotic cellular region.
Figure 1: Construction of a high-throughput optical coherence tomography (HT-OCT) system for tumor spheroid imaging. (A) Schematics of the HT-OCT system. A diagram of the 96-well plate is plotted next to the OCT system. Five wells (D2, D11, B6, D6, G6) labeled in yellow are used for the fine adjustment of the stages in (D). (B) The actual configuration of HT-OCT system. See Table of Materials for optical components used for each part of the system. (C) Spectrometer design for the HT-OCT system. (D) Stage setup for the HT-OCT system. Proper alignment of the 6-axis stage and synchronization between the OCT acquisition and the stage movement are required for high-throughput imaging. (E) and (F) show the effects of rotation and tilting on final image of different wells. Rotation causes the OCT images of different wells to shift horizontally while tilting will lead to vertical shifting of different wells. Please click here to view a larger version of this figure.
Figure 2: Data Processing for OCT images of tumor spheroids. (A) Flowchart of general post-processing steps for OCT data. (B) Flowchart of morphological quantification of the tumor spheroid. (C) Flowchart of dead cell region detection of the tumor spheroid. Scale bar: 100 µm for all the subfigures. Please click here to view a larger version of this figure.
Figure 3: High-throughput OCT scanning of a 96 well plate containing U-87 MG tumor spheroids. (A) The actual setup with the 96-well plate under the objective. (B) Flow Chart of the software implementation of HT-OCT system. Collages of 96 en face (C), cross-sectional (D) and 3D rendered maximum intensity projection (MIP) (E) OCT images of Day 3 HCT 116 spheroids were generated from the processed data. Scale Bar: 200 µm for all the subfigures. Please click here to view a larger version of this figure.
Figure 4: Longitudinal Morphological and Physiological Quantification of Tumor Spheroids with 3D OCT data. (A) Obtained 3D OCT structural images of a tumor spheroid after general OCT post-processing. From the OCT data, we can generate a 3D surface plot and XZ, YZ and XY orthogonal slices to visualize the structure of the tumor spheroid in any direction (B). We can perform longitudinal monitoring of a single tumor spheroid (C), characterizing its diameter, height and voxel-based volume (listed in the Table of Materials) and plotting the growth curves in size and volume during the 21-day development. In the example, as the spheroid developed, it became disrupted on day 11 and fully collapsed on day 21. We can further monitor the physiological status of a tumor spheroid longitudinally based on the optical intrinsic attenuation contrast (D). 3D rendered images of a tumor spheroid showed the appearance and growth of dead-cell regions from day 7 to day 14. The high-attenuation-labeled dead-cell areas in red were matched with histological and immunohistochemical (IHC) results. OCT attenuation map, H&E, and TUNEL result in Figure 4D are modified from Ref. 42. Scale bars: 100 µm for all the subfigures. Please click here to view a larger version of this figure.
Video 1: High-throughput OCT imaging of tumor spheroids. A workflow of 3D OCT imaging, basic OCT processing and stage movement was presented in the video with a 5x speed. Previews of processed OCT structural images of spheroids were also presented. Please click here to view this video. (Right-click to download.)
Tumor activity is highly relevant to its morphological structure. Similar to monitoring characteristic growth curve for 2D cell cultures, tracking the growth curve for 3D tumor spheroids is also a conventional approach to characterize the long-term spheroid growth behavior for different cell lines. Notably, we can characterize the drug response by analyzing tumor degradation or tumor regrowth directly reflected in the growth curve. Therefore, quantitative assessment of 3D tumor spheroids, including the size and volume, to derive the growth curve, is of great importance for the characterization of tumor spheroids and the evaluation of compound effect. Currently, imaging platforms based on bright field, phase contrast or fluorescent imaging have been established for routine imaging and analysis of morphology or functions of the 3D tumor spheroids8,9,18,22. However, they are unable to resolve the entire, large tumor structure due to limited depth penetration as well as low-resolution depth-resolvability. In the representative results, we have demonstrated OCT to visualize the entire 3D structure of the tumor spheroid developing over time. 3D OCT imaging could provide the view of the spheroid in any orientation and any cross-section with high-resolvability, which was not available in conventional imaging modalities that lack the resolution along the depth. Furthermore, voxel-based volume quantification based on 3D OCT data yielded an accurate quantification of spheroid volumes without assuming their original shapes. Therefore, we have demonstrated that OCT is a robust imaging modality for 3D morphology characterization of tumor spheroids, which ensures accurate measurements of characteristic growth patterns for different cell lines and potentially can serve as an alternative for drug response evaluation.
Viability tests using fluorescent staining remain a popular approach for functional analyses of tumor spheroids, especially for drug screening18. However, the disruptive nature of fluorescent dyes indicates that these tests are only suitable for end-point studies. In our representative results (Figure 4D), we demonstrated an alternative method that can characterize cell viability within the entire spheroid. Our results have shown that OCT could distinguish the dead-cell region from the viable region in the spheroid based on intrinsic optical attenuation contrast. In addition, with 3D imaging capability and non-destructive nature of the OCT system, quantitative evaluation of the dead-cell distributions and in situ monitoring of the progression of dead-cell regions within the spheroid are feasible, which potentially provide more valuable information of the spheroid growth pattern. However, we should note that, in our representative results, we are not able to differentiate different types of cell death modes, such as apoptosis and necrosis, in the binary OCT attenuation map.
Since a drug compound library can be extensive (>10,000), a high-throughput and robust system to characterize tumor spheroids in multi-well plates during drug screening is imperative. The current high-throughput imaging system can achieve a screening of the whole 96-well plate in <5 min in 2D scan mode22. OCT can be adapted for high-throughput screening purpose, with the aid of a motorized stage. One can also obtain a commercially available OCT system (See Table of Materials for a list of commercial OCT systems) with a similar performance to our custom OCT system, and incorporate a motorized stage into the system. However, efforts must be taken to modify the commercial OCT system to integrate the motorized stage. Also, custom software implementation to realize the synchronization between the OCT acquisition trigger and stage movement trigger is required. For our prototype HT-OCT system, it took 2-18 seconds to acquire one 3D OCT data from a single tumor spheroid, depending on the choice of camera speed. Thus, the total acquisition time can be as short as ~3.2 min for a 96-well plate using HT-OCT system. However, the intermediate steps for the current HT-OCT system, including data processing, reading and writing data on hard drives, and stage movements, remained time consuming. Additional ~18 min would be needed on top of the ~3.2 min minimum data acquisition time. The total imaging time can be further reduced in several aspects: use state-of-the-art OCT systems equipped with a high-speed tunable laser source50,51; optimized the workflow by arranging critical steps (data acquisition, data processing, writing, stage movement) working in parallel; employ a parallel OCT imaging with a space-division multiplexing setup52. With system optimization, the high-throughput OCT system can be utilized in cancer drug discovery as well as characterization of other 3D bio-fabricated samples (e.g., 3D tissue organoids) for various biomedical applications.
The authors have nothing to disclose.
This work was supported by NSF grants IDBR (DBI-1455613), PFI:AIR-TT (IIP-1640707), NIH grants R21EY026380, R15EB019704 and R01EB025209, and Lehigh University startup fund.
Custom Spectral Domain OCT imaging system | Developed in our lab | ||
Superluminescent Diode (SLD) | Thorlabs | SLD1325 | light source |
2×2 single mode fused fiber coupler, 50:50 splitting ratio | AC Photonics | WP13500202B201 | |
Reference Arm | |||
Lens Tube | Thorlabs | ||
Adapter | Thorlabs | ||
Collimating Lens | Thorlabs | AC080-020-C | |
Focusing Lens | Thorlabs | ||
Kinematic Mirror Mount | Thorlabs | ||
Mirror | Thorlabs | ||
1D Translational Stage | Thorlabs | ||
Continuous neutral density filter | Thorlabs | ||
Pedestrial Post | Thorlabs | ||
Clamping Fork | Thorlabs | ||
Sample Arm | |||
Lens Tube | Thorlabs | ||
Adapter | Thorlabs | ||
Collimating Lens | Thorlabs | AC080-020-C | |
Galvanometer | Thorlabs | ||
Relay Lens | Thorlabs | AC254-100-C | two Relay lens to make a telescope setup |
Triangle Mirror Mount | Thorlabs | ||
Mirror | Thorlabs | ||
Objective | Mitutoyo | ||
Pedestrial Post | Thorlabs | ||
Clamping Fork | Thorlabs | ||
Polarization Controller | Thorlabs | ||
30mm Cage Mount | Thorlabs | ||
Cage Rod | Thorlabs | ||
Stage | |||
3D motorized translation stage | Beijing Mao Feng Optoelectronics Technology Co., Ltd. | JTH360XY | |
2D Tilting Stage | |||
Rotation Stage | |||
Plate Holder | 3D printed | ||
Spectrometer | |||
Lens Tube | Thorlabs | ||
Adapter | Thorlabs | ||
Collimating Lens | Thorlabs | AC080-020-C | |
Grating | Wasatch | G = 1145 lpmm | |
F-theta Lens | Thorlabs | FTH-1064-100 | |
InGaAs Line-scan Camera | Sensor Unlimited | SU1024-LDH2 | |
Name | Company | Catalog Number | Comments |
Cell Culture Component | |||
HCT 116 Cell line | ATCC | CCL-247 | |
Cell Culture Flask | SPL Life Sciences | 70025 | |
Pipette | Fisherbrand | 14388100 | |
Pipette tips | Sorenson Bioscience | 10340 | |
Gibco GlutaMax DMEM | Thermo Fisher Scientific | 10569044 | |
Fetal Bovine Serum, certified, US origin | Thermo Fisher Scientific | 16000044 | |
Antibiotic-Antimycotic (100X) | Thermo Fisher Scientific | 15240062 | |
Corning 96-well Clear Round Bottom Ultra-Low Attachment Microplate | Corning | 7007 | |
Gibco PBS, pH 7.4 | Thermo Fisher Scientific | 10010023 | |
Gibco Trypsin-EDTA (0.5%) | Thermo Fisher Scientific | 15400054 | |
Forma Series II 3110 Water-Jacketed CO2 Incubators | Thermo Fisher Scientific | 3120 | |
Gloves | VWR | 89428-750 | |
Parafilm | Sigma-Aldrich | P7793 | |
Transfer pipets | Globe Scientific | 138080 | |
Centrifuge | Eppendorf | 5702 R | To centrifuge the 15 mL tube |
Centrifuge | NUAIRE | AWEL CF 48-R | To centrifuge the 96-well plate |
Microscope | Olympus | ||
Name | Company | Catalog Number | Comments |
Histology & IHC | |||
Digital slide scanner | Leica | Aperio AT2 | Obtain high-resolution histological images |
Histology Service | Histowiz | Request service for histological and immunohistological staining of tumor spheroid | |
Name | Company | Catalog Number | Comments |
List of Commerical OCTs | |||
SD-OCT system | Thorlabs | Telesto Series | |
SD-OCT system | Wasatch Photonics | WP OCT 1300 nm | |
Name | Company | Catalog Number | Comments |
Software for Data Analyses | |||
Basic Image Analysis | NIH | ImageJ | Fiji also works. |
3D Rendering | Thermo Fisher Scientific | Amira | Commercial software. Option 1 |
3D Rendering | Bitplane | Imaris | Commercial software. Option 2. Used in the protocol |
OCT acquisition software | custom developed in C++. | ||
Stage Control | Beijing Mao Feng Optoelectronics Technology Co., Ltd. | MRC_3 | Incorporated into the custom OCT acquisition code |
OCT processing software | custom developed in C++. Utilize GPU. Incorporated into the custom OCT acquisition code. | ||
Morphological and Physiological Analysis | custom developed in MATLAB |