The spread of cancer from the primary tumor to distant organs is caused by the invasion of cancer cells into the vasculature or lymphatic system. This is followed by eventual deposition and growth in a preferred location in tissue.1 Research has indicated that the concentration of these circulating tumor cells (CTCs) is a significant prognostic factor.23.–4 The biologically relevant concentration of these CTCs (0 to of blood) is much smaller than the number of erythrocytes ( to ) and leukocytes ( to ) in blood, complicating the task of detection.5
Photoacoustic flow cytometry (PAFC) has been used both in vivo and in vitro to detect circulating melanoma cells, which are naturally optically absorbing.6,7 Most other cancers are nonpigmented and inaccessible to PAFC without cellular alteration. Viator et al.8 demonstrated the ability to add gold nanoparticles (AuNPs) to prostate cancer cells through immunochemistry and detect them using the in vitro cytometry method. Magnetic particles have also been used for in vivo photoacoustic (PA) cytometry for both optical contrast and to aggregate CTCs for increased signal-to-noise ratio.9 Additionally, nanoparticle attachment to melanoma cells has been considered in order to increase signal contrast and consistency, since melanin expression is susceptible to biological variability.1011.–12 Finally, Bhattacharyya et al.13 demonstrated attachment of antibody-conjugated AuNPs to breast cancer cells and consequent detection in saline with in vitro PAFC.
Significant research has been performed on in vivo PAFC14 with applications of CTC detection in the mouse model. However, the method of translation to human studies remains unclear due to the larger blood volumes, increased light scattering due to deeper tissue positions of vasculature, and difficulty in validating false positives. On the other hand, CTCs detected with in vitro PAFC can be validated to be cancer cells and captured for further testing. O’Brien et al.15 demonstrated the ability to capture circulating melanoma cells with in vitro two-phase PAFC. This technique of in vitro two-phase PAFC was also used by Gupta et al.16 to detect and capture circulating melanoma cells in an induced mouse melanoma model. In two-phase PAFC, the aqueous cell suspension to be tested is flowed into the flow chamber along with air. When the two immiscible fluids are flowed simultaneously into the same chamber, they form alternating compartments or slugs of each fluid, referred to as the two phases.17
However, oil, in place of air, can also be used for the two-phase PAFC since oil still does not mix with aqueous cell solutions, which are mostly water. In this study, two-phase PAFC with oil and nanoparticle enrichment of cancer cells are combined to create the complete in vitro CTC detection and capture system. Feature extraction from emitted PA signals and classification algorithms are used for the automated detection of cancer cells. Healthy blood is spiked with breast cancer cells, tested with the PAFC, and detected cells are isolated. Finally, a micromanipulator is used to extract single CTCs.
Material and Methods
First, of streptavidin coated 320 nm red flourescent latex nanoparticles (Bangs Laboratories, Fishers, Indiana) were added to of biotin conjugated anti-EpCAM monoclonal antibody (Pierce Antibodies, Rockford, Illinois) in of PBS. This suspension was incubated at 22°C for 1 h. Then, of neutravidin-conjugated 54-nm spherical AuNPs was added to the suspension with an additional of PBS and allowed to incubate for 20 min. Finally, of anti-EpCAM in another of PBS was added to the suspension and allowed to incubate for 30 min. The resulting suspension was centrifuged at for 8 min in order to pellet the nanoparticles and leave the excess antibodies in suspension. The supernatant was removed and the nanoparticle pellet was resuspended in 1 mL of PBS.
Blood Centrifugation Procedure
In a 15-mL conical cuvette, 1 mL of blood drawn from a healthy individual was layered onto 3 mL of Histopaque 1077 (Sigma Aldrich, St. Louis, Missouri), which is widely used to separate leukocytes from erythrocytes, and centrifuged at for 30 min. After centrifugation, the peripheral mononuclear blood cell (PBMC) layer was withdrawn with a transfer pipette and placed in another conical cuvette. The cancer cells in blood are expected to be in the PBMC layer due to their similarity in density with mononuclear leukocytes. This was diluted with 10 mL of PBS and centrifuged at for 15 min. The supernatant was removed and the cell pellet was resuspended in of PBS and of red blood cell lysis solution (G-Biosciences St. Louis, Missouri). This solution was agitated every 2 to 3 min for 10 min to aid in lysing the remaining erythrocytes, which may cause false-positive signals due to hemoglobin pigment. After 10 min, this solution was diluted in 2 mL of PBS and centrifuged at for 10 min. The supernatant was removed and the cell pellet was resuspended in 1 mL of PBS.
Nanoparticle Attachment to Breast Cancer Cells
The following protocols were developed to investigate if the nanoparticles were just attaching to the EpCAM surface receptors exclusively on breast cancer cells or unknown targets, which were not specific to the anti-EpCAM antibody. Additionally, confocal imaging was used to determine if nanoparticles were internalized by the cells.
Specific labeling to EpCAM surface receptor
Cultured T47D breast cancer cells were diluted into a concentration of in 1 mL of PBS and incubated with of biotin conjugated anti-EpCAM monoclonal antibody for 1 h. An excess of anti-EpCAM antibody was used to ensure that all EpCAM receptors on the cells were blocked. Then, the cell suspension was centrifuged at for 10 min in order to remove the excess unbound antibodies. The supernatant was removed and the breast cancer cell pellet was resuspended in 1 mL of PBS and of the fluorescent-AuNP suspension was added to the cell suspension. This solution was allowed to incubate for 1 h and then centrifuged at for 10 min to remove excess nanoparticles. The cell pellet was resuspended in 1 mL of PBS.
The cells were then imaged with a Zeiss LSM 510 META NLO confocal microscope. A tetramethylrhodamine (TRITC) filter and excitation scheme was used to image the flourescent-gold particles attached to the breast cancer cells. A three-dimensional (3-D) z-stack was acquired from multiple cells with step size with variable depths depending on the geometry of the cells.
Affinity to breast cancer cells in the presence of leukocytes
Cultured T47D breast cancer cells were diluted into a concentration of in 1 mL of PBS and incubated with of 4’,6-diamidino-2-phenylindole (DAPI) for 10 min in order to label the cancer cell nuclei for easy identification. Then the cell suspension was centrifuged at for 10 min in order to remove the excess DAPI. The supernatant was removed and the breast cancer cell pellet was resuspended in the resulting leukocyte suspension from Sec. 2.2. After which, of the fluorescent-AuNP suspension was added to the previously mentioned cell suspension. This solution was allowed to incubate for 1 h and then centrifuged at for 10 min to remove excess nanoparticles. The cell pellet was resuspended in 1 mL of PBS.
This suspension was then prepared on a slide and imaged with an Olympus IX-70 microscope with wide-field fluorescent capability and a air objective. For the DAPI nuclear stain and the red-fluorescent AuNPs, the DAPI and TRITC filters were used, respectively. An exposure of 5 ms was used to capture the DAPI fluorescence, while a 100 ms exposure was required to capture the fluorescence from the fluorescent AuNP.
Intracellular distribution of nanoparticle labeling
Cultured T47D breast cancer cells were diluted into a concentration of in 1 mL of PBS and incubated with of the fluorescent-AuNP suspension. This solution was allowed to incubate for 1 h and then centrifuged at for 10 min to remove excess nanoparticles. The cell pellet was resuspended in 1 mL of PBS.
The cells were then imaged with a Zeiss LSM 510 META NLO confocal microscope. A TRITC filter and excitation scheme was used to image the flourescent-gold particles attached to the breast cancer cells. A 3-D z-stack was acquired from multiple cells with step size with variable depths depending on the geometry of the cells.
Photoacoustic Flow Cytometry Detection
The in vitro two-phase PAFC system is composed of two programmable syringe pumps (Braintree Scientific, Braintree, Massachusetts), one with optical grade mineral oil (Fisher Scientific) and the other with the prepared cell sample to be tested. Both fluids were pumped at simultaneously into a 1.5-mm diameter, wall-thickness quartz flow chamber, creating alternating droplets of cell sample and oil, each in volume. A 1-mm optical fiber delivered 5 ns, 532 nm pulsed light transversely to a portion of the chamber (Fig. 1). The laser irradiated of fluid with every pulse and the radiant exposure at the chamber was . An acoustic transducer with a 20-MHz bandwidth (Olympus, Waltham, Massachusetts) was coupled with the irradiated portion of the flow chamber to capture the PA emission. The analog signal was amplified 50 dB (Ritec Inc., Warwick, Rhode Island) and digitized with a LabVIEW integrated PXI card (National Instruments, Austin, Texas). The time resolution was 1 ns and the voltage resolution was .
Breast cancer cells were tagged with nanoparticles using the same parameters as defined in Sec. 2.3 but not in the presence of leukocytes, so as to eliminate any false positives from nonspecific attachment. Then they were diluted into a concentration of in 1 mL of PBS. For this concentration of cells, approximately two cells are expected to be irradiated with each laser pulse, since is the irradiated volume. This suspension was tested with the PAFC under the flow parameters previously described.
Detection from centrifuged blood
About 2 mL of blood from a donor was separated into two vials with 1 mL each. About 35 cancer cells with DAPI labeled nuclei and tagged with nanoparticles were added to one vial. Both vials of blood were centrifuged according to the procedure described in Sec. 2.2. The resulting solutions were tested with the PAFC detection system.
The vial without cancer cells was first run through the system. The PA signals from multiple droplets were used as a training set in order to determine the parameters of classification, following which, the vial with cancer cells was tested under the same parameters. The droplets that tested positive for cancer according to the classification algorithm were captured on a slide. Then a light microscope was used to visualize the cancer cells and withdraw them with a micromanipulator.
Large fluorescent spheres (300 nm) and smaller gold spheres (50 nm) were conjugated with Streptavidin–Biotin binding through anti-EpCAM antibodies. Fluorescent AuNPs were imaged with a scanning electron microscope to ensure that the expected results were achieved with the conjugation chemistry. Figure 2 has images of the fluorescent AuNPs showing that the protocol described in Sec. 2.1 was successful in attaching the gold nanospheres to the larger fluorescent latex particles. Figure 2(b) demonstrates that clusters of particles also formed, which were on the scale.
Nanoparticle Attachment to Cancer Cells
Breast cancer cells with DAPI-labeled nuclei were tagged with fluorescent-gold in the presence of leukocytes. Figure 3(b) demonstrates the high specificity of the fluorescent AuNP to the breast cancer cells, which are much larger than the surrounding leukocytes and platelets. However, there is some nonspecific attachment of the nanoparticles to the leukocytes, as shown by those that have been circled in yellow in Fig. 3(b). Approximately, 5% to 10% of leukocytes are nonspecifically labeled with nanoparticles. Those that are labeled have 6 to 10 times fewer nanoparticles on them compared to breast cancer cells, as can be determined through fluorescent intensity measurements based on pixel values (data not shown).
Nanoparticle attachment was both localized and diffuse in the breast cancer cell, as seen in Fig. 4. The dim red fluorescence from the entire cell is likely due to the single fluorescent-gold complexes, as shown in Fig. 2(a), whereas the brighter spots within the cell are from the clusters of nanoparticle complexes, as shown in Fig. 2(b).
Figure 5 compares labeling of a normal breast cancer cell with that of a cell that has EpCAM receptors blocked before exposure to the anti-EpCAM labeled nanoparticle complex. This verifies that the nanoparticle complex was attaching to the EpCAM receptors on the cancer cell.
Photoacoustic Flow Cytometry Detection
The PA signals captured with each irradiation was data with time and voltage values corresponding to acoustic pressure arriving at the ultrasound transducer at that moment. The chamber lasted in time since it was 1.5 mm in diameter and the speed of sound in water is . Figure 6(a) shows a plot of the PA signal from within the chamber when there is just PBS and Fig. 6(b) shows the PA signal from within the chamber when there are tagged breast cancer cells.
Two separate samples were tested with the PAFC system: PBS and tagged breast cancer cells in PBS at a concentration of . The PA waveform captured for each irradiation was saved and two features were extracted in postprocessing. One feature was the integral of the PA signal from within the flow chamber after every value was squared. This integral was adjusted by subtracting from it the squared-integral of a portion of the voltage trace outside the chamber, which was considered noise.
The second feature was a count within the chamber of the number of voltage values, which reached above a positive noise threshold or below a negative one. The noise thresholds were determined by a portion of the voltage trace outside the chamber as the boundaries of the noise, where denotes standard deviation. The count of values above or below the noise threshold was adjusted by subtracting from it the count from a portion of the voltage trace outside the chamber.
Figure 7 is a plot in feature space of each waveform collected from the two groups. There are 92 points in the control or PBS group, while there are 64 points in the breast cancer cell group. The red boundary is an example of the boundary of the control group assuming the distribution is bivariate normal. There is 1 control point outside the boundary and 43 out of 64 (67%) breast cancer cell points outside the boundary. This threshold can be used as a binary classifier and if the threshold is changed, then the true and false positive rates also change.
The boundaries of the threshold are calculated with Eq. (1). Starting with circle, , centered around the origin with a radius of 2.5 and transforming it into the ellipse, is seen in Fig. 7:
Figure 8 shows the receiver operating characteristic (ROC) curve for the data presented in Fig. 7. The ROC curve shows the relationship of the threshold to the true and false positive rates for the binary classifier of an irradiated volume either having or lacking cancer cells. The ROC curve was generated by changing the threshold from to in increments and the results are plotted from right to left.
Detection from Centrifuged Blood
In the final part of this study, 1 mL of blood was centrifuged to remove erythrocytes and the remaining leukocytes were resuspended in 1 mL of PBS. About of this solution was flowed through the system in order to parametrize the PA response of just leukocytes according to the two features used for classification. These parameters were then used for the automated identification of cancer cells, which produce PA signals with different features, among leukocytes. About 35 cancer cells were added to 1 mL of whole blood, the centrifugation procedure was used to remove erythrocytes, and the remaining leukocytes and cancer cells were resuspended in 1 mL of PBS. The PAFC system was used to detect and capture the added cancer cells.
Figure 9 demonstrates the PA signals seen by the researchers during the flow tests. The trace in red is characteristic of when there are only leukocytes being irradiated and the blue trace shows the presence of a robust PA emitter between 850 and 900 ns. Even with cancer cells added, most PA signals resembled the red trace since the presence of a cancer cell in the chamber was an unlikely occurrence due to the small concentration.
This is especially supported by Fig. 10, which is a feature plot of every PA waveform acquired during the flow test for both samples. The waveforms from the initial sample of just leukocytes are plotted as squares (). The red dot at the center of the ellipse is the mean of that distribution and the ellipse is the boundary. The blue circles () are all of the waveforms acquired during the PAFC test of the second sample, which was spiked with cancer cells. Most waveforms are within the red boundary and classified as not being different from just leukocytes. However, outside of the red boundary, toward the top-right, there are 17 outliers, which were classified by the automated system as containing cancer cells. These droplets were captured and imaged to find isolated cancer cells. The outliers at the bottom-left were not classified as containing cancer cells because their features were smaller than the mean and the voltage traces did not contain PA signals from cancer cells.
Figures 11(a) and 11(b) are light micrographs of 2 out of 15 total captured cancer cells, which were found in 13 of the 17 captured droplets. The remaining four droplets were not found to contain cancer cells. Since the cancer cell nuclei were prelabeled with DAPI before they were added to the blood, researchers looked for cells with fluorescent blue nuclei. Out of the approximate 35 cancer cells added, 15 were successfully retrieved.
Finally, Fig. 12 demonstrates through a photo montage how a breast cancer cell can be captured and transported with a micromanipulator. Breast cancer cells were identified with DAPI fluorescence under microscopy and transported using a glass micropipette to a different dish with cytophilic wells, allowing for further immunohistochemical or genetic analyses.
Nanoparticle Attachment and Specificity
The conjugation chemistry used to create the antibody-bound nanoparticles formed single groups of gold nanospheres attached to one fluorescent sphere, as well as large clusters of gold nanospheres attached to multiple fluorescent spheres, as evident in Fig. 2. These clusters may actually work to the benefit of the researchers as they provide large, concentrated doses of optical contrast to the cell. Moreover, from an acoustic standpoint, the 20-MHz transducer being used is unable to resolve and differentiate between ultrasound signals emitted from these larger clusters as opposed to single groups since they are both at or below . Therefore, the presence of clusters should not change the detected frequency components of the PA signal.
The widefield microscopy used to generate the images for Fig. 3 is unable to optically resolve these level clusters but it does demonstrate specificity of the nanoparticles to the cancer cells over leukocytes. However, there is nonspecific attachment of nanoparticles to leukocytes, as demonstrated by Fig. 3(b). About 5% to 10% of leukocytes are nonspecifically labeled with 10% to 17% of the average fluorescent intensity found on the average breast cancer cell. There are to luekocytes in 1 mL of human blood while there will be at most cancer cells in 1 mL.3,4 Even though they are not as photoacoustically active, this creates a significant number of background absorbers (about 200,000 to 500,000). Currently, blocking buffers composed of bovine serum albumin and mouse serum are being explored to reduce nonspecific labeling with promising results. Initially, blocking agents were not used in the labeling procedure in order to create the simplest possible technique. For these reasons of nonspecific labeling of leukocyte, cancer cells that were prelabeled with DAPI and nanoparticles were added to the vial of blood for the results presented in Sec. 3.4.
Detection and Classification of Breast Cancer Cells
The two features computed from the PA waveforms collected from each irradiation measure the presence of large variations from the mean. Both of these time-domain features are fairly intuitive since one is an integral and the other is a count of values above noise thresholds (positive and negative).
The method of classification being used with this feature set is threshold based since the distribution of features from the control of PBS group parametrizes well to a bivariate normal distribution. Other parametric methods were difficult to pursue as the feature distribution from the breast cancer cell group was not easily described by mathematically tractable distributions.
It is important to note that though the expected value is , this will not be the case for every occurrence. Table 1 uses the Poisson distribution to calculate the probability of finding a specific number of cells in the irradiated volume of given that the average rate is two cells. Since there will be 0 cells in the irradiated volume about 13.5% of the time, we should expect an ideal true positive rate of 86.5%, which has been marked with horizontal dashed green line in Fig. 8. However, to detect one cancer cell in 1 mL with certainty, the false positive rate must be reduced to below 1 per 1 mL of sample. Therefore, the maximum acceptable false positive rate must be since the irradiated volume is . The red vertical dashed line in Fig. 8 marks this maximum acceptable false positive rate, which is paired with an approximate true positive rate of 67% with a threshold larger than . Moreover, the slope of the ROC curve changes from greater than 1 to less than 1 at the threshold of , implying that for every incremental increase in the false positive rate, there is a smaller increase in the true positive rate. This demonstrates that there are diminishing returns when the threshold is reduced beyond .
Expected value of two cells.
|Number of cells||Probability|
According to the Poisson distribution, the probability of finding two or more cells in the irradiated volume is 0.592 and a larger proportion of 0.67 is classified as true positive, implying a detection limit between one and two cells for the flow system. This is understandable since Bhattacharyya et al.12 have shown that the nanoparticle load from one cancer cell to another can vary.12 Moreover, the position of the cell in the beam spot can also change the sensitivity of the system as the energy profile and acoustic sensitivity are not uniform across the irradiated area or volume.
Detection and Retrieval of Cancer Cells from Blood
About 35 cancer cells were added to 1 mL of blood and 15 were isolated, located, and imaged. However, the initial number of cancer cells added to blood was probably not exactly 35, since a small volume of stock breast cancer cell suspension was added to the blood. It is safer to assume that the initial number of cancer cells added was within 25 to 45 cells if the variance is assumed to be equal to the expected value when drawing from a large sample.
Moreover, the centrifugation procedure used to remove erythrocytes causes 10% to 30% cell loss for leukocytes and cancer cells.18 This implies that the total number of cancer cells had been reduced to 17 to 40 cells even before the solution was tested with the PAFC. The automated system classified 17 signals as being different from leukocytes but only 15 cells were found in 13 of the droplets. This may be because some cells were lost due to the inefficiency of the capturing protocol. For this study, droplets were captured manually but this procedure is being automated to provide more consistent results. When the system detected a cancer cell, the droplet was captured downstream of where it was irradiated after a specific amount of time, which allowed the droplet to reach the end of the chamber. It is also possible that the remaining four droplets never contained any cancer cells and were false positives. This is slightly higher than the expected number of false positives (), which could be caused by stray erythrocytes.
Since 15 cells were captured and about 35 cells were added, the retrieval rate of the entire protocol with the in vitro PAFC is . Nonetheless, it is difficult to determine an exact retrieval rate since there could have been as few as 25 cancer cells added to the blood. If the retrieval rate is indeed 43%, at least four cancer cells would have to be present in 1 mL of blood to have about a 90% chance of retrieving one cell. Though Giuliano et al.4 have found with the CellSearch system that of blood has significant prognostic value, Kirby et al.5 have argued with the geometrically enhanced differential immunocapture system that CellSearch underestimates the cancer cell concentrations by 2 to 400 times. From these sources it can be inferred that the in vitro PAFC system detects clinically relevant concentrations of cancer cells in blood.
The advantages of this technology over existing CTC technologies is that the cells remain in suspension and are never attached to a substrate like many antibody technologies. Additionally, it does not depend on size or mechanical stiffness like many technologies, as these parameters have a huge biological variability.
However, it is important to remember that the cancer cells had been prelabeled with both DAPI and nanoparticles before they were added to the blood. In a clinical blood test, the cancer cells in the blood cannot be prelabeled but must be specifically labeled in the presence of leukocytes. As mentioned in Sec. 4.2, the antibody mediated labeling procedure is currently being developed since there is some nonspecific labeling of leukocytes with nanoparticles. However, once nonspecific labeling is reduced, similar retrieval rates are expected since the PAFC is able to detect the presence of one to two cancer cells in the chamber.
Figure 11 shows images of two different captured cancer cells surrounded by leukocytes. The red fluorescence from the fluorescent-gold particle was bleached when the cells were irradiated with the PAFC, making DAPI the only characteristic stain on the cancer cells. Though this method of identification was adequate for this study, a different method of identifying cancer cells will have to be used for a clinically relevant test. The captured breast cancer cells could be labeled with fluorescent antibodies for human epidermal growth factor receptor 2 or human milk fat globulin for secondary verification and visual identification, as these are known to be overexpressed in breast cancers.19,20
Finally, Fig. 12 demonstrates how the identified cell could then be removed from the solution with a micromanipulator controlled micropippette and transported to another location. In this case, the cells are transported to a cytophilic well in a different slide. However, this new location can be the cap of a polymerase chain reaction (PCR) tube filled with cell lysis buffer to start the first step to an quantitative reverse transcription PCR method or it could be a Petri dish filled with culture media in order to culture this captured cancer cell.
This technique can be modified to apply to other cancers by changing the antibody conjugated to the nanoparticles used for PA contrast. This study demonstrates the ability to capture single circulating cells and is a path toward the possibilities that are sure to transform oncology in the next few decades.
We acknowledge the support of the National Cancer Institute and the National Institutes of Health under award number 1-RO1-CA161367-01. Cells were cultured by the Cell and Immunology Core at the University of Missouri. All images were acquired at the Molecular Cytology Core at the University of Missouri. The authors would like to thank Dr. Anand Chandrasekhar for lending his micropipette puller. Additional acknowledgments are due to the machine shop in the Biological Engineering Department for machining portions of the flow cytometer. Finally, this work would have been impossible without all of the blood donations from members of the Viator Lab.
I. R. Hart and I. J. Fidler, “Role of organ selectivity in the determination of metastatic patterns of B16 melanoma,” Cancer Res. 40, 2281–2287 (1980).CNREA80008-5472Google Scholar
P. Paterlini-Berchot and N. L. Benali, “Circulating tumor cells (CTC) detection: clinical impact and future directions,” Cancer Lett. 253, 180–204 (2007).CALEDQ0304-3835http://dx.doi.org/10.1016/j.canlet.2006.12.014Google Scholar
M. C. Miller, G. V. Doyle and L. W. M. M. Terstappen, “Significance of circulating tumor cells detected by the CellSearch system in patients with metastatic breast colorectal and prostate cancer,” J. Oncol. 2010, 617421 (2010).http://dx.doi.org/10.1155/2010/617421Google Scholar
M. Giuliano et al., “Circulating tumor cells as prognostic and predictive markers in metastatic breast cancer patients receiving first-line systemic treatment,” Breast Cancer Res. 13, R67 (2011).BCTRD6http://dx.doi.org/10.1186/bcr2907Google Scholar
B. J. Kirby et al., “Functional characterization of circulating tumor cells with a prostate-cancer-specific microfluidic device,” PLoS One 7, e35976 (2012).POLNCL1932-6203http://dx.doi.org/10.1371/journal.pone.0035976Google Scholar
E. Galanzha et al., “In vivo, noninvasive, label-free detection and eradication of circulating metastatic melanoma cells using two-color photoacoustic flow cytometry with a diode laser,” Cancer Res. 69, 7926–7934 (2009).CNREA80008-5472http://dx.doi.org/10.1158/0008-5472.CAN-08-4900Google Scholar
G. Gutierrez-Juarez et al., “Detection of melanoma cells in vitro using an optical detector of photoacoustic waves,” Lasers Surg. Med. 42, 274–281 (2010).LSMEDI0196-8092http://dx.doi.org/10.1002/lsm.v42:3Google Scholar
J. Viator et al., “Gold nanoparticle mediated detection of prostate cancer cells using photoacoustic flowmetry with optical reflectance,” J. Biomed. Nanotechnol. 6, 187–191 (2010).http://dx.doi.org/10.1166/jbn.2010.1105Google Scholar
E. Galanzha et al., “In vivo magnetic enrichment and multiplex photoacoustic detection of circulating tumour cells,” Nat. Nanotechnol. 4, 855–860 (2009).NNAABX1748-3387http://dx.doi.org/10.1038/nnano.2009.333Google Scholar
D. McCormack et al., “Enhanced photoacoustic detection of melanoma cells using gold nanoparticles,” Lasers Surg. Med. 43, 333–338 (2011).LSMEDI0196-8092http://dx.doi.org/10.1002/lsm.v43.4Google Scholar
D. Nedosekin et al., “In vivo ultra-fast photoacoustic flow cytometry of circulating human melanoma cells using near-infrared high-pulse rate lasers,” Cytometry 79A, 825–833 (2011).1552-4922http://dx.doi.org/10.1002/cyto.a.v79a.10Google Scholar
K. Bhattacharyya et al., “Gold nanoparticle-mediated detection of circulating cancer cells,” Phys. Med. Biol. 60, 3081–3096 (2015).PHMBA70031-9155http://dx.doi.org/10.1088/0031-9155/60/8/3081Google Scholar
E. I. Galanzha and V. P. Zharov, “Circulating tumor cell detection and capture by photoacoustic flow cytometry in vivo and ex vivo,” Cancers 5, 1691–1738 (2013).http://dx.doi.org/10.3390/cancers5041691Google Scholar
C. O’Brien et al., “Capture of circulating tumor cells using photoacoustic flowmetry and two phase flow,” J. Biomed. Opt. 17, 061221 (2012).JBOPFO1083-3668http://dx.doi.org/10.1117/1.JBO.17.6.061221Google Scholar
M. Ishii, “Thermo-fluid dynamic theory of two-phase flow,” in NASA STI/Recon Technical Report A, Vol. 75, p. 29657, NASA Scientific and Technical Information Program, Hampton, Virginia (1975).Google Scholar
R. Gertler et al., “Detection of circulating tumor cells in blood using an optimized density gradient centrifugation,” Recent Results Cancer Res. 162, 149–155 (2003).RRCRBU0080-0015http://dx.doi.org/10.1007/978-3-642-59349-9Google Scholar
J. Baselga et al., “HER2 overexpression and paclitaxel sensitivity in breast cancer: therapeutic implications,” Oncology 11, 43–48 (1997).Google Scholar
J. Burchell et al., “Development and characterization of breast cancer reactive monoclonal antibodies directed to the core protein of the human milk mucin,” Cancer Res. 47, 5476–5782 (1987).CNREA80008-5472Google Scholar
Kiran Bhattacharyya is a doctoral candidate at Northwestern University in the Biomedical Engineering Department. He received his BS and MS degrees in bioengineering from the University of Missouri. His research interests include applying biomedical optics and nanotechnology methods to cancer detection, cellular analysis and manipulation, and neuroscience.
Benjamin S. Goldschmidt is an assistant professor of biomedical engineering at Duquesne University. He received his BS and PhD degrees in bioengineering from the University of Missouri in Columbia, where he worked on photoacoustic flow cytometry and developed total internal reflection methods using photoacoustics. He is actively involved in biomedical optics, particularly in the areas of photoacoustics and materials characterization. He also has experience in commercialization of optical technologies for human health.
John A. Viator is a professor and founding director of the Biomedical Engineering Program at Duquesne University. He is also an adjunct professor of bioengineering at the University of Pittsburgh. He received his PhD in electrical engineering from Oregon Health & Science University. He has held positions at the University of California, Irvine, and the University of Missouri prior to arriving at Duquesne. His research interests are in biomedical optics, specifically in photoacoustic sensing and imaging.