Quantitative Microscopy in Medicine =================================== * Alexandre Matov * Shayan Modiri ## ABSTRACT Methods for personalizing medical treatment are the focal point of contemporary biomedical research. In cancer care, we can analyze the effects of therapies at the level of individual cells. Quantitative characterization of treatment efficacy and evaluation of why some individuals respond to specific regimens, whereas others do not, requires additional approaches to genetic sequencing at single time points. Methods for the analysis of changes in phenotype, such as *in vivo* and *ex vivo* morphology and localization of cellular proteins and organelles can provide important insights into patient treatment options. Novel therapies are needed to extend survival in metastatic castration-resistant prostate cancer (mCRPC). Prostate-specific membrane antigen (PSMA), a cell surface glycoprotein that is commonly overexpressed by prostate cancer (PC) cells relative to normal prostate cells, provides a validated target. We developed a software for image analysis designed to identify PSMA expression on the surface of epithelial cells in order to extract prognostic metrics. In addition, our software can deliver predictive information and inform clinicians regarding the efficacy of PC therapy. We can envisage additional applications of our software system, beyond PC, as PSMA is expressed in a variety of tissues. Our method is based on image denoising, topologic partitioning, and edge detection. These three steps allow to segment the area of each PSMA spot in an image of a coverslip with epithelial cells. Our objective has been to present the community with an integrated, easy to use by all, tool for resolving the complex cytoskeletal organization and it is our goal to have such software system approved for use in the clinical practice. Key words * Circulating Tumor Cells (CTCs) * Prostate Specific Membrane Antigen (PSMA) * Stationary Wavelet Transform * Watershed Algorithm * Active Contour ## INTRODUCTION Prostate cancer (PC) affects hundreds of thousands of men each year in the United States and around the World. Despite advances in diagnostic and treatment strategies, PC is the second most common cause of cancer mortality in men (*1*). Though treatment options for metastatic castration resistant prostate cancer (mCRPC) have expanded over the past decade, highly proliferative phenotypes frequently emerge at the time of progression on androgen signaling inhibitors. Though initially effective in reducing tumor burden for some patients, resistance to systemic is universal and approximately one-third of tumors are primarily refractory to this treatment approach. This is clinically highly significant as mCRPC refractory to tubulin inhibitors is uniformly fatal within 12-18 months (*2*). Currently, mCRPC chemotherapy is limited to FDA-approved docetaxel and cabazitaxel, and could be potentially impactful. However, even for the FDA-approved taxanes, there is currently no good mechanistic understanding of drug action and, for instance, docetaxel is always used as first line without having to examine patient cells *ex vivo* for susceptibility. For patients with progressive mCRPC that was refractory to androgen receptor pathway inhibitors and had received or been deemed ineligible for taxane chemotherapy, there are not many treatment options available. For patients with mutated BRACA1/2, treatment with olaparib offers a benefit (*3*) and novel therapies are needed to extend survival. Prostate-specific membrane antigen (PSMA), a cell surface antigen overexpressed in PC, provides a validated target. 225Ac-J591, anti-PSMA monoclonal antibody J591 radiolabeled with the alpha emitter actinium-225 is being investigated in the context of safety, efficacy, maximum tolerated dose (MTD), and recommended for phase II dose (RP2D) (*4*). PSMA expression is found in higher grade PC as well as mCRPC (*5*), while its expression is low in normal tissues (*6*), and recent clinical trials have validated PSMA as a target for therapy for high-risk and advanced PC (*7*), demonstrating it may offer benefits over the use of tubulin inhibitors (*8*). This highlights the need to develop methods to inform PSMA-targeting therapeutic strategies, which is the focus of this contribution. We have developed an image analysis method for segmenting the cellular areas with cell surface expression of PSMA. Phenotype analysis can confirm the presence of significant cell surface expression of PSMA before initiating treatment. Further, during treatment, a determination of an effective drug-target engagement on PSMA, evidenced by changes in image metrics, may allow early detection of molecular response to treatment, or lack thereof molecular response. This will improve the precision of therapy and its customization to the individual. ### PSMA IMAGE SEGMENTATION The raw PSMA imaging data, shown on Fig. 1A, shows proximity of cells in clusters (see the encircled area for an example of a cluster). That presented us with two problems - to (1) correctly identify the four cells in the cluster and (2) precisely segment the cellular areas for each cell. To this end, we first processed the raw image with a stationary wavelet transform (*9*) and this step identified clusters of pixels with high intensity values, which we call seeds. In the same region, highlighted with a red circle on Fig. 1B, it detected four bright pixel clusters. As our next step, we wanted to outline the precise contours based on the detected seeds and the areas were refined using an active contour algorithm (*10*). This step, however, fused three of the areas, as seen within the red circle on Fig. 1C. We then applied a watershed transformation (*11*) based on the seeds shown on Fig. 1B. The watershed results are presented on Fig. 1D and we can see the topological lines representing the borders of the four cellular areas discussed above (shown in the green circle). For visualization purposes and to appreciate the segmentation result, we show on Fig. 1E the overlay of Fig. 1C and Fig. 1D, both with reversed intensities, and the reader can appreciate the accuracy of the separation between the four cells within the green circle. In order to obtain the final segmentation, we performed a logical conjunction (a logical “and” operation) between the image shown on Fig. 1C and the black borders shown on Fig. 1D. This allowed to improve the precision of the contours from Fig. 1C by introducing pixels with zero intensity as borders (Fig. 1D) between any fused cellular areas. The improved segmentation for the four discussed cells and can be seen in the white circle on Fig. 1F. It depicts a correct representation of the segmentation masks of the raw image of the four cells highlighted with the white circle on Fig. 1A. This way, we can precisely identify the expression levels of PSMA in patient cells collected from a peripheral blood draw before and during therapy. This method can aid in the drug selection and any changes in the regimen based on the *ex vivo* analysis of patient cells. ![Figure 1.](http://medrxiv.org/http://medrxiv.stage.highwire.org/content/medrxiv/early/2024/08/14/2024.07.31.24311304/F1.medium.gif) [Figure 1.](http://medrxiv.org/content/early/2024/08/14/2024.07.31.24311304/F1) Figure 1. Strategy for the analysis of organoid morphology after drug treatment. (A) Raw image of PSMA labeling in CTCs. The white circle shows four cells in high proximity. The inset shows the 8 mm coverslip. The white square marks the zoomed area in (A). (B) The initial image segmentation is accomplished by stationary wavelet transform, which identifies bright clusters in noisy images; we use this step as “seeding”. (C) Active contour, as our next step, identifies precisely the edges of the image features based on the seeds. (D) Watershed of the seeding step in (B). (E) Overlay of (C) and (D), reversed intensities. (F) Logical “and” of (C) and (D) identifies the area and their exact borders. ## CONCLUSIONS About a third of the patients progressing after treatment with hormonal therapies have intrinsic resistance to systemic therapy. The two-thirds which respond, suffer visible side effects from drug toxicity and rapidly acquire resistance to therapy. Overall, the treatment of advanced PC is hampered by the lack of known molecular markers for reliable targeted therapies. In this context, the PSMA offers a much-needed target and clinical trials are demonstrating that patients experience low levels of toxicity and manageable side effects. In this contribution, we have proposed a method to evaluate the expression of PSMA in patient cells and inform therapy. Besides PC, the expression of PSMA is elevated in other epithelial tumors, such as tumors in the kidney and the breast, which shows an applicability of the approach beyond diseases of the prostate. We compared our method, based on the analysis of PSMA labeling only, to analysis of multiplex microscopy imaging and generated also datasets for which PSMA images were acquired in combination with positive epithelial cell nuclear identifiers (DAPI staining) and negative leukocyte cell identifiers (CD45 staining). However, the three-color datasets did not offer any computational benefit and did not improve on our ability to correctly identify the tumor cells and the areas with PSMA expression. This highlights the ability of this clinical method to detect disease and monitor drug activity based on a peripheral blood draw, a simple processing step, and a straightforward quantitative microscopy method. ## MATERIALS AND METHODS ### Coverslip Processing Peripheral blood previously drawn from mRPC patients (the samples were de-identified) receiving docetaxel-based first line chemotherapy was thawed and processed using ficolling (Ficoll-Paque®) to separate cells in the buffy coat from the blood. The cells were isolated by centrifugation, fixed in phenol, and blocked on 8 mm coverslips. In brief, the ficolling method, through centrifugation, isolates Peripheral Blood Mononuclear Cells (PBMCs) and plasma from granulocytes and erythrocytes. The mononuclear layer contains lymphocytes, monocytes, and CTCs. These cells are plated onto an 8 mm coverslip and subsequently stained for biomarkers (PSMA, CD45, and DAPI). PSMA labeling was done using 177Lu -J591, anti-PSMA monoclonal antibody J591. ### Microscopy Imaging The type of microscopy used was widefield epi-fluorescence with 10x objective. The acquisition was accomplished using an ORCA-Flash 4.0 (6.5 µm/pixel) camera. ### Image Analysis All image analysis programs for PSMA segmentation and graphical representation of the results were developed in MATLAB (Mathworks, MA) and C/C++. The wavelet transform method used, spotDetector, was described and validated in (*9*). The computer code is available for download at: [https://github.com/amatov/SegmentationBiomarkerCTC](https://github.com/amatov/SegmentationBiomarkerCTC). ## Data Availability All data produced in the present study are available upon reasonable request to the authors [https://github.com/amatov/SegmentationBiomarkerCTC](https://github.com/amatov/SegmentationBiomarkerCTC) ## Ethics Declaration Not applicable. ## Funding No funding. ## ACKNOWLEDGEMENTS We thank Guang-Yu Lee for processing coverslips from a clinical study with IRB protocols 0804009740 at Cornell Medicine and acquiring images. ## Footnotes * Corrected minor formatting issues with the manuscript. * Received July 31, 2024. * Revision received August 13, 2024. * Accepted August 14, 2024. * © 2024, Posted by Cold Spring Harbor Laboratory This pre-print is available under a Creative Commons License (Attribution-NonCommercial-NoDerivs 4.0 International), CC BY-NC-ND 4.0, as described at [http://creativecommons.org/licenses/by-nc-nd/4.0/](http://creativecommons.org/licenses/by-nc-nd/4.0/) ## REFERENCES 1. 1. F. Bray et al., Global cancer statistics 2018: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA: a cancer journal for clinicians 68, 394–424 (2018). [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.3322/caac.21492&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=30207593&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F08%2F14%2F2024.07.31.24311304.atom) 2. 2. E. Francini et al., Impact of new systemic therapies on overall survival of patients with metastatic castration-resistant prostate cancer in a hospital-based registry. Prostate cancer and prostatic diseases 22, 420–427 (2019). [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1038/s41391-018-0121-2&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=30643173&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F08%2F14%2F2024.07.31.24311304.atom) 3. 3. M. De Santis et al., Feasibility of Indirect Treatment Comparisons Between Niraparib Plus Abiraterone Acetate and Other First-Line Poly ADP-Ribose Polymerase Inhibitor Treatment Regimens for Patients with BRCA1/2 Mutation-Positive Metastatic Castration-Resistant Prostate Cancer. Advances in therapy (2024). 4. 4. S. T. Tagawa et al., Prostate-Specific Membrane Antigen-Targeting Alpha Emitter via Antibody Delivery for Metastatic Castration-Resistant Prostate Cancer: A Phase I Dose-Escalation Study of (225)Ac-J591. Journal of clinical oncology : official journal of the American Society of Clinical Oncology 42, 842–851 (2024). 5. 5. D. G. Bostwick, A. Pacelli, M. Blute, P. Roche, G. P. Murphy, Prostate specific membrane antigen expression in prostatic intraepithelial neoplasia and adenocarcinoma: a study of 184 cases. Cancer 82, 2256–2261 (1998). [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1002/(SICI)1097-0142(19980601)82:11<2256::AID-CNCR22>3.0.CO;2-S&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=9610707&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F08%2F14%2F2024.07.31.24311304.atom) [Web of Science](http://medrxiv.org/lookup/external-ref?access_num=000073756900022&link_type=ISI) 6. 6. D. A. Silver, I. Pellicer, W. R. Fair, W. D. Heston, C. Cordon-Cardo, Prostate-specific membrane antigen expression in normal and malignant human tissues. Clinical cancer research : an official journal of the American Association for Cancer Research 3, 81–85 (1997). 7. 7. O. Sartor et al., Lutetium-177-PSMA-617 for Metastatic Castration-Resistant Prostate Cancer. The New England journal of medicine 385, 1091–1103 (2021). [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1056/NEJMoa2107322&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=http://www.n&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F08%2F14%2F2024.07.31.24311304.atom) 8. 8. M. S. Hofman et al., [(177)Lu]Lu-PSMA-617 versus cabazitaxel in patients with metastatic castration-resistant prostate cancer (TheraP): a randomised, open-label, phase 2 trial. Lancet (London, England) 397, 797–804 (2021). 9. 9. J.-C. Olivo-Marin, Extraction of spots in biological images using multiscale products. Pattern Recogn 35, 1989–1996 (2002). 10. 10. V. Caselles, R. Kimmel, G. Sapiro, Geodesic Active Contours. International Journal of Computer Vision 22, 61–79 (1997). 11. 11. F. Meyer, Topographic distance and watershed lines. Signal Processing 38, 113–125 (1994). [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1016/0165-1684(94)90060-4&link_type=DOI) [Web of Science](http://medrxiv.org/lookup/external-ref?access_num=A1994NX65900010&link_type=ISI)