Mimicking the "Seed and Soil:" The State of PDX Models of Breast Cancer Metastases

Why do the quaintest analogies stick to biology’s most morbid phenomena? Look no further than English surgeon Stephen Paget’s “seed and soil” model for breast cancer metastasis for a prototypical example. In his 1889 publication in the Lancet, Paget compares the spread of cancer from primary to secondary sites (which ultimately accounts for 90% of cancer deaths) and plant germination. “When a plant goes to seed,” Paget explains, “its seeds are carried in all directions; but they can only live and grow if they fall on congenial soil.”¹  

It is the same, Paget argues, for metastatic cancer cells. He came to this conclusion through his analysis of 735 cases of breast cancer, determining that where and when secondary tumors grow is not dictated merely by anatomic factors, but must require some favorable interaction between breast cancer cells and specific host organs.² 

Modern Models of Metastasis 

Centuries later, as researchers and clinicians have teased apart the molecular and cellular mechanisms of metastasis, the “seed and soil” model still resonates. And it accurately describes why the spread of primary tumors (the “seed’) to specific tissues is not due to chance. Accordingly, evidence now suggests that target organs may develop pre-metastatic niches, predisposing them to metastatic growth by promoting tumor cell homing and subsequent seeding.² 

Both in vitro and in vivo models of cancer metastasis have helped to form the foundation of our current knowledge. But there is still a lot that researchers do not know about the timing and location of metastasis. Our understanding of this complex process is still evolving, and the development and use of improved, clinically accurate models that can mimic the process of metastasis, as a whole, is needed. Clinically accurate models are needed not only to understand the molecular inner workings of metastasis, but to develop efficacious therapeutic strategies that target the metastatic cascade. 

Exploring Metastatic Breast Cancer  

In the last three decades, the age-adjusted death rates from breast cancer have decreased by 42%, according to the AACR Cancer Progress Report. However, the 5-year survival rate for metastatic breast cancer remains steadily bleak, at 29% for those whose cancer has spread to a distant part of the body. According to Cancer.net, nearly 342,000 people in the US (99% of them women) will be diagnosed with breast cancer this year.^3,4 For women who receive an early diagnosis with breast cancer that is only localized to the breast tissue (based on the SEER database), the 5-year relative survival rate is very good at 99%.⁵  In part, this is due to the encouragement of mammographic screening and early diagnosis of breast cancer and administration of efficacious adjuvant therapy, which includes chemotherapy and/or endocrine therapy following surgical removal and/or radiotherapy of the primary tumor.⁶ 

Yet, when breast cancers go undiagnosed and metastasize to other areas of the body, clinical outcomes change significantly. Metastatic breast cancer is the second leading cause of death from cancer (second to skin cancer) in women and an estimated 43,780 Americans (43,250 women and 530 men) will die in 2022 as a result.⁴ While new treatments are being developed and approved, there is currently a lack of highly efficacious treatments, and the five-year relative survival rate for metastatic breast cancer is 29%.⁵  

Gaps in Understanding the Metastatic Cascade 

While many insights about breast cancer metastasis have been made, the inter-tumoral heterogeneity of the primary tumors makes it challenging to generalize about the timing and location of metastasis. Different breast cancer molecular subtypes change the nature of both the “seed” and “soil.” Intra-tumoral heterogeneity adds another confounding factor. 

For instance, the expression of hormone receptors, either estrogen receptor (ER) or progesterone receptor (PR), is associated with metastasis to the bone or lymph nodes.⁷ Triple-negative breast cancer (TNBC) preferentially metastasizes to the lungs, liver, and/or brain, while human epidermal growth factor receptor type 2 positive (HER2+) breast cancers often metastasize to the brain following relapse from treatment with HER2-targeting therapies, such as trastuzumab or pertuzumab.^7,8 

Metastasis requires a series of steps, including 1) invasion and intravasation from the primary tumor, 2) survival in circulation in the vasculature, 3) extravasation, 4) seeding, and 5) metastatic outgrowth.7 But it is unclear just how linear or discrete these steps are and how they are shaped by the accumulation of additional mutations or epigenetic signaling.7 Without such knowledge, it can be difficult to predict recurrence, metastatic potential, or organ tropism. 

These complications place several barriers in the path of drug developers and beg the question, what is the best model for drug development in metastatic breast cancer? 

The Pros and Cons of Preclinical Models of Metastatic Breast Cancer 

Like everything in biology, the answer to, what is the best model for drug development in metastatic breast cancer? is, “it depends.” Mimicking the various mechanisms of metastasis is critical for any preclinical model. But no model perfectly mimics each step in secondary growth as it occurs in the clinic. There are advantages and drawbacks for the models discussed below and some combination is likely necessary to get a holistic view of how different therapeutic interventions affect primary or metastatic tumors.⁷ 

In Vitro Models of Early Breast Cancer Metastasis 

The complicated process of metastasis is difficult to study fully using an in vitro model, yet the careful control of genetic backgrounds and culturing conditions afforded by these models is attractive. It enables the study of early steps in the metastatic process. For instance, migration is a critical step in metastasis, and 2-D cell culture using a scratch or wound healing assay can be used to untangle the molecular steps or identify drugs that inhibit migration.⁷ Likewise, chemotactic studies using a transwell system have proven valuable for the study of metastatic breast cancer and even led to the establishment of useful cell lines, such as the bone-tropic mouse mammary cell line, 4T1.7 Co-culturing systems with macrophages or endothelial cells and the addition of extracellular matrix (ECM) can also model some aspects of the tumor microenvironment in a controlled manner.⁷ 

The use of in vitro models has been facilitated by the characterization of several human breast cancer cell lines, that represent the common clinical subtypes, including MCF-7 (ER/PR+ subtype), MDA-MB-231 (TNBC), HCC202 (HER2+), and many others. Due to their extensive use in vitro and in vivo (see below), they can be used predictably, however, there are significant drawbacks to their use for drug development or R&D. These cell lines are derived from aggressive tumors and have been passaged for decades in various labs, leading to genetic drift and the generation of subtypes that behave differently from the parental cell line.⁷  

3-D models have emerged more recently, as an in vitro system that maintains the control of culturing conditions seen with 2-D models and is more physiologically relevant for metastatic studies.⁷ 3-D organoid biobanks of patient tumors that represent the phenotypic and molecular heterogeneity seen in breast cancer have been created.^9,10 3-D organoid models have also been used to study the role of ECM in the migration of metastatic breast cancer cells and develop organotypic cultures, such as the Bone-In-Culture-Array, that can help serve as a bone metastasis model for breast cancer.⁷ 

Advantages⁷ 

• Control of cellular and ECM inputs and culturing variables 
• Scalable testing 
• Low cost and rapid testing 
• Easy to reproduce and manipulate for assays 
• Closely mimic single aspects of in vivo physiology 
• Accommodates cell lines, patient-derived tumors  

Disadvantages⁷ 

• Missing relevant components of complex in vivo environment, including tumor microenvironment, ECM, etc. 
• Lack of intra-tumoral heterogeneity 
• Cannot replicate late stages of metastases 
• Genetic drift over longitudinal passaging 

In Vivo Cell-Derived Xenograft (CDX) Models of Early Breast Cancer Metastasis 

Mouse xenograft models hold a lot of potential and several studies have demonstrated how powerful these models are for mimicking metastasis.^7,11,14 The simplest (and thus, most widely used) in vivo model of metastatic breast cancer is the CDX model.¹⁴ Yet, for all of the reasons (i.e., lack of heterogeneity, genetic drift) about the use of cell lines mentioned above, there are several caveats to using these models for drug development. 

There are additional considerations as well, including whether to use a subcutaneous or orthotopic CDX model. Subcutaneous injection is often considered the easiest and tumor growth can be measured easily using calipers, however, the environment in which the tumor is grown is dissimilar from the orthotopic site, the mammary fat pad.¹⁴ Orthotopic injection, is often viewed as more challenging, but tumor growth can be easily measured in a mouse model using calipers as well. One major drawback of CDX models, even those cell lines implanted orthotopically, is that not all cell lines can metastasize.^7,14 For those that do, it is difficult to determine how closely they mimic the clinical course of the original patients (i.e., human cell lines) from which they were originally derived. 

 Advantages¹⁴ 

• Reproducible growth 
• Several tumor models, representative of breast cancer subtypes, are readily available 
• Trackable with bioluminescent or fluorescent models 
• Rapid and cost-effective 

Disadvantages¹⁴ 

• Immunodeficient host 
• Cannot replicate early stages of metastases (for many cell lines) 
• Does not accurately represent intra-tumoral heterogeneity 
• Genetic drift can change cell line behavior 

In Vivo Models for Middle and Late-Stage Breast Cancer Metastasis 

One common model for the middle to later, post-intravasation stages of metastatic breast cancer uses the breast cancer cell lines described above for injection into the vasculature of an immunodeficient mouse model.^7,11 Injection can be done in several ways, including:⁷ 

• Tail vein injections, which results in lung metastases 
• Intracardiac injection, which results in brain and bone metastases 
• Intracarotid injection, which results in brain metastases 
• Intra-iliac artery injections, which preferentially seed bone metastasis 

Derivatives of the MDA-MB-231 TNBC cell line that preferentially metastasize to the lungs are widely used for these types of experiments, but derivatives that home to the brain and bone are also used.⁷ Metastasis can be sensitively tracked through the use of cell lines that stably express bioluminescent or fluorescent markers, such as luciferase or GFP, respectively.^12,13 This model is also amenable to the use of patient-derived tumors, though unless genetically engineered, can be challenging to track the locations of secondary growth.⁷ Despite these shortcomings, these models have been successfully used to gain a better understanding of metastatic fitness and test drug efficacy. 

 Advantages⁷ 

• Reproducible, site-specific metastases 
• Several tumor models, representative of breast cancer subtypes, are readily available 
• Trackable with bioluminescent or fluorescent models 
• Cell lines can be used rapidly 

Disadvantages⁷ 

• Human cell lines or use of patient-derived tumors require the use of immunodeficient mouse models 
• Cannot replicate early stages of metastases 
• Cell lines do not accurately represent intra-tumoral heterogeneity 
• Cell lines experience genetic drift as they are passaged 

In Vivo Patient-Derived Xenograft (PDX) Models of All Stages of Breast Cancer Metastasis 

An ideal preclinical metastatic breast cancer model should accurately mimic each stage of metastasis seen in the clinic. Thus far, the models we’ve discussed fall short of that. 

Many of the shortcomings of the models described above can be overcome through the use of PDX models, implanted into immunodeficient mice. Though more expensive and time-consuming to establish, PDX models more accurately capture the intra-tumoural heterogeneity, tumor microenvironment, and disease progression, including growth kinetics and histology.^7,14,15  

In the past decade, PDX collections, both public and private, have made establishing a reproducible PDX model for translational research, much easier.^14,15 For instance, Dobrolecki et al. have established, “...over 500 stably transplantable PDX models representing all three clinical subtypes of breast cancer…”¹⁵ These models are “stably transplantable” and treatment responses, metastatic behavior, and genomic, transcriptomic, and proteomic characteristics have been analyzed.¹⁵ This resource and others like it are indispensable for creating a more accurate preclinical model for breast cancer. 

In terms of accurately modeling metastatic breast cancer, the site of implantation can play a major factor. There is growing evidence that orthotopic PDX models (O-PDX) offer a more clinically relevant model than subcutaneous PDX models. The subcutaneous vs. O-PDX debate has continued on for some time and for metastatic breast cancer, there is not an industry-wide consensus: Some studies have found that O-PDX models accurately recapitulate metastases, treatment efficacy and prognosis seen in the clinic and others do not.^7,14,16-18 As with the breast cancer cell lines discussed above, researchers are beginning to establish bioluminescent subcutaneous PDX and O-PDX models, which will enable more sensitive detection of secondary growth and precursors to metastatic growth, such as circulating tumor cells, enabling a better understanding of the clinical relevance of each model.^7,19 

 Advantages^7,14 

• Models the entire metastatic cascade 
• Uses clinically-relevant tumors 
• Capture intra- and inter-tumoural heterogeneity 
• Mimics tumor microenvironment (only for orthotopic implantation) 
• Easy to measure primary (using calipers) and secondary (using bioluminescence) tumor growth 

Disadvantages^7,14 

• Immunodeficient host 
• High cost 
• Model establishment can take a long time (~9-12 months) 

Breast Cancer PDX Co-Clinical Trials

Because PDX models capture important tumor characteristics more accurately than other models, and because of their established value as a preclincal breast cancer model, they are now being used prospectively in co-clinical trials. Co-clinical trials are preclinical trials conducted in parallel with a human clinical trial. ^20-23 A co-clinical trial uses fundamental clinical, biological, and pharmacological information from a preclincal trial to predict a treatment response in patients enrolled in the clinical trial.^20-23 A co-clinical trial with a PDX model involves creating a PDX model from the tumor (or tumors) of a corresponding enrolled patient; treating the PDX model with either the same clinical protocol that the enrolled patient receives or a new drug or drug combination; analyzing the response in the PDX model; and screening for biomarkers that replicate or predict the corresponding patient’s clinical response to the administered drug(s).^20-23 

According to clinicaltrials.gov, at least 8 co-clinical trials for metastatic breast cancer that involve creating PDX models are either recruiting or have been completed.²⁴  Six trials involve TNBC (NCT04745975, NCT04608357, NCT04526587, NCT04133077, NCT02732860, and NCT02247037), one involves HER2+ breast cancer (NCT03765983), and one involves ER+ HER2-negative breast cancer (NCT02752893).  Each trial varies in how extensively the PDX model contributes to the study, but several of the trials emphasize PDX model development and analysis (NCT04133077, NCT02752893, NCT02732860, and NCT02247037). These are promising developments for PDX models of breast cancer metastasis, and additional co-clinical trials will undoubtedly follow in the future.

Certis is Advancing O-PDX-Derived, Luciferase-Tagged Cell Cultures and OPX Models for Metastatic Breast Cancer 

As metastasis and metastatic breast cancer continue to be a major source of mortality for cancer patients, Certis is dedicated to finding the most clinically relevant “seed and soil” model for drug developers. We believe that O-PDX is that model and in conjunction with other models, we can help drive the development of novel treatment strategies forward.  

Certis is developing a variety of O-PDX models of breast cancer, primed to tackle difficult-to-treat secondary growths. For instance, 15% of breast cancers metastasize to the brain, and drugs that cross the blood-brain barrier are needed.²⁵ Certis is positioned to help establish O-PDX models that can identify preclinical drug candidates that reach these difficult metastatic tumors.  

In addition, Certis is developing O-PDX-derived, luciferase tagged cell cultures that can be utilized for late-stage metastasis analysis via tail vein, intracardiac, and intracarotid injections.  

To learn more about our drug development services, email us at busdev@certisoncology.com. 

You can also: 

• Learn more about our custom model development for translational research 
• Register for Cancer Model Database access  
• Read our promising clinical results 

References 

1. Paget S. Distribution of Secondary Growths in Cancer of the Breast. Lancet. 1889;133(3421):571-573.  
2. Akhtar M, Haider A, Rashid S, Al-Nabet ADMH. Paget's "Seed and Soil" Theory of Cancer Metastasis: An Idea Whose Time has Come. Adv Anat Pathol. 2019;26(1):69-74. 
3. Breast Cancer: Statistics. Cancer.Net website. Published January 2022. Accessed September 13, 2022. 
4. Breast Cancer - Metastatic: Statistics. Cancer.Net website. Published January 2022. Accessed September 13, 2022. 
5. Survival Rates for Breast Cancer. American Cancer Society website:  Published January 27, 2021. Accessed July 9, 2021. 
6. Weigelt B, Peterse JL, van 't Veer LJ. Breast Cancer Metastasis: Markers and Models. Nat Rev Cancer. 2005;5(8):591-602. 
7. Roarty K, Echeverria GV. Laboratory Models for Investigating Breast Cancer Therapy Resistance and Metastasis. Front Oncol. 2021;11:645698.  
8. Treatment of Stage IV (Metastatic) Breast Cancer. American Cancer Society website:  Published December 18, 2020. Accessed July 13, 2021.  
9. Sachs N, de Ligt J, Kopper O, Gogola E, Bounova G, Weeber F, et al. A Living Biobank of Breast Cancer Organoids Captures Disease Heterogeneity. Cell. 2018;172:373–86 
10. Rosenbluth JM, Schackmann RCJ, Gray GK, Selfors LM, Li CM, Boedicker M, et al. Organoid Cultures from Normal and Cancer-Prone Human Breast Tissues Preserve Complex Epithelial Lineages. Nat Commun. 2020;11:1711. 
11. Rashid OM, Maurente D, Takabe K. A Systematic Approach to Preclinical Trials in Metastatic Breast Cancer. Chemotherapy (Los Angeles). 2016;5(3):204.  
12. Jenkins DE, Hornig YS, Oei Y, Dusich J, Purchio T. Bioluminescent Human Breast Cancer Cell Lines that Permit Rapid and Sensitive In Vivo Detection of Mammary Tumors and Multiple Metastases in Immune-Deficient Mice. Breast Cancer Res. 2005;7(4):R444-R454. 
13. Wang K, Xie S, Ren Y, Xia H, Zhang X, He J. Establishment of a Bioluminescent MDA-MB-231 Cell Line for Human Triple-Negative Breast Cancer Research. Oncol Rep. 2012;27(6):1981-1989. 
14. Holen I, Speirs V, Morrissey B, Blyth K. In Vivo Models in Breast Cancer Research: Progress, Challenges and Future Directions. Dis Model Mech. 2017;10(4):359-371.  
15. Dobrolecki LE, Airhart SD, Alferez DG, et al. Patient-Derived Xenograft (PDX) Models in Basic and Translational Breast Cancer Research. Cancer Metastasis Rev. 2016;35(4):547-573.  
16. Whittle JR, Lewis MT, Lindeman GJ, Visvader JE. Patient-Derived Xenograft Models of Breast Cancer and their Predictive Power. Breast Cancer Res. 2015;17(1):17. Published 2015 Feb 10.  
17. DeRose YS, Wang G, Lin YC, et al. Tumor Grafts Derived from Women with Breast Cancer Authentically Reflect Tumor Pathology, Growth, Metastasis and Disease Outcomes. Nat Med. 2011;17(11):1514-1520. 
18. DeRose YS, Gligorich KM, Wang G, et al. Patient-Derived Models of Human Breast Cancer: Protocols for In Vitro and In Vivo Applications in Tumor Biology and Translational Medicine. Curr Protoc Pharmacol. 2013; Chapter 14: Unit14.23.  
19. Echeverria GV, Powell E, Seth S, et al. High-Resolution Clonal Mapping of Multi-Organ Metastasis in Triple Negative Breast Cancer. Nat Commun. 2018;9(1):5079.  
20. Nardella C, Lunardi A, Patnaik A, Cantley LC, Pandolfi PP. The APL Paradigm and the “Co-Clinical Trial” Project. Cancer Discov. 2011;1(2):108-116.
21. Kim HR, Kang HN, Shim HS et al. Co-clinical Trials Demonstrate Predictive Biomarkers for Dovitinib, an FGFR Inhibitor, in Lung Squamous Cell Carcinoma. Ann Oncol. 2017;28(6):1250-1259.
22. Jung J, Seol HS, Chang S. The Generation and Application of Patient-Derived Xenograft Model for Cancer Research. Cancer Res Treat. 2018;59(1):1-10.
23. Koga Y, Ochiai A. Systematic Review of Patient-Derived Xenograft Models for Preclinical Studies of Anti-Cancer Drugs in Solid Tumors. Cells. 2019;8(5):418-431.
24. Clinicaltrials.gov. U.S. National Library of Medicine website. Accessed September 13, 2022. 
25. Suh JH, Kotecha R, Chao ST, Ahluwalia MS, Sahgal A, Chang EL. Current Approaches to the Management of Brain Metastases. Nat Rev Clin Oncol. 2020;17(5):279-299.  

ABOUT THE AUTHOR

Elie Diner has a PhD in bioengineering and 12 years of research experience in microbiology, synthetic biology and immunology. During his time at the bench, he developed a passion for effective science communication and eventually transitioned into a career as a professional science and content writer. He's authored 12 peer-reviewed scientific publications and numerous blogs, whitepapers, and eBooks for life science companies.