“Is this real, or did we mislabel the tubes?” That was our first reaction.
We had received a surgical specimen from a patient with an aggressive multicentric breast cancer. As part of the sample collection, the hospital provided both tumor tissue and tissue considered healthy based on the pathological assessment of the surgical margins. We cultured organoids from the two samples separately, expecting a relatively straightforward comparison: tumor-derived organoids should retain stronger proliferative and stem-like properties, whereas organoids derived from a histologically tumor-free margin should behave more like normal tissue. But that was not what we saw.
The tumor-derived organoids looked exhausted. They were altered, differentiated and showed limited long-term growth. The organoids derived from the supposedly clean margin, however, behaved very differently. They were capable of sustained growth and displayed functional properties associated with stem-like and basal-like states, including self-renewal and migration.
Our first thought was not that we had discovered something unexpected about breast cancer. It was that we had probably made a mistake: Maybe the tubes had been labelled incorrectly, or maybe the tumor and margin samples had somehow been switched.
So we checked, and then we checked again. As more multicentric and multifocal breast cancer samples arrived, we became increasingly careful about sample identification and processing. We also began sampling tissue at different distances from the tumor margins. If what we were seeing was simply a local effect caused by the immediate proximity of tumor cells, we expected the phenotype to gradually disappear as we moved away from the lesion.
But it didn't. The same striking functional behavior continued to appear in organoids derived from histologically tumor-free margins, including samples taken at increasing distances from the tumor. At that point, the possibility that we had simply mislabeled a tube became increasingly difficult to sustain.
We still needed a proper comparison. Fortunately, unifocal breast cancer samples were arriving in parallel. These became our most reassuring negative controls. In these cases, the organoids behaved much more as we had initially expected: tumor-derived organoids more frequently displayed basal-like and self-renewal-associated properties, while margin-derived organoids tended to show reduced proliferative capacity.
In other words, our organoid system was not simply turning every breast tissue sample into an aggressive culture: the difference appeared to be associated with the biology of the disease itself.
And this was when the question changed. We had started by asking whether we had labelled the tubes correctly, but we were now asking something much more interesting: What if the tumor is not the only part of the tissue carrying the biology of the disease?
Images created by our designer, Mateo Casas
A different functional landscape
This observation became the starting point for our study comparing organoid phenotypes derived from multicentric/multifocal breast cancer (MMBC) and unifocal breast cancer (UFBC). The comparison revealed distinct functional landscapes.
In unifocal disease, tumor-derived organoids more frequently exhibited basal-like features, whereas margin-derived organoids generally showed reduced proliferative capacity. In multicentric/multifocal disease, the distribution was different: tumor-derived organoids more often displayed differentiated phenotypes and limited long-term growth, while margin-derived organoids more frequently exhibited stem-like and basal-like characteristics.
The most intriguing observation was therefore not simply that the margins could generate organoids. It was that the functional properties of tumor and margin-derived organoids were fundamentally different in multicentric disease from those observed in unifocal disease. This distinction matters because histology and functional assays are asking different questions. Histopathology provides morphological and architectural information that remains the foundation of breast cancer diagnosis and surgical decision-making. Our organoid assay was asking something else. Once cells were isolated from the tissue and placed in a defined experimental environment, what were they capable of doing? Could they self-renew? Could they maintain long-term growth? Could they migrate? Did they preferentially adopt basal-like or more differentiated phenotypes? These are functional properties, and they are not necessarily visible in a histological section.
The paradox of “clean” margins
This led us to reconsider the meaning of the word clean. A clean margin is a histological definition: no tumor is detected at the evaluated resection border. But our observations raised the possibility that tissue can be histologically tumor-free while still displaying biological heterogeneity that is not captured by conventional morphology.
This is consistent with the broader concept of a tumor “field effect”, in which tissue surrounding a tumor can exhibit molecular or functional alterations despite appearing non-malignant.
We are not suggesting that an organoid with stem-like or migratory properties proves the presence of residual tumor in the patient. Nor does our study establish that these functional phenotypes predict recurrence or treatment failure. Those are important questions for future studies. What our findings suggest is more modest, but potentially important: histologically clean does not necessarily mean functionally homogeneous.
This is why we see organoids not as a replacement for histopathology, but as a complementary experimental tool. Histology tells us what we can see in the tissue, while functional organoid-based assays may reveal what cells from that tissue are capable of doing under controlled conditions.
From tumor lesions to tumor ecosystems
The findings also made us think differently about multicentric breast cancer. MMBC disease is usually described in terms of multiple tumor lesions. Clinically and pathologically, this makes sense: the lesions can be identified, mapped and characterized. However, biologically, perhaps the relevant unit is more complex.
Our results raise the possibility that multicentric breast cancer involves a spatially heterogeneous tissue ecosystem in which different regions can occupy different functional states. The dominant tumor lesion may not necessarily represent the full spectrum of cellular states present within the patient's breast tissue.
The exhausted phenotype of tumor-derived organoids in our multicentric samples was particularly intriguing in this context. Rather than assuming that the most biologically competent cells must reside within the visible tumor, our observations suggested that functional potential might be distributed differently across the tumor and its surrounding tissue.
This is where concepts such as cellular plasticity and eco-evolutionary dynamics become useful, as frameworks for generating new hypotheses. Breast tumors and their surrounding tissues are exposed to continuously changing environments. Hypoxia, extracellular matrix remodeling, immune interactions and systemic treatments can all act as perturbations. In patients receiving neoadjuvant therapy, these pressures may be particularly relevant.
Our samples were obtained both after neoadjuvant treatment and from upfront resections. We therefore cannot disentangle, within the current study, intrinsic tumor biology from treatment-associated adaptation, but this limitation also points towards an important next step. If cellular states in tumor and margin-associated tissues are dynamic, when do they arise? Are stem-like properties already present in the margin? Are particular cell populations selected by treatment? Are these states induced by the tumor microenvironment? Or could several of these mechanisms operate simultaneously?
What happens next?
For us, this study is therefore less about providing a definitive answer than about opening a different line of investigation.
A larger and more systematic characterization of multicentric breast cancer could examine, apart from the dominant tumor lesions, additional lesions and the surrounding peritumoral tissue. Single-cell and spatial approaches could help determine which cell populations underlie the functional phenotypes observed in organoid cultures and whether these states are already present in the tissue of origin.
Prospective studies will also be needed to determine whether any of these functional features are associated with clinically meaningful outcomes such as recurrence or treatment response.
And perhaps one of the most interesting questions is whether organoids can eventually become a useful experimental bridge between morphology and function: specially to investigate the biological information that the word clean may leave out.
For now, we think one of the most important conclusions is simply that the unexpected behavior of those first organoids was worth taking seriously. Sometimes the most useful experiment begins with the suspicion that you have labelled the tubes incorrectly... And sometimes, after checking everything again and again, the result is still there.