Ductal carcinoma in situ (DCIS) may progress to ipsilateral invasive breast cancer (iIBC), but often never will. Because DCIS is treated as early breast cancer, many women with harmless DCIS face overtreatment. To identify features associated with progression, we developed an artificial intelligence-based DCIS morphometric analysis pipeline (AIDmap) on hematoxylin-eosin-stained (H&E) tissue sections. We analyzed 689 digitized H&Es of pure primary DCIS of which 226 were diagnosed with subsequent iIBC and 463 were not. The distribution of 15 duct morphological measurements was summarized in 55 morphometric variables. A ridge regression classifier with cross validation predicted 5-years-free of iIBC with an area-under the curve of 0.67 (95% CI 0.57–0.77). A combined clinical-morphometric signature, characterized by small-sized ducts, a low number of cells and a low DCIS/stroma ratio, was associated with outcome (HR = 0.56; 95% CI 0.28–0.78). AIDmap has potential to identify harmless DCIS that may not need treatment.

Abstract The current standard of care for glioblastoma (GBM), encompassing surgery, chemotherapy, and radiation, fails to extend patient survival beyond ~12-15 months. Even advanced immunotherapies, such as checkpoint blockade and chimeric antigen receptor-armored T-cell therapies, are ineffective in GBM due to tissue-specific niche-dependent escape strategies employed by glioma cells. Immune effectiveness, a finely regulated spatial and context-dependent process, underpins such a dismal state of the present therapeutic options. Herein, we applied CO-Detection by indEXing (CODEX), a state-of-the-art multiplex imaging platform, to elucidate the immune cell networks of the tumor microenvironment. By harnessing computational tools on multiplexed regions of interest across whole-slide images, we characterize the immune repertoire of GBMs by studying intratumor heterogeneity across space-histological territories- and time, matching pre- and post-treatment tissue sections of GBMs from seven patients under treatment (Stupp protocol). We aimed to assess changes in the interactions between the two main compartments of the immune system, myeloid and lymphoid, pre- and post-treatment using spatial statistics. We designed a computational workflow for the georeferencing, classification, and phenotyping individual cells of CODEX images across 105 regions of interest, representative of a mosaic of histologically defined cellular tumor (CT) or infiltrating tumor. We defined six major immune phenotypes and evaluated their spatial association with Cd45-Nestin+Olig2+ glioma cells. We modeled the GBM ecosystem and the cells’ spatial coexistence as spatial point processes that allowed the projection of coexistence networks. After stringent quality control, we detected and phenotyped 2.3M cells. Despite interpatient heterogeneity, some patients maintained an even immune composition, while others showed an enrichment in microglia, particularly in CT. Spatial network analyses revealed an increased coexistence between myeloid and lymphoid cells after treatments despite T cells being canonically considered moderately abundant. Quantitatively, the immune meta-network displayed higher edge density and lower modularity post-treatment compared with pre-treatment, reflecting increased lymphoid-myeloid engagement. In summary, we detected an increased spatial myeloid-lymphoid engagement in GBM undergoing chemo-radiation treatments. Spatiotemporal rearrangement of tumor-immune interactions indicates mechanisms implicated in disease recurrence and resistance to standard treatment, opening the frontiers for developing new targeted immunotherapies. Citation Format: Simon P. Castillo, Afrooz Jahedi, Pravesh Gupta, Jason T. Huse, Kasthuri Kannan, Yinyin Yuan, Krishna P. Bhat. Evolution of the spatial myeloid-lymphoid engagement in glioblastoma under temozolomide treatment [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 6171.

Patient characteristics: paired diagnostic and posttreatment samples.

Performance evaluation of MoSaicNet and AwareNet deep learning models: A, The ROC curves and AUC values of the MoSaicNet superpixel classifier. The values in brackets indicate the 95% CI. B, Two-dimensional mapping of superpixels using MoSaicNet learned 200-dimensional features after dimensionality reduction by UMAP. C, The ROC curves and AUC values of single-cell classifier model on separately held test data. The values in brackets indicate the 95% CI. D, UMAP features visualization of deep learned features by AwareNet single-cell classifier CNN. E and F, Validation of AwareNet model using correlation of density of CD8+ (E) and CD4+ (F) cells in panel 1 and panel 2.

Abstract Background. Ductal carcinoma in situ (DCIS) is a frequently found precursor of invasive breast cancer (IBC). However, the majority of DCIS will never progress to IBC. As we cannot distinguish yet which DCIS will remain indolent (‘harmless’) from those that might progress in the future to IBC, almost all women with DCIS are intensively treated by surgery, often followed by radiotherapy. This brings an urgent clinical need to learn distinguishing harmless from potentially progressive DCIS to save many women with indolent DCIS the burden of unnecessary overtreatment. Aim. We aimed to investigate if the geometry and spatial configuration of DCIS ducts in Hematoxylin-Eosin (H&E) stained tissue sections are related to the risk of progression of DCIS to ipsilateral IBC (iIBC). Methods. We obtained data from a population-based cohort of women diagnosed with primary DCIS between 1989 and 2004 in the Netherlands, treated with breast conserving surgery (BCS) only and a median follow-up time of 12 years. A nested case-control study (n=689) was designed in which patients diagnosed with iIBC recurrences during follow-up were considered as “cases” (n= 226) and those with no subsequent iIBC as “controls” (n=463). The DCIS and stroma regions were digitally annotated on H&E-stained whole slide images (WSIs) by a pathologist as ground truth for the deep learning neural network of HALO AI module (IndicaLabs). We developed a computational pipeline to automatically detect and measure stroma areas, DCIS ducts, and the nucleus of their cells. We validated the accuracy of DCIS detection in H&Es WSIs from an external study, in which DCIS regions were digitally annotated by an independent pathologist (Translational Breast Cancer Research Consortium, TBCRC). We classified cases and controls according to morphological measurements using logistic ridge regression with double-loop cross-validation, followed by hierarchical clustering. The risk of subsequent iIBC after primary DCIS diagnosis was evaluated by multivariate Cox proportional hazards models. Results. The accuracy of DCIS detection performed by the computational pipeline was compared with pathologist annotations in 20 slides from the TBCRC study. Results showed a satisfactory DCIS overlap area agreement of 0.76 (0.68 – 0.83). We applied the DCIS computational pipeline on the case-control series. We obtained 15 morphological measurements for each DCIS duct, such as duct area, cell density, distance between ducts, average nucleus area, etc. We calculated 8 distribution parameters from each measurement in each WSI, including median and range. After leaving out redundant variables, 55 unique morphometric variables were obtained, representing the heterogeneity of DCIS ducts per WSI. The c lassifier revealed a median area-under the curve (AUC) of 0.66 (0.55-0.77) to predict 5-years free of iIBC, 0.59 (0.50-0.67) to predict 10-years and 0.60 (0.52-0.68) to predict 15-years. The 30 variables with the highest association with outcome were used to build four morphometric signatures. Signature number 1, which is characterized by lesions with small-sized ducts, a lower number of cells and a lower DCIS/stroma area ratio, showed a significant lower risk of developing iIBC compared to the other three signatures in a multivariate Cox regression model including grade, ER, COX-2 and HER2 expression: HR = 0.56 (0.28-0.78 95%CI). Conclusion. We developed a computational pipeline able to detect and measure DCIS ducts in H&E WSIs with high accuracy and reproducibility. DCIS lesions presenting the morphometric signature of small-sized DCIS ducts have a very low chance to progress to invasive breast cancer. After successful validation, our morphometric method will serve as a robust and easy to implement biomarker for de-escalation strategies in DCIS, and as such, could limit unnecessary overtreatment in the near future. Citation Format: Marcelo Sobral-Leite, Simon Castillo, Shiva Vonk, Xenia Melillo, Noomie Lam, Brandi de Bruijn, Yeman Hagos, Joyce Sanders, Mathilde Almekinders, Lindy Visser, Emilie Groen, Carolien Van der Borden, Petra Kristel, Ercan Caner, Leyla Azarang, Yinyin Yuan, Renee Menezes, Esther Lips, Jelle Wesseling, Grand Challenge PRECISION Consortium. Morphometric signature identifies ductal carcinoma in situ of the breast with low risk of progression to invasive breast cancer [abstract]. In: Proceedings of the 2023 San Antonio Breast Cancer Symposium; 2023 Dec 5-9; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2024;84(9 Suppl):Abstract nr PS03-07.

Selecting regions of interest (ROIs) in whole-slide histology images (WSIs) is a crucial step for spatial molecular profiling. As a general practice, pathologists manually select ROIs within each WSI based on morphological tumor markers to guide spatial profiling, which can be inconsistent and subjective. To enhance reproducibility and avoid inter-pathologist variability, we introduce a novel immune-guided end-to-end pipeline to automate the ROI selection in multiplex immunofluorescence (mIF) WSIs stained with three cell markers (Syto13, CD45, PanCK). First, we estimate immune infiltration (CD45 ^+ expression) scores at the grid level in each WSI. Then, we incorporate the Pathology Language and Image Pre-Training (PLIP) foundational model to extract features from each grid and further select a subset of grids representative of the whole slide that comparatively matches pathologists’ assessment. Further, we implement state-of-the-art detection models for ROI detection in each grid, incorporating learning from pathologists’ ROI selection. Our study shows a significant correlation between our automated method and pathologists’ ROI selection across five different types of carcinomas, as evidenced by a significant Spearman’s correlation coefficient (> 0.785, p < 0.001), substantial inter-rater agreement (Cohen’s κ > 0.671), and the ability to replicate the ROI selection made by independent pathologists with excellent average performance (0.968 precision and 0.991 mean average precision at a 0.5 intersection-over-union). By minimizing manual intervention, our solution provides a flexible framework that potentially adapts to various markers, thus enhancing the efficiency and accuracy of digital pathology analyses.

Accurate cell detection in multiplex immunofluorescence (mIF) is crucial for quantifying and analyzing the spatial distribution of complex cellular patterns within the tumor microenvironment. Despite its importance, cell detection in mIF is challenging, primarily due to difficulties obtaining comprehensive annotations. To address the challenge of limited and unevenly distributed annotations, we introduced a streamlined semi-supervised approach that effectively leveraged partially pathologist-annotated single-cell data in multiplexed images across different cancer types. We assessed three leading object detection models, Faster R-CNN, YOLOv5s, and YOLOv8s, with partially annotated data, selecting YOLOv8s for optimal performance. This model was subsequently used to generate pseudo labels, which enriched our dataset by adding more detected labels than the original partially annotated data, thus increasing its generalization and the comprehensiveness of cell detection. By fine-tuning the detector on the original dataset and the generated pseudo labels, we tested the refined model on five distinct cancer types using fully annotated data by pathologists. Our model achieved an average precision of 90.42%, recall of 85.09%, and an F1 Score of 84.75%, underscoring our semi-supervised model's robustness and effectiveness. This study contributes to analyzing multiplexed images from different cancer types at cellular resolution by introducing sophisticated object detection methodologies and setting a novel approach to effectively navigate the constraints of limited annotated data with semi-supervised learning.

Abstract Bone marrow trephine biopsy is crucial for the diagnosis of multiple myeloma. However, the complexity of bone marrow cellular, morphologic, and spatial architecture preserved in trephine samples hinders comprehensive evaluation. To dissect the diverse cellular communities and mosaic tissue habitats, we developed a superpixel-inspired deep learning method (MoSaicNet) that adapts to complex tissue architectures and a cell imbalance aware deep learning pipeline (AwareNet) to enable accurate detection and classification of rare cell types in multiplex immunohistochemistry images. MoSaicNet and AwareNet achieved an AUC of >0.98 for tissue and cellular classification on separate test datasets. Application of MoSaicNet and AwareNet enabled investigation of bone heterogeneity and thickness as well as spatial histology analysis of bone marrow trephine samples from monoclonal gammopathies of undetermined significance (MGUS) and from paired newly diagnosed and posttreatment multiple myeloma. The most significant difference between MGUS and newly diagnosed multiple myeloma (NDMM) samples was not related to cell density but to spatial heterogeneity, with reduced spatial proximity of BLIMP1+ tumor cells to CD8+ cells in MGUS compared with NDMM samples. Following treatment of patients with multiple myeloma, there was a reduction in the density of BLIMP1+ tumor cells, effector CD8+ T cells, and regulatory T cells, indicative of an altered immune microenvironment. Finally, bone heterogeneity decreased following treatment of patients with multiple myeloma. In summary, deep learning–based spatial mapping of bone marrow trephine biopsies can provide insights into the cellular topography of the myeloma marrow microenvironment and complement aspirate-based techniques. Significance: Spatial analysis of bone marrow trephine biopsies using histology, deep learning, and tailored algorithms reveals the bone marrow architectural heterogeneity and evolution during myeloma progression and treatment.

Density of immune T cells and plasma cells in MGUS, NDMM, and posttreatment samples. A–G, Box plots showing the difference in density of FOXP3+CD4+ (A), the density of CD8+ (B), the density of FOXP3−CD4+ (C), FOXP3+CD4+:FOXP3−CD4+ ratio (D), FOXP3+CD4+:CD8+ ratio (E), density of BLIMP1+ (F), and CD8+:BLIMP1+ ratio (G) between paired NDMM samples and posttreatment samples (n = 10 pairs). H–J, Box plot showing the difference in density of FOXP3+CD4+(H), the density of BLIMP1+ cells (I), and CD8+:BLIMP1+ cells (J) between MGUS and NDMM samples (n = 19). K and L, Sample images showing the reduction of the density of FOXP3+CD4+ and CD8+ cells (K) and BLIMP1+ cells (L) at posttreatment compared with paired NDMM samples. The cell density is presented per 1 mm2 tissue area.

PURPOSE:Tumor-infiltrating lymphocytes (TILs) have prognostic significance in several cancers, including breast cancer. Despite interest in combining radiation therapy with immunotherapy, little is known about the effect of radiation therapy itself on the tumor-immune microenvironment, including TILs. Here, we interrogated longitudinal dynamics of TILs and systemic lymphocytes in patient samples taken before, during, and after neoadjuvant radiation therapy (NART) from PRADA and Neo-RT breast clinical trials. METHODS AND MATERIALS:We manually scored stromal TILs (sTILs) from longitudinal tumor samples using standardized guidelines as well as deep learning-based scores at cell-level (cTIL) and cell- and tissue-level combination analyses (SuperTIL). In parallel, we interrogated absolute lymphocyte counts from routine blood tests at corresponding time points during treatment. Exploratory analyses studied the relationship between TILs and pathologic complete response (pCR) and long-term outcomes. RESULTS:Patients receiving NART experienced a significant and uniform decrease in sTILs that did not recover at the time of surgery (P < .0001). This lymphodepletive effect was also mirrored in peripheral blood. Our SuperTIL deep learning score showed good concordance with manual sTILs and importantly performed comparably to manual scores in predicting pCR from diagnostic biopsies. The analysis suggested an association between baseline sTILs and pCR, as well as sTILs at surgery and relapse, in patients receiving NART. CONCLUSIONS:This study provides novel insights into TIL dynamics in the context of NART in breast cancer and demonstrates the potential for artificial intelligence to assist routine pathology. We have identified trends that warrant further interrogation and have a bearing on future radioimmunotherapy trials.

Metastasis is a nonrandom process with varying degrees of organotropism-specific source-acceptor seeding. Understanding how patterns between source and acceptor tumors emerge remains a challenge in oncology. We hypothesize that organotropism results from the macronutrient niche of cells in source and acceptor organs. To test this, we constructed and analyzed a metastatic network based on 9303 records across 28 tissue types. We found that the topology of the network is nested and modular with scale-free degree distributions, reflecting organotropism along a specificity/generality continuum. The variation in topology is significantly explained by the matching of metastatic cells to their stoichiometric niche. Specifically, successful metastases are associated with higher phosphorus content in the acceptor compared to the source organ, due to metabolic constraints in proliferation crucial to the invasion of new tissues. We conclude that metastases are codetermined by processes at source and acceptor organs, where phosphorus content is a limiting factor orchestrating tumor ecology.

Characterizing the spatial structure of taxonomic and functional diversity (FD) of marine organisms across regional and latitudinal scales is essential for improving our understanding of the processes driving species richness and those that may constrain or enhance the set of species traits that define the functional structure of communities. Here, we present the functional diversity of coastal invertebrate macrofaunal species along the south‐eastern Pacific from 7°N to 56°S, describe spatial variation of species traits, and examine the relationship with environmental variables. For that, we defined the functional traits and distribution ranges of 2350 marine macroinvertebrates calculated eight metrics of FD. Random forest regression was applied to identify significant relationships between FD and six environmental variables. Finally, functional β‐turnover was estimated to detect alongshore shifts in functional structure and their coincidence with biogeographical domains. Our results show, in contrast with taxonomic richness that measures of trait differences, functional space and functional specialisation increase with latitude, while functional evenness exhibits a non‐linear shape, peaking at mid latitudes. Functional redundancy decreased significantly poleward, while indicators of vulnerability increase. In contrast to taxonomic richness, FD was tightly connected to variables indicative of stress and productivity, such as dissolved oxygen and nutrients. Sea surface temperature and coastal area best explained the increased FD redundancy and richness towards the tropics. The high spatial correlation between taxonomic and functional turnover suggests environmental filters play an important role in the functional structure of the seascape. Our findings suggest that processes favouring taxonomic richness are latitudinally divergent from those favouring functional diversity. Correlations with environmental variables suggest that increased sea surface temperature and measures of stability increase redundancy, while variations in dissolved oxygen and nutrients positively affect functional diversification. Moreover, the functional diversity patterns suggest low resilience of high latitude coastal ecosystems, which are heavily exploited and threatened by climate change, hence highlighting the urgent need for effective conservation policies.

Earth’s biosphere is currently undergoing drastic reorganisation as a consequence of the sixth mass extinction brought on by the Anthropocene. Impacts of local and regional extirpation of species have been demonstrated to propagate through the complex interaction networks they are part of, subsequently leading to secondary extinctions, exacerbating biodiversity loss. Contemporary ecological theory has developed several measures to analyse the structure and robustness of ecological networks under biodiversity loss. However, a toolbox for direct simulation and quantification of extinction cascades and the creation of novel interactions (i.e. rewiring) remains absent. Here, we present NetworkExtinction - a novel R package which we have developed to explore the propagation of species extinctions sequences through ecological networks as well as quantify the effects of rewiring potential in response to primary species extinctions. With NetworkExtinction we have integrated ecological theory and computational simulations to develop functionality with which users may analyze and visualize the structure and robustness of ecological networks. The core functions introduced with NetworkExtinction focus on simulations of sequential primary extinctions and associated secondary extinctions while allowing for user-specified secondary extinction thresholds and realisation of rewiring potential. With the package NetworkExtinction, users can estimate the robustness of ecological networks after performing species extinction routines based on several algorithms. Moreover, users can compare the number of simulated secondary extinctions against a null model of random extinctions. In-built visualizations enable graphing topological indices calculated by the deletion sequence functions after each simulation step. Finally, the user can define the degree distribution of the network by fitting different common distributions. Here, we illustrate the use of the package and its outputs by analyzing a Chilean coastal marine food web. NetworkExtinction is a compact and easy-to-use R package with which users can quantify changes in ecological network structure in response to different patterns of species loss, thresholds, and rewiring potential. Therefore, this package is particularly useful to evaluate ecosystem responses to anthropogenic and environmental perturbations that produce non-random species extinctions.

Cancers occur across species. Understanding what is consistent and varies across species can provide new insights into cancer initiation and evolution, with significant implications for animal welfare and wildlife conservation. We build a pan-species cancer digital pathology atlas (panspecies.ai) and conduct a pan-species study of computational comparative pathology using a supervised convolutional neural network algorithm trained on human samples. The artificial intelligence algorithm achieves high accuracy in measuring immune response through single-cell classification for two transmissible cancers (canine transmissible venereal tumour, 0.94; Tasmanian devil facial tumour disease, 0.88). In 18 other vertebrate species (mammalia = 11, reptilia = 4, aves = 2, and amphibia = 1), accuracy (range 0.57–0.94) is influenced by cell morphological similarity preserved across different taxonomic groups, tumour sites, and variations in the immune compartment. Furthermore, a spatial immune score based on artificial intelligence and spatial statistics is associated with prognosis in canine melanoma and prostate tumours. A metric, named morphospace overlap, is developed to guide veterinary pathologists towards rational deployment of this technology on new samples. This study provides the foundation and guidelines for transferring artificial intelligence technologies to veterinary pathology based on understanding of morphological conservation, which could vastly accelerate developments in veterinary medicine and comparative oncology.

Quiescence is a state of cell cycle arrest, allowing cancer cells to evade anti-proliferative cancer therapies. Quiescent cancer stem cells are thought to be responsible for treatment resistance in glioblastoma, an aggressive brain cancer with poor patient outcomes. However, the regulation of quiescence in glioblastoma cells involves a myriad of intrinsic and extrinsic mechanisms that are not fully understood. In this review, we synthesise the literature on quiescence regulatory mechanisms in the context of glioblastoma and propose an ecological perspective to stemness-like phenotypes anchored to the contemporary concepts of niche theory. From this perspective, the cell cycle regulation is multiscale and multidimensional, where the niche dimensions extend to extrinsic variables in the tumour microenvironment that shape cell fate. Within this conceptual framework and powered by ecological niche modelling, the discovery of microenvironmental variables related to hypoxia and mechanosignalling that modulate proliferative plasticity and intratumor immune activity may open new avenues for therapeutic targeting of emerging biological vulnerabilities in glioblastoma.

Glioblastoma (GBM) recurrence originates from invasive margin cells that escape surgical debulking, but to what extent these cells resemble their bulk counterparts remains unclear. Here, we generated three immunocompetent somatic GBM mouse models, driven by subtype-associated mutations, to compare matched bulk and margin cells. We find that, regardless of mutations, tumors converge on common sets of neural-like cellular states. However, bulk and margin have distinct biology. Injury-like programs associated with immune infiltration dominate in the bulk, leading to the generation of lowly proliferative injured neural progenitor-like cells (iNPCs). iNPCs account for a significant proportion of dormant GBM cells and are induced by interferon signaling within T cell niches. In contrast, developmental-like trajectories are favored within the immune-cold margin microenvironment resulting in differentiation toward invasive astrocyte-like cells. These findings suggest that the regional tumor microenvironment dominantly controls GBM cell fate and biological vulnerabilities identified in the bulk may not extend to the margin residuum.

Despite the important role of preclinical experiments to characterize tumor biology and molecular pathways, there are ongoing challenges to model the tumor microenvironment, specifically the dynamic interactions between tumor cells and immune infiltrates. Comprehensive models of host-tumor immune interactions will enhance the development of emerging treatment strategies, such as immunotherapies. Although in vitro and murine models are important for the early modelling of cancer and treatment-response mechanisms, comparative research studies involving veterinary oncology may bridge the translational pathway to human studies. The natural progression of several malignancies in animals exhibits similar pathogenesis to human cancers, and previous studies have shown a relevant and evaluable immune system. Veterinary oncologists working alongside oncologists and cancer researchers have the potential to advance discovery. Understanding the host-tumor-immune interactions can accelerate drug and biomarker discovery in a clinically relevant setting. This review presents discoveries in comparative immuno-oncology and implications to cancer therapy.

Tumour infiltrating lymphocytes (TIL) influence the prognosis of Ductal carcinoma in situ (DCIS). Currently, manual assessment of TIL by expert pathologists is considered a gold standard. However, there are issues with a shortage of expert pathologists and inter-observer variability. A reliable automated scoring method is yet to be developed due to the inherent complexity of DCIS duct morphology and the assessment strategy. We developed a new deep learning and spatial analysis pipeline to automatically score DCIS stromal TIL (AI-TIL) from 243 diagnostic haematoxylin and eosin-stained whole slide images from 127 patients. To automatically identify and segment DCIS ducts, we implemented a generative adversarial network. To identify lymphocytes, we used a pre-trained deep learning model. Our DCIS segmentation model achieved a dice overlap of 0.94 ( ±0.01 ) and the cell classifier model achieved 92 6.0 × 10^-7 , W = 0.3 mm) compared with smaller boundary (r = 0.23, p = 0.12, W = 0.03 mm). Using multivariate analysis, a low AI-TIL score was associated with an increased risk of recurrence independent of age, grade, estrogen receptor (ER) status, progesterone receptor (PR) status, and necrosis (hazard ratio = 0.14, 95

Nothing in cancer makes sense except in the light of evolution. Mel Graves (2018) BMC Biology 16 (22) The origin and fate of cancer cells touch every aspect of biology, from the genetic scale to the organism's multicellular organisation and its relationship with the environment. From cells and their microenvironment to their harbouring organism and its environment, tumour progression is, by all means, an evolutionary phenomenon. Nordling[1] was one of the first taking a step towards evolution in cancer, originating the notion of cancer onset as a multi-stage accumulation of mutations. That process entails transforming a cell from a healthy phenotype towards a cancer one resulting from a collection of mutations, which sooner or later co-emerge with the local microenvironment, reproducing the well-known cancer hallmarks.[2] How that complex process is triggered and evolves remains a puzzle. However, any approach should consider the mechanistic integration between evolutionary history and the myriad hallmarks of cancer plus the interaction between genes and the immediate tumour microenvironment as essential ingredients for evolution. In this issue, Lineweaver and colleagues[3] contribute to filling the gap. They propose a novel hypothesis that moves one step forward the atavistic model of cancer, that is, the idea that cancer in multicellular organisms resembles an atavistic cellular machinery characteristic of ancestral unicellular organisms. Their model, named the Serial Atavism Model, presents the idea that cancer progression is not a one-shot reversion towards a quasi-unicellular state of cells; instead, it emerges from a series of ordered steps that erode the multicellular organisation of metazoans. The authors invite us to embrace their disruptive and novel idea and to think of tumour progression as an ordered and regular phenomenon across species and cancer types. They hypothesise that the sequence of changes from a healthy phenotype to cancer and its further evolution may follow a similar trajectory but in a reverse direction relative to the transition from single-cell organisms to multicellularity. They also extend their model to reaching eukaryogenesis, oxidative phosphorylation and the transition to adaptive immunity. The new proposal made by Lineweaver et al. is thought-provoking, and it is worthy of attention and evaluation. However, a serial steps consideration such as theirs constrains the possible evolutionary trajectories towards hallmarks deeply encrypted in the evolutionary history. But, their perspective fires up new questions. For example, do cancer cells find short-cuts in their evolution towards hallmarks that, despite an ancient origin, present adaptive advantages in the local context, or are they prisoners of their evolutionary history? How does this hypothesis explain the emergence of tumour heterogeneity? And how is that linked to the dynamic local environment or local ecology?[4] Further efforts need to consider that the order of events (intra or extracellular) matters in cancer evolution.[5] Although the authors did not intend to answer these fundamental questions, their hypothesis and suggested evaluation could point to that direction deserving further testing; above all, because, through the understanding of the fundamental process of tumour progression, we move the frontier of the unknown to gain clarity to improve patients’ survival. This article comments on the hypothesis paper by Charles H. Lineweaver et al., https://doi.org/10.1002/bies.202000305 The author declares no conflict of interest.