{"id":"9a7335ca-4979-45ec-8931-b4f4a78c4d56","arxiv_id":"2412.01285","paper_version":1,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":7,"one_line_summary":"Invasive breast cancer's fragmented cluster pattern is explained as a steady state of active nematic cell clusters breaking up and merging inside a passive nematic fiber matrix, and a model-derived activity measure correlates with patient survival.","lead":"Using computer simulations and the H&E slides of more than 2,000 breast cancer patients, this paper argues that the scattered, irregular clusters of invasive breast cancer are created by cancer cells moving together in liquid-crystal-like streams that stretch, split, and fuse inside the surrounding fibrous matrix.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Causal attribution to cluster motility is not secured: because the model sets ECM spontaneous order S0_0=0 and ECM activity ζ0=0, the simulated ECM order is flow-induced by construction, so the matching defect and anchoring statistics cannot rule out fibroblast-driven desmoplastic alignment.","rationale":"Reader's weakest_assumption correctly identifies the load-bearing point: the central conclusion depends on the claim that ECM nematic order and topological defects are generated by active flows from cancer clusters. In the simulations this premise is guaranteed, because Methods 'Simulation Parameters' sets ζ0=0 and S0_0=0, so no ECM order exists without cluster activity. The histological data are static; the exponential defect-distance decay and parallel anchoring are the key observations used to infer the mechanism. Cancer-associated fibroblasts are known to remodel and align collagen in desmoplastic tumors independently of cluster motility, producing similar local collagen organization. The healthy-tissue control (green curve, Fig. 5e) shows a flat defect distribution, but it does not rule out desmoplastic remodeling because that remodeling is cancer-specific and absent in healthy tissue. The survival marker max(ΔDC)/ξ would lose its mechanistic interpretation if the defects are pre-existing. The proposed simulation test—repeating the observables with an intrinsically ordered, passive ECM—can discriminate the two causal stories. This does not change the reader's CONDITIONAL verdict; it reinforces the conditionality: the paper should be accepted only if the test shows that the flow-generated pattern is necessary. The other issues (univariate survival analysis, power-law versus exponential ambiguity for shape, hand-set cutoffs) are secondary and addressable, but the causal identification is the load-bearing concern.","tokens_in":22035,"tokens_out":4082,"duration_ms":41076,"concrete_test":"Simulate the same cluster configurations with an intrinsically ordered ECM (S0_0 ≈ 0.5, matching the healthy-tissue value measured in the paper) and with the clusters kept passive (ζ1 = 0), while keeping all other parameters unchanged. If the exponential defect-distance decay (Fig. 5e) and the parallel anchoring (Fig. 5f) still emerge from the passive, pre-aligned ECM near cluster boundaries, then the histological agreement does not uniquely support the active-cluster mechanism; if those patterns vanish, the concern is settled in the paper's favor.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's central conclusion is that invasive breast cancer morphology is a self-organized steady state driven by cancer-cell motility, with the ECM acting as a passive nematic that is aligned and defected by active cluster flows. This causal claim rests on the premise that ECM nematic order and topological defects are generated by the active clusters. The model explicitly encodes this premise: in Methods 'Simulation Parameters', the ECM has zero spontaneous order (S0_0 = 0) and zero activity (ζ0 = 0), so all ECM alignment in the simulations is a consequence of the active clusters by construction. The patient histology, however, is a set of static snapshots; the two observables used to argue for the mechanism—the exponential decay of defect density away from cluster boundaries (Fig. 5e) and the parallel anchoring of ECM fibers at boundaries (Fig. 5f)—are equally consistent with pre-existing desmoplastic remodeling by cancer-associated fibroblasts, which is known to align collagen around invasive tumors and does not require cluster motility. The healthy-tissue control (Fig. 5e, green curve) is not discriminative because desmoplastic remodeling is cancer-specific and would also be absent in healthy tissue. If the ECM alignment is primarily fibroblast-driven, then the claimed 'interplay' of active and passive nematic dynamics is not established, and the prognostic marker max(ΔDC)/ξ loses its mechanistic interpretation as a readout of cluster activity.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper combines 2D continuum simulations of active nematic droplets (cancer cell clusters) embedded in a passive viscoelastic nematic ECM with histological analysis of 2012 invasive breast cancer patients, 87 DCIS patients, and 32 healthy tissue samples. The simulations produce a dynamical steady state of cluster breakup, motion, collision, and re-formation. The quantitative comparisons include a cluster area distribution P(A) ~ A^(-2.38), a shape-size correlation (Spearman R = 0.80 in histology vs. 0.71 in simulation), an exponential decay of ECM defect density with distance from cluster boundaries, and parallel ECM anchoring at cluster surfaces. The paper further uses the cluster shape index and a normalized maximum defect-cluster distance as model-derived activity proxies and reports that threshold-based risk groups stratify disease-free survival in a training/test split. The conclusion is that invasive breast cancer morphology is a mechanically self-organized steady state driven by cancer cell motility, with the ECM acting as a passive nematic whose alignment and defects are generated by the active clusters.","tokens_in":22244,"tokens_out":7130,"duration_ms":69401,"significance":"If the causal interpretation were secured, this would be a substantial contribution: it would connect active nematic physics to a clinically important histopathological signature using a large patient cohort, and it reports quantitative agreement across several independent observables (size exponent, shape-size correlation, defect-distance decay, anchoring angle, DCIS contrast). The study also has notable strengths: extensive histological data, mechanical parameter inputs from AFM and traction-force literature, five different activity sampling schemes (Supplementary Note 1.3), checks against two different initial conditions, and a train/test split for the survival claim. The main weakness is that the central causal attribution to cluster motility is partly built into the model by setting the ECM spontaneous order and activity to zero, and the robustness of the results to a key elastic parameter is not tested. The phenomenology is likely valuable even if the mechanistic direction remains underdetermined, but the paper currently overstates what the evidence can distinguish.","major_comments":[{"comment":"The central conclusion that ECM nematic order and topological defects are generated by active cancer cluster flows is not secured because the model encodes this premise. The ECM is set to have zero spontaneous order (S0_0 = 0) and zero activity (zeta_0 = 0), so all simulated ECM alignment, defects, and anchoring are flow-induced by the active clusters by construction. The patient histology provides only static snapshots, and the two observables used as mechanism evidence (exponential defect-distance decay in Fig. 5e and parallel anchoring in Fig. 5f) are equally consistent with pre-existing desmoplastic remodelling by cancer-associated fibroblasts. The healthy-tissue control is not discriminative because desmoplasia is cancer-specific and would be absent in healthy tissue; moreover, the authors' own comparison in Supplementary Note 1.4 and Fig. S8 shows that healthy ECM has comparable or even higher nematic order and defect densities at small length scales, which is difficult to reconcile with the assumption S0_0 = 0. A test that does not assume the answer would be, for example, comparing defect statistics and anchoring in regions of the same tumours with high versus low fibroblast/myofibroblast density, or repeating the simulations with finite S0_0 or zeta_0 and showing that the histological observables change in a falsifiable manner.","section":"Methods 'Simulation Parameters'; Fig. 5e-f; Supplementary Note 1.4"},{"comment":"The cancer cluster elastic modulus is set to E1 = 0.001 (5 Pa) in simulation units, but the AFM measurements reported in Supplementary Note 1.2 give a low-frequency shear modulus around G' ~ 200 Pa for dense cancer clusters. This is a roughly 40-fold discrepancy with no direct justification; the statement that cancer clusters fluidize the ECM in their vicinity does not explain why the cluster modulus itself should be as low as 5 Pa. The robustness study in Supplementary Note 1.3 varies the ECM elasticity, surface tension, area fraction, and activity, but it does not vary E1 or the cancer-nematic parameters (C1, K1_LC, S1_0, lambda_1, Gamma). Because cluster deformability and breakup control the size and shape distributions shown in Fig. 3, the quantitative agreement with histology could depend on tuning E1 downward. The authors should either use the independently measured cluster modulus or demonstrate that the distributions in Fig. 3 remain essentially unchanged over an E1 range spanning the AFM measurement.","section":"Methods 'Simulation Parameters'; Supplementary Note 1.2, Fig. S5a; Supplementary Note 1.3"},{"comment":"The prognostic claim is based on thresholds selected by maximizing the log-rank statistic in a 30% training subset of the same cohort, with no adjustment for established prognostic factors such as grade, nodal status, ER/PR status, or tumour size, and no external validation cohort. The test-set p-values are encouraging, but the threshold-search procedure is a form of selective inference on the training set, and univariate survival splits could be confounded by tumour stage or grade. The manuscript should either provide a multivariable analysis with the model-derived activity proxies as covariates or soften the conclusion that increased activity estimates are associated with poorer prognosis to a statement about association that remains to be confirmed in independent data.","section":"Fig. 6; Methods 'Survival analyses, prognostic observables'"}],"minor_comments":[{"comment":"The axes of the state diagram are labeled 'mechanical resistance of the ECM' and 'activity of the cluster'; please specify the underlying simulation parameters (e.g., E0 and zeta_1) and give their ranges in physical units.","section":"Fig. 4 caption"},{"comment":"The figure labels in Fig. S8 are duplicated: two panels are labeled 'b' while the third is 'c'. Please renumber the panels.","section":"Supplementary Note 1.4, Fig. S8"},{"comment":"The text states that the model uses surface tension parameters A_phi = K_phi = 1 and then reports a calibration from spheroid fusion, but the relationship between the LB-unit parameters and the resulting surface tension gamma is not stated numerically; adding the implied gamma value would help reproducibility.","section":"Methods, 'Simulation Parameters'"},{"comment":"The power-law fit range and the filtering of small clusters are described in the text and Methods, but the figure itself does not show the fit range or the lower cutoff; adding these to the figure would make the comparison easier to evaluate.","section":"Fig. 3a"},{"comment":"The phrase 'over 2,000 breast cancer patients' in the abstract is consistent with the Methods cohort size, but the main text reports 2012 patients in one place and 2170 in another; please reconcile these numbers.","section":"Abstract and Introduction"}],"recommendation":"major_revision","confidential_remarks":"The paper is a good fit for a soft-matter/active-matter or biophysics-oriented journal and the quantitative phenomenology is impressive. The main risk is that the causal narrative is underdetermined by the evidence as presented. I would encourage the editor to request the additional analysis on ECM spontaneous order/activity and the E1 calibration before acceptance; the survival analysis is secondary but should also be framed more cautiously."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Quick take: this is a serious, data-rich paper that couples active nematic droplet simulations to a very large breast cancer histology cohort. The quantitative matches are impressive: cluster area power law with exponent near -2.38, a strong size-shape correlation, exponential decay of ECM defect density away from cluster boundaries, parallel ECM anchoring, and a clean DCIS contrast. The survival stratification with a train/test split is methodologically sound as far as it goes. The paper deserves a serious referee.\n\nWhat's genuinely new is the scale and integration: more than 2,000 patients' H&E slides analyzed for nematic order and cluster morphology, compared to a two-phase active-passive nematic model with parameters anchored to independent AFM and traction force measurements. The activity-averaging robustness checks in Supplement Note 1.3 are real work, and the DCIS comparison is a good control: in situ tumors show no scaling, which fits the motility-driven breakup picture. Citations are appropriate; they build on the droplet fission/fusion literature and their own prior cell unjamming work.\n\nThe main soft spot is causal attribution. In the model, ECM spontaneous order and activity are set to zero, so all ECM alignment and defect generation is flow-induced by the active clusters by construction. The matching statistics therefore establish sufficiency—this mechanism can produce the observed patterns—but they do not establish that cluster motility is the actual driver in patients. Pre-existing desmoplastic remodeling by cancer-associated fibroblasts could produce similar static images, and the healthy-tissue control does not discriminate because desmoplasia is cancer-specific. I would not call this fatal; the model is plausible, but the abstract and conclusion overstate what is proven. The authors could soften the language, or better, test a distinguishing prediction, for example whether cluster-level activity proxies correlate with local collagen alignment at single-cluster resolution.\n\nThe other issues are addressable and less severe. The survival analysis is univariate; grade, nodal status, and receptor status are in the cohort, so multivariate adjustment and hazard ratios should be added. The power-law exponent is fit over hand-set area cutoffs, and the shape distribution cannot distinguish a power law from an exponential—the authors concede this. No code or data is released, and the cluster segmentation relies on unpublished prior work. None of these undermine the core comparison; they limit reproducibility.\n\nWho is this for: soft matter physicists working on active nematics, and cancer biologists willing to take a mechanics-based view of lesion morphology. I'd bring it to a reading group. I'd send it to peer review with a request for revision, not a desk reject.","headline":"Serious, high-volume active nematic/histology comparison that deserves refereeing, but the causal claim about cluster-driven ECM alignment is stronger than the model actually supports.","tokens_in":22979,"tokens_out":4965,"would_cite":true,"duration_ms":40753,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"Invasive breast cancer's cluster pattern arises from active self-organization driven by cell motility, with cluster sizes and shapes matching a power law P(A) ∼ A−2.38.","keywords":["active nematics","breast cancer","extracellular matrix","topological defects","cell motility","self-organization","power-law scaling","tumor histology"],"falsifier":"Treat an ex vivo breast tumour explant with a motility inhibitor (e.g., ROCK inhibitor) and follow cluster statistics and ECM defect distribution over time; the active-nematic mechanism predicts that cluster splitting slows, the cluster-size power law steepens or collapses toward a single large cluster, and the exponential defect halo around clusters fades. Alternatively, a spatial map of ECM alignment in patient tissue that shows strong nematic order with defects located far away from any cluster boundary would contradict the claim that cluster activity is the source of ECM order.","tokens_in":21672,"feed_emoji":"🔬","tokens_out":5565,"duration_ms":47134,"temperature":0.7,"pith_summary":"This paper argues that the patchwork of cancer cell clusters embedded in fibrous tissue that defines invasive breast cancer is not the result of progressive disorder, but a self-organized steady state produced by the motility of cancer cells. The authors simulate cell clusters as active nematic droplets that exert stresses on a surrounding passive nematic matrix representing the extracellular matrix (ECM), and show clusters continuously split, move, collide and merge. Histological images from more than 2,000 breast cancer patients match the simulation in quantitative detail: cluster sizes follow a power law P(A) ~ $A^{-2}$.38, larger clusters are more irregular in shape, ECM defects cluster near cluster boundaries, and ECM fibres align parallel to those boundaries. The same activity measures that drive the dynamics separate patients with good and poor disease-free survival. If the mechanism is right, tumour morphology itself carries a mechanical signature of how motile the cancer is.","feed_headline":"Tumor clusters move, split and re-form as a steady state","feed_subtitle":"Simulations + 2,000-patient histology trace tumor architecture to active nematic dynamics and a survival signal.","key_machinery":"The central object is a coupled two-phase liquid crystal model: cancer cell clusters are an active nematic (director-aligned motile cells producing active stress −ζQ), embedded in a passive nematic ECM that obeys Landau-de Gennes free energy with zero spontaneous order, so all ECM alignment is generated by flows from the clusters. Phase separation is described by a Cahn-Hilliard field with surface tension, and the ECM is a Kelvin-Voigt viscoelastic fluid. The mechanism that carries the argument is the dynamical balance between active droplet break-up and coalescence, which yields the steady-state scaling of cluster sizes and shapes and generates the characteristic defect distribution in the ECM.","core_discovery":"The paper claims that the characteristic structure of invasive breast cancer—many small, irregular clusters of cancer cells embedded in dense, aligned extracellular matrix—is a dynamic steady state maintained by active nematic forces rather than a passive consequence of dedifferentiation. Treating cancer clusters as active nematic inclusions and the surrounding ECM as a passive, viscoelastic nematic phase, the authors show that active stresses from cell motility stretch clusters until they overcome surface tension, causing them to split; the resulting fragments move, collide and fuse, so the population reaches a scale-free cluster-size distribution P(A) ~ $A^{-2}$.38 that matches histology. The same dynamical state produces the observed exponential decay of ECM topological-defect density with distance from cluster boundaries and the parallel anchoring of ECM fibres at cluster surfaces. As prognostic markers, the average cluster shape index and the normalized defect–cluster distance both separate patients by disease-free survival (log-rank p < 0.001 in the test cohort), connecting the physics directly to clinical outcome.","pith_inferences":["The analogy to active nematic droplets suggests that tumour clusters, like liquid crystal droplets, should exhibit size-selection behaviour that depends on ECM stiffness and activity; one could test this by measuring cluster-size distributions across patients with different stromal density, predicting steeper exponents in stiffer matrices.","The prognostic thresholds (1.62 and 0.81) were derived from a 30% training set; an independent prospective cohort would settle whether the thresholds are universal or patient-population specific.","If the mechanism extends beyond breast cancer, other invasive carcinomas with desmoplastic stroma (e.g., pancreatic ductal adenocarcinoma) may show similar power-law cluster-size distributions; the pancreatic precursor data cited in the paper already hints at this.","The model treats ECM as a passive nematic with zero spontaneous order; a natural extension is to include a fraction of pre-aligned collagen contributed by cancer-associated fibroblasts and ask how much pre-alignment can be tolerated before the active signature is masked."],"forward_implications":["Invasive tumour morphology is a mechanically self-organized steady state in which cluster fission and fusion set the size and shape statistics, so the architecture itself reports the level of cancer cell motility.","The power-law exponent α = −2.38 for cluster areas and the size–shape correlation (Spearman R = 0.80) provide quantitative standards for comparing any proposed model of invasion against patient histology.","Because proliferation-driven spreading is negligible for 99% of clusters (motility dominates), anti-migratory therapies that reduce active stress should shift the steady state toward larger, rounder clusters—the low-risk morphology—rather than merely preventing single-cell escape.","The same activity proxies (shape index above 1.62; normalized defect distance below 0.81) can be computed from routine H&E slides and separate patients by disease-free survival, making the physics directly usable as a prognostic readout.","The defect halo and parallel anchoring of ECM fibres are predictions of cluster-generated stress fields; measuring them in tissue gives a direct test of whether a lesion is in the active regime."],"supporting_citations":[{"why":"Supplies the measured cell motility velocities and the cluster detection algorithm used on histology.","marker":"[3]"},{"why":"Provides the active-nematic stress framework used to model cancer clusters.","marker":"[8]"},{"why":"Provides cell and nucleus shape as fluidity indicator and spheroid fusion timescale for surface-tension calibration.","marker":"[11]"},{"why":"Supplies measured ECM stiffness range used to set the ECM elastic modulus in simulations.","marker":"[13]"},{"why":"Provides traction stress magnitudes that set the activity scale ζ.","marker":"[15]"},{"why":"Shows spontaneous division and motility in active nematic droplets, the basis for cluster break-up in the model.","marker":"[16]"},{"why":"Supplies biphasic lyotropic active nematic droplet behaviour underlying fission/fusion dynamics.","marker":"[18]"},{"why":"Reports power-law size distributions in pancreatic precursor lesions, supporting power-law scaling as a general tumour feature.","marker":"[19]"},{"why":"Shows cluster boundaries give a selective advantage to new aggressive phenotypes, used to interpret why activity correlates with progression.","marker":"[28]"}],"fun_headline_variants":["Active nematic forces shape invasive breast cancer clusters","Tumor clusters split and fuse in active self-organized state","Cancer cell motion drives tumor cluster architecture","Nematic dynamics explain breast cancer cluster patterns","Tumor structure from active nematic self-organization"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The claim rests on the assumption that the alignment and defects of the extracellular matrix are generated by the cancer clusters' own active flows, rather than being pre-arranged by fibroblasts during desmoplastic remodeling; if the matrix is aligned independently of cluster motility, the same histology could arise without active nematic self-organization.","fun_headline_variants_meta":{"raw":{"variants":["Active nematic forces shape invasive breast cancer clusters","Tumor clusters split and fuse in active self-organized state","Cancer cell motion drives tumor cluster architecture","Nematic dynamics explain breast cancer cluster patterns","Tumor structure from active nematic self-organization"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00015,"raw_usage":{"total_tokens":1176,"prompt_tokens":901,"completion_tokens":275,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":517,"completion_tokens_details":{"reasoning_tokens":204}},"tokens_in":517,"tokens_out":275,"duration_ms":2914,"temperature":1.0,"reasoning_tokens":204,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T04:32:07.296505+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Treat an ex vivo breast tumour explant with a motility inhibitor (e.g., ROCK inhibitor) and follow cluster statistics and ECM defect distribution over time; the active-nematic mechanism predicts that cluster splitting slows, the cluster-size power law steepens or collapses toward a single large cluster, and the exponential defect halo around clusters fades. Alternatively, a spatial map of ECM alignment in patient tissue that shows strong nematic order with defects located far away from any cluster boundary would contradict the claim that cluster activity is the source of ECM order.","supporting_citations":[{"cited_title":"& author Oskarsson, T","cited_arxiv_id":null,"evidence_quote":"Provides cell and nucleus shape as fluidity indicator and spheroid fusion timescale for surface-tension calibration."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies measured ECM stiffness range used to set the ECM elastic modulus in simulations."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides traction stress magnitudes that set the activity scale ζ."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Shows spontaneous division and motility in active nematic droplets, the basis for cluster break-up in the model."},{"cited_title":"& author Ramaswamy, S","cited_arxiv_id":null,"evidence_quote":"Supplies biphasic lyotropic active nematic droplet behaviour underlying fission/fusion dynamics."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Reports power-law size distributions in pancreatic precursor lesions, supporting power-law scaling as a general tumour feature."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Shows cluster boundaries give a selective advantage to new aggressive phenotypes, used to interpret why activity correlates with progression."}],"review_version":1}