Submit to this Journal Review for this Journal Propose a Special Issue

Article Menu

Share Help Cite Discuss in SciProfiles

Open AccessReview

Peer-Review Record

Pre-Adaptive States and Evolutionary Trajectories in Breast Cancer Drug Resistance: From Drug-Tolerant Persisters to Clonal Evolution

Cells 2026, 15(9), 756; https://doi.org/10.3390/cells15090756

by Hye Young Choi^1,2,3

, Mi Jung Park^1,2

, Seung-Jun Lee⁴, Jeongyun Hwang⁴, Ho-Cheol Choi²

and Young-Sool Hah^3,5,*

Reviewer 1: Anonymous

Reviewer 2: Anonymous

Cells 2026, 15(9), 756; https://doi.org/10.3390/cells15090756

Submission received: 30 March 2026 / Revised: 17 April 2026 / Accepted: 21 April 2026 / Published: 23 April 2026

(This article belongs to the Section Cell Signaling)

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

This is a well-written and comprehensive review on a highly relevant topic. The idea of integrating drug-tolerant persister (DTP) biology with clonal evolution into a “Resistance Continuum Model” is interesting and, overall, the manuscript does a good job of bringing together a large and complex body of literature. However, I think that the manuscript would benefit from some reorganization to remove redundancies and improve clarity.

Major Comments

In some instances, the manuscript presents the main ideas as more novel than they actually are. Some of the underlying concepts (e.g., DTPs as intermediates, evolutionary trajectories of resistance) have already been described in the literature. I would suggest slightly softening these claims.
Some concepts, such as the “epigenetic memory ratchet”, are very interesting but quite speculative. It should be clear in the text what is well supported by studies in the field from what is more hypothetical and the authors' own model.
The manuscript is quite long and dense in parts (particularly Sections 3 and 4), with some repetition of key DTP features. It would help to eliminate such repetitions and shorten the text so that it is easier to follow.
The figure legends do not cite original sources even though many elements (e.g., KDM5 activity) are taken from other studies. If some subfigures are adapted from published figures, this should be clearly stated and permission should be acquired.

Minor comments

The figures are informative but seem very "packed" . I would suggest simplifying them so that it is easier for the readers to capture the main ideas.
Although the review is presented as breast cancer–specific, several sections describe in detail evidence from other tumor types. These parts should be shortened and related breast cancer–specific studies should be highlighted.

Author Response

Reviewer 1

General comment: "This is a well-written and comprehensive review on a highly relevant topic... However, I think that the manuscript would benefit from some reorganization to remove redundancies and improve clarity."

We thank Reviewer 1 for the positive evaluation of our manuscript and for the constructive suggestions. We agree that the manuscript benefited from a more restrained framing, clearer separation of evidence tiers, a more concise organization, and greater emphasis on breast cancer. We have revised the manuscript accordingly, as detailed below.

Major Comment 1

Comment: In some instances, the manuscript presents the main ideas as more novel than they actually are. Some of the underlying concepts have already been described in the literature. I would suggest slightly softening these claims.

Response:

We thank the Reviewer for this important comment and fully agree that several concepts discussed in our manuscript build upon prior work in drug-tolerant persister biology, non-genetic adaptation, and evolutionary models of resistance. Our intention was not to imply that all individual elements of the proposed framework are entirely unprecedented, but rather to integrate them into a breast cancer-focused conceptual synthesis.

In response, we have revised the manuscript more broadly to moderate claims of novelty and to better acknowledge prior conceptual foundations. Specifically, we revised the Highlights, Abstract, Introduction, Figure 1 legend, Section 2.1, and the Conclusion to replace over-assertive language such as “unified model,” “complete resistance trajectory,” “novel strategy,” and “paradigm shift” with more measured wording. Across these sections, the Resistance Continuum is now framed as a breast cancer-focused conceptual framework or working scaffold for organizing current evidence, rather than as a wholly new or definitive model. We also clarified that the proposed trajectory is canonical but not universal, and that several aspects of the framework remain context-dependent and incompletely resolved.

In Section 2.1 specifically, we further revised the comparison with prior frameworks to explicitly acknowledge the conceptual contributions of Laplane and Maley [17], Marine et al. [18], and Pu et al. [20], and we replaced “five principal conceptual advances” with the more appropriate phrase “five conceptual extensions to prior frameworks.”

Changes:

Softened novelty-related language in the Highlights, Abstract, Introduction, Figure 1 legend, Section 2.1, and Conclusion
Replaced over-assertive phrasing such as “unified model,” “complete trajectory,” “novel strategy,” and “paradigm shift”
Reframed the Resistance Continuum as a breast cancer-focused conceptual framework / working scaffold
Clarified that the proposed trajectory is canonical but not universal
Revised Section 2.1 to explicitly acknowledge prior conceptual foundations and to replace “five principal conceptual advances” with “five conceptual extensions to prior frameworks”

Modified Manuscript Text (Highlight, Page 1, Line 18-40)

Before: Drug resistance in breast cancer progresses through a five-stage... by real-time circulating tumor DNA (ctDNA) monitoring of clonal evolution.

Revised:

What are the main findings?

We propose a breast cancer-focused Resistance Continuum as a conceptual framework linking treatment-naïve heterogeneity, pre-adaptive priming, reversible drug-tolerant persister (DTP) states, cycling persisters, and genetically stabilized resistant clones. Rather than implying a universally linear pathway, this framework is intended to organize a canonical resistance trajectory while allowing for subtype-specific and branched evolutionary routes.
Available evidence across breast cancer subtypes supports an important role for epigenetic and metabolic plasticity in sustaining DTP states. We further discuss epigenetic memory as an emerging, hypothesis-generating concept that may help explain how repeated episodes of persistence could facilitate later resistance in selected contexts.

What are the implications of the main findings?

Mapping candidate intervention points along the continuum highlights opportunities to intercept resistance, including preventing persister emergence, targeting DTP-associated vulnerabilities, and disrupting the transition to stable resistance. However, the strength of supporting evidence is uneven across subtypes and intervention classes, and several strategies remain preclinical or conceptual.
A technology–stage matrix identifies major evidence gaps in breast cancer, including limited single-cell epigenomic and spatial characterization of cycling persisters and limited tools for directly monitoring non-genetic persistence in patients. In this context, ctDNA is best viewed as a clinically useful monitor of clonal and genetic evolution, whereas direct tracking of non-genetic DTP states remains an important unresolved challenge.

Modified Manuscript Text (Abstract)

Before: Drug resistance remains the principal cause of treatment failure and cancer-related mortality in breast cancer, yet the conventional mutation-centric paradigm fails to explain clinically observed phenomena such as re-sensitization after drug holidays… This framework argues for a paradigm shift from reactive treatment switching to proactive resistance prevention in breast cancer.

Revised: Drug resistance is a major cause of treatment failure in breast cancer, yet mutation-centered models do not fully explain delayed resistance, reversible tolerance, or re-sensitization after treatment interruption. Here, we synthesize recent findings in drug-tolerant persister (DTP) biology, clonal evolution, and tumor ecosystem dynamics to propose a breast cancer-focused Resistance Continuum as a conceptual framework for organizing the transition from initial therapy to stable resistance across ER-positive, HER2-positive, and triple-negative disease. Here, we synthesize recent findings in drug-tolerant persister (DTP) biology, clonal evolution, and tumor ecosystem dynamics to propose a breast cancer-focused Resistance Continuum as a conceptual framework for organizing the transition from initial therapy to stable resistance across ER-positive, HER2-positive, and triple-negative disease. This framework describes a canonical, but not universal, trajectory spanning treatment-naïve heterogeneity, pre-adaptive priming, reversible DTP states, cycling persisters, and genetically stabilized resistant clones. We discuss how epigenetic and metabolic plasticity may sustain persistence, and we present epigenetic memory as an emerging hypothesis linking repeated non-genetic persistence to facilitated resistance in selected contexts. We also compare subtype-specific features of DTP biology, outline a multi-omics roadmap for interrogating the continuum, and highlight therapeutic opportunities for resistance interception. Overall, the Resistance Continuum is intended as a working scaffold to integrate current evidence and guide future mechanistic and translational studies.

Modified Manuscript Text (Introduction, Page 3, Line 135-140)

Before: In this review, we propose a unified "Resistance Continuum Model" that traces the complete trajectory from treatment-naïve cellular states to stable genetic resistance in breast cancer (Figure 1). We systematically examine (i) the pre-adaptive states that seed the persister cell pool prior to treatment; (ii) the molecular hallmarks of DTP cells across ER+, HER2+, and TNBC subtypes; (iii) the epigenetic and metabolic plasticity mechanisms that sustain the persister state and generate "epigenetic memory"; (iv) the transition from non-genetic persistence to genetically fixed resistance through clonal evolution; and (v) the tumor microenvironment interactions that shape persister cell fate.

Revised: In this Review, we present the Resistance Continuum as a breast cancer-focused conceptual framework for organizing the progression from treatment-naïve cellular diversity to stable drug resistance (Figure 1). Our goal is not to imply that all tumors traverse an identical, strictly linear sequence of states, but rather to synthesize evidence supporting a canonical trajectory that may vary in timing, branching structure, and molecular implementation across subtypes and treatment contexts.

Modified Manuscript Text (Figure 1 legend)

Before: Figure 1. The Resistance Continuum Model in breast cancer. A unified conceptual framework depicting the five-stage evolutionary trajectory of drug resistance...

Revised: Figure 1. The Resistance Continuum in breast cancer as a conceptual framework. A schematic representation of a canonical resistance trajectory linking five cellular states that may arise during breast cancer treatment: (I) treatment-naïve heterogeneity; (II) pre-DTP priming; (III) reversible drug-tolerant persister (DTP) state; (IV) cycling persister state; and (V) genetically stabilized resistant clone. This model is intended to organize current evidence into a temporally oriented framework and should not be interpreted as implying that all tumors follow a uniform or obligatory linear sequence. Depending on subtype, therapy class, and tumor ecosystem context, transitions may be branched, overlapping, reversible, or partially bypassed. Four parallel tracks illustrate candidate molecular dimensions associated with each state, including epigenetic remodeling, metabolic adaptation, clonal dynamics, and population behavior. The dashed red rectangle indicates a putative “window of vulnerability” between the emergence of non-genetic persistence and the dominance of stable genetic resistance. The temporal axis spans hours (Stage II) to months (Stage V). Gradient color coding from green (sensitive) through amber, orange, and purple to red (resistant) reflects the progressive acquisition of resistance features. Several transitions—particularly those linking persistence to later resistant states—remain incompletely resolved and should be interpreted as working hypotheses rather than uniformly established pathways.

Modified Manuscript Text (Section 2, page 4, Line 169-173)

Before: … we propose a unified Resistance Continuum Model comprising five stages (Figure 1).

Revised: ... we present the Resistance Continuum as a conceptual framework comprising five stages (Figure 1). This framework is intended to organize recurrent patterns observed across the literature and should be interpreted as a canonical, but non-universal, trajectory rather than a definitive sequence shared by all tumors.

Modified Manuscript Text (Section 2, Page 5, Line 214-221)

Before: This comparison highlights the five principal conceptual advances of the present model: the formally defined five-stage architecture with temporal dynamics; the epigenetic memory ratchet concept; the integrated epigenetic–metabolic circuit framework; breast cancer subtype-stratified analysis; and the technology–stage evidence gap matrix.

Revised: This comparison highlights the five conceptual extensions to prior frameworks offered by the present model: the explicit temporal staging architecture with hours-to-months dynamics, extending the two-phase tolerance-to-resistance trajectory articulated by Laplane and Maley [17]; the epigenetic memory ratchet hypothesis (Section 5.4); the integrated epigenetic–metabolic circuit framework, consolidating observations previously treated separately by Marine et al. [18] and Pu et al. [20]; breast cancer subtype-stratified analysis; and the technology–stage evidence gap matrix.

Modified Manuscript Text (Section 10.1, Page 34, Line 1395-1398 and 10.3, Line 1433-1435)

Before: In this review, we have proposed and systematically elaborated a unified Resistance Continuum Model for breast cancer drug resistance.

...

The Resistance Continuum Model argues for a paradigm shift toward proactive resistance prevention—intercepting the resistance trajectory before it reaches irreversibility.

Revised: In this review, we have outlined the Resistance Continuum as a breast cancer-focused framework for organizing evidence across treatment-naïve heterogeneity, pre-adaptive priming, reversible drug-tolerant persister (DTP) states, cycling persisters, and genetically stabilized resistant clones.

...

The framework developed here suggests a complementary perspective: in some settings, resistance may be more effectively addressed by identifying and intercepting earlier adaptive states before stable clonal dominance is established.

Major Comment 2

Comment: Some concepts, such as the "epigenetic memory ratchet", are very interesting but quite speculative. It should be clear what is well supported from what is more hypothetical.

Response:

We thank the Reviewer for this insightful comment and agree that the original manuscript did not always sufficiently distinguish between evidence-supported observations and hypothesis-generating concepts. We particularly agree that the “epigenetic memory ratchet” should be framed more cautiously.

In response, we revised the manuscript to explicitly present epigenetic memory as an emerging, hypothesis-generating concept rather than as an established general mechanism in breast cancer. We strengthened the discussion of limitations, including the current lack of longitudinal epigenomic datasets, incomplete causal validation, and reliance on indirect or cross-cancer evidence in some parts of the argument. We also revised the corresponding figure legends and translational discussion to clearly separate established observations from forward-looking conceptual proposals.

To make this distinction consistent throughout the manuscript, we revised the Highlights, Abstract, Section 5 opening paragraph, Figure 4 legend, Section 5.4 opening paragraph, Section 5.4.2, Section 5.4 closing statement, Section 9.4, and Figure 7 legend. Across these sections, we now distinguish between (i) observations directly supported by breast cancer data, (ii) plausible extensions supported by indirect evidence, and (iii) predictions of the ratchet model that remain untested.

Changes:

Reframed the epigenetic memory ratchet as an emerging / hypothesis-generating concept
Added clearer distinction between evidence-supported mechanisms and speculative model components
Expanded limitations regarding longitudinal validation, progressive accumulation, and causal evidence
Revised Highlights and Abstract to reduce overstatement
Revised Figure 4 and Figure 7 legends to reduce mechanistic and translational overstatement
Repositioned therapeutic “memory erasure” in Section 9.4 as a forward-looking experimental strategy rather than a clinically established approach

Modified Manuscript Text (Highlight, Page 1, Line 24-28)

Before: The DTP state is sustained by a self-reinforcing epigenetic–metabolic circuit in which metabolic intermediates (α-ketoglutarate, SAM, FAD) serve as cofactors for chromatin-modifying enzymes, while each passage through the DTP state deposits residual chromatin "scars" that progressively lower the barrier to future resistance (epigenetic memory ratchet).

Revised: Available evidence across breast cancer subtypes supports an important role for epigenetic and metabolic plasticity in sustaining DTP states. We further discuss epigenetic memory as an emerging, hypothesis-generating concept that may help explain how repeated episodes of persistence could facilitate later resistance in selected contexts.

Modified Manuscript Text (Abstract, Page 2, Line 54-57)

Before: We further introduce the concept of "epigenetic memory," whereby residual chromatin modifications from prior DTP episodes progressively lower the barrier to subsequent resistance.

Revised: We discuss how epigenetic and metabolic plasticity may sustain persistence, and we present epigenetic memory as an emerging hypothesis linking repeated non-genetic persistence to facilitated resistance in selected contexts.

Modified Manuscript Text (Section 5, Page 16, Line 647-650)

Before: We conclude by introducing the concept of “epigenetic memory,” which provides a mechanistic bridge between the non-genetic DTP state and the emergence of stably resistant clones.

Revised: We conclude by introducing epigenetic memory as an emerging, hypothesis-generating concept that may provide a mechanistic bridge between the non-genetic DTP state and the emergence of stably resistant clones.

Modified Manuscript Text (Figure 4 legend, Page 17, Line 652-660, 662-666)

Before: (Top) Drug exposure triggers two interconnected adaptive programs... creating a self-reinforcing regulatory circuit.

(Bottom) The concept of epigenetic memory: each passage through the DTP state deposits residual chromatin modifications (“scars”) that progressively lower the barrier to DTP re-entry...

Revised: (Top) Drug exposure triggers two interconnected adaptive programs... The “self-reinforcing regulatory circuit” framing, therefore, represents a mechanistic inference that requires direct functional validation in breast cancer DTP systems.

(Bottom) Epigenetic memory as an emerging hypothesis: prior passage through the DTP state may leave residual chromatin modifications that incompletely reset after drug withdrawal and could, in selected contexts, facilitate subsequent DTP re-entry or later resistance. The durability, causality, and clinical relevance of these changes remain to be established in longitudinal breast cancer models and patient-linked studies.

Modified Manuscript Text (Section 5.4, Page 20, Line 803-810)

Before: Perhaps the most consequential feature of the epigenetic dimension of the DTP state is its potential capacity to generate "epigenetic memory." Although the DTP state is operationally defined by its reversibility...

Revised: Perhaps the most consequential feature of the epigenetic dimension of the DTP state is its potential capacity to generate “epigenetic memory.” We note at the outset that the epigenetic memory ratchet concept presented in this section is a hypothesis-generating framework rather than an established feature of breast cancer resistance. Throughout this section, we distinguish between (i) observations directly supported by breast cancer data, (ii) plausible extensions supported by indirect or cross-cancer evidence, and (iii) predictions of the ratchet model that remain untested. Although the DTP state is operationally defined by its reversibility...

Modified Manuscript Text (Section 5.4.2, Page 21, Line 840-847)

Before: Building on the evidence for residual chromatin marks described above, we propose the "epigenetic memory ratchet" hypothesis...

Revised: Building on the evidence for residual chromatin marks described above, we propose the “epigenetic memory ratchet” hypothesis. We emphasize, however, that this model rests on several important evidentiary limitations: current support is derived largely from single-episode observations, much of the literature is indirect or cross-cancer, and causal links between retained chromatin marks and accelerated DTP re-entry remain to be demonstrated. With these limitations stated explicitly, the ratchet framework can be used to generate testable predictions about how prior DTP experience might influence future resistance trajectories.

Modified Manuscript Text (Section 5.4, Page 22, Line 909-911)

Before: [No equivalent sentence]

Revised: Taken together, the epigenetic memory ratchet should currently be used to generate mechanistic and translational hypotheses for future study rather than to inform near-term clinical decision-making.

Modified Manuscript Text (Section 9.4, Page 32, Line 1324-1328)

Before: The concept of epigenetic memory introduced in Section 5.4.3 opens a novel therapeutic frontier... ...locus-specific epigenetic erasure may become feasible within the next decade...

Revised: The concept of epigenetic memory raises the possibility that resistance prevention might eventually extend beyond eliminating persister cells to modulating the long-term chromatin consequences of prior drug exposure. At present, however, this idea should be regarded as a forward-looking experimental hypothesis rather than a clinically established therapeutic strategy.

Modified Manuscript Text (Figure 7 legend, Page 31, Line 1244-1247)

Before: Strategy 4 (red): A novel conceptual strategy targeting epigenetic memory erasure...

Revised: Strategy 4 (red) represents epigenetic memory modulation, a forward-looking conceptual strategy aimed at reducing the long-term persistence-favoring effects of prior drug exposure; this approach remains speculative and requires substantial mechanistic and translational validation.

Major Comment 3

Comment: The manuscript is quite long and dense in parts, with some repetition of key DTP features. It would help to eliminate such repetitions and shorten the text so that it is easier to follow.

Response:

We thank the Reviewer for this helpful suggestion and agree that the original manuscript became overly dense in some sections, particularly where common features of drug-tolerant persister (DTP) biology were reiterated across the discussion of pre-adaptive states and subtype-specific persister mechanisms.

In response, we substantially streamlined Sections 3 and 4 to improve readability and reduce redundancy. In Section 3, we shortened the discussion of transcriptional and metabolic pre-adaptation and condensed the synthesis paragraph on multi-layered heterogeneity so that these subsections focus more directly on the conceptual role of pre-adaptive states in seeding the persister pool, without repeatedly restating hallmark features of the fully established DTP state. In Section 4, we reduced repeated definitions of reversibility, quiescence, and plasticity, strengthened internal cross-referencing, and shifted more of the comparative burden to Table 2, so that shared versus subtype-specific mechanisms do not need to be fully re-explained in each subtype subsection.

We also inserted a short terminology clarification paragraph at the end of Section 4.1 to distinguish DTP cells, cycling persisters, and stably resistant clones more cleanly. Overall, these revisions were intended to make the manuscript more concise, easier to follow, and more clearly organized around stage-specific and subtype-specific distinctions rather than repeated restatement of shared DTP characteristics.

Changes:

Shortened and streamlined Sections 3 and 4
Condensed Section 3.3–3.5 to reduce repeated explanation of DTP hallmarks already elaborated later in the manuscript
Reduced repeated descriptions of DTP reversibility, quiescence, and plasticity in Section 4
Strengthened internal cross-referencing and shifted comparative discussion to Table 2
Added a short terminology clarification paragraph at the end of Section 4.1

Modified Manuscript Text (Section 3.3, Page 9, Line 363-370)

Before: Importantly, Shaffer et al. demonstrated in melanoma—a finding with strong conceptual parallels to breast cancer—that rare cells expressing high levels of resistance-associated markers prior to drug exposure were not genetically distinct but rather occupied transient transcriptional states that stochastically arose and dissipated within the population [27]. These transiently primed cells were disproportionately likely to survive drug treatment and enter the persister state. In TNBC specifically, Baudre et al. recently identified that the hallmark transcriptional features of the DTP state—including high expression of basal keratins and activation of AP-1/NF-κB/IRF-STAT networks—could be detected at low levels in a fraction of pre-treatment cells, suggesting that the persister program is not entirely induced by therapy but is partially pre-encoded in the transcriptomic landscape [23]. Moreover, computational models of non-genetic heterogeneity have captured the emergence of reversible drug resistance in ER+ breast cancer cells [70], further supporting the bet-hedging interpretation.

Revised: In breast cancer, single-cell and computational studies support the view that pre-treatment transcriptional heterogeneity contributes to later persister emergence. In TNBC, Baudre et al. identified low-level pre-treatment expression of core persister-associated programs, including basal keratins and AP-1/NF-κB/IRF-STAT signaling [23]. In ER+ disease, modeling studies of non-genetic heterogeneity likewise support a bet-hedging interpretation of reversible drug tolerance [70]. Pan-cancer observations, including the melanoma study by Shaffer et al. [27], provide additional conceptual support for this framework.

Modified Manuscript Text (Section 3.4, Page 9-10, Line 372-381)

Before: Metabolic heterogeneity represents a third dimension of pre-adaptation. While many rapidly proliferating breast cancer cells rely predominantly on aerobic glycolysis (the Warburg effect), a subpopulation of slow-cycling cells within treatment-naïve tumors exhibits elevated mitochondrial oxidative phosphorylation (OXPHOS) and enhanced antioxidant capacity [44,71,72]. These metabolically distinct cells appear pre-conditioned for therapeutic survival, as they already display the OXPHOS-oriented metabolic profile and upregulated antioxidant defenses—including NRF2 targets, GPX4, and fatty acid oxidation pathways—that characterize the fully established DTP state (detailed in Section 5.2).

The functional significance of this metabolic preconditioning was underscored by Hangauer et al., who demonstrated that DTP cells across multiple cancer types, including breast cancer models, exhibit a selective vulnerability to GPX4 inhibition—a finding that would be inexplicable if persister cells did not rely on pre-established redox defense programs [43]. Furthermore, PINK1-mediated mitophagy has been shown to sustain OXPHOS activity and redox homeostasis in drug-tolerant breast cancer cells, with elevated PINK1 expression detectable in a subset of treatment-naïve cells [73]. These observations indicate that the metabolic infrastructure required for drug tolerance is, at least in part, assembled before therapy begins.

Revised: Metabolic heterogeneity represents a third dimension of pre-adaptation. Within treatment-naïve breast tumors, a subset of slow-cycling cells exhibits elevated mitochondrial oxidative phosphorylation (OXPHOS) and enhanced antioxidant capacity [44,71,72], suggesting that some cells are metabolically preconditioned for therapeutic survival before drug exposure. Functional support for this interpretation comes from studies showing that persister-like cells are selectively vulnerable to GPX4 inhibition [43] and from evidence that PINK1-mediated mitophagy sustains OXPHOS activity and redox homeostasis in drug-tolerant breast cancer cells [73]. Together, these findings suggest that part of the metabolic infrastructure required for drug tolerance may already be present before therapy begins.

Modified Manuscript Text (Section 3.5, Page 10, Line 383-389)

Before: The preceding subsections describe genetic, epigenetic, transcriptomic, and metabolic dimensions of pre-adaptation as though they were independent layers. These dimensions are deeply intertwined. A single cell may simultaneously harbor a resistance-predisposing copy-number alteration, a bivalently chromatinized stress-response locus, elevated expression of AP-1 transcription factors, and a metabolic profile tilted toward OXPHOS—with each layer compounding its probability of surviving therapeutic challenge [74]. While no single study has simultaneously profiled all four dimensions in the same individual breast cancer cells, the multi-layered nature of pre-adaptation is supported by converging evidence from studies addressing each layer independently. Multi-modal single-cell technologies that simultaneously profile the genome and transcriptome of the same cells, such as the recently developed wellDR-seq platform [75], are beginning to resolve these multi-layered pre-adaptive architectures in individual breast tumors.

Revised: The preceding subsections describe genetic, epigenetic, transcriptomic, and metabolic dimensions of pre-adaptation separately, but these layers are likely to coexist within the same cells. Although no single study has yet resolved all four dimensions simultaneously in individual breast cancer cells, converging evidence supports a multi-layered architecture in which several pre-adaptive features may compound the probability of therapeutic survival [74]. Emerging multi-modal single-cell platforms such as wellDR-seq [75] may help resolve these integrated pre-adaptive states more directly in future studies.

Modified Manuscript Text (Section 3.5, Page 10, Line 390-396)

Before: A key conceptual distinction emerging from this body of evidence is the difference between deterministic and stochastic models of pre-adaptation (Figure 2D). In the deterministic model, specific cells are "programmed" for survival by their fixed genetic or stable epigenetic features—for instance, a cell harboring a pre-existing PIK3CA amplification that confers resistance [76]. In the stochastic model, survival capacity arises from transient transcriptional or metabolic fluctuations that are not heritable and cannot be predicted from any single molecular feature [27,69]. The current evidence suggests that both models operate simultaneously in breast tumors: a genetically defined subclonal architecture provides a stable scaffold of pre-adaptation, while stochastic epigenetic and transcriptional fluctuations generate additional, transient windows of survival potential within and across these subclones. This dual architecture implies that strategies aimed at preventing resistance emergence must address both the stable and dynamic components of the pre-adaptive landscape—a challenge we revisit in Section 9.

Revised: A key conceptual distinction emerging from this body of evidence is the difference between deterministic and stochastic models of pre-adaptation (Figure 2D). In practice, current evidence suggests that both likely operate together in breast tumors: stable subclonal architecture may provide a scaffold of pre-adaptive risk, whereas stochastic epigenetic and transcriptional fluctuations may create additional transient windows of survival potential. This combined view helps explain why pre-adaptation is best understood as a layered landscape rather than as a single, discrete state.

Modified Manuscript Text (Section 4.1, Page 12, Line 477-482)

Before: [Did not exist]

Revised: To promote terminological clarity throughout the review, we use the following operational definitions. Drug-tolerant persister (DTP) cells are cells that survive initial therapeutic exposure through reversible, predominantly non-genetic mechanisms. Cycling persister cells are a subset of persister-derived cells that have re-entered the cell cycle while retaining drug tolerance under continued therapy. Stably resistant clones are cells that harbor fixed, heritable alterations that confer durable resistance across cell divisions.

Modified Manuscript Text (Section 4.3, Page 13, Line 517-523)

Before: HER2+ breast cancer, constituting approximately 15–20% of all breast cancers, is treated with anti-HER2 agents including trastuzumab, pertuzumab, and increasingly, antibody–drug conjugates (ADCs) such as trastuzumab deruxtecan (T-DXd) [94]. The DTP state in this context is shaped by the unique biology of HER2 signaling and the pharmacological properties of these targeted agents.

Revised: HER2+ breast cancer provides an important but still less fully resolved setting for the study of drug tolerance within the Resistance Continuum. Compared with ER+ and TNBC disease, where contemporary single-cell transcriptomic and epigenomic studies have more directly characterized persister-associated states, HER2+ DTP biology remains defined primarily by targeted mechanistic studies, bulk profiling, and preclinical models. Accordingly, the HER2+ framework presented here should be interpreted as an emerging working model rather than as a comparably mature map of persister biology.

Modified Manuscript Text (Section 4.3, Page 14, Line574-585)

Before: Despite these advances, a critical gap remains in HER2+ DTP biology: unlike the ER+ and TNBC settings, where systematic single-cell transcriptomic and epigenomic characterization of the persister state has been performed [23,24,87], the DTP phenotype in HER2+ breast cancer has not yet been defined at single-cell resolution using contemporary multi-omics approaches. Consequently, the transcriptional programs, chromatin states, and metabolic dependencies that distinguish HER2+ DTPs from resistant clones remain incompletely resolved. Furthermore, the mechanisms of ADC-specific tolerance—including the kinetics and regulation of lysosomal sequestration, the threshold of HER2 surface density required for effective ADC internalization, and the potential role of drug efflux transporters in payload resistance—have been studied primarily in bulk cell populations and preclinical models, with limited patient-derived validation [100]. Addressing this gap through single-cell profiling of HER2+ tumors before, during, and after anti-HER2 therapy—including ADC-treated specimens—represents one of the highest-priority experimental directions in the field (Table 3, Section 8).

Revised: Taken together, available data support the view that HER2+ tumors can enter reversible drug-tolerant states characterized by RTK switching, metabolic rewiring, and non-genetic adaptation under anti-HER2 pressure. However, several important features remain insufficiently resolved at single-cell resolution, including the transcriptional heterogeneity of HER2+ DTPs, the chromatin programs that distinguish persisters from later resistant clones, and the degree to which ADC tolerance shares versus diverges from conventional anti-HER2 persistence mechanisms. Proposed ADC-tolerance mechanisms—such as lysosomal sequestration, altered intracellular trafficking, and modulation of HER2 surface density—have thus far been defined largely in bulk systems and preclinical settings. For these reasons, HER2+ persistence should currently be regarded as a high-priority area for longitudinal, patient-linked single-cell and spatial profiling, rather than a fully resolved subtype within the continuum.

Modified Manuscript Text (Section 4.5, Page 15, Line 620-625)

Modified Manuscript Text (Section 4.5, Page 15, Line 630-636)

Before: However, critical subtype-specific differences exist (Table 2). In ER+ breast cancer, the DTP state is intimately linked to the dissociation of cell survival from ER-dependent transcription, mediated by KDM5B and alternative RTK signaling. In HER2+ breast cancer, RTK switching from HER2 to compensatory receptors (AXL, IGF-1R, MET) is the dominant adaptive mechanism, with ADC-specific tolerance pathways emerging as a new frontier. In TNBC, the persister program is dominated by basal keratin upregulation, the AP-1/NF-κB/IRF-STAT transcriptional axis, and a particularly pronounced reliance on P-gp-mediated detoxification.

Revised: However, critical subtype-specific differences exist (Table 2). In ER+ breast cancer, the DTP state is closely linked to dissociation of survival from ER-dependent transcription, mediated by KDM5B and alternative RTK signaling. In HER2+ breast cancer, RTK switching from HER2 to compensatory receptors emerges as a major adaptive mechanism, although the overall HER2+ persister landscape remains less well resolved than in ER+ and TNBC disease. In TNBC, the persister program is defined more prominently by basal keratin upregulation, the AP-1/NF-κB/IRF-STAT transcriptional axis, and marked stress-adaptive plasticity.

Major Comment 4

Comment: The figure legends do not cite original sources even though many elements (e.g., KDM5 activity) are taken from other studies. If some subfigures are adapted from published figures, this should be clearly stated and permission should be acquired.

Response:

We thank the Reviewer for raising this important point. We agree that the original figure legends did not sufficiently identify the literature sources underlying specific mechanistic elements shown in the schematics.

In response, we revised the figure legends to more clearly indicate when specific molecular features are based on prior published studies and to provide source citations where appropriate. We also reviewed the major conceptual figures to clarify whether they should be understood as direct adaptations of prior published material or as author-generated integrative schematics based on multiple cited studies. In the revised manuscript, we now explicitly state in the relevant legend that the figure is an author-generated integrative summary rather than a direct reproduction of any single prior figure.

Because Reviewer 1 specifically highlighted mechanistic elements such as KDM5 activity, we focused our revision most directly on Figure 4, where these features are discussed most explicitly. There, we added supporting citations for epigenetic and metabolic components and clarified that the diagram is a synthesized representation of published findings. If any figure panel is ultimately retained as a direct adaptation during final figure preparation, we will explicitly indicate this in the legend and obtain any required permission in accordance with the journal’s policy.

Changes:

Added source citations to figure legends where specific mechanistic elements are derived from published studies
Clarified that the revised schematic is an author-generated integrative summary
Revised the legend language to distinguish literature-supported features from author-level conceptual synthesis
Reviewed figures for possible adaptation status and aligned the legend language accordingly

Modified Manuscript Text (Figure 4 legend, Page 17)

Before: (Top) Drug exposure triggers two interconnected adaptive programs: epigenetic remodeling (left, indigo) involving KDM5A/B-mediated H3K4me3 loss... and metabolic reprogramming (right, teal) involving glycolysis-to-OXPHOS switching...

(Bottom) The concept of epigenetic memory: each passage through the DTP state deposits residual chromatin modifications (“scars”) that progressively lower the barrier to DTP re-entry, shorten the duration of therapeutic response, and accelerate the trajectory toward stable resistance...

Revised: (Top) Drug exposure triggers two interconnected adaptive programs: epigenetic remodeling (left, indigo), including KDM5A/B-associated H3K4me3 remodeling and chromatin changes linked to DTP formation [21,65,108], and metabolic reprogramming (right, teal), including OXPHOS, FAO, and ferroptosis-defense programs reported in persister models [42,43,118,119]. These two arms are hypothesized to be linked through a bidirectional crosstalk bridge in which metabolic intermediates serve as biochemically established cofactors for chromatin-modifying enzymes, while epigenetic modifications are thought to contribute to the expression of metabolic genes. The “self-reinforcing regulatory circuit” framing, therefore, represents a mechanistic inference that requires direct functional validation in breast cancer DTP systems. This schematic is an author-generated integrative summary based on published findings cited in the text and legend, rather than a direct reproduction of any single prior figure.

Minor Comment 1

Comment: The figures are informative but seem very “packed”. I would suggest simplifying them so that it is easier for the readers to capture the main ideas.

Response: We thank the Reviewer for this constructive suggestion and agree that several of the conceptual figures in the original submission were visually dense. In particular, we recognized that the combination of multiple mechanistic layers, extended in-figure annotations, and a highly compact layout reduced the accessibility of the main take-home messages.

In response, we revised the major conceptual figures to improve readability and visual hierarchy. Specifically, we simplified the presentation of Figures 1, 6, and 7 by reducing excessive in-figure text, consolidating labels where possible, and shifting some explanatory detail from the artwork into the legends. We also revised the corresponding legends so that each figure more clearly communicates its primary function: Figure 1 as a conceptual overview of the Resistance Continuum, Figure 6 as a technology–stage evidence map, and Figure 7 as a stage-informed, yet still exploratory, intervention framework.

Because visual simplification is only partially evident in the manuscript text, we note that the revised figure files have also been edited directly to reduce crowding and improve the clarity of the visual message.

Changes:

Simplified the visual presentation of Figures 1, 6, and 7
Reduced excessive in-figure text and consolidated labels where possible
Shifted explanatory detail from figure artwork into the legends
Revised the legends to make the main message of each figure easier to grasp
Clarified distinctions between evidence-supported versus more speculative components in the conceptual figures

Modified Manuscript Text (Figure 1 legend, Page 4, Line 149-160)

Before: Figure 1. The Resistance Continuum Model in breast cancer. A unified conceptual framework depicting the five-stage evolutionary trajectory of drug resistance…

Revised: Figure 1. The Resistance Continuum in breast cancer as a conceptual framework. A schematic representation of a canonical resistance trajectory linking five cellular states that may arise during breast cancer treatment: (I) treatment-naïve heterogeneity; (II) pre-DTP priming; (III) reversible drug-tolerant persister (DTP) state; (IV) cycling persister state; and (V) genetically stabilized resistant clone. This model is intended to organize current evidence into a temporally oriented framework and should not be interpreted as implying that all tumors follow a uniform or obligatorily linear sequence. Depending on subtype, therapy class, and tumor ecosystem context, transitions may be branched, overlapping, reversible, or partially bypassed. Four parallel tracks illustrate candidate molecular dimensions associated with each state, including epigenetic remodeling, metabolic adaptation, clonal dynamics, and population behavior. The dashed red rectangle indicates a putative “window of vulnerability” between the emergence of non-genetic persistence and the dominance of stable genetic resistance.

Modified Manuscript Text (Figure 6 legend, Page 28)

Before: Figure 6. Technology roadmap for dissecting the Resistance Continuum in breast cancer. Matrix mapping the applicability of eight single-cell and …

Revised: Figure 6. Technology roadmap for dissecting the Resistance Continuum in breast cancer. Matrix mapping the applicability of key single-cell and molecular profiling technologies across the five stages of the Resistance Continuum. Evidence levels are color-coded as stronger support, emerging support, or a major evidence gap. Notable gaps include the limited application of scCUT&Tag across most stages, the inability of ctDNA to directly detect non-genetic DTP states during the early continuum, and the paucity of spatial and single-cell data at the cycling persister stage. Column header colors correspond to the Resistance Continuum stages defined in Figure 1.

Modified Manuscript Text (Figure 7 legend, Page 30-31)

Before: Figure 7. Therapeutic strategies targeting the window of vulnerability in the Resistance Continuum. The top bar shows the five stages of ...

Revised: Figure 7. Candidate therapeutic intervention points along the Resistance Continuum.

The schematic maps four classes of intervention onto the proposed Resistance Continuum, with emphasis on the interval between reversible persistence and stable resistance. These strategies should be interpreted as stage-informed therapeutic hypotheses rather than uniformly validated clinical options. Strategy 1 (green) represents prevention of persister emergence through upfront combinations designed to suppress pre-adaptive entry programs. Strategy 2 (purple/teal/amber) represents elimination of established drug-tolerant persister (DTP) cells through epigenetic, metabolic, or dual-circuit targeting based largely on preclinical evidence. Strategy 3 (pink) represents efforts to delay or disrupt the transition toward resistant clonal dominance through adaptive scheduling and mutation-informed treatment adjustment, for which ctDNA is most relevant as a monitor of clonal evolution rather than a direct readout of DTP biology. Strategy 4 (red) represents epigenetic memory modulation, a forward-looking conceptual strategy aimed at reducing the long-term persistence-favoring effects of prior drug exposure; this approach remains speculative and requires substantial mechanistic and translational validation. The lower panel summarizes major barriers to clinical translation, including biomarker scarcity, subtype heterogeneity, uncertain timing of intervention, and limited tools for monitoring non-genetic persistence in patients.

Minor Comment 2

Comment: Although the review is presented as breast cancer–specific, several sections describe in detail evidence from other tumor types. These parts should be shortened and related breast cancer–specific studies should be highlighted.

Response: We thank the Reviewer for this valuable comment and agree that the breast cancer-specific focus should be more consistently maintained throughout the manuscript. While selected pan-cancer studies were originally included to provide conceptual context, we recognize that some sections gave these studies disproportionate space relative to direct breast cancer evidence.

In response, we shortened several cross-cancer discussions and revised the text to foreground breast cancer-specific studies wherever possible. Pan-cancer examples are now used more selectively, primarily when they provide essential mechanistic context in areas where direct breast cancer data remain limited. To make this shift more explicit, we revised the comparative synthesis in Section 4.5, the metabolic framing in Section 5.2, the mutagenesis discussion in Section 6.2, and the translational monitoring discussion in Section 6.5 so that breast cancer data are introduced first and non-breast cancer literature is used in a supporting rather than leading role.

Changes:

Shortened pan-cancer examples across multiple sections
Re-prioritized breast cancer-specific evidence in the main narrative
Retained non-breast cancer studies only when necessary for the mechanistic context
Strengthened breast cancer framing in Sections 4.5, 5.2, 6.2, and 6.5

Modified Manuscript Text (Section 4.5, Page 15, Line 620-628)

Before: Across the three major breast cancer subtypes, several core features of the DTP state are conserved: cell cycle arrest or quiescence, the glycolysis-to-OXPHOS metabolic switch and fatty acid oxidation upregulation (detailed in Section 5.2), epigenetic reprogramming involving histone demethylases (particularly the KDM5 family), upregulation of anti-oxidant defenses, and transcriptional activation of stress-response programs. These shared features suggest that the DTP state represents a convergent cellular strategy for surviving diverse therapeutic insults—a conclusion reinforced by the observation that similar persister hallmarks have been identified in non-breast malignancies, including lung, colorectal, and melanoma.

Revised: Across the three major breast cancer subtypes, several core features of the DTP state are conserved: cell cycle arrest or quiescence, recurrent metabolic rewiring, epigenetic reprogramming involving histone demethylases, upregulation of antioxidant defenses, and transcriptional activation of stress-response programs. For clarity and to avoid repetition across Sections 4.2–4.4, Table 2 serves as the primary reference for side-by-side comparison of shared versus subtype-specific DTP mechanisms. These shared features suggest that the DTP state represents a convergent cellular strategy for surviving diverse therapeutic insults. Pan-cancer parallels provide useful context, but the present discussion is centered on direct evidence from breast cancer whenever such evidence is available.

Modified Manuscript Text (Section 5.2, Page 18, Line 707-713)

Revised: In parallel with epigenetic remodeling, DTP cells undergo profound metabolic reprogramming. In breast cancer specifically, Echeverria et al. first demonstrated that chemotherapy-tolerant TNBC cells undergo a marked shift from glycolysis to OXPHOS [43]. We emphasize, however, that the metabolic phenotype of DTP cells is best understood through a dual framework of metabolic dependency and metabolic adaptability, and that OXPHOS dependency—though frequently observed—represents only one of several metabolic strategies that persister cells may employ.

Modified Manuscript Text (Section 6.2, Page 24, Line956-969)

Before: A landmark discovery linking DTP biology to clonal evolution was the identification of therapy-induced mutagenesis as an active process within persister cells. Isozaki et al. demonstrated that drug-tolerant persister cells in multiple cancer types, including breast cancer models, upregulate the APOBEC3A cytidine deaminase, generating a burst of C-to-T and C-to-G mutations at APOBEC signature motifs (TCW context) during the DTP state [30]. This therapy-induced APOBEC3A activity provides a direct mechanistic link between the non-genetic DTP state and the generation of genetic diversity from which resistant clones can be selected. The finding is particularly relevant to breast cancer, where APOBEC-signature mutations are among the most prevalent mutational processes and have been implicated in the evolution of resistance to multiple therapeutic modalities [119,120]. Furthermore, blocking genomic instability has been shown to prevent acquired resistance to MAPK inhibitor therapy in melanoma, suggesting that similar strategies could be applicable in breast cancer [49].

Revised: APOBEC-signature mutagenesis is especially relevant to breast cancer, where it represents one of the most prevalent mutational processes and has been implicated in clonal diversification and therapeutic resistance across multiple treatment contexts [119,120]. In this disease-specific context, a landmark discovery linking DTP biology to clonal evolution was the identification of therapy-induced mutagenesis as an active process within persister cells. Isozaki et al. demonstrated that drug-tolerant persister cells in multiple cancer types, including breast cancer models, upregulate the APOBEC3A cytidine deaminase, generating a burst of C-to-T and C-to-G mutations at APOBEC signature motifs (TCW context) during the DTP state [30]. This therapy-induced APOBEC3A activity provides a direct mechanistic link between the non-genetic DTP state and the generation of genetic diversity from which resistant clones can be selected. Furthermore, blocking genomic instability has been shown to prevent acquired resistance to MAPK inhibitor therapy in melanoma, suggesting that related anti-mutagenic strategies may also warrant investigation in breast cancer [49].

Modified Manuscript Text (Section 6.5, Page 25, Line1016-1021)

Before: 6.5. Liquid Biopsy and ctDNA as Real-Time Monitors of Clonal Evolution

Revised: 6.5. Liquid Biopsy and ctDNA as Clinical Monitors of Clonal and Genetic Evolution

Reviewer 2 Report

Comments and Suggestions for Authors

This manuscript presents an ambitious and well-structured “Resistance Continuum Model” describing the transition from reversible drug tolerance (DTP state) to stable genetic resistance in breast cancer. The review is comprehensive, timely, and conceptually integrative, effectively linking epigenetic remodeling, metabolic reprogramming, and clonal evolution.

The manuscript is generally well-written and supported by extensive literature. Notably, it introduces original and thought-provoking concepts, such as the “epigenetic memory ratchet”, and proposes testable experimental frameworks.

However, several conceptual overextensions, evidentiary gaps, and clarity issues should be addressed before publication.

- Overgeneralization of DTP Metabolic Phenotypes

While the manuscript acknowledges that the glycolysis-to-OXPHOS shift is not universal, it still implicitly frames OXPHOS dependency as a dominant paradigm.

The concept of metabolic flexibility is introduced but not fully integrated into the model.
The distinction between:
- metabolic dependency
- metabolic adaptability

remains underdeveloped.

-“Epigenetic Memory Ratchet” Hypothesis Requires Stronger Support

This is one of the most innovative aspects of the manuscript but also the most speculative.

Limitations include:

Reliance on indirect and cross-cancer evidence
Lack of longitudinal epigenomic data
Absence of demonstrated:
- progressive accumulation
- causal relationships

Moderate the tone and clearly define this as a hypothesis/framework.

-Causality Between Metabolism and Epigenetics

The bidirectional crosstalk between metabolic and epigenetic programs is well articulated but may overstate causality.

Much of the current evidence is correlative
Functional validation in DTP-specific contexts is limited

Include examples of causal perturbation studies, if available

-Limited Clinical Translation Framework

Although the therapeutic implications are intriguing, they remain somewhat abstract.

Concepts such as:

“window of vulnerability”
adaptive therapy
epigenetic memory erasure

are not sufficiently operationalized.

-Minor points

Some sentences are overly long and dense
Repeated use of emphasis terms (e.g., “critically,” “importantly”)
Key ideas (e.g., DTP reversibility, plasticity) are repeated multiple times
Terms such as:
- DTP
- cycling persisters
- resistant clones
  are sometimes used with overlapping meaning

Author Response

Reviewer 2

General comment: "This manuscript presents an ambitious and well-structured 'Resistance Continuum Model'... The review is comprehensive, timely, and conceptually integrative..."

We sincerely thank Reviewer 2 for the generous evaluation and incisive critiques, which substantially strengthened the rigor, balance, and translational clarity of the manuscript.

Comment 1: Overgeneralization of DTP Metabolic Phenotypes

Comment: While the manuscript acknowledges that the glycolysis-to-OXPHOS shift is not universal, it still implicitly frames OXPHOS dependency as a dominant paradigm. The concept of metabolic flexibility is introduced but not fully integrated into the model. The distinction between metabolic dependency and metabolic adaptability remains underdeveloped.

Response:

We thank the Reviewer for this important observation and agree that the original narrative implicitly privileged OXPHOS as the canonical DTP metabolic phenotype without sufficiently distinguishing metabolic dependency from metabolic adaptability.

In response, we revised the metabolic discussion to explicitly integrate this dual framework. We now define metabolic dependency as state-specific reliance on a pathway that is functionally non-redundant in a given context, whereas metabolic adaptability refers to the capacity of DTP cells to rewire metabolism dynamically in response to treatment, nutrient conditions, or microenvironmental constraints. We also revised the framing of subtype comparisons and therapeutic implications to make clear that OXPHOS is an important and recurrent DTP-associated phenotype, but not a universal or subtype-invariant metabolic target.

To implement this more consistently, we revised the opening of Section 5.2, added a new subsection (Section 5.2.1), refined the shared-metabolism wording in the comparative text for Figure 3 and Table 2, and added a short translational clarification in Section 9.2 to carry the dependency-versus-adaptability distinction into therapeutic interpretation.

Changes:

Revised Section 5.2 opening to foreground breast cancer-specific metabolic evidence and to reduce OXPHOS overgeneralization
Added Section 5.2.1 (new): Metabolic Dependency versus Metabolic Adaptability: A Dual Framework
Revised comparative subtype framing in Section 4.5 / Table 2-linked discussion to avoid implying universal OXPHOS dependence
Revised Figure 3 legend to clarify that recurrent metabolic rewiring does not imply a single universal metabolic phenotype
Added a short translational clarification in Section 9.2 to distinguish dependency-dominant versus adaptability-dominant metabolic targeting contexts

Modified Manuscript Text (Section 5.2, Page 18, Line 707-713)

Before: In parallel with epigenetic remodeling, DTP cells undergo profound metabolic reprogramming. The dominant metabolic shift is from aerobic glycolysis—the hallmark metabolic mode of rapidly proliferating cancer cells—to mitochondrial oxidative phosphorylation (OXPHOS). This glycolysis-to-OXPHOS switch has been documented in DTP cells across breast cancer subtypes and therapeutic contexts, and appears to serve dual functions: supporting the reduced proliferative rate of persister cells and providing the bioenergetic flexibility required for survival under therapeutic stress.

Modified Manuscript Text (Section 5.2.1, Page 19, Line 751-767)

Before: [Did not exist]

Revised: 5.2.1. Metabolic Dependency versus Metabolic Adaptability: A Dual Framework

The distinction between metabolic dependency and metabolic adaptability carries direct therapeutic implications. Here, we use metabolic dependency to refer to a state in which DTP survival relies on a pathway that is both essential and relatively non-redundant in a given therapeutic context, whereas metabolic adaptability refers to the capacity of persister cells to switch between fuel sources or energetic programs under changing selective pressures. This distinction matters because pathway inhibition may be effective in dependency-dominant settings but less durable in highly adaptable persister populations [119,120].

Breast cancer subtypes likely differ in the balance between these two dimensions. ER+ persisters may, in selected therapeutic contexts, exhibit greater reliance on OXPHOS-associated programs, particularly under endocrine therapy and CDK4/6 inhibition [43-45,73,115]. TNBC persisters may exhibit greater metabolic adaptability, including flexibility across OXPHOS, fatty acid oxidation, and antioxidant defense states [25,43,118]. HER2+ persister metabolism is currently supported by more limited evidence, with available data suggesting OXPHOS and FAO adaptation but with less comprehensive resolution than in ER+ and TNBC settings [99].

Modified Manuscript Text (Figure 3 legend)

Before: The central overlap identifies six core mechanisms conserved across all three subtypes: cell cycle arrest/quiescence, glycolysis-to-OXPHOS metabolic shift, KDM5 family epigenetic activity, NRF2/GPX4 antioxidant defense, fatty acid oxidation upregulation, and transcriptional plasticity...

Revised: (A) ER-positive DTP cell. Endocrine therapy and CDK4/6 inhibitor exposure promote a persist-er-associated state. Labeled arrows summarize representative mechanisms associated with ER+ persistence, including KDM5B-mediated H3K4me3 remodeling at estrogen-responsive loci (KDM5B↑), CDK2 upregulation and Cyclin E1 amplification enabling bypass of the RB1-dependent G1 checkpoint (CDK2/Cyclin E1), OXPHOS-associated metabolic adaptation (amber), and senescence bypass through PI3K/AKT-linked survival signaling. (B) HER2-positive DTP cell. Anti-HER2 monoclonal antibodies and antibody–drug conjugates (T-DXd) induce a persister-associated state characterized by receptor tyrosine kinase (RTK) switching to AXL, IGF-1R, and MET (RTK switch), lysosomal ADC sequestration limiting payload delivery (ADC sequest.), payload-specific resistance via TOP1 alterations (TOP1 mutation), and OXPHOS/glutamine-adaptive metabolism (amber). (C) Triple-negative breast cancer (TNBC) DTP cell. Chemotherapy or immune checkpoint inhibitor (ICI) exposure promotes a persistence program associated with basal keratin upregulation (KRT5/14/17↑), H3K27me3 remodeling at survival loci (H3K27me3 remodel), diapause-like G0/G1 arrest (G0 diapause), and FAO/detoxification-linked metabolic adaptation (amber). (D) Venn diagram synthesizing recurrent versus subtype-enriched DTP features. The central overlap identifies recurrently observed DTP features shared across subtypes, including quiescence, epigenetic/redox plasticity, and metabolic rewiring. Non-overlapping regions indicate subtype-enriched mechanisms, whereas pairwise overlaps highlight partially shared adaptive programs between two subtypes. Metabolic labels are intended to summarize major patterns of adaptive rewiring rather than to imply that any single metabolic phenotype, including OXPHOS dependence, is universal across all breast cancer persister states. Color coding: blue (ER+), teal (HER2+), coral (TNBC), amber (representative metabolic adaptations).

Modified Manuscript Text (Section 4.5, Page 15, Line 620-625)

Modified Manuscript Text (Section 9.2, Page 31-32, Line 1287-1291)

Before: [No equivalent sentence]

Revised: The therapeutic implications of metabolic targeting depend on whether a given persister population is dominated by metabolic dependency or by metabolic adaptability. OXPHOS inhibition may be effective in dependency-dominant settings, whereas highly adaptable DTP populations may require combination or sequential metabolic strategies to prevent compensatory rewiring.

Comment 2: Epigenetic Memory Ratchet Hypothesis

Comment: This is one of the most innovative aspects of the manuscript but also the most speculative.

Limitations include:

Reliance on indirect and cross-cancer evidence

Lack of longitudinal epigenomic data

Absence of demonstrated:

progressive accumulation

causal relationships

Moderate the tone and clearly define this as a hypothesis/framework.

Response:

We thank the Reviewer for this thoughtful and important comment and fully agree that the epigenetic memory ratchet is one of the most conceptually interesting yet also most speculative elements of the manuscript. Our original wording did not sufficiently distinguish the concept's attractive explanatory value from the still-limited evidentiary base supporting it.

In response, we revised the manuscript to consistently frame the epigenetic memory ratchet as an emerging, hypothesis-generating framework rather than an established mechanism of breast cancer resistance. We also made the underlying limitations explicit. Specifically, we now state more clearly that current support relies in part on indirect and cross-cancer evidence, that longitudinal epigenomic datasets in breast cancer are still lacking, and that the model’s predicted features—particularly progressive accumulation across repeated DTP episodes and causal linkage between residual chromatin marks and accelerated resistance—have not yet been directly demonstrated.

Changes:

Reframed the epigenetic memory ratchet as an emerging, hypothesis-generating framework
Added explicit statements regarding the current reliance on indirect / cross-cancer evidence
Added explicit statements regarding the lack of longitudinal epigenomic data
Clarified that progressive accumulation and causal linkage remain unproven
Revised Abstract, Highlights, Section 5 opening, Section 5.4, Figure 4 legend, Section 9.4, and Figure 7 legend to reduce overstatement
Repositioned therapeutic memory erasure as a forward-looking experimental concept rather than a clinically actionable strategy

Modified Manuscript Text (Highlight, Page 1, Line 24-28)

Modified Manuscript Text (Abstract, Page 2, Line 54-57)

Before: We further introduce the concept of "epigenetic memory," whereby residual chromatin modifications from prior DTP episodes progressively lower the barrier to subsequent resistance.

Modified Manuscript Text (Section 5, Page 16, Line 647-650)

Before: We conclude by introducing the concept of “epigenetic memory,” which provides a mechanistic bridge between the non-genetic DTP state and the emergence of stably resistant clones.

Modified Manuscript Text (Figure 4 legend, Page 17, Line 662-666)

Before: (Bottom) The concept of epigenetic memory: each passage through the DTP state deposits residual chromatin modifications (“scars”) that progressively lower the barrier to DTP re-entry, shorten the duration of therapeutic response, and accelerate the trajectory toward stable resistance...

Revised: Epigenetic memory as an emerging hypothesis: prior passage through the DTP state may leave residual chromatin modifications that incompletely reset after drug withdrawal and could, in selected contexts, facilitate subsequent DTP re-entry or later resistance. The durability, causality, and clinical relevance of these changes remain to be established in longitudinal breast cancer models and patient-linked studies.

Modified Manuscript Text (Section 5.4, Page 20, Line 803-810)

Modified Manuscript Text (Section 5.4.2, Page 21, Line 840-847)

Before: Building on the evidence for residual chromatin marks described above, we propose the "epigenetic memory ratchet" hypothesis: that each cycle of treatment and DTP-state traversal does not fully reset the chromatin landscape but instead deposits cumulative epigenetic "scars"...

Revised: Building on the evidence for residual chromatin marks described above, we propose the “epigenetic memory ratchet” hypothesis. We emphasize, however, that this model rests on several important evidentiary limitations: current support is largely based on single-episode observations, much of the literature is indirect or cross-cancer, and causal links between retained chromatin marks and accelerated DTP re-entry remain to be demonstrated. With these limitations stated explicitly, the ratchet framework can be used to generate testable predictions about how prior DTP experience might influence future resistance trajectories.

Modified Manuscript Text (Section 5.4,3, Page 22, Line 909-911)

Before: [No equivalent sentence]

Modified Manuscript Text (Section 9.4, Page 32, Line 1324-1328)

Before: The concept of epigenetic memory introduced in Section 5.4.3 opens a novel therapeutic frontier. If each passage through the DTP state deposits residual chromatin marks that lower the barrier to future resistance, then strategies aimed at erasing this memory could slow the progressive acceleration of resistance observed across successive lines of therapy.

Modified Manuscript Text (Section 9.4, Page 33, Line 1374-1380)

Before: While CRISPR-based epigenome editing is not yet clinically available... the rapid pace of CRISPR technology development suggests that locus-specific epigenetic erasure may become feasible within the next decade. In the interim, identifying specific memory loci... will be essential for guiding future targeted interventions.

Revised: Accordingly, therapeutic “memory erasure” is best framed as a research agenda for future breast cancer studies rather than an intervention ready for clinical implementation. Priority questions include whether persistence-associated chromatin states are causally linked to re-resistance, whether they can be selectively identified in patient-derived models, and whether partial resetting can be achieved without compromising lineage identity or increasing phenotypic plasticity elsewhere in the tumor ecosystem.

Modified Manuscript Text (Figure 7 legend, Page 31, Line 1244-1247)

Before: Strategy 4 (red): A novel conceptual strategy targeting epigenetic memory erasure through early-line DNA methylation inhibitors and HDAC inhibitors, aimed not at killing DTP cells directly but at preventing the progressive accumulation of chromatin scars that lower the barrier to resistance across successive treatment cycles.

Comment 3: Causality Between Metabolism and Epigenetics

Comment: The bidirectional crosstalk between metabolic and epigenetic programs is well articulated but may overstate causality. Much of the current evidence is correlative. Include examples of causal perturbation studies, if available.

Response:

We thank the Reviewer for this important point and fully agree that the original wording risked overstating causality. In the initial version, some parts of Section 5.3 and Figure 4 could be read as implying that the metabolism–epigenetics interface has already been demonstrated as a fully validated causal circuit in breast cancer DTP systems, whereas much of the current evidence is better interpreted as biochemical plausibility plus partial functional support.

In response, we revised Section 5.3 to distinguish more explicitly between biochemically established cofactor relationships and functionally demonstrated causal interactions in DTP contexts. Specifically, we moderated causal language in both the opening and the “Conversely” paragraph, replacing direct causal statements with more cautious phrasing such as “may be linked,” “are thought to contribute,” “may itself be,” and “if experimentally validated.” We also added a new paragraph summarizing the currently limited causal perturbation literature relevant to this interface, including evidence from Roesch et al. [109] in melanoma and breast cancer-relevant metabolic perturbation studies such as Luo et al. [70] and Fox et al. [118], while explicitly noting that these studies do not yet directly establish chromatin rewiring by metabolic perturbation in breast cancer DTP cells.

Finally, we revised the Figure 4 legend so that the “self-reinforcing regulatory circuit” is now described as a mechanistic inference requiring direct functional validation, rather than as an already established causal framework.

Changes:

Revised Section 5.3 opening paragraph to distinguish biochemical cofactor relationships from functional causal validation
Revised the Section 5.3 “Conversely” paragraph to moderate causal language
Added a new paragraph in Section 5.3 summarizing the currently limited causal perturbation literature
Revised Figure 4 legend so that the epigenetic–metabolic circuit is presented as a mechanistic inference rather than an established causal loop

Modified Manuscript Text (Section 5.3, Page 20, Line 769-780)

Before: A central insight emerging from recent DTP research is that epigenetic and metabolic reprogramming are not independent adaptations but are linked through bidirectional biochemical crosstalk (Figure 4, center bridge). Metabolites produced by the rewired DTP metabolism serve as essential cofactors and substrates for chromatin-modifying enzymes. α-ketoglutarate (α-KG), generated through the tricarboxylic acid (TCA) cycle, is a required co-substrate for the Jumonji-domain histone demethylases, including KDM5A/B and KDM6A/B; S-adenosylmethionine (SAM), derived from one-carbon metabolism, is the universal methyl donor for both DNA methyltransferases and histone methyltransferases; and flavin adenine dinucleotide (FAD) serves as a cofactor for the lysine-specific demethylase LSD1/KDM1A [72,121,122]. Thus, the metabolic rewiring that characterizes the DTP state directly modulates the epigenetic machinery that sustains it, creating a self-reinforcing regulatory circuit.

Revised: A central insight emerging from recent DTP research is that epigenetic and metabolic reprogramming are not independent adaptations but may be linked through bidirec-tional biochemical crosstalk (Figure 4, center bridge). Metabolites produced by rewired DTP metabolism serve as essential cofactors and substrates for chromatin-modifying enzymes. α-ketoglutarate (α-KG), generated through the tricarboxylic acid (TCA) cycle, is a required co-substrate for the Jumonji-domain histone demethylases, including KDM5A/B and KDM6A/B; S-adenosylmethionine (SAM), derived from one-carbon me-tabolism, is the universal methyl donor for both DNA methyltransferases and histone methyltransferases; and flavin adenine dinucleotide (FAD) serves as a cofactor for the lysine-specific demethylase LSD1/KDM1A [72,121,122]. These relationships are bio-chemically established, although their precise functional integration in breast cancer DTP systems remains incompletely defined.

Modified Manuscript Text (Section 5.3, Page 20, Line 781-790)

Before: Conversely, epigenetic modifications control the expression of metabolic genes. The opening of chromatin at promoters and enhancers of OXPHOS and FAO genes during the DTP transition is itself an epigenetically regulated event, driven by the same histone demethylase and acetyltransferase activities that reshape the broader persister transcriptome. This bidirectional coupling implies that perturbation of either the epigenetic or metabolic arm of the circuit could destabilize the entire persister state—a principle with direct therapeutic relevance that is explored further in Section 9.

Revised: Conversely, epigenetic modifications are thought to contribute to the expression of metabolic genes. The opening of chromatin at promoters and enhancers of OXPHOS and FAO genes during the DTP transition may itself be an epigenetically regulated event, plausibly driven by the same histone demethylase and acetyltransferase activities that reshape the broader persister transcriptome [21,65,108]. If experimentally validated, this bidirectional coupling would imply that perturbation of either the epigenetic or metabolic arm of the circuit could destabilize the persister state—a potential principle with thera-peutic relevance that is explored further in Section 9. At present, however, the circuit model remains a mechanistic hypothesis rather than a fully established causal framework in breast cancer DTP biology.

Modified Manuscript Text (Section 5.3, Page 20, Line791-801)

Before: [Did not exist]

Revised: The available causal perturbation studies examining this interface in DTP-related contexts remain sparse. In non-breast cancer systems, Roesch et al. demonstrated that blocking mitochondrial respiration in slow-cycling JARID1B-high melanoma cells dis-rupted the drug-tolerant phenotype [109], supporting the functional importance of metabolic state in persistence. In breast cancer-relevant settings, Luo et al. showed that manipulation of redox signaling altered the equilibrium of breast cancer stem cell states [70], and Fox et al. demonstrated that NRF2 activation governs redox and nucleotide metabolism during dormant tumor cell recurrence [118]. These studies support the bio-logical plausibility of metabolism-linked state regulation in breast cancer, but they do not directly interrogate epigenetic–metabolic coupling in DTP cells. Thus, current evidence supports a compelling mechanistic model, but not yet a fully validated causal circuit.

Modified Manuscript Text (Figure 4 legend, Page 17, Line 655-660)

Before: These two arms are linked through a bidirectional crosstalk bridge (center, amber) in which metabolic intermediates (α-ketoglutarate, S-adenosylmethionine, flavin adenine dinucleotide) serve as essential cofactors for chromatin-modifying enzymes, while epigenetic modifications control the expression of metabolic genes, creating a self-reinforcing regulatory circuit.

Revised: These two arms are hypothesized to be linked through a bidirectional crosstalk bridge (center, amber) in which metabolic intermediates serve as biochemically established cofactors for chromatin-modifying enzymes, while epigenetic modifications are thought to contribute to the expression of metabolic genes. The “self-reinforcing regulatory circuit” framing, therefore, represents a mechanistic inference that requires direct functional validation in breast cancer DTP systems.

Comment 4: Limited Clinical Translation Framework

Comment: Although the therapeutic implications are intriguing, they remain somewhat abstract.

Concepts such as:

“window of vulnerability”

adaptive therapy

epigenetic memory erasure

are not sufficiently operationalized.

Response:

We thank the Reviewer for this important comment and agree that the translational framework in the original manuscript was conceptually appealing but not yet sufficiently operationalized. We agree that terms such as the “window of vulnerability,” adaptive therapy, and epigenetic memory erasure require more explicit clarification of their biological meanings, what is currently measurable in patients, and which components remain preclinical or speculative.

In response, we revised the translational sections to provide more explicit operational framing. First, we revised Section 6.5 to define the current clinical role of ctDNA more precisely, emphasizing that ctDNA is informative for clonal and genetic evolution, but does not directly capture non-genetic DTP states. Second, we revised the opening of Section 8 to explicitly acknowledge the asymmetric performance of currently available monitoring tools across the continuum. Third, we revised the opening of Section 9 to define the “window of vulnerability” more cautiously as a putative interval whose timing and detectability are likely to vary across subtypes and treatment contexts. Fourth, we expanded the therapeutic sections to better distinguish strategies that are currently actionable, early translational, or still conceptual. We revised Section 9.3 to clarify the operational requirements for adaptive therapy approaches, and we revised Section 9.4 to frame epigenetic memory modulation as a forward-looking experimental hypothesis rather than a clinically established strategy. Finally, we revised Figure 7 and the concluding section to make the evidentiary maturity of these strategies more transparent.

Changes:

Revised Section 6.5 to clarify the actual clinical scope and limitations of ctDNA
Revised the opening of Section 8 to connect the technology map with asymmetric clinical monitoring capability across the continuum
Revised the opening of Section 9 to define the “window of vulnerability” more cautiously
Added an operational clarification to Section 9.3 regarding the practical requirements for adaptive or intermittent dosing strategies
Revised Section 9.4 to frame epigenetic memory modulation as a forward-looking experimental hypothesis rather than a clinically established therapeutic strategy
Revised Figure 7 legend to clarify that the mapped strategies are stage-informed hypotheses rather than uniformly validated clinical options
Revised Section 10.3 to present the translational outlook as a gradual, evidence-based shift rather than an already mature clinical framework

Modified Manuscript Text (Section 6.5, Page 25, Line 1016-1021)

Before: 6.5. Liquid Biopsy and ctDNA as Real-Time Monitors of Clonal Evolution

The clinical utility of the Resistance Continuum Model depends on the ability to monitor the trajectory in real time. Circulating tumor DNA (ctDNA) analysis has emerged as the most clinically accessible tool for tracking clonal evolution during breast cancer treatment [131,132]. In the PALOMA-3 trial, serial ctDNA profiling revealed that clonal evolution occurs frequently during CDK4/6 inhibitor therapy, with new driver mutations in PIK3CA and ESR1 emerging in both treatment arms, while RB1 mutations appeared exclusively in the palbociclib-treated arm [14].

Revised: 6.5. Liquid Biopsy and ctDNA as Clinical Monitors of Clonal and Genetic Evolution

The translational value of the Resistance Continuum depends in part on the ability to monitor resistance dynamics during therapy. In current clinical practice, circulating tumor DNA (ctDNA) is the most accessible tool for tracking clonal and genetic evolution in breast cancer, particularly the emergence and expansion of resistance-associated mutations under treatment pressure [131,132]. In the PALOMA-3 trial, serial ctDNA profiling revealed that clonal evolution occurs frequently during CDK4/6 inhibitor therapy, with new driver mutations in PIK3CA and ESR1 emerging in both treatment arms, while RB1 mutations appeared selectively in the palbociclib-treated arm [14].

Modified Manuscript Text (Section 6.5, Page 25, Line 1029-1035)

Before: [Did not exist]

Revised: At the same time, the interpretive scope of ctDNA should not be overstated. ctDNA is well-suited to monitoring genetically encoded resistance trajectories, but it does not directly capture non-genetic drug-tolerant persister (DTP) states, especially during early persistence phases, when chromatin remodeling, quiescence, and metabolic rewiring predominate in the absence of stable genomic alterations. Thus, within the Resistance Continuum, ctDNA is best positioned to inform the later transition from persistence toward clonal outgrowth rather than to serve as a direct biomarker of the DTP state itself.

Modified Manuscript Text (Section 8, Page 28, Line 1139-1150)

Before: The Resistance Continuum Model posits that drug resistance unfolds through five distinct cellular states, each governed by different molecular programs. Translating this conceptual framework into experimentally testable hypotheses requires technologies capable of capturing the genomic, epigenomic, transcriptomic, and spatial dimensions of the resistance trajectory at single-cell resolution. In this section, we map the state-of-the-art multi-omics toolkit onto each stage of the continuum, identify gaps in current technological coverage, and highlight emerging platforms of promise for breast cancer resistance research (Figure 6).

Revised: The Resistance Continuum Model posits that drug resistance unfolds through five distinct cellular states, each governed by different molecular programs. Translating this conceptual framework into experimentally testable hypotheses requires technologies capable of capturing the genomic, epigenomic, transcriptomic, and spatial dimensions of the resistance trajectory at single-cell resolution. In this section, we map the state-of-the-art multi-omics toolkit onto each stage of the continuum, identify gaps in current technological coverage, and highlight emerging platforms of promise for breast cancer resistance research (Figure 6; Table 3). Importantly, the technology map underscores that current clinical monitoring tools are asymmetrically informative across the continuum: ctDNA is comparatively powerful for tracking clonal and mutational evolution at later stages, whereas direct detection of pre-DTP and DTP states remains a major unmet need in breast cancer.

Modified Manuscript Text (Section 9, Page 29, Line 1225-1232)

Before: The Resistance Continuum Model identifies a “window of vulnerability”—the interval between DTP emergence and the fixation of stable genetic resistance—during which the resistance trajectory may still be reversed or disrupted (Section 2, Figure 1). This concept has important therapeutic implications: it suggests that the optimal time to intervene against resistance may not be after resistant clones have emerged, but rather during the transitional persister stages when the process remains epigenetically and metabolically plastic.

The Resistance Continuum suggests that there may be an interval between the emergence of non-genetic persistence and the dominance of stable resistant clones, during which resistance trajectories are more modifiable than at later stages. We refer to this interval as a putative “window of vulnerability,” while acknowledging that its timing, duration, and even existence are likely to vary across breast cancer subtypes, treatment modalities, and individual patients. In this section, we discuss candidate therapeutic strategies that may be relevant at different points along this interval, organized by their stage of intervention on the Resistance Continuum (Figure 7).

Modified Manuscript Text (Section 9.3, Page 32, Line 1317-1322)

Before: [No equivalent sentence]

Revised: For adaptive or intermittent dosing approaches to become clinically actionable within the Resistance Continuum framework, future studies will need to define operational parameters such as trigger biomarkers, ctDNA VAF thresholds, drug-holiday duration, and stopping rules for re-initiation. Without such operational anchors, adaptive therapy remains conceptually attractive but difficult to standardize prospectively in breast cancer.

Modified Manuscript Text (Section 9.4, Page 32, Line 1324-1328)

Revised: The concept of epigenetic memory raises the possibility that resistance prevention might eventually extend beyond eliminating persister cells to modulating the long-term chromatin consequences of prior drug exposure. At present, however, this idea should be regarded as a forward-looking experimental hypothesis, not as a clinically established therapeutic strategy.

Modified Manuscript Text (Section 9.4, Page 33, Line 1374-1380)

Before: While CRISPR-based epigenome editing is not yet clinically available... the rapid pace of CRISPR technology development suggests that locus-specific epigenetic erasure may become feasible within the next decade. In the interim, identifying specific memory loci and validating them in breast cancer models will be essential for guiding future targeted interventions.

Modified Manuscript Text (Figure 7 legend)

Before: Figure 7. Therapeutic strategies targeting the window of vulnerability in the Resistance Continuum. The top bar shows the five stages of the Resistance Continuum (colors as in Figure 1), with the window of vulnerability (dashed red rectangle) spanning Stages III–IV. Four intervention strategies are mapped to their respective points along the continuum. Strategy 1 (green): Upfront combination of standard-of-care therapy with persister-blocking agents (...). Strategy 2 (...). Strategy 3 (...). Strategy 4 (red): A novel conceptual strategy targeting epigenetic memory erasure (...). The bottom panel highlights key challenges in clinical translation.

Revised: Figure 7. Candidate therapeutic intervention points along the Resistance Continuum. The schematic maps four classes of intervention onto the proposed Resistance Continuum, with emphasis on the interval between reversible persistence and stable resistance. These strategies should be interpreted as stage-informed therapeutic hypotheses rather than uniformly validated clinical options. Strategy 3 (pink) represents efforts to delay or disrupt transition toward resistant clonal dominance through adaptive scheduling and mutation-informed treatment adjustment, for which ctDNA is most relevant as a monitor of clonal evolution rather than a direct readout of DTP biology. Strategy 4 (red) represents epigenetic memory modulation, a forward-looking conceptual strategy aimed at reducing the long-term persistence-favoring effects of prior drug exposure; this approach remains speculative and requires substantial mechanistic and translational validation. The lower panel summarizes major barriers to clinical translation, including biomarker scarcity, subtype heterogeneity, uncertain timing of intervention, and limited tools for monitoring non-genetic persistence in patients.

Modified Manuscript Text (Section 10.3, Page 34-35, Line 1431-1447)

Revised: The current clinical paradigm for managing drug resistance in breast cancer remains largely reactive: therapy is typically changed after resistant disease has become clinically apparent. The framework developed here suggests a complementary perspective: in some settings, resistance may be more effectively addressed by identifying and intercepting earlier adaptive states before stable clonal dominance is established. Realizing this goal will require more than conceptual reframing. It will depend on longitudinal patient-linked datasets, improved biomarkers for non-genetic persistence, clearer subtype-specific models, and careful integration of mechanistic and clinical evidence.

We emphasize, however, that these enabling capabilities vary in maturity. ctDNA-based monitoring of clonal and mutational evolution is already clinically actionable in selected settings, whereas persister-targeting strategies remain largely preclinical or early translational, and individualized trajectory-prediction frameworks are still aspirational. We therefore view the Resistance Continuum not as a finalized map of breast cancer resistance, but as a scaffold for future testing that may help guide a gradual shift from retrospective description of resistance to earlier, evidence-based intervention. This framework is intended to organize current evidence and identify testable translational priorities, rather than to define a uniform or immediately deployable clinical roadmap.

Comment 5: Minor Style and Terminology Points

Comment: Some sentences are overly long and dense. Repeated use of emphasis terms (e.g., “critically,” “importantly”). Key ideas (e.g., DTP reversibility, plasticity) are repeated multiple times. Terms such as DTP, cycling persisters, and resistant clones are sometimes used with overlapping meaning.

Response:

We thank the Reviewer for these helpful stylistic and terminological observations and agree that the original manuscript would benefit from a more disciplined presentation at the sentence, paragraph, and terminology levels.

In response, we performed a broad copy-editing and consistency pass throughout the manuscript. We shortened or split overly long sentences, reduced repeated emphasis terms, removed repeated descriptions of core DTP hallmarks where these had already been introduced elsewhere, and clarified the operational distinction between drug-tolerant persister (DTP) cells, cycling persisters, and stably resistant clones. To improve consistency, we also added a brief terminology clarification paragraph at the end of Section 4.1, where these concepts are first discussed in a comparative manner. In parallel, we streamlined several dense summary paragraphs—particularly in Section 4.5 and Section 10.3—to reduce repetition and improve readability.

Changes:

Performed a broad copy-editing pass to shorten long and dense sentences
Reduced repeated emphasis terms throughout the manuscript
Removed repeated descriptions of DTP reversibility, quiescence, and plasticity where these concepts had already been established
Added an operational terminology clarification paragraph in Section 4.1
Streamlined selected summary paragraphs, including Section 4.5 and Section 10.3

Modified Manuscript Text (Section 4.1, Page 12, Line 477-482)

Before: [Did not exist]

Modified Manuscript Text (Section 4.5, Page 15, Line 620-627)

Revised: Across the three major breast cancer subtypes, several core features of the DTP state are conserved: cell cycle arrest or quiescence, recurrent metabolic rewiring, epigenetic reprogramming involving histone demethylases, upregulation of antioxidant defenses, and transcriptional activation of stress-response programs. For clarity and to avoid repetition across Sections 4.2–4.4, Table 2 serves as the primary reference for side-by-side comparison of subtype-specific versus recurrent metabolic features. These shared features suggest that the DTP state represents a convergent cellular strategy for surviving diverse therapeutic insults.

Modified Manuscript Text (Section 10.3, Page 34-35, Line 1431-1438)

Before: The current clinical paradigm for managing drug resistance in breast cancer is fundamentally reactive: therapy is administered until resistance manifests as clinical progression, at which point treatment is switched to the next available agent. The Resistance Continuum Model argues for a paradigm shift toward proactive resistance prevention—intercepting the resistance trajectory before it reaches irreversibility. Realizing this vision will require the convergence of three capabilities: (i) real-time monitoring of the resistance trajectory using ctDNA and other minimally invasive biomarkers; (ii) a repertoire of persister-targeting therapeutic agents validated in subtype-stratified clinical trials; and (iii) computational frameworks capable of predicting individual resistance trajectories and recommending personalized intervention timing. The technological and conceptual foundations for each of these capabilities are now in place.

Modified Manuscript Text (Global style revision; multiple locations throughout the manuscript)

Before: Repeated emphasis terms such as “critically,” “importantly,” and repeated restatement of DTP reversibility and plasticity appeared across several sections.

Revised: These emphasis terms were reduced throughout the manuscript, and repeated definitional statements regarding DTP reversibility, cycling persisters, and subtype-shared plasticity were consolidated so that they were introduced once and then referenced more selectively in later sections.

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

Thank you for taking into consideration my suggestions and revising your manuscript accordingly. I recommend it for publication.

Reviewer 2 Report

Comments and Suggestions for Authors

The authors have satisfactorily addressed all the revisions requested during the peer review process. The changes implemented are clear, appropriate, and have significantly improved the overall quality of the manuscript. In its current form, the work meets the standards of Cells and is suitable for publication.

Article Menu

Pre-Adaptive States and Evolutionary Trajectories in Breast Cancer Drug Resistance: From Drug-Tolerant Persisters to Clonal Evolution

- Overgeneralization of DTP Metabolic Phenotypes

-“Epigenetic Memory Ratchet” Hypothesis Requires Stronger Support

-Causality Between Metabolism and Epigenetics

-Limited Clinical Translation Framework

-Minor points

Further Information

Guidelines

MDPI Initiatives

Follow MDPI