Dynamic Fitness Landscapes and Evolutionary Constraint in Prostate Cancer
Why Oscillating Environments Slow Tumour Adaptation
A Note on This Paper
This paper is a theoretical synthesis rather than a conventional literature review. It draws on evolutionary biology, cancer ecology, and clinical oncology to build an argument about why oscillating environments constrain tumour adaptation. References are provided selectively at the four points where specific external findings are foundational to the argument, the clonal evolution and metastatic seeding literature, the adaptive lag literature, the adaptive therapy clinical data, and the ecological disturbance framework, and are not provided for the paper's own synthesis and application of those concepts to the Quiet Biology framework. That synthesis is the contribution of this paper, and it stands or falls on its own coherence.
This paper sits alongside the Chronic Activation vs Oscillation paper in the series. Where that paper argues the oscillation case from the cellular signalling level, why pulsatile signals and chronic signals produce different cellular outcomes, this paper argues the same case from the evolutionary population level: why oscillating environments prevent tumour populations from reaching stable adaptive optima. The two arguments are complementary and mutually reinforcing.
Cancer progression is increasingly understood as an evolutionary process occurring within a dynamic ecological environment. Tumour populations adapt to selective pressures imposed by the host microenvironment and by therapeutic interventions. Traditional cancer treatments often apply continuous maximal pressure with the objective of eliminating malignant cells. While this approach can produce rapid short-term responses, it frequently accelerates evolutionary adaptation by creating selective environments in which resistant phenotypes become dominant. This paper explores the implications of dynamic fitness landscapes for prostate cancer management. It argues that oscillating environmental conditions, whether arising from therapeutic cycling, metabolic modulation, or physiological perturbations such as exercise, may constrain tumour adaptation by preventing stable evolutionary optimisation. An important qualification runs throughout: oscillating environments do not inevitably produce adaptive lag. Under some conditions, they may instead favour generalist or highly plastic phenotypes. The paper addresses both outcomes and specifies why the framework's oscillation parameters are hypothesised to favour lag over generalism. The paper concludes with a set of predictions and experimental questions that would allow the framework to be tested or falsified.
01Cancer as Evolution in a Fitness Landscape
Evolutionary biology describes adaptation using the concept of a fitness landscape: a theoretical representation of how different phenotypes perform within a given environment. In this landscape, each point corresponds to a specific biological state, and elevation corresponds to reproductive success or evolutionary fitness. Populations tend to evolve toward fitness peaks, phenotypes that maximise survival and reproduction under prevailing environmental conditions. Importantly, these peaks are not intrinsic properties of the organism. They are defined by the environment in which the organism exists.[1]
In cancer, tumour cells occupy positions within a similar evolutionary landscape. Their fitness is determined by interactions with nutrient availability, oxygen gradients, immune surveillance, stromal regulation, endocrine signalling, and therapeutic intervention. Changes in any of these variables reshape the fitness landscape, altering which tumour phenotypes possess the greatest evolutionary advantage. Within this framework, tumour progression is not simply the result of accumulating mutations. It is the process by which tumour populations move through a changing evolutionary landscape toward phenotypes that best exploit their environment.
This framing connects directly to the metabolic argument developed across the Quiet Biology series. The MDM2 Convergence paper established that chronic AKT activation, driven by insulin excess, simultaneously suppresses p53 and dysregulates AR turnover. In evolutionary terms, this is a stable fitness landscape feature: a persistently elevated insulin environment selects for tumour phenotypes that have adapted to exploit it. Correcting the metabolic field does not merely reduce a molecular signal. It reshapes the fitness landscape in which the tumour population is evolving.
02Therapeutic Pressure and Landscape Reshaping
Conventional oncologic therapy often applies continuous, maximal pressure with the aim of eliminating malignant populations as rapidly as possible. While this approach may produce substantial reductions in tumour burden, it also reshapes the fitness landscape in ways that favour resistant phenotypes. When therapy eliminates sensitive tumour cells, the ecological niches those cells occupied become available. Resistant variants that previously existed as minor populations may now occupy the highest available fitness peaks in the new therapeutic environment. What appears clinically as treatment failure is often the predictable outcome of evolutionary selection under stable selective pressure.[2]
This dynamic is well illustrated in androgen deprivation therapy for prostate cancer. Androgen-sensitive tumour cells dominate the initial tumour ecology because they exploit the androgen-rich hormonal environment. When androgen signalling is suppressed pharmacologically, this ecological niche collapses. Cells capable of surviving in low-androgen conditions, through receptor hypersensitivity, ligand independence, or lineage plasticity, gain a strong selective advantage. The tumour that emerges following prolonged suppression is therefore not merely the original tumour returning. It is a new evolutionary community shaped by the selective pressures imposed during treatment.[2a]
The BAT paper in this series documents the clinical consequence of this process and its reversal. By cycling between supraphysiological testosterone and castration, BAT destabilises the adapted CRPC population and restores sensitivity to androgen deprivation. In evolutionary terms, BAT is a landscape-reversal strategy: it removes the stable low-androgen environment that CRPC cells have optimised for and replaces it with a rapidly alternating environment that no single phenotype can optimise for simultaneously. Critically, BAT does not merely reverse the landscape. It does so faster than a CRPC-adapted population can re-optimise in the opposite direction. This is the adaptive lag mechanism in clinical operation: the oscillation rate exceeds the population's adaptive response rate, leaving the tumour chronically maladapted to whichever environment currently prevails. The clinical response to BAT is therefore not simply a pharmacological phenomenon. It is an evolutionary one.
03Adaptive Therapy and Competitive Suppression
Recognition of the evolutionary consequences of continuous therapy has led to the development of adaptive therapy, an approach that modulates treatment intensity in order to preserve ecological competition within tumours. Rather than attempting complete elimination of sensitive tumour cells, adaptive therapy maintains these populations so that they continue to compete with resistant clones for resources and space. By preserving competitive suppression, resistant populations are prevented from expanding unchecked.[3]
Clinical trials in metastatic prostate cancer have demonstrated that adaptive therapy strategies can extend time to progression while using substantially less total drug exposure than conventional continuous dosing, findings that provide empirical support for the ecological interpretation of tumour evolution.[4]
Adaptive therapy, however, addresses only one dimension of evolutionary management: population competition. A second dimension arises from the dynamics of the evolutionary landscape itself. Preserving sensitive cell populations to suppress resistant ones is a strategy for managing an existing tumour ecology within a given landscape. Dynamic fitness landscapes offer a complementary strategy: making the landscape itself inhospitable to stable evolutionary optimisation, regardless of which population is currently dominant.
04Dynamic Fitness Landscapes and Adaptive Lag
In classical evolutionary theory, populations evolve toward local fitness peaks when the environment remains stable. When the environment changes slowly, populations can track these changes and remain near the moving optimum. When the environment changes rapidly, however, populations may fail to keep pace. The result is adaptive lag, a state in which organisms remain maladapted because the optimal phenotype shifts faster than evolutionary processes can track it.[5]
Mathematically, this condition arises when the rate of environmental change exceeds the rate of adaptive genetic or phenotypic adjustment. In such systems, populations continually pursue an adaptive peak that has already moved elsewhere. The evolutionary distance between the current population state and the optimal state remains large, not because the population is failing to evolve, but because the target is moving faster than evolution can follow.[6]
This principle has important implications for cancer evolution. Tumour populations often adapt efficiently to stable environments, evolving phenotypes optimised for specific metabolic, hormonal, or therapeutic conditions. If those conditions fluctuate continuously and rapidly enough, however, tumour populations may never reach a stable evolutionary optimum. Instead, they may remain in a state of incomplete adaptation, chronically maladapted, and therefore constrained in their capacity for progressive evolutionary escape.
Many forms of prostate cancer exhibit sufficiently slow evolutionary dynamics that they may be particularly amenable to landscape-based management strategies. Tumour populations that evolve slowly are more vulnerable to environmental fluctuation because their rate of adaptive response is limited relative to the rate at which the landscape can be shifted. This is not a universal feature of prostate cancer, metastatic castration-resistant disease can evolve rapidly, lineage plasticity may emerge abruptly, and genomic instability varies substantially between patients. The argument applies most directly to indolent and early-stage disease where evolutionary tempo is genuinely constrained. In such settings, the rate asymmetry between environmental oscillation and adaptive response is most likely to be exploitable.
05Oscillating Environments as Evolutionary Constraint: the Limits
Dynamic fitness landscapes can arise in tumours through multiple mechanisms. Intermittent therapeutic intervention introduces cycles of selective pressure and relief that continuously shift which phenotypes are favoured. Cyclical metabolic environments alter the energetic substrates available to tumour cells, favouring different metabolic configurations at different times. Fluctuating hormonal conditions reshape the endocrine signalling landscape. Immune activation and suppression cycles alter the immunological niche. Physiological disturbances such as structured exercise introduce transient systemic perturbations across multiple of these axes simultaneously.
In prostate cancer, endocrine signalling represents a particularly powerful environmental variable. Oscillations in androgen signalling, whether through physiological fluctuation or therapeutic cycling, may prevent stable adaptation to either androgen-rich or androgen-depleted states. The four-signal protocol architecture described in the Signal, Stress, and Selection paper, stress, inspection, stabilisation, reconstruction, in temporal sequence, is, in evolutionary terms, a landscape-cycling strategy. Each phase creates a different selective environment. The framework hypothesises that no single phenotype is likely to be optimally adapted across all four phases simultaneously. The sequence denies the tumour population the stable environment that stable adaptive optimisation requires.
A critical qualification is required here. Oscillating environments do not inevitably constrain adaptation. Under some circumstances, fluctuating environments may instead select for generalist phenotypes capable of tolerating a wide range of conditions, for bet-hedging strategies in which populations maintain phenotypic diversity as an adaptive insurance against environmental uncertainty, or for highly plastic phenotypes that can rapidly switch between adaptive states. Cancer populations are capable of all three responses. The existence of lineage plasticity in late-stage prostate cancer, the capacity to transition between luminal and neuroendocrine phenotypes, is a clinical demonstration of exactly this kind of adaptive breadth.
Whether the oscillation parameters of the framework favour adaptive lag rather than generalism is therefore an empirical question rather than a theoretical guarantee. The framework's hypothesis is that the rate and magnitude of environmental change matter critically: oscillations that are too slow or too narrow in amplitude may select for generalists; oscillations that are sufficiently rapid and broad may exceed the adaptive capacity of even plastic phenotypes, producing genuine lag. The protocol's temporal architecture, phases measured in days to weeks, spanning metabolic, mitochondrial, hormonal, and inflammatory axes simultaneously, is designed to operate in the lag-favouring rather than the generalism-favouring regime. Whether it succeeds in doing so is among the most important empirical questions the framework generates.
06Ecological Disturbance and Tumour Stability
Ecological systems outside oncology provide instructive analogies. In natural ecosystems, periodic disturbances, fires, floods, grazing events, prevent the dominance of single species and maintain biodiversity. The intermediate disturbance hypothesis, proposed by Connell in 1978 and influential though not without subsequent debate regarding its universality, suggests that moderate, periodic disturbance may maximise species diversity by preventing competitive exclusion while not destroying populations entirely.[7] Stable environments, by contrast, often allow competitive exclusion, in which one species monopolises available resources.
Tumour ecosystems may behave similarly. Stable metabolic or hormonal environments allow tumour populations to specialise and optimise for those conditions. Disturbances disrupt this specialisation, forcing populations to adapt repeatedly to changing conditions rather than converging on a single dominant phenotype. The analogy is instructive but should not be overstated: ecological diversity per se is not the goal in tumour management, and the conditions under which the intermediate disturbance hypothesis operates in tumour ecosystems remain poorly characterised.
Within this framework, physiological processes such as structured exercise may function as systemic ecological disturbances. Exercise induces transient fluctuations in metabolic substrates, oxygen distribution, immune activity, and endocrine signalling. While these perturbations are well tolerated by healthy tissues with full metabolic flexibility, tumour populations may be less able to accommodate repeated environmental shifts if they are operating near the limits of metabolic adaptation. Such disturbances do not eliminate tumours directly. They alter the evolutionary landscape in which tumours evolve, and may impose a meaningful constraint on evolutionary convergence in populations whose adaptive reserve is limited.
07Implications for Prostate Cancer Management
This perspective does not reject conventional oncologic therapies. Rather, it reframes their role within a broader evolutionary logic. Aggressive interventions remain appropriate when disease demonstrates clear biological autonomy or rapid progression, when the evolutionary tempo of the tumour has escaped the range in which landscape management is viable. In indolent or early-stage disease, however, preserving ecological containment while avoiding evolutionary acceleration may represent a more rational management strategy than applying continuous maximal pressure.
The metabolic field correction that the Quiet Biology protocol pursues is, in evolutionary terms, a landscape-level intervention. By reducing chronic insulin excess, restoring mTOR oscillation, improving mitochondrial quality, and modulating the hormonal environment, the protocol alters the fitness landscape across multiple axes simultaneously. It does not target a single phenotypic vulnerability, which would simply select for resistant variants that have lost that vulnerability. It changes the environmental conditions that define which phenotypes are viable at all.
It should be stated explicitly: the evolutionary management framework described here is a mechanistic hypothesis grounded in cancer ecology and evolutionary theory. While supporting evidence exists from adaptive therapy trials and experimental models, prospective clinical validation of oscillatory constraint strategies in prostate cancer remains limited. The rationale for this framework rests on biological plausibility and convergent evidence from multiple fields, and should be understood as such.
08Predictions and Experimental Questions
A framework paper is most useful when it generates hypotheses that can be tested. The following predictions follow directly from the evolutionary architecture described above. Each is falsifiable by available or feasible experimental methods. Taken together, they constitute the empirical agenda that would allow the adaptive lag model to be supported, refined, or rejected.
Hypothesis 1: Oscillatory interventions reduce resistant clone emergence relative to stable interventions
If adaptive lag constrains tumour evolution under oscillating conditions, then cell populations exposed to cycling selective environments should show reduced emergence of resistant phenotypes compared to populations exposed to continuous selective pressure at equivalent total dose. This is testable in vitro by comparing resistant clone frequency, tracked by whole-population sequencing or resistance marker expression, under stable versus oscillating androgen, metabolic, or therapeutic conditions. Failure to observe reduced resistance emergence under oscillating conditions would challenge the core adaptive lag hypothesis.
Hypothesis 2: Tumour populations exposed to oscillating environments exhibit measurable adaptive lag
If the lag mechanism operates, then cell populations subjected to environmental cycling should show persistent maladaptation relative to populations in stable environments, measurable as reduced fitness (proliferation rate, metabolic efficiency, survival under stress) during each successive phase compared to populations that have been held in that phase continuously. Adaptive lag predicts a systematic fitness deficit in cycled populations that diminishes if cycling stops. This is directly testable by comparing growth kinetics and stress responses in cycling versus stable-condition cell lines across multiple experimental cycles.
Hypothesis 3: Prostate cancer populations with slower evolutionary tempo derive greater benefit from landscape cycling
If the vulnerability to adaptive lag is rate-dependent, then tumour populations with lower mutational burden, lower genomic instability, and slower growth kinetics should show greater constraint under oscillating conditions than more genomically unstable, rapidly-evolving populations. In clinical terms, this predicts that landscape-cycling strategies will show greater benefit in indolent, early-stage disease than in rapidly-evolving metastatic CRPC. This prediction aligns with the adaptive therapy literature and is testable by correlating baseline evolutionary tempo markers (mutational burden, genomic instability, PSA doubling time) with response to oscillatory protocols.
Hypothesis 4: Oscillation rate and amplitude determine whether lag or generalism dominates
This hypothesis addresses the generalism alternative directly. If oscillation parameters determine which evolutionary outcome predominates, then there should be identifiable thresholds below which oscillation selects for generalists and above which it produces lag. In preclinical models, this predicts that slow-cycling environments (weeks to months between phase changes) should enrich for phenotypically plastic, broadly tolerant clones, while rapid-cycling environments (days to weeks) should produce populations with persistent fitness deficits across phases, the lag signature. Mapping the relationship between oscillation rate and the generalism-versus-lag outcome would provide the empirical basis for optimising protocol timing.
Hypothesis 5: Structured landscape cycling prolongs time to progression relative to stable suppression
The ultimate clinical prediction of the framework: patients receiving oscillatory metabolic and hormonal management should show longer time to biochemical progression than patients receiving equivalent interventions applied continuously or in unstructured combinations. This prediction is not yet testable at population scale, but is amenable to n-of-1 structured case series, tracking PSA dynamics, evolutionary markers, and metabolic parameters across cycling and non-cycling periods within individual patients, as a preliminary signal prior to formal trial design.
These hypotheses are not exhaustive. They are the predictions that most directly test the novel claims of the framework: the lag hypothesis, the rate-dependence hypothesis, the tempo-sensitivity hypothesis, and the generalism boundary. Falsification of any of them would require revision of the relevant section of the framework. Confirmation of all of them would not prove the framework correct, but would substantially increase its credibility and justify progression to prospective clinical evaluation.
A framework is not a conclusion. It is a structured set of claims about how things might be connected, accompanied by the predictions that would follow if the connections are real. The predictions above are what distinguish this framework from a sophisticated evolutionary essay.
Cancer evolution occurs within dynamic ecological landscapes shaped by environmental conditions and therapeutic interventions. Traditional cancer therapies often reshape these landscapes in ways that favour resistant phenotypes, accelerating evolutionary escape despite initial tumour regression. This is not treatment failure in any simple sense — it is the predictable outcome of applying stable selective pressure to an evolving population.
Evolutionary theory suggests an alternative possibility. When environmental conditions fluctuate rapidly and across sufficient amplitude, populations may fail to converge on stable adaptive states, a phenomenon of adaptive lag that may constrain evolutionary optimisation and maintain tumour populations in chronically maladapted configurations. This outcome is not guaranteed. Oscillating environments can equally select for generalists, bet-hedgers, and highly plastic phenotypes. The framework's hypothesis is that oscillation at sufficient rate and breadth, targeting populations with limited adaptive tempo, may favour lag over generalism. The predictions in Section 8 are what that hypothesis requires of the evidence.
For patients navigating indolent or slowly evolving disease, this framework suggests that therapeutic success may depend not only on the potency of individual interventions but on how those interventions shape the evolutionary landscapes in which tumours evolve. The aim is not to eliminate tumour populations immediately, but to govern the ecological conditions under which their evolution unfolds, constraining the range of available adaptive solutions, and extending the window of biological stability. This is the evolutionary rationale underlying the Quiet Biology approach.
A stable environment is an invitation to adapt.
An oscillating environment is a constraint.
Keep the landscape moving and the tumour cannot find its footing.
- Wright S. The roles of mutation, inbreeding, crossbreeding and selection in evolution. Proceedings of the Sixth International Congress of Genetics. 1932;1:356–366. Origin of the fitness landscape concept. Also: Gavrilets S. Fitness Landscapes and the Origin of Species. Princeton University Press. 2004. The definitive modern treatment of fitness landscape theory and its evolutionary implications.
- Greaves M, Maley CC. Clonal evolution in cancer. Nature. 2012;481(7381):306–313. doi:10.1038/nature10762. Establishes the evolutionary framework for understanding cancer progression, treatment failure, and resistance as outcomes of clonal selection within dynamic tumour ecosystems.
- Gundem G et al. The evolutionary history of lethal metastatic prostate cancer. Nature. 2015;520(7547):353–357. doi:10.1038/nature14347. Whole-genome sequencing of spatially distinct metastatic samples within individual patients; demonstrates polyclonal seeding, metastasis-to-metastasis migration, and treatment as a selection pressure driving resistant subclonal spatial redistribution, the empirical prostate cancer evidence for the clonal selection framework established in [2].
- Gatenby RA, Brown JS. Integrating evolutionary dynamics into cancer therapy. Nature Reviews Clinical Oncology. 2020;17(11):675–686. doi:10.1038/s41571-020-0411-1. The primary theoretical framework paper for adaptive therapy and evolutionary management of cancer, including the competitive suppression rationale.
- Zhang J, Cunningham JJ, Brown JS, Gatenby RA. Integrating evolutionary dynamics into treatment of metastatic castrate-resistant prostate cancer. Nature Communications. 2017;8(1):1816. doi:10.1038/s41467-017-01968-5. Clinical demonstration of adaptive therapy in metastatic prostate cancer, showing extended time to progression with substantially reduced drug exposure relative to standard continuous dosing.
- Bürger R, Lynch M. Evolution and extinction in a changing environment: a quantitative-genetic analysis. Evolution. 1995;49(1):151–163. doi:10.2307/2410312. Foundational mathematical treatment of adaptive lag, the condition in which the rate of environmental change exceeds the rate of adaptive genetic response, producing chronically maladapted populations.
- Kopp M, Matuszewski S. Rapid evolution of quantitative traits: theoretical perspectives. Evolutionary Applications. 2014;7(1):169–191. doi:10.1111/eva.12127. Mathematical analysis of the conditions under which populations track versus lag behind moving fitness optima, including the rate-dependence of adaptive lag.
- Connell JH. Diversity in tropical rain forests and coral reefs. Science. 1978;199(4335):1302–1310. doi:10.1126/science.199.4335.1302. The original statement of the intermediate disturbance hypothesis, that moderate, periodic disturbance maximises biodiversity by preventing competitive exclusion. Applied here as an ecological analogy for tumour phenotypic diversity under oscillating selective pressure.
- IndolentRead phase →
- ActiveRead phase →