The New Blueprint for Cancer Treatment: Overcoming a Heterogeneous, Shapeshifting Disease (2026)
Quick Answer
Cancer treatment has produced genuine breakthroughs — checkpoint immunotherapy, CAR-T cells, antibody-drug conjugates, targeted therapy, and now personalized mRNA cancer vaccines. Yet outcomes for advanced and metastatic disease still lag behind the pace of innovation, largely because of two intertwined biological problems that no single breakthrough has solved: tumor heterogeneity (a tumor contains many genetically and functionally different cell populations) and phenotypic plasticity — cancer's "shapeshifting" capacity to change cellular identity and escape a treatment aimed at a fixed target. The new blueprint for cancer treatment reorganizes therapy around these two problems through six design principles: continuous molecular surveillance instead of a single biopsy, simultaneous multi-axis targeting instead of one pathway at a time, evolution-aware adaptive dosing instead of maximum-dose treat-to-progression schedules, therapies engineered to tolerate heterogeneity, active monitoring for identity-level transformation, and AI-guided modeling to adapt treatment as the tumor evolves. The new blueprint is therefore an integrated multi-modal approach that combines precision oncology, systemic therapy, immunotherapy, local control, metabolic and microenvironmental insights, resistance management, continuous monitoring and adaptive treatment. This is an educational framework describing where oncology is heading, not a treatment protocol — treatment decisions should always be made with a qualified oncology team.
Table of Contents
- The Breakthrough Paradox
- Problem One: One Cancer, Many Diseases
- Problem Two: The Shapeshifter — Phenotypic Plasticity
- The Old Blueprint vs. the New Blueprint
- The New Blueprint: Six Design Principles
- Pillar 1 — Continuous Surveillance, Not a Single Snapshot
- Pillar 2 — Multi-Axis Simultaneous Targeting
- Pillar 3 — Evolution-Aware, Adaptive Dosing
- Pillar 4 — Therapies Engineered to Tolerate Heterogeneity
- Pillar 5 — Targeting Plasticity Itself
- Pillar 6 — AI-Guided Modeling and Real-Time Adaptation
- Supporting Layer: Whole-Patient Optimization
- What's Established, Emerging, and Experimental Today
- A Practical Blueprint Checklist
- Frequently Asked Questions
- Related OneDayMD Research
- Conclusion
The Breakthrough Paradox
Modern oncology has delivered more genuine breakthroughs in the past fifteen years than in the previous fifty. Checkpoint immunotherapy has produced durable, sometimes decade-long remissions in cancers — metastatic melanoma, a subset of non-small cell lung cancer — that were once treated as uniformly fatal within a year or two. CAR-T cell therapy has cured a meaningful fraction of patients with refractory leukemia and lymphoma who had exhausted every other option. Targeted therapies transformed EGFR- and ALK-driven lung cancer, BRCA-associated cancers, and chronic myeloid leukemia from rapidly fatal diagnoses into manageable, sometimes chronic, conditions. Antibody-drug conjugates have added meaningful survival across breast, lung, and urothelial cancers. And as of this month, an individualized mRNA cancer vaccine has, for the first time, met its endpoints in a Phase 3 trial — a milestone the field has pursued for over a decade.
And yet, for a large share of advanced and metastatic solid tumors, survival gains have been real but incremental rather than transformative. Relapse after an initial response remains the norm, not the exception, across chemotherapy, targeted therapy, and immunotherapy alike. This is the breakthrough paradox: oncology now has more anticancer tools than at any point in history, and in aggregate, cancer is still out-adapting many of them.
The reason is not that any individual breakthrough failed. It is that most of these breakthroughs were designed, tested, and delivered around an implicit assumption — that a cancer is a reasonably fixed target that can be hit once, hard, at a single point in time. The systems oncology framework describes why that assumption breaks down biologically: cancer behaves as an evolving network of genetics, metabolism, immunity, and microenvironment rather than a static lesion. The OneDayMD Master Oncology Guide lays out the ten interconnected layers of modern cancer care, from prevention through emerging therapies, that this biology demands.
This article asks the next, more practical question: given that biology, what does treatment design itself need to look like? What follows is a six-pillar blueprint — not a replacement for established oncology, but a synthesis of where the evidence and the research pipeline are converging to directly address the two problems that have proven hardest to solve.
Problem One: One Cancer, Many Diseases
Cancer is not one disease, and it is rarely a static target. A tumor is rarely a single, uniform population of identical cells. Even within one patient, one organ, and one biopsy, a solid tumor typically contains multiple genetically distinct subclones, cells at different points in the cell cycle, cells with different metabolic states, and cells with different levels of immune visibility. Different regions of the same tumor mass can carry different driver mutations. Metastases can diverge further still from the primary tumor and from each other.
This matters enormously for treatment design because any therapy exerts selection pressure. When a drug eliminates the cell populations that depend on its target, it does not touch populations that do not depend on that target — and those surviving populations are, by definition, the ones that go on to define the recurrence. Oncologists increasingly describe this dynamic using a term borrowed from ecology: competitive release. Drug-sensitive cells that would normally compete with resistant cells for space, nutrients, and blood supply are wiped out, freeing the resistant population to expand without competition. A treatment can look highly effective on a scan — a large reduction in tumor burden — while simultaneously performing a selection experiment that hands the future of the disease to whichever subclone was least affected.
The practical consequence is that a single biopsy, taken from a single region of a single tumor at a single point in time, is a snapshot of a moving, branching process. It can miss a minority subclone that is biologically irrelevant today and clinically decisive in eighteen months. This is the first problem the new blueprint has to design around, and it is discussed in more mechanistic depth in the site's dedicated review of cancer treatment resistance.
Problem Two: The Shapeshifter — Phenotypic Plasticity
Heterogeneity explains how cancer survives treatment by already containing the right cells for the job. A second, distinct problem explains how cancer survives treatment by becoming the right cells for the job after the fact. This is lineage plasticity, sometimes called phenotypic plasticity: the capacity of a cancer cell to change its fundamental cellular identity — not simply acquire a new mutation, but shift its developmental program — in response to treatment pressure.
Three well-documented examples illustrate why this is such a difficult problem for treatment design:
- Neuroendocrine transformation in prostate cancer. Androgen deprivation and androgen-receptor-signaling inhibitors are a cornerstone of advanced prostate cancer treatment. In a meaningful share of cases — estimated around one in four advanced prostate cancers treated with modern AR-targeted therapy — the tumor does not simply mutate the androgen receptor; it transdifferentiates into neuroendocrine prostate cancer, a subtype that no longer depends on androgen receptor signaling at all. The drug is still working exactly as designed. The disease has simply stopped being the kind of cancer that drug treats.
- Histologic transformation in EGFR-mutant lung cancer. Lung adenocarcinomas driven by EGFR mutations can, under sustained EGFR-tyrosine-kinase-inhibitor pressure, transform into small-cell lung cancer or squamous cell carcinoma — entirely different histologic subtypes with different biology and different treatment approaches, arising from what was originally the same tumor.
- Drug-tolerant persister states. Some resistance is not genetic at all. Chemotherapy can induce a transient, reversible drug-tolerant phenotype in cancer cells that can revert back toward drug sensitivity once treatment pressure is removed — a form of non-genetic, reversible plasticity that current genomic testing is not designed to detect.
Recent lineage-tracing work has gone further, identifying a minority subpopulation within some lung tumors — a "high-plasticity cell state" — that appears disproportionately capable of jumping between cell states and seeding both early and advanced disease. Plasticity is increasingly discussed in the cancer biology literature as a distinct, emerging hallmark of cancer in its own right, alongside genomic instability and immune evasion.
The strategic implication is significant: resistance is not one mechanism to solve once. It is several independent biological strategies — genetic mutation, epigenetic reprogramming, microenvironmental protection, immune escape, and now identity-level transformation — any one of which can independently produce the same clinical outcome: treatment failure. A blueprint built to counter only genetic resistance will eventually run into a tumor that resists by changing what it is rather than what mutations it carries. This dynamic is explored further in the site's reviews of immunotherapy resistance and the evolution of cancer resistance.
The Old Blueprint vs. the New Blueprint
Framed side by side, the shift in treatment design philosophy becomes clear:
| Old Blueprint | New Blueprint |
|---|---|
| One biopsy at diagnosis defines the treatment plan | Longitudinal molecular surveillance tracks the tumor as it changes |
| Treat continuously at the maximum tolerated dose until progression | Adjust dose and timing based on measured tumor-burden dynamics where evidence supports it |
| One drug, one pathway, one mechanism at a time | Multiple biological axes (driver, immune, microenvironment, metabolism) addressed together where rational and evidence-based |
| Therapies designed around one shared, fixed target | Therapies engineered to tolerate heterogeneity: multi-antigen, patient-specific, or self-adapting |
| Resistance is investigated only after relapse occurs | Plasticity and transformation are actively monitored for, especially at atypical progression |
| Treatment adjustments rely on periodic imaging and clinician judgment alone | AI-assisted modeling helps integrate multiple data streams to inform adjustment, alongside clinical judgment |
None of this makes the old blueprint wrong — it produced most of the breakthroughs described above, and it remains the foundation of evidence-based oncology. The new blueprint is better understood as the next layer being built on top of it.
The New Blueprint: Six Design Principles
The six pillars below are not a treatment protocol. They are a synthesis of where current clinical trial design, drug development, and research infrastructure are converging in response to heterogeneity and plasticity. Several build directly on the ten-layer OneDayMD Master Oncology Guide, particularly its combination-architecture and resistance layers, applied here specifically through the lens of "how do we design around a moving, shapeshifting target."
Continuous Surveillance, Not a Single Snapshot
Because a single tissue biopsy captures one region at one moment, the new blueprint pairs it with longitudinal, blood-based monitoring wherever the evidence supports it. Circulating tumor DNA (ctDNA) analysis — a form of liquid biopsy — can detect molecular residual disease and emerging resistance mutations from a blood draw, often before recurrence is visible on imaging.
This is no longer purely theoretical. In stage II colon cancer, a randomized trial found that using post-surgical ctDNA status to guide adjuvant chemotherapy decisions reduced chemotherapy use without compromising recurrence-free survival compared with standard clinicopathological decision-making. In muscle-invasive bladder cancer, the Phase III IMvigor011 trial used ctDNA-detected molecular residual disease after surgery to identify patients who benefited from adjuvant immunotherapy, improving disease-free and overall survival specifically in the ctDNA-positive group. In hormone-receptor-positive metastatic breast cancer, the PADA-1 trial used ctDNA monitoring of emerging ESR1 mutations to trigger an early treatment switch, demonstrating the feasibility of molecularly-triggered, adaptive treatment escalation in real time.
Consensus groups in lung cancer and other solid tumors are now actively developing standardized frameworks for ctDNA-guided monitoring, and the site's own review of the 2026 ASCO ctDNA guidelines covers current clinical recommendations in more depth. Molecular tumor boards that re-review a patient's biology at key decision points — not only at initial diagnosis — are becoming a companion practice to this monitoring, particularly when treatment stops working as expected.
Multi-Axis Simultaneous Targeting
If heterogeneity means a tumor contains subpopulations that survive on different biological axes — some driven by an oncogenic mutation, others protected by the tumor microenvironment, others invisible to the immune system — then a therapy addressing only one axis will always leave a door open. The Master Oncology Guide's combination architecture frames this as attacking multiple layers at once: controlling the primary tumor, targeting the dominant molecular driver, activating antitumor immunity, addressing known resistance mechanisms, and optimizing the metabolic and physical health of the host, all while monitoring and adapting over time.
This does not mean simply stacking more drugs together. Biological synergy observed in laboratory experiments does not automatically translate into clinical benefit, and combining multiple active therapies can increase toxicity, drug interactions, and cost substantially. The new blueprint's version of this principle is rational, evidence-based multi-axis design: identifying which independent biological escape routes a given cancer's heterogeneity is likely to use, and addressing more than one of them deliberately, rather than adding therapies simply because each has an interesting mechanism in isolation.
Evolution-Aware, Adaptive Dosing
Standard oncology dosing philosophy is built around the maximum tolerated dose, given continuously until progression, on the logic that killing as many cancer cells as fast as possible is always the goal. Evolutionary biologists working in oncology have argued this can be self-defeating precisely because of competitive release: maximal cell kill eliminates the drug-sensitive cells that would otherwise competitively suppress the growth of resistant cells, handing the resistant population an unconstrained growth advantage.
Adaptive (or evolutionary) therapy modulates dose or timing based on measured tumor-burden response, deliberately maintaining a population of drug-sensitive cells to keep resistant cells in check. In a pilot clinical trial in metastatic castration-resistant prostate cancer, abiraterone was paused once PSA fell substantially below baseline and resumed once it rose again; researchers reported meaningfully longer time to progression than historical continuous-dosing controls, using less than half the cumulative drug dose. Related adaptive-dosing research is underway in platinum-sensitive ovarian cancer (the ACTOv trial) and in PARP-inhibitor maintenance therapy, and intermittent-dosing concepts have also been explored in BRAF-mutant melanoma. A July 2026 mathematical modeling study published in Genetics extended this reasoning further, suggesting that timed switching between multiple therapies — before a tumor regrows — could outperform maximum-dose treatment by exploiting the fitness costs that resistance often carries.
At a larger scale, the U.S. Advanced Research Projects Agency for Health (ARPA-H) has launched ADAPT, a nationwide initiative aiming to build the infrastructure to detect tumor evolution in real time and use interpretable AI within an evolutionary clinical trial platform to guide treatment adjustments as a patient's cancer changes.
Important caveat
Adaptive and evolutionary dosing trials remain investigational, are typically studied one cancer type at a time, and are run as structured, biomarker-triggered protocols within clinical trials — not as a general recommendation to reduce, pause, or skip standard treatment on one's own. Existing pilot data, while promising, mostly comes from small, single-arm studies rather than large randomized trials, and should be interpreted accordingly.
Therapies Engineered to Tolerate Heterogeneity
A parallel strategy accepts that heterogeneity cannot always be out-monitored or out-dosed, and instead builds therapies that are less vulnerable to it by design.
Multi-antigen and bispecific approaches
Dual-antigen CAR-T constructs, such as those simultaneously targeting CD19 and CD22 in B-cell malignancies, are designed specifically to reduce antigen-loss relapse — the scenario where a tumor escapes single-target therapy simply by losing or down-regulating that one target. In solid tumors, where CAR-T efficacy has historically lagged behind blood cancers because of antigen heterogeneity, an immunosuppressive microenvironment, and physical tumor barriers, next-generation strategies are emerging: cytokine-armored CAR-T cells, logic-gated "SynNotch" constructs that require two signals before activating, and CAR-T cells engineered to locally secrete bispecific T-cell engagers. In an early glioblastoma trial, CAR-T cells targeting one EGFR variant while also secreting an antibody fragment targeting standard EGFR produced rapid tumor regression in a small initial cohort — a proof of concept for addressing antigen heterogeneity within a single engineered therapy, though based on very few patients to date.
Tumor-infiltrating lymphocyte (TIL) therapy
Lifileucel (brand name Amtagvi), FDA-approved in February 2024 for metastatic melanoma, harvests and expands a patient's own tumor-reactive T cells directly from their tumor. Because these T cells are already naturally adapted to that specific patient's antigen landscape rather than engineered against one predetermined target, TIL therapy is inherently more tolerant of heterogeneity than a single-antigen approach.
Personalized neoantigen mRNA vaccines
This is arguably the most direct answer to heterogeneity available today: instead of targeting one antigen shared across patients, the tumor is sequenced and a vaccine is manufactured against that individual patient's unique mutational fingerprint. Two recent developments illustrate how quickly this field is moving:
- On August 19, 2026, Merck and Moderna announced that their individualized neoantigen mRNA therapy, combined with pembrolizumab, met its primary and secondary endpoints in a Phase 3 trial for resected, high-risk melanoma — reducing cancer recurrence and distant spread compared with pembrolizumab alone. Outside oncologists described it as a landmark first for the individualized-vaccine field. This result was announced by the companies; full peer-reviewed data and overall-survival follow-up are still pending, and the same platform is now being tested across lung, bladder, kidney, gastric, and pancreatic cancers.
- A separate, much smaller Phase 1 program from BioNTech, Genentech, and Memorial Sloan Kettering tested an individualized mRNA vaccine in resected pancreatic cancer — a cancer notorious for immune resistance. About half of the sixteen patients mounted a strong T-cell response, and long-term follow-up found that most of those immune responders remained alive years later, though the cohort is small, uncontrolled, and hypothesis-generating. A Phase 2 trial is now underway to test the finding in a larger, controlled population.
These approaches share a common logic: rather than betting that one drug hits enough of a heterogeneous tumor's cells, the therapy is built to match the specific patient's cancer from the outset.
Targeting Plasticity Itself — The Next Resistance Frontier
Pillars one through four largely address heterogeneity: the problem of a tumor already containing multiple cell populations. Lineage plasticity is a different and, at present, harder problem, because the tumor's identity itself can shift after treatment begins. A therapy exquisitely matched to a tumor's biomarkers at diagnosis can become irrelevant if the tumor later transdifferentiates into something biologically distinct.
This pillar is deliberately the least mature part of the blueprint. No therapy is currently approved to prevent or reverse lineage plasticity in humans. What is emerging in the research literature includes:
- Treating unexpected or atypical progression, especially in cancers with known plasticity risk such as EGFR-mutant lung cancer and AR-targeted prostate cancer, as a trigger for re-biopsy to check for histologic transformation rather than assuming the original cancer has simply become resistant in place.
- Investigating the signaling pathways that drive plasticity itself — for example, research implicating JAK and FGFR signaling in prostate cancer lineage plasticity — as potential co-targets that might delay or blunt transformation rather than only treating the cancer that results from it.
- Recognizing lineage plasticity as an emerging hallmark of cancer in its own right in ongoing cancer biology research, distinct from, but related to, genomic instability and immune evasion.
The practical message for now is awareness rather than a specific intervention: a tumor that stops responding in an unusual or unexpectedly aggressive way is not always simply "resistant" in the conventional sense — it may have changed what it is.
AI-Guided Modeling and Real-Time Adaptation
The first five pillars generate a great deal of data: longitudinal ctDNA, imaging, multi-omic tumor profiling, treatment-response trajectories, and re-biopsy results at key decision points. Pillar six is the connective tissue that makes this practically usable rather than overwhelming.
Digital twin models — dynamic, continuously updated virtual representations of an individual patient's cancer that integrate imaging, genomic, and treatment-response data — are an active and rapidly expanding research area, with applications proposed across treatment selection, radiotherapy planning, and immuno-oncology response prediction. Separately, machine-learning tools trained on large multi-omic datasets are being developed specifically to help decode and anticipate resistance mechanisms across chemotherapy, targeted therapy, and immunotherapy. ARPA-H's ADAPT initiative, discussed in Pillar 3, explicitly pairs real-time evolution detection with interpretable AI as part of its national infrastructure goal.
Important caveat
As with every AI application in oncology, a model identifying a statistical association in data is not the same as a clinically validated decision tool. Digital twins and AI-guided treatment-selection algorithms remain investigational and require prospective clinical validation before they should guide real-world treatment decisions outside of research settings.
Supporting Layer: Whole-Patient Optimization
None of the six pillars above matter if a patient cannot physically tolerate the treatment built to deliver them. Nutrition, preserved muscle mass, physical function, metabolic health, sleep, smoking cessation, and psychological support remain foundational to whether a sophisticated, multi-axis treatment plan can actually be completed as designed. This is covered in depth in Layers 7 and 9 of the OneDayMD Master Oncology Guide, and is treated here as a supporting foundation rather than repeated in full.
What's Established, Emerging, and Experimental Today
Evidence strength is graded using the same hierarchy defined in the OneDayMD Master Oncology Guide (Tier 1, established clinical evidence, through Tier 5, anecdotal evidence), broadly consistent with CEBM-style evidence levels used elsewhere on this site.
| Approach | Current evidence tier | Notes |
|---|---|---|
| ctDNA/MRD-guided adjuvant therapy (colon, bladder, selected breast cancer) | Tier 1–2 | Multiple randomized trials in specific settings; not yet standardized across all cancers |
| CAR-T cell therapy for hematologic malignancies | Tier 1 | Multiple FDA-approved products; established standard of care in specific indications |
| TIL therapy (lifileucel) for metastatic melanoma | Tier 1–2 | FDA-approved February 2024 |
| Adaptive/evolutionary dosing (prostate, ovarian, melanoma) | Tier 3 | Small pilot and early-phase trials; not yet standard of care |
| Individualized mRNA neoantigen vaccine, adjuvant melanoma | Tier 3 | Phase 3 endpoints met per August 2026 company announcement; full publication and OS data pending |
| Individualized mRNA vaccines in pancreatic and other solid tumors | Tier 4 | Small Phase 1–2 cohorts; hypothesis-generating |
| CAR-T / bispecific antibodies for solid tumors | Tier 3–4 | Multiple early trials across targets; largely not yet approved for solid tumors |
| AI digital twins / resistance-prediction algorithms | Tier 4 | Active computational and early clinical-validation research |
| Therapies directly targeting lineage plasticity | Tier 4–5 | Mechanistic and preclinical; no approved human therapy yet |
A Practical Blueprint Checklist
For patients and clinicians thinking through how this framework applies to an individual case, useful questions include:
- Has the tumor been profiled more than once, or only at the time of diagnosis?
- Does the current treatment plan address more than one biological axis (driver, immune, microenvironment), where evidence supports doing so?
- Is there a monitoring plan for emerging resistance — imaging combined with blood-based ctDNA testing where clinically appropriate?
- If progression occurs unexpectedly, atypically, or faster than the original cancer's known behavior, has re-biopsy to check for histologic or lineage transformation been discussed?
- Are there relevant clinical trials for this specific cancer type involving adaptive dosing, personalized neoantigen vaccines, or multi-antigen cellular therapy?
- Is whole-patient health — nutrition, muscle mass, physical function — being actively supported alongside tumor-directed treatment?
Frequently Asked Questions
It is a framework describing how cancer treatment is being redesigned around two biological problems that individual breakthroughs have not solved on their own: tumor heterogeneity and phenotypic plasticity. The blueprint centers on continuous molecular surveillance, multi-axis simultaneous targeting, evolution-aware adaptive dosing, heterogeneity-tolerant therapies, plasticity monitoring, and AI-guided real-time adaptation.
Most major breakthroughs were built to hit a single, well-defined target identified at one point in time. Because tumors are heterogeneous and evolving, a subpopulation that does not depend on that target, or that later changes identity, can survive and repopulate the tumor. Progress has been substantial, but a single-mechanism breakthrough rarely eliminates every subpopulation in a heterogeneous, adapting tumor.
It refers to lineage plasticity: cancer cells changing their cellular identity, not just accumulating new mutations, in response to treatment pressure. Documented examples include neuroendocrine transformation in prostate cancer after androgen-receptor-targeted therapy and transformation of EGFR-mutant lung adenocarcinoma into small-cell or squamous lung cancer.
Generally no. It remains investigational, studied within specific trials for cancers such as metastatic castration-resistant prostate cancer and ovarian cancer, and run as a structured, biomarker-triggered protocol — not a self-directed decision to reduce or pause standard treatment.
Evidence is early but encouraging. An individualized neoantigen mRNA therapy combined with pembrolizumab met its primary and secondary endpoints in a Phase 3 trial for resected high-risk melanoma in August 2026, though full peer-reviewed results and overall survival data are pending. A separate small Phase 1 program in pancreatic cancer has shown durable responses in a very small number of patients. These are promising, active areas of investigation rather than broadly established treatments outside melanoma's adjuvant setting.
Not as a standalone decision-maker. Digital twin models and AI-based prediction tools are an active research area for anticipating treatment response and resistance, but they require prospective clinical validation, and current treatment decisions should be made by a qualified oncology team.
It means a single biopsy may not represent the whole cancer. It does not make biopsies useless; it is why oncology increasingly combines tissue biopsy with longitudinal liquid biopsy and, where relevant, multi-region sampling.
It is the ability of a cancer cell to shift from one cellular identity to another, often under treatment pressure, letting the tumor escape a therapy without needing a new mutation in the original target. It is increasingly discussed as a distinct, emerging hallmark of cancer.
No. It is an educational framework describing the direction oncology research is moving, not a treatment protocol. Many approaches discussed remain investigational for most cancers. Treatment decisions should always be individualized by a qualified oncology team.
Whether the tumor has been profiled more than once; whether the current plan addresses more than one biological pathway; whether there is a plan to monitor for emerging resistance; whether re-biopsy would be considered if progression looks atypical; and whether relevant clinical trials exist for this specific cancer type.
Conclusion: A Blueprint for a Moving Target
Cancer has proven so difficult to solve outright not because oncology lacks powerful tools, but because most of those tools were built to hit a target that does not hold still. Heterogeneity means the target was never singular to begin with. Plasticity means the target can change what it is after treatment begins. Together, these two properties explain why breakthrough after breakthrough has extended survival and cured some patients without yet closing the door on relapse and resistance across the board.
The new blueprint does not claim to have solved this. It represents where the evidence, the clinical trial pipeline, and the underlying biology are converging: toward continuous rather than single-point information, toward multiple biological axes addressed together rather than one at a time, toward dosing informed by tumor evolution rather than fixed schedules, toward therapies built to tolerate a moving target rather than assume a fixed one, toward active vigilance for the tumor changing its identity altogether, and toward computational tools that can hold and act on all of this complexity faster than a human team alone. Most of these pillars remain investigational for most cancers today. Together, they describe the direction modern oncology is moving — from static treatment of a disease to dynamic engagement with an evolving one.
Last reviewed: August 2026.
References and Evidence Sources
- Zhang J, Cunningham JJ, Brown JS, Gatenby RA. Integrating evolutionary dynamics into treatment of metastatic castrate-resistant prostate cancer. Nature Communications. 2017;8:1816.
- Noble R, et al. Mathematical modeling of timed therapy switching and evolution of drug resistance in cancer. Genetics. Published July 22, 2026.
- Advanced Research Projects Agency for Health (ARPA-H). The ADAPT learning cancer treatment system: a nationwide initiative to detect and respond to tumor evolution in real time. 2026.
- Enabling evolutionary therapy in metastatic cancer lacking serum biomarkers (ACTOv trial, NCT05080556, platinum-sensitive ovarian cancer). Preprint, 2026.
- Moffitt Cancer Center. Adaptive PARP-inhibitor maintenance dosing in ovarian cancer. Cell Systems. 2026.
- Tie J, et al. Circulating tumor DNA analysis guiding adjuvant therapy in stage II colon cancer. New England Journal of Medicine. 2022.
- IMvigor011 investigators. ctDNA-guided adjuvant atezolizumab following cystectomy in molecular-residual-disease-positive bladder cancer (Phase III). Summarized in ESMO Daily Reporter, 2026, and New England Journal of Medicine.
- Asian Thoracic Oncology Research Group. Consensus statement on ctDNA minimal residual disease testing in early-stage non-small cell lung cancer: a Delphi study. 2026.
- Jamroze A, Liu X, Tang DG. Treatment-induced stemness and lineage plasticity in driving prostate cancer therapy resistance. 2024.
- Lau LYE, Skanderup AJ, Tan AC. Lineage plasticity and histologic transformation in EGFR-TKI-resistant lung cancer. International Journal of Molecular Sciences. 2025/2026.
- Lineage plasticity and histological transformation: tumor histology as a spectrum. Cell Research. 2025.
- Critical role for a high-plasticity cell state in lung cancer. 2025/2026.
- Lineage plasticity in prostate cancer depends on JAK/STAT and FGFR inflammatory signaling. Science.
- Merck & Co. and Moderna. Phase 3 trial of individualized neoantigen therapy (mRNA-4157/V940) plus pembrolizumab in resected high-risk melanoma: topline results. Company announcement, August 19, 2026.
- Sethna Z, Guasp P, Reiche C, et al. RNA neoantigen vaccines prime long-lived CD8+ T cells in pancreatic cancer. Nature. 2025;639:1042–1051.
- Memorial Sloan Kettering Cancer Center. Long-term follow-up of individualized mRNA neoantigen vaccine (autogene cevumeran) in resected pancreatic cancer, Phase 1 cohort. 2026.
- Application of CAR-T therapy in solid tumors: current opportunities and challenges. Cancer Gene Therapy. 2026;33:479–491.
- Advancing CAR-T therapy for solid tumors: from barriers to clinical progress. 2026.
- Efficacy and safety of approved cellular therapies and bispecific antibodies in solid tumors: the current state. Discover Oncology. 2026.
- Digital twins in oncology: from predictive modelling to personalised treatment strategies. 2026.
- Advancing AI for multi-omics and clinical data integration in basic and translational cancer research. Nature Reviews Cancer. 2026.
.png)
.png)
.png)
Comments
Post a Comment