Part XVII · Research methods, evidence, and emerging therapy · Chapter 97

Emerging treatment strategies

Strategies that change when and whether a drug is given, rather than which drug it is.

1 · Molecular interception and prevention trials

Interception sits between prevention and treatment. Prevention reduces the chance a lesion arises. Treatment addresses established cancer. Interception targets a lesion that exists and has not become invasive, or a field that is primed to produce one.

Endocrine prevention established the principle. Tamoxifen and aromatase inhibitors reduce hormone-receptor-positive breast cancer incidence in high-risk women, in NSABP P-1, the IBIS-II prevention trial¹ and related trials. After local excision of DCIS, IBIS-II DCIS found anastrozole non-inferior to tamoxifen for all recurrence, including new contralateral tumours². The problem is not efficacy. It is uptake, because the population offered prevention is well, the benefit is a reduction in a probability, and the toxicity is immediate and certain. The framing problem that creates is the one in Communication and shared decision-making.

The current direction is to make interception molecular rather than hormonal. Three questions organise it. Which DCIS lesions will progress, which is the question COMET, LORD and LORIS were built to address and which Active surveillance trials and what they will and will not answer discusses. What distinguishes a benign high-risk field from one that has begun clonal expansion. And whether an intervention short of endocrine therapy can act on a defined molecular state.

The trial design problem is severe and worth stating. An interception trial in a healthy population requires very large numbers, long follow-up, and a rare endpoint. That is the setting where the methodological constraints in Clinical trials and evidence interpretation bite hardest, and it is why this field moves slowly.

2 · Residual disease-guided intervention

The logic is direct. If residual disease after definitive treatment can be detected, and if detecting it earlier permits an intervention that works, then outcomes should improve. Each clause is a separate empirical claim and the evidence for them differs.

Detection is established. Circulating tumour DNA identifies patients at high risk of relapse with long lead times, as set out in Liquid biopsy and longitudinal monitoring. DARE and SURVIVE test detection-guided strategies.

Actionability is not established, and one trial defines the problem precisely. In c-TRAK TN, patients with triple-negative breast cancer were monitored after surgery and treated with pembrolizumab on ctDNA detection. Most patients found to be ctDNA-positive already had detectable metastatic disease on imaging at the time of detection, and pembrolizumab did not clear ctDNA³.

That result separates two failures. The lead time may be shorter than assumed, so detection is not occurring in a true minimal residual disease state. And the intervention may be ineffective against residual disease even when the timing is right.

The strategic consequence is that residual disease detection is a strong prognostic tool and is not yet a treatment strategy. Using it as one outside a trial creates anxiety and expense without demonstrated benefit.

3 · Dormancy-directed strategies

Dormancy is where the largest theoretical gain in this disease lies and where the least is established. The biology is in Dormancy, residual disease, and late recurrence. Disseminated tumour cells persist for years in a non-proliferative state, survive adjuvant therapy that targets dividing cells, and account for late recurrence in hormone-receptor-positive disease.

Three strategies are logically available and each has a different problem.

Maintain dormancy indefinitely. This requires knowing what sustains the dormant state and being able to reinforce it for decades. Extended endocrine therapy is the only current example, and its benefit is real and small relative to its burden.

Awaken and kill. Force dormant cells to cycle, then treat them with cytotoxic therapy. The obvious hazard is a failure of the second step, which converts a dormant cell population into active metastatic disease. No one has demonstrated control of the awakening reliably enough to accept that risk outside a trial.

Kill without awakening. Target a survival dependency specific to the dormant state, such as autophagy. CLEVER took this approach in patients with persistent disseminated tumour cells after standard therapy⁴.

The binding constraint on all three is measurement. Dormant cell burden is not measurable with adequate sensitivity in routine practice, and bone marrow aspiration is not a screening test. A strategy aimed at a population that cannot be identified cannot be tested efficiently, and that is the bottleneck rather than the drugs.

4 · Adaptive and evolution-informed treatment

Standard practice gives maximum tolerated dose continuously until progression. Evolutionary theory says this selects hardest for resistance, because it applies maximal and sustained pressure and removes the sensitive population that would otherwise compete with resistant clones.

Adaptive therapy proposes to dose to a target tumour burden rather than to maximum effect, deliberately maintaining a sensitive population to suppress resistant clones. The theory is coherent and the clinical evidence in breast cancer is minimal. The requirements are demanding: a reliable frequently measurable burden marker, resistance that carries a fitness cost, and a tumour where control rather than eradication is the goal.

Evolution-informed timing is the version that has now worked, and it is worth being exact about what it achieved. SERENA-6 monitored ctDNA during first-line aromatase inhibitor and CDK4/6 inhibitor therapy and switched to camizestrant when an ESR1 mutation emerged, before radiological progression. 3256 patients were tested and 315 were randomised. Median progression-free survival was 16.8 months with camizestrant plus the CDK4/6 inhibitor against 9.2 months with the aromatase inhibitor continued, with a hazard ratio of 0.45 and an interval from 0.34 to 0.59. Median second progression-free survival was 25.7 against 19.1 months, with a hazard ratio of 0.63 and an interval from 0.46 to 0.86⁵.

Overall survival is not established. What has been demonstrated is that switching earlier delays progression, twice over. That is a progression benefit and it is not evidence of longer life, which is the distinction Regulatory pathways, accelerated approval, and post-marketing evidence insists on and the one to hold when describing this trial to a patient.

This is not adaptive dosing. It is acting on a detected evolutionary event at the moment it appears rather than waiting for its clinical consequence. It is the first demonstration in breast cancer that the timing itself can be the intervention.

The result matters beyond the specific drug. It establishes that a molecularly detected resistance event can be treated as an actionable moment, which is the premise the whole ctDNA programme in Liquid biopsy and longitudinal monitoring rests on, and which c-TRAK TN had left in doubt.

5 · Rational combinations and sequence optimization

Combination development has been dominated by a weak heuristic: combine two active agents and expect more activity. The results have been additive toxicity more often than additive efficacy.

A rational combination has a stated mechanistic premise, and the premises are distinguishable. Block a resistance pathway before it is used, which is the logic of adding a PI3K pathway inhibitor to endocrine therapy. Exploit synthetic lethality, which is the PARP inhibitor logic in Targeted therapy. Convert a non-immunogenic tumour into an immunogenic one, which is the premise of chemotherapy with checkpoint blockade in Immunotherapy. Each premise implies a different trial design and a different biomarker.

Sequencing is the more consequential and less studied question. A patient with metastatic hormone-receptor-positive disease may receive six or more lines. The order changes the resistance landscape each subsequent line faces, and almost all evidence compares agents within a line rather than orders across lines.

SONIA is the model for how to ask the sequencing question. It compared CDK4/6 inhibition first line against second line. There was no significant difference in progression-free survival from randomisation to second progression, at 31.0 against 26.8 months⁶, and overall survival was 47.9 against 48.1 months⁷. The deferred arm spent substantially less time on the drug with less toxicity⁶. The finding that the sequence did not matter for outcome is precisely as important as a finding that it did.

6 · Central nervous system directed innovation

Central nervous system disease has been treated as a problem of drug delivery across the blood-brain barrier. That framing is incomplete and is now changing.

The delivery framing produced real gains. Tucatinib crosses adequately and improved intracranial outcomes^8,9, and trastuzumab deruxtecan demonstrated intracranial activity that the molecule's size was not expected to permit¹⁰. That second result is the one that broke the framing, since a conjugate of that size should not cross an intact barrier. The barrier in established metastases is not intact.

The second framing is biological. Brain metastases differ from their primaries in receptor status and in driver alterations¹¹, which means CNS disease can be a different disease treated on the primary's biology. The brain microenvironment is immunologically distinct, with resident microglia and astrocytes that actively support metastatic colonisation¹².

Leptomeningeal disease remains the least tractable presentation, with distinct biology and very poor outcomes^13,14. Intrathecal delivery bypasses the barrier and reaches only the compartment it is given into.

The open strategic question is prevention rather than treatment. Patients at high risk of CNS relapse are identifiable by subtype and stage. Whether screening asymptomatic high-risk patients improves outcome, rather than advancing diagnosis, is unresolved and is a lead-time question of the kind Lead time and the unresolved question of whether acting early helps describes.

7 · Biomarker-guided escalation, de-escalation, and treatment cessation

The three decisions share one structure. Each requires a marker that predicts what happens under the alternative, not merely what happens under current treatment. That is the prognostic-predictive distinction in Causal inference, bias, and external validation, and it is the reason genomic assays required randomised trials rather than validation cohorts^15,16.

Escalation is the best developed. Residual disease after neoadjuvant therapy identifies patients for intensification, which KATHERINE and CREATE-X established. The marker is a treatment response, which is a more reliable predictor than a baseline measurement.

De-escalation is advancing fastest and carries the methodological risk. LUMINA identified a population with 5-year local recurrence of 2.3% without radiotherapy¹⁷. SENOMAC and SOUND address axillary surgery, and RxPONDER and TAILORx address chemotherapy. Every one of these rests on a margin argument, and the margin discipline in Designing de-escalation trials and the non-inferiority margin problem is what separates a sound de-escalation from a hopeful one.

Cessation is the least developed and the most requested by patients. When can endocrine therapy stop. When can a patient in prolonged remission on maintenance therapy stop. POSITIVE addressed temporary interruption for pregnancy and found it did not worsen short-term outcomes, which is the closest existing evidence.

The barrier is the same in all three. A test that says a patient is at low risk on treatment does not say they would be at low risk off it. Until a marker is validated against the counterfactual, every cessation decision is a judgement made without the evidence that would settle it.

References

All 17 references

1. Cuzick J, Sestak I, Forbes JF, et al. Anastrozole for prevention of breast cancer in high-risk postmenopausal women (IBIS-II): an international, double-blind, randomised placebo-controlled trial. Lancet 2014 383:1041-1048. PMID 24333009
2. Forbes JF, Sestak I, Howell A, et al. Anastrozole versus tamoxifen for the prevention of locoregional and contralateral breast cancer in postmenopausal women with locally excised ductal carcinoma in situ (IBIS-II DCIS): a double-blind, randomised controlled trial. Lancet 2016 387:866-873. PMID 26686313
3. Turner NC, Swift C, Jenkins B, et al. Results of the c-TRAK TN trial: ctDNA mutation tracking to detect molecular residual disease and trigger intervention in moderate- and high-risk early-stage triple-negative breast cancer. Ann Oncol 2023 34:200-211. PMID 36423745
4. DeMichele A, Clark AS, Shea E, et al. Targeting dormant tumor cells to prevent recurrent breast cancer: a randomized phase 2 trial. Nat Med 2025 31:3464-3474. PMID 40897974
5. Turner NC, Mayer EL, Park YH, et al. Switching to camizestrant at ESR1 mutation emergence before disease progression during first-line treatment of hormone receptor-positive advanced breast cancer (SERENA-6). Lancet Oncol 2026. PMID 42442380
6. Sonke GS, van Ommen-Nijhof A, Wortelboer N, et al. Early versus deferred use of CDK4/6 inhibitors in advanced breast cancer. Nature 2024 636:474-480. PMID 39604725
7. Wortelboer N, van Ommen-Nijhof A, Konings IR, et al. Overall survival with first-line vs second-line CDK4/6 inhibitor use in advanced breast cancer: a randomized clinical trial. JAMA Oncol 2026 12:375-383. PMID 41712229
8. Murthy RK, Loi S, Okines A, et al. Tucatinib, trastuzumab, and capecitabine for HER2-positive metastatic breast cancer. N Engl J Med 2020 382:597-609. PMID 31825569
9. Lin NU, Borges V, Anders C, et al. Intracranial efficacy and survival with tucatinib plus trastuzumab and capecitabine for previously treated HER2-positive breast cancer with brain metastases in the HER2CLIMB trial. J Clin Oncol 2020 38:2610-2619. PMID 32468955
10. Harbeck N, Ciruelos E, Jerusalem G, et al. Trastuzumab deruxtecan in HER2-positive advanced breast cancer with or without brain metastases: a phase 3b/4 trial. Nat Med 2024 30:3717-3727. PMID 39271844
11. Priedigkeit N, Hartmaier RJ, Chen Y, et al. Intrinsic subtype switching and acquired ERBB2/HER2 amplifications and mutations in breast cancer brain metastases. JAMA Oncol 2017 3:666-671. PMID 27926948
12. Valiente M, Obenauf AC, Jin X, et al. Serpins promote cancer cell survival and vascular co-option in brain metastasis. Cell 2014 156:1002-1016. PMID 24581498
13. Fitzpatrick A, Iravani M, Mills A, et al. Assessing CSF ctDNA to improve diagnostic accuracy and therapeutic monitoring in breast cancer leptomeningeal metastasis. Clin Cancer Res 2022 28:1180-1191. PMID 34921020
14. Yang JT, Wijetunga NA, Pentsova E, et al. Randomized phase II trial of proton craniospinal irradiation versus photon involved-field radiotherapy for patients with solid tumor leptomeningeal metastasis. J Clin Oncol 2022 40:3858-3867. PMID 35802849
15. Sparano JA, Gray RJ, Makower DF, et al. Adjuvant chemotherapy guided by a 21-gene expression assay in breast cancer. N Engl J Med 2018 379:111-121. PMID 29860917
16. Kalinsky K, Barlow WE, Gralow JR, et al. 21-gene assay to inform chemotherapy benefit in node-positive breast cancer. N Engl J Med 2021 385:2336-2347. PMID 34914339
17. Whelan TJ, Smith S, Parpia S, et al. Omitting radiotherapy after breast-conserving surgery in luminal A breast cancer. N Engl J Med 2023 389:612-619. PMID 37585627