Precision Oncology 2026: Cancer Genomics, Biomarkers, Targeted Therapy and ESMO Updates

Updated 27 October 2026 · Includes ESMO Congress 2026, Madrid, Spain (23–27 October), and selected 2026 clinical, regulatory and peer-reviewed developments.

Evidence and safety note: This guide is for education, not an individual treatment recommendation. Conference results and company announcements can precede full peer-reviewed publication or regulatory review. A genomic finding does not, by itself, establish that a treatment will work or that it is approved for a particular patient.

Quick Answer

Precision oncology uses information about a cancer's molecular characteristics—such as gene mutations, gene fusions, copy-number changes, protein expression, immune markers and sometimes circulating tumour DNA (ctDNA)—to refine diagnosis, estimate risk, select treatments and monitor disease. It adds a molecular layer to conventional cancer classification; it does not replace the importance of cancer type, histology, stage, pathology, patient health and treatment goals.

Next-generation sequencing (NGS) can identify potential targets, but not every detected alteration is clinically actionable. The key question is whether strong evidence supports a specific biomarker–treatment match in the cancer type and clinical setting being considered. ESMO's updated ESCAT v2 framework, published in October 2026, provides a more detailed way to communicate the actionability of biomarkers and the evidence for biomarkers associated with resistance.

At ESMO Congress 2026 in Madrid, notable developments included phase III data for savolitinib plus osimertinib in selected EGFR-mutated, MET-overexpressing non-small cell lung cancer (SANOVO), and phase III results for an investigational personalised neoantigen mRNA therapy plus pembrolizumab after melanoma resection (INTerpath-001). These are important results in defined populations—not proof that one treatment or testing strategy works for every cancer.

Key Takeaways

• Genetics and genomics are related, but different in scope. Inherited germline testing can identify hereditary cancer risk; tumour profiling assesses acquired and sometimes inherited changes relevant to diagnosis or treatment.
• Actionability depends on context. A mutation's exact variant, cancer type, assay, co-alterations, prior therapy and clinical evidence all matter.
• ESCAT v2 strengthens clinical interpretation. The ESMO framework distinguishes levels of certainty for actionable targets and introduces a parallel scale for resistance biomarkers.
• SANOVO showed a progression-free survival benefit. In the trial's treatment-naive EGFR-mutated, MET-overexpressing NSCLC population, savolitinib plus osimertinib improved progression-free survival versus osimertinib alone. MET protein overexpression and MET gene amplification are not interchangeable tests.
• INTerpath-001 advanced personalised cancer vaccines into phase III evidence. Intismeran autogene plus pembrolizumab improved recurrence-free and distant metastasis-free survival after complete resection of high-risk melanoma. Overall-survival data were immature at the reported analysis, and the therapy remains investigational.
• ctDNA and minimal residual disease (MRD) have specific, not universal, roles. In the United States, Signatera CDx was authorised in 2026 to identify ctDNA-MRD-positive patients with resected muscle-invasive bladder cancer who may benefit from adjuvant atezolizumab. This does not validate every MRD test for every cancer.
• More sensitive detection is not automatically better screening. The NHS-Galleri randomized trial did not meet its primary endpoint of reducing combined stage III/IV diagnoses, although fewer stage IV cancers were observed in later screening rounds.
• AI is an aid, not an oncologist. Clinical tools need suitable validation, transparency and evaluation in the population where they will be used.
• Access and interpretation remain central barriers. Comprehensive profiling, specialist review, matched drugs and clinical trials are not uniformly available or reimbursed.

On this page

• Introduction: what precision oncology can—and cannot—do
• Genetics vs. genomics
• Common genomic alterations
• Tumour heterogeneity and resistance
• NGS and genomic testing methods
• Actionable biomarkers and ESCAT v2
• Clinical applications
• ESMO Congress 2026: Madrid findings
• Other important 2026 developments
• Evidence snapshot
• Limitations and access
• Questions to ask before testing or treatment
• What comes next?
• Frequently asked questions
• Using AI to understand a cancer report
• Selected references

1. Introduction: Why Precision Oncology Matters

Cancer is a collection of related diseases rather than one uniform condition. Tumours can arise in the same organ and still differ substantially in their mutations, gene expression, immune environment, growth patterns and sensitivity to treatment. Conversely, a molecular alteration found in different tumour types may sometimes point to a shared treatment option, although the evidence and benefit can differ by cancer type.

Precision oncology combines pathology, clinical findings and molecular information to make cancer care more specific to the patient and the disease. Depending on the clinical question, this may involve tumour NGS, RNA fusion testing, immunohistochemistry (IHC), in-situ hybridisation, germline testing, PD-L1 testing, mismatch-repair testing or a blood-based liquid biopsy.

Precision medicine does not mean that every cancer can be matched to a targeted drug. A result can be biologically interesting but clinically unproven; a drug may work only in a narrow subgroup; a tumour may develop resistance; or the potential benefit may not outweigh toxicity. The aim is not to order the largest possible test in every case. It is to choose a test that can answer a useful clinical question and interpret the result in context.

2. Genetics vs. Genomics: Key Differences

The terms overlap in everyday use, but they often describe different clinical questions. Germline testing examines inherited variants that may affect cancer susceptibility or influence treatment. Somatic tumour profiling looks for acquired changes in cancer cells. Broad genomic profiling can assess many genes simultaneously, while some focused tests assess one gene, one protein or one specific biomarker.

Cancer genetics

Inherited risk and family implications

Usually evaluates germline variants in blood, saliva or another validated sample. Results may affect screening, risk-reducing options, relatives' risk assessment and, in some cases, treatment.

Examples: BRCA1/BRCA2, mismatch-repair genes associated with Lynch syndrome, and TP53 in Li-Fraumeni syndrome.

Cancer genomics

The tumour's molecular profile

Examines cancer-related changes in tumour tissue or tumour-derived material in blood. Depending on the assay, this may include mutations, fusions, copy-number changes and selected biomarkers.

Examples: EGFR alterations in lung cancer, BRAF V600 alterations in melanoma, HER2 alterations in selected cancers, and NTRK gene fusions in eligible tumour types.

A tumour test can occasionally identify a variant that may also be inherited. When this possibility matters, confirmatory germline testing and genetic counselling may be appropriate. Tumour-only results should not automatically be interpreted as proof that a variant was inherited.

3. The Genomic Hallmarks of Cancer

Cancer evolves as cells acquire changes that promote growth, survival, immune escape and spread. The hallmarks framework is useful for understanding the biology, but no single mutation explains every behaviour of a tumour. Commonly discussed processes include:

• Persistent growth signalling: activating changes in pathways such as EGFR, RAS–RAF–MEK or PI3K–AKT can help drive cell division.
• Loss of tumour-suppressor activity: alterations in genes such as TP53 or RB1 may disrupt normal controls on cell growth.
• Resistance to cell death: tumour cells can acquire changes that make them less susceptible to apoptosis or treatment-induced damage.
• Immune evasion: changes in tumour and immune cells can reduce immune recognition or promote an immunosuppressive microenvironment.
• Replicative capacity and genome instability: altered telomere maintenance and DNA repair can help cells persist while generating further variation.
• Invasion and metastasis: additional changes and interactions with surrounding tissue allow some tumour cells to spread to distant organs.

These mechanisms help researchers identify potential drug targets. However, a pathway's biological importance does not prove that inhibiting it will improve patient outcomes. Clinical trials are needed to establish efficacy, safety and the population most likely to benefit.

4. Common Types of Genomic Alterations

Point mutations and small insertions/deletions

Changes in one or a few DNA bases. Some activate an oncogene, inactivate a tumour suppressor or alter how a drug binds. The exact variant matters: two mutations in one gene may have very different implications.

Copy-number changes

Gains or losses of DNA segments. Amplification can increase the number of copies of a gene, such as ERBB2/HER2 in certain cancers. Testing methods differ in how well they assess amplification.

Gene fusions and rearrangements

DNA breaks and rearrangements can join parts of genes, creating a fusion with diagnostic or treatment relevance. Some NTRK, ALK, ROS1, RET and other fusions have matched therapies in defined settings.

DNA repair and mutational signatures

Patterns of mutation or defects in repair systems can indicate tumour biology and sometimes treatment sensitivity. Examples include mismatch-repair deficiency and homologous-recombination repair alterations, interpreted within the relevant tumour and assay context.

Epigenetic changes

Changes in DNA methylation, histone marks or chromatin regulation can alter gene activity without changing the DNA sequence itself. Some are important diagnostically or therapeutically, but many remain research tools.

Gene expression and protein biomarkers

Not every clinically useful biomarker is a DNA mutation. RNA expression signatures, protein expression measured by IHC, hormone receptors and PD-L1 may be important for treatment selection in specific diseases.

For a molecular result to guide care, the report should identify the alteration or marker, the method used to detect it, relevant limitations, and the level of evidence supporting the proposed clinical interpretation. Variant classification systems and drug-label criteria may differ across laboratories and jurisdictions.

5. Tumour Heterogeneity, Clonal Evolution and Resistance

A cancer is not always a uniform population of cells. Different regions of the same tumour can contain different mutations, and separate metastases may evolve in different directions. This is called tumour heterogeneity. Under treatment pressure, sensitive cells may die while resistant subclones survive and become more prominent.

Resistance can be present before treatment or emerge later. Mechanisms may include a second alteration that restores pathway signalling, activation of a bypass pathway, transformation into a different histology, changes in drug transport or metabolism, and immune escape. A resistance biomarker can be as important as a biomarker of sensitivity: it may explain why a previously effective medicine stops working or why a particular drug should not be chosen.

This biology helps explain why a biopsy from years earlier may not fully describe a cancer after multiple therapies. When the disease progresses, clinicians may consider a new tissue biopsy, plasma ctDNA testing or other targeted tests if the result could alter management. These approaches are not interchangeable: plasma testing can miss alterations when little tumour DNA is shed, and tissue sampling can miss variation elsewhere in the disease.

6. Next-Generation Sequencing and Other Testing Methods

Next-generation sequencing (NGS) reads many DNA or RNA fragments in parallel. The type of test should match the clinical question, sample quality, suspected alteration and available treatment options.

Targeted gene panel

Analyses a selected set of genes known to be relevant to cancer care. Often practical for routine profiling, but may miss changes outside the panel or alterations the assay was not designed to detect.

Whole-exome sequencing

Focuses on protein-coding regions of the genome. It can support broad analysis but is not equivalent to whole-genome sequencing and does not detect every clinically relevant alteration equally well.

Whole-genome sequencing

Analyses a much broader proportion of the DNA, including non-coding regions. It can provide wider discovery potential but may require greater resources and complex interpretation.

RNA-based testing

Can be especially useful for detecting or confirming gene fusions and measuring gene expression. Some DNA-only panels may not detect all fusions reliably.

Immunohistochemistry and in-situ hybridisation

Evaluate proteins or selected DNA/RNA features in tissue. These tests remain important for markers such as hormone receptors, PD-L1, mismatch-repair proteins and HER2, depending on the cancer.

Liquid biopsy

Uses blood or another body fluid to look for tumour-derived material, including ctDNA. It can complement tissue profiling and is useful in selected settings, but a negative result does not always rule out a target.

There is no single “best” test for every patient. A broader assay may produce more findings, but additional findings can include variants of uncertain significance, incidental results and changes that have no established treatment implication. Test selection should consider whether the result is likely to change a decision, sample availability, turnaround time, cost and local access to matched therapies or trials.

7. From a Genomic Result to an Actionable Treatment

The central question is not simply whether a mutation was detected. It is whether the result is sufficiently reliable and clinically relevant to support a particular action for this patient at this point in the disease course.

A useful interpretation separates at least five questions:

1. Analytic validity: Can the assay reliably detect the alteration in this specimen and at the reported level?
2. Clinical validity: Is the biomarker associated with the diagnosis, prognosis, treatment response or resistance being claimed?
3. Clinical utility: Does using the result to guide care improve a meaningful outcome or otherwise support a sound decision?
4. Regulatory and guideline status: Is the treatment approved or recommended for this biomarker and clinical setting in the relevant country?
5. Patient-level fit: Do the expected benefits, risks, previous treatments, co-existing conditions, preferences and treatment goals support the option?

ESMO ESCAT v2: A clearer actionability and resistance framework

In October 2026, ESMO published version 2 of the ESMO Scale for Clinical Actionability of molecular Targets (ESCAT), updating the framework first introduced in 2018. ESCAT provides a common language for ranking how strongly a molecular alteration supports a treatment choice. It is an evidence framework, not an approval list or a guarantee that a drug will work for an individual patient.

Tier I

Actionability supported for clinical use

ESCAT v2 distinguishes proven, inferred and extrapolated evidence within Tier I. The clinical directive remains to act when the evidence supports a match, while the subcategory communicates how directly the evidence applies.

Tier II and lower tiers

Evidence strength and uncertainty

The framework refines how prospective interventional data, high-quality non-prospective or real-world evidence, and replicated case-level evidence are represented. Lower tiers signal greater uncertainty or a need for further evidence.

Tumour-agnostic matches

More explicit eligibility criteria

ESCAT v2 incorporates formal criteria aligned with ESMO's Tumour-Agnostic Classifier and Screener (ETAC-S) when assessing tumour-agnostic therapeutic activity.

Resistance biomarkers

A parallel scale

A separate four-level resistance framework ranges from clinically validated resistance to preclinical or computational evidence. ESCAT v2 also distinguishes alterations tested and found not to predict benefit from matches where the therapy has not shown meaningful activity.

Why the distinction matters: The same gene can contain numerous variants; not all variants have the same functional effect. A biomarker may be actionable in one cancer type but not another, or may matter only after a specific treatment. Conversely, a resistance biomarker can affect whether continuing a drug makes sense. The exact report wording, assay and supporting evidence must be reviewed rather than relying on a gene name alone.

ESCAT is also different from Oxford Centre for Evidence-Based Medicine (CEBM) evidence levels. CEBM describes the design and strength of evidence behind a clinical claim; ESCAT evaluates the clinical actionability of a biomarker–therapy relationship. They answer related but distinct questions.

Source: The ESMO Scale for Clinical Actionability of Molecular Targets, version 2, Annals of Oncology (2026); see also ESMO's ESCAT v2 explanation.

8. Where Precision Oncology Is Used in Clinical Care

8.1 Inherited cancer-risk assessment

Germline testing can identify variants associated with hereditary cancer syndromes. Results may guide screening intensity, risk-reducing interventions, family counselling and, for some cancers, selection of a treatment. Testing decisions should account for personal and family history, informed consent, possible implications for relatives and access to genetic counselling. A negative test does not eliminate all inherited risk.

8.2 Diagnostic classification

Molecular testing can help classify tumours whose appearance under the microscope is ambiguous, distinguish molecular subtypes, identify certain rare cancers and clarify whether a tumour may have originated elsewhere. Genomics complements expert pathology and clinical evaluation; it does not independently replace them.

8.3 Targeted therapy

Targeted therapies interfere with a molecular driver or pathway. Examples include EGFR inhibitors in selected lung cancers, HER2-directed treatments in eligible cancers, BRAF/MEK combinations for selected BRAF-altered tumours, and therapies directed at specific gene fusions. A listed drug may be approved only for a specific alteration, cancer type, stage or prior-treatment setting. Results from one tumour type should not automatically be transferred to another.

8.4 Immunotherapy selection

Biomarkers such as mismatch-repair deficiency/microsatellite instability-high status (dMMR/MSI-H), PD-L1 expression and tumour mutational burden (TMB) can help guide immunotherapy in defined contexts. Their predictive value varies by tumour type, assay, treatment regimen and regulatory label. A high TMB or positive PD-L1 result does not guarantee response, and a low or negative result does not always rule out benefit.

8.5 Treatment intensity and genomic-guided de-escalation

Precision testing can sometimes help identify patients unlikely to gain meaningful additional benefit from more intensive treatment. At ASCO 2026, the phase III OPTIMA trial reported results in clinically high-risk, ER-positive/HER2-negative early breast cancer using the Prosigna (PAM50) gene-expression assay to guide chemotherapy decisions. About two-thirds of the patients were classified as having a low genomic risk-of-recurrence score, and outcomes in the test-guided group were reported as comparable with standard chemoendocrine management.

These findings are relevant to the concept of avoiding unnecessary treatment, but they should not be turned into a blanket rule to omit chemotherapy. The assay, eligibility criteria, follow-up, local guidelines and individual clinical features matter. Patients should not change adjuvant therapy based only on a headline or an isolated genomic score.

Further reading: The ASCO Post's coverage of the phase III OPTIMA trial.

8.6 Liquid biopsy, ctDNA and minimal residual disease

A liquid biopsy detects cancer-derived material in a fluid sample, most often blood. Depending on the assay, it may be used to identify genomic alterations in advanced disease, investigate acquired resistance, monitor molecular changes over time or look for ctDNA after definitive treatment. MRD refers to cancer that remains at a level below what routine clinical assessment or imaging can reliably detect; tumour-informed ctDNA assays may help assess this in selected diseases.

In May 2026, the US Food and Drug Administration (FDA) approved Signatera CDx as a companion diagnostic to identify patients with muscle-invasive bladder cancer after cystectomy who have ctDNA MRD and may benefit from adjuvant atezolizumab. The approval is linked to a specific cancer, assay, treatment and clinical context. It is not an approval of all ctDNA tests for all cancers, and it should not be described as a universal MRD-based treatment-de-escalation rule.

The FDA also approved Guardant360 Liquid CDx in May 2026 with a broader genomic footprint for selected treatment-selection uses. As with tissue profiling, the report should be interpreted according to the specific alteration, the assay's intended use and the relevant drug indication.

Important limitations: A negative plasma result can be uninformative when a tumour sheds little ctDNA. A positive result needs interpretation within the clinical context and sometimes confirmation. The prognostic value of MRD detection is not the same as proof that changing treatment based on the result improves outcomes. Only use MRD results to change treatment where robust evidence and relevant guidance support that decision.

Sources: FDA approval of adjuvant atezolizumab for ctDNA-MRD-positive muscle-invasive bladder cancer; FDA approval record for Guardant360 Liquid CDx.

8.7 Cancer of unknown primary (CUP)

In cancer of unknown primary, the original site cannot be established despite diagnostic work-up. Comprehensive genomic profiling may identify a targetable alteration and support a discussion of molecularly guided therapy for selected patients. However, not all CUP cases contain an actionable alteration, and the evidence for each matched treatment differs.

The updated randomized phase II CUPISCO analysis, published in the Journal of Clinical Oncology in October 2026, reported longer median progression-free survival with molecularly guided therapy after disease control on induction platinum-based chemotherapy than with continued chemotherapy (6.1 versus 4.4 months; hazard ratio 0.75, 95% CI 0.59–0.95). Median overall survival was 15.2 versus 12.8 months, but that difference was not statistically significant in the reported analysis (hazard ratio 0.79, 95% CI 0.61–1.02; p=0.0689).

This supports comprehensive profiling and specialist review for appropriately selected patients; it does not establish that genomic-guided treatment is superior for every person with CUP or that a result can replace standard diagnostic evaluation.

Source: Krämer et al., updated CUPISCO efficacy and safety, Journal of Clinical Oncology (2026).

9. ESMO Congress 2026 in Madrid: The Data and Their Meaning

The European Society for Medical Oncology (ESMO) Congress 2026 took place in Madrid, Spain, from 23 to 27 October. The following developments are especially relevant to precision oncology. For clarity, results presented at a congress are identified as such; conference reporting is not the same as regulatory approval or a completed guideline update.

9.1 SANOVO: Targeting EGFR and MET in advanced non-small cell lung cancer

The phase III SANOVO trial evaluated savolitinib, a MET inhibitor, plus osimertinib, an EGFR inhibitor, in treatment-naive patients with locally advanced or metastatic EGFR-mutated non-small cell lung cancer (NSCLC) whose tumours also showed MET protein overexpression by immunohistochemistry. The reported intention-to-treat population included MET IHC 2+ or IHC 3+ disease.

Primary endpoint · Median PFS

24.9 months

Savolitinib + osimertinib

Versus 13.8 months with placebo + osimertinib

Relative effect

HR 0.56

95% CI 0.43–0.74; p<0.0001

A 44% relative reduction in the risk of progression or death in the reported trial population.

Overall survival

44.5 vs 34.9 months

Reported median OS for the combination versus osimertinib alone.

Reported HR 0.48 (95% CI 0.33–0.68); long-term follow-up and full publication remain important.

In the high-MET subgroup reported as IHC 3+ in more than 90% of tumour cells, median PFS was 24.8 versus 11.1 months (HR 0.45, 95% CI 0.31–0.66; p<0.0001).

Precision-oncology interpretation: This trial illustrates how a co-occurring resistance pathway may matter in a tumour already defined by a driver mutation. Testing for EGFR alone may not capture all relevant biology when MET is also clinically important.

Limits: These are conference-reported results and the source is a sponsor announcement. The study population was specific, and the trial was conducted in China. MET protein overexpression by IHC is not the same as MET gene amplification, and neither should be casually equated with MET exon 14 skipping. Clinicians must check the assay, full paper, local approval status, guidelines, safety profile and patient's situation before considering a treatment change.

Source: SANOVO phase III results reported for ESMO 2026, abstract LBA69.

9.2 INTerpath-001: A personalised neoantigen mRNA therapy after melanoma surgery

INTerpath-001 is a randomized, double-blind phase III trial of intismeran autogene (also called intismeran; formerly V940 or mRNA-4157), an investigational personalised mRNA-based neoantigen therapy, combined with pembrolizumab. The study enrolled 1,137 patients with completely resected high-risk stage IIB–IV cutaneous melanoma and compared the combination with pembrolizumab plus placebo.

At the reported primary analysis, with median follow-up of 24.6 months:

Recurrence-free survival

HR 0.58

95% CI 0.45–0.75; p<0.001.

A 42% relative reduction in risk of recurrence or death compared with pembrolizumab alone.

Distant metastasis-free survival

HR 0.62

95% CI 0.45–0.85; p=0.002.

A 38% relative reduction in risk of distant metastasis or death.

Two-year RFS rate

82.2% vs 70.0%

Combination versus pembrolizumab alone.

Median RFS had not been reached in either group at the reported analysis.

Two-year DMFS rate

88.5% vs 80.2%

Combination versus pembrolizumab alone.

Median DMFS had not been reached in either group at the reported analysis.

Overall-survival data were immature at the reported analysis. The study remains ongoing to evaluate overall survival and other endpoints. Sponsor-reported safety data showed grade 3–5 treatment-related adverse events in 14.2% of the combination group and 13.4% of the pembrolizumab-alone group; the benefit–risk profile requires full clinical review.

Why it matters: A personalised neoantigen therapy is designed using tumour-sequencing information to identify mutation-derived targets specific to an individual patient's cancer. INTerpath-001 provides phase III evidence for this approach in a defined post-surgery melanoma setting.

What it does not mean: Intismeran is investigational, not a universal cancer vaccine or a proven treatment for every cancer. The results apply to the trial population and regimen studied. Long-term survival data, regulatory review and local availability remain important.

Sources: Merck and Moderna's INTerpath-001 results announcement; The ASCO Post's report on the ESMO 2026 presentation.

9.3 How INTerpath-001 differs from the five-year KEYNOTE-942 data

The phase IIb KEYNOTE-942 study and phase III INTerpath-001 study evaluate the same personalised therapy platform in different trial populations and with different levels of evidence. The five-year KEYNOTE-942 follow-up, published in the Journal of Clinical Oncology in 2026, reported a recurrence-free survival hazard ratio of 0.51 and a distant metastasis-free survival hazard ratio of 0.411 for intismeran plus pembrolizumab compared with pembrolizumab alone. The five-year analyses were descriptive and involved 157 randomized patients.

Those phase IIb long-term data are supportive but should not be combined numerically with the larger phase III results as if they were one study. Differences in sample size, enrolment criteria, follow-up and analysis affect interpretation. INTerpath-001's phase III result is a separate, confirmatory data set; overall survival was not mature at the reported phase III analysis.

Source: Five-year KEYNOTE-942 update, Journal of Clinical Oncology (2026).

9.4 What the ESMO developments mean in practice

• Test with a decision in mind: establish whether the result may change diagnosis, treatment, resistance assessment, trial eligibility or monitoring.
• Read the exact biomarker: a variant, amplification, fusion, protein-expression result and ctDNA signal are different kinds of findings.
• Look at the evidence level: ESCAT v2 can structure actionability discussions, while the trial design and endpoints indicate how much clinical benefit has been demonstrated.
• Do not confuse relative and absolute effects: a 42% relative reduction in recurrence risk is not the same as a 42-percentage-point increase in the chance of remaining recurrence-free.
• Check availability and recommendations: presentation at ESMO does not automatically equal regulatory approval, reimbursement or a new standard of care.

10. Other 2026 Developments in Precision Oncology

10.1 NHS-Galleri: multi-cancer early detection and the importance of primary endpoints

The NHS-Galleri randomized trial evaluated annual multi-cancer early detection (MCED) blood testing alongside standard screening in more than 142,000 participants in England. Results presented at ASCO 2026 did not meet the primary endpoint of a statistically significant reduction in combined stage III/IV diagnoses for the 12 prespecified cancers. After three screening rounds, the incidence-rate ratio for combined stage III/IV cancers was 1.03 (95% CI 0.92–1.14; p=0.6324).

The trial did report fewer stage IV cancers overall in the intervention arm (342 versus 397; incidence-rate ratio 0.86, 95% CI 0.744–0.998) and more stage I/II diagnoses. This is an encouraging signal for further study, but the primary endpoint was not met. The study has not established that MCED screening reduces cancer mortality, and the test should not be presented as a replacement for established screening programmes or diagnostic evaluation of symptoms.

This is a useful example of why precision and early-detection reporting must distinguish a secondary endpoint or subgroup finding from the prespecified primary outcome, and why longer follow-up—including outcomes such as mortality, harms, false positives, overdiagnosis and cost-effectiveness—matters.

Sources: NHS-Galleri ASCO 2026 abstract; NHS-Galleri trial update.

10.2 Artificial intelligence in pathology and treatment selection

AI systems are being developed to analyse pathology slides, combine imaging with clinical variables, estimate prognosis, and identify features associated with treatment response. Their clinical maturity varies. For example, the US FDA cleared ArteraAI Breast in 2026 as a digital-pathology-based risk-stratification tool for defined patients with early-stage hormone receptor-positive, HER2-negative invasive breast cancer. The clearance defines an intended use; it does not mean that the model can independently select treatment or replace standard pathology and clinical assessment.

Other AI models reported at conferences may remain retrospective, exploratory or preclinical. A model's high accuracy in one dataset does not ensure performance in another hospital, population, scanner, laboratory or treatment era. For clinical decision support, external validation, calibration, assessment of bias, reproducibility and prospective evidence of clinical usefulness are important.

Source: ArteraAI Breast FDA clearance announcement; FDA 510(k) summary and intended use.

10.3 Multi-omics and single-cell analysis

Genomics measures DNA alterations; transcriptomics measures RNA expression; proteomics measures proteins; metabolomics measures small molecules involved in cellular metabolism. Combining these data may help clarify why two tumours with similar DNA findings behave differently, or why a resistant subpopulation persists. Single-cell methods can characterise cell-to-cell variation and interactions with the tumour microenvironment that are obscured when analysing bulk tissue.

These technologies are valuable research tools, but more data do not automatically improve care. Clinical adoption requires reproducible assays, interpretable results, prospective validation and evidence that the information changes management in a way that benefits patients.

10.4 Personalised neoantigen vaccines beyond melanoma

Personalised cancer vaccines are being studied in several tumour types. Early-stage research in pancreatic cancer has reported immune responses and encouraging follow-up in some patients, but those data come from small, selected cohorts and do not establish routine efficacy. Trial phase, sample size, control group, endpoint and follow-up should be reviewed before comparing these results with randomized phase III melanoma findings.

Terminology matters: an individualised therapeutic cancer vaccine is different from a vaccine intended to prevent an infectious disease or to prevent cancer in the general population. The therapy is designed to direct immune responses against selected tumour-associated neoantigens, often in combination with another treatment.

11. Evidence Snapshot: How Strong Are the Current Findings?

These categories are summaries of study design and evidence status, not a universal ranking of how much a treatment helps every patient. ESCAT actionability categories should not be substituted for trial-phase labels or evidence levels.

ESCAT v2

Evidence framework

Published framework, not a therapeutic trial. Helps classify biomarker actionability, tumour-agnostic matches and resistance evidence.

SANOVO

Phase III randomized trial

ESMO 2026 conference-reported results show improved PFS with savolitinib plus osimertinib in selected EGFR-mutated, MET-overexpressing NSCLC. Full publication and regulatory review remain important.

INTerpath-001

Phase III randomized trial

Improved RFS and DMFS reported for investigational intismeran plus pembrolizumab after resection of high-risk melanoma. OS is immature; regulatory status remains separate from conference efficacy results.

KEYNOTE-942

Phase IIb randomized trial

Five-year follow-up provides longer-term supportive data for personalised intismeran plus pembrolizumab in a smaller resected melanoma population. The five-year analyses were descriptive.

CUPISCO

Randomized phase II trial

Updated results support molecularly guided treatment in a selected CUP population, with improved PFS; the updated OS difference was not statistically significant.

NHS-Galleri

Large randomized screening trial

The primary endpoint for combined stage III/IV cancers was not met. Fewer stage IV diagnoses were observed; mortality, harms and cost-effectiveness need continued assessment.

Signatera CDx + atezolizumab

Regulatory-authorized specific use

FDA-authorized companion diagnostic for a defined MRD-positive muscle-invasive bladder cancer setting. Do not extrapolate to all MRD tests or malignancies.

AI prediction models

Tool-specific evidence

Some tools have regulatory clearance for specific intended uses; many other models remain investigational. Clearance does not equal universal clinical utility.

How to read conference news: look for the study design, number and characteristics of participants, prespecified primary endpoint, effect estimate and confidence interval, follow-up duration, safety, sponsor involvement, peer-reviewed publication status and any subsequent regulatory or guideline decision. A statistically significant result may still have modest absolute benefit, uncertain long-term effects or limited applicability outside the studied group.

12. Limitations and Challenges

Not every finding is actionable

Many tumours carry alterations for which no proven matched therapy exists. A variant of uncertain significance is not a treatment instruction. Experimental matches may be appropriate for a clinical trial discussion, but they should not be presented as equivalent to established treatment options.

Sample quality and assay limitations

Small biopsies, low tumour content, degraded DNA/RNA and the biology of ctDNA shedding can affect detection. A negative result may mean that no alteration was detected by that test in that sample; it does not always mean that the tumour lacks the alteration. A clinician may recommend repeat sampling or an alternative method where the result could change care.

Resistance and time

A tumour's molecular profile can change after treatment. A historical molecular report may no longer represent the dominant disease, particularly after several lines of therapy. Repeat testing is most useful when there is a realistic possibility that the result will change a decision.

Cost and unequal access

Comprehensive genomic profiling, specialist molecular tumour boards, matched drugs and clinical trials are not uniformly available or reimbursed. Access differs by health system, location, insurance coverage and tumour type. The most extensive test is not automatically the best choice if its result cannot be acted on.

Privacy and inherited findings

Genomic data may reveal information about both the person tested and biological relatives. Patients should understand how their data and samples may be stored or shared, and whether a tumour test could raise a question that requires germline confirmation. Direct-to-consumer testing is not a replacement for validated clinical testing or counselling.

Conflicts of interest and evidence communication

Many oncology trials and diagnostic platforms are developed by commercial sponsors. Sponsor-funded evidence can still be scientifically valuable, but readers should check whether results are reported by investigators, a peer-reviewed journal, a conference abstract or a company announcement. Independent replication, complete reporting and transparent disclosures improve confidence.

13. Questions to Ask Before Genomic Testing or a Matched Treatment

• What clinical decision could this test change?
• Is the test intended for this cancer type and treatment setting?
• Does it assess DNA, RNA, protein expression, germline variants, ctDNA or another marker?
• Is the detected alteration established as actionable, supported by emerging evidence, or still uncertain?
• What treatment is actually matched to this finding, and is it approved or guideline-supported where I am treated?
• What are the expected benefits, important harms and reasonable alternatives?
• Could a clinical trial be relevant if no standard matched treatment is available?
• Would a new biopsy, liquid biopsy or additional pathology test change the interpretation?
• Will the result have implications for relatives or require germline counselling?
• Who will review the report with me, and how will uncertain or incidental findings be handled?

14. The Future of Precision Oncology

Precision oncology is moving toward more integrated interpretation of clinical data, pathology, DNA and RNA sequencing, protein biomarkers, immune context and serial blood-based measurements. More mature tumour-agnostic evidence, resistance biomarkers, AI-assisted pathology and carefully validated MRD applications may make treatment selection more specific in some settings.

The major challenge is implementation. A useful precision-oncology system must provide a reliable assay, an interpretable report, expert review, an evidence-based treatment option, access to that option and follow-up to determine whether it is helping. If one of these links is missing, the presence of a genomic result may not improve patient outcomes.

The guiding question is therefore not simply, “What mutation does this tumour have?” It is: “Which result is reliable, which treatment or decision does it support, what evidence shows that this choice benefits patients like me, and what should happen if the cancer changes?”

15. Frequently Asked Questions

What is precision oncology?

Precision oncology uses molecular and clinical information to refine cancer diagnosis, estimate risk, select treatment and monitor disease. It complements, rather than replaces, pathology, cancer type, stage and the patient's overall clinical situation.

What is the difference between cancer genetics and cancer genomics?

Cancer genetics often refers to inherited variants that affect cancer risk or treatment, while tumour genomics examines a broader molecular profile that may include acquired mutations and other alterations. Both can be relevant to care, and tumour testing can sometimes raise a question that requires confirmatory germline testing.

Does every mutation have a targeted treatment?

No. Many alterations have no proven matched therapy, and the same gene can contain variants with different meanings. A result must be interpreted in the context of cancer type, exact variant, assay, treatment history and clinical evidence.

What is ESCAT v2?

ESCAT v2 is ESMO's updated framework for ranking the clinical actionability of molecular targets and describing evidence associated with resistance. It helps communicate evidence strength but is not, by itself, a drug approval or a guarantee of benefit.

What did ESMO 2026 show about lung cancer precision treatment?

The phase III SANOVO trial reported improved progression-free survival with savolitinib plus osimertinib compared with osimertinib alone in selected treatment-naive patients with EGFR-mutated, MET-overexpressing advanced NSCLC. The finding should be interpreted using the trial's MET IHC criteria and in light of full publication and local treatment guidance.

Is the personalised melanoma mRNA vaccine approved?

Intismeran autogene remains investigational as of the date of this article. The phase III INTerpath-001 trial reported improved recurrence-free and distant metastasis-free survival when it was added to pembrolizumab after melanoma resection, but overall-survival data were immature at the reported analysis. Regulatory approval and availability must be checked separately.

Can a liquid biopsy replace a tissue biopsy?

Not always. Blood-based ctDNA testing can be useful when tissue is unavailable or to identify changes over time, but some tumours shed little DNA into blood and not every alteration is detectable by every assay. A negative result does not always exclude a target.

Does a positive MRD test mean that cancer will definitely return?

No. MRD results indicate risk or evidence of tumour-derived material in a defined context, but their meaning depends on the assay, cancer type and evidence base. The decision to treat based on an MRD result should follow evidence and guidance for that specific setting.

Can AI accurately choose the best cancer treatment?

Some AI tools have regulatory clearance for narrowly defined uses, but many models remain in research. Performance in one dataset does not guarantee reliability in another population or clinic. AI output should be interpreted alongside established diagnostics and specialist judgement.

Should everyone with cancer get comprehensive genomic profiling?

Not necessarily. It is most useful when the result could guide an evidence-supported therapy, refine diagnosis, affect a relevant clinical-trial option or answer another actionable question. The decision depends on cancer type, stage, prior treatment, sample availability, cost and local access.

16. Using AI Assistants to Understand a Cancer Genomics Report

AI tools can help explain technical terminology, organise questions and summarise cited sources, but they may misread a report, confuse similarly named variants, cite outdated approvals or overstate early trial results. Do not upload identifiable medical records unless you understand the service's privacy settings and terms.

A safer way to use an AI assistant is to remove identifying details and ask it to explain the report without making a treatment recommendation. For example:

Example prompt

“Explain the attached cancer genomic report in plain language. For each finding, distinguish the exact alteration, whether it is somatic or may require germline confirmation, the evidence for clinical actionability, and whether the evidence applies to this cancer type and treatment setting. Separate approved treatment options from clinical-trial hypotheses and variants of uncertain significance. Cite primary sources or regulatory guidance, state the publication date, and flag any uncertainty. Do not recommend that I start, stop or change treatment; list questions to discuss with my oncology team.”

An oncologist, molecular tumour board or certified genetic counsellor can help determine whether a finding is meaningful for an individual's care. AI should be used to prepare for that conversation—not replace it.

17. Selected References and Further Reading

The sources below prioritise peer-reviewed publications, official regulatory records and primary trial or congress reports. Some ESMO 2026 efficacy data were initially reported at a scientific meeting and through sponsor announcements; publication and regulatory status should be rechecked as evidence evolves.

1. ESMO. The ESMO Scale for Clinical Actionability of Molecular Targets, version 2 (ESCAT v2): an evidence-based framework for ranking alterations as actionability and resistance biomarkers in precision oncology. Annals of Oncology. 2026. doi:10.1016/j.annonc.2026.09.008.
2. ESMO. ESMO updates ESCAT framework for precision oncology: new guidance on actionability and resistance biomarkers. 22 October 2026. ESMO summary.
3. ESMO. ESMO Congress 2026: registration and meeting information. Official congress dates and details.
4. HUTCHMED. SANOVO phase III data at ESMO. ESMO Congress 2026, abstract LBA69. Sponsor announcement.
5. Merck and Moderna. INTerpath-001 phase III results for intismeran autogene plus pembrolizumab at ESMO 2026. Trial announcement.
6. The ASCO Post. High-Risk Melanoma: Personalized mRNA Vaccine Plus Immunotherapy Reduces Risk of Recurrence After Surgery. October 2026. Report.
7. Khattak A, et al. Intismeran Autogene Plus Pembrolizumab Versus Pembrolizumab Alone in High-Risk Resected Melanoma: 5-Year Update of the Randomized Phase IIb KEYNOTE-942 Study. Journal of Clinical Oncology. 2026;44:2482–2489. doi:10.1200/JCO-26-00835.
8. Krämer A, et al. Molecularly Guided Therapy Versus Continued Chemotherapy in Unfavorable Cancer of Unknown Primary: Updated Efficacy and Safety From the Randomized, Phase II CUPISCO Study. Journal of Clinical Oncology. 2026;44:2895–2902. doi:10.1200/JCO-25-02974.
9. US Food and Drug Administration. FDA approves atezolizumab for adjuvant treatment of muscle-invasive bladder cancer in patients with molecular residual disease. 15 May 2026. FDA announcement.
10. US Food and Drug Administration. Signatera CDx Premarket Approval record. FDA device record.
11. US Food and Drug Administration. Guardant360 Liquid CDx Premarket Approval record. FDA device record.
12. NHS-Galleri Trial investigators. Primary results from a randomised controlled trial to assess the clinical utility of a multi-cancer early detection test in population screening. ASCO Annual Meeting 2026, abstract LBA100. ASCO abstract.
13. NHS-Galleri trial. First full results published in leading journals. 23 September 2026. Trial update.
14. Stein RC, et al. First results from the OPTIMA phase III randomized non-inferiority trial of test-directed chemotherapy in ER-positive, HER2-negative early breast cancer. Journal of Clinical Oncology. 2026;44(suppl 16):500. ASCO Post coverage.
15. Artera. Artera Receives U.S. FDA Clearance for ArteraAI Breast. 6 May 2026. Company announcement; FDA 510(k) summary.
16. Merker JD, et al. Circulating Tumor DNA Analysis in Patients With Cancer: American Society of Clinical Oncology and College of American Pathologists Joint Review. Journal of Clinical Oncology. 2018. doi:10.1200/JCO.2017.76.8671.
17. Wan JCM, et al. Liquid biopsies come of age: towards implementation of circulating tumour DNA. Nature Reviews Cancer. 2017. doi:10.1038/nrc.2017.7.
18. Hanahan D, Weinberg RA. Hallmarks of cancer: the next generation. Cell. 2011;144:646–674. doi:10.1016/j.cell.2011.02.013.
19. Vogelstein B, et al. Cancer genome landscapes. Science. 2013;339:1546–1558. doi:10.1126/science.1235122.
20. Alexandrov LB, et al. Signatures of mutational processes in human cancer. Nature. 2013;500:415–421. doi:10.1038/nature12477.

Medical disclaimer: This article is for general educational purposes only and does not constitute medical advice, diagnosis or treatment recommendations. Genomic testing, targeted therapy, immunotherapy, MRD monitoring and clinical-trial eligibility are individual decisions that depend on cancer type, stage, prior treatment, test limitations and overall health. Discuss results with a qualified oncology team and, where appropriate, a certified genetic counsellor. Do not start, stop or delay treatment on the basis of this article or an AI-generated interpretation alone.

Editorial transparency: This guide synthesises peer-reviewed publications, official regulatory sources, society materials, congress reports and sponsor announcements. Where a cited data set was presented at a meeting or reported by a sponsor, the article labels that status and distinguishes it from established clinical recommendations.