Abstract Management of multiple primary tumours is an increasing clinical burden as the population of cancer survivors rises globally. Despite standard-of-care histological, radiological and panel sequencing diagnostics, identifying multiple primary cancers remains clinically challenging. We characterize multiple primary tumours in the TRACERx study. We integrate baseline and follow up clinical data, centrally-reviewed histopathological assessments with whole-exome sequencing (WES) to identify patients with multiple primary tumours from lung and other sites. Where multi-region sample collection was available, we assessed the genomic relatedness of multiple lesions from the same patient through shared clonal mutations. 205 (24% of 844) TRACERx patients had clinically-diagnosed or genomically-discovered multiple primary cancers. Of these, 82 patients had a prior cancer diagnosis, 56 patients had synchronous primaries at study enrollment, and 95 developed new primaries during follow up. Lung was the most common organ site for additional primaries, with 98 patients having two or more lung primary tumours, including the lung tumour profiled for the TRACERx study. High resolution tumour WES revealed clinical misclassification in 21 of 138 patients (15.2%) with multiple sequenced lesions. At baseline, clinical classification favoured relatedness between lesions: two patients diagnosed with intrapulmonary metastases actually had independent primary lung cancers. In addition, eight patients with lesions staged histopathologically as single tumours, some with microscopic evidence of histological heterogeneity, were confirmed through WES to be collision tumours composed of 2 to 3 independent lineages. One patient was discovered to have had synchronous lung primaries, undetectable clinically, through WES sequencing of lymph node metastases. During follow up, clinical classification tended to favour independent primary designation, with 6 of 10 discrepancies due to clinically-diagnosed metachronous primary tumours found through WES to be metastases. Together, the TRACERx multiple primary cohort represents one of the largest multi-region cohorts of multiple primary lung cancers analyzed to date. These data demonstrate the clinical utility of WES in diagnosing independent primary cancers. Ongoing analysis will investigate the association between history of multiple primary cancers and germline cancer susceptibility and patient outcome. Deep, multi-region molecular characterization, available for 58 patients with multiple primary lung cancers, will also allow detailed exploration of the molecular mechanisms driving the evolution of multiple primary cancers, informing potential biomarkers to highlight patients with elevated risk of secondary primary tumours and improve subsequent clinical management. Citation Format: Lydia Y. Liu, Charlotte Grieco, Adriana Salcedo, Takahiro Karasaki, Cristina Naceur-Lombardelli, Oliver Shutkever, Ariana Huebner, Carlos Martinez Ruiz, Sara Waise, Aino-Maija Leppä, Olivia Lucas, Tej Pandya, Selvaraju Veeriah, Sophia Ward, Kerstin Haase, Alexander A. Azizi, Woody Z. Zhang, Emma Hazelwood, Alexandar M. Frankell, TRACERx Consortium, Allan Hackshaw, Nicholas McGranahan, David A. Moore, Mariam Jamal-Hanjani, Charles Swanton. Clinico-genomic characteristics of multiple primary cancers in TRACERx [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2026; Part 1 (Regular Abstracts); 2026 Apr 17-22; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2026;86(7 Suppl):Abstract nr 6793.

Limited understanding of the biological processes that govern metastatic dissemination hinders its prevention and treatment1. Here, using 501 longitudinally collected primary and metastatic tumour samples from 24 patients with non-small cell lung cancer (NSCLC) enrolled in the TRACERx lung study and PEACE autopsy programme, we infer tumour evolution from diagnosis to death. With DNA-sequencing data encompassing 70% of the metastases that were radiologically detected before death and paired multi-region sampled primary tumours, we show that the genomes of metastases diverge markedly from those of their ancestral primary tumour, with additional driver alterations and genome doubling events occurring after metastatic dissemination. In 62.5% of patients, multiple primary tumour subclones disseminated, each founding a distinct metastasis. These metastases served as sources of onward spread: more than half of the metastases sampled were seeded by other metastases. The duration that metastases existed in situ influenced their likelihood of seeding further metastases. Most metastatic migrations started and ended in the same anatomical cavity. The few subclones that exited the thorax to seed metastases disseminated widely and were enriched for somatic copy-number alterations, suggesting that chromosomal instability may facilitate extrathoracic spread. This spatial and temporal evolutionary analysis sheds light on the extent of metastatic diversity and seeding in advanced NSCLC-which tends to be underestimated in single metastasis biopsies-and identifies genomic and clinical mediators of metastatic progression.

Whether support from artificial intelligence (AI) models improves breast screening is a topic of research in national cancer screening programmes and clinical oncology. Three studies now show that AI tools can assist radiologists with evaluating mammograms, substantially reducing workloads and possibly improving screening performance.

BACKGROUND:Locally advanced rectal cancer is routinely treated with neoadjuvant radiotherapy. Concomitant systemic anticancer therapy with standard fluoropyrimidines has not generally improved outcomes. Small studies reported high pathological complete response rates and acceptable toxicity using irinotecan and fluoropyrimidine chemoradiation. We aimed to explore the effect of the addition of concomitant irinotecan to standard-of-care chemoradiotherapy in patients with locally advanced rectal cancer. METHODS:ARISTOTLE is a multicentre, open-label, parallel-design, phase 3, randomised controlled trial conducted at 75 UK hospital sites treating patients with rectal cancer. Patients were eligible if they were aged at least 18 years with MRI-defined, locally advanced rectal cancer threatening or involving resection margins without metastases. Patients were randomly assigned (1:1) to preoperative radiotherapy 45 Gy in 25 daily fractions, combined with either oral capecitabine 900 mg/m2 alone twice daily on weekdays (standard-of-care group) or oral capecitabine 650 mg/m2 twice daily on weekdays plus intravenous irinotecan 60 mg/m2 once weekly in weeks 1-4 (irinotecan group). Randomisation was done centrally, with allocation concealed to investigators; masking of patients and treating clinicians was not feasible. Stratification was by radiotherapy centre, mesorectal fascia involvement, and presence or absence of equivocal metastatic disease. The primary endpoint was disease-free survival, with analysis done on a modified intention-to-treat basis (ie, all eligible patients who were randomly assigned to treatment). Safety analyses were conducted in all patients who started protocol treatment (as-treated population). For time-to-event endpoints, patients without an event were censored at the date they were last known to be alive; missing radiological and pathological response assessments were classified as non-response. The study is registered with EudraCT, 2008-005782-59, and is complete. FINDINGS:Between Oct 25, 2011, and July 5, 2018, 589 patients were randomly assigned to treatment; after exclusions, 564 were included in the modified intention-to-treat population (284 in the standard-of-care group and 280 in the irinotecan group). 370 (66%) patients were male, 194 (34%) were female, and median age was 61 years (IQR 54-68); ethnicity data were not collected. Staging in both groups was similar: 223 (79%) patients in the standard-of-care group and 212 (76%) in the irinotecan group had mrT3 tumours; 44 (15%) and 45 (16%) had mrT4 tumours, respectively. Patients in the irinotecan group were less likely to receive 45 Gy radiotherapy than in the standard-of-care group (208 [75%] of 276 vs 251 [89%] of 283) or at least 90% of the planned capecitabine dose (187 [68%] of 276 vs 253 [89%] of 283). With a median follow-up was 78 months (95% CI 75-86), 36-month disease-free survival was 68% (95% CI 63-73) in the irinotecan group versus 67% (61-72) in the standard-of-care group (hazard ratio 0·91 [95% CI 0·68-1·23]; p=0·54). Deaths occurred in 88 (31%) patients in the irinotecan group and 92 (32%) in the standard-of-care group; rectal cancer was the leading cause, occurring in 59 (67%) patients in the irinotecan group and 67 (73%) in the standard-of-care group. There were five deaths related to protocol treatment: three in the irinotecan group (two had a thromboembolic event and one had multiorgan failure plus sepsis) and two in the standard-of-care group (one cardiac arrest and one febrile neutropenia). Grade 3 or worse were more frequent in patients receiving irinotecan than those receiving standard of care (215 [78%] vs 148 [52%]), including haematological investigations (235 [85%] vs 109 [39%]) and gastrointestinal events (57 [21%] vs 35 [12%]), notably diarrhoea (38 [14%] vs ten [4%]), lymphocyte count decreased (181 [66%] vs 100 [35%]), and neutrophil count decreased (27 [10%] vs three [1%]). INTERPRETATION:For MRI-defined, high-risk, locally advanced rectal cancer, the addition of irinotecan to standard of care did not improve outcomes and should not be used in combination with radiotherapy plus capecitabine. FUNDING:Cancer Research UK.

Abstract Background: Sensitive and scalable detection of circulating tumour DNA (ctDNA) is essential for minimal residual disease (MRD) assessment following curative-intent treatment in non-small cell lung cancer (NSCLC). While personalised, tumour-guided approaches are highly sensitive, they typically rely on bespoke panel design, limiting scalability and constraining the breadth of biological signals that can be interrogated. Error-corrected whole-genome sequencing (WGS) offers the potential to overcome these limitations by enabling personalised MRD detection without custom sequencing panels, while supporting genome-wide and tumour-agnostic analyses. Methods: We analysed personalised WGS-based ctDNA detection using Ultima Genomics’ ppmSeq in a TRACERx pilot cohort comprising 45 post-operative plasma samples from individuals with early-stage NSCLC during the landmark period. Landmark was previously defined as between days 10 and 120 post-operatively, prior to the start of adjuvant therapy or disease recurrence. Tumour-specific variants derived from tumour WGS were queried in genome-wide matched plasma. Plasma WGS was performed to a target coverage of 150x. Analytical sensitivity, ctDNA fraction distributions, and associations with relapse-free survival were assessed. Results: Error-corrected WGS enabled detection of tumour-informed variants at parts per million (ppm) allele fractions without bespoke panel design. Among MRD-positive samples, 13% exhibited ctDNA fractions below 10 ppm, and 72% were below 100 ppm, highlighting sensitivity across the ultra-low ctDNA range. The lowest ctDNA fraction called was 2.2 ppm. Landmark ctDNA detection by WGS was significantly associated with relapse-free survival, with a median relapse-free survival of 23.2 months in landmark-positive individuals (n = 37) compared with 98.7 months in landmark-negative individuals (n = 8, log-rank p = 0.042). Conclusions: These data provide early validation of personalised, tumour-guided ppmSeq ctDNA detection in early-stage NSCLC, with ctDNA quantification to parts per million. This approach is now being expanded to include up to 400 individuals from TRACERx (>1,500 plasma timepoints) to evaluate prognostic performance and scalability, positioning plasma WGS as a next-generation MRD platform that unifies personalised and tumour-agnostic monitoring from a single assay, including genome-wide error-corrected mutations, copy number, and fragmentation features for residual disease and relapse monitoring. Citation Format: Jonathan Wan, Sophie Ward, Alexander Azizi, Hila Benjamin, Raju Veeriah, Siân Harries, Jeanette Kittel, Nika Iremadze, Itai Rusinek, Gat Kreiger, Ariel Jaimovich, Jacqui Shaw, Allan Hackshaw, Nnennaya Kanu, Elena Helman, Mariam Jamal-Hanjani, Charles Swanton. Error-corrected plasma whole-genome sequencing for personalised and tumor-agnostic minimal residual disease detection in NSCLC [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2026; Part 2 (Late-Breaking, Clinical Trial, and Invited Abstracts); 2026 Apr 17-22; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2026;86(8_Suppl):Abstract nr LB116.

BACKGROUND:Smokers are eligible for lung cancer screening using low dose CT (LDCT) if they have a sufficiently high risk of lung cancer. In England, for example, people are eligible if their estimated risk using either of two validated models (PLCOm2012 and Liverpool Lung Project) exceeds a certain threshold. Risk is often estimated using only cigarette consumption (duration and number per day), yet many smokers also use other cancer-causing tobacco products. METHODS:SUMMIT is a prospective cohort study of a LDCT screening service. We evaluated the impact of cigars, pipes, and cigarillos on lung cancer risk. RESULTS:Among 13,000 participants who were current/former cigarette smokers, 511 (1 in 25) were also regular users of cigars, pipes, or cigarillos. The 5-year cumulative lung cancer incidence in people who currently smoked both types of tobacco products was 8.85%, versus 5.74% for cigarette only current smokers. Among these 511 participants, the percentage whose risk was below the screening eligibility threshold ignoring other tobacco products which then exceeded the threshold including these products was 0.6% using LLPv2, and 8.2% using PLCOm2012. CONCLUSIONS:All tobacco consumption should be considered when determining screening eligibility to avoid missing those who can benefit from LDCT. CLINICAL TRIAL REGISTRATION:Clinicaltrials.gov no: NCT03934866.

The prognosis of newly diagnosed metastatic prostate cancer is highly variable. The primary objective of the PARADIGM prospective cohort study was to evaluate predictors of survival in blood collected at the start of each of the first six treatment cycles from 114 patients with high-volume metastatic prostate cancer (biologically male) who were starting androgen deprivation therapy in combination with docetaxel or an androgen receptor pathway inhibitor. Here circulating tumor DNA (ctDNA) was detected in 29% of patients after 6-12 weeks of combination therapy (compared to 70% before any treatment) and associated with 12 month overall survival of 73% versus 99% for patients who were ctDNA-negative and 24 month survival of 50% versus 85%. The secondary objective was to test ctDNA with serum prostate-specific antigen (PSA). In multivariable models, both were independent risk factors on combination treatment with a hazard ratio of death of 20.34 for the poorest prognosis group, but only ctDNA was associated with shorter survival on androgen deprivation before the start of combination therapy. Using ctDNA with serum PSA and clinical characteristics can improve the accuracy of survival prediction and should be evaluated for ctDNA-informed treatment modification. ClinicalTrials.gov: NCT04067713 .

Although immunotherapy has revolutionized cancer treatment, many patients still experience limited benefit, highlighting the urgent need for improved biomarkers1. Although immunotherapy is founded on unleashing T cells2, most existing biomarkers remain tumour-centric and mainly overlook host immune competence. The thymus is a key immune organ that is crucial for T cell maturation, and we hypothesized that thymic functionality is associated with immunotherapy outcomes3. Here we show that thymic health, a radiographic measure of thymic functionality, is strongly associated with immunotherapy outcomes across several cancer types. Using a deep-learning framework applied to routine computed tomography images, we quantified thymic health in a pan-cancer cohort of 3,476 patients receiving immune checkpoint inhibitors. In patients with non-small cell lung cancer, higher thymic health was associated with reduced risks of progression and all-cause mortality. These associations remained significant across clinically relevant levels of programmed death ligand 1 (PD-L1) and tumour mutation burden. In the prospective TRACERx lung cancer study, thymic health was positively associated with T cell receptor diversity and T cell receptor excision circles, and correlated with immune-system signalling pathways, supporting radiographic thymic health as a proxy for thymic activity and adaptive immune competence. Analysis across patients with melanoma, breast cancer or renal cancer demonstrated pan-cancer relevance. Together, these findings identify thymic health as a previously unrecognized, tumour-agnostic determinant of immunotherapy efficacy, with potential implications for patient stratification, treatment timing and the development of immune-rejuvenating strategies in precision immuno-oncology.

BACKGROUND:Lung cancer screening is now widely adopted across healthcare systems. Screening typically occurs in asymptomatic individuals. As symptoms of lung cancer overlap with those of chronic conditions, defining asymptomatic screening is challenging. This study outlines the frequency of symptomatic participants in a lung cancer screening trial. RESEARCH QUESTION:How common are respiratory and red flag symptoms in lung cancer screening participants and how do these symptoms impact screening outcomes? STUDY DESIGN AND METHODS:SUMMIT is a prospective observational cohort study to assess the implementation of Low-Dose Computed Tomography (LDCT) screening for lung cancer. Baseline clinical assessments collected self-reported symptoms, medical history, demographics, and spirometry. Haemoptysis (in last year) and unintentional weight loss (≥5kg in 3 months) were classified as red flag symptoms, and cough (acute: onset less than six weeks, or chronic: >6 weeks) and dyspnoea (modified Medical Research Council scale≥1) classified as non-specific symptoms. Lung cancer diagnoses within one year were ascertained. Multivariable logistic regression was used to assess associations between symptoms and lung cancer. RESULTS:Among 13,035 participants eligible for a baseline LDCT 76% (N=9,859) reported at least one symptom. Cough was present in 36% (N=4,707) and 66% of participants reported dyspnoea. Only 6% of participants reported red flag symptoms, including hemoptysis and weight loss. The presence of any of these symptoms was associated with higher likelihood of lung cancer diagnosis in the year following assessment (OR 1.45, p=0.03, adjusted for other baseline factors). INTERPRETATION:Symptoms are commonly reported in those undergoing lung cancer screening and those with symptoms are more likely to be diagnosed with lung cancer.

Background: Patients with advanced NSCLC and ECOG PS 3 are often excluded from clinical trials, resulting in limited randomised evidence to guide treatment decisions. We evaluated survival and toxicity outcomes in patients with advanced NSCLC and ECOG PS 3 at diagnosis who were ineligible for chemotherapy and enrolled in the TOPICAL trial. Methods: We performed a subgroup analysis of TOPICAL, a phase 3 multicentre, double-blind, placebo-controlled randomised trial conducted at 78 hospitals in the UK between April 2005 and April 2009. Eligible patients had stage IIIB or IV NSCLC, were chemotherapy-naive, and were considered unsuitable for chemotherapy because of frailty, poor performance status, or comorbidity. This analysis included 192 of 670 randomised patients (29%) with ECOG PS 3 at diagnosis. Subgroup analyses by performance status were prespecified in the statistical analysis plan. Patients received oral erlotinib 150 mg daily or matched placebo, both with best supportive care. Outcomes included overall survival (OS), progression-free survival (PFS), treatment delivery, and adverse events. An additional subgroup analysis examined outcomes according to the development of first-cycle rash within 28 days. Findings: Among 192 patients with ECOG PS 3 (102 assigned erlotinib and 90 placebo), baseline characteristics were well balanced between groups. Median age was 76 years, and most were current or former smokers. In the intention-to-treat population, median OS was 1.9 months with erlotinib vs 2.1 months with placebo (HR 0.91; 95% CI, 0.69-1.22; p= 0.55). Median PFS was 1.6 months vs 1.5 months (HR 0.79; 95% CI, 0.59-1.06; P= 0.12). Twenty-four patients (12%) deteriorated or died before starting treatment (16 of 102 [16%] erlotinib; 8 of 90 [9%] placebo). Grade 3 or 4 adverse events were similar (57 of 102 [56%] vs 54 of 90 [60%]). Among rash-evaluable patients, first-cycle rash with erlotinib was associated with longer OS than placebo (HR 0.60; 95% CI, 0.39-0.92; P= 0.02) and longer PFS (HR 0.54; 95% CI, 0.35-0.82; P=0.004). Interpretation: Patients with advanced NSCLC and ECOG PS 3 had very poor survival, with a median about OS of about 2 months, highlighting profound frailty and a very narrow therapeutic window for intervention. Although erlotinib did not improve OS in the intention-to-treat subgroup, a subset of patients might still derive benefit from prompt, tolerable systemic therapy. These results support the inclusion of patients with ECOG PS 3 in carefully designed clinical trials and treatment pathways that prioritise rapid treatment initiation.

Rare paediatric central nervous system (CNS) tumours comprise a biologically-heterogeneous group of low-incidence diseases that present significant challenges to both therapeutic development and clinical research, limiting the availability of high-quality evidence to guide clinical care. The UK3CR Children’s Cancer Research Group (CRG) CNS Tumours Subgroup convened a multidisciplinary workshop to identify research priorities and establish a strategic framework for advancing clinical trials in this setting. Four high-priority tumour groups—craniopharyngioma, choroid plexus carcinoma (CPC), very high-risk medulloblastoma (VHR-MB), and rare embryonal and sarcomatous tumours (REST), including embryonal tumour with multilayered rosettes (ETMR)—were evaluated. Key outcomes included the feasibility of a UK-led craniopharyngioma trial, the need for enhanced national CPC data collection, and support for European early-phase platform trials in VHR-MB. REST highlighted transnational regulatory complexity requiring coordinated registries and parallel trial models. Cross-cutting barriers included regulatory discordance, contracting delays, fragmented funding, pharmacovigilance differences, data-sharing constraints, and interoperable infrastructures. The workshop emphasised innovative methodologies, including Bayesian/adaptive designs, platform trials, and external control arms, to maximise efficiency in small populations, alongside strengthened patient involvement and international collaboration. Collectively, these findings define a collaborative and internationally-aligned strategic roadmap to accelerate clinical research and improve outcomes in rare paediatric CNS tumours.

To determine if a pragmatic, conservative approach to managing < 3 cm anterior mediastinal lesions in lung cancer screening (LCS) is safe. 55- to 77-year-old current or former smokers underwent low-dose computed tomography (LDCT) screening. Anterior mediastinal lesions < 3 cm at baseline were managed conservatively with annual LDCT follow-up for up to 2 years. Lesions ≥ 3 cm at baseline, growing during follow-up (based on visual assessment), or demonstrating concerning radiological characteristics were referred for further assessment. Outcomes for all anterior mediastinal lesions were assessed using follow-up LDCT images, electronic health records and the national cancer registry. Descriptive frequencies were calculated for all reported outcomes. The baseline prevalence of anterior mediastinal lesions was 0.7

Abstract Background: Immune checkpoint inhibitors (ICIs) depend on effective T-cell priming and renewal, yet current biomarkers focus primarily on tumor-intrinsic features (PD-L1, TMB). Because the thymus maintains naïve T-cell output and repertoire diversity, we hypothesized that an individual's thymic health represents a host-intrinsic determinant of ICI benefit. Methods: We developed a self-supervised deep learning model to quantify thymic function, here termed thymic health, from routine CT scans. Thymic health was independently evaluated in 3,476 patients treated with ICI for tumors of various entities. Associations with PFS and OS were assessed using multivariable Cox models adjusting for key epidemiological and clinical factors. Sensitivity analyses evaluated the incremental contribution of thymic health. Biological validation was performed in treatment-naïve NSCLC patients from TRACERx (n=464) using T-cell receptor (TCR) sequencing and plasma proteomics. Results: In patients with NSCLC (n=1,218), high thymic health was associated with significantly reduced risk of progression (HR 0.65; 95% CI, 0.54-0.77) and death (HR 0.56; 95% CI, 0.46-0.68); preserved after full multivariate adjustments for sex, age, PD-L1, TMB, ECOG, histology, treatment line and stratification by treatment type (P<0.001). Adding thymic health improved model fit beyond demographic, clinical, and tumor-intrinsic variables (likelihood ratio P<0.001). Importantly, including thymic health did not attenuate PD-L1 or TMB effects, indicating complementary, non-redundant prognostic information. Across PD-L1 and TMB strata, higher thymic health was consistently associated with improved PFS and OS (P<0.03). In TRACERx, thymic health was positively associated with T cell receptor excision circles, i.e., T cell output, and correlated with greater peripheral and intratumoral TCR diversity and with proteomic signatures of adaptive immune activation, biologically supporting its role as a radiographic proxy for immune competence. Pan-cancer analyses across 2,258 ICI-treated patients (melanoma, renal, breast, and others) showed consistent positive associations between thymic health and improved outcomes. Conclusions: Thymic health was consistently associated with improved immunotherapy outcomes across cancer types and added independent information beyond PD-L1, TMB, and standard clinical variables. These findings support thymic health as a possible non-invasive, host-intrinsic biomarker with potential to refine patient stratification and advance precision immuno-oncology. Clinical trial identification: ClinicalTrials.gov: TRACERx, NCT01888601 Citation Format: Simon Bernatz, Vasco Prudente, Suraj Pai, Asbjørn Kjær, Alessandro Di Federico, Andrew Rowan, Selvaraju Veeriah, Lars Dyrskjøt, Leonard Nürnberg, João Victor M. Alessi, Patrick Ott, Elad Sharon, Allan Hackshaw, Nicholas McGranahan, Christopher Abbosh, Raymond H. Mak, Danielle S. Bitterman, Mark M. Awad, Biagio Ricciuti, Charles Swanton, Mariam Jamal-Hanjani, Nicolai Juul Birkbak, Hugo JWL Aerts. Radiographic thymic health as host-intrinsic determinant of immunotherapy outcomes across cancer types [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2026; Part 1 (Regular Abstracts); 2026 Apr 17-22; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2026;86(7 Suppl):Abstract nr 1185.

Abstract Background: Circulating tumor DNA (ctDNA) offers a minimally invasive approach for response monitoring of cancer immunotherapy treatment and early prediction of therapeutic outcomes. However, the clinical utility of ctDNA-based liquid biopsy faces a critical challenge: reliable ctDNA detection in low-shedding tumors and in patients with low molecular residual disease (MRD) following treatment response. We employed an ultrasensitive ctDNA assay to address this technical limitation, enabling precise longitudinal monitoring essential for optimizing IO therapy. Methods: We analyzed longitudinal plasma samples from 41 patients with non-small cell lung cancer (NSCLC) [adenocarcinoma (n=19), non-adenocarcinoma (n=22)], who received IO monotherapy (n=33) or combined IO and chemotherapy (n=8) in the Deciphering Anti-tumour Response and Resistance With INtratumour Heterogeneity (DARWIN 2) trial nested within the TRACERx study. Using NeXT Personal®, an ultra-sensitive personalized liquid biopsy approach, we tracked up to ∼1,800 patient-specific somatic variants per case across 233 plasma samples. Results: The median limit of detection across all tests was 1.52 parts per million (PPM), enabling ctDNA detection across six orders of magnitude (range 2.1-309,673 PPM), with 21% of positive ctDNA detections in the ultrasensitive range (<100 PPM), with that increasing to 28% while on treatment. Histological subtype had no impact on detection with this assay. Early molecular response (mR), defined as either >50% reduction in ctDNA or sustained ctDNA negativity from pre-treatment baseline to the subsequent plasma sample (median interval: 43.5 days), was significantly associated with improved clinical outcomes. Patients achieving early mR exhibited superior progression-free survival (PFS; HR = 0.33, 95% CI 0.14-0.77, p = 0.010) and overall survival (OS; HR = 0.31, 95% CI 0.14-0.70, p = 0.005). All patients with complete response by RECIST criteria achieved early mR (sensitivity = 100%). Conversely, disease in all patients lacking mR progressed within 15 months. Furthermore, durable molecular complete response (dmCR), defined as ctDNA negativity maintained for ≥180 days, was strongly associated with improved survival outcomes (2-year PFS: 50% vs. 9%, HR = 0.31 95%CI 0.11-0.90, p = 0.032; 2-year OS: 86% vs. 13%, HR = 0.06, 95% CI 0.01-0.48, p = 0.007 for dmCR vs. non-dmCR, respectively). Conclusions: Early ctDNA kinetics serve as a robust predictor of long-term immunotherapy outcomes in patients with advanced NSCLC. The ability to detect ultra-low ctDNA levels allowed for accurate assessment of minimal residual disease, irrespective of lung cancer histology. These findings establish ultrasensitive ctDNA monitoring as a valuable tool for precise, real-time evaluation of immunotherapy response, with implications for clinical decision-making. Citation Format: Kishen R. Patel, Bailiang Li, Charles W. Abbott, Cristina Naceur-Lombardelli, Sadegh Saghafina, Sevasti Galani, James R. M. Black, Wing Liu, Nicola Steele, Gillian Price, Shobhit Baijal, Dean Fennell, Matthew G. Krebs, Tanya Ahmad, Alexandra Pender, Siow M. Lee, Mariam Jamal-Hanjani, Nicholas McGranahan, Allan Hackshaw, Sean Michael Boyle, Richard O. Chen, Charles Swanton, Crispin T. Hiley. Ultrasensitive ctDNA monitoring predicts early response of immunotherapy in recurrent metastatic non-small cell lung cancer [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2026; Part 1 (Regular Abstracts); 2026 Apr 17-22; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2026;86(7 Suppl):Abstract nr 3851.

The standard treatment for stage I lung adenocarcinoma is surgical resection, in most cases without additional systemic adjuvant treatment. A significant proportion of stage I cases recur with a less than 50% 5-year survival rate. There are clinical data suggesting that adjuvant treatment may improve survival in such recurrent cases. However, previously evaluated predictors such as the IASLC grading system from histological sections and transcriptomic profiles have not been sufficiently accurate and consistent for risk stratification and to guide therapeutic interventions. We hypothesized that these previously investigated diverse diagnostic measurements carry complementary information that may provide higher prognostic power when combined. Here we describe a multimodal deep learning method, PATH-ORACLE. This biomarker is built on top of the prospectively validated transcriptomic-based ORACLE score with the addition of routine histological sections processed by pre-trained foundation models. PATH-ORACLE predicts recurrence with an accuracy of over 85% in two independent cohorts. Given further validation this predictor could be used to prioritize stage IB patients for adjuvant chemotherapy in a more consistent fashion. Furthermore, for stage IA cases, PATH-ORACLE, combined with liquid biopsy-based monitoring may help identify high-risk patients suitable for adjuvant targeted therapy.