Collaborative networks and initiatives

The United Kingdom is well-placed to lead translational research and innovative trials with global impacts on patient outcomes. The NHS offers a unified healthcare service covering a population of over 60 million, with existing links between cancer centres, neuroscience centres and academic units. Almost all patients are diagnosed within the NHS allowing for excellent capture and integration of imaging, pathology and clinical data. Clinical trials are embedded within care pathways, and access to trials is increasing via initiatives like NIHR's ‘be part of research’ [10]. UK trials provide true standard of care (SOC) comparator arms in almost all patients owing to the harmonised nature of UK training and clinical practice, including minimal off-label patient-funded drugs, and testing and treatment without requiring health insurance coverage. Primary and post-primary care integration with limited points of entry allows complex queries to be addressed, including patient-oriented research questions and pre-diagnosis journeys.

Since 2018, the United Kingdom has developed several clinical/research collaborations. The Tessa Jowell Brain Cancer Mission (TJBCM) is a national initiative supporting clinical studies to provide platforms for facilitating patient enrolment in biomarker-driven trials. Two examples are BRAIN-MATRIX [11] and the Minderoo Precision Brain Tumour Programme [12]. BRAIN-MATRIX is a 10-centre trial platform (with 4 more centres planned) including advanced molecular profiling, which has recruited 395 patients and provided the basis for several clinical trials (ARISTOCAT, DETERMINE and 5G). The Minderoo Precision Brain Tumour Programme [12] enrolled 230 patients in the first 2 years, exceeding the target of 125 patients, with whole genome and transcriptome sequencing data provided with a 3-week turnaround and a second arm now opening. Other TJBCM programmes include the Brain Tumour Research Novel Therapeutics Accelerator (BTR-NTA), which launched in 2023 and aims to de-risk drug or device development by offering up to 240 h of free (to academics), systematic multidisciplinary evaluation and feedback [13]; NHS clinical neuro-oncology service Centres of Excellence, a designation awarded to 17 UK centres between 2020 and 2022 (next application round in 2024) to acknowledge standards of excellence in clinical practice, patient care, staff training opportunities, access to clinical trials and research opportunities, which go beyond today's existing guidelines [14]; and a dedicated NHS clinical fellowship training programme, which awarded two fellowships in the first round in 2023. Neuro-oncology Research Centres of Excellence have also been funded by CRUK (n = 2) and BTR (n = 4, with plans for 3 more) [15-17]. International networks for pre-clinical and clinical studies include UK members. The global Glioma Longitudinal AnalySiS (GLASS) consortium [18] analyses longitudinal datasets to refine molecular profiling and tumour evolution and includes 3 UK centres, and the Brain Liquid Biopsy Consortium [19] was co-founded in the United Kingdom and aims to accelerate research and translation of neuro-oncology biofluid biomarkers. The EORTC Brain Tumour Group is a European-led clinical trial collaborative with UK representation on 6 of its 11 dedicated committees, from which The ROAM/EORTC1308 trial for atypical meningioma was facilitated: a UK-led inter-group trial across 59 sites in the United Kingdom, EORTC and Australia/New Zealand (Trans-Tasmin Radiation Oncology Group [TROG]) [20].

National neuro-oncology conferences are well attended although ideologically segregated—principally oriented towards clinicians (e.g. British Neuro-Oncology Society [BNOS] Annual Conference) or scientists (e.g. CRUK Brain Tumour Conference). Patient and public involvement and engagement (PPIE) in the community is essential. Initiatives such as brainstrust's Patient Research Involvement Movement (PRIME) bring people closer to research and research closer to funding [21].

Brain tumour research funding

Despite recently increasing funding levels for brain cancer research, this disease site remains relatively underfunded. Annual NCRI partner [22] funding for brain tumour research increased by £7.4m between 2017 (£10.2m) and 2021 (£17.6m) on par with the increase in funding for breast (£7.0m), bowel (£8.7m) and lung (£6.4m) cancer in the same period (Figure 3A) [23, 24]. However, the funding allocated to brain cancer in 2021 still only constituted 5.5% of the total NCRI partner annual spend on cancer research, having risen from 3.7% in 2017 (Figure 3B) [23]. Compare this with breast, bowel and lung cancer for which the allocation has remained consistently high at circa 16%, 12% and 11% of the total budget, respectively (Figure 3B) [23]. Although Figure 3A indicates that funding allocation is proportional to prevalence, this does not take into account the malignancy of each cancer subtype. Indeed, when funding allocation is plotted according to the average years of life lost, brain cancer is a clear outlier [23, 25] (Figure 3C). Inspecting how funding is allocated within cancer subtype, according to the Common Scientific Outline (a 6-tier classification of types of cancer research), we see that a relatively large portion of neuro-oncology research is still focused on understanding the basic biology of the disease, where the more well-funded cancers have more money allocated to earlier detection and prevention research (Figure 3D) [23]. This reflects the complexity of tumours of the brain and of the organ itself. Numerous factors, including cell type diversity and idiosyncratic aspects of systems biology, have meant that an in-depth knowledge of the human brain still alludes us. Focused, specific research is still very much needed to understand the human brain and its pathologies, including cancer.

Neuro-oncology training (scientific and clinical)

Training in scientific neuro-oncology research faces many challenges: brain cancer biology is uniquely complex; the relative disease rarity and accessibility of fresh and fixed tissue limits research samples; and there is no suitable single experimental model nor successful bench-to-bedside trajectory. All 47 UK masters-level biology programmes with ‘cancer’ or ‘oncology’ in the title [26] cover generalised elements of pan-cancer research (genomics, immunology, the tumour microenvironment [TME]). More specialised cancer-specific research training occurs at the doctoral level, where funding is disproportionally allocated to other cancers. This lack of specific training in, and exposure to, basic neuro-oncology research, combined with lower funding opportunities, produces fewer desirable careers for cancer researchers aiming for independence.

Comparable challenges face clinical training. Increasingly complex management of brain tumours requires surgery, radiotherapy and chemotherapy. Advances in these fields necessitate additional ongoing training and development involving multiple specialities. Beyond neurosurgery, where the pathway is well-defined, there is a paucity of training opportunities for neuro-oncology clinicians. UK brain tumour management has, historically, been led by clinical oncologists, with limited time and opportunities to interact with research. Neuro-oncology is not mandatory in the medical oncology curriculum, leading to a scarcity of early-phase trialists and clinical drug developers with the expertise to truly accelerate the development of novel therapeutics for brain tumours.

Priority 1: Prompter diagnosis

In many cancers, the notion of an ‘early diagnosis’ pertains to identifying the disease in a less mature state (at a lower ‘stage’ or ‘grade’), which can lead to less intrusive/toxic and/or more effective treatment. In brain cancer, it is debatable whether diagnosing at earlier disease stages impacts treatment decisions and prognosis. However, it is widely accepted that a prompter diagnosis, that is, shorter time between the development of symptoms of a tumour, irrespective of its stage or grade and clinical confirmation of the presence and type of tumour, is beneficial for many reasons [28-30]. Brain tumours are challenging to diagnose, with idiosyncrasies and barriers at each level from initially detecting a brain tumour through to the diagnosis of subtype [31]. Presenting symptoms are driven both by tumour anatomical location and more global effects of tumour growth. The former may produce stereotypical motor, visual or speech deficits, but the latter are non-specific and secondary to raised intracranial pressure or regional changes caused by the tumour, for example, headaches, nausea/vomiting, lethargy, behavioural changes or seizures. The commonality of some non-specific symptoms often delays patients visiting a doctor until symptoms escalate. Once consulted, medical practitioners often pursue other more common diagnoses, delaying definitive investigations. Rationing of investigations such as brain imaging also delays diagnosis. Approximately 2/3 of brain tumours are diagnosed after an emergency admission to hospital often preceded by several primary care consultations [32]. Only 1% of patients are diagnosed through the designated NHS England 2-week wait suspected cancer pathway [33]. Campaigns such as ‘HeadSmart’ (The Brain Tumour Charity), ‘Brain Tumour Awareness Month’ and ‘Wear a Hat Day’ (Brain Tumour Research) are increasing awareness of brain tumour symptoms with the aspiration of leading to prompter diagnosis.

Priority 2: Identify actionable target drivers of malignancy

Whilst molecular testing is being adopted for the diagnostic classification of brain tumours (Priority 1), the results do not routinely inform treatment decisions because of limited therapeutically actionable molecular biomarkers. This results from a limited understanding of genomics of brain tumours, and the (historical) exclusion of patients with brain tumours from precision medicine targeted trials.

Access to high-quality, well-annotated patient biosamples is essential for identifying target drivers of malignancy, particularly when co-occurring driver genes typically activate different collaborating oncogenic pathways. Integrating genomic, epigenomic, transcriptomic, proteomic and neuroimaging data will be critical to reveal vulnerabilities most amenable to therapeutic targeting. Disease rarity makes neuro-oncology biobanking relatively costly because the infrastructure needed is disproportionate to the sample volumes. The resulting sample scarcity for research causes issues of ownership and access to existing collections. Furthermore, brain bio-banking is often under-resourced, leading to deficits in: processing to maximise sample usage; collection beyond the tumour (host, blood, cerebral spinal fluid [CSF]); associated clinical metadata with follow-up; and generation of associated patient-derived models (see Priority 3). This promotes a negative perception of myriad biobanked samples sitting unavailable for research, when samples are either not known about, are inaccessible or lack sufficient clinical annotation for utility. Even where additional research-allocated samples cannot be collected, making the genetic data resulting from clinical practice accessible to basic science researchers, alongside linked clinical metadata and imaging data, would be hugely valuable.

In the United Kingdom, several initiatives aim to tackle this. BRAIN UK (BRain Archive Information Network UK) [36, 37] is a virtual biobank across a network of NHS Neuropathology Centres, exemplifying the unique UK ability to leverage NHS connectivity. BRAIN UK has generic ethics needed to approve projects and coordinate and grant access to archival surplus brain material. However, this is mostly limited to fixed tissue and retrospectively collated, centre-specific clinical data owing to a dearth of local infrastructure for greater provision. BRAIN MATRIX [38] includes resources to perform a more limited collection of frozen adult glioma samples, specifically, and molecularly profile them via NHS England Genomic Hubs with linked imaging and clinical data. Whilst centralised tissue cannot be repurposed, there is no barrier to using fresh tissue at the site for complementary research techniques such as single-cell analyses. Again, this is dependent on local infrastructure. Alongside these national efforts, multiple autonomous UK research tissue banks include neuro-oncology collections. These independent efforts vary with regard to consenting procedures, types of samples and data collected, access, processing, governance and application requirements. Their coordination would better facilitate higher-impact, larger-scale research.

Priority 3: Use suitable preclinical models and assays

Experimental models are needed to (1) validate the direct involvement of aberrant molecules and/or mechanisms in pathogenesis as causative rather than consequent for rational prioritisation of drug development; (2) screen novel therapeutic interventions. Both require the experimental system to mirror patient biology, or the specific aspect being tested, and this poses a major challenge for brain tumours [40]. The continued failure of neuro-oncology clinical trials is partly attributable to difficulties in experimentally modelling brain tumour biology, that is, tumour heterogeneity; tumour microenvironment (TME); the blood–brain barrier (BBB); and response to SOC [3, 41]. Advances in brain cancer cell culture techniques have led to cell lines that more closely mirror the originating tumour [42]. These can be used in 2D and 3D systems, with scaffolds and co-cultures to incorporate the TME, and in vivo, but each system models different aspects of tumour biology, and increasing complexity increases time and cost, forcing trade-offs [43-46]. Organoids and microfluidic ex vivo and BBB models offer great promise for modelling complexity at scale [47-49]. Patient-derived xenotransplants (PDX) models usually do not fully recapitulate the TME.

Priority 4: Provide sufficient evidence of therapeutic opportunity

The adoption of temozolomide as the SOC for glioblastoma occurred almost 20 years ago [2], demonstrating the translational failure that casts neuro-oncology as a ‘graveyard’ for novel therapeutics. Among legion contributors, inter- and intra-patient heterogeneity of brain cancer and the BBB, which modulates drug delivery, represent major obstacles [54]. Academic research is key to identifying new drug targets (Priority 2), including understanding target biology and links between targets and disease states (Priority 3). However, academic credit and pharmaceutical company value structures do not align. Academic progression prioritises publication and grant funding, often predicated on novelty, and industry prioritises understanding the ‘right target’, which requires thorough, standardised validation (or de-validation) of a scientific hypothesis throughout the lifetime of a project. Furthermore, the ability to de-risk a promising drug target is dependent on the clinical annotation, quantity/quality of patient tissue and accuracy of the model(s) used in its validation/de-validation. There are problems in both aspects of neuro-oncology research.

Several biopharma companies have adopted the 5R framework (‘the right target, right tissue, right safety, right patient, and right commercial potential’) to tackle R&D productivity issues [55, 56]. To deliver impactful data packages that can serve as a platform of evidence for the next stages of drug development, research must progress from purely academic exploration to the initiation of efforts to interrogate the drug candidate in the context of pharmacokinetic/pharmacodynamic properties, establishing proof of concept as well as safety/tolerability [55, 57, 58].

Priority 5: Develop accessible, innovative and evidence-based clinical trials

Clinical trials realise translation of novel interventions arising from Priorities 2–4. First-in-man phase 1 trials evaluate safety and test pharmacokinetics with escalated dosing to ascertain the appropriate prescription. Phase 2 trials apply this to a larger cohort to assess safety and indicate activity. Large, randomised phase 3 trials test promising interventions, usually against SOC. This pipeline has limitations for rarer cancers, as reflected in the poor conversion of promising early brain cancer trial results to phase 3 outcomes, and the lack of improvement in overall survival since 2005 (Table 2). Some contributing factors are relevant to all clinical trials with others brain cancer specific.

Firstly, patients with brain tumours are excluded from the majority of early phase trials and tumour agnostic basket trials with <1% of UK recruiting trials listed on the EC trial finder website [66] permitting enrolment of patients with brain tumours. This has historically been attributed to a poor understanding of the BBB (and its leakiness) and uncertainty about whether novel agents can achieve meaningful concentrations in the brain. Phase 0 window of opportunity trials that can quantify brain exposure to novel agents, as well as provide pharmacodynamic evidence of pathway modulation, will help to identify active drugs more efficiently, but they are challenging to deliver.

Early phase trials, particularly single-arm trials, typically have small sample sizes that risk selection and sampling bias and increased risk of false positives. If surrogate endpoints do not correlate with clinical outcomes, they can mislead causing premature and inappropriate inclusion/exclusion of candidate interventions. Surrogate biomarkers are lacking, and there is variability of surgery and radiotherapy, varying by tumour location and proximity to eloquent brain and organs at risk, which limits comparator arm comparability. Given the heterogeneity of brain cancers, even where targeted agents have been trialled in brain cancer patients, and progressed to later-stage registration trials, these have been in an unselected patient population and failed to meet their endpoints (Table 2). Even with an adaptive clinical trial strategy such as those used in the international Phase 2/3 platform GBM AGILE trial (NCT03970447), evaluating multiple regimes in unselected patients has been disappointing thus far with the initial regimes tested not meeting interim efficacy for transition to Phase 3 [67]. This suggests an urgent and ambitious need for bespoke novel clinical trial designs to specifically overcome the challenges specific to brain tumour trials incorporating a seamless transition from Phase 0 surgical trials to biomarker-defined early-phase hypotheses testing to later-stage efficacy testing. The MHRA-approved 5G (An AGile Next Generation Genomically Guided Glioblastoma Trial) adaptive platform trial (conceived following the NCRI Brain Strategic Workshops in 2021) will utilise genomic and transcriptomic data to stratify patients into molecular hypotheses testing subprotocols, allowing for agile and rapid in-flight course correction and refinement of molecular hypotheses as investigators learning as much as they can directly from patients enrolled on this platform.

Priority 6: Treat every patient as a research patient

Only 5% of brain tumour patients are entering the limited number of trials available, partly from a lack of up-to-date clinical trial databases but also the variability in access. The latter results from cross-centre variation in infrastructure, resources and capacity, including time allocation for the trial leads and research nurse support. Improving outcomes needs the right people to drive change, requiring sufficient time allocation and remuneration. This is unsustainable: recruitment and retention of (clinical) academics requires suitable rewards. In addition, although some may not be eligible for trials, every patient should be offered to opportunity to donate samples, imaging and clinical metadata to research.

The analysis and interpretation of outcome measures, low adherence and missing data are methodological challenges. The current focus on system-wide delivery and outcome measurement loses sight of the person living with the brain tumour and devalues what matters to them. Patients are more than their clinical data: for example, their perception of their health, what motivates or negates behaviour changes or how other life events and stressors confound the maintenance of health and well-being. Yet, patient involvement in research remains fragmented and lacks strategic overview. The multiplication of therapies means more trials, necessitating a paradigm shift in the measurement of health-related quality of life (HRQoL). The disproportionate focus on outcomes limits understanding of what individual patients want to achieve. COBRA and COSMIC are patient-centred clinical trials co-developed with patient and carer stakeholders that are starting to move these goalposts, ensuring that outcome sets are truly meaningful to patients in the real world [69, 70]. With personalised medicine, patients experience different clinical journeys: one size no longer fits all.

Priority 7: Facilitate living beyond a brain tumour

The United Kingdom is strategically well-placed to contribute to and lead research into survivorship, quality of life and patient-reported outcomes [73]. Several centres have produced world-leading outputs in the last decade with international collaborators. The James Lind Alliance produced a consensus priority list highlighting ‘quality of life’ questions about lifestyle factors, interval scanning, early referral to palliative care, the study of late effects, interventions for carers and strategies for managing fatigue [4]. Numerous routes for grant funding exist: The Brain Tumour Charity's dedicated Quality of Life research grant call funded BT-LIFE, an innovative UK pilot trial of lifestyle interventions for fatigue that recently published positive results [74] and the NIHR-funded SPRING, a phase 3 trial of levetiracetam prophylaxis of epilepsy in seizure-naive patients with newly diagnosed glioma [75].

Notwithstanding these UK initiatives, survivorship and outcomes research received just 5% of total NCRI partner spend on brain tumour research in 2021 (Figure 3D), potentially limiting improvements. Increasing proportional spending requires a shift away from low-impact observational studies. Although single-centre observational studies are more accessible to trainees or non-career academics, their analysis is typically confounded by the high number of variables and small sample sizes. The clinical impact of observational studies is limited, and these proposals struggle to attract funding. Large-scale, collaborative epidemiology or data-linkage studies and randomised controlled trials (RCTs) are robust to these limitations and should be prioritised. Glioma patients also have cognitive impairment, fatigue and complex often toxic treatments that can directly and indirectly affect quality of life. Challenges to clinical trials in these areas require strong mentorship and guidance to support and improve the methodological quality of proposals.