RDC Biomarker Study

Background

The proposed research will focus on Rapid Diagnostic Centres (RDCs) already established across multiple sites in England. Since 2019-2020, RDCs, have begun rolling out nationally across all cancer alliances. They are a single point of access for cancer diagnostics to allow personalised and rapid diagnosis of patients presenting with non-specific but concerning symptoms of cancer (NSCS).

Cancer patients with NSCS typically have a longer interval in primary care to referral and have more advanced disease once a diagnosis is made, yet approximately 50% of cancer patients present with NSCS. These patients are poorly served by existing cancer referral pathways which can be circuitous when a clear pattern of symptoms is not present.

The objectives of RDCs are to:

• Support earlier and faster cancer diagnosis
• Create increased capacity through more efficient diagnostic pathways
• Deliver a better, personalised diagnostic experience for patients
• Reduce unwarranted variation in referral for, access to and in the reliability of relevant diagnostic tests

RDCs Limitations

Although patients seen in RDCs are presumed to receive faster diagnosis and treatment, there is still very little known about how their vague symptoms developed over time and how these are linked to underlying health conditions, diagnosis, health outcomes and RDC effectiveness.

Most RDC patients will not have cancer. Therefore, accurate and rapid triage will be essential and depend upon improved methods of risk stratification and cancer detection. There is both a significant need and opportunity to integrate quantitative research into RDCs to develop and validate novel diagnostic assays, and risk scores based on this unique population of patients. These patients both represent a primary care community, but the enriched cancer prevalence within RDCs could be utilised to facilitate more efficient biomarker development.

Rationale

Developing pathways to streamline care in this population group is important not only due to the importance of timely detection and treatment, but most patients in these pathways both with and without cancer are more likely to be from more deprived backgrounds or prone to being less engaged with healthcare provision. The implementation specification from NHS England highlights the need for RDCs to tackle health inequalities in their roll out. This study will allow us to generate a sample set and database to examine the potential to:

1. discover high risk groups in RDC populations and ii) to detect cancer-specific signals in symptom profiling data; routine clinical and imaging data; and blood and non-blood biomarker tools.

The primary aims focus on using descriptive analysis, to identify the following:

1. The number and clinical traits of Benign versus Malignant Diagnoses
2. The number and clinical traits of cancer subtypes (e.g. lung vs ovarian or other cancer) The primary aims also focus on describing how the variability between participants can delineate cancer versus benign pathology using molecular; germline risk, symptom-profiling; and clinical data from each of the four main work packages, independently or in combination using AI-based multi-parametric analysis of these traits: Descriptive analyses will provide feasibility data to develop and validate models using i) Peripheral blood and non-blood (e.g. breath and saliva) molecular signatures that can delineate between cancer and benign pathology. ii) Patient 'symptome' signatures delineating between cancer and benign pathology. iii) Digital health record signature from routinely collected clinical electronic health record (EHR) and/or imaging data and/or the 'standard' RDC dataset that delineate between cancer and benign pathology. iv) Risk stratification tools (e.g. epidemiological and genetic/PRS) that inform risk of cancer in RDC attendees and primary care.

Patient Selection

Overall, we expect an initial sample size of 1000 patients that will allow several exploratory analyses and simple model development/validation (in distinct datasets) within an observational platform study design. The target recruitment will be increased by the Trial Steering Group (TSG) if new relevant assays are identified, and extended funding is secured. This is supported by previous work which suggests that 1000 patients are more than sufficient to identify candidate biomarkers for further studies, it is a pragmatic approach and is not the result of power calculations. A target goal will be for 75% of cancer cases to have histological confirmation i.e. ground truth of final diagnosis to help power diagnostic accuracy and reliability for any future use of histopathological data Patients recruited will be removed from analysis where ground truth cannot be established by the time of completion of initial investigations or a period of clinical surveillance (not to exceed 12 months of surveillance after initial diagnostic pathway completed).

The local Principal Investigator (PI), or members of the delegated clinical care team at participating trusts (e.g. Consultant physicians, nurse specialists, clinical fellows or admin staff), will identify suitable patients via RDC clinic appointments, RDC multidisciplinary team (MDT) lists or by RDC service evaluations, supported where necessary by clinical informatics approaches (e.g. structured query language searches) or local research team.

We estimate that each participating hospital will see approximately 20-50 new RDC patients per week, thus we expect that we could recruit large numbers of patients from a small number of RDCs. This estimation may evolve considering the number of participating sites, as well as the evolving nature of how RDCs are utilised, particularly given the impact of the COVID-19 pandemic on cancer diagnosis. Recruitment will be consecutive in the order in which patients are seen within each respective RDC. Allocation of participants to a particular sample type/assay will be pragmatic based on site geography/capability of the RDC at which the patient is recruited.

Study Duration

The study will end when complete outcome or surveillance data is available for all participants, or after 12 months following recruitment ± completion of clinical assessment of the last patient, whichever occurs sooner. This duration is necessary to ensure confirmation of diagnosis, or stability after discharge and therefore ground truth. It is expected that recruitment will complete within two years. Patient data and samples will be stored for a total of ten years. The ten-year duration of storage is necessary to allow us to utilise more advanced laboratory and data analysis techniques, which may become available with evolving technology. Newer approaches may require very large datasets, and this timeframe would allow for a protocol amendment for expansion of our data and further research to be performed as the field evolves. Upon completion of the study, we will establish a research database for future work and will store patient data for a total of ten years.

Assessment of imaging data quality and integrity and suitability of research specimens for laboratory analysis will be contemporaneous to patient recruitment/sample receipt and ongoing thereafter. Analysis will typically be performed on patients with confirmed ground truth by biopsy or sufficient clinical surveillance if benign e.g. 12 months (see above) but scans performed after this period may be accessed if future relevant data become available.

Enrolment and Consent

Patient enrolment into the study will occur once consent is taken and eligibility criteria is met. A trial number will be allocated to each patient for the identification research samples.

In most cases, patients will donate a blood specimen on one visit (Baseline). Where recurrent visits to the hospital for surveillance CT scans are made over a longer period, then up to 3 blood specimens will be collected in total with a minimum of 21 days between each research specimen collection or a treatment intervention. This will allow the consideration of evolution of the biomarker signal in question as a determinant of cancer presence or absence (using the logic that a given signal is likely to increase alongside tumour growth and therefore be easier to detect in comparison with samples from patients without malignancy).

All participants are free to withdraw at any time from the protocol treatment without giving reasons and without prejudicing further treatment. The patient can request that any remaining samples donated are destroyed. All members of the delegated research team will ensure that patient confidentiality is maintained in compliance with the UK Data Protection Act 2018 and General Data Protection Regulation

Patient Samples, Acquisition and Processing

The laboratory manual will provide further details of the collection, processing, and storage of specimens.

Study Oversite

A Trial Steering Group (TSG) will meet on a regular basis to provide overall study oversight and clinical or scientific steer - particularly for academic/research considerations.

In addition, a Trial Management Group (TMG) will meet quarterly to discuss patient recruitment, data collection, data management, data analysis, reporting for adverse events, and change management. The minutes of these meetings will be made available, and a rolling agenda will highlight any ongoing actions or concerns.

In addition, the trial management team will provide day-to-day support for trial administration, including all rules and regulations concerning research studies in the UK. This includes: UK Policy Framework for Health and Social Care Research, Data Protection Act 2018 and UK General Data Protection Regulations

Data Analysis and Statistical Considerations

Statistical guidance and study design oversight will be provided by RM-ICR guided by the CI and Lead Statistician/Data Scientist and will be undertaken in partnership with academic partner institutions named in the protocol ± support from other academic centres. Data sharing agreements will be in place in each case. A full Data Analysis Plan will be assembled and ratified by the Trial Steering Group at its creation and on any subsequent amendment.

Clinical Data analysis

The main analysis, including clinical data values, will be descriptive and will be focused to inform future prospective studies. For continuous variables, the mean and standard deviation will be presented, together with the mean between-group difference, and 95% confidence interval. For binary outcomes, the percentage and frequency of patients in the outcome category of interest will be presented. When necessary intracluster correlation coefficients will be reported, together with 95% confidence interval. Where appropriate p-values will be presented.

Baseline characteristics, collected at the time of commencing the study will provide an overview of the study population, both at the RDC and site level. It is expected that patients may differ based on which RDC they are recruited at and described with summary statistics including the index of multiple deprivation score for the RDC postcode, percentage of patients with comorbidities, and the cancer type. At the individual patient level, variables can include gender, age at baseline data collection, individual Index of Multiple Deprivation (IMD) values, baseline measures of all physiological measurements, comorbidity. Baseline characteristics will be summarised as the mean, standard deviation and range for continuous, approximately symmetric variables; medians, interquartile range and range for continuous, skewed variables; frequencies/percentages of patients/RDC in each category for categorical variables.

Laboratory Data Analysis

Analysis of laboratory data will lie in the first instance with the clinical and scientific lead for each work package with further details provided in the laboratory manual and local laboratory protocols. The assays will give a read out for each patient that will be assessed using receiver operating characteristic curve analysis against each endpoint. Firstly, this will be performed using a single biological parameter for cancer versus non-cancer, or most likely cancer subtype, and then in combination with other laboratory and clinical endpoints as a multiparametric signature for the same endpoints as outlined in the protocol.