Service evaluation of a blood test which uses AI to produce a calibrated probability that a symptomatic patient has cancer

The West Yorkshire and Harrogate Cancer Alliance is one of 19 alliances set up around the country to deliver the significant improvements set out in the report of the independent national cancer taskforce Achieving World Class Cancer Outcomes: A Strategy for England, 2015 to 2020 and the NHS Long Term Plan.

The alliance works with local health and care partners, including clinical commissioning groups (CCGs), local acute Trusts, GP practices in West Yorkshire and Harrogate, the West Yorkshire Association of Acute Trusts, local authorities, and voluntary sector organisations including Macmillan Cancer Support. It also works with local patients through a patient panel drawn from the West Yorkshire and Harrogate area and managed by Health Watch.

Key objectives were to design and deliver the transformation of services and care, to reduce variations in the availability of good care and treatment, to deliver ongoing improvements in the future, to pilot innovative approaches and to share best practice.

Situation

The urgent referral pathways for cancer (also known as 2-week wait or 2WW) are a particular focus for the alliance, due to the challenges of an ongoing 10% year-on-year increase in 2WW referral rates.

In 2019 to 2020, 111,548 symptomatic patients in West Yorkshire and Harrogate Cancer Alliance were referred to an urgent cancer referral pathway by GPs in 397 practices across 6 CCGs. Of these patients, 6,934 received a cancer diagnosis. This means that 94% of patients referred in the region did not have cancer, but all underwent extensive diagnostic investigations. This aligns with statistics from other cancer alliances and NHS England as a whole.

While the 2WW referral pathways are very successful as a way of identifying patients with cancer, the number of referrals in West Yorkshire and Harrogate alone have grown 10% year-on-year since 2010 and were expected to grow to more than 150,000 a year by 2025. The large number of patients without cancer who are referred is now placing an unsustainable pressure on cancer diagnostic services and subjecting large numbers of patients to unnecessary stress and medical procedures.

Aspiration

An NHS rapid response service to help prioritise cancer referrals and make the diagnosis process more streamlined.

Solution and impact

The PinPoint Test is a clinical decision support tool that produces a calibrated probability of cancer for a patient on a given 2WW referral pathway. It combines signals from a range of blood measurements (using a machine learning algorithm), thereby providing an additional source of relevant information for secondary care clinicians and GPs to use when making clinical decisions about the patient. The results, combined with basic patient information, can indicate with very high accuracy a patient’s chance of having cancer.

The principal use-case for the PinPoint Test during the COVID-19 pandemic is enabling the triage of patients into high, medium and low risk. This allows clinicians to red flag those patients who most urgently require further investigation. This is of particular importance during the crisis due to the necessity of identifying those at very high risk of cancer who need immediate investigations to avoid stage shift.

The PinPoint Test has already been validated using historical samples of patients referred down urgent suspected cancer pathways in Leeds Teaching Hospitals NHS Trust. The current phase of work (phase 1) involves a service evaluation to check the feasibility of GPs carrying out the test in practice and getting the results to where they need to be. To this end, the project is currently deployed to GP practices in North Kirklees CCG and Wakefield CCG. Approximately half the GP practices are now participating. Mid Yorkshire Hospitals NHS Trust provides the pathology hub and has contributed greatly to creating National Pathology Exchange (NPEx) pathways to carry the data flow to and from the pathology laboratory. Leeds Teaching Hospitals NHS Trust is currently mobilising to take bloods in secondary care starting with the breast cancer team.

Functionality

The PinPoint Test is ordered exactly like any standard NHS blood test. It is a software solution deployed inside the NHS and which runs entirely using blood measurements already available from any NHS pathology laboratory. This makes it safe by design with patient data, highly affordable, and highly scalable across the entire NHS.

As part of the test, the patient is asked by their GP to provide a blood sample during their urgent cancer referral. The amount of blood collected will not exceed 8.5ml (less than 2 teaspoons) and obtaining the sample takes about 5 minutes. These bloods are sent to an NHS pathology laboratory and the PinPoint panel of measurements made. These measurements (as well as age and sex) are input into the PinPoint software running on an NHS IT system. The software returns the probability result to the clinician’s standard IT system, along with recommended guidance text (as decided by NHS clinicians).

The process involves the use of machine learning to combine signals from multiple blood analytes and basic patient information into a single clinical decision support tool. This allows clinicians to easily identify those patients who should be referred to specialists for cancer diagnosis, red flag particularly urgent cases and identify those who can be safely investigated for other possible causes of their symptoms.

Capabilities

The PinPoint Test is designed to be used alongside standard testing for suspected cancer (that is, it is a clinical decision support tool). It provides an additional source of information to help identify whether or not a patient has cancer.

The test is CE marked for use in the following 9 2WW cancer pathways:

• breast (including breast symptoms)
• lower gastrointestinal
• upper gastrointestinal
• gynaecological
• lung
• urological (excluding testicular)
• haematological (excluding acute leukaemia)
• head and neck
• skin

Use of the test:

• saves unnecessary stress or medical risk for patients
• eases workload on secondary care clinicians
• saves the NHS money and releases capacity
• load balances the pathway (smooths out peaks and troughs in the rate of referral, for example due to COVID or winter pressures)

The PinPoint Test cannot currently be used as a general test to rule out cancer for non-symptomatic patients in the wider population, as this is outside the currently validated indications for use.

Scope

The PinPoint Test is available to primary and/or secondary care clinicians as a blood test for assessing the risk that a given 2WW patient has cancer. Its intended use is as a decision support tool, to help as part of the overall clinical judgement as to whether the patient should continue with a 2WW referral.

1. Rule-out test: use of the test to identify 20% of 2WW patients who have very low cancer risk, so that they can be safely ruled-out of the 2WW pathway and other possible causes of their symptoms investigated.
2. Triage test: use of the test to identify patients with a high risk of cancer, so that their onward clinical journey can be expedited.

Key learning points

IT integration

Hosting in a trusted organisation within the NHS firewall under NHS control reduces the possibility of data mismanagement by the non-NHS organisation and makes seamless integration easier to achieve.

Trust IT services are stretched so using standard interfaces (such as HL7: international standards for transfer of clinical and administrative data) and allowing time to work with Trust IT staff is essential to achieve integration. Senior support is also needed to ensure sign off does not slow down the process.

Information governance

Establishing appropriate data flows and information governance requires careful and detailed understanding of all processes.

Agreements between the different parties, privacy statement and information sheets for patients and clinicians involves multiple stakeholders, and when working with multiple Trusts, engagement with different information governance departments.

Stakeholder engagement

It was recognised that early and ongoing engagement with stakeholders is a vital element of the project. Engaging patients and establishing views on the test was achieved through Healthwatch.

Members of the Cancer Alliance Community/ Patient Panel have been involved, including the development of specific patient-facing information.

Engagement with local medical committees (LMCs) was part of the key to gaining GP support. Engaging GPs through face-to-face sessions has worked well. Engagement sessions and short videos are likely to be the best way to get understanding and buy in the future.

Commencing an innovation in the middle of a pandemic

• PinPoint was identified as a COVID-19 priority project by the Cancer Alliance board because of the benefits it could bring to ensure patients at the highest risk could be prioritised on the 2WW pathway
• IT support was required at time when IT was the only means of communication across the NHS, as hundreds of staff were asked to work from home. This increased the time taken to secure IT support and time to complete the work
• Starting the project during a pandemic raised questions about ‘why now?’ This required constant explanation, highlighting the need for good communications
• Timing of the launch coincided with the vaccination programme and this meant a slower uptake by GP Practices. GP practices attended 30-minute lunchtime engagement slots which resulted in a gradual increase in bloods being taken

Key figures/quotes

“We are committed to bringing in new technologies to help the cancer diagnostic system, PinPoint is ready to help us now.”
Professor Sean Duffy, strategic lead cancer clinician and director, Leeds Cancer Programme, and former national clinical director for cancer, NHS England

“This will help systems return to normal post pandemic helping over 100,000 patients a year in West Yorkshire and Harrogate alone experience improved cancer services.”
Professor Richard Neal, professor of primary care oncology at University of Leeds and hub lead for the Department of Health policy research unit for cancer screening, awareness and early diagnosis

PinPoint Data Science has developed and validated their algorithm using historical samples of patients referred down urgent suspected cancer pathways in Leeds Teaching Hospitals NHS Trust.

Algorithms were developed and validated for 9 2WW pathways using retrospective data from 371,799 referrals to Leeds Teaching Hospitals NHS Trust (development set 2011 to 2016, validation set 2017 2019). In use-case 1, the algorithms correctly identify 20% of patients who do not have cancer and may not need an urgent 2WW referral. In use-case 2, they identify 90% of cancer cases with a high probability of cancer that could be prioritised for review.

Find out more

Read more about the West Yorkshire and Harrogate Cancer Alliance case study

Find out more about PinPoint

Key contact

Helen Ryan, West Yorkshire and Harrogate Cancer Alliance Programme Manager, New Models of Care and Innovation

helen.ryan7@nhs.net

Rich Savage, chief scientist, PinPoint

rich.savage@pinpointdatascience.com