Our publications

A plain language summary of:
Addressing the variation in adjuvant chemotherapy treatment for colorectal cancer: Can a regional intervention promote national change?  John C. Taylor, Daniel Swinson, Jenny F. Seligmann, Rebecca J. Birch, Alice Dewdney, Victoria Brown, Joanna Dent, Hannah L. Rossington, Philip Quirke, Eva J. A. Morris, YCR BCIP Study Group

Background:
Surgery is the main treatment for bowel cancer. In order to try and reduce the chances of the cancer coming back, some patients are also offered additional treatment either before or after they have surgery. This is usually done by chemotherapy, which is where a patient is given a drug or a combination of drugs over a number of months. When chemotherapy is given after surgery, this is referred to as adjuvant chemotherapy. There are guidelines in place which help doctors decide when to offer this treatment to a patient.

The aim of this study is to identify which patients received adjuvant chemotherapy and look at whether treatment is more common in any particular groups of patients or at any particular hospital. This information could then be used to improve the guidance provided to doctors therefore improving the care of patients.

What did we do?
The differences in the use of adjuvant chemotherapy was investigated using data from all hospitals treating colorectal cancer in England. A more detailed investigation then took place focusing just on the hospitals within the Yorkshire and Humber region.

Alongside the investigation into the data, questionnaires about the uses of adjuvant chemotherapy were sent to the doctors who prescribe the chemotherapy, known as Oncologists, in the Yorkshire and Humber region.

A series of meetings for all Oncologists from the Yorkshire and Humber region were held to address the differences in adjuvant chemotherapy use both at a regional and national level. A further round of questionnaires were then sent to the Oncologists to see if the differences identified in the previous meetings had been addressed.

A final meeting took place during which a new set of guidelines was agreed in order to reduce the differences in treatment across the Yorkshire and Humber region.

What did we find?
The study found that there are large differences in the use of adjuvant chemotherapy for bowel cancer both nationally and regionally. Through bringing this information to the attention of the Oncologists in the Yorkshire and Humber region, all hospitals in the region agreed to a new set of guidelines which aims to reduce the differences in care that patients receive. This process could be copied in other regions in England.

A plain language summary of:
Standardised reports with a template format are superior to free text reports:  the case for rectal cancer reporting in clinical practice, P.J. Brown, H. Rossington, J. Taylor, D.M.J. Lambregts, E. Morris, N.P. West, P. Quirke, D. Tolan on behalf of the Yorkshire Cancer Research Bowel Cancer Improvement Programme Study Group

Background:
The most accurate way of staging and assessing rectal cancer before an operation is through Magnetic Resonance Imaging (MRI).

Primary--staging MRI is used to determine how a cancer is managed by the clinical team, including whether radiotherapy or chemotherapy is given before surgery. A re-staging MRI can help determine the operation technique or alternatively what other treatment options could be offered. MRI reports greatly influence the decision on what treatment is used, so it is very important that they are accurate.

There is growing interest amongst radiologists to have a more structured reporting system in order to improve the communication of the findings and to generate consistent reports. The European Society of Gastrointestinal Abdominal Radiology (ESGAR) and Society of Abdominal Radiology (SAR) recommend the use of report templates for primary staging and re-staging MRI reports.

MRI reporting templates have been produced however they are not widely used.

What did we do?
This study evaluated the standard of primary staging rectal cancer MRI reports.

Sixteen UK colorectal cancer Multi-disciplinary Teams (MDT’s) were invited to take part in the study. Each MDT was asked to submit ten consecutive primary staging reports for each of their radiologists all which had to be within a 12 month period from January 2016 to January 2017. Four hundred and ten primary staging reports were submitted. Out of these, 360 were analysed. Eighty one were reported using a standardised template and 279 were reported as free text.

Twenty-two key tumour descriptors were evaluated within each report using a simple reporting scoring system.

What did we find?
The study found that important information describing tumours is consistently omitted from primary staging rectal cancer reports. The use of a template in primary staging MRI reporting significantly increased the inclusion of key tumour descriptors when compared to free-text reporting.

A plain language summary of:
Radiologist and MDT clinician opinions on the quality of MRI rectal cancer staging reports: how are we doing? P.J. Brown, H. Rossington, J. Taylor, D.M.J. Lambregts, E.J.A. Morris, N.P. West, P. Quirke, D. Tolan, YCR BCIP Study Group

Background:
The most accurate way of staging and assessing rectal cancer before surgery is through Magnetic Resonance Imaging (MRI). MRI reports, which are written by radiologists, contain information called key tumour features which are used by specialist clinicians in the colorectal cancer multidisciplinary team to personalise a patient’s treatment.

The standard of rectal cancer staging MRI reports are not consistent and as a result guidelines have been issued by the European Society of Gastrointestinal Abdominal Radiology (ESGAR) and Society of Abdominal Radiology (SAR) which recommend the use of structured report templates.

Structured report templates are increasingly being thought to improve the communication of imaging and pathological findings however they are not widely used by radiologists as many still prefer the free style reporting.

As rectal cancer staging MRI techniques have increased, so have the number of key tumour features that it is recommended to include in the reports. The demand to include these features is usually led by colorectal cancer specialists in order to optimise and personalise a patient’s treatment. So therefore the input and opinion from colorectal cancer clinicians on the quality and contents of rectal cancer staging reports could guide radiologists.

What did we do?
This study evaluated the standard of, and satisfaction with, rectal cancer MRI reports in the UK.

Sixteen UK colorectal cancer Multi-Disciplinary Teams (MDT’s) were invited to take part in the study. The colorectal cancer MDT lead clinician in each hospital was asked to send out a questionnaire about the quality of the MRI reports to colleagues in their MDT. This was sent via email to participants which included surgeons, medical oncologists, clinical oncologists, histopathologists and clinical nurse specialists. A different version of the questionnaire was sent to the MDT clinical radiologists.

The questionnaire included 25 questions, 22 of these were closed questions and three were open questions which asked for more feedback and suggestions.

What did we find?
The study found that both groups of people recognised that key tumour descriptors are often missing from the rectal cancer MRI reports.  Both groups recognised that rectal cancer MRI scans can be difficult to interpret, even by specialist consultant radiologists, which potentially leads to different opinions on the key tumour features.

Conclusion:
The implementation of a standardised report template may improve the completeness and clarity of MRI reports for rectal cancer as well as lead to an improvement in the clinical management and outcomes of rectal cancer. Further work is required to assess the variation in interpretation of the key tumour features by different consultant gastrointestinal radiologists.

A plain language summary of:
Time to surgery following short-course radiotherapy in rectal cancer and its impact on post-operative outcomes. A population-based study across the English NHS, 2009-2014. B.A.Levick, A.J.Gilbert, K.L.Spencer, A.Downing, J.C.Taylor, P.J.Finan, D.J.Sebag-Montefiore, E.J.A.Morris, Clinical Oncology (2019).

What was the study about?
Short course radiotherapy (SCRT) before an operation is an effective treatment for rectal cancer. Previous studies have shown that it reduces the risk of the disease returning in the same location. The recommended gap between completing SCRT and having surgery varies widely, with some studies suggesting that it should be less than four days. This study examined the time between SCRT and surgery and related it to post-operative outcomes up to a year after surgery.

What did we do?
Information was obtained on all rectal cancer patients who underwent surgery and received SCRT in the English NHS between April 2009-December 2014. Post-operative outcomes such as thirty day post-operative mortality, returns to theatre, length of stay and one year survival were investigated in relation to the amount of time between SCRT and surgery. Patients who had a gap of over 27 days were not included in the study.

What did we find?
The study found that although the majority of patients were operated on within seven days of SCRT, there was huge variation across the English NHS. The study did not find any associations between time to surgery following SCRT and surgical outcome.

Acknowledgements
This work involves patient-level information collected by the NHS that has either been provided by, or derived from, patients as part of their care and support. The data are collated, maintained and quality assured by the National Cancer Registration and Analysis Service, which is part of Public Health England (PHE). Access to the data was facilitated by the PHE Office for Data Release. The data used for this study are available from the National Cancer Registration and Analysis Service via the PHE Office for Data Release, subject to relevant approvals. This work was supported by the Bobby Moore Fund/Cancer Research UK (grant number C23434/A23706) and Yorkshire Cancer Research (grant number L394). It was underpinned by the Leeds MRC Medical Bioinformatics Centre (grant number MR/L01629X/1). This study was approved by the South West – Central Bristol Research Ethics Committee (ref 18/SW/0134).

A Plain language summary of:
Additional loss of MSH2 and MSH6 expression in sporadic deficient mismatch repair colorectal cancer due to MLH1 promoter hypermethylation. Alice Westwood, Amy Glover, Gordon Hutchins, Caroline Young, Scarlet Brockmoeller, Rachel Robinson, Lisa Worrilow, Dave Wallace, Kate Rankeillor, Julian Adlard, Philip Quirke, Nicholas West, Journal of Clinical Pathology 2019

Background:
3% of patients with bowel cancer are found to have an inherited mutation in the genes that repair DNA damage, known as Lynch syndrome. An inherited mutation is a change in the gene of a sperm or egg cell that is passed on from parents to their children. The passed on gene becomes part of the DNA of all the cells in the child.

Cancer is often caused by gene mutations that cause cells to grow too quickly; these mutations can happen randomly when a cell is growing and dividing and would normally be corrected by mismatch repair proteins. Individuals with Lynch syndrome are not able to repair the damaged cells due to faulty DNA mismatch repair genes therefore they have a higher chance of developing cancer, specifically bowel cancer.

The UK National Institute for Health and Care Excellence (NICE) guidance published in February 2017 recommends that all patients diagnosed with bowel cancer are screened for Lynch syndrome by testing their cancer for signs of deficient mismatch repair.

What did we do?
Lynch screening was made available to all patients newly diagnosed with bowel cancer in all 16 hospitals in the Yorkshire and Humber region from May 2017 as part of the Yorkshire Cancer Research Bowel Cancer Improvement Programme.

What did we find?
In the first 829 cases that were screened, 94 cases showed evidence of deficient mismatch repair in the tumour of which 20 patients showed possible features of Lynch syndrome requiring further testing.

Conclusion
It is important that Lynch screening results are interpreted correctly so that patients are not referred unnecessarily for further testing, which could lead to anxiety for them and their family as well as unnecessary costs for the healthcare system.

A plain language summary of:
‘Variation in the use of resection for colorectal cancer liver metastases across the English NHS’. Hayley Fenton, John Taylor, Peter Lodge, Giles Toogood, Paul Finan, Alastair Young, Eva Morris, Annals of Surgery, 2019.

What was the study about?
Bowel cancer is a common disease in the UK with over 42,000 people diagnosed every year. If at diagnosis the cancer is found to have spread then patient outcomes can be poor. A common site for such spread is the liver. When this happens, it is called secondary liver cancer (the primary cancer being in the bowel). The National Institute for Health and Care Excellence (NICE) recommends surgery to remove the cancer from the liver where possible (the surgery needs to leave enough of the liver to support the patient). Around 25% of people who have such surgery survive for at least ten years. Previous studies, some of which date back 10 years, have shown that use of liver surgery varies significantly across the English NHS. This study looked at more recent data to see whether this variation still exists.

What did we do?
The study identified all patients who underwent bowel cancer surgery in an NHS hospital between 2005-2012. All patients who received liver surgery within three years of their bowel surgery were then identified.

What did we find?
There were significant differences in the rate of liver surgery between NHS Trusts. Also, patients whose bowel surgery took place in a hospital with an onsite liver specialist team were more likely to receive a liver operation. This may mean more patients could be benefiting from surgery for secondary liver cancer.

More information about secondary liver cancer can be found on the Macmillan website.

A plain language summary of:
Variation in post-colonoscopy colorectal cancer across colonoscopy providers in the English National Health Service Nicholas E Burr, Edmund Derbyshire, John Taylor, Simon Whalley, Venkataraman Subramanian, Paul J Finan, Matthew D Rutter, Roland Valori, Eva J A Morris, British Medical Journal, 2019.

Background
Around 40,000 people are diagnosed with bowel cancer every year in England. As well as being the main test to detect cancer, colonoscopies can also prevent cancer. Unfortunately, colonoscopies are not perfect and sometimes a person develops bowel cancer after having a colonoscopy. This is referred to as a post-colonoscopy colorectal cancer (PCCRC).

What did we do?
We looked at all individuals who had undergone a colonoscopy in the English NHS between 2005 and 2013 who then went on to be diagnosed with bowel cancer within the three years following the colonoscopy. Cancers diagnosed 6-36 months after a colonoscopy were categorised as PCCRC.

To see if some individuals were more at risk of developing a PCCRC, we looked at the rates in relation to age, sex, socio-economic status and whether they had any pre-existing health conditions, including a history of inflammatory bowel disease or diverticular disease, whether they had ever previously had bowel cancer and whether they had previously undergone a colonoscopy.

The rate of PCCRC was then calculated for each colonoscopy provider in the English NHS.

What did we find? 
The overall rate of PCCRC has fallen from 9% in 2005 to 6.5% in 2013. PCCRCs were more frequent in females, those with pre-existing health conditions, in people with inflammatory bowel disease and diverticular disease, in people who had undergone multiple colonoscopies and in people with a previous diagnosis of bowel cancer. PCCRCs were also more common in older patients.

The lowest rates of PCCRC were found in those performed as part of the Bowel Cancer Screening Programme. Colonoscopies performed at private providers for the NHS had much higher rates.

In the years 2011-2013 the rate of PCCRCs varied substantially across NHS Trusts with rates ranging from 3.3% in the best performing to 13.4% in the worst performing.

If all providers during the eight years of the study period had achieved the same rate of PCCRC as the colonoscopies performed as part of the Bowel Cancer Screening Programme, 3900 cases of colorectal cancer could have been diagnosed earlier or even prevented.

Conclusion
Although the rates of PCCRC are improving, significant difference between providers exists. There is a need for all providers of colonoscopy to examine their cases of PCCRC, and local reasons for them.

From this, providers may be able to identify measures which once put into practice will lead to improved rates and improved outcomes for patients.

The lower rates of PCCRC found in the Bowel Cancer Screening Programme colonoscopies is noteworthy. All colonoscopies performed as part of the Bowel Cancer Screening Programme take place within accredited screening centres by colonoscopists who have undergone an accreditation test and who adhere to strict performance criteria. This shows that where strict standards are applied, quality is better.

Key Findings

• There is significant variation in the rate of post-colonoscopy colorectal cancers (PCCRCs) in the English NHS from 13.4% in the worst performing to 3.3% in the best.
• Colonoscopies performed as part of the Bowel Cancer Screening Programme have the lowest rate of PCCRCs.
• If all providers during the eight years of the study period had achieved the same rate of PCCRC as the colonoscopies performed as part of the Bowel Cancer Screening Programme, 3900 cases of colorectal cancer could have been diagnosed earlier or even prevented.
• High risk groups were identified: patients with a previous diagnosis of cancer, patients who have had multiple colonoscopies, and patients with inflammatory bowel diseases.

A plain language summary of:
Regional multidisciplinary team intervention programme to improve colorectal cancer outcomes: study protocol for the Yorkshire Cancer Research Bowel Cancer Improvement Programme (YCR BCIP)

What is this study about?
Bowel cancer is a common disease in the UK with over 41,000 people diagnosed each year and 16,000 people dying from it.

Although the outcomes for patients with bowel cancer in England are getting better, they are still worse than those seen in many comparable countries.

In order to help address this, Yorkshire Cancer Research have funded a programme of work called the Yorkshire Cancer Research Bowel Cancer Improvement Programme (YCR BCIP). This programme aims to improve the outcomes of bowel cancer in the Yorkshire and Humber region.

This will be done through working with multidisciplinary teams (MDT’s) to understand any differences in practice and also to develop education workshops to help minimise the differences and improve outcomes across the region.

What will we do?
Routinely generated health data will be analysed to understand the differences in the demographics, treatments and outcomes of patients across Yorkshire and the Humber. The Programme will also collect Patient Related Outcome Measures (PROMS) data as well as tissue samples in order to help create a more complete picture of the care that takes place in the region.

Using this data, the Programme will engage with the MDT’s and will provide educational workshops, studies and audits in order to optimise practice.

A plain language summary of:
Current concepts in imaging for local staging of advanced rectal cancer. P.J.Brown, R. Hyland, A.J.Quyn, N.P.West, D. Sebag-Montefiore, D. Jayne, P.Sagar, D.J.Tolan.

Background:
Many advancements in the treatment of rectal cancer using surgery were made in the 1980’s and 1990’s. The practice of staging a tumour before surgery and identifying any high risk tumours, has enabled clinicians to be more selective in how the tumour is managed ahead of treatment and improved overall outcomes.

What did we do?
This review looks at the importance of the features of locally advanced rectal cancer using standard magnetic resonance imaging (MRI) and the potential impact of functional MRI techniques. Standard MRI scans produce 3D images of an area of the body whereas, Functional MRI scans produce images of activity within the specific area of the body.

What did we find?
MRI remains the best method to stage rectal cancer and to decide upon the best way of managing the tumour however, despite advancements in the quality of imaging, it is still a challenge for clinicians to accurately categorise tumour characteristics.

The paper identifies the need for further developments and education to gain a better understanding of the risks to the patient and improve personalised therapies. In addition to this the paper also identifies that work is required to improve the accuracy of rectal cancer staging in routine practice.

A plain language summary of:
What factors determine specimen quality in colon cancer surgery? - A cohort study. Kheng-Seong NG, Nicholas. P. West, Nigel Scott, Melanie Holzgang, Phil Quirke and David,G. Jayne

Background:
There is evidence that a patient’s survival from colon cancer is influenced by the quality of the surgery.

It is undecided, however, what factors determine good quality surgery and whether the quality of surgery can influence the future health of a patient and the risk of death.

This study looks at the quality of colon cancer surgery by looking at the section of the bowel that is removed during surgery, also known as the pathological specimen. This is done to look for factors that influence the quality of surgery and assess whether the quality of surgery affects the occurrence of complications afterwards. Clinical data was also used in the assessment.

What did we do?
The study included all patients who had undergone surgery to remove a primary colon cancer at St James’s Hospital, Leeds between January 2015 and December 2017.

The area of the colon that was removed during surgery was assessed. The quality of the specimen was assessed by looking at the area that was removed known as the plane of mesocolic dissection as well as the area and length of the tissue that was removed. Whether any complication occurred after surgery was also measured.

What did we find?
The study found that the majority of good quality specimens were from open surgery which is often associated with a bigger risk of complications occurring after surgery.

The study found that there is still a need for routine photographs and assessments of the pathological specimen to be taken in order to help understand what surgeons do with colon cancer in the 21^st century. It found that there is room for improvement in colon surgery, in particular in the quality of laproscopic surgery (also known as key hole surgery which is when a surgeon operates through smaller openings in the body instead of larger cuts), as well as surgeries that are performed in an emergency.

A plain language summary of:
Improving outcome prediction in individuals with colorectal cancer and diabetes by accurate assessment of vascular complications: Implications for clinical practice. Rebecca J. Birch, Amy Downing, Paul J. Finan, Simon Howell, Ramzi A Ajjan, Eva J. A. Morris

Background:
Up to 20% of people with cancer also have diabetes. Patients with both cancer and diabetes are more likely to have worse outcomes after surgery. For patients with bowel cancer and diabetes there is also a poorer chance of receiving treatment to cure the cancer, a higher rate of chemotherapy related toxicity as well as an increased risk of dying after surgery.
In order to understand what the relationship is between diabetes and the outcomes for patients with cancer, it is first important to understand the difference between patients with complicated (a person with other underlying health conditions which are caused by the diabetes) and uncomplicated diabetes (a person who has no other underlying health conditions that are linked to their diabetes). This would then lead to a better understanding of whether diabetes itself is associated with a higher risk or whether it is the complications that arise from diabetes that cause more risks for patients.

The aim of this study is to identify a reliable and practical system to categorise people with complicated and uncomplicated diabetes in order to better predict outcomes for patients with diabetes following bowel cancer surgery.

What did we do?
We looked at individuals who had been diagnosed with primary bowel cancer in England between 2005 and 2016 who underwent surgery to remove some or all of their bowel. From this group of people we then identified the individuals with diabetes. The status of the individual’s diabetes was determined using a measure called the adapted Diabetes Complications Severity Index (aDCSI) which is designed to measure how severe a patient’s diabetes is. We put individuals into three categories, no diabetes, uncomplicated diabetes and complicated diabetes and looked at the outcomes of patients in each of the three categories.

What did we find?
We found that patients with complicated diabetes have worse outcomes after bowel cancer surgery than patients with uncomplicated diabetes. We also found that patients with uncomplicated diabetes have similar outcomes as patients who do not have diabetes. This finding suggests that poor outcomes after surgery are linked to the complications arising from diabetes rather than from the diabetes itself. The results of this study show the importance of accurately defining diabetes complications when assessing the risks associated with bowel cancer surgery for patients.

A plain language summary of: Lynch syndrome screening in colorectal cancer: results of a prospective two-year regional programme validating the NICE diagnostics guidance pathway across a 5.2 million population Nicholas P West, Niall Gallop, Danny Kaye, Amy Glover, Caroline Young, Gordon G A Hutchins, Scarlet F Brockmoeller, Alice C Westwood, Hannah Rossington, Philip Quirke

Background:
In 2017 the National Institute for Health and Care Excellence (NICE) released new guidance which recommended that all patients newly diagnosed with bowel cancer in the UK should be screened for Lynch Syndrome.

Lynch Syndrome is a hereditary genetic condition that significantly increases the risk of a person developing bowel cancer as well as other types of cancers such as womb and stomach among others. It is one of the most common hereditary cancer syndromes and it is estimated that around 1 in 300 people may be carriers of a gene associated with Lynch Syndrome.

As Lynch Syndrome is genetic, after a patient has received a diagnosis, testing is offered to other family members. This testing is known as germline testing where tests are carried out on cells that do not currently have cancer to see if a person has a gene mutation such as Lynch Syndrome. If an individual is found to have a gene mutation such as Lynch syndrome, further screening, such as regular colonoscopies can be offered in order to help identify early tumours or prevent their development, therefore improving the outcomes for patients and saving lives.

The Yorkshire Cancer Research Bowel Cancer Improvement Programme is a programme of work with the primary aim to improve bowel cancer outcomes for patients in Yorkshire and the Humber. The programme included funding to carry out Lynch Syndrome screening for all newly diagnosed patients with bowel cancer over the age of 50 in the Yorkshire region until a time when the screening could be commissioned through the NHS. This paper reports the results of the two-year study as validation of the 2017 NICE guidance.

What did we do:
Lynch screening was offered to all bowel cancer Multidisciplinary Teams (MDT’s) in the Yorkshire and Humber region. All patients over the age of 50 who were diagnosed with bowel cancer between 1^st April 2017 and 31^st March 2019 were eligible for testing. Testing was not offered for patients under the age of 50 due to other screening guidance in place prior to 2017 which recommends the routine screening of bowel cancer patients under the age of 50.

The screening was carried out at the University of Leeds and involved a tissue sample removed at the time of surgery as part of the routine clinical care pathway being sent to the central laboratory for testing.

What did we find?
Over 3,000 individual cancers were tested during the two year study period. Of the 3,000 cancers tested, 81 patients were found to potentially have Lynch Syndrome and would require germline testing to confirm it.

As the largest UK regional study to validate the 2017 guidance on Lynch Screening to date, it was found that the guidance is achievable at scale and can be carried out routinely following engagement with local commissioners to ensure that the pathology departments have enough funding and resource.

In addition to screening for Lynch Syndrome, the results from this study are being used to inform the treatment of patients including the use of chemotherapy and immunotherapy.

A plain language summary of: Influence of age on surgical treatment and postoperative outcomes of patients with colorectal cancer in Denmark and Yorkshire, England John C. Taylor, Lene H. Iverson, Dermot Burke, Paul J. Finan, Simon Howell, Lars Pedersen, Mark M.Iles, Eva J.A. Morris, Philip Quirke, The YCR BCIP Study Group

Background:
The survival rate of patients with bowel cancer varies across Europe. Previous studies showed that the survival rate for patients with bowel cancer in Denmark and England were lower in the 1990’s and 2000’s when compared to many other countries with similar populations and health care systems. Since these studies were published, both Denmark and England have undergone changes in how they treat and manage bowel cancer in order to try and improve the outcomes of patients.

Some of the changes made to how bowel cancer is treated and managed are the same in both countries.  However, Denmark, unlike England, chose to analyse the data that is routinely collected when a patient interacts with the health care system.  This allows patterns of practice to be looked at, areas of concern identified, and changes implemented to address resulting concerns.  There is strong evidence that these changes have had a big effect on the management of bowel cancer in Denmark and have improved the outcomes of patients.

In 2016, a similar programme of work began in Yorkshire. The programme analyses routinely collected data in order to get a detailed picture of the patterns of care and outcomes of patients with bowel cancer in the region. Changes in practice can then be made directly by clinicians in order to address any variation in how bowel cancer patients are treated and managed in order to improve outcomes.

One of the key areas that has been looked at is bowel cancer surgery. Major surgical resection, which is surgery to remove the tumour and surrounding tissue, is the main treatment for bowel cancer. It has been suggested that the survival differences of patients with bowel cancer between England and Denmark may be due to more older patients in Denmark undergoing major bowel cancer surgery. This study aims to identify differences in the management of bowel cancer surgery and outcomes following surgery according to a patient’s age in Demark and Yorkshire.

What did we do?
The study analysed data on patients who were diagnosed with bowel cancer between 2005 and 2016 in Yorkshire and Denmark. The number of patients who had a major surgical resection, the number of patients who died following surgery and the number of patients who survived were compared between Yorkshire and Denmark across several age groups.

What did we find?
The study found that there are major differences in how elderly patients with bowel cancer are managed in Yorkshire and Denmark. More patients in older age groups were selected for major surgical resection in Denmark than in Yorkshire. The study concluded that with improvements in the care given to older patients both before and after surgery, more older patients could be found to be suitable for surgery in Yorkshire, improving longer-term outcomes for patients.

Author - Katie McSweeney and Emily Boldison

Lynch Syndrome - Hidden in Plain sight is a patient editorial published alongside the paper Lynch Syndrome screening in colorectal cancer: results of a prospective two-year regional programme validating the NICE diagnostic guidance pathway across a 5.2 million population.

A link to the paper is available here.

A plain language summary of: Differences in the management of patients requiring an emergency resection for colon cancer in two European populations

Background:
Between 15-30% of patients with bowel cancer have their cancer discovered by visiting emergency services and requiring emergency care. Patients who do not need emergency bowel cancer surgery often are referred through other routes such as a GP referral or the screening programme, known as elective care.

Comparing how bowel cancer patients are treated in different populations with similar demographics can help to find different and improved practices which may improve the patients’ outcomes. Denmark and Yorkshire are two populations which are very similar in population size and demographics and so they are a good comparator in how bowel cancer patients are treated.

New alternative treatments have been introduced which may reduce the number of patients who need to have emergency surgery. One example is self-expanding metallic stents (SEMS), which are used to stabilize patients to avoid them having to have emergency surgery enabling the option of planned, lower risk, elective surgery instead. However, due to the level of expertise needed to perform this treatment and given that there are no fixed guidelines on their use, the use of SEMS is likely to vary across regions. Denmark issued guidelines to recommend treating patients with SEMS for left-sided bowel tumours, in the place of emergency surgery, in 2009.

The aim of this study is to compare the use of emergency operations and elective operations for bowel cancer and the outcomes, after surgery, between Denmark and Yorkshire.

What did we do?
We identified all patients who received surgery to remove some or all of their bowel (known as a major resection) between 2005 and 2016 in Denmark and in Yorkshire. We then compared the number of patients who had received emergency surgery, the number of patients who received a stenting procedure before their surgery and the number of patients who had died up to 30 days following their surgery between the two populations.

What did we find?
This study found that emergency surgery in Denmark decreased by 3.7% from the period 2005-2007 to 2014-2016, whereas in Yorkshire, emergency surgery increased by 3.3%. It also found that Danish patients with left-sided tumours were less likely to have emergency surgery which coincides with an increase in the use of stents at this time. The rate of patients who died within 30 days of receiving any kind of surgery (emergency or elective) decreased in both Denmark and Yorkshire however a larger decrease was seen in Denmark.

Conclusion:
This study shows that, following the implementation of the SEMS guidelines in 2009, Denmark increased the number of stenting procedures and reduced the number of emergency operations. Some of the potential emergency operations became elective operations which, in turn, can lead to better outcomes for patients.

A plain language summary of: Artificial intelligence for detection of microsatellite instability in colorectal cancer - a multicentric analysis of a pre-screening tool for clinical application A Echle, N Ghaffari Laleh, P Quirke, H I Grabsch, H S Muti, O L Saldanha, S F Brockmoeller, P A van den Brandt, G G A Hutchins, S D Richman, K Horisberger, C Galata, M P Ebert, M Eckardt, M Boutros, D Horst, C Reissfelder, E Alwers, T J Brinker, R Langer, J C A Jenniskens, K Offermans, W Mueller, R Gray, S B Gruber, J K Greenson, G Rennert, J D Bonner, D Schmolze, J Chang-Claude, H Brenner, C Trautwein, P Boor, D Jaeger, N T Gaisa, M Hoffmeister, N P West, J N Kather

Background:
According to medical guidelines, all patients who are diagnosed with bowel cancer should be screened for a genetic condition called Lynch Syndrome. Lynch Syndrome is a condition which can be passed down through families and increases the chances of a person developing certain types of cancer such as bowel cancer. It is therefore important that people with Lynch Syndrome are identified so that family members can also be tested for it to see if they are also at an increased risk.

Screening for Lynch syndrome starts with testing the cancer tissue for abnormalities in its genetic makeup. This is known as mismatch repair testing. Around one in six bowel cancers show defective mismatch repair, of which a small number will have Lynch syndrome. As well as finding patients with possible Lynch syndrome, bowel cancer patients with defective mismatch repair may be offered a different treatment pathway to those with normal mismatch repair because the cancers behave differently and respond to different drug treatments.

Defective mismatch repair is usually diagnosed by conducting relatively expensive tests in a laboratory on cancer samples taken during biopsy or surgery. These tests also take several days to complete.. This study aimed to show whether defective mismatch repair can be diagnosed by developing a computer algorithm using Artificial Intelligence (AI) that can read standard pathology slides used for diagnosis without the need for additional expensive and slow tests.

What did we do?
We collected standard pathology slides from 8342 patients with bowel cancer from different countries and ethnicities with a known mismatch repair status found using current standard of care tests. We then trained AI models to detect defective mismatch repair.

What did we find?
We found that the AI system we developed is excellent at diagnosing defective mismatch repair with equivalent sensitivity to standard of care testing currently in use across the NHS.

Conclusion:
By using the AI system, defective mismatch Repair can be safely ruled out in up to half of patients. This would reduce the number of patients needing standard of care testing therefore reducing the costs and the turnaround time involved.

A plain language summary of: Swarm learning for decentralized artificial intelligence in cancer histopathology Oliver Lester Saldanha, Philip Quirke, Nicholas P. West, Jacqueline A. James, Maurice B. Loughrey, Heike I. Grabsch, Manuel Salto-Tellez, Elizabeth Alwers, Didem Cifci, Narmin Ghaffari Laleh, Tobias Seibel, Richard Gray, Gordon G. A. Hutchins, Hermann Brenner, Marko van Treeck, Tanwei Yuan, Titus J. Brinker, Jenny Chang-Claude, Firas Khader, Andreas Schuppert, Tom Luedde, Christian Trautwein, Hannah Sophie Muti, Sebastian Foersch, Jakob Nikolas Kather

Background:
Artificial Intelligence (AI) is expected to revolutionise medicine and healthcare over the next 10 years. Medical imaging, which refers to technologies like X-ray and MRI, used to view the body in order to manage medical conditions, has already been transformed by the use of AI. As well as being able to automate the interpretation of medical imaging, AI can also see information that is not visible to the human eye. This could also be very useful in cancer pathology, when cancer tissue taken from a biopsy or during surgery is examined by a pathologist under a microscope.

To ensure that an AI model is useful to clinicians, it needs access to a lot of data about patients and usually this data is stored in one place. However, when training AI systems on data from patients across many hospitals and even countries, the data that is needed is usually produced in many different places and there are a lot of practical, ethical and legal challenges in bringing this data together.

These challenges can potentially be overcome through a practice called Swarm Learning. Swarm learning is where a number of different partners, in this case, those who hold the data that is needed, jointly train the same AI model, so that the actual patient data does not need to be shared or transferred to a central place.

In this study we aimed to look at whether Swarm Learning can be used to train AI models to predict DNA changes in cancer tissue using routinely produced pathology slides.

What did we do?
In this study, using Swarm Learning, we trained AI models on three patient datasets from Northern Ireland, Germany and the United States of America.

What did we find?
In this study we show that AI models trained used Swarm Learning can accurately predict the presence of genetic changes within the cancer of patients with bowel cancer including mutations in the BRAF gene and microsatellite instability directly from routine pathology slides. This shows that  Swarm Learning can be used successfully to train AI models to analyse pathology slides and that they are effective and work efficiently.

Conclusion:
In the future, Swarm Learning can be used to train AI models for any pathology image analysis tasks which will overcome the need for patient data to be transferred between partners.

A plain language summary of: Process, structural and outcome quality indicators to support perioperative opioid stewardship: a rapid review C. Thomas, M. Ayres, K. Pye, D. Yassin, S.J.Howell and S Alderson

Background:
Opioids are a group of drugs which are used to relieve pain. Although opioids are very effective painkillers, they can also cause harm to patients. Prescribing opioids to patients inappropriately is a problem and is considered to be a threat to the population and a challenge to health care services internationally. Despite increasing awareness of the risks of opioid misuse, prescription opioid use is still high in both Europe and the US.

Opioids are both effective and commonly used painkillers around the time of surgery. However they can have unwanted effects and if opioid use continues in the long term, problems including slower recovery, addiction, dependence and an increased risk of death may occur.

Opioids must be used safely. Opioid stewardship includes the use of the best opioid for a particular patient, in the correct dose and prescribed for the correct duration of time. Effective opioid stewardship is therefore important when opioids are given around the time of surgery. A need to improve this has been identified.  Improving opioid stewardship around the time of surgery for patients with bowel cancer has the potential to improve recovery, lead to faster discharge, improve outcomes and most importantly, prevent patient harm.

Health care quality indicators are a type of performance measure that evaluate aspects of quality of care. They are used to monitor and improve the quality of healthcare.

What did we do?
A rapid review was performed to find research and review papers where quality indicators for opioid use around the time of surgery are described. The review included 628 abstracts and 118 publications. In total 128 indicators were extracted. Duplicates were merged, leaving a final total of 24separate indicators.

What will we do next?
The 24 quality indicators will be used to develop a toolkit to guide and improve the safety of opioid use around the time of surgery. The next step will involve an expert clinical panel. They will agree which of the quality indicators will be most valuable in the Yorkshire and Humber region for the management of opioids for patients having bowel cancer surgery.

A plain language summary of: Generalized biomarker prediction from cancer pathology slides with self-supervised deep learning: A retrospective multi-centric study. Jan Moritz Niehues, Philip Quirke, Nicholas P West, Heike I Grabsch, Marko van Treeck, Yoni Schirris, Gregory P Veldhuizen, Gordon G A Hutchins, Susan D Richman, Sebastian Foersch, Titus J Brinker, Junya Fukuoka, Andrey Bychkov, Wataru Uegami, Daniel Truhn, Hermann Brenner , Alexander Brobeil, Michael Hoffmeister, Jakob Nikolas Kather

Background:
Deep Learning is a type of Artificial Intelligence in which computers are taught to process large volumes of data, often in a superior way to humans. Deep Learning models are able to recognise complicated patterns in pictures, text, sounds and other data to provide insights and predictions.

In pathology, Deep Learning can be used to predict common DNA changes in cancers from routinely collected histopathology tissue slides. In bowel cancer, these include Microsatellite Instability (MSI) and KRAS/NRAS/BRAF mutation status . These insights enable doctors to diagnose hereditary cancer syndromes and make decisions on a patient’s treatment.

This paper examines whether Deep Learning can be used to accurately predict these DNA changes in patients with bowel cancer.

What did we do?
Using tissue slides from patients with bowel cancer, we compared six different Deep Learning models to predict key biomarkers required in the treatment pathway including MSI and mutations in BRAF, KRAS, NRAS and PIK3CA.

What did we find?
Our study found that Deep Learning can be used to accurately predict MSI and BRAF mutations. Current standard of care testing is expensive and time consuming. A move to Deep Learning for these biomarkers would save both time and money. However, even with powerful Deep Learning models, using current technology it is not possible to predict the mutation status of KRAS, NRAS and PIK3CA accurately enough to change testing pathways for these markers.

A plain language summary of:  National variation in pulmonary metastasectomy for colorectal cancer. Fenton HM, Finan PJ, Milton R, Shackcloth M, Taylor JC, Treasure T, Morris EJA.

Background:

Bowel cancer is the fourth most common cancer in the UK. At diagnosis, around 20% of patients (which subsequently increases to 50% of patients) with bowel cancer are told that their cancer has metastasised. This means that the cancer has spread to another part of the body. One place that the cancer often spreads is to the lungs. It is uncommon for the cancer to spread only to the lungs in patients with bowel cancer, with over 80-85% of patients also having metastases in other areas, most commonly the liver.

There are a number of recommendations on what surgical practice should be taken in the NHS when dealing with bowel cancer lung metastases. One such recommendation is a pulmonary metastasectomy which is the surgical removal of tumours in the lungs. Despite this being a recommendation, there does not appear to be any large scale trials looking into the practice of pulmonary metastasectomy.

This study aims to investigate the use of pulmonary metastasectomy in the bowel cancer population across the English NHS and to find the extent of any differences in practice amongst NHS hospitals and differences in outcomes for patients.

What did we do?
All patients with a primary bowel cancer (this means that the cancer originated in the bowel) and who had received surgery to remove the tumour at an NHS hospital between 1 January 2005 and 31 December 2013 were identified. From here, all patients who had then undergone a pulmonary metastasectomy within three years of the original cancer surgery were identified.

What did we find?
2% of patients who had surgery to remove a tumour in the bowel went on to have a pulmonary metastasectomy within three years. The number of patients receiving a pulmonary metastasectomy varied significantly between hospital trusts.

Plain language summary of: Encrypted federated learning for secure decentralized collaboration in cancer image analysis. Daniel Truhn, Soroosh Tayebi Arasteh, Oliver Lester Saldanha, Gustav
Müller-Franzes, Firas Khader, Philip Quirke, Nicholas P. West, Richard Gray, Gordon G. A. Hutchins, Jacqueline A. James, Maurice B. Loughrey,Manuel Salto-Tellez, Hermann Brenner, Alexander Brobeil,
Tanwei Yuan, Jenny Chang-Claude, Michael Hoffmeister, Sebastian
Foersch, Tianyu Han, Sebastian Keil, Maximilian Schulze-Hagen,
Peter Isfort, Philipp Bruners, Georgios Kaissis, Christiane Kuhl,
Sven Nebelung, Jakob Nikolas Kather

Background:

Artificial Intelligence (AI) and machine learning techniques are having a huge impact on cancer imaging and cancer research. They have been used in the detection and diagnosis of cancer to classify the type of cancer, to improve the cancer treatment as well as to analyse radiology images and digital pathology slides.

For AI models to be able to work they need to be trained. This requires large amounts of data. Often the data that is needed is spread out in a number of different locations. Collecting the data to bring it all into one centralized site is not often possible due to the issues which come with ensuring that data is kept secure, private, accurate, available and useable, this is known as data governance.

In order to overcome the data governance obstacles, models such as Federated Learning and Swarm Learning can be used. Federated Learning allows several sites that hold data to jointly train machine learning models without needing to bring all of the data together to one central location. Only the summary (or aggregated) data is shared. However, as the data is still transferred to the aggregated server to merge the datasets, there is still a possibility for data security breaches.

Homomorphically Encrypted Federated Learning (HEFL) is a process where data is in a plain, unreadable coded format which means it can not be read or understood by anyone who does not have the code to decipher it. This keeps the data secure.

In this study, we aim to show that using HEFL can be used to train AI models to analyse cancer imaging on radiology and histopathology datasets.

What did we find?

Our study found that AI models for cancer image processing can be trained securely on data from different centres ensuring patient privacy. This will allow joint working between researchers from different centres (such as different universities or hospitals) which can lead to the development of advanced AI models which are clinically useful.

A plain language summary of: Transformer-based biomarker prediction from colorectal cancer histology: A large-scale multicentric study. Sophia J Wagner, Daniel Reisenbüchler, Nicholas P West, Jan Moritz Niehues, Jiefu Zhu, Sebastian Foersch, Gregory Patrick Veldhuizen, Philip Quirke, Heike I Grabsch, Piet A van den Brandt, Gordon G A Hutchins, Susan D Richman, Tanwei Yuan, Rupert Langer, Josien C A Jenniskens, Kelly Offermans, Wolfram Mueller, Richard Gray, Stephen B Gruber, Joel K Greenson, Gad Rennert, Joseph D Bonner, Daniel Schmolze, Jitendra Jonnagaddala, Nicholas J Hawkins, Robyn L Ward, Dion Morton, Matthew Seymour, Laura Magill, Marta Nowak, Jennifer Hay, Viktor H Koelzer, David N Church; TransSCOT consortium; Christian Matek, Carol Geppert, Chaolong Peng, Cheng Zhi, Xiaoming Ouyang, Jacqueline A James, Maurice B Loughrey, Manuel Salto-Tellez, Hermann Brenner, Michael Hoffmeister, Daniel Truhn, Julia A Schnabel, Melanie Boxberg, Tingying Peng, Jakob Nikolas Kather

Background:
Deep Learning is a type of Artificial Intelligence in which computers are taught to process large volumes of data, often in a superior way to humans. Deep Learning models are able to recognise complicated patterns in pictures, text, sounds and other data to provide insights and predictions.

In pathology, Deep Learning can be used to speed up the process to predict common DNA changes in cancers from routinely collected histopathology tissue slides. In bowel cancer, these include Microsatellite Instability (MSI) and KRAS/NRAS/BRAF mutation status. These insights enable doctors to diagnose hereditary cancer syndromes and make key decisions on a patient’s treatment.

Current processes using Deep Learning rely on convolutional neural networks (an algorithm that is used to recognise patterns in data) and have mostly been tested on small numbers of patients. In this study we develop a new way of flowing data to predict common DNA changes in cancer from histopathology tissue slides.

What did we do?
We used a new type of Deep Learning with a transformer-based pipeline to test common DNA changes in over 13,000 patients with colorectal cancer across 16 cohorts. This was compared to current state-of-the-art algorithms.

What did we find?
Our transformer-based approach greatly improved the performance, generalizability, data efficiency and interpretability or predicting DNA changes in cancer from routinely collected pathology slides when compared to other current algorithms. For the prediction of MSI, the sensitivity was over 99%.

A plain language summary of: Clinical-Grade Detection of Microsatellite Instability in Colorectal Tumors by Deep Learning.  Ms. Amelie Echle Prof. Heike Irmgard Grabsch, Prof. Philip Quirke, Prof. Piet A. van den Brandt, Dr. Nicholas P. West, Dr. Gordon G. A. Hutchins, Dr. Lara R. Heij, Ms. Xiuxiang Tan, Dr. Susan D. Richman, Mr. Jeremias Krause, Dr. Elizabeth Alwers, Ms. Josien Jenniskens, Ms. Kelly Offermans, Prof. Richard Gray, Prof. Hermann Brenner, Prof. Jenny Chang-Claude, Prof. Christian Trautwein, Prof. Alexander T. Pearson, Prof. Peter Boor, Prof. Tom Luedde, Prof. Nadine Therese Gaisa, Prof. Michael Hoffmeister, and Dr. Jakob Nikolas Kather

Background:
Mismatch repair deficiency (dMMR) is found in 10-20% of patients with bowel cancer and causes Microsatellite Instability (MSI). People who have dMMR and MSI tend to have very distinct types of bowel cancers which need to be treated using treatment options that may be different from standard treatment.  It is therefore very important that patients with bowel cancer are tested for dMMR when they are diagnosed so that the appropriate treatment is given.

At present, dMMR and MSI are identified using expensive testing procedures, however deep learning could provide a faster and more cost-effective alternative.

In this study, we aim to use deep learning technology to detect microsatellite instability (MSI) and mismatch-repair deficiency (dMMR) in bowel cancer tumours tumors.

What did we do?
A large number of bowel cancer tissues slides were collected from five different countries. We used deep learning (an AI method which teaches computers to process data in a way that is the same as the human brain) to train a system to identify the tissue samples with MSI.

What did we find?
The deep-learning system successfully identified tissue specimens with dMMR and MSI and so demonstrates the successful development of a deep-learning system capable of detecting bowel cancer specimens with dMMR or MSI using standard tissue slides.