THE WYNN DATABASE - METABOLISM AND MORTALITY

The purpose of the Wynn Database is to provide a resource for analyses of the roles of metabolic risk factors in the pathogenesis of chronic disease including cardiovascular disease, diabetes and cancer. Mortality status and cause of death is sought from NHS England to be incorporated into the Wynn Database.

The Wynn Database preserves a unique body of electronic and paper records that was accumulated between the years 1965 and 2000 under the direction of the metabolic medicine research team at St Mary’s Hospital Medical School (one of the constituent schools of Imperial College London). This work was then migrated to what became the Wynn Institute of Imperial College London (ICL) in 1998.

The Wynn database continues to be a useful resource in understanding the pathophysiology of diabetes, heart disease and cancer. The Wynn Database comprises 29,244 records of metabolic information for 14,615 individuals. The data derives from studies carried out on ethnically diverse, healthy volunteers (recruited by advertisement, magazine articles or personal contact), and clinic patients receiving anabolic steroid therapy, anti-androgen therapy, oral contraceptives, postmenopausal hormone replacement therapy. Coronary heart disease patients, heart failure patients, lipid clinic patients, obesity clinic patients, endocrine clinic patients are also represented.

Imperial College London request permission from NHS England to link the Wynn database to the pseudonymised Civil Registration Deaths data on an annual basis. This linkage to mortality information held by NHS England, will enable extensive retrospective and prospective analyses of relationships between metabolic risk factors, survival time and cause of death. Provision of age at death and cause of death information for participants represented in the Wynn Database will add key information to this exceptionally rich data collection. Linkage of baseline metabolic information recorded many decades ago to eventual age and cause of death will enable better discrimination of known predictors of chronic disease and, potentially, identify novel predictors or prediction profiles. The information recorded in the Wynn Database is of particular relevance to diabetes, vascular disease and cancer and can provide for a deeper understanding of the role of metabolic risk factors in the development of such chronic diseases. Analysis of the associations between baseline metabolic measures, survival time and cause of death will be a major aspect of work involving the Database.

It is anticipated that inclusion of mortality information in The Wynn database will enable testing of a range of novel hypotheses. Examples of objectives that can be immediately addressed include the following:

1. Does individual variability in metabolic risk indicators relate to mortality outcomes?

Metabolic risk factors for cardiovascular disease are generally evaluated on the basis of single measurements. However, preliminary studies in the course of the Heart Disease and Diabetes Research Indicators in a Screened Cohort (HDDRISC) study have shown that increased variability in risk factor levels may indicate increased risk of adverse health outcomes. The Wynn Database holds data for multiple serial evaluations in single individuals of a variety of metabolic risk indicators. Around 2,000 participants have information on fasting plasma glucose and insulin and serum lipids from which measures of metabolic instability may be derived. With mortality outcomes in each individual, it will be possible to explore further the clinical implications of metabolic variation.

2. Can oral glucose tolerance test glucose and insulin profile features provide novel information regarding cardiovascular mortality risks?

The risk of developing cardiovascular disease is increased in impaired glucose tolerance and diabetes but there may be further variations in risk among normoglycemic individuals. There is, currently, interest in the possibility that sub-clinical variations in oral glucose tolerance test (OGTT) glucose and insulin profiles may have health implications. However, investigations into this possibility have been limited by lack of suitable data and of clinical outcome information. In a Wynn Database pilot analysis of 397 normoglycemic individuals participating in the HDDRISC study (unpublished), we have used cluster analysis of OGTT glucose and insulin measurement sets to classify key profile features into 10 discrete categories. Independently of individual characteristics, insulin resistance and beta cell function, associations were found between profile features and lipid and blood pressure risk factors for cardiovascular disease, including diastolic blood pressure, triglycerides, insulin and high-density lipoprotein (HDL) cholesterol. The complete Wynn database can provide OGTT glucose and insulin profile information for 7107 individuals. These data can provide for a substantially expanded analysis of relationships between OGTT profile features and cardiovascular disease risk factors and, with mortality follow-up for each participant, could enable profile features to be related to risk and cause of death.

3. Does the metabolic response to oral contraceptive use relate to subsequent mortality and cause of death?

There have been many studies of the effects of oral contraceptive use on mortality in cohorts of women who, on recruitment, were either taking or not taking oral contraceptives. These studies identified increased risks of venous thromboembolic and arterial disease associated with oral contraceptive use. Combined oral contraceptives can have marked adverse effects on glucose and lipid homeostasis that are consistent with these risks. However, no study has investigated the relationship between metabolic response to oral contraceptive use and long-term mortality. With the Wynn Database, relationships between oral contraceptive-induced changes in metabolic risk indicators and mortality can be studied in 3311 women who were oral contraceptive users at the time of receiving an OGTT. As a reference group, there are 4,996 women aged 45 or less who were not oral contraceptive users at the time of receiving an OGTT.

While the recruiting and consenting, methods used at the time would not meet the standards of today, ICL maintain that the data was collected without deception, and the nature of how the personal data is processed has been communicated openly and honestly by maintaining the study’s website. (https://www.imperial.ac.uk/metabolism-digestion-reproduction/research/diabetes-endocrinology-metabolism/metabolic-medicine/wynnmet/). Consent by participants represented in the database for continued holding of the data and linkage to mortality information by NHS England is not feasible since last known addresses date back 20 years or further and are, in any case, only available for a minority of participants (~15 percent). Accordingly, Section 251 support for continuing Database studies has been sought from and granted by the Confidentiality Advisory Group. A condition of this support is that once the identifiable cohort data held at Imperial College London has been supplied to NHS England for linkage purposes the identifiers (full name and last known postcode) held by Imperial College London will be destroyed. The destruction of identifiers is estimated to be complete within two years of CAG approval.

At present the following information is contained within with Wynn database includes (this includes personal identifiers)

• surname

• forename

• last known postcode (recorded for ~15% of participants)

• date of visit and age at visit (age available for 13,777 participants)

• Gender

• height

• weight

• anthropometric measurements such as skinfold thicknesses

• blood pressure

• current and past diagnoses

• medication history

• a broad range of blood measurements relating to risks of diabetes, heart disease and cancer.

Once the linkage is complete, ICL will delete all identifiable information from the database to implement pseudonymised working. This includes name, date of birth, sex and last known postcode. Of those, only year of birth will be retained as this provides potentially useful information on the time period of the individual's early years. ICL will retain individual characteristic information, which will included age in years, sex, ethnicity, height and weight and the extensive de-identified clinical, anthropometric and biochemical information the Database holds.

All data contained within the Wynn database was initially obtained based on consent, however the consent material did not explicitly reference the further analysis outlined within this Agreement, hence on 26/04/2021 the study received support under section 251 of the NHS Act 2006. A condition of the section 251 support is that once the identifiable cohort data held at Imperial College London has been supplied to NHS England for linkage purposes the identifiers (full name and last known postcode) held by Imperial College London will be destroyed. The destruction of identifiers will be completed within two years of receiving said approval.

To address the GDPR Principle of Data Minimisation only fields necessary for carrying out the planned analyses have been requested, this has been determined to be age at death and cause of death only.

With linkage to NHS number and pseudonymised provision of age at death and cause of death by NHS England, specific metabolic variables or metabolic variables combined in so-called 'metabolic signatures' can be assessed as predictors of survival and cause of death. Previous studies, including studies based on a subset of the information held in the Wynn Database, have demonstrated the potential relevance of the data to the pathogenesis and risks of diabetes, cardiovascular disease and cancer. The greatly expanded numbers of observations made available in the full Database hope to enable confirmation and extension of these previous studies and has the potential to identify novel associations between metabolic disturbance and long-term health outcomes.

Details of the intended operating procedures and of the use of a 'database number', entirely unrelated to 'member number' to enable continuity with previous analyses as well as member number-free exports of selected items of data, if necessary, are as follows:

Currently, the Wynn Database is contained in two electronic files at ICL. One contains non-identifiable clinical information, with individual participants distinguished by a unique database number, and includes age, year of birth, sex and ethnicity. The second contains identifiable information (name, date of birth, sex, and for limited numbers, last known postcode) and is currently securely stored in a Section of Metabolic Medicine-specific, project-specific, limited-access encrypted folder provided and managed by Imperial College's Information & Communication Technologies Division. The encryped file of personal information is held in isolation from the clinical dataset, again with participants distinguished by the unique database number. For uploading to NHS England for linkage and provision of mortality information, the identifiable data file in the encryption folder will be migrated to the project SEFT account. Individuals included in the identifiable data file to be sent to NHS England will then be distinguished by 'member number' and mortality information provided by NHS England will be by member number. A file of correspondences between member number and the current 'database number' identifier will be retained in the project-specific, encryption folder to enable merging of mortality information with Individual database information. Analyses of the data, continuity of data analyses and data exports will distinguish individuals by 'database number'. 'Member number' will not appear outside the project-specific, limited-access encrypted folder and only NHS England will hold correspondences between identifiable information and member number. The difference, therefore, between member number and database number is that member number will only appear in the records held by NHS England and in the section's project-specific, limited-access encrypted folder and will only be used as the identifier for the pseudonymised mortality information provided by NHS England. Database number, on the other hand will be used as the identifier for on-going analyses and any data exports that may be considered necessary.

The combination of the unique metabolic information held in the database and the mortality information provided by NHS England will, therefore, enable a range of different hypotheses regarding antecedents of health outcomes to be tested, and the expanded scope of the Wynn database with mortality information added is likely to connect with the research interests of other groups. Access to database information will be granted via collaborative research proposals developed by or in consultation with the Head of the Section of Metabolic Medicine, with review by and discussion with other members of the Management Group. All proposals are expected to fall within the scope of existing protocol and ethical approval for the study database.

As the database is relevant to the research specialities of the Section of Metabolic Medicine, it is expected that the great majority of proposals will be generated within the Section of Metabolic Medicine at ICL, if proposals are generated outside of Imperial College an amendment will be required to this Agreement.

As the database will be pseudonymised it will not be possible for any investigators, including the applicants, to link either member number or database number to personal identifiers. There is no requirement to re-identify participants once the database has been fully pseudonymised.

Imperial College London is the sole data controller and the only organisation that will process data under the proposed data sharing agreement.

The lawful basis under Article 6 of the UK General Data Protection Regulation (GDPR) 2018 for processing the data is that it is a Public Interest Task 6(1)(e), ‘ the processing is necessary for you to perform a task in the public interest or for your official functions, and the task or function has a clear basis in law.’ As a not-for-profit, national and international medical research and educational organisation, Imperial College London can rely on this basis.

The condition under Article 9 of UK GDPR for processing the data as special category data is purpose 9(2)(j), ‘Archiving, research and statistics (with a basis in law)’ and the processing of data is in accordance with Section 1 Part 1 paragraph 4 of the Data Protection Act 2018.

All organisations party to this agreement must comply with the Data Sharing Framework Contract requirements, including those regarding the use (and purposes of that use) by “Personnel” (as defined within the Data Sharing Framework Contract i.e: employees, agents and contractors of the Data Recipient who may have access to that data).

Section 251 support for this agreement covers the following in terms of processing:

• ICL to retain the Wynn database, which includes confidential patient information, between the date CAG support is provided, and the date identifiers deleted (which the applicants estimate will take a period of 2 years).

• ICL to send confidential patient information from the Wynn database held at Imperial College London (ICL) to NHS England for the purposes of linking with Personal Demographics Service (PDS) to find NHS Number, and then to Civil Registration Deaths (mortality) data in order to provide the applicants with a pseudonymised dataset, annually, for the purpose of incorporating it in to the existing Wynn database, and updating it annually with mortality data,

• The flow of data from NHS England back to the applicants will not be complete until the participant identifiers, retained in the Wynn Database, are destroyed as the applicant would be able to reidentify the participants at this stage.

• NHS England to retain the key between the identifiers and member number, to facilitate the annual linkages with mortality data.

Data Flow:

ICL will securely flow the following identifying information to NHS England for the purpose of supporting linkage to pseudonymised Civil Registration Deaths data:

• WD Personal Member Number (Study ID)

• Surname

• Forename 1, Forename 2, Forename 3

• Other given name

• Date of Birth

• Gender

• Last Known Postcode - only applicable for 15% of the cohort

NHS England will link to Civil Registration Death data using the following identifiers:

- Forename

- Surname

- Other given name

- Date of Birth

- Gender

- Last known postcode

NHS England will inform the Wynn Database team when this is complete and the Wynn Database team will establish pseudonymisation by destroying all personally identifiable information relating to the Wynn Database by ICL .

Upon receipt of the NHS England data the Wynn Database team will destroy the following identifiers:

Surname

Forename 1

Forename 2

Forename 3

Date of birth

Last known postcode

There will be no further recruitment into the cohort. ICL will receive an initial pseudonymised dissemination from NHS England for cohort members who have already passed away, followed by pseudonymised Civil Registration Death data on an annual basis. NHS England will retain a copy of the cohort flagged to produce the data for future disseminations.

The Wynn Database identifiable information file is held in a Section of Metabolic Medicine-specific, limited access encrypted folder, provided and managed by Imperial College's Information & Communication Technologies Division (ISO 27001 certificate number 210676: next review due 31/10/2025). Encryption is by Symantec Encryption Desktop and no cloud-based systems are used.

All individuals processing the data will be a) substantive employees of ICL or b) ICL students with appropriate contractual measures in place. It is expected that around 2-4 students will be processing that data disseminated under this Agreement within any given year. All users need to provide evidence of Data Protection and Information Security Awareness training in order to be granted an access to secure environment.

Findings of analyses utilising the Wynn Database are expected to be published in peer-reviewed scientific journals (subject to acceptance) and in summary form on the study website: (https://www.imperial.ac.uk/metabolism-digestion-reproduction/research/diabetes-endocrinology-metabolism/metabolic-medicine/wynnmet/ ). All such outputs will report aggregated results only with small numbers suppressed, and no individual will ever be identified. Journals in which analyses may be published are likely to be those in which previous analyses employing data included in the Wynn Database have been published, for example: Diabetic Medicine, International Journal of Obesity, Quarterly Journal of Medicine, Cancer Causes and Control, Disease Markers, European Journal of Endocrinology (for full list, see the Wynn Database website)

The Wynn database is exceptionally rich in combinations of potentially informative metabolic variables, specifically those relating to glucose-stimulated insulin secretion, fat metabolism, blood pressure and anthropometric variables, including various measures of adiposity.

Except for mortality information, the Wynn Database is complete and no further recruitment or data acquisition is required for analyses to begin. Moreover, by 2022, a large number of participants will likely be deceased, meaning that informative analyses can begin as soon as linkage to mortality data is complete. Moreover, with approval from the HRA Confidentiality Advisory Group, analyses employing Wynn Database information have already begun, with one manuscript currently in draft. In principle, analyses could continue until all those represented in the Database have died. By 2022 the youngest individual represented in the database would have been 39 years of age. The useful life of the Database may, therefore, be appreciable. Published papers with findings from the analyses based on the mortality information have a target date of 2024.

All analyses generated by the Wynn Database are expected to, subject to acceptance, be published in high impact peer review journals with a focus on risk factors for diabetes, cardiovascular disease and cancer and the target audience for these publications aim to primarily be the scientific community, medical professionals, and the general public.

Given that a particular strength of the Wynn Database is the number and quality of glucose and insulin measurements made, particular attention is expected to come from those interested in diabetes and its complications, which include cardiovascular disease and may include certain types of cancer. Presentations outlining findings hope to, therefore, be appropriate for national and international conferences attended by diabetes clinical care and research professionals, specifically the annual conferences of Diabetes UK, the European Association for the Study of Diabetes and the American Diabetes Association.

Findings may also be of interest to the lay community both in general and among those living with diabetes. Lay summaries of on-going research and new findings expect to be published on the Wynn Database website. To date, public and patient involvement and engagement has included canvasing of opinions via from the ICL Section of Metabolic Medicine Patient and Public Involvement (PPI) group and the Chief Investigator's Twitter feed (https://mobile.twitter.com/mortdecai). Importantly, the Imperial College Public Experience Research Group organised a 90-minute online discussion group session in July 2021 focusing on issues relating to the Wynn Database and its use for research of personal information unconsented for further analyses. Twenty-five members of the public participated. Attendees were patients or carers contacted through existing contacts and patient groups linked to Imperial College's Section of Metabolic Medicine PPI and Diabetes Technology Groups; the Guy’s and St Thomas’s diabetes peer support group individuals with relevant lived experience within the VOICE North West London Research Involvement Network. The full 35-page report of the meeting is publicly available (http://hdl.handle.net/10044/1/94126). It is intended that the Wynn Database Management Group work with Imperial's Patient Experience Research Group to establish a Data Access Committee, with public representation, who check that requests to use the data for research purposes are appropriate and for public benefit.

The primary contributions the Wynn Database analyses hope to make to public health concern is a better understanding of the pathogenesis of chronic diseases and the possibility of identifying novel risk measurement profiles with which significant reductions in disease incidence may be projected. Whilst neither better understanding nor projected reductions in disease incidences directly impact on health, there is no doubt that both can inform prevention strategies and, importantly, justify prevention trials designed to determine whether projected incidence reductions can be truly achieved. Such trials can then have measurable and substantial impacts on health. Public interest concerns regarding chronic diseases such as diabetes, vascular disease and cancer that work with the Wynn Database can contribute to primarily involve possibilities for more focused disease prevention messages. The economic costs of chronic disease, particularly in relation to the current diabetes epidemic, for example, mandate investigation of any data resource that can extend our understanding of disease pathogenesis. Accordingly, dissemination of findings from the Wynn Database has the potential to support and reinforce promotion of health as a consequence of their focus on early metabolic indicators of long-term health risks. Variation in metabolic indicators underlies many of the connections between individual characteristics and lifestyles and their potential health consequences. The broad range of metabolic measurements held in the Database and exceptionally large number of observations recorded is hoped to strengthen conclusions and, therefore, the certainty with which health promotion messages generated by the analyses can be communicated.

Moreover, a better understanding of disease pathogenesis and the identification of more discriminatory indices for evaluating disease risk may contribute to and reinforce public perceptions of measures needed to reduce long-term morbidities, with all the advantages that has in relation to individual wellbeing and public health expenditure. On-going public involvement in decisions regarding work with the Database hopes to enable monitoring of how health promotion messages generated by Wynn Database analyses are likely to be received by the wider public.

Publications arising from analyses employing the Wynn Database may help inform healthcare professionals as to contributing causes of major chronic diseases, as well as the relative importance of different metabolic measures in identifying likely individual health outcomes in the long-term. Analyses hope to have the potential for identifying candidate measurements for better-focused risk assessments and preventive measures that can either be readily implemented or become the subject of dedicated trials in at-risk groups.

Expected benefits accruing from the Wynn Database are hoped to be achieved primarily by dissemination of the findings in scientific journals, which are expected to be read and acted upon by the wider healthcare and research community, according to the directions the findings give to chronic disease risk evaluation and prevention. In this respect, citations of published analyses from the Wynn Database hope to give an indication of the impact that findings have had. Given that the diseases in question constitute some of the most common causes of long-term morbidity, in particular diabetes and diabetes-related complications, Wynn Database analyses could, therefore, have the potential for broad public health impacts

Since 1966, the Wynn Database has made unique contributions to our understanding of the metabolic antecedents of diabetes as well as cardiovascular disease-related metabolic effects of oral contraceptives and postmenopausal hormone replacement therapy, and has contributed to knowledge of metabolic and inflammatory risk factors for cardiovascular disease in men.

Most recently, the Wynn Database is contributing to our understanding of key physiological influences on the principal investigative procedures that have been used for many years in research into the pathophysiology of diabetes.