Lymphoma is a blood cancer, the fifth most common cancer type in the UK. It can affect all areas of the body, at any age. The two main types are: Hodgkin (HL) and Non-Hodgkin lymphoma (NHL). Lymphoma can only be detected through specialist, secondary care hospital services, accessed through general practitioner (GP) referrals.
Lymphoma patients often experience delays in diagnosis, which are associated with advanced disease and poor survival. Lymphoma patients are more likely to have multiple GP consultations before hospital referral, less likely to have an urgent GP referral and have an increased risk of diagnosis after emergency presentation. The reasoning for delayed diagnosis is three-fold. Firstly, lymphoma symptoms are non-specific and can often be mistaken for other less serious illnesses; these include tiredness and fevers. Secondly, there are inadequate tests to detect lymphoma. Thirdly, GPs have limited exposure to lymphoma, and the current referral guidelines exclude the wider range of symptoms experienced in patients.
Our first study will determine patterns that occur prior to a lymphoma diagnosis, exploring the differences in features (symptoms, blood test results and prescribed medications) of healthy patients and lymphoma patients in general practice. The second study will create a prediction tool using these features to calculate the risk of patients developing lymphoma. We will obtain data from GPs of patients of all ages and utilize Statistical modelling analysis methods.
This research will develop diagnostic tools to alert GPs to a patient’s risk of lymphoma, allowing timely referrals and expedited diagnosis.
Lymphoma is a haematological cancer that develops in the lymphatic system. It is the fifth most common cancer, with over 14,000 people diagnosed annually in the UK. Lymphoma is classified into Hodgkin lymphoma and the more common subtype Non-Hodgkin lymphoma.
Lymphoma diagnosis is often delayed. Lymphoma is frequently diagnosed after multiple GP referrals and following emergency presentation, associated with poor survival. Diagnostic delays within primary care can be attributed to the non-specific presenting symptoms, also associated with common benign conditions and the lack of reliable investigative tests.
This project aims to expedite lymphoma diagnosis by identifying symptoms, tests and medications (clinical features) to guide GP referrals for suspected lymphoma. Incident lymphoma diagnosis will be measured for patients of all ages.
A case-control study using conditional logistic regression models will examine the association between clinical features and a lymphoma diagnosis compared to age and sex matched controls, deriving odds ratios for up to 5 years prior to date of lymphoma diagnosis. Diagnostic accuracy statistics including sensitivity, specificity and positive and negative likelihood ratios will be calculated for individual and combined clinical features.
Using a cohort design, we will derive and validate a prediction model including the clinical features identified in the case-control study. Cox proportional hazard models will be used to estimate the risk of a lymphoma diagnosis within 5 years.
Derivation and validation datasets will be created based on geographical region. Calibration and discrimination statistics will assess model performance in the validation dataset, these include R2 statistics and ROC curves.
Study 1:
Incident diagnosis of Lymphoma recorded in either CPRD, HES, or NCRAS, between 1st January 2001 and 1st January 2023
Study 2
Incident diagnosis of Lymphoma in recorded in either CPRD, HES or NCRAS, between 1st January 2001 to 1st January 2023
Both studies will record all-cause mortality and socioeconomic/deprivation status using ONS and IMD datasets on top of CPRD. HES will also be used to obtain hospitalisation data, frequency of admissions.
Clare Bankhead - Chief Investigator - University of Oxford
Tara Seedher - Corresponding Applicant - University of Oxford
Brian Nicholson - Collaborator - University of Oxford
Cynthia Wright Drakesmith - Collaborator - University of Oxford
Margaret Smith - Collaborator - University of Oxford
Micheal McKenna - Collaborator - University of Oxford
Robert Williams - Collaborator - University of Oxford
HES Accident and Emergency;HES Admitted Patient Care;HES Outpatient;NCRAS Cancer Registration Data;ONS Death Registration Data;Patient Level Index of Multiple Deprivation;CPRD Aurum Ethnicity Record;NCRAS Tumour / Treatment data