Background
Previous studies have showed the importance of blood tests ordered by General Practitioners (GPs) in helping to detect cancer earlier. One in four adults have a full blood test over a year and using these results to look for unusual signs could lead to earlier diagnosis and better outcomes.
This study will build on the researchers' existing work by researching a link between different types of cancers and the multiple blood counts, including “neutrophil to lymphocyte ratio”. Neutrophils and lymphocytes are different types of white blood cells; the body produces them when fighting infections.
Abnormal results can indicate inflammation in the body which could be caused by an undetected cancerous tumour. Previous studies have indicated that the “neutrophil to lymphocyte ratio” could be of particular importance when reviewing survival for patients already diagnosed with several cancer types, however it has never been considered as a tool to aid diagnosis of these cancers.
The aim of this study is io understand how abnormal blood test results predict multiple cancer types. Blood test results will be reviewed to understand what types of results are followed by the patient developing a cancer. The outcomes of this study will provide valuable information and advice to GPs, who may order routine blood tests which revel a high “neutrophil to lymphocyte ratio” and from here understand the likelihood of cancer.
Several blood tests including neutrophil to lymphocyte ratio (NLR) have been shown to be useful prognostic markers of several cancer types. We have experience of studying blood tests (including thrombocytosis and microcytosis) as diagnostic markers of cancer and integrating these findings into practice. This application will extend that work to NLR and other routine blood tests.
Aims
1. Explore the distribution of the neutrophil-to-lymphocyte ratio.
2. Quantify the association between the neutrophil-to-lymphocyte ratio and an incident cancer diagnosis.
3. Determine the neutrophil-to-lymphocyte ratio that confers an excess cancer risk of 3% or greater.
4. Quantify a hierarchy of cancer risk by site, over and above that expected by age and sex, in patients
with an abnormal neutrophil-to-lymphocyte ratio, to inform cancer investigative strategies in these
patients.
5. Explore what additional risk, if any, is conferred by the presence of cancer symptoms recorded in the 28
days before the blood test.
6. Quantify the association between neutrophil-to-lymphocyte ratio and cancer outcomes, such as stage
and 1-year survival.
7. Explore whether the neutrophil-to-lymphocyte ratio adds additional utility in cancer risk prediction over
known cancer biomarkers.
8. Repeat the above for other blood tests.
Methods
Retrospective cohort study. Patients with a blood test from 01/01/2012-31/12/2018 will be included. Exposure: value of the neutrophil-to-lymphocyte ratio. Outcomes: incident cancer diagnosis. Statistical analysis: descriptive analysis of cancer
incidence and neutrophil to lymphocyte counts in the cohort. Mixed-effects logistic regression, stratified by
age and sex, will test associations between the neutrophil-to-lymphocyte ratio and an incident cancer
diagnosis (or stage at diagnosis, depending on the aim being addressed), with random effect for general
practice. One-year net survival will be modelled using the non-parametric Pohar-Perme estimator.
The data sources requested are described in the relevant section below.
Outcomes: The main outcomes for the research aims above are:
Aim 1. The neutrophil-to-lymphocyte ratio, determined from the neutrophil and lymphocyte counts.
Aims 2 and 7. An incident cancer diagnosis recorded in NCRAS within 1 year of the index date. In this context, incident means first ever record of a cancer in NCRAS, excluding non-melanoma skin cancer. We will include all cancers except non-melanoma skin cancer (ICD10 C44), and testicular and cervical cancers (whose peak incidence occurs ≤40 years).
Aim 3. The probability of any cancer diagnosis given the neutrophil-to-lymphocyte ratio value, and change in that value over time, for all patients and for those with selected comorbidities.
Aim 4. The probability of an individual cancer diagnosis given the neutrophil-to-lymphocyte ratio value, and change in that value over time, for all patients and for those with selected comorbidities.
Aim 5. The probability of any cancer diagnosis given the neutrophil-to-lymphocyte ratio value and the presence or absence of cancer symptom(s) in the 28 days before the index date
Aim 6 (a). Cancer stage at diagnosis (dichotomised as advanced (stages 3 or 4) or early (stages 1 or 2)) Aim 6 (b). Death within 1 year of the index date, determined from ONS and CPRD variables.
Aim 7. See above (Aim 2 and 7)
Aim 8. As above, for other elements of a full blood count including red cell count, haemoglobin levels, white cell counts, platelets count, red cell distribution width, and mean corpuscular volume.
Update July 2024:
Cancer incidence in patients with an abnormal test result in this context is equivalent to the positive predictive value (PPV) of the test, which equates to risk.
We have updated the data linkage section. We had a misunderstanding about the source of deprivation data and have updated the entire eRAP to reflect that we will use ONS for data of death for survival analysis, and IMD for deprivation data.
Sarah Bailey - Chief Investigator - University of Exeter
Tanimola Martins - Corresponding Applicant - University of Exeter
David Shotter - Collaborator - University of Exeter
Elizabeth Shephard - Collaborator - University of Exeter
Luke Mounce - Collaborator - University of Exeter
Sarah Price - Collaborator - University of Exeter
HES Admitted Patient Care;NCRAS Cancer Registration Data;ONS Death Registration Data;Patient Level Index of Multiple Deprivation;CPRD Aurum Ethnicity Record;NCRAS Tumour / Treatment data