Search/filter projects:
- Professor Carol Joinson - Dr Christina Dardani -
Observational studies have found evidence that depression and anxiety are prospectively associated with lower urinary tract symptoms (LUTS) in women. Although prospective studies provide evidence of the direction of observed associations, they are limited by unmeasured and residual confounding. Observational studies that rely on self-report questionnaires to assess depression and anxiety exposures are also limited by measurement error. Polygenic risk scores (PRS) can be used estimate an individual’s underlying genetic liability to complex traits such as depression and anxiety. PRS should not be associated with genetic or environmental confounders at a population level, which can bias observational studies.
This project, based on data from the ALSPAC mothers’ cohort, will examine the relationship between depression, anxiety and LUTS using PRS for depression and anxiety derived from genetic variants identified in published genome wide association studies. The analysis will examine evidence for differential associations between the depression/anxiety PRS and LUTS including different subtypes of urinary incontinence (UI) (stress UI, urgency UI, mixed UI), urinary urgency, and nocturia.
Logistic regression analysis will be used to examine the associations between the PRS for depression/anxiety and binary variables for the LUTS phenotypes. Analyses will be adjusted for the first ten principal components of the ALSPAC genotype data.
Als TD et al. Nat Med. 2023 Jul;29(7):1832-1844.
Purves, K.L., et al. Molecular Psychiatry, 2020. 25(12): p. 3292-3303.
Felde G, Ebbesen MH, Hunskaar S. Neurourol Urodyn 2017 Feb;36(2):322-328.
- Nancy McBride (lead) - Dr Carolina Borges -
Our knowledge of the mechanisms that underpin gestational age (GA), and related pregnancy disorders, such as pre term birth (PTB – births < 37 weeks gestation), and very preterm birth (vPTB, births < 34 weeks gestation), is still poor (1).
PTB affects around 15 million births worldwide each year, and is the largest contributor to fetal mortality and morbidity (2). While we know there are clinical factors and other co-morbid disorders of pregnancy that increase risk of PTB, we still do not know what a woman’s gestational duration will be when she first attends clinic, and often until she signs symptoms of spontaneous pre term labour.
Metabolomics may help us elucidate more about these mechanisms. The NMR platform quantifies metabolic traits. The targeted metabolic traits measured by the platform represent a broad molecular signature of metabolism including routine lipids, lipoprotein subclass profiling, fatty acid composition and several low-molecular metabolites, including amino acids, ketone bodies and gluconeogenesis-related metabolites (3).
Metabolomics can improve our understanding of pregnancy-related disorders because metabolite levels are known to change markedly during pregnancy and may reflect underlying pathophysiology (4-6). Changes in metabolite profiles have been previously associated with other pregnancy-related disorders as well as adverse cardio-metabolic outcomes that associate with pregnancy related disorders (4, 7).
We have access to pregnancy and birth cohorts of mothers and offspring within which this can be explored, within the MR-PREG consortium. In a previous study of 7,440 pregnant participants of the Born in Bradford cohort, maternal dyslipidaemia in the second trimester was associated with a shorter GA at birth (8). These associations may be explained by residual confounding and reverse causality. Mendelian randomization (MR) is a causal inference technique which can help to elucidate whether these associations are causal (9). We hope to identify the metabolic changes underlying the physiological onset of labour to help develop new clinically useful strategies and identify potential drug targets (1).
The aim of this study is to explore the causal effect of maternal metabolites on the risk of PTB, it’s subtypes, and gestational age, using MR.
This study will be undertaken within the MR-PREG collaboration, which aims to explore causes and consequences of different pregnancy and perinatal outcomes. The PhD candidate will have access to the largest genome-wide association studies (GWAS) currently available for NMR-profiled metabolites (using data from ~300,000 UK Biobank individuals) and large meta-analyses conducted in European cohorts of GA/PTB (births <37 weeks, and subtypes such as very-pre-term birth, <34 weeks). We will use these to identify genetic instruments for metabolites and PTB / gestational age, and use two-sample Mendelian randomisation to probe the causal role of metabolites on gestational age and PTB.
1. Following Spontaneous Labour at Term in Humans Using Untargeted Metabolomics Analysis: A Pilot Study. LID - 10.3390/ijerph16091527 [doi] LID - 1527. (1660-4601 (Electronic)).
2. Stock SJ, Horne M, Bruijn M, White H, Boyd KA, Heggie R, et al. Development and validation of a risk prediction model of preterm birth for women with preterm labour symptoms (the QUIDS study): A prospective cohort study and individual participant data meta-analysis. PLOS Medicine. 2021;18(7):e1003686.
3. Dunn WB, Bailey NJC, Johnson HE. Measuring the metabolome: current analytical technologies. Analyst. 2005;130(5):606-25.
4. McBride N, White SL, Farrar D, Poston L, Sattar N, Nelson SM, et al. Do nuclear magnetic resonance (NMR)-based metabolomics improve the prediction of pregnancy-related disorders? medRxiv. 2020:2020.06.22.20134650.
5. Souza RT, Galvão RB, Leite DA-O, Passini R, Jr., Baker P, Cecatti JA-O. Use of metabolomics for predicting spontaneous preterm birth in asymptomatic pregnant women: protocol for a systematic review and meta-analysis. (2044-6055 (Electronic)).
6. Considine EC, Khashan AS, Kenny LC. Screening for Preterm Birth: Potential for a Metabolomics Biomarker Panel. Metabolites. 2019;9(5):90.
7. White SL, Pasupathy D, Sattar N, Nelson SM, Lawlor DA, Briley AL, et al. Metabolic profiling of gestational diabetes in obese women during pregnancy. (1432-0428 (Electronic)).
8. Birchenall KA. Investigating the trigger for human parturition using metabolomic and phosphoproteomic techniques within case-control and cohort studies: University of Bristol; 2020.
9. Skrivankova VW, Richmond RC, Woolf BAR, Davies NM, Swanson SA, VanderWeele TJ, et al. Strengthening the reporting of observational studies in epidemiology using mendelian randomisation (STROBE-MR): explanation and elaboration. BMJ. 2021;375:n2233.
- Dr Anna Hurley-Wallace (lead) - Dr Katie Whale -
Chronic musculoskeletal pain is a common issue experienced by adolescents and young adults (AYAs) [1,2,3], which can have a significant impact on their wellbeing [4,5]. Conditions include rheumatoid arthritis, hypermobility disorders, and Ehlers-Danlos Syndromes. A key challenge for AYAs is accessing, and subsequently engaging with appropriate multidisciplinary healthcare services for chronic musculoskeletal pain [6]. Musculoskeletal pain management programmes usually focus on physical or occupational therapy [7,8], with adjuvant psychological therapy and medications as appropriate. Barriers preventing AYAs from accessing and engaging with musculoskeletal pain management programmes remain unclear. By understanding these barriers, interventions and service improvements can be implemented to improve programme engagement.
Primary aim: Explore and summarise young people’s (16 to 24-years) experiences of accessing and engaging with musculoskeletal chronic pain programmes offered in secondary care. Objectives: (i) identify barriers to accessing musculoskeletal pain programmes, (ii) identify barriers and facilitators to successful engagement with musculoskeletal pain programmes, (iii) explore contextual factors that may be contributing to AYA levels of engagement with musculoskeletal pain programmes, (iv) draft a set of key recommendations for a co-intervention to improve AYA’s engagement with musculoskeletal pain programmes, focusing on engagement with the physical therapy component.
A secondary analysis of qualitative interview data, which was originally collected to capture young people’s experiences of online resources for chronic pain management[9]. The aim for this analysis is to explore and summarise experiences of accessing and engaging with pain management programmes within the musculoskeletal pain participant sub-set (n = 8), with potential expansion to the post-injury pain sub-set (n = 3). Thematic analysis will be conducted using NVivo [10]. Results should present the themes and suggest key recommendations for a co-intervention to improve AYA’s engagement with musculoskeletal pain programmes. Original transcripts and field notes are available.
1. King, S., Chambers, C. T., Huguet, A., MacNevin, R. C., McGrath, P. J., Parker, L., & MacDonald, A. J. (2011). The epidemiology of chronic pain in children and adolescents revisited: a systematic review. Pain, 152(12), 2729-2738. https://doi.org/10.1016/j.pain.2011.07.016
2. Kastelein, M., Luijsterburg, P. A. J., Heintjes, E. M., van Middelkoop, M., Verhaar, J. A. N., Koes, B. W., & Bierma-Zeinstra, S. M. A. (2015). The 6-year trajectory of non-traumatic knee symptoms (including patellofemoral pain) in adolescents and young adults in general practice: a study of clinical predictors. British Journal of Sports Medicine, 49(6), 400-405. http://dx.doi.org/10.1136/bjsports-2014-093557
3. Hanvold, T. N., Veiersted, K. B., & Wærsted, M. (2010). A prospective study of neck, shoulder, and upper back pain among technical school students entering working life. Journal of Adolescent Health, 46(5), 488-494. https://doi.org/10.1016/j.jadohealth.2009.11.200
4. Myrtveit, S. M., Sivertsen, B., Skogen, J. C., Frostholm, L., Stormark, K. M., & Hysing, M. (2014). Adolescent neck and shoulder pain—the association with depression, physical activity, screen-based activities, and use of health care services. Journal of Adolescent Health, 55(3), 366-372. https://doi.org/10.1016/j.jadohealth.2014.02.016
5. Rathleff, M. S., Holden, S., Straszek, C. L., Olesen, J. L., Jensen, M. B., & Roos, E. M. (2019). Five-year prognosis and impact of adolescent knee pain: a prospective population-based cohort study of 504 adolescents in Denmark. BMJ open, 9(5), e024113. http://dx.doi.org/10.1136/bmjopen-2018-024113
6. Slater, H., Jordan, J. E., Chua, J., Schütze, R., Wark, J. D., & Briggs, A. M. (2016). Young people's experiences of persistent musculoskeletal pain, needs, gaps and perceptions about the role of digital technologies to support their co-care: a qualitative study. BMJ open, 6(12), e014007. https://doi.org/10.1136/bmjopen-2016-014007
7. Caes, L., Fisher, E., Clinch, J., & Eccleston, C. (2018). Current evidence-based interdisciplinary treatment options for pediatric musculoskeletal pain. Current Treatment Options in Rheumatology, 4, 223-234. https://doi.org/10.1007/s40674-018-0101-7
8. Van Meulenbroek, T., Conijn, A. E. A., Huijnen, I. P. J., Engelbert, R. H. H., & Verbunt, J. A. (2020). Multidisciplinary Treatment for Hypermobile Adolescents with Chronic Musculoskeletal Pain. Journal of rehabilitation medicine. Clinical communications, 3, 1000033. https://doi.org/10.2340/20030711-1000033
9. Hurley-Wallace, A., Kirby, S., & Bishop, F. (2022). Trusting in the online 'community': An interview study exploring internet use in young people with chronic pain. British journal of pain, 16(3), 341–353. https://doi.org/10.1177/20494637211061970
10. Clarke, V., Braun, V., & Hayfield, N. (2015). Thematic analysis. Qualitative psychology: A practical guide to research methods, 3, 222-248.
- Dr Neil Goulding - Prof Abigail Fraser -
The aetiology of the hypertensive disorders of pregnancy (HDP; preeclampsia and gestational hypertension) is likely complex, involving both maternal vascular and inflammatory components. We hypothesize that women who experience a HDP will have a more adverse inflammatory profile later in life.
To investigate the associations of HDP with 92 individual inflammatory proteins measured in women at mean (standard deviation; SD) age 47.5 (4.4) in a population based prospective pregnancy cohort study (ALSPAC).
1. Conduct a review of the literature to gain insight into the pathogenesis of HDP.
2. Identify potential confounders of the relationship of interest.
3. Conduct multivariable regression analyses to estimate the relationships between HDP and the 92 proteins measured using the Olink Target 96 Inflammation Panel (Olink, Uppsala, Sweden).
4. Write up a report and if the student wishes to, prepare a manuscript for publication.
LA Magee et al. N Engl J Med 2022;386:1817-1832.
Fraser et. al. Int J Epidemiol 2013;97-110.
- Dr Mark Gormley - Dr Rebecca Richmond -
Ankyloglossia (‘tongue-tie’) is a common condition characterised by a short lingual frenulum. It has been associated with feeding difficulties, predominantly among breastfed infants, which can lead to inadequate nutrition and failure to thrive.
Tongue-tie is being more frequently diagnosed, particularly in high income countries. This may be due to increased breastfeeding or awareness of the condition. Another hypothesis is that rates are increasing due to folic acid (FA) supplementation during pregnancy, but the evidence remains insufficient.1
FA is protective against the development of neural tube defects and is recommended during the periconceptional period. It would not be advisable for expectant mothers to stop taking FA since the benefits outweigh the potential harms. Nonetheless, if tongue-tie was found to be influenced by maternal FA intake, this would raise further awareness of the condition and potential for early treatment through frenectomy, a simple and low risk procedure.
This project will investigate the relationship between maternal folate status and offspring tongue tie using data from mothers and children in the Avon Longitudinal Study of Parents and Children (ALSPAC).
Exposures will be maternal FA supplementation at 17 and 32 weeks of pregnancy, folate intake in the diet at 32 weeks of pregnancy and the methyltetrahydrofolate reductase (MTHFR) genotype.2
Outcomes will include reported tongue tie, infant feeding difficulties, breastfeeding cessation, and weight gain in infancy.
The analysis will involve multivariable linear and logistic regression analysis with adjustment for potential confounding factors (including maternal smoking, alcohol, body mass index, social class, education, dietary intake).
MTHFR genotype will be used to assess the causal role of maternal folate status in relation to tongue tie and related outcomes within a Mendelian randomization framework.2
Analyses will be further stratified by mode of infant feeding (breastfeeding exclusivity and duration).
1. Rubin, G., Stewart, C., McGowan, L., Woodside, J., Barrett, G., Godfrey, K., & Hall, J. (2023). Maternal folic acid supplementation and the risk of ankyloglossia (tongue-tie) in infants; a systematic review. Authorea, doi. 10.22541/au.168007870.09307215/v1.
2. Lewis, S. J., Leary, S., Smith, G. D., & Ness, A. (2009). Body composition at age 9 years, maternal folate intake during pregnancy and methyltetrahydrofolate reductase (MTHFR) C677T genotype. British Journal of Nutrition, 102(4), 493-496.
- Dr Tom Bond - Dr Apostolos Gkatzionis -
Studies of pregnancy outcomes are often conducted in samples that are restricted to women who are currently pregnant. This involves conditioning on pregnancy “incidence”, analogous to conditioning on disease incidence in studies of disease progression. In this situation pregnancy becomes a collider variable, which could induce spurious associations between causes of pregnancy that are truly independent in the source population. Furthermore, if any of the causes of pregnancy also cause the pregnancy outcome of interest then this could result in biased estimates of the true causal associations between risk factors and the pregnancy outcome. This is an example of index event bias (1), and the extent to which this could affect Mendelian randomization (MR) studies of pregnancy outcomes is currently unknown. This mini project will inform and influence a large number of MR investigations that we are currently conducting within the MR-PREG consortium.
1. Explore, via simulations and empirical analyses, the extent to which index event bias may affect MR studies of pregnancy outcomes.
2. Establish, via simulations and empirical analyses, whether existing methods can mitigate index event bias in MR studies of pregnancy outcomes.
We will conduct simulation studies and (where possible) real data analyses, to explore the extent to which two sample MR estimates of causal effects on pregnancy outcomes may be affected by index event bias under realistic assumptions. There will also be scope to conduct further simulation studies, along with applied example analyses, to test whether existing methods can mitigate index event bias in a two sample MR study of pregnancy outcomes. Relevant methods will include Slope-Hunter (2) and the method of Dudbridge et al. (3), which requires a GWAS of pregnancy incidence.
1. Mitchell RE, et al. PLOS Genetics. 2023;19(2):e1010596.
2. Mahmoud O, et al. Nature Communications. 2022;13(1):619.
3. Dudbridge F, et al. Nature Communications. 2019;10(1):1561.
- Dr Eleanor Sanderson - Dr Gareth Griffith -
Mendelian randomization (MR) uses genetic variants to estimate the causal effect of an exposure on an outcome of interest in a way that aims to overcome bias from unobserved confounding. The genetic variants used in MR are often selected as those most strongly associated with the exposure from a genome-wide association study (GWAS). These GWAS are conducted on datasets such as UK Biobank where many of the individuals are on medication such as blood pressure (BP) lowering medication. This medication use may bias effect estimates from GWAS studies or induce associations between SNPs for other characteristics and blood pressure, in turn biasing any MR studies which include BP as an exposure or an outcome.
It is not clear how to adjust for such bias in MR studies; excluding all individuals who report taking BP medication could induce selection bias as individuals taking medication will have had higher BP prior to medication use than those who do not. However, ignoring the potential issue is also likely to bias MR effect estimates. Alternative strategies include applying a BP adjustment to those reporting medication use to account for the effect of medication or restricting the sample to only younger individuals who are less likely to be taking medication.
The aim of this project is to explore how the different potential adjustments for medication use that could be applied to GWAS studies affect the results obtained from summary data MR analyses.
In this project the student will run GWAS on UK Biobank using the following adjustments to account for reported blood pressure medication
o Increasing the measured blood pressure by a set amount for those reporting medication use.
o Excluding those reporting medication use
o Restricting the sample to those under 50 where medication use is lower.
Each of these set of GWAS results will then be used in the same summary data MR analyses to compare the results obtained from the different adjustments. Separate MR analyses will be conducted with blood pressure as the exposure and the outcome. The results for each analysis will be compared to each other and to results with no adjustment for medication use to understand how these different adjustments bias the results from MR studies.
- Natalia Lewis - Dr Elizabeth Cook -
Intimate partner violence has a negative impact on social and economic outcomes. However, it has not been established how long those social and economic consequences last and how the duration varies with the type of violence. Duration of such effects is needed to cost the social and economic harms from violence.
To quantify the average length of time for which social and economic outcomes are affected by intimate partner violence, by type of violence (sexual, physical, emotional, financial).
A focused systematic review of studies reported in peer reviewed literature that recruited adults, had multiple time points, a social and/or economic outcome(s) and where intimate partner violence was a predictor, independent variable, or inclusion criterion.
Support will be provided by supervisors and subject librarian.
Training on quantitative evidence synthesis is available via short course and guided self-learning.
Patton SC, Szabo YZ, Newton TL. Mental and Physical Health Changes Following an Abusive Intimate Relationship: A Systematic Review of Longitudinal Studies. Trauma Violence Abuse. 2022 Oct;23(4):1079-1092. doi: 10.1177/1524838020985554. Epub 2021 Jan 20. PMID: 33468040.
Page, M.J., McKenzie, J.E., Bossuyt, P.M. et al. The PRISMA 2020 statement: an updated guideline for reporting systematic reviews. Syst Rev 10, 89 (2021). https://doi.org/10.1186/s13643-021-01626-4
- Dr Natalia Lewis - Professor Gene Feder -
Theory of change (programme theory) describes how an intervention can produce long-term outcomes through a logical sequence of intermediate outcomes. Theories of change have been widely used to design and evaluate health interventions. The development of a theory of change typically occurs through a co-production process, requiring stakeholders to reflect on how their programmes can bring change. There is comparatively little literature on applying a theory of change approach to multidisciplinary programmes of research addressing complex public health and clinical problems.
The Violence, Health, and Society (VISION) consortium is a 5-year UKPRP-funded multi-disciplinary collaboration which aims to improve the measurement of data on violence to influence policy and practice and reduce violence and the health inequalities that result. VISION worked with partner organisations and other professional stakeholders to co-produce theories of change for research strands within the consortium.
This study aims to evaluate the process of developing theories of change for research strands within the VISION consortium. The research questions are: 1) How were the theories of change co-produced? 2) What factors were important in facilitating or hindering the co-production?
This is a qualitative process evaluation with a case study comprising of participant observations of stakeholders’ workshops, analysis of physical artefacts and documents, semi-structured interviews with workshops attendees. Ethics approval for the study has been obtained. We will provide observation fieldnotes, artefacts, and documents. The student will conduct up to 20 semi-structured qualitative interviews with VISION researchers and stakeholders who took part in the development of theories of change. The interview topic guide will explore participants’ perspective on the stakeholders’ workshops on co-producing theories of change including any things which drew them to it initially, how they experienced it, what factors facilitated and hindered the process of theories development, and whether there are things that could have been done differently. Interviews will be audio recorded and professionally transcribed. The student will use the principles of co-production and framework approach for organising primary data into codes, candidate themes, and analytical themes.
1. Breuer et al 2016.Using theory of change to design and evaluate public health interventions: a systematic review. doi: 10.1186/s13012-016-0422-6.
2. Boblin et al 2013. Using Stake’s Qualitative Case Study Approach to Explore Implementation of Evidence-Based Practice. doi:10.1177/1049732313502128
- Professor Carol Joinson - Dr Kimberley Burrows -
There is robust evidence that adverse childhood experiences (ACEs) have enduring impacts on a range of negative health outcomes in adulthood. A growing number of studies have reported a link between ACEs and adult urinary symptoms including incontinence, overactive bladder, and nocturia. Limitations of these studies include a retrospective design, the use of small clinic samples without control groups, and a focus on specific types of ACEs (e.g. sexual abuse), rather than the total burden of ACEs.
This project, based on data from the ALSPAC mothers cohort, will examine prospective relationships between ACEs (e.g. maladaptive family functioning, parental mental illness, sexual abuse) and urinary symptoms in middle-aged women. The study will use self-reported data on ACEs in questionnaires administered during pregnancy and 4 years after recruitment to ALSPAC. Urinary symptoms were assessed using validated self-report questionnaires at 10 and 19 years after recruitment.
Multivariable logistic regression analysis will be used to examine the associations between ACEs and urinary symptoms, adjusted for confounders.
Epperson CN, Duffy KA, Johnson RL, Sammel MD, Newman DK. Enduring impact of childhood adversity on lower urinary tract symptoms in adult women. Neurourol Urodyn 2020; 39: 1472–81.
Brady SS, Arguedas A, Huling JD, et al. Adverse Childhood Experiences and Lower Urinary Tract Symptoms and Impact Among Women. Journal of Urology 2023; 209: 1167–75.
Selai C, Elmalem MS, Chartier-Kastler E, et al. Systematic review exploring the relationship between sexual abuse and lower urinary tract symptoms. Int Urogynecol J 2023; 34: 635–53.
- Rebecca Richmond - Mr Benji Woolf -
In previous work, we found that a large proportion of the Mendelian randomization literature to be poorly reported (1). The STROBE-MR checklist was subsequently introduced as a guideline to improve reporting of studies using Mendelian randomization (2). The effect that issuing the STROBE-MR guidelines has had on reporting quality in MR studies requires evaluation.
To compare reporting quality of Mendelian randomization studies published before and after issuing of the STROBE-MR guidelines to see whether this has improved reporting of MR studies.
There may also be scope for considering if there are useful changes which should be made to the STROBE-MR checklist – e.g. simplifications or additional items.
Study selection and evaluation will use similar systematic review methods as those in the previous study (1). Rather than evaluating all published studies, we expect that this project will use random sampling to select a more manageable number of studies. The actual evaluation will use a difference-in-difference approach, which is an econometric method for evaluating the effect of an intervention (here issuing of the STROBE-MR guidelines) after selecting suitable control group (e.g. published randomised controlled trails).
transparency of reporting in two-sample summary data Mendelian randomization studies using the MR-Base platform. Int J Epidemiol. 2022 Apr 6;dyac074.
2. Skrivankova VW, Richmond RC, Woolf BAR, Davies NM, Swanson SA, VanderWeele TJ, et al. Strengthening the reporting of observational studies in epidemiology using mendelian randomisation (STROBE-MR): explanation and elaboration. BMJ. 2021 Oct 26;375:n2233.
- Dr Eleanor Sanderson - Dr Eleanor Sanderson -
Chronotype (evening preference) has previously been shown to have a causal effect on breast cancer through univariable Mendelian randomization (MR), although the mechanisms underlying this effect are unclear. Chronotype is correlated with alcohol consumption and may have a shared genetic aetiology or there may be a causal effect between these traits. MR uses genetic variants to estimate the causal effect of an exposure on an outcome of interest in a way that aims to overcome bias from unobserved confounding. However, MR is biased if the genetic variants are associated with the outcome through a pathway that does not include the exposure, such as if genetic variants associated with chronotype are also associated with alcohol consumption. Multivariable MR is an extension of MR that can include multiple exposures and so can account for such pathways. Multivariable MR can also be used to establish the extent to which alcohol consumption mediates or confounders the effect of chronotype on breast cancer.
The aim of this project is to estimate to what extent the effect of chronotype on breast cancer can be explained by alcohol consumption.
This project will use individual level data from UK Biobank and apply multivariable MR method to estimate the extent to which the effect of chronotype on breast cancer is explained by alcohol consumption. Multivariable MR can also be biased by pathways from the genetic instruments to the outcome that do not act through the exposure and therefore this project will apply a novel multivariable MR that estimates which genetic variants are likely to have pleiotropic effects and obtains effect estimates that are robust to that pleiotropy.
Richmond, Rebecca C., et al. "Investigating causal relations between sleep traits and risk of breast cancer in women: Mendelian randomisation study." bmj 365 (2019).
Sanderson, Eleanor, et al. "An examination of multivariable Mendelian randomization in the single-sample and two-sample summary data settings." International journal of epidemiology 48.3 (2019): 713-727.
- Dr Rebecca Richmond - Dr Eleanor Sanderson -
Insomnia has previously been shown to have a causal effect on glycaemic traits and diabetes through univariable Mendelian randomization (MR), although the mechanisms underlying this effect are unclear. Insomnia is correlated with depression and may have a shared genetic aetiology or there may be a causal effect between these traits. MR uses genetic variants to estimate the causal effect of an exposure on an outcome of interest in a way that aims to overcome bias from unobserved confounding. However, MR is biased if the genetic variants are associated with the outcome through a pathway that does not include the exposure, such as if genetic variants associated with insomnia are also associated with depression. Multivariable MR is an extension of MR that can include multiple exposures and so can account for such pathways. Multivariable MR can also be used to establish the extent to which depression mediates or confounders the effect of insomnia on glycaemic traits and type 2 diabetes.
The aim of this project is to estimate to what extent the effect of insomnia on glycaemic traits and type 2 diabetes can be explained by depression.
This project will use individual level data from UK Biobank and apply multivariable MR method to estimate the extent to which the effect of insomnia on glycaemic traits and type 2 diabetes is explained by depression. Multivariable MR can also be biased by pathways from the genetic instruments to the outcome that do not act through the exposure and therefore this project will apply a novel multivariable MR that estimates which genetic variants are likely to have pleiotropic effects and obtains effect estimates that are robust to that pleiotropy.
Liu, James, et al. “Assessing the causal role of sleep traits on glycated hemoglobin: a Mendelian Randomization study”. Diabetes Care 45.4 (2022): 772-781.
Sanderson, Eleanor, et al. "An examination of multivariable Mendelian randomization in the single-sample and two-sample summary data settings." International journal of epidemiology 48.3 (2019): 713-727.
- Dr. Chin Yang Shapland - Dr. Christina Dardani -
There is increasing evidence suggesting that there is a bi-directional causal relationship between alcohol use disorder (AUD) and major depressive disorder (MDD) (Treur et al. 2021) However, most research utilising genetically informed approaches has been based on samples of European ancestry.
Little is known on the relationships between AUD and MDD in non-European ancestry samples and elucidating them can lead towards a better understanding of the mechanisms underlying AUD and MDD risk within and across ancestries. For example, one possibility is that differences in the relationships between AUD and MDD might be attributed to differences in vulnerability rather than susceptibility across populations. This means that some groups might be exposed to different attitudes and behaviours towards alcohol drinking and mental health, rather than being genetically predisposed to AUD or MDD.
Mendelian randomization (MR) is a causal inference technique that avoids unmeasured confounding by using genetic instruments (Bowden et al. 2018). Multi-ancestry MR is complex, as each ancestry could potentially have different GWAS hits, for example, due to differences in allele frequency and linkage disequilibrium.
The aim of this study is to understand the relationship between AUD and MDD in non-European ancestry individuals. Specifically, we will focus on two research questions:
1. Is there a bi-directional causal link between AUD and MDD?
2. Can we disentangle susceptibility and vulnerability?
The PhD candidate will have access to GWAS summary statistics for alcohol intake and depression from European and non-European ancestry samples. We will use these to identify genetic instruments, assess the transferability of these instruments (i.e. are the genetic instrument from each ancestry in linkage disequilibrium?), perform bi-directional two-sample Mendelian randomisation and Linkage Disequilibrium Score Regression (LDSC) for each ancestry. Finally, perform meta-analyses to see whether each ancestry gives the same casual effect. Once the main analysis is complete, with continued support from the multi- disciplinary supervisory team, PhD candidate will have the opportunity to explore other analyses aiming to address Aim (2).
By the end of the project, the student will have gained knowledge and skills in using GWAS summary data, and applying MR, genetic correlation, and meta-analysis methods. Analyses will be conducted in R, and high-performance computing environments. The finished work could feasibly be submitted to a relevant journal, such as Addiction .
1 Treur, J. L., Munafò, M. R., Logtenberg, E., Wiers, R. W. & Verweij, K. J. Using Mendelian randomization analysis to better understand the relationship between mental health and substance use: a systematic review. Psychological medicine 51, 1593-1624 (2021).
2 Bowden, J. et al. Improving the accuracy of two-sample summary data Mendelian randomization: moving beyond the NOME assumption. 159442 (2018).
- Prof Evie Stergiakouli - Prof Sarah Lewis -
Cleft of the lip and/or palate (CL/P) is one of the most common congenital anomalies requiring corrective surgery within the first year of life. Children born with CL/P have 3.2 admissions and spend 13.2 days in hospital in the first two years of life. Despite treatment to repair the cleft in infancy, being born with a cleft frequently results in multiple adverse outcomes across the lifespan, including facial disfigurement, impaired speech and low intelligibility, with potentially poor educational, vocational, social, mental and physical health outcomes. We have previously shown, using data from a large cohort of children born with cleft, the Cleft Collective, that children born with CL/P have higher levels of behavioural problems than children in the general population at 5 and 10 years (1). However, we do not know whether there is genetic correlation between cleft and psychiatric disorders or whether factors associated to the cleft phenotype cause psychiatric disorders in children born with cleft. We have previously used Mendelian randomization to show that being born with cleft does not cause children to underperform at school (2).
AIMS & OBJECTIVES: In this project, we will use data from a large meta-analysis GWAS of cleft (including Cleft Collective data) to explore the links between cleft and psychiatric disorders (ADHD, autism spectrum disorder (ASD), anxiety, depression, schizophrenia).
In this project, we will use data from a large meta-analysis GWAS of cleft (including Cleft Collective data) to explore the links between cleft and psychiatric disorders (ADHD, autism spectrum disorder (ASD), anxiety, depression, schizophrenia).
A. We will use two-sample Mendelian randomization (3) to investigate any causal links between genetic liability to cleft of the lip and/or palate, and ADHD, ASD, anxiety, depression, schizophrenia bidirectionally
B. Apply sensitivity analyses including weighted median, weighted mode, MR-Egger regression, MR-PRESSO and colocalization analyses to assess and adjust for pleiotropy
C. We will perform Linkage Disequilibrium (LD)-score regression (4) to estimate any genetic correlation between cleft of the lip and/or palate, and ADHD, ASD, anxiety, depression, schizophrenia
1. Berman et al. Prevalence and Factors Associated with Behavioral Problems in 5-Year-Old Children Born with Cleft Lip and/or Palate from the Cleft Collective. Cleft Palate Craniofac J. 2024 Jan;61(1):40-51. doi: 10.1177/10556656221119684.
2. Dardani et al. Is genetic liability to ADHD and ASD causally linked to educational attainment?, Int J Epidemiol. 2022 Jan 6;50(6):2011-2023. doi: 10.1093/ije/dyab107.
3. Davey Smith, G. & Hemani, G. Mendelian randomization: genetic anchors for causal inference in epidemiological studies. Hum. Mol. Genet. 23, R89–R98 (2014)
4. Bulik-Sullivan et al. LD Score regression distinguishes confounding from polygenicity in genome-wide association studies. Nat Genet 2015;47:291–95.
- Dr Clare French - Professor Julian Higgins -
Vaccines are an extremely powerful and highly cost-effective public health intervention. However, there have been concerning declines in the uptake of vaccinations in the UK in recent years (1,2). Various interventions have been used to encourage people to get vaccinated, including reminders, education, media campaigns, and improving access.
The project supervisors are conducting a broad evidence review on the effectiveness of interventions to increase vaccine uptake among the general population (3). The literature searches run for this review captured studies conducted among all population groups, but the main analysis excludes studies on individuals belonging to clinical risk groups. However, this is of course an important group in which to optimise uptake of vaccinations.
Review the effectiveness of interventions to increase vaccine uptake among specific clinical risk groups – for example, people with conditions such as chronic obstructive pulmonary disease, diabetes, and cancer. There is potential for the choice of clinical risk group(s) to be tailored to the student’s interests.
The student will undertake a systematic review, guided by the Cochrane Handbook for Systematic Reviews of Interventions (4). The literature searches for this review have already been conducted, and the bulk of title/abstract and full text screening has been done. In consultation with the supervisors, the student will determine which clinical risk group(s) to focus their review on. They will then identify relevant studies from our existing database of eligible randomised controlled trials. The student will extract data from these trials, code/classify the intervention used, assess risk of bias using the Cochrane risk of bias 2 (RoB2) tool (5), and synthesize the data - likely using meta-analysis.
The student will have the opportunity to write an abstract for submission to a suitable conference, and to prepare a manuscript for publication in a peer-reviewed journal.
1. Bedford H, Skirrow H. Action to maximise childhood vaccination is urgently needed. BMJ. 2023 Oct 24;383:2426.
2. UK Health Security Agency. Prenatal pertussis vaccination coverage in England from January to March 2024 and annual coverage for 2023 to 2024. https://www.gov.uk/government/publications/pertussis-immunisation-in-pregnancy-vaccine-coverage-estimates-in-england-october-2013-to-march-2014/prenatal-pertussis-vaccination-coverage-in-england-from-january-to-march-2024-and-annual-coverage-for-2023-to-2024
3. French CE, Caldwell DC, McGrath C et al. Interventions to increase vaccine uptake: Systematic review and component network meta-analysis to identify the most effective strategies. International prospective register of systematic reviews. 2023. https://www.crd.york.ac.uk/prospero/display_record.php?RecordID=369139.
4. Higgins JPT, Thomas J, Chandler J, Cumpston M, Li T, Page MJ, Welch VA (editors). Cochrane Handbook for Systematic Reviews of Interventions version 6.3 (updated February 2022). Cochrane, 2022. Available from: https://www.training.cochrane.org/handbook
5. Sterne JAC, Savovic J, Page MJ, Elbers RG, Blencowe NS, Boutron I, et al. RoB 2: a revised tool for assessing risk of bias in randomised trials. BMJ. 2019;366:l4898.
- Dr Rebecca Richmond - Dr Bethan Lloyd-Lewis -
Breast mammographic density (MD), the proportion of fibroglandular tissue as detected by mammography or magnetic resonance imaging (MRI), is strongly associated with breast cancer risk [1]. Our recent work indicated that early-life adiposity decreases breast cancer risk largely through reducing breast MD in adulthood (age 39-80 years) [2]. However, whether the effect of early-life exposures such as adiposity influence MD at an earlier age is not established. Further, the mechanisms through which increased breast density leads to increased cancer risk are not well understood. Addressing this gap in knowledge is critically important as progress in this area has the potential to transform precision cancer prevention approaches in women with elevated breast MD.
This project aims to explore links between early life exposures, breast MD in early adulthood and breast cancer risk. The student will map the contribution of common genetic variation to breast density as detected by MRI in a subset of young (age 20-22 years) nulliparous women available from the Avon Longitudinal Study of Parents and Children (ALSPAC) birth cohort [3]. This unique dataset includes detailed metabolic, hormone and body fat mass measurements taken at multiple timepoints, which can be used to identify novel traits associated with breast MD to influence later life breast cancer risk.
1) Conduct a review of the literature on breast MD and breast cancer risk
2) Run a GWAS of breast MD in young nulliparous women using available MRI data from ALSPAC
3) Conduct longitudinal analysis to assess associations between metabolic, hormone and body fat mass measurements at repeated timepoints and MD at age 20-22 years.
4) Use findings from the MD GWAS and others to evaluate causal effects of metabolic, hormone and body fat mass measurements on MD using Mendelian randomization,
5) Use findings from the MD GWAS and a large meta-analysis for breast cancer (BCAC) in a Mendelian randomization framework
6) Integrate findings from 4) and 5) to evaluate the extent to which MD mediates the relationship between metabolic, body fat and hormone traits in early life on later-life breast cancer risk, using MR approaches for mediation.
[1] Sherratt, M.J., McConnell, J.C. & Streuli, C.H. Raised mammographic density: causative mechanisms and biological consequences. Breast Cancer Res 18, 45 (2016).
[2] Vabistsevits, M., Davey Smith, G., Richardson, T.G. et al. Mammographic density mediates the protective effect of early-life body size on breast cancer risk. Nat Commun 15, 4021 (2024).
[3] Denholm, R., et al. Growth Trajectories, Breast Size, and Breast-Tissue Composition in a British Prebirth Cohort of Young Women. American Journal of Epidemiology, 187 (2018)
- Prof Emma Clark - Dr Jess Harris -
Vfrac is a research project developed in Bristol (https://www.bristol.ac.uk/translational-health-sciences/research/musculoskeletal/rheumatology/research/vfrac-study.html) that has produced a clinical checklist for healthcare professionals to use during a consultation with an older person with back pain. It provides an evidence-based approach to understand if a spinal radiograph is required.
As part of the development work for Vfrac we need to understand what usually happens to older people consulting in primary care with back pain. This information will provide pilot data for a future project, as well as clarifying the definition of ‘standard care’ for the control arm of a future randomised controlled trial (RCT). To facilitate this we have a CPRD extract with data covering consultations for back pain and subsequent referrals, investigations and prescriptions.
The aims of this mini project are
1. to gain further experience in data handling, cleaning and refinement using a CPRD extract
2. to describe the usual healthcare pathway for older people consulting with back pain in primary care, focusing on referral for a spinal radiograph (X-ray) after an index consultation for back pain in adults aged 50+
3. to identify important subgroups where differences in the healthcare pathway exist
4. to gain further experience in dissemination through production of abstracts for conference submissions and first drafts of papers
Aim 1: exploration of the CPRD data using simple descriptives, with further refinement of variables according to pre-determined hypotheses
Aim 2: utilising an index consultation for back pain, a pathway diagram will be populated, with proportions of people moving down each arm/sub-arm identified
Aim 3: statistical analyses will be undertaken to identify variance from the pathway according to practice level descriptives (deprivation, size) and patient level characteristics (age, gender, comorbidities)
Yu D et al (2022) Trends in the annual consultation incidence and prevalence of low back pain and osteoarthritis in England from 2000 to 2019: Comparative estimates from two clinical practice databases. Clinical Epidemiology 14:179-189
Khalid T et al (2023) An online clinical decision tool to screen for vertebral fragility fractures (Vfrac) in older women presenting with back pain in general practice: protocol for a feasibility study in preparation for a future cluster randomized controlled trial. Archives of Osteoporosis OSIN-D-23-00756R1
NICE Guideline 59: Low back pain and sciatica in over 16s: assessment and management
- Prof Evie Stergiakouli - Prof Sarah Lewis -
Cleft of the lip and/or palate (CL/P) is one of the most common congenital anomalies requiring corrective surgery within the first year of life. Children born with CL/P have 3.2 admissions and spend 13.2 days in hospital in the first two years of life. Despite treatment to repair the cleft in infancy, being born with a cleft frequently results in multiple adverse outcomes across the lifespan, including facial disfigurement, impaired speech and low intelligibility, with potentially poor educational, vocational, social, mental and physical health outcomes. We have previously shown, using data from a large cohort of children born with cleft, the Cleft Collective, that children born with CL/P have higher levels of behavioural problems than children in the general population at 5 and 10 years (1). However, we do not know whether there is genetic correlation between cleft and psychiatric disorders or whether factors associated to the cleft phenotype cause psychiatric disorders in children born with cleft. We have previously used Mendelian randomization to show that being born with cleft does not cause children to underperform at school (2).
In this project, we will use data from a large meta-analysis GWAS of cleft (including Cleft Collective data) to explore the links between cleft and psychiatric disorders (ADHD, autism spectrum disorder (ASD), anxiety, depression, schizophrenia).
We will use two-sample Mendelian randomization (3) to investigate any causal links between genetic liability to cleft of the lip and/or palate, and ADHD, ASD, anxiety, depression, schizophrenia bidirectionally
Apply sensitivity analyses including weighted median, weighted mode, MR-Egger regression, MR-PRESSO and colocalization analyses to assess and adjust for pleiotropy
We will perform Linkage Disequilibrium (LD)-score regression (4) to estimate any genetic correlation between cleft of the lip and/or palate, and ADHD, ASD, anxiety, depression, schizophrenia
1. Berman et al. Prevalence and Factors Associated with Behavioral Problems in 5-Year-Old Children Born with Cleft Lip and/or Palate from the Cleft Collective. Cleft Palate Craniofac J. 2024 Jan;61(1):40-51. doi: 10.1177/10556656221119684.
2. Dardani et al. Is genetic liability to ADHD and ASD causally linked to educational attainment?, Int J Epidemiol. 2022 Jan 6;50(6):2011-2023. doi: 10.1093/ije/dyab107.
3. Davey Smith, G. & Hemani, G. Mendelian randomization: genetic anchors for causal inference in epidemiological studies. Hum. Mol. Genet. 23, R89–R98 (2014)
4. Bulik-Sullivan et al. LD Score regression distinguishes confounding from polygenicity in genome-wide association studies. Nat Genet 2015;47:291–95.
- Prof Evie Stergiakouli - Prof Sarah Lewis -
Cleft of the lip and/or palate (CL/P) is one of the most common congenital anomalies requiring corrective surgery within the first year of life. Despite treatment to repair the cleft in infancy, being born with a cleft frequently results in multiple adverse outcomes across the lifespan, including facial disfigurement, impaired speech and low intelligibility, with potentially poor educational, vocational, social, mental and physical health outcomes. We have previously shown, using data from a large cohort of children born with cleft, the Cleft Collective, that children born with CL/P have higher levels of behavioural problems than children in the general population at 5 and 10 years (1). We will now extend this work to investigate more mental health and neurodevelopmental traits in the Cleft Collective and determine the contribution of common genetic variants to mental health traits in cleft.
We have recently conducted one of the largest trio GWAs studies of cleft using data from the Cleft Collective and wish to use these data to explore association of cleft and mental health.
In this project, we will compare trajectories of neurodevelopmental and mental health traits in the Cleft Collective with those from children in the general population. We will use data from a large meta-analysis GWAS of cleft (including Cleft Collective data) to explore the links between cleft and psychiatric disorders (ADHD, autism spectrum disorder (ASD), anxiety, depression, schizophrenia).
We will describe the distribution and correlation of each mental health trait (SDQ, MFQ, ASQ, SCARED) in the Cleft Collective and examine these separately by gender and cleft subtype. We will model individual trait trajectories for each of the scales using all available time points. We will then compare the trajectories between the Cleft Collective and children from the general population.
Building on our findings showing that CL/P does not directly cause educational underachievement (2), we will use common genetic variants across the genome to determine whether neurodevelopmental and psychiatric traits/disorders share genetic risk factors with CL/P. Within cleft, we will generate polygenic risk scores (PRS) for psychiatric disorders and mental health traits and test if PRS are associated with mental health problems in cleft. We will also test if PRS (common genetic variants) are also important in children with cleft and large genetic mutations (such as copy number variants).
1. Berman et al. Prevalence and Factors Associated with Behavioral Problems in 5-Year-Old Children Born with Cleft Lip and/or Palate from the Cleft Collective. Cleft Palate Craniofac J. 2024 Jan;61(1):40-51. doi: 10.1177/10556656221119684.
2. Dardani et al. Is genetic liability to ADHD and ASD causally linked to educational attainment?, Int J Epidemiol. 2022 Jan 6;50(6):2011-2023. doi: 10.1093/ije/dyab107.
- Dr Ana Goncalves Soares - Dr Lucy Goudswaard -
The increased risk of cardiovascular morbidity and mortality related to traffic air pollution and noise has been increasingly recognised.1,2,3 Inflammation is one of the postulated mechanisms via which these environmental exposures might affect cardiovascular risk.1,2 Proteomics can help to characterise the complex biological processes between these environmental exposures and inflammation.4,5 Studies suggest that children are more vulnerable to the effects of air pollution,5,6 therefore investigating this age group is of particular importance.
The aim of this study is to assess the association of road traffic noise and air pollution with inflammatory proteomics in 9-year-old children.
Data from the 9-year follow-up of the Avon Longitudinal Study of Parents and Children (ALSPAC) will be used. We will assess the association of road traffic noise (whole day noise, [Lden] and night time noise [Lnight]) and air pollution exposure (particulate matter <2.5 [PM2.5] and <10 µg [PM10], nitrogen dioxide [NO2] and black carbon [BC]) with 92 circulating inflammatory proteins.7 Multivariate linear regressions, adjusted for possible confounders, will be used.
1. Münzel T, Molitor M, Kuntic M, Hahad O, Röösli M, Engelmann N, Basner M, Daiber A, Sørensen M. Transportation Noise Pollution and Cardiovascular Health. Circ Res. 2024 Apr 26;134(9):1113-1135. doi: 10.1161/CIRCRESAHA.123.323584
2. Rajagopalan S, Brook RD, Salerno P, Bourges-Sevenier B, Landrigan P, Nieuwenhuijsen MJ, Munzel T, Deo SV, Al-Kindi S: Air pollution exposure and cardiometabolic risk. Lancet Diabetes Endocrinol 2024, 12(3):196-208; doi: 10.1016/S2213-8587(23)00361-3.
3. Khoshakhlagh AH, Mohammadzadeh M, Gruszecka-Kosowska A, Oikonomou E. Burden of cardiovascular disease attributed to air pollution: a systematic review. Global Health. 2024 May 3;20(1):37. doi: 10.1186/s12992-024-01040-0. PMID: 38702798
4. Casella C, Kiles F, Urquhart C, Michaud DS, Kirwa K, Corlin L. Methylomic, Proteomic, and Metabolomic Correlates of Traffic-Related Air Pollution in the Context of Cardiorespiratory Health: A Systematic Review, Pathway Analysis, and Network Analysis. Toxics. 2023 Dec 12;11(12):1014. doi: 10.3390/toxics11121014
5. He S, Klevebro S, Baldanzi G, Pershagen G, Lundberg B, Eneroth K, Hedman AM, Andolf E, Almqvist C, Bottai M et al: Ambient air pollution and inflammation-related proteins during early childhood. Environ Res 2022, 215(Pt 2):114364; doi: 10.1016/j.envres.2022.114364.
6. Fitton C.A., Cox, B., Stewart, M., Chalmers, J., Belch, J.J.F. (2023). Respiratory Admissions Linked to Air Pollution in a Medium Sized City of the UK: A Case-crossover Study. Aerosol Air Qual. Res. 23, 230062. https://doi.org/10.4209/aaqr.230062
7. Goulding N, Goudswaard LJ, Hughes DA, Corbin LJ, Groom A, Ring S, Timpson NJ, Fraser A, Northstone K, Suderman M: Inflammation proteomics datasets in the ALSPAC cohort. Wellcome Open Res 2024, 7(277); doi: 10.12688/wellcomeopenres.18482.2.
- Dr Ana Goncalves Soares - Dr David Hughes -
Studies have shown that transportation noise increases the risk for cardiovascular morbidity and mortality.1,2 Traffic noise, especially at night time, causes fragmentation and shortening of sleep, elevation of stress hormone levels, and increased oxidative stress.1 These factors can promote vascular (endothelial) dysfunction, inflammation, and arterial hypertension, thus elevating cardiovascular risk.1,3 Whilst the effects of road traffic noise on cardiovascular health have been more widely studied in adults, much less evidence is available in children. Furthermore, traffic noise is highly correlated with air pollution, given transportation is the major source of both,4 and understanding their interplay is of great importance to disentangle their individual and combined effects.
The aim of this study is to assess the association between road traffic noise and several cardiometabolic risk factors and inflammatory markers in children.
Data from the 7-year follow-up of the Avon Longitudinal Study of Parents and Children (ALSPAC) will be used. We will assess the association of road traffic noise exposure (whole day noise [Lden] and night time noise [Lnight]) with cardiovascular risk factors, namely blood pressure, pulse rate, body mass index (BMI), glucose, and blood lipids. We will also assess the association of road traffic noise with C-reactive protein (CRP), as a proxy of inflammation. Multivariate linear regressions, adjusted for possible confounders, will be used. Interactions with air pollution exposure will also be explored.
1. Münzel T, Molitor M, Kuntic M, Hahad O, Röösli M, Engelmann N, Basner M, Daiber A, Sørensen M. Transportation Noise Pollution and Cardiovascular Health. Circ Res. 2024 Apr 26;134(9):1113-1135. doi: 10.1161/CIRCRESAHA.123.323584
2. Fu X, Wang L, Yuan L, Hu H, Li T, Zhang J, Ke Y, Wang M, Gao Y, Huo W, Chen Y, Zhang W, Liu J, Huang Z, Zhao Y, Hu F, Zhang M, Liu Y, Sun X, Hu D. Long-Term Exposure to Traffic Noise and Risk of Incident Cardiovascular Diseases: a Systematic Review and Dose-Response Meta-Analysis. J Urban Health. 2023 Aug;100(4):788-801. doi: 10.1007/s11524-023-00769-0
3. Kupcikova Z, Fecht D, Ramakrishnan R, Clark C, Cai YS. Road traffic noise and cardiovascular disease risk factors in UK Biobank. Eur Heart J. 2021 Jun 1;42(21):2072-2084. doi: 10.1093/eurheartj/ehab121.
4. Eminson K, Cai YS, Chen Y, Blackmore C, Rodgers G, Jones N, Gulliver J, Fenech B, Hansell AL. Does air pollution confound associations between environmental noise and cardiovascular outcomes? - A systematic review. Environ Res. 2023 Sep 1;232:116075. doi: 10.1016/j.envres.2023.116075.
- Dr Chloe Slaney - Professor Golam Khandaker -
C-reactive protein (CRP) is an acute-phase protein often used as a marker of systemic inflammation. Whilst CRP has been shown to be associated with many clinical conditions (e.g., cardiovascular disease, psychiatric conditions), it is unclear whether it plays a causal role.
Interestingly, Mendelian randomization studies have suggested that higher CRP levels may have a protective effect on schizophrenia (i.e., higher CRP reducing risk of schizophrenia) (Hartwig et al., 2017; Ligthart et al., 2018; Prins et al., 2016; Said et al., 2022). It is unclear through what plausible mechanism this could occur. One proposed mechanism is that genetic predisposition for higher CRP may reduce risk of early-life infection, which in turn may reduce risk of schizophrenia (Hartwig et al., 2017). Although this has been proposed in several studies, to our knowledge it has not been directly tested.
In this project, we will test the association between genetically proxied CRP on early-life infection.
Study Aim: Test the potential causal role of CRP on early-life infection (based on antibody and self-report data) in a population-based cohort (Avon Longitudinal Study of Parents and Children; ALSPAC).
Objectives:
1. Generate CRP genetic risk score within the ALSPAC cohort.
2. Generate early-life infection outcome variables (e.g., presence of early life-infection; number of early-life infections) using self-report and antibody data within ALSPAC.
3. Test the potential causal role of CRP on early-life infection using Mendelian randomization.
This project will use individual-level phenotypic and genetic data in a large population-based cohort (ALSPAC). Specifically, it will involve use of (1) publicly available GWAS data to create a CRP genetic risk score within the ALSPAC cohort, and (2) self-report and antibody data to assess early-life infection within the ALSPAC cohort (e.g., presence of an infection in early life; number of early-life infections). Following this, Mendelian randomization will be applied to test potential causal relationships between CRP and early-life infection outcomes.
The PhD candidate will gain valuable skills in working with population-based cohort data, conducting genetic analyses (using GWAS summary data, creating genetic risk scores, Mendelian randomization), experience with statistical software (R) and high-performance computing.
Hartwig, F. P., Borges, M. C., Horta, B. L., Bowden, J., & Davey Smith, G. (2017). Inflammatory biomarkers and risk of schizophrenia: A 2-sample mendelian randomization study. JAMA Psychiatry, 74(12), 1226–1233.
Ligthart, S., Vaez, A., Võsa, U., Stathopoulou, M. G., de Vries, P. S., Prins, B. P., Van der Most, P. J., Tanaka, T., Naderi, E., Rose, L. M., Wu, Y., Karlsson, R., Barbalic, M., Lin, H., Pool, R., Zhu, G., Macé, A., Sidore, C., Trompet, S., … Alizadeh, B. Z. (2018). Genome Analyses of >200,000 Individuals Identify 58 Loci for Chronic Inflammation and Highlight Pathways that Link Inflammation and Complex Disorders. American Journal of Human Genetics, 103(5), 691–706.
Prins, B. P., Abbasi, A., Wong, A., Vaez, A., Nolte, I., Franceschini, N., Stuart, P. E., Guterriez Achury, J., Mistry, V., Bradfield, J. P., Valdes, A. M., Bras, J., Shatunov, A., Lu, C., Han, B., Raychaudhuri, S., Bevan, S., Mayes, M. D., Tsoi, L. C., … Alizadeh, B. Z. (2016). Investigating the Causal Relationship of C-Reactive Protein with 32 Complex Somatic and Psychiatric Outcomes: A Large-Scale Cross-Consortium Mendelian Randomization Study. PLOS Medicine, 13(6), e1001976.
Said, S., Pazoki, R., Karhunen, V., Võsa, U., Ligthart, S., Bodinier, B., Koskeridis, F., Welsh, P., Alizadeh, B. Z., Chasman, D. I., Sattar, N., Chadeau-Hyam, M., Evangelou, E., Jarvelin, M.-R., Elliott, P., Tzoulaki, I., & Dehghan, A. (2022). Genetic analysis of over half a million people characterises C-reactive protein loci. Nature Communications 2022 13:1, 13(1), 1–10.
- Grace M. Power - Chrissy Hammond -
Affecting nearly 2 billion people globally, musculoskeletal conditions can result in severe and lasting consequences. Furthermore, the annual worldwide cost of hip fractures alone is projected to reach US$132 billion by 2050, highlighting the major social and economic burden of these health states and underscoring the importance of identifying pathways for prevention and improving intervention options.
We have recently published novel evidence that a higher body size in childhood reduces the risk of fractures later in life through its influence on increased estimated bone mineral density (eBMD). This relationship is complex from a public health perspective, however, as adult obesity remains a significant risk factor for various adverse health conditions. That being said, quantifying the impact of body size at different stages of life on musculoskeletal health in later years is important as it helps us to (i) identify modifiable pathways to musculoskeletal health for potential intervention targets, and (ii) improve the predictive value of body size at various life stages in assessing musculoskeletal health.
Employing a lifecourse Mendelian randomisation (MR) approach by using genetic instruments to separate effects at different life stages, this project sets out to explore how prepubertal and adult body size independently influence a range of musculoskeletal health conditions, including osteoarthritis and osteoporosis. It will additionally investigate whether plausible risk factors for musculoskeletal health outcomes serve as intermediate variables (mediators) on the causal pathway between childhood body size and selected outcomes.
Using data from the UK Biobank and other publicly available resources, univariable and multivariable MR will be conducted to simultaneously estimate the effects of age-specific genetic proxies for body size on several musculoskeletal traits. A two-step MR framework will additionally be applied to elucidate potential mediators.
The student will have the opportunity to draft an abstract for submission to both national and international conferences and expand their project into a research paper suitable for journal submission. Results from MR analyses will be used for future functional testing in the genetically tractable zebrafish animal model across their lifecourse, and the student can learn more about the link between genetic epidemiological approaches and functional translation via wet lab studies if they are interested.
Power, GM, Tobias, JH, Frayling, TM et al. Age-specific effects of weight-based body size on fracture risk in later life: a lifecourse Mendelian randomisation study. Eur J Epidemiol 38, 795–807 (2023).
Kemp JP, Sayers A, Davey Smith G, Tobias JH, Evans DM. Using mendelian randomization to investigate a possible causal relationship between adiposity and increased bone mineral density at different skeletal sites in children. Int J Epidemiol. 2016;45(5):1560–72.
Richardson TG, Sanderson E, Elsworth B, Tilling K, Davey Smith G. Use of genetic variation to separate the effects of early and later life adiposity on disease risk: mendelian randomisation study. BMJ. 2020;369:m1203.
- Dr Robyn Wootton - Professor Marcus Munafò -
Cigarette smoking is a risk factor for many health outcomes (e.g. cancer, mental illness, cardiovascular diseases), and is often an important confounder. Mendelian randomisation (MR) is a causal inference technique designed to answer these respective questions using genetic instruments. For smoking, ideally smoking heaviness SNPs would be used in a stratified sample of smokers and contrasted with the negative control of non-smokers. However, often MR is conducted with summary level data, hence the lifetime smoking genetic instrument was created (Wootton et al., 2019). It captures initiation, heaviness, duration, and cessation and can be used in unstratified samples. To date, there has been no systematic review of these applications, nor how suitably the genetic instrument is being applied. A review is timely given: 1) a rapid increase in MR studies published in predatory journals, 2) recent evidence of widespread pleiotropy for lifetime smoking (Reed et al., 2023).
1. Summarise the applications of the lifetime smoking genetic instrument in downstream analyses
2. Quantify and summarise the risk of bias for each MR study (including smoking specific best practice)
3. Make recommendations for future applications of this genetic instrument based on our review of the evidence and recent developments in the field.
Methods to achieve aim 1: Since its publication, the lifetime smoking index paper has been cited 208 times. These 208 papers citing the lifetime smoking index will first be classified and summarised based on the methods they have used for their downstream analysis and which other types of traits they explored. If the papers use MR, we will summarise whether smoking was an exposure or a confounder/mediator.
Methods to achieve aim 2: Papers which have used Mendelian randomisation analysis will be assessed for risk of bias using the Mamluk Risk of Bias Tool (Mamluk et al., 2020). If there are a large number of papers, we may take a random subsample instead of assessing for all. We will additionally include some smoking specific modifications (e.g. did they have individual level data available, if so did they also incorporate smoking heaviness and the negative control of non-smokers).
Methods to achieve aim 3: the findings of this work, and recent publications in the field will be written up in a publication providing guidance and advice for employing the lifetime smoking genetic instrument based upon the current state of the evidence.
Wootton RE, Richmond RC, Stuijfzand BG, Lawn RB, Sallis HM, Taylor GMJ, Hemani G, Jones HJ, Zammit S, Davey Smith G, Munafò MR. Evidence for causal effects of lifetime smoking on risk for depression and schizophrenia: a Mendelian randomisation study. Psychol Med. 2020 Oct;50(14):2435-2443. doi: 10.1017/S0033291719002678. Epub 2019 Nov 6. PMID: 31689377; PMCID: PMC7610182.
Reed, ZE, Wootton, RE, Khouja, JN, Richardson, G, Sanderson, E, Davey Smith, G, Munafò MR. Exploring pleiotropy in Mendelian randomisation analyses: What are genetic variants associated with “cigarette smoking initiation” really capturing? MedRxiv. 2023. doi: https://doi.org/10.1101/2023.08.04.23293638
Loubaba Mamluk, Timothy Jones, Sharea Ijaz, Hannah B Edwards, Jelena Savović, Verity Leach, Theresa H M Moore, Stephanie von Hinke, Sarah J Lewis, Jenny L Donovan, Deborah A Lawlor, George Davey Smith, Abigail Fraser, Luisa Zuccolo, Evidence of detrimental effects of prenatal alcohol exposure on offspring birthweight and neurodevelopment from a systematic review of quasi-experimental studies, International Journal of Epidemiology, Volume 49, Issue 6, December 2020, Pages 1972–1995, https://doi.org/10.1093/ije/dyz272
- Dr Yi Liu - Dr Zhaozhen Xu -
Forecasting time series data can be approached by classical statistical methods such as ARIMA-GARCH, as well as the new statistical learning or machine learning methods such as Prophet (Taylor and Lethan, 2018) or CatBoost (Prokhorenkova et al., 2018).
The recent few years have seen ground-breaking improvement in machine learning from large foundation models (also known as Large Language Models) and their associated methods, e.g. the Transformer architecture, the Attention mechanism, and transfer learning / pretraining.
Although they are typically widely known for the explosive success in Natural Language Processing (e.g. ChatGPT), new Transformer time series methods are becoming available, e.g. TimesFM (Das et al., 2023), which claim to have similar potential to improve over traditional methods.
Nevertheless the application of them on public health problems (e.g. forecasting blood glucose level from insulin intake, or detecting anomalies in heart rates) remains to investigated.
The purpose of this project is to investigate the performance of time series Transformer models on biomedical time series data and provide insights on directions where these next-gen methods should be appropriately used on public health problems.
Specifically, we aim to investigate:
1. The accuracy and robustness of the candidate Transformer models (or models using their underlying mechanisms) in time series forecasting, as compared to other well established methods.
2. To what extent and why time series forecasting benefits from Transformers and underlying mechanisms, i.e. if a pretrained time series foundation model is able to offer substantial benefit over ARIMA then we would like to know more about the underlying factors.
3. Additionally, if the forecasting can be applied to change point detection and regime changes which can offer further insights on public health problems.
The student will apply cutting-edge Transformer-based time series methods (such as TimesFM) to forecast time series data on such as the MIMIC-III Blood Glucose Management dataset or the RR interval time series dataset. The student will also apply some of the classical statistical time series methods (from the Python sktime and prophet frameworks) to compare the results with the Transformer models.
We expect the main computation to be carried out on University of Bristol HPCs such as BluePebble or Isambard-AI (when it is available).
The student is expected to use Python and its libraries / frameworks (such as Pytorch, Huggingface Transformers and Pandas) to carry out the research. For student without prior experiences in Python but wants to learn, this mini project will be a good opportunity to learn Python.
Das, A., Kong, W., Sen, R., & Zhou, Y. (2023). A decoder-only foundation model for time-series forecasting. arXiv preprint arXiv:2310.10688.
Taylor, S. J., & Letham, B. (2018). Forecasting at scale. The American Statistician, 72(1), 37-45.
Prokhorenkova, L., Gusev, G., Vorobev, A., Dorogush, A. V., & Gulin, A. (2018). CatBoost: unbiased boosting with categorical features. Advances in neural information processing systems, 31.
MIMIC-III Blood Glucose Management dataset https://physionet.org/content/glucose-management-mimic/1.0.1/
RR interval time series dataset https://physionet.org/content/rr-interval-healthy-subjects/1.0.0/
- Grace M. Power - Eleanor Sanderson -
Conventional epidemiolocal studies have reported a graded, inverse relationship between high-density lipoprotein cholesterol (HDL-C) and both cardiovascular disease (CVD). However, findings from clinical trials and Mendelian randomization (MR) analyses point to a negligible effect of HDL-C on CVD risk. More recently a causal inverse association of HDL-C at low levels with CVD has been reported using stratified nonlinear MR methods to guide analyses. However, there are notifiable issues with these approaches. For example, the stratification process required to perform this analysis may lead to enhanced heritable confounding (or correlated pleiotropic) effects driven by cytokines such as interleukin 6 (IL-6), in the strata. This project sets out to investigate the causal effect of HDL-C on CVD risk using a range of novel genetic epidemiological methods to fully interrogate this relationship.
This project, using data from the UK Biobank cohort, will examine the relationship between HDL-C and CVD risk using several novel methodological approaches developed to account for potential heritable confounding (otherwise known as correlated pleitropy) within a Mendelian randomization (MR) framework.
In this project the student will be required to run several analyses to interrogate the potential for heritable confounding, including:
- Generating a polygenic risk score (PRS) and deriving a residualised measure of HDL-C from this
- Generating a genome-wide association study (GWAS)
- Estimating the SNP effects from the GWAS onto the residualised HDL-C
- Univariable and multivariable Mendelian randomization
- Steiger filtering
The student will have the opportunity to expand their project into a research paper suitable for journal submission.
Hamilton, F.W., Hughes, D.A., Spiller, W. et al. Non-linear Mendelian randomization: detection of biases using negative controls with a focus on BMI, Vitamin D and LDL cholesterol. Eur J Epidemiol 39, 451–465 (2024). doi.org: 10.1007/s10654-024-01113-9
Hamilton F, Pedersen KM, Ghazal P, Nordestgaard BG, Smith GD. Low levels of small HDL particles predict but do not influence risk of sepsis. Crit Care. 2023 Oct 9;27(1):389. doi: 10.1186/s13054-023-04589-1.
Chen JX, Li Y, Zhang YB, Wang Y, Zhou YF, Geng T, Liu G, Pan A, Liao YF. Nonlinear relationship between high-density lipoprotein cholesterol and cardiovascular disease: an observational and Mendelian randomization analysis. Metabolism. 2024 May;154:155817. doi: 10.1016/j.metabol.2024.155817.
- Qian Yang - Carolina Borges -
Hypertensive disorders of pregnancy (HDP) affect approximately 10% of all pregnancies, contribute to maternal and fetal mortality, and are associated with the development of chronic hypertension. HDP encompasses a spectrum of conditions, incl. gestational hypertension, preeclampsia and eclampsia, and identifying their causal metabolites could help understand their pathologies and pave the way for targeted interventions. Nuclear magnetic resonance (NMR) metabolomics enables profiling of the lipidome, some amino acids and other small-molecule metabolites. Circulating NMR metabolites have been associated with complex diseases, incl. hypertension and HDP. We have observed associations of maternal pregnancy NMR metabolites (fatty acids, amino acids) with a higher risk of preeclampsia in Born in Bradford. These associations may be explained by residual confounding and reverse causality. Mendelian randomization (MR) provides an alternative approach to probe causal associations of NMR metabolites with HDP. Previous MR studies have identified some NMR metabolites affect blood pressure, outside of pregnancy and offspring birthweight.
The aim of this study is to explore the causal effect of maternal metabolites on the risk of HDP and its subtypes using MR.
This study will be undertaken within the MR-PREG collaboration, which aims to explore causes and consequences of different pregnancy and perinatal outcomes. The PhD candidate will have access to the largest genome-wide association studies (GWAS) currently available for metabolites and HDP. The student will select genetic variants strongly associated with NMR metabolites using data from ~300,000 UK Biobank individuals. For the selected genetic variants, the student will extract female-specific GWAS data for NMR metabolites (N ~150,000), as well as for GWAS meta-analyses of HDPs (N cases: ~26,000, N controls: ~416,000), gestational hypertension (N cases: ~17,000, N controls: ~405,000), preeclampsia (N cases: ~14,000, N controls: ~474,000).
The student will have the opportunity to use, and further develop, methods that account for high levels of biological and environmental correlation between metabolites. For example, we have shown how univariable and multi-variable MR can be used to account for correlation between metabolites, and the student would likely use this method for some of the work. This was used only after prior selection based on biology and strong correlations. The student will also be encouraged to explore the possibility of instrumenting more metabolites with higher correlations using newly developed MR methods.
1. Taylor K, Ferreira DLS, West J, Yang T, Caputo M, Lawlor DA. Differences in Pregnancy Metabolic Profiles and Their Determinants between White European and South Asian Women: Findings from the Born in Bradford Cohort. Metabolites. 2019;9.
2. Zhao J, Stewart ID, Baird D, Mason D, Wright J, Zheng J, et al. Causal effects of maternal circulating amino acids on offspring birthweight: a Mendelian randomisation study. EBioMedicine. 2023;88:104441.
3. Barry CS, Lawlor DA, Shapland CY, Sanderson E, Borges MC. Using Mendelian Randomisation to Prioritise Candidate Maternal Metabolic Traits Influencing Offspring Birthweight. Metabolites. 2022;12.
4. Yang Q, Borges MC, Sanderson E, Magnus MC, Kilpi F, Collings PJ, et al. Associations between insomnia and pregnancy and perinatal outcomes: Evidence from mendelian randomization and multivariable regression analyses. PLoS Med. 2022;19:e1004090.
- Dr Zhaozhen Xu - Dr Yi Liu -
Screening relevant studies for a systematic review is labour-intensive. It usually takes two reviewers about a month for one review topic. Therefore, researchers seek machine learning and natural language processing approaches to automate this process.
In recent years, some Large Language Models (LLMs), such as ChatGPT, achieved solid results in study screening [1]. In the previous research, we fine-tuned a biomedical LLM, BlueBERT [2], to screen studies for systematic reviews on cancer incidence and survival studies. Even though it obtains a good performance, some misclassified examples remain in the database. Introducing a more complicated and computationally expensive LLM could further improve the screening accuracy of these examples.
Alongside the development of LLMs, prompt engineering [3] provides a new and more accessible approach to maximise the model performance by crafting the input prompts. The model can learn and think like a reviewer with an effective prompt.
In this project, we will teach the LLM to classify the studies as inclusion/exclusion for a given systematic review topic using prompt engineering. For example, a basic prompt for a breast cancer incidence review is shown below.
“Please review the following abstracts and determine their eligibility for inclusion in a breast cancer incidence systematic review.”
Nevertheless, this prompt can be further improved by giving more specific and detailed instructions, such as type of studies, exposures, inclusion examples, etc. Different prompts will lead to different outputs generated from LLM.
The student will tackle the misclassified examples from the previous study screening results produced by a trained model. More specifically, we aim to
1. Investigate the construction of prompts for study screening
2. Improve the screening accuracy on the false negative and false positive examples.
The student will apply prompt engineering methods on one of the open-sourced LLM: LLaMA [4].
We will provide a list of misclassified studies for each systematic review topic, including study title, abstract, and human labelling of inclusion or exclusion. The student will need to design effective prompts for LLaMA (which can be accessed from the Python Huggingface library) to generate an inclusion/exclusion decision based on additional information, such as screening criteria. The results generated from different prompts will be compared using evaluation metrics such as accuracy, precision, and recall.
The main part of the project will be carried out at the University of Bristol HPC (BluePebble).
The student will need to use Python and its libraries (e.g. Pandas and Huggingface) for this project. It can be a good opportunity for students who are interested in learning Python and machine learning basics but do not have previous experience.
[1] Guo E, Gupta M, Deng J, Park YJ, Paget M, Naugler C. Automated Paper Screening for Clinical Reviews Using Large Language Models: Data Analysis Study. J Med Internet Res. 2024 Jan 12;26:e48996. doi: 10.2196/48996. PMID: 38214966; PMCID: PMC10818236.
[2] Peng Y, Yan S, Lu Z. Transfer learning in biomedical natural language processing: an evaluation of BERT and ELMo on ten benchmarking datasets. arXiv preprint arXiv:1906.05474. 2019 Jun 13.
[3] https://platform.openai.com/docs/guides/prompt-engineering/six-strategies-for-getting-better-results
[4] Touvron H, Lavril T, Izacard G, Martinet X, Lachaux MA, Lacroix T, Rozière B, Goyal N, Hambro E, Azhar F, Rodriguez A. Llama: Open and efficient foundation language models. arXiv preprint arXiv:2302.13971. 2023 Feb 27.
- Dr Charlotte James - Dr Louise Millard -
In the UK, of the people known to have been infected with COVID-19, an estimated 2 million (10% of COVID-19 infections, 3% of the population) report symptoms that persist longer than 4 weeks.[1] Of these, 1.4 million (74%) have post-COVID-19 syndrome: symptoms develop during or after infection and continue for more than 12 weeks.[1] When symptoms of COVID-19 persist for longer than 4 weeks, or new symptoms develop after infection, an individual is considered to have Long COVID. As Long COVID is a new syndrome, to improve diagnostic accuracy and quality of life for those infected, understanding which symptoms characterise the syndrome is paramount.
An existing project in the UK Longitudinal Linkage Collaboration (UKLLC, https://ukllc.ac.uk ), has used Latent Class Analysis (LCA) to identify symptom patterns associated with recent and historic COVID-19 infection.
The aim of this project is to use machine learning (ML) algorithms to find clusters of symptoms following COVID-19 infection and compare these clusters with the symptom patterns identified using LCA.
The student will work within the UKLLC,* using a curated dataset of self-reported symptoms from 10 longitudinal population studies. Following dimensionality reduction using, for example, PCA, they will apply different ML clustering algorithms (e.g. k-means, DBSCAN) to self-reported symptoms and compare the clusters to the latent classes already obtained.
The student will gain skills working in Trusted Research Environments; applying methods for dimensionality reduction; applying and evaluating ML clustering algorithms in R/Python; and using git and GitLab for version control.
*To work within the UKLLC, the student will be required to complete ONS Safe Researcher training.
[1] Presence of ongoing symptoms following coronavirus (COVID-19) infection in the UK: 7 July 2022. https://www.ons.gov.uk/peoplepopulationandcommunity/healthandsocialcare/conditionsanddiseases/bulletins/prevalenceofongoingsymptomsfollowingcoronaviruscovid19infectionintheuk/7july2022
- Dr Monika Halicka - Professor Deborah Caldwell -
Cannabis use disorder (CUD) is diagnosed based on habitual use, craving, and inability to stop consuming cannabis despite it causing physical or psychological harm. With increasing incidence and prevalence of CUD globally, more and more people are seeking treatment. There is considerable variability in how the effectiveness of interventions for CUD is measured in clinical trials, and lack of consensus on what represents a meaningful therapeutic change. This poses a challenge for comparing or pooling results across trials to identify the most effective treatments. Abstinence from cannabis tends to be the most commonly used indicator of intervention effectiveness, however, a clinical benefit from the patient perspective could also be conceptualized as reduced symptoms, severity, or remission of CUD. Loflin et al. (2020) recommended a core standard assessment toolkit for CUD, emphasizing that trials should include a symptom checklist corresponding to the DSM-5 diagnostic criteria for CUD.
(1) To compare the frequency of reporting symptom checklist outcomes in CUD trials before and after publication of the recommendations for a core standard assessment toolkit. (2) To characterize the methodology of assessing symptom checklist, as consensus is still needed on the optimal measurement protocol. This addresses life course epidemiology in the following ways: Heavy cannabis use has been associated with adverse outcomes throughout life course, such as increased likelihood of developing dependence on other illicit drugs, nicotine, and alcohol, mental health problems, and poorer socio-economic wellbeing. Evaluating the reporting of relevant patient outcomes in the context of treatments for cannabis use disorder could aid further research in identifying the most beneficial interventions. Effectively treating CUD may help to reduce the risk of adverse outcomes later in life.
This project is embedded in ongoing systematic reviews of the effectiveness and safety of pharmacological and psychosocial interventions for CUD (NIHR167862; NIHR165373; ESG, 2024). Firstly, studies will be screened for reporting of outcomes linked to Loflin’s symptom checklist, including assessments of CUD diagnostic status, number of symptoms, and/or severity of the disorder. The frequency of reporting these outcomes will be compared over the time period since the publication of the recommendations (Loflin et al., 2020) versus the same duration of time prior to publication. Secondly, details of how the symptom checklist outcomes were assessed will be extracted from the identified studies, including specific tools used, timescales (e.g. past 3 months), type of diagnostic criteria (e.g. ICD, DSM) and adaptations to changes in these criteria over time (e.g. DSM-IV, DSM-5). This information will be synthesized narratively and critically evaluated.
Loflin et al. Drug Alcohol Depend;2020;212:107993.
NIHR167862. https://fundingawards.nihr.ac.uk/award/NIHR167862
NIHR165373. https://fundingawards.nihr.ac.uk/award/NIHR165373
ESG, 2024. https://tinyurl.com/esg-cud
- Professor Sarah Lewis - Dr Lucy Goudswaard -
Although, agricultural workers generally have a lower cancer risk than the general population, the risk of melanoma, multiple myeloma and prostate cancer was found to be elevated for this group (Togawa et al, 2021). A systematic review of 25 studies of pesticide exposure on prostate cancer risk, found some evidence of an effect, which was greater in studies with expert level exposure measurements (Ohlander et al, 2022). Polymorphisms in genes involved in the metabolism and detoxification of chemicals can help to shed light on the potential of a compound to cause harm. For example a study of sheep dippers found that the health effects of exposure to sheep dip were associated with polymorphisms in the serum paraoxonase (PON1) gene, which is involved in the hydrolysis of organophosphates (Povey et al, 2005).
Aim: To determine whether pesticide exposure is responsible for the increased risk of multiple myeloma and prostate cancer among agricultural workers.
Objectives:
1) Identify genes involved in metabolism and detoxification of pesticides
2) Identify functional variants in those genes
3) Use available GWAS data to determine whether functional variants in genes related to pesticides are over-represented among cases with prostate cancer and multiple myeloma
4) Use non-functional variants in the same genes as negative control exposures
5) If time allows, determine whether functional variation in genes linked to pesticide exposure are associated with any of the key characteristics of cancer risk.
1) Conduct a literature review to identify genes which are involved in the metabolism, hydrolysis or detoxification of commonly used pesticides.
2) Use databases such as dbSNP (https://www.ncbi.nlm.nih.gov/projects/SNP/) to identify functional and non-functional polymorphisms in the genes identified in 1).
3) Using existing genome wide association studies of prostate cancer and multiple myeloma extract the results for the polymorphisms identified in 2) in relation to cancer risk.
4) Statistically evaluate whether there is an excess of associations with functional versus non-functional variants in genes involved in pesticide exposure.
Togawa K et al. Cancer incidence in agricultural workers: Findings from an international consortium of agricultural cohort studies (AGRICOH). Environ Int. 2021 Dec;157:106825. doi: 10.1016/j.envint.2021.106825.
Ohlander J. Impact of occupational pesticide exposure assessment method on risk estimates for prostate cancer, non-Hodgkin's lymphoma and Parkinson's disease: results of three meta-analyses. Occup Environ Med. 2022 Aug;79(8):566-574. doi: 10.1136/oemed-2021-108046.
Povey AC et al. Paraoxonase polymorphisms and self-reported chronic ill-health in farmers dipping sheep. Occup Med (Lond). 2005 Jun;55(4):282-6. doi: 10.1093/occmed/kqi128.
- Professor Sarah Lewis - Dr Carolina Borges -
Orofacial clefts are the most common type of birth defects worldwide and occur at a rate of one in 650 live births in the UK. The majority of clefts follow a multifactorial model with both genetic and prenatal environmental risk factors that are still largely unknown. Whilst the cleft is usually repaired in the first year of life, often further surgeries are required as the child grows, coupled with other interventions to improve speech, hearing and psychological outcomes amongst those born with a cleft. A better understanding of the causes of orofacial clefts is essential to inform prevention. This project will use genome wide data from the Cleft Collective, the world’s largest cohort study of individuals affected by cleft and their families based in the UK to conduct Mendelian randomization studies to determine whether modifiable risk factors cause cleft.
In this project, we will use genetic data from mothers of children born with a cleft to identify modifiable risk factors which may cause cleft.
A. We will use two-sample Mendelian randomization (3) to investigate any causal links between cleft and maternal diabetes, BMI, folate intake, vitamin A intake and smoking during pregnancy.
B. Apply sensitivity analyses including weighted median, weighted mode, MR-Egger regression, MR-PRESSO and colocalization analyses to assess and adjust for pleiotropy.
Tyrrell J et al. Early Growth Genetics (EGG) Consortium. Genetic variation in the 15q25 nicotinic acetylcholine receptor gene cluster (CHRNA5-CHRNA3-CHRNB4) interacts with maternal self-reported smoking status during pregnancy to influence birth weight. Hum Mol Genet. 2012 Dec 15;21(24):5344-58. doi: 10.1093/hmg/dds372.
Davey Smith G, Ebrahim S. What can mendelian randomisation tell us about modifiable behavioural and environmental exposures? BMJ. 2005 May 7;330(7499):1076-9. doi: 10.1136/bmj.330.7499.1076.
Davies AJV, Humphries K, Lewis SJ, Ho K, Sandy JR, Wren Y. The Cleft Collective: protocol for a longitudinal prospective cohort study. BMJ Open. 2024 Jul 5;14(7):e084737. doi: 10.1136/bmjopen-2024-084737.
- Professor Sarah Lewis - Dr Kaitlin Wade -
Thousands of epidemiological studies have investigated the impact of diet on health outcomes. However, very few of the resulting observed associations represent robust causal effects due to a high susceptibility to measurement error, reverse causation and confounding. In recent years, acknowledging that nutrients are not consumed in isolation, researchers have investigated dietary patterns as exposures rather than individual nutrients or food items. However, as populations tend to adopt dietary patterns due to social, cultural and health reasons, and are influenced by the geographical availability of foods, studies of population-level dietary patterns can still be subject to confounding.
To overcome the issues of measurement error, confounding and reverse causation, there have been several Mendelian randomization (MR) studies that have investigated the effects of nutrients on disease risk. However, instrumenting dietary patterns for MR is more complex.
As an exemplar dietary pattern, this project will aim to develop a genetic instrument for the DASH diet, which was developed to reduce hypertension, by using metabolites identified to be influenced by the diet itself. The instrument will then be applied to determine whether metabolites specific to the DASH diet reduces blood pressure in a proof of principle analysis, which may be expanded to other such diets and dietary patterns.
1) Use the feeding RCT of the DASH diet to identify metabolites influenced by the DASH dietary intervention and use publicly available genome-wide association studies of metabolites to identify genetic variants associated with these metabolites.
2) Using the instrument identified above, apply MR methodology to determine whether the metabolites influenced by the DASH diet (and thus, the DASH diet itself) reduces blood pressure, sodium levels, biomarkers of cardiovascular disease and cardiovascular disease.
3) Conduct MR sensitivity analyses to test the robustness of the above findings.
Advanced steps (if time allows):
4) Compare the estimates from the MR analysis above to one using a genetic instrument based on self-reported dietary patterns
5) Compare the estimates from the MR analysis conducted above with an MR analysis using a genetic instrument for metabolites relating to the self-reported dietary patterns.
6) Compare the estimates from the MR analysis with an MR analysis which uses metabolites that are associated with BP lowering in response to DASH in a DASH feeding RCT
Wade KH, Yarmolinsky J, Giovannucci E, et al. Applying Mendelian randomization to appraise causality in relationships between nutrition and cancer. Cancer Causes Control. 2022;33(5):631-652. doi:10.1007/s10552-022-01562-1
Theodoridis X, Triantafyllou A, Chrysoula L, et al. Impact of the Level of Adherence to the DASH Diet on Blood Pressure: A Systematic Review and Meta-Analysis. Metabolites. 2023;13(8):924. Published 2023 Aug 7. doi:10.3390/metabo13
Rebholz CM, Lichtenstein AH, Zheng Z, Appel LJ, Coresh J. Serum untargeted metabolomic profile of the Dietary Approaches to Stop Hypertension (DASH) dietary pattern. Am J Clin Nutr. 2018 Aug 1;108(2):243-255. doi: 10.1093/ajcn/nqy099.
- Professor Sarah Lewis - Dr Lucy Goudswaard -
Worldwide there were over 4.5 million newly diagnosed cancers of the breast, colorectum, prostate and endometrium in 2018 [1]. A systematic review conducted by World Cancer Research Fund (WCRF) concluded that there is strong evidence that more physical activity decreases the risk of cancers of the breast, endometrium, colorectum [2]. Another study using 9 prospective cohorts demonstrated that undertaking recommended amounts of exercise was associated with a lower risk of these three cancers, along with reducing the risk of kidney and liver cancers, myeloma and non-Hodgkin’s lymphoma [3]. Despite this, the mechanisms linking physical activity and cancer risk are unclear. An epigenome-wide study of DNA methylation (EWAS) study using the Melbourne Collaborative Cohort Study (MCCS) indicated that physical activity may alter DNA methylation in peripheral blood [4], however It is unclear whether methylation sites associated with physical activity also associate with cancer outcomes.
To explore whether DNA methylation (CpG) sites altered by physical activity are associated with cancer risk
The student will work on individual level data from the Melbourne Collaborative Cohort Study (MCCS). Participants included in the proposed analysis will have been recruited as part of the cancer nested case-control studies (with breast cancer, prostate cancer, colorectal cancer, urothelial cell carcinoma, kidney cancer, gastric cancer, lung cancer and mature B-cell neoplasm outcomes) and had DNA methylation data generated from peripheral blood. A list of DNA methylation sites reported to be altered with physical activity by van Roekel et al. will be explored in relation to cancer status using logistic regression, accounting for covariates. The student will have the option of submitting this work as a manuscript to a peer-reviewed journal.
1. Bray, F., et al., Global cancer statistics 2018: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin, 2018. 68(6): p. 394-424.
2. World Cancer Research Fund/American Institute for Cancer Research Continuous Update Project Third expertreport: Diet, nutrition, physical activity and cancer: a global perspective.
3. Matthews, C.E., et al., Amount and Intensity of Leisure-Time Physical Activity and Lower Cancer Risk. J Clin Oncol, 2020. 38(7): p. 686-697.
4. VAN Roekel, E.H., et al., Physical Activity, Television Viewing Time, and DNA Methylation in Peripheral Blood. Med Sci Sports Exerc, 2019. 51(3): p. 490-498.
- Dr Venexia Walker - Professor Simon Satchell -
Kidney function is a critical measure of health used in clinical practice and research, measured using estimated glomerular filtration rate (eGFR) based on either creatinine or cystatin C. Previous work has highlighted the complexity of using creatinine and cystatin C measurements to calculate eGFR, as they are influenced by different factors. (1) For instance, hyperthyroidism affects creatinine and cystatin C measurements in different directions. (2) Recently, there have been several papers showing that eGFR derived from cystatin-C measurements is a stronger predictor of cardiovascular disease than eGFR derived from creatinine measurements. (3,4) However, the epidemiology of this phenomenon is unclear and there are currently no proposed mechanisms that explain it. Through this project, we want to characterise the epidemiology of creatinine and cystatin C measurements so that we can better understand why eGFR derived from these different measurements may have different properties.
The aim of this project is to explore the features of creatinine and cystatin C measurements, both of which are used to calculate estimated glomerular filtration rate (eGFR).
The specific objectives are:
(1) To calculate creatinine / cystatin C ratio.
(2) To describe the characteristics of people with different creatinine / cystatin C ratios.
(3) To conduct a genome-wide association study (GWAS) of creatinine / cystatin C ratio.
(4) To perform two-sample Mendelian randomization (MR) to estimate the causal effect of creatinine / cystatin C ratio on relevant phenotypes including albuminuria, kidney disease and cardiovascular disease.
This project will provide be conducted in UK biobank so will provide the researcher with experience analysing genetic and phenotypic data from a large biobank study. (5,6) Specifically, the researcher will have the opportunity to perform data cleaning, conduct a GWAS (7), and perform two-sample MR (8–10).
1. Lees JS, Welsh CE, Celis-Morales CA, Mackay D, Lewsey J, Gray SR, et al. Glomerular filtration rate by differing measures, albuminuria and prediction of cardiovascular disease, mortality and end-stage kidney disease. Nat Med. 2019 Nov;25(11):1753–60.
2. Karawajczyk M, Ramklint M, Larsson A. Reduced cystatin C-estimated GFR and increased creatinine-estimated GFR in comparison with iohexol-estimated GFR in a hyperthyroid patient: A case report. J Med Case Reports. 2008 Feb 28;2:66.
3. Malmgren L, Öberg C, den Bakker E, Leion F, Siódmiak J, Åkesson A, et al. The complexity of kidney disease and diagnosing it – cystatin C, selective glomerular hypofiltration syndromes and proteome regulation. Journal of Internal Medicine. 2023;293(3):293–308.
4. He D, Gao B, Wang J, Yang C, Zhao MH, Zhang L. The Difference Between Cystatin C- and Creatinine-Based Estimated Glomerular Filtration Rate and Risk of Diabetic Microvascular Complications Among Adults With Diabetes: A Population-Based Cohort Study. Diabetes Care. 2024 May 1;47(5):873–80.
5. Allen NE, Sudlow C, Peakman T, Collins R, UK Biobank. UK biobank data: come and get it. Sci Transl Med. 2014 Feb 19;6(224):224ed4.
6. Bycroft C, Freeman C, Petkova D, Band G, Elliott LT, Sharp K, et al. The UK Biobank resource with deep phenotyping and genomic data. Nature. 2018 Oct;562(7726):203–9.
7. Elsworth B, Mitchell R, Raistrick C, Paternoster L, Hemani G, Gaunt T. MRC IEU UK Biobank GWAS pipeline version 2 [Internet]. 2019. Available from: https://doi.org/10.5523/bris.pnoat8cxo0u52p6ynfaekeigi
8. Davey Smith G, Ebrahim S. ‘Mendelian randomization’: can genetic epidemiology contribute to understanding environmental determinants of disease? Int J Epidemiol. 2003 Jan 2;32(1):1–22.
9. Davey Smith G, Hemani G. Mendelian randomization: genetic anchors for causal inference in epidemiological studies. Hum Mol Genet. 2014 Sep 15;23(R1):R89–98.
10. Lawlor DA. Commentary: Two-sample Mendelian randomization: opportunities and challenges. Int J Epidemiol. 2016 Jun;45(3):908–15.
- Dr Joshua Burton - Professor Tom Gaunt -
Molecular quantitative trait loci (molQTL) can provide functional evidence on the mechanisms underlying phenotype-genotype associations and are increasingly used in drug target validation and safety assessment, with protein QTLs (pQTLs) and gene expression QTLs (eQTLs) being the most commonly used. However, questions remain on how to best consolidate results from pQTLs and eQTLs for target validation.
This project will look at combining eQTL and pQTL data to form QTL pairs representing genes and their products. Enrichment analyses will then be performed to identify features of the QTL pairs that provide consistent evidence for drug targets based on the concordance of the direction of effect of both the pQTL and the eQTL.
For QTL pairs that are discordant, what is the underlying biology that can explain this effect and how does it link back to drug target validation.
The aim of this project is to evaluate the concordance rates of QTL pairs and provide a biological interpretation for the presence of discordant pairs by investigating the drug target enrichment, tissue type, and genomic region of the associated gene and product.
Extract conditionally independent pQTLs and match to eQTLs, allowing for proxy SNP search. Harmonise QTL pairs and perform a naïve sign test on both beta estimates, with Bayesian winner’s curse correction. Enrichment using Drug-Gene Interaction database and annotate genes which encode for druggable proteins.
Finan C, Gaulton A, Kruger FA, Lumbers RT, Shah T, Engmann J, Galver L, Kelley R, Karlsson A, Santos R, Overington JP, Hingorani AD, Casas JP. The druggable genome and support for target identification and validation in drug development. Sci Transl Med. 2017 Mar 29;9(383):1166.
Mizuno A, Okada Y. Biological characterization of expression quantitative trait loci (eQTLs) showing tissue-specific opposite directional effects. Eur J Hum Genet. 2019 Nov;27(11):1745-1756.
Freshour SL, Kiwala S, Cotto KC, Coffman AC, McMichael JF, Song JJ, Griffith M, Griffith OL, Wagner AH. Integration of the Drug-Gene Interaction Database (DGIdb 4.0) with open crowdsource efforts. Nucleic Acids Res. 2021 Jan 8;49(D1):D1144-D1151.
Pietzner M, Wheeler E, Carrasco-Zanini J, Cortes A, Koprulu M, Wörheide MA, Oerton E, Cook J, Stewart ID, Kerrison ND, Luan J, Raffler J, Arnold M, Arlt W, O'Rahilly S, Kastenmüller G, Gamazon ER, Hingorani AD, Scott RA, Wareham NJ, Langenberg C. Mapping the proteo-genomic convergence of human diseases. Science. 2021 Nov 12;374(6569):1541.
- Dr Anthony Manyara - Professor Dann Mitchell -
Climate change is one the leading global health threats while multimorbidity is expected to increase with increasing life expectancy. Climate change will lead to increase in both infectious and noncommunicable disease consequently increasing multimorbidity. It remains unknown on what is known or being done on these two important topics and gaps in evidence to date.
1. To describe empirical research on climate change and multimorbidity (e.g., design, setting, outcomes)
2. To appraise the quality of included research
3. To map the empirical research against the WHO global research priorities for protecting human health from climate change, https://iris.who.int/bitstream/handle/10665/44133/9789241598187_eng.pdf?sequence=1
4. To critically identify gaps for future research
Use the frameworks by Arksey and O’Malley, Levac et al and the JBI guidance to search, screen, and chart literature and then critically interpret the evidence and write up the results following the PRISMA extension for scoping reviews, https://www.prisma-statement.org/scoping