array(41) {
  ["project_title"]=>
  string(120) "Predicting response under hypothetical interventions in patients with rheumatoid arthritis: a methodological exploration"
  ["project_narrative_summary"]=>
  string(700) "In the management of rheumatoid arthritis (RA), finding the right treatment for patients follows a trial-and-error-like approach. Following UK guidelines, patients with RA are typically prescribed methotrexate (MTX) as their first line therapy, but ~40% of patients do not respond by 6 months(1). Research has focused on identifying which patients are at high-risk of not responding to MTX, but there is no guarantee that these patients will respond better to alternative treatments. We aim to explore methods that answer ?What if? questions around treatment allocation (e.g., what will happen to this patient if we prescribe a drug different to MTX?). This could benefit decision making in practice."
  ["project_learn_source"]=>
  string(9) "colleague"
  ["project_learn_source_exp"]=>
  string(0) ""
  ["project_key_personnel"]=>
  array(5) {
    [0]=>
    array(6) {
      ["p_pers_f_name"]=>
      string(4) "Glen"
      ["p_pers_l_name"]=>
      string(6) "Martin"
      ["p_pers_degree"]=>
      string(3) "PhD"
      ["p_pers_pr_affil"]=>
      string(123) "Centre for Health Informatics, Division of Informatics, Imaging and Data Sciences, University of Manchester, Manchester, UK"
      ["p_pers_scop_id"]=>
      string(11) "57192176105"
      ["requires_data_access"]=>
      string(0) ""
    }
    [1]=>
    array(6) {
      ["p_pers_f_name"]=>
      string(7) "Matthew"
      ["p_pers_l_name"]=>
      string(7) "Sperrin"
      ["p_pers_degree"]=>
      string(3) "PhD"
      ["p_pers_pr_affil"]=>
      string(123) "Centre for Health Informatics, Division of Informatics, Imaging and Data Sciences, University of Manchester, Manchester, UK"
      ["p_pers_scop_id"]=>
      string(11) "26535288300"
      ["requires_data_access"]=>
      string(0) ""
    }
    [2]=>
    array(6) {
      ["p_pers_f_name"]=>
      string(5) "Kimme"
      ["p_pers_l_name"]=>
      string(6) "Hyrich"
      ["p_pers_degree"]=>
      string(6) "MD PhD"
      ["p_pers_pr_affil"]=>
      string(176) "Centre for Epidemiology Versus Arthritis, Centre for Musculoskeletal Research, Division of Musculoskeletal and Dermatological Sciences, University of Manchester, Manchester, UK"
      ["p_pers_scop_id"]=>
      string(10) "7801500474"
      ["requires_data_access"]=>
      string(0) ""
    }
    [3]=>
    array(6) {
      ["p_pers_f_name"]=>
      string(7) "Suzanne"
      ["p_pers_l_name"]=>
      string(10) "Verstappen"
      ["p_pers_degree"]=>
      string(3) "PhD"
      ["p_pers_pr_affil"]=>
      string(176) "Centre for Epidemiology Versus Arthritis, Centre for Musculoskeletal Research, Division of Musculoskeletal and Dermatological Sciences, University of Manchester, Manchester, UK"
      ["p_pers_scop_id"]=>
      string(10) "6603191436"
      ["requires_data_access"]=>
      string(0) ""
    }
    [4]=>
    array(6) {
      ["p_pers_f_name"]=>
      string(5) "Jamie"
      ["p_pers_l_name"]=>
      string(8) "Sergeant"
      ["p_pers_degree"]=>
      string(3) "PhD"
      ["p_pers_pr_affil"]=>
      string(109) "Centre for Biostatistics, Manchester Academic Health Science Centre, University of Manchester, Manchester, UK"
      ["p_pers_scop_id"]=>
      string(11) "38562164200"
      ["requires_data_access"]=>
      string(0) ""
    }
  }
  ["project_ext_grants"]=>
  array(2) {
    ["value"]=>
    string(65) "External grants or funds are being used to support this research."
    ["label"]=>
    string(65) "External grants or funds are being used to support this research."
  }
  ["project_funding_source"]=>
  string(37) "Versus Arthritis (grant number 21755)"
  ["project_assoc_trials"]=>
  array(1) {
    [0]=>
    object(WP_Post)#3826 (24) {
      ["ID"]=>
      int(1295)
      ["post_author"]=>
      string(4) "1363"
      ["post_date"]=>
      string(19) "2023-08-05 04:46:07"
      ["post_date_gmt"]=>
      string(19) "2023-08-05 04:46:07"
      ["post_content"]=>
      string(0) ""
      ["post_title"]=>
      string(230) "NCT00236028 - A Randomized, Double-blind, Trial of Anti-TNFa Chimeric Monoclonal Antibody (Infliximab) in Combination With Methotrexate Compared With Methotrexate Alone for the Treatment of Patients With Early Rheumatoid Arthritis"
      ["post_excerpt"]=>
      string(0) ""
      ["post_status"]=>
      string(7) "publish"
      ["comment_status"]=>
      string(4) "open"
      ["ping_status"]=>
      string(4) "open"
      ["post_password"]=>
      string(0) ""
      ["post_name"]=>
      string(194) "nct00236028-a-randomized-double-blind-trial-of-anti-tnfa-chimeric-monoclonal-antibody-infliximab-in-combination-with-methotrexate-compared-with-methotrexate-alone-for-the-treatment-of-patients-w"
      ["to_ping"]=>
      string(0) ""
      ["pinged"]=>
      string(0) ""
      ["post_modified"]=>
      string(19) "2024-06-20 15:37:56"
      ["post_modified_gmt"]=>
      string(19) "2024-06-20 19:37:56"
      ["post_content_filtered"]=>
      string(0) ""
      ["post_parent"]=>
      int(0)
      ["guid"]=>
      string(243) "https://dev-yoda.pantheonsite.io/clinical-trial/nct00236028-a-randomized-double-blind-trial-of-anti-tnfa-chimeric-monoclonal-antibody-infliximab-in-combination-with-methotrexate-compared-with-methotrexate-alone-for-the-treatment-of-patients-w/"
      ["menu_order"]=>
      int(0)
      ["post_type"]=>
      string(14) "clinical_trial"
      ["post_mime_type"]=>
      string(0) ""
      ["comment_count"]=>
      string(1) "0"
      ["filter"]=>
      string(3) "raw"
    }
  }
  ["project_date_type"]=>
  string(91) "Individual Participant-Level Data, which includes Full CSR and all supporting documentation"
  ["property_scientific_abstract"]=>
  string(2571) "Background: Methotrexate (MTX) is the recommended first line therapy for patients with rheumatoid arthritis (RA). However, response to MTX varies, with around 40% of patients not responding by six months in terms of the EULAR (European Alliance of Associations for Rheumatology) response criteria. Clinical prediction models (CPMs) have previously been developed to identify patients at high-risk of non-response, with the goal to enable earlier prescription of biologic therapies. A limitation of these CPMs is that they cannot quantify an individual?s risk of non-response or adverse outcomes if they were given an alternative treatment, because existing CPMs do not have a causal structure to them.
In current UK clinical practice, which is captured in observational data, patients are required to have failed MTX prior to the prescription of a biologic. This group of patients will be referred to as DMARD (disease-modifying anti-rheumatic drug)-inadequate responders (DIR). Conversely, randomised controlled trials (RCTs) of treatments for RA often enrol patients that are DMARD-nave (DN), which is our target population where a causal CPM would be applied.
Objective: The aim of this study is to explore methods that enable risk predictions under hypothetical interventions in patients with RA. It will comprise two main Objectives: 1) developing a CPM for response to infliximab (INF) vs MTX using real-world data on DIR patients. As this data does not precisely capture the target population, estimates of individual patient response to INF vs MTX may therefore be biased. To assess this, the CPM will be 2) validated in the YODA RCT data of DN patients.
Study Design: The YODA data will be used for methodological exploration of a causal prediction approach, rather than proposing a model for clinical use per se. The study will focus on model validation, which aims to determine how transferrable the predictions obtained in DIR patients are to a DN population, and therefore, whether causal prediction for RA using observational data is reasonable. For the purpose of our methodological research, we may also explore developing a CPM in DN patients and validating in DIR patients.
Participants: Participants of interest are DN adults (?18 years) with a diagnosis of RA for ?3 months and ?3 years at time of screening and pre-infusion with the drug or control.
Primary Outcome Measure: The primary outcome measure is response defined by achieving remission (binary) at 6 and 12 months (a 28-joint disease activity score (DAS28)" ["project_brief_bg"]=> string(3184) "Rheumatoid arthritis (RA) is a chronic, heterogenous condition that affects around 0.46% of the global population (3). Early diagnosis and treatment of RA is vital in controlling disease activity and preventing permanent joint damage (4). As recommended by major guidelines including EULAR (European Alliance of Associations for Rheumatology) (2), ACR (American College for Rheumatology)(5) and NICE (National Institute for Health and Care Excellence)(6), methotrexate (MTX) is commonly prescribed as first line therapy for patients with RA. However, around 40% of patients do not respond to MTX by six months (1). Clinical prediction models (CPMs) are statistical tools/algorithms that can estimate the probability of an individual either having (diagnostic models) or developing in the future (prognostic models) a particular outcome (7). Previously developed and validated CPMs of MTX therapy outcomes in RA are at high risk of bias due to methodological shortcomings, which may be hindering their impact on clinical practice (8). A further limitation of these models is that, although they could potentially identify patients at high-risk of non-response to MTX, there is no indication of which alternative treatments may be better suited to an individual (9).
To influence guidelines and clinical practice, we need to strengthen the evidence base around risk prediction under hypothetical interventions. Due to a lack of randomised treatment assignment in available observational data, it is challenging to correctly estimate hypothetical risks of different treatment options (9,10). Conversely, randomised data such as from clinical trials typically only capture a small sample of a population, often with strict inclusion criteria and treatment protocols, and therefore does not fully represent the wider patient population. This is suboptimal when developing a CPM. Therefore, causal inference methods for observational data are increasingly being applied (9,11,12). However, as per current UK clinical guidelines (6), patients starting a biologic DMARD (bDMARD) must have previously failed a conventional DMARD, such as MTX. This poses a challenge when using observational data to develop a CPM that estimates a patient?s risk if they were given a drug different to MTX, as the population of interest for the application of such a model is treatment nave, not those who have already failed MTX.
To address this, we propose developing a CPM for INF vs MTX response in observational data (where INF patients are DIRs), for which a large sample size is available, and validating this model in the target population of DN patients, for which we intend to use the YODA data. For our proof-of-concept study, only INF vs MTX will be used, which in future, could be extended to more biologics, biosimilars, combination therapies, and variability in dosage. This will contribute a novel methodological approach to predicting treatment outcomes in the field of rheumatology, as previous CPMs of MTX outcomes did not consider hypothetical interventions. This has the potential to impact clinical decision making and providing more personalised care for patients with RA." ["project_specific_aims"]=> string(735) "The specific aim of this project is to explore causal prediction methods that enable risk predictions under hypothetical interventions in patients with RA. It will consist of two main stages: 1) developing a causal CPM for predicting INF vs MTX response using real-world data in which INF patients are DIRs, followed by 2) validating this model in DN-INF patients using data from the early RA INF trial on the YODA project. This will evaluate whether developing a CPM in a population that previously failed MTX, with likely higher grades of joint erosions and disability, is transferrable to the target population of treatment nave RA patients. We are applying for access to YODA data to enable us to undertake Stage 2 of the project." ["project_study_design"]=> array(2) { ["value"]=> string(8) "meth_res" ["label"]=> string(23) "Methodological research" } ["project_study_design_exp"]=> string(0) "" ["project_purposes"]=> array(1) { [0]=> array(2) { ["value"]=> string(50) "Research on clinical prediction or risk prediction" ["label"]=> string(50) "Research on clinical prediction or risk prediction" } } ["project_purposes_exp"]=> string(0) "" ["project_software_used"]=> array(2) { ["value"]=> string(7) "RStudio" ["label"]=> string(7) "RStudio" } ["project_software_used_exp"]=> string(0) "" ["project_research_methods"]=> string(420) "The inclusion criteria for our study are adults (?18 years) with a diagnosis of RA for ?3 months and ?3 years before screening and pre-infusion of the drug. Patients in our study sample will receive MTX (start at 7.5mg/week, increased to 20mg/week by week 8) and the study medication (3mg/kg INF or placebo). We will exclude patients taking 6mg/kg INF as by current UK guidelines, 3mg/kg is prescribed in adults with RA." ["project_main_outcome_measure"]=> string(73) "The primary outcome measure will be remission defined using the DAS28-CRP" ["project_main_predictor_indep"]=> string(509) "The predictor variables in the CPM are: age (years), sex, BMI (kg/m2), DAS28, Health Assessment Questionnaire (HAQ), rheumatoid factor (RF), symptom duration (months). The DAS28 will be based on four components: TJC28, SJC28, PGA, CRP (see section above for formula). The HAQ score, based on a patient reported questionnaire using the Disability Index, ranges from 0-3. RF status is a binary variable and positivity is defined by RF values >14 IU/mL. BMI is calculated using the height and weight of patients." ["project_other_variables_interest"]=> string(0) "" ["project_stat_analysis_plan"]=> string(2004) "The requested clinical trial data will be summarised using descriptive statistics and will be used for the purposes of validating a CPM. We will follow the TRIPOD guidelines for best practice of model development, validation, and reporting (16,17).
A baseline table of patient characteristics will be produced, per treatment arm (INF + MTX vs MTX). We will summarise continuous variables using the median and interquartile range, and categorical variables as frequencies of occurrence. This summary will capture information on demographic, clinical, and treatment related variables, such as age, sex, swollen and tender joint count, RF, CRP, disease duration, and a disability score quantified using the HAQ.
The CPM to be validated is based on a logistic regression. The model?s intercept and regression coefficients will be used to obtain the linear predictor in the YODA data. This linear predictor will then be used to obtain individual patient predictions.
The CPM?s predictive performance will be quantified using calibration, which is the agreement between predicted and observed risks, as well as discrimination, the model?s ability to distinguish between patients that experience the outcome from those that do not. The total variance explained in the model will be assessed using the Nagelkerke R2. We will also assess net benefit for the treatment decision made using the CPM. This will involve calculating net benefit given specific treat/not treat thresholds and plotting this on a decision curve (13?15).
If our findings suggest that the individual predictions of response obtained in the real-world data are not well calibrated with the target population, one option is to use the treatment effect from the INF trial and insert this into the CPM using the offset method (18). There are various ways to estimate the average treatment effect in a clinical trial (all producing very similar results), such as using contingency tables or regression techniques (19,20)." ["project_timeline"]=> string(238) "Anticipated project start date: February 2022
Analysis completion date: June 2023
Date manuscript drafted and first submitted for publication: September 2023
Date results reported back to the YODA project: September 2023" ["project_dissemination_plan"]=> string(493) "We plan to submit an abstract of this work to the International Society for Clinical Biostatistics 2023. Due to the methodological nature of this work, we anticipate submitting a manuscript to a methodology-focused peer-reviewed scientific journal, for example, the Journal of Diagnostic and Prognostic Research, Journal of Clinical Epidemiology, or BMC Medical Research Methodology. The work will also appear in the University of Manchester PhD thesis of the lead applicant, Celina Gehringer." ["project_bibliography"]=> string(4184) "

1. Sergeant JC, Hyrich KL, Anderson J, Kopec-Harding K, Hope HF, Symmons DPM, et al. Prediction of primary non-response to methotrexate therapy using demographic, clinical and psychosocial variables: Results from the UK Rheumatoid Arthritis Medication Study (RAMS). Arthritis Res Ther. 2018;20(1):1?11.
2. Smolen JS, Landew RBM, Bijlsma JWJ, Burmester GR, Dougados M, Kerschbaumer A, et al. EULAR recommendations for the management of rheumatoid arthritis with synthetic and biological disease-modifying antirheumatic drugs: 2019 update. Ann Rheum Dis. 2020;79(6):S685?99.
3. Almutairi K, Nossent J, Preen D, Keen H, Inderjeeth C. The global prevalence of rheumatoid arthritis: a meta-analysis based on a systematic review. Rheumatol Int. 2021 May;41(5):863?77.
4. Heidari B. Rheumatoid Arthritis: Early diagnosis and treatment outcomes. Casp J Intern Med. 2011;2(1):161?70.
5. Fraenkel L, Bathon JM, England BR, St.Clair EW, Arayssi T, Carandang K, et al. 2021 American College of Rheumatology Guideline for the Treatment of Rheumatoid Arthritis. Arthritis Care Res. 2021 Jul;73(7):924?39.
6. Recommendations | Rheumatoid arthritis in adults: management | Guidance | NICE [Internet]. NICE; [cited 2022 Mar 2]. Available from: https://www.nice.org.uk/guidance/ng100/chapter/Recommendations
7. Steyerberg EW, Moons KGM, van der Windt DA, Hayden JA, Perel P, Schroter S, et al. Prognosis Research Strategy (PROGRESS) 3: Prognostic Model Research. PLoS Med. 2013 Feb 5;10(2):e1001381.
8. Gehringer CK, Martin GP, Hyrich KL, Verstappen SMM, Sergeant JC. Clinical prediction models for methotrexate treatment outcomes in patients with rheumatoid arthritis: A systematic review and meta-analysis. Semin Arthritis Rheum. 2022 Oct 1;56:152076.
9. Lin L, Sperrin M, Jenkins DA, Martin GP, Peek N. A scoping review of causal methods enabling predictions under hypothetical interventions. Diagn Progn Res. 2021;5(1).
10. Dickerman BA, Hernn MA. Counterfactual prediction is not only for causal inference. Eur J Epidemiol. 2020 Jul 1;35(7):615?7.
11. Sperrin M, Martin GP, Pate A, Van Staa T, Peek N, Buchan I. Using marginal structural models to adjust for treatment drop-in when developing clinical prediction models. Stat Med. 2018 Dec 10;37(28):4142?54.
12. Lin L, Poppe K, Wood A, Martin G, Peek N, Sperrin M. Making predictions under hypothetical interventions: a case study from the PREDICT-CVD cohort in New Zealand primary care [Internet]. In Review; 2022 Sep [cited 2022 Nov 23]. Available from: https://www.researchsquare.com/article/rs-1824359/v2
13. Vickers AJ, Kattan MW, Sargent D. Method for evaluating prediction models that apply the results of randomized trials to individual patients. Trials. 2007 Jun 5;8:14.
14. Vickers AJ, Calster BV, Steyerberg EW. Net benefit approaches to the evaluation of prediction models, molecular markers, and diagnostic tests. BMJ. 2016 Jan 25;352:i6.
15. Chalkou K, Vickers AJ, Pellegrini F, Manca A. Decision curve analysis for personalized treatment choice between multiple options. :24.
16. Collins GS, Reitsma JB, Altman DG, Moons KGM. Transparent reporting of a multivariable prediction model for individual prognosis or diagnosis (TRIPOD): The TRIPOD Statement. BMC Med. 2015;13(1):1?10.
17. Moons KGM, Altman DG, Reitsma JB, Ioannidis JPA, Macaskill P, Steyerberg EW, et al. Transparent Reporting of a multivariable prediction model for Individual Prognosis Or Diagnosis (TRIPOD): Explanation and Elaboration. Ann Intern Med. 2015 Jan 6;162(1):W1?73.
18. van Amsterdam WAC, Ranganath R. Conditional average treatment effect estimation with treatment offset models [Internet]. arXiv; 2022 [cited 2022 Dec 22]. Available from: http://arxiv.org/abs/2204.13975
19. Rombach I, Knight R, Peckham N, Stokes JR, Cook JA. Current practice in analysing and reporting binary outcome data?a review of randomised controlled trial reports. BMC Med. 2020 Jun 8;18(1):147.
20. J T, L B, T H, J R, M W, M H. Different ways to estimate treatment effects in randomised controlled trials. Contemp Clin Trials Commun. 2018 Jun 1;10:80?5.

" ["project_suppl_material"]=> bool(false) ["project_coi"]=> array(6) { [0]=> array(1) { ["file_coi"]=> array(21) { ["ID"]=> int(8971) ["id"]=> int(8971) ["title"]=> string(6) "coi_kh" ["filename"]=> string(10) "coi_kh.pdf" ["filesize"]=> int(29606) ["url"]=> string(59) "https://yoda.yale.edu/wp-content/uploads/2018/01/coi_kh.pdf" ["link"]=> string(52) "https://yoda.yale.edu/data-request/2022-5056/coi_kh/" ["alt"]=> string(0) "" ["author"]=> string(4) "1363" ["description"]=> string(0) "" ["caption"]=> string(0) "" ["name"]=> string(6) "coi_kh" ["status"]=> string(7) "inherit" ["uploaded_to"]=> int(6134) ["date"]=> string(19) "2023-07-31 15:19:28" ["modified"]=> string(19) "2023-08-01 00:55:56" ["menu_order"]=> int(0) ["mime_type"]=> string(15) "application/pdf" ["type"]=> string(11) "application" ["subtype"]=> string(3) "pdf" ["icon"]=> string(62) "https://yoda.yale.edu/wp/wp-includes/images/media/document.png" } } [1]=> array(1) { ["file_coi"]=> array(21) { ["ID"]=> int(8965) ["id"]=> int(8965) ["title"]=> string(6) "coi_js" ["filename"]=> string(10) "coi_js.pdf" ["filesize"]=> int(20777) ["url"]=> string(59) "https://yoda.yale.edu/wp-content/uploads/2019/08/coi_js.pdf" ["link"]=> string(52) "https://yoda.yale.edu/data-request/2022-5056/coi_js/" ["alt"]=> string(0) "" ["author"]=> string(4) "1363" ["description"]=> string(0) "" ["caption"]=> string(0) "" ["name"]=> string(6) "coi_js" ["status"]=> string(7) "inherit" ["uploaded_to"]=> int(6134) ["date"]=> string(19) "2023-07-31 15:19:10" ["modified"]=> string(19) "2023-08-01 00:55:56" ["menu_order"]=> int(0) ["mime_type"]=> string(15) "application/pdf" ["type"]=> string(11) "application" ["subtype"]=> string(3) "pdf" ["icon"]=> string(62) "https://yoda.yale.edu/wp/wp-includes/images/media/document.png" } } [2]=> array(1) { ["file_coi"]=> array(21) { ["ID"]=> int(8938) ["id"]=> int(8938) ["title"]=> string(6) "coi_gm" ["filename"]=> string(10) "coi_gm.pdf" ["filesize"]=> int(20441) ["url"]=> string(59) "https://yoda.yale.edu/wp-content/uploads/2016/12/coi_gm.pdf" ["link"]=> string(52) "https://yoda.yale.edu/data-request/2022-5056/coi_gm/" ["alt"]=> string(0) "" ["author"]=> string(4) "1363" ["description"]=> string(0) "" ["caption"]=> string(0) "" ["name"]=> string(6) "coi_gm" ["status"]=> string(7) "inherit" ["uploaded_to"]=> int(6134) ["date"]=> string(19) "2023-07-31 15:17:35" ["modified"]=> string(19) "2023-08-01 00:55:56" ["menu_order"]=> int(0) ["mime_type"]=> string(15) "application/pdf" ["type"]=> string(11) "application" ["subtype"]=> string(3) "pdf" ["icon"]=> string(62) "https://yoda.yale.edu/wp/wp-includes/images/media/document.png" } } [3]=> array(1) { ["file_coi"]=> array(21) { ["ID"]=> int(9003) ["id"]=> int(9003) ["title"]=> string(6) "coi_ms" ["filename"]=> string(10) "coi_ms.pdf" ["filesize"]=> int(19871) ["url"]=> string(59) "https://yoda.yale.edu/wp-content/uploads/2019/12/coi_ms.pdf" ["link"]=> string(52) "https://yoda.yale.edu/data-request/2022-5056/coi_ms/" ["alt"]=> string(0) "" ["author"]=> string(4) "1363" ["description"]=> string(0) "" ["caption"]=> string(0) "" ["name"]=> string(6) "coi_ms" ["status"]=> string(7) "inherit" ["uploaded_to"]=> int(6134) ["date"]=> string(19) "2023-07-31 15:21:11" ["modified"]=> string(19) "2023-08-01 00:55:56" ["menu_order"]=> int(0) ["mime_type"]=> string(15) "application/pdf" ["type"]=> string(11) "application" ["subtype"]=> string(3) "pdf" ["icon"]=> string(62) "https://yoda.yale.edu/wp/wp-includes/images/media/document.png" } } [4]=> array(1) { ["file_coi"]=> array(21) { ["ID"]=> int(9049) ["id"]=> int(9049) ["title"]=> string(6) "coi_sv" ["filename"]=> string(10) "coi_sv.pdf" ["filesize"]=> int(20093) ["url"]=> string(59) "https://yoda.yale.edu/wp-content/uploads/2018/04/coi_sv.pdf" ["link"]=> string(52) "https://yoda.yale.edu/data-request/2022-5056/coi_sv/" ["alt"]=> string(0) "" ["author"]=> string(4) "1363" ["description"]=> string(0) "" ["caption"]=> string(0) "" ["name"]=> string(6) "coi_sv" ["status"]=> string(7) "inherit" ["uploaded_to"]=> int(6134) ["date"]=> string(19) "2023-07-31 15:23:34" ["modified"]=> string(19) "2023-08-01 00:55:56" ["menu_order"]=> int(0) ["mime_type"]=> string(15) "application/pdf" ["type"]=> string(11) "application" ["subtype"]=> string(3) "pdf" ["icon"]=> string(62) "https://yoda.yale.edu/wp/wp-includes/images/media/document.png" } } [5]=> array(1) { ["file_coi"]=> array(21) { ["ID"]=> int(8741) ["id"]=> int(8741) ["title"]=> string(6) "coi_cg" ["filename"]=> string(10) "coi_cg.pdf" ["filesize"]=> int(21073) ["url"]=> string(59) "https://yoda.yale.edu/wp-content/uploads/2019/04/coi_cg.pdf" ["link"]=> string(52) "https://yoda.yale.edu/data-request/2022-5056/coi_cg/" ["alt"]=> string(0) "" ["author"]=> string(4) "1363" ["description"]=> string(0) "" ["caption"]=> string(0) "" ["name"]=> string(6) "coi_cg" ["status"]=> string(7) "inherit" ["uploaded_to"]=> int(6134) ["date"]=> string(19) "2023-07-31 15:08:03" ["modified"]=> string(19) "2023-08-01 00:55:56" ["menu_order"]=> int(0) ["mime_type"]=> string(15) "application/pdf" ["type"]=> string(11) "application" ["subtype"]=> string(3) "pdf" ["icon"]=> string(62) "https://yoda.yale.edu/wp/wp-includes/images/media/document.png" } } } ["data_use_agreement_training"]=> bool(true) ["certification"]=> bool(true) ["project_send_email_updates"]=> bool(true) ["project_status"]=> string(7) "ongoing" ["project_publ_available"]=> bool(true) ["project_year_access"]=> string(4) "2023" ["project_rep_publ"]=> array(1) { [0]=> array(1) { ["publication_link"]=> string(0) "" } } ["project_assoc_data"]=> array(0) { } ["project_due_dil_assessment"]=> array(21) { ["ID"]=> int(10401) ["id"]=> int(10401) ["title"]=> string(47) "yoda_project_due_diligence_assessment_2022-5056" ["filename"]=> string(51) "yoda_project_due_diligence_assessment_2022-5056.pdf" ["filesize"]=> int(89321) ["url"]=> string(100) "https://yoda.yale.edu/wp-content/uploads/2023/08/yoda_project_due_diligence_assessment_2022-5056.pdf" ["link"]=> string(93) "https://yoda.yale.edu/data-request/2022-5056/yoda_project_due_diligence_assessment_2022-5056/" ["alt"]=> string(0) "" ["author"]=> string(4) "1363" ["description"]=> string(0) "" ["caption"]=> string(0) "" ["name"]=> string(47) "yoda_project_due_diligence_assessment_2022-5056" ["status"]=> string(7) "inherit" ["uploaded_to"]=> int(6134) ["date"]=> string(19) "2023-08-09 16:54:50" ["modified"]=> string(19) "2023-08-09 19:11:18" ["menu_order"]=> int(0) ["mime_type"]=> string(15) "application/pdf" ["type"]=> string(11) "application" ["subtype"]=> string(3) "pdf" ["icon"]=> string(62) "https://yoda.yale.edu/wp/wp-includes/images/media/document.png" } ["project_title_link"]=> array(21) { ["ID"]=> int(10403) ["id"]=> int(10403) ["title"]=> string(44) "yoda_project_protocol_-_2022-5056_-_23-01-23" ["filename"]=> string(48) "yoda_project_protocol_-_2022-5056_-_23-01-23.pdf" ["filesize"]=> int(31720) ["url"]=> string(97) "https://yoda.yale.edu/wp-content/uploads/2023/08/yoda_project_protocol_-_2022-5056_-_23-01-23.pdf" ["link"]=> string(90) "https://yoda.yale.edu/data-request/2022-5056/yoda_project_protocol_-_2022-5056_-_23-01-23/" ["alt"]=> string(0) "" ["author"]=> string(4) "1363" ["description"]=> string(0) "" ["caption"]=> string(0) "" ["name"]=> string(44) "yoda_project_protocol_-_2022-5056_-_23-01-23" ["status"]=> string(7) "inherit" ["uploaded_to"]=> int(6134) ["date"]=> string(19) "2023-08-09 16:54:56" ["modified"]=> string(19) "2023-08-09 19:11:18" ["menu_order"]=> int(0) ["mime_type"]=> string(15) "application/pdf" ["type"]=> string(11) "application" ["subtype"]=> string(3) "pdf" ["icon"]=> string(62) "https://yoda.yale.edu/wp/wp-includes/images/media/document.png" } ["project_review_link"]=> array(21) { ["ID"]=> int(10397) ["id"]=> int(10397) ["title"]=> string(36) "yoda_project_review_-_2022-5056_site" ["filename"]=> string(40) "yoda_project_review_-_2022-5056_site.pdf" ["filesize"]=> int(1315704) ["url"]=> string(89) "https://yoda.yale.edu/wp-content/uploads/2023/08/yoda_project_review_-_2022-5056_site.pdf" ["link"]=> string(82) "https://yoda.yale.edu/data-request/2022-5056/yoda_project_review_-_2022-5056_site/" ["alt"]=> string(0) "" ["author"]=> string(4) "1363" ["description"]=> string(0) "" ["caption"]=> string(0) "" ["name"]=> string(36) "yoda_project_review_-_2022-5056_site" ["status"]=> string(7) "inherit" ["uploaded_to"]=> int(6134) ["date"]=> string(19) "2023-08-09 16:54:41" ["modified"]=> string(19) "2023-08-09 19:11:18" ["menu_order"]=> int(0) ["mime_type"]=> string(15) "application/pdf" ["type"]=> string(11) "application" ["subtype"]=> string(3) "pdf" ["icon"]=> string(62) "https://yoda.yale.edu/wp/wp-includes/images/media/document.png" } ["project_highlight_button"]=> string(0) "" ["search_order"]=> string(5) "-8400" } data partner
array(1) { [0]=> string(15) "johnson-johnson" }

pi country
array(1) { [0]=> string(14) "United Kingdom" }

pi affil
array(1) { [0]=> string(8) "Academia" }

products
array(1) { [0]=> string(8) "remicade" }

num of trials
array(1) { [0]=> string(1) "1" }

res
array(1) { [0]=> string(1) "3" }

2022-5056

General Information

How did you learn about the YODA Project?: Colleague

Conflict of Interest

Request Clinical Trials

Associated Trial(s):
  1. NCT00236028 - A Randomized, Double-blind, Trial of Anti-TNFa Chimeric Monoclonal Antibody (Infliximab) in Combination With Methotrexate Compared With Methotrexate Alone for the Treatment of Patients With Early Rheumatoid Arthritis
What type of data are you looking for?:

Request Clinical Trials

Data Request Status

Status: Ongoing

Research Proposal

Project Title: Predicting response under hypothetical interventions in patients with rheumatoid arthritis: a methodological exploration

Scientific Abstract: Background: Methotrexate (MTX) is the recommended first line therapy for patients with rheumatoid arthritis (RA). However, response to MTX varies, with around 40% of patients not responding by six months in terms of the EULAR (European Alliance of Associations for Rheumatology) response criteria. Clinical prediction models (CPMs) have previously been developed to identify patients at high-risk of non-response, with the goal to enable earlier prescription of biologic therapies. A limitation of these CPMs is that they cannot quantify an individual?s risk of non-response or adverse outcomes if they were given an alternative treatment, because existing CPMs do not have a causal structure to them.
In current UK clinical practice, which is captured in observational data, patients are required to have failed MTX prior to the prescription of a biologic. This group of patients will be referred to as DMARD (disease-modifying anti-rheumatic drug)-inadequate responders (DIR). Conversely, randomised controlled trials (RCTs) of treatments for RA often enrol patients that are DMARD-nave (DN), which is our target population where a causal CPM would be applied.
Objective: The aim of this study is to explore methods that enable risk predictions under hypothetical interventions in patients with RA. It will comprise two main Objectives: 1) developing a CPM for response to infliximab (INF) vs MTX using real-world data on DIR patients. As this data does not precisely capture the target population, estimates of individual patient response to INF vs MTX may therefore be biased. To assess this, the CPM will be 2) validated in the YODA RCT data of DN patients.
Study Design: The YODA data will be used for methodological exploration of a causal prediction approach, rather than proposing a model for clinical use per se. The study will focus on model validation, which aims to determine how transferrable the predictions obtained in DIR patients are to a DN population, and therefore, whether causal prediction for RA using observational data is reasonable. For the purpose of our methodological research, we may also explore developing a CPM in DN patients and validating in DIR patients.
Participants: Participants of interest are DN adults (?18 years) with a diagnosis of RA for ?3 months and ?3 years at time of screening and pre-infusion with the drug or control.
Primary Outcome Measure: The primary outcome measure is response defined by achieving remission (binary) at 6 and 12 months (a 28-joint disease activity score (DAS28)

Brief Project Background and Statement of Project Significance: Rheumatoid arthritis (RA) is a chronic, heterogenous condition that affects around 0.46% of the global population (3). Early diagnosis and treatment of RA is vital in controlling disease activity and preventing permanent joint damage (4). As recommended by major guidelines including EULAR (European Alliance of Associations for Rheumatology) (2), ACR (American College for Rheumatology)(5) and NICE (National Institute for Health and Care Excellence)(6), methotrexate (MTX) is commonly prescribed as first line therapy for patients with RA. However, around 40% of patients do not respond to MTX by six months (1). Clinical prediction models (CPMs) are statistical tools/algorithms that can estimate the probability of an individual either having (diagnostic models) or developing in the future (prognostic models) a particular outcome (7). Previously developed and validated CPMs of MTX therapy outcomes in RA are at high risk of bias due to methodological shortcomings, which may be hindering their impact on clinical practice (8). A further limitation of these models is that, although they could potentially identify patients at high-risk of non-response to MTX, there is no indication of which alternative treatments may be better suited to an individual (9).
To influence guidelines and clinical practice, we need to strengthen the evidence base around risk prediction under hypothetical interventions. Due to a lack of randomised treatment assignment in available observational data, it is challenging to correctly estimate hypothetical risks of different treatment options (9,10). Conversely, randomised data such as from clinical trials typically only capture a small sample of a population, often with strict inclusion criteria and treatment protocols, and therefore does not fully represent the wider patient population. This is suboptimal when developing a CPM. Therefore, causal inference methods for observational data are increasingly being applied (9,11,12). However, as per current UK clinical guidelines (6), patients starting a biologic DMARD (bDMARD) must have previously failed a conventional DMARD, such as MTX. This poses a challenge when using observational data to develop a CPM that estimates a patient?s risk if they were given a drug different to MTX, as the population of interest for the application of such a model is treatment nave, not those who have already failed MTX.
To address this, we propose developing a CPM for INF vs MTX response in observational data (where INF patients are DIRs), for which a large sample size is available, and validating this model in the target population of DN patients, for which we intend to use the YODA data. For our proof-of-concept study, only INF vs MTX will be used, which in future, could be extended to more biologics, biosimilars, combination therapies, and variability in dosage. This will contribute a novel methodological approach to predicting treatment outcomes in the field of rheumatology, as previous CPMs of MTX outcomes did not consider hypothetical interventions. This has the potential to impact clinical decision making and providing more personalised care for patients with RA.

Specific Aims of the Project: The specific aim of this project is to explore causal prediction methods that enable risk predictions under hypothetical interventions in patients with RA. It will consist of two main stages: 1) developing a causal CPM for predicting INF vs MTX response using real-world data in which INF patients are DIRs, followed by 2) validating this model in DN-INF patients using data from the early RA INF trial on the YODA project. This will evaluate whether developing a CPM in a population that previously failed MTX, with likely higher grades of joint erosions and disability, is transferrable to the target population of treatment nave RA patients. We are applying for access to YODA data to enable us to undertake Stage 2 of the project.

Study Design: Methodological research

What is the purpose of the analysis being proposed? Please select all that apply.: Research on clinical prediction or risk prediction

Software Used: RStudio

Data Source and Inclusion/Exclusion Criteria to be used to define the patient sample for your study: The inclusion criteria for our study are adults (?18 years) with a diagnosis of RA for ?3 months and ?3 years before screening and pre-infusion of the drug. Patients in our study sample will receive MTX (start at 7.5mg/week, increased to 20mg/week by week 8) and the study medication (3mg/kg INF or placebo). We will exclude patients taking 6mg/kg INF as by current UK guidelines, 3mg/kg is prescribed in adults with RA.

Primary and Secondary Outcome Measure(s) and how they will be categorized/defined for your study: The primary outcome measure will be remission defined using the DAS28-CRP

Main Predictor/Independent Variable and how it will be categorized/defined for your study: The predictor variables in the CPM are: age (years), sex, BMI (kg/m2), DAS28, Health Assessment Questionnaire (HAQ), rheumatoid factor (RF), symptom duration (months). The DAS28 will be based on four components: TJC28, SJC28, PGA, CRP (see section above for formula). The HAQ score, based on a patient reported questionnaire using the Disability Index, ranges from 0-3. RF status is a binary variable and positivity is defined by RF values >14 IU/mL. BMI is calculated using the height and weight of patients.

Other Variables of Interest that will be used in your analysis and how they will be categorized/defined for your study:

Statistical Analysis Plan: The requested clinical trial data will be summarised using descriptive statistics and will be used for the purposes of validating a CPM. We will follow the TRIPOD guidelines for best practice of model development, validation, and reporting (16,17).
A baseline table of patient characteristics will be produced, per treatment arm (INF + MTX vs MTX). We will summarise continuous variables using the median and interquartile range, and categorical variables as frequencies of occurrence. This summary will capture information on demographic, clinical, and treatment related variables, such as age, sex, swollen and tender joint count, RF, CRP, disease duration, and a disability score quantified using the HAQ.
The CPM to be validated is based on a logistic regression. The model?s intercept and regression coefficients will be used to obtain the linear predictor in the YODA data. This linear predictor will then be used to obtain individual patient predictions.
The CPM?s predictive performance will be quantified using calibration, which is the agreement between predicted and observed risks, as well as discrimination, the model?s ability to distinguish between patients that experience the outcome from those that do not. The total variance explained in the model will be assessed using the Nagelkerke R2. We will also assess net benefit for the treatment decision made using the CPM. This will involve calculating net benefit given specific treat/not treat thresholds and plotting this on a decision curve (13?15).
If our findings suggest that the individual predictions of response obtained in the real-world data are not well calibrated with the target population, one option is to use the treatment effect from the INF trial and insert this into the CPM using the offset method (18). There are various ways to estimate the average treatment effect in a clinical trial (all producing very similar results), such as using contingency tables or regression techniques (19,20).

Narrative Summary: In the management of rheumatoid arthritis (RA), finding the right treatment for patients follows a trial-and-error-like approach. Following UK guidelines, patients with RA are typically prescribed methotrexate (MTX) as their first line therapy, but ~40% of patients do not respond by 6 months(1). Research has focused on identifying which patients are at high-risk of not responding to MTX, but there is no guarantee that these patients will respond better to alternative treatments. We aim to explore methods that answer ?What if? questions around treatment allocation (e.g., what will happen to this patient if we prescribe a drug different to MTX?). This could benefit decision making in practice.

Project Timeline: Anticipated project start date: February 2022
Analysis completion date: June 2023
Date manuscript drafted and first submitted for publication: September 2023
Date results reported back to the YODA project: September 2023

Dissemination Plan: We plan to submit an abstract of this work to the International Society for Clinical Biostatistics 2023. Due to the methodological nature of this work, we anticipate submitting a manuscript to a methodology-focused peer-reviewed scientific journal, for example, the Journal of Diagnostic and Prognostic Research, Journal of Clinical Epidemiology, or BMC Medical Research Methodology. The work will also appear in the University of Manchester PhD thesis of the lead applicant, Celina Gehringer.

Bibliography:

1. Sergeant JC, Hyrich KL, Anderson J, Kopec-Harding K, Hope HF, Symmons DPM, et al. Prediction of primary non-response to methotrexate therapy using demographic, clinical and psychosocial variables: Results from the UK Rheumatoid Arthritis Medication Study (RAMS). Arthritis Res Ther. 2018;20(1):1?11.
2. Smolen JS, Landew RBM, Bijlsma JWJ, Burmester GR, Dougados M, Kerschbaumer A, et al. EULAR recommendations for the management of rheumatoid arthritis with synthetic and biological disease-modifying antirheumatic drugs: 2019 update. Ann Rheum Dis. 2020;79(6):S685?99.
3. Almutairi K, Nossent J, Preen D, Keen H, Inderjeeth C. The global prevalence of rheumatoid arthritis: a meta-analysis based on a systematic review. Rheumatol Int. 2021 May;41(5):863?77.
4. Heidari B. Rheumatoid Arthritis: Early diagnosis and treatment outcomes. Casp J Intern Med. 2011;2(1):161?70.
5. Fraenkel L, Bathon JM, England BR, St.Clair EW, Arayssi T, Carandang K, et al. 2021 American College of Rheumatology Guideline for the Treatment of Rheumatoid Arthritis. Arthritis Care Res. 2021 Jul;73(7):924?39.
6. Recommendations | Rheumatoid arthritis in adults: management | Guidance | NICE [Internet]. NICE; [cited 2022 Mar 2]. Available from: https://www.nice.org.uk/guidance/ng100/chapter/Recommendations
7. Steyerberg EW, Moons KGM, van der Windt DA, Hayden JA, Perel P, Schroter S, et al. Prognosis Research Strategy (PROGRESS) 3: Prognostic Model Research. PLoS Med. 2013 Feb 5;10(2):e1001381.
8. Gehringer CK, Martin GP, Hyrich KL, Verstappen SMM, Sergeant JC. Clinical prediction models for methotrexate treatment outcomes in patients with rheumatoid arthritis: A systematic review and meta-analysis. Semin Arthritis Rheum. 2022 Oct 1;56:152076.
9. Lin L, Sperrin M, Jenkins DA, Martin GP, Peek N. A scoping review of causal methods enabling predictions under hypothetical interventions. Diagn Progn Res. 2021;5(1).
10. Dickerman BA, Hernn MA. Counterfactual prediction is not only for causal inference. Eur J Epidemiol. 2020 Jul 1;35(7):615?7.
11. Sperrin M, Martin GP, Pate A, Van Staa T, Peek N, Buchan I. Using marginal structural models to adjust for treatment drop-in when developing clinical prediction models. Stat Med. 2018 Dec 10;37(28):4142?54.
12. Lin L, Poppe K, Wood A, Martin G, Peek N, Sperrin M. Making predictions under hypothetical interventions: a case study from the PREDICT-CVD cohort in New Zealand primary care [Internet]. In Review; 2022 Sep [cited 2022 Nov 23]. Available from: https://www.researchsquare.com/article/rs-1824359/v2
13. Vickers AJ, Kattan MW, Sargent D. Method for evaluating prediction models that apply the results of randomized trials to individual patients. Trials. 2007 Jun 5;8:14.
14. Vickers AJ, Calster BV, Steyerberg EW. Net benefit approaches to the evaluation of prediction models, molecular markers, and diagnostic tests. BMJ. 2016 Jan 25;352:i6.
15. Chalkou K, Vickers AJ, Pellegrini F, Manca A. Decision curve analysis for personalized treatment choice between multiple options. :24.
16. Collins GS, Reitsma JB, Altman DG, Moons KGM. Transparent reporting of a multivariable prediction model for individual prognosis or diagnosis (TRIPOD): The TRIPOD Statement. BMC Med. 2015;13(1):1?10.
17. Moons KGM, Altman DG, Reitsma JB, Ioannidis JPA, Macaskill P, Steyerberg EW, et al. Transparent Reporting of a multivariable prediction model for Individual Prognosis Or Diagnosis (TRIPOD): Explanation and Elaboration. Ann Intern Med. 2015 Jan 6;162(1):W1?73.
18. van Amsterdam WAC, Ranganath R. Conditional average treatment effect estimation with treatment offset models [Internet]. arXiv; 2022 [cited 2022 Dec 22]. Available from: http://arxiv.org/abs/2204.13975
19. Rombach I, Knight R, Peckham N, Stokes JR, Cook JA. Current practice in analysing and reporting binary outcome data?a review of randomised controlled trial reports. BMC Med. 2020 Jun 8;18(1):147.
20. J T, L B, T H, J R, M W, M H. Different ways to estimate treatment effects in randomised controlled trials. Contemp Clin Trials Commun. 2018 Jun 1;10:80?5.