PROTOCOL FORMAT SYSTEMATIC REVIEW ANIMAL INTERVENTION STUDIES Item # Section/topic General 1. Title of the review BY SYRCLE (WWW.SYRCLE.NL) VERSION 0.9 (APRIL 2014) Description Ischemic preconditioning in animal models of myocardial infarction (in no particular order) Check for approval 2. 3. Authors (name, affiliation, contribution) Other contributors (name, affiliation, contribution) KE Wever, SYRCLE and dept. of surgery, Radboudumc, Nijmegen, The Netherlands C Hooijmans, SYRCLE, Radboudumc M Ritskes-Hoitinga, SYRCLE, Radboudumc M Warlé, Dept. of surgery, Radboudumc N Riksen, Dept. of internal medicine, Radboudumc, Nijmegen, The Netherlands D Yellon, Hatter Cardiovascular Institute, London, UK E Sena, CAMARADES, Edinburgh, UK A Tillema, librarian, Radboudumc T Sterenborg, student, Radboudumc I Frenay, student, Radboudumc B Bakker, student, Radboudumc L Clement, student, Radboudumc S De Kruif, student, Radboudumc L Van Beek, student, Radboudumc M Ergün, student, Radboudumc 4. Contact person + e-mail address K Wever kim.wever@radboudumc.nl 5. Date of protocol registration 02-01-2014 Background Coronary heart disease (CHD) is the leading cause of disability and death worldwide, with numbers on the rise in low income countries. Most CHD patients suffer from the detrimental effects of acute myocardial ischemiareperfusion injury, a condition typically arising after myocardial infarction (MI). Novel strategies to reduce infarct size after MI are needed. 6. What is already known about this disease/ model/ intervention? Why is it important to do this review? Ischemic preconditioning (IPC) is a protective strategy in which brief bursts of ischemia induce protection against a prolonged ischemic insult. Since its discovery in the dog heart in 1986 [1], the protective effects of IPC have been demonstrated in hundreds of experiments in rodents, dogs and pigs [2 4]. Several clinical trials have investigated IPC as a protective strategy against MI in humans, but compared to the preceding animal studies, the results have been disappointing (recently reviewed in [4]). Thus, in spite of promising results in animal studies, ischemic conditioning has not yet been successfully SYRCLE SYSTEMATIC REVIEW PROTOCOL V1.0 JULY 2013 1/8
translated into clinical practice [5]. Several important discrepancies between the animal studies on IPC and the trials performed in patients may hamper translation. Firstly, there is no consensus on how many IPC stimuli should be applied for optimal protection, and what the duration of the ischemic and intermediate reperfusion periods should be. It is also unclear whether the so called early or late window of protection is most effective. Furthermore, it is unknown which patient related factors such as age, gender or co morbidities play a role, since the vast majority of animal studies are performed in healthy, young adult males. As a result, the translational transparency of the animal models is limited and the IPC stimulus used in clinical trials could have been suboptimal or incorrectly applied, or unsuitable for the patient population. The aim of this systematic review is to collectively review all animal studies on IPC as a protective strategy for MI. Together, these studies may hold vital clues on how successful translation to the clinical setting can finally be achieved. The analyses will clarify and improve the translational value of these animal models, and can be used to inform the design of future clinical trials. Objectives of this SR Specify the disease / health problem Myocardial infarction, ischemia-reperfusion injury (IRI) of 7. of interest the myocardium Specify the population /species 8. All animal species studied 9. Specify the intervention/exposure Any type of ischemic preconditioning Animals with myocardial infarction, without treatment or 10. Specify the control population with vehicle/sham treatment 11. Specify the outcome measures Myocardial infarct size Does ischemic preconditioning reduce myocardial infarct State your research question (based 12. size in animal models of myocardial infarction, when on point 7-11) compared to untreated animals? Methods: Search and study identification 13. 14. 15. 16. Identify literature databases to search (e.g. Pubmed, Embase, Web of science) Define electronic search strategies (e.g. use the step by step search guide [1] and animals search filters [2, 3]) Identify other sources for study identification Define search strategy for these other sources Pubmed Web of Science SCOPUS EMBASE Other, namely [type here] Specific journal(s), namely [type here] Please add a supplementary file containing your search strategy: <search strategy cardiac IPC.pdf> NA Reference lists of included studies Books Reference lists of relevant reviews Conference proceedings, namely [type here] Contacting authors/ organisations, namely [type here] Other, namely [type here] SYRCLE SYSTEMATIC REVIEW PROTOCOL V1.0 JULY 2013 2/8
17. 18. Study selection procedure Define screening phases (e.g. prescreening based on title/abstract, full text screening, both) Specify number of observers per screening phase Study selection criteria. Define all inclusion and exclusion criteria based on: 19. Type of study (design) 20. 21. Type of animals/ population (e.g. age, gender, disease model) Type of intervention (e.g. dosage, timing, frequency) 22. Outcome measures 23. Language restrictions 24. Publication date restrictions 25. Other 1. Prescreening based on title and abstract (if available) 2. Full-text screening for final inclusion Two observers will screen each reference Inclusion criteria: studies in which a sham or vehicle control (MI only) is compared to an ischemic preconditioned group (IPC + MI) Exclusion criteria: absence of an appropriate control group Inclusion criteria: models in which cardiac IRI is applied to induce myocardial infarction, without co-interventions, in healthy animals (of all ages, sexes and species), or in animals with relevant co-morbities related to risk factors for cardiovascular disease (e.g. hypertension, diabetes, senescence, obesity and cardiac hypertrophy) Exclusion criteria: 1. studies conducted in humans 2. studies conducted ex vivo, in vitro or in silico 3. models inducing permanent coronary occlusion, cardiac transplantation, or no IRI at all 4. models in which the duration of coronary occlusion is too short to induce MI (so-called stunning models) 5. animals with co-morbidities not related to cardiovascular risk factors 6. genetically modified animals 7. models including any co-interventions (e.g. cardiopulmonary bypass) Inclusion criteria: all types of IPC protocols, regardless of duration, timing or site of preconditioning Exclusion criteria: 1. no form of conditioning applied at all 2. postconditioning 3. perconditioning 4. non-ischemic preconditioning Inclusion criteria: myocardial infarct size, expressed as a percentage of the area at risk (commonly: IS/AAR%) Exclusion criteria: all other outcome measures, infarct size expressed in a unit of measurement which cannot be converted to IS/AAR% Inclusion criteria: all languages Exclusion criteria: none Inclusion criteria: all dates Exclusion criteria: none Inclusion criteria: is case of data duplication (two or more papers reporting the same data), we will include the paper SYRCLE SYSTEMATIC REVIEW PROTOCOL V1.0 JULY 2013 3/8
26. Sort and prioritize your exclusion criteria per selection phase reporting the largest dataset / most outcomes/ most details Exclusion criteria: in case of data duplication (two or more papers reporting the same data), we will exclude the paper reporting the smallest dataset / fewest outcomes/ fewest details Selection phase: screening on title and abstract 1. Duplicate from another database 2. No original article (abstract, review, comment, editorial, letter to the editor) 3. Study in humans / patients 4. In vitro / in silico model 5. No cardiac IRI model (e.g. liver, kidney, spinal cord, skin flap) 6. Ex vivo model (e.g. isolated heart) 7. No IRI applied 8. No form of conditioning applied 9. Ischemic postconditioning only 10. Ischemic perconditioning only 11. No ischemic conditioning applied Selection phase: full text screening 1. No appropriate control group 2. Knock-out animals only or animals with nonrelevant co-morbidities 3. Models with co-intervention(s) (e.g. cardiopulmonary bypass, cardiac transplantation) 4. No relevant outcome measure 5. Irretrievable article Study characteristics to be extracted (for assessment of external validity, reporting quality) 27. Study ID (e.g. authors, year) Authors, year, title, language 28. 29. 30. Study design characteristics (e.g. experimental groups, number of animals) Animal model characteristics (e.g. species, gender, disease induction) Intervention characteristics (e.g. intervention, timing, duration) 31. Outcome measures 32. Other (e.g. drop-outs) number of animals in control and experimental groups, type of control group Species, strain, inbred/outbred, age, gender, weight, type of anaesthesia during IPC/IRI, class of anaesthetic, pre-op opioid use, co-morbidity, site of coronary occlusion, duration of ischemia, duration of reperfusion, body temperature control during surgery preconditioning protocol (n*x min ischemia, y min reperfusion), time between IPC&IRI (min), timing of infarct size measurement, site of IPC, if RIPC: remote organ Mean, SD or SEM and n of the infarct size as a percentage of the area at risk, for control and experimental groups In case of repeated use of control group, correct n for multiple comparisons (corrected n = n/#comparisons) Number of excluded animals, reason for exclusion, reporting of a animal welfare statement, conflict of SYRCLE SYSTEMATIC REVIEW PROTOCOL V1.0 JULY 2013 4/8
33. 34. 35. Risk of bias assessment (internal validity) Define criteria to assess the internal validity of included studies (e.g. selection, performance, detection and attrition bias) Collection of outcome data For each outcome measure, define the type of data to be extracted (e.g. continuous/ dichotomous, unit of measurement) Methods for data extraction/retrieval (e.g. extraction from graphs, contacting authors) interest statement and reporting of a power analysis By use of SYRCLE Risk of Bias tool By use of SYRCLE Risk of Bias tool, adapted as follows: additional questions on reporting of randomisation and blinding, reporting of a power calculation and conflict of interest statement. other, namely [type here] Infarct size: continuous, expressed as a percentage of the area at risk (IS/AAR%), extracted as mean ± SD and n in control and experimental groups. To calculate standard deviation (SD) from standard error (SE) of the mean, we will used the formula SD = SE x N. 1. Extract numerical data from text 2. In case of graphical data only: extract from graph using ImageJ software (http://imagej.nih.gov/ij/) 3. In case of missing data: contact authors. 4. In case of no response from authors: use conservative assumption if possible. In case of unclear n, we will use the most conservative estimate (lowest n for SD, highest n for SEM). In case of unclear unit of variance we will assume it to be the SEM, which will produce the most conservative estimate when converted to SD. 5. If no conservative assumption can be made, data will be excluded from analysis. 36. 37. 38. Data analysis/synthesis Specify how you are planning to combine/compare the data (e.g. descriptive summary, meta-analysis) Specify how the decision as to whether a meta-analysis is appropriate will be made If a meta-analysis seems feasible/sensible: Specify the effect measure to be used (e.g. mean difference, standardized In case of unreported data duplication detected by eyeballing, the data will be verified with the authors. In case of no response, suspected duplicate data will be excluded from analysis. Meta-analysis including subgroup analysis We expect the number of included studies to be in the order of 400+. Due to the exploratory nature of the included animal studies and related differences in their experimental design, we expect high heterogeneity and aim to investigate sources of this heterogeneity through meta-analysis. Meta-analysis will not be performed if the number of included studies is unexpectedly low (<20), combined with high heterogeneity. Mean difference SYRCLE SYSTEMATIC REVIEW PROTOCOL V1.0 JULY 2013 5/8
mean difference, risk ratio, odds ratio) We will use the Mean Difference (MD) as our effect measure. This measure can be used when all the outcomes are measured on an identical scale. In our case, all outcome measure data are in the same unit of measurement, namely the infarct size as a percentage of the area at risk of infarction (IS/AAR%). This unit of measurement is relative and corrects the infarct size for the size of the animal, the heart and the AAR. It is therefore on the same scale for each animal. The meaning of e.g. a 20% reduction of the IS/AAR% is considered to be of equal importance in each species. An alternative effect measure, the Standardized Mean Difference (SMD), can combine outcomes measuring the same underlying effect, but on different scales. The method assumes that differences in standard deviations (SD) between studies reflect differences in measurement scales, and not real differences in variability between trial populations. This makes the SMD less suitable for our analyses, since all outcomes are measured on an identical scale and any differences in are a reflection of variation between trial populations. Furthermore, the SMD is obtained by dividing the mean difference in each study by that study's SD. This works well in studies with large group sizes, where the sample SD is close to the population SD. However, because animal studies generally have small group sizes, the SD is often skewed, which unfortunately has a large effect on the SMD, leading to an over- or underestimation of the effect size. Furthermore, use of the SMD in subgroup analysis containing small numbers of studies is associated with low power. Another effect measure which can be used to transform data to the same scale is the Normalized Mean Difference (NMD), which normalizes each effect size to its baseline value. The baseline is the value the outcome measure would have in a healthy animal under baseline circumstances, i.e. without induction of disease model or intervention. This effect measure therefore requires (for each study) a measurement of the IS/AAR% in a sham operated animal, not undergoing myocardial infarction. However, preliminary screening shows that these data are not reported in the majority of studies. Furthermore, use of the NMD in studies with small group sizes analyses is associated with a high false-positive rate. We have therefore decided not to use this effect measure. 39. Specify statistical model of analysis (e.g. random or fixed effects model) A fixed effects model assumes that the data from each study are drawn from the same population with the same true population mean. It therefore assumes that factors that could influence the effect size are identical in all studies. In a fixed effects model, random error is SYRCLE SYSTEMATIC REVIEW PROTOCOL V1.0 JULY 2013 6/8
considered to be the only cause for variation in effect size between the studies. We do not think this is a correct assumption for our field of study. The factors influencing the effect size may differ between studies, because of the exploratory nature of animal experiments. Differences in study design, quality, animal model and intervention may influence the effect size in different ways for different studies. Because of this expected heterogeneity between studies and the explorative nature of the experiments we will use a random effects model. We will use T 2 as our main statistic to assess and describe heterogeneity. T 2 is calculated as the excess variance corrected for the observed variance. The advantage of T 2 is that it is expressed in on the same scale as the effect measure, and therefore reflects the absolute amount of variation on that scale. Because we will use a Mean Difference as our effect measure, T 2 can be interpreted more easily. 40. Specify statistical methods to assess heterogeneity (e.g. I 2, Q) Q is a measure for the total variance of the studies in the meta-analysis, and can be used to calculate the excess variance (Q-k, where k is the degrees of freedom). A p- value can be calculated for Q, giving an indication of whether all studies share a common effect size (p<0.05) or not (p>0.05). However, because Q does not describe the actual magnitude of heterogeneity, we will use it as an informative statistic. 41. Specify which study characteristics will be examined as potential source of heterogeneity (subgroup analysis) I 2 describes the proportion of the observed variance that reflects true differences in effect size between studies (i.e. out of the total variance, which % is caused by true differences between studies?). We will use this as an informative statistic. Factors influencing IPC efficacy: 1. Species (large vs small) 2. Species (compare all species) 3. Sex (male vs female vs mixed) 4. Duration of IPC protocol (total ischemia time) 5. Duration of IPC protocol (ischemia time per cycle) 6. Timing of IPC protocol (divide into categories) 7. Repetition of IPC protocol (compare number of cycles) 8. Site of preconditioning (local vs remote vs both) 9. Co-morbidity (compare classes) 10. Opioid used preoperative (Y vs N) Factors regarding study quality: 11. Reported randomisation (Y vs N/?) 12. Reported blinding (Y vs N/?) SYRCLE SYSTEMATIC REVIEW PROTOCOL V1.0 JULY 2013 7/8
42. Specify the method for assessment of risk of publication bias 43. Sensitivity 13. Risk of bias items, all separately Y vs N vs NR 14. Within large animals vs. small animals o Randomization o Blinding 15. Overall risk of bias (per total # of items) Additional analyses/graphs: 16. Reported quality vs Risk of Bias o Levels of blinding o Levels of randomization 17. Quality stratified per language 18. Quality stratified per year Funnel plot assessment, Egger s regression, trim and fill analysis using STATA and/or CMA software shifting the categories for IRI delay shifting the categories for IPC total duration removing "LAD to carotis bypass" models from analysis Approval by (names, affiliations): KE Wever, C Hooijmans and M Ritskes-Hoitinga, SYRCLE E Sena, CAMARADES Date 29-04-2014 SYRCLE SYSTEMATIC REVIEW PROTOCOL V1.0 JULY 2013 8/8