XML Template (2014) [17.9.2014–12:49pm] //blrnas3.glyph.com/cenpro/ApplicationFiles/Journals/SAGE/3B2/CEPJ/Vol00000/140155/APPFile/SG-CEPJ140155.3d
(CEP)
[1–14] [PREPRINTER stage]
Special Issue Article
Adverse cardiovascular events associated with triptans and ergotamines for treatment of migraine: Systematic review of observational studies
Cephalalgia 0(0) 1–14 ! International Headache Society 2014 Reprints and permissions: sagepub.co.uk/journalsPermissions.nav DOI: 10.1177/0333102414550416 cep.sagepub.com
G Roberto1, E Raschi1, C Piccinni1, V Conti2, L Vignatelli3, R D’Alessandro4, F De Ponti1 and E Poluzzi1 Abstract Background: Apart from the underlying cardiovascular (CV) risk associated with migraine, both triptans and ergotamines can induce vasoconstriction and potentially increase the risk of serious ischemic events. Because of the low frequency of such events in eligible patients, randomized controlled trials are not exhaustive to assess the drug-related CV risk. Observational studies are, therefore, an essential source of information to clarify this matter of concern. Aim: The aim of this study was to systematically review the available published observational studies investigating the risk of serious CV events in triptan or ergotamine users, as compared to unexposed migraineur controls. Methods: We systematically searched MEDLINE and EMBASE electronic databases for cohort or case-control studies up to December 1, 2013. Studies retrieved from CDSR, DARE and HTA databases of the Cochrane Library were used for snowballing. Studies investigating the risk of any CV outcome in patients with a migraine diagnosis and exposed to triptans or ergotamines were considered for inclusion. Selection of studies, data extraction, and risk of bias assessment were conducted independently by two reviewers. Pooled odds ratios (ORs) with 95% confidence interval (95% CI) were computed using a random-effects model for studies and outcomes judged eligible for quantitative data synthesis. Results: From a total of 3370 citations retrieved, after duplicate removal and screening, only four studies met the inclusion criteria (three nested case-control analyses and one retrospective cohort study). These studies investigated the risk of different CV outcomes associated with either the recency or the intensity of exposure to the studied drugs. As for the intensity of use, the pooled OR of serious ischemic events was 2.28 (95% CI 1.18–4.41; I2 ¼ 0%) for ergotamine use (two studies), whereas for triptans (three studies) it was 0.86 (95% CI 0.52–1.43; I2 ¼ 24.5%). Recent use of ergotamines was not significantly associated with any CV outcome (only one available study). Two studies investigated the risk of stroke related to recent triptan use: the first study reported an OR of 0.90 (0.64–1.26), and the second one suggested an increased risk of 2.51 (1.10–5.71). In this case, because of the high degree of heterogeneity, results were not pooled. Conclusions: To date, few comparative observational studies have investigated the CV safety of migraine-specific drugs in clinical practice. Evidence gathered here suggests that intense consumption of ergotamines may be associated with an increased risk of serious ischemic complications. As for triptans, available studies do not suggest strong CV safety issues, although no firm conclusions can be drawn. In particular, evidence on stroke risk is conflicting. However, if an increase of the absolute stroke risk in recently exposed patients does actually exist, it must be small. Overall, residual uncontrolled confounding factors reduce the confidence in the risk estimates collected from the included studies. Further investigations are needed to better define the risk for rare but serious CV events related to triptan and ergotamine use for treatment of migraine. Keywords CV risk associated with migraine, ergotamine, triptans, migraine Date received: 25 April 2014; revised: 9 June 2014; 29 July 2014; accepted: 15 August 2014
1
Department of Medical and Surgical Sciences, University of Bologna, Italy Regional Centre for Pharmacovigilance, Lombardy Region, Italy 3 Local Health Trust of Bologna, Italy; Health and Social Regional Agency of Emilia-Romagna Region, Italy 4 IRCCS, Institute of Neurological Sciences of Bologna, Italy 2
Corresponding author: Elisabetta Poluzzi, Department of Medical and Surgical Sciences, University of Bologna, Via Irnerio 48, I-40126 Bologna, Italy. Email: [email protected]
XML Template (2014) [17.9.2014–12:49pm] //blrnas3.glyph.com/cenpro/ApplicationFiles/Journals/SAGE/3B2/CEPJ/Vol00000/140155/APPFile/SG-CEPJ140155.3d
(CEP)
[1–14] [PREPRINTER stage]
2 Introduction Migraine is a chronic and debilitating neurological disorder affecting more than 10% of the worldwide adult population (1). Different pharmacological approaches are currently available to treat acute migraine attacks. These are usually classified into two main categories: non-specific therapies with analgesic effect, such as nonsteroidal anti-inflammatory drugs (NSAIDs), paracetamol, barbiturate-containing analgesics and opiates; and migraine-specific drugs, including triptans and ergotamines (1,2). Drug choice in clinical practice is generally based on headache features (e.g. onset rapidity, severity, early vomiting) as well as patients’ preferences and characteristics (e.g. inter-individual variability of response/ tolerability, comorbidities, risk factors, co-medications) (2–4). As for migraine-specific drugs, underlying cardiovascular (CV) diseases or risk factors represent the major limitation for their use in practice. In fact, both triptans and ergotamines can induce, through heterogeneous mechanisms, vasoconstriction that may potentially increase the risk of serious ischemic events (2,4–6). Ergotamine and dihydroergotamine were the first specific anti-migraine agents introduced in therapy, respectively, in 1926 and 1945 (7). Their efficacy in the treatment of migraine is thought to be related to the affinity with 5HT1b/d receptors, through which these drugs act on central serotoninergic neurons, inhibit neurogenic inflammation and cause cerebral vasoconstriction (8). However, both ergotamine and dihydroergotamine also activate other 5HT receptors subtypes and a-adrenoceptors that, in turn, are mostly responsible for side effects (7–9), such as rare but potentially fatal vasospastic reactions (e.g. ergotism, gangrene) (5). In the early 1990s, sumatriptan, the first selective agonist of the 5HT1b/d receptors, was specifically developed and marketed for the acute treatment of migraine, with the promise of a better safety profile than ergotamines. In the last two decades, a total of six other sumatriptan analogs, commonly called ‘‘triptans,’’ have been licensed worldwide (1). During these years, a large number of randomized clinical trials have demonstrated the efficacy of triptans, and no major CV issues have been reported from such studies (3,10,11). However, results from pre-approval clinical trials have a recognized limited power in detecting rare adverse reactions, as well as little generalizability toward ‘‘real-world’’ populations (12). During post-marketing experience, different serious ischemic vascular events have been reported in association with the use of triptans (6,13). Nevertheless, on the basis of case reports only, the risk of such events
Cephalalgia 0(0) cannot be estimated and, in most of cases, a possible causal relationship with triptan exposure cannot be definitively excluded (3). In this context, post-marketing observational studies can provide important information on the CV safety profile of migraine-specific drugs. In fact, by involving large and unselected populations from a real-world setting, observational studies may represent the best source for estimating the risk of rare adverse drug reactions in clinical practice (12). Therefore, the aim of this study was to systematically review the available published literature on observational studies (case-control and cohort) investigating the risk of severe CV events associated with triptan or ergotamine use among migraine patients.
Methods Protocol registration The study protocol was submitted in advance to the International Prospective Register of Systematic Reviews (PROSPERO) (14).
Characteristics of studies eligible for inclusion Observational cohort or case-control studies investigating the risk of any CV outcome in patients exposed at least to a triptan (i.e. sumatriptan, rizatriptan, zolmitriptan, naratriptan, eletriptan, almotriptan, frovatriptan) or an ergotamine compound (i.e. ergotamine or dihydroergotamine) for the acute treatment of migraine were considered for inclusion. Since migraine itself may represent an independent risk factor for serious CV outcomes (15,16), we decided to exclude studies involving non-migraineurs as the control group. No restriction with respect to age or gender was applied.
Search strategy Studies entered up to December 1, 2013, in PubMed, EMBASE, Cochrane Database of Systematic Reviews (CDSR), Database of Abstract of Review of Effects (DARE) and Health Technology Assessment (HTA) database of the Cochrane Library were retrieved using a combination of exploded terms and free text. As for PubMed, the terms ‘‘Serotonin Receptor Agonists,’’ Medical Subject Heading (MeSH), ‘‘triptan*,’’ ‘‘sumatriptan*,’’ ‘‘rizatriptan*,’’ ‘‘naratriptan*,’’ ‘‘zolmitriptan*,’’ ‘‘eletriptan*,’’ ‘‘almotriptan*,’’ ‘‘frovatriptan*,’’ ‘‘ergotamine*,’’ ‘‘dihydroergotamine*’’ AND ‘‘cardiovascular diseases’’ (MeSH) OR ‘‘all the cardiovascular events associated to triptan use as reported by Roberto et al. (13)’’ were used in combination with the search limit ‘‘humans.’’ A similar
XML Template (2014) [17.9.2014–12:49pm] //blrnas3.glyph.com/cenpro/ApplicationFiles/Journals/SAGE/3B2/CEPJ/Vol00000/140155/APPFile/SG-CEPJ140155.3d
(CEP)
[1–14] [PREPRINTER stage]
3
Roberto et al. approach was used also for searching in EMBASE (full search strategy for all databases concerned is available in the online Supplementary file 1). CSDR, DARE and HTA databases were searched in order to identify other systematic reviews possibly useful for snowballing: In this case, in order to be as sensible as possible, the exposure of interest only was included in the search string, i.e. ‘‘Serotonin Receptor Agonists,’’ MeSH ‘‘triptan*,’’ ‘‘sumatriptan*,’’ ‘‘rizatriptan*,’’ ‘‘naratriptan*,’’ ‘‘zolmitriptan*,’’ ‘‘eletriptan*,’’ ‘‘almotriptan*,’’ ‘‘frovatriptan*,’’ ‘‘ergotamine*’’ and ‘‘dihydroergotamine*.’’ Retrieved references were managed with Reference Manager 11.
Study selection Two pairs of investigators (GR-ER and LV-EP), working in duplicates, independently examined all titles and abstracts of retrieved references. When characteristics for inclusion/exclusion were not clearly reported within the title/abstract, or the abstract was not available, studies were in any case retained for subsequent full-text screening. Retrieval of full-text studies was performed with the help of the document delivery system’s librarians of the University of Bologna. Only studies in English, Spanish, French or German were screened on full-text. Included studies as well as any other full paper that could include useful references (e.g. review, case-reports with literature review, guidelines, bulletin) were hand-searched for further studies of interest.
Data extraction and quality appraisal Data from included studies were extracted by GR, ER, CP, VC working independently and in duplicate. A pre-established form concerning information on study design, source of data, population characteristics, outcomes of interests, matching criteria and confounding factors considered for statistical adjustment was used. EP and LV were consulted for discrepancies on extracted data. The risk of bias of included studies was assessed by using the Newcastle-Ottawa Scale (NOS) (17). In order to produce a graphical representation of the assessment results, the three NOS domains (selection, comparability and exposure or outcome) were imported in the Risk of Bias tool of the Review Manager software (version 5.2) provided by the Cochrane Collaboration (18). We further customized the Risk of Bias tool by adding two fields concerning ‘‘migraine ascertainment’’ and ‘‘company funding,’’ since these aspects are not specifically addressed within the NOS scale, but they should be taken into account when interpreting study results.
Data synthesis According to the Meta-analysis Of Observational Studies in Epidemiology (MOOSE) guidelines, we used broad inclusion criteria (19). For those studies and outcomes judged eligible for data synthesis, the most adjusted relative risks, odds ratios (ORs) or hazard ratios were used, with 95% confidence intervals (95% CIs), assuming that ORs, relative risks and hazard ratios were all comparable measures in the presence of relatively rare events (20). Statistical heterogeneity among studies was evaluated using Q statistic (with p < 0.10 considered significant), and I-squared statistic (I2) together with visual inspection of graphs were used to estimate the proportion of total variation contributed by between-study variance. Pooled ORs with 95% CIs were computed for the comparisons of interest, using both a fixed-effect model (inverse-variance method) (20) and a random-effects model (DerSimonian and Laird method) (21). In order to take into account differences between observational studies, the random-effects model was preferred for the presentation of results, since it represents a more conservative approach for the calculation of aggregated effect size. All statistical analyses were performed with Comprehensive Meta-Analysis Version 2 software. As recently suggested by Barbui et al. (22), the GRADE pro Version 3.6 software was used to transparently grade the strength of evidence collected on the CV profile of triptans and ergotamines, and generate a summary of findings table.
Results Literature search results The literature search strategy retrieved 3370 citations (Figure 1). After automated duplicate removal and screening on title and abstract of the retrieved studies, a total of 189 full-text articles were further assessed for inclusion. Only four studies met the pre-specified inclusion criteria (23, 23–26). Four otherwise eligible casecontrol studies were excluded because of the use of nonmigraineurs as controls (27–30). Reasons for exclusion of the remaining screened full papers are reported in Figure 1.
Characteristics of included studies A total of three nested case-control and one retrospective cohort study were included. Design, population, intervention and outcome characteristics are shown in Table 1. The included studies were all published between 2004 and 2007. For only two out of three case-control studies was information on sample size
XML Template (2014) [17.9.2014–12:49pm] //blrnas3.glyph.com/cenpro/ApplicationFiles/Journals/SAGE/3B2/CEPJ/Vol00000/140155/APPFile/SG-CEPJ140155.3d
(CEP)
[1–14] [PREPRINTER stage]
4
Cephalalgia 0(0)
PubMed: 1518 EMBASE: 1805 CDSR: 14 DARE: 26 HTA: 7
Duplicate references removed
Total records retrieved n = 3370
References screened through title/abstract n = 2929
Full-text articles assessed for eligibility n = 189
n = 441
Records excluded n = 2740
Full-text articles excluded n =185 Reasons for exclusion:
Studies included in qualitative synthesis n=4 One cohort study Three case-control studies
Studies included in quantitative synthesis (meta-analysis) n=3
Case reports: 24 Case series: 12 Case-control studies: -Abstract only (Ongoing study): 1 -No exposure of interest: 3 -No migraineur controls: 4 Cohort studies -No outcome of interest: 1 -No exposure of interest: 1 Comments/letters: 9 Descriptive studies: 25 Guidelines: 2 In vitro studies: 2 Interventional studies: 24 Meta-analysis of interventional studies: 1 Newsletters: 2 Review: 69 Treatment-related cost analysis with no data of interest: 5
Figure 1. Selection process for study inclusion in the systematic review. CDSR: Cochrane Database of Systematic Reviews; DARE: Database of Abstract of Review of Effects; HTA: Health Technology Assessment Database.
reported (about 800 individuals). As for the only included cohort study, sample size was 130,411. Included studies addressed the risk of CV outcomes associated with either the recency or the intensity (i.e. number of prescribed/dispensed doses) of exposure to the studied drugs. Extracted results were presented and pooled accordingly.
Studies estimating the risk of CV events related to the intensity of migraine-specific drug use Two out of three included case-control studies investigated the risk of ischemic complications and adverse cardiac events associated with the intensity of use of
migraine-specific drugs (Table 2(a) and (b)). However, one such study concerned triptans only. Within the cohort study from Velentgas et al. (23), the risk of stroke associated with the intensity of use of triptans and ergotamines was estimated. Concerning triptan treatment, none of the three available studies reported a statistically significant increased risk for the CV outcome under study when the highest frequency of exposure category was considered. However, statistically significant stroke risk emerged for the intermediate intensity category in the study by Velentgas et al. (Table 2(a)). As for ergotamines, only one of the two available studies reported an increased risk for ischemic complications in ‘‘overuser’’
France (outpatients)
UK (outpatients) General Practice Research Database (medical records)
Lugardon (2007)
Becker (2007)
1) First-time migraine diagnosis 2) Date of drug prescription
1) First prescription of triptans 2) Date of drug dispensing
Likely incident
Incident
Without
With and without
Without
1) Patients with >1 prescription of a triptan or ergotamine 2) Date of drug dispensing Prevalent
Without
Type of users
With or without CV diseases
1) Migraine diagnosis Prevalent (ICD9) and/or triptan dispensation 2) Date of drug dispensing
1) Cohort inclusion 2) Exposure assessment
Company funding
Cardiac eventsi
Age ( 1 years), Triptansh (126) sex, index date
Age, sex, general Triptans (not practice, and specified) calendar time
>18 years (50.1 years) 69%
8
>0 to
(CEP)
[1–14] [PREPRINTER stage]
Special Issue Article
Adverse cardiovascular events associated with triptans and ergotamines for treatment of migraine: Systematic review of observational studies
Cephalalgia 0(0) 1–14 ! International Headache Society 2014 Reprints and permissions: sagepub.co.uk/journalsPermissions.nav DOI: 10.1177/0333102414550416 cep.sagepub.com
G Roberto1, E Raschi1, C Piccinni1, V Conti2, L Vignatelli3, R D’Alessandro4, F De Ponti1 and E Poluzzi1 Abstract Background: Apart from the underlying cardiovascular (CV) risk associated with migraine, both triptans and ergotamines can induce vasoconstriction and potentially increase the risk of serious ischemic events. Because of the low frequency of such events in eligible patients, randomized controlled trials are not exhaustive to assess the drug-related CV risk. Observational studies are, therefore, an essential source of information to clarify this matter of concern. Aim: The aim of this study was to systematically review the available published observational studies investigating the risk of serious CV events in triptan or ergotamine users, as compared to unexposed migraineur controls. Methods: We systematically searched MEDLINE and EMBASE electronic databases for cohort or case-control studies up to December 1, 2013. Studies retrieved from CDSR, DARE and HTA databases of the Cochrane Library were used for snowballing. Studies investigating the risk of any CV outcome in patients with a migraine diagnosis and exposed to triptans or ergotamines were considered for inclusion. Selection of studies, data extraction, and risk of bias assessment were conducted independently by two reviewers. Pooled odds ratios (ORs) with 95% confidence interval (95% CI) were computed using a random-effects model for studies and outcomes judged eligible for quantitative data synthesis. Results: From a total of 3370 citations retrieved, after duplicate removal and screening, only four studies met the inclusion criteria (three nested case-control analyses and one retrospective cohort study). These studies investigated the risk of different CV outcomes associated with either the recency or the intensity of exposure to the studied drugs. As for the intensity of use, the pooled OR of serious ischemic events was 2.28 (95% CI 1.18–4.41; I2 ¼ 0%) for ergotamine use (two studies), whereas for triptans (three studies) it was 0.86 (95% CI 0.52–1.43; I2 ¼ 24.5%). Recent use of ergotamines was not significantly associated with any CV outcome (only one available study). Two studies investigated the risk of stroke related to recent triptan use: the first study reported an OR of 0.90 (0.64–1.26), and the second one suggested an increased risk of 2.51 (1.10–5.71). In this case, because of the high degree of heterogeneity, results were not pooled. Conclusions: To date, few comparative observational studies have investigated the CV safety of migraine-specific drugs in clinical practice. Evidence gathered here suggests that intense consumption of ergotamines may be associated with an increased risk of serious ischemic complications. As for triptans, available studies do not suggest strong CV safety issues, although no firm conclusions can be drawn. In particular, evidence on stroke risk is conflicting. However, if an increase of the absolute stroke risk in recently exposed patients does actually exist, it must be small. Overall, residual uncontrolled confounding factors reduce the confidence in the risk estimates collected from the included studies. Further investigations are needed to better define the risk for rare but serious CV events related to triptan and ergotamine use for treatment of migraine. Keywords CV risk associated with migraine, ergotamine, triptans, migraine Date received: 25 April 2014; revised: 9 June 2014; 29 July 2014; accepted: 15 August 2014
1
Department of Medical and Surgical Sciences, University of Bologna, Italy Regional Centre for Pharmacovigilance, Lombardy Region, Italy 3 Local Health Trust of Bologna, Italy; Health and Social Regional Agency of Emilia-Romagna Region, Italy 4 IRCCS, Institute of Neurological Sciences of Bologna, Italy 2
Corresponding author: Elisabetta Poluzzi, Department of Medical and Surgical Sciences, University of Bologna, Via Irnerio 48, I-40126 Bologna, Italy. Email: [email protected]
XML Template (2014) [17.9.2014–12:49pm] //blrnas3.glyph.com/cenpro/ApplicationFiles/Journals/SAGE/3B2/CEPJ/Vol00000/140155/APPFile/SG-CEPJ140155.3d
(CEP)
[1–14] [PREPRINTER stage]
2 Introduction Migraine is a chronic and debilitating neurological disorder affecting more than 10% of the worldwide adult population (1). Different pharmacological approaches are currently available to treat acute migraine attacks. These are usually classified into two main categories: non-specific therapies with analgesic effect, such as nonsteroidal anti-inflammatory drugs (NSAIDs), paracetamol, barbiturate-containing analgesics and opiates; and migraine-specific drugs, including triptans and ergotamines (1,2). Drug choice in clinical practice is generally based on headache features (e.g. onset rapidity, severity, early vomiting) as well as patients’ preferences and characteristics (e.g. inter-individual variability of response/ tolerability, comorbidities, risk factors, co-medications) (2–4). As for migraine-specific drugs, underlying cardiovascular (CV) diseases or risk factors represent the major limitation for their use in practice. In fact, both triptans and ergotamines can induce, through heterogeneous mechanisms, vasoconstriction that may potentially increase the risk of serious ischemic events (2,4–6). Ergotamine and dihydroergotamine were the first specific anti-migraine agents introduced in therapy, respectively, in 1926 and 1945 (7). Their efficacy in the treatment of migraine is thought to be related to the affinity with 5HT1b/d receptors, through which these drugs act on central serotoninergic neurons, inhibit neurogenic inflammation and cause cerebral vasoconstriction (8). However, both ergotamine and dihydroergotamine also activate other 5HT receptors subtypes and a-adrenoceptors that, in turn, are mostly responsible for side effects (7–9), such as rare but potentially fatal vasospastic reactions (e.g. ergotism, gangrene) (5). In the early 1990s, sumatriptan, the first selective agonist of the 5HT1b/d receptors, was specifically developed and marketed for the acute treatment of migraine, with the promise of a better safety profile than ergotamines. In the last two decades, a total of six other sumatriptan analogs, commonly called ‘‘triptans,’’ have been licensed worldwide (1). During these years, a large number of randomized clinical trials have demonstrated the efficacy of triptans, and no major CV issues have been reported from such studies (3,10,11). However, results from pre-approval clinical trials have a recognized limited power in detecting rare adverse reactions, as well as little generalizability toward ‘‘real-world’’ populations (12). During post-marketing experience, different serious ischemic vascular events have been reported in association with the use of triptans (6,13). Nevertheless, on the basis of case reports only, the risk of such events
Cephalalgia 0(0) cannot be estimated and, in most of cases, a possible causal relationship with triptan exposure cannot be definitively excluded (3). In this context, post-marketing observational studies can provide important information on the CV safety profile of migraine-specific drugs. In fact, by involving large and unselected populations from a real-world setting, observational studies may represent the best source for estimating the risk of rare adverse drug reactions in clinical practice (12). Therefore, the aim of this study was to systematically review the available published literature on observational studies (case-control and cohort) investigating the risk of severe CV events associated with triptan or ergotamine use among migraine patients.
Methods Protocol registration The study protocol was submitted in advance to the International Prospective Register of Systematic Reviews (PROSPERO) (14).
Characteristics of studies eligible for inclusion Observational cohort or case-control studies investigating the risk of any CV outcome in patients exposed at least to a triptan (i.e. sumatriptan, rizatriptan, zolmitriptan, naratriptan, eletriptan, almotriptan, frovatriptan) or an ergotamine compound (i.e. ergotamine or dihydroergotamine) for the acute treatment of migraine were considered for inclusion. Since migraine itself may represent an independent risk factor for serious CV outcomes (15,16), we decided to exclude studies involving non-migraineurs as the control group. No restriction with respect to age or gender was applied.
Search strategy Studies entered up to December 1, 2013, in PubMed, EMBASE, Cochrane Database of Systematic Reviews (CDSR), Database of Abstract of Review of Effects (DARE) and Health Technology Assessment (HTA) database of the Cochrane Library were retrieved using a combination of exploded terms and free text. As for PubMed, the terms ‘‘Serotonin Receptor Agonists,’’ Medical Subject Heading (MeSH), ‘‘triptan*,’’ ‘‘sumatriptan*,’’ ‘‘rizatriptan*,’’ ‘‘naratriptan*,’’ ‘‘zolmitriptan*,’’ ‘‘eletriptan*,’’ ‘‘almotriptan*,’’ ‘‘frovatriptan*,’’ ‘‘ergotamine*,’’ ‘‘dihydroergotamine*’’ AND ‘‘cardiovascular diseases’’ (MeSH) OR ‘‘all the cardiovascular events associated to triptan use as reported by Roberto et al. (13)’’ were used in combination with the search limit ‘‘humans.’’ A similar
XML Template (2014) [17.9.2014–12:49pm] //blrnas3.glyph.com/cenpro/ApplicationFiles/Journals/SAGE/3B2/CEPJ/Vol00000/140155/APPFile/SG-CEPJ140155.3d
(CEP)
[1–14] [PREPRINTER stage]
3
Roberto et al. approach was used also for searching in EMBASE (full search strategy for all databases concerned is available in the online Supplementary file 1). CSDR, DARE and HTA databases were searched in order to identify other systematic reviews possibly useful for snowballing: In this case, in order to be as sensible as possible, the exposure of interest only was included in the search string, i.e. ‘‘Serotonin Receptor Agonists,’’ MeSH ‘‘triptan*,’’ ‘‘sumatriptan*,’’ ‘‘rizatriptan*,’’ ‘‘naratriptan*,’’ ‘‘zolmitriptan*,’’ ‘‘eletriptan*,’’ ‘‘almotriptan*,’’ ‘‘frovatriptan*,’’ ‘‘ergotamine*’’ and ‘‘dihydroergotamine*.’’ Retrieved references were managed with Reference Manager 11.
Study selection Two pairs of investigators (GR-ER and LV-EP), working in duplicates, independently examined all titles and abstracts of retrieved references. When characteristics for inclusion/exclusion were not clearly reported within the title/abstract, or the abstract was not available, studies were in any case retained for subsequent full-text screening. Retrieval of full-text studies was performed with the help of the document delivery system’s librarians of the University of Bologna. Only studies in English, Spanish, French or German were screened on full-text. Included studies as well as any other full paper that could include useful references (e.g. review, case-reports with literature review, guidelines, bulletin) were hand-searched for further studies of interest.
Data extraction and quality appraisal Data from included studies were extracted by GR, ER, CP, VC working independently and in duplicate. A pre-established form concerning information on study design, source of data, population characteristics, outcomes of interests, matching criteria and confounding factors considered for statistical adjustment was used. EP and LV were consulted for discrepancies on extracted data. The risk of bias of included studies was assessed by using the Newcastle-Ottawa Scale (NOS) (17). In order to produce a graphical representation of the assessment results, the three NOS domains (selection, comparability and exposure or outcome) were imported in the Risk of Bias tool of the Review Manager software (version 5.2) provided by the Cochrane Collaboration (18). We further customized the Risk of Bias tool by adding two fields concerning ‘‘migraine ascertainment’’ and ‘‘company funding,’’ since these aspects are not specifically addressed within the NOS scale, but they should be taken into account when interpreting study results.
Data synthesis According to the Meta-analysis Of Observational Studies in Epidemiology (MOOSE) guidelines, we used broad inclusion criteria (19). For those studies and outcomes judged eligible for data synthesis, the most adjusted relative risks, odds ratios (ORs) or hazard ratios were used, with 95% confidence intervals (95% CIs), assuming that ORs, relative risks and hazard ratios were all comparable measures in the presence of relatively rare events (20). Statistical heterogeneity among studies was evaluated using Q statistic (with p < 0.10 considered significant), and I-squared statistic (I2) together with visual inspection of graphs were used to estimate the proportion of total variation contributed by between-study variance. Pooled ORs with 95% CIs were computed for the comparisons of interest, using both a fixed-effect model (inverse-variance method) (20) and a random-effects model (DerSimonian and Laird method) (21). In order to take into account differences between observational studies, the random-effects model was preferred for the presentation of results, since it represents a more conservative approach for the calculation of aggregated effect size. All statistical analyses were performed with Comprehensive Meta-Analysis Version 2 software. As recently suggested by Barbui et al. (22), the GRADE pro Version 3.6 software was used to transparently grade the strength of evidence collected on the CV profile of triptans and ergotamines, and generate a summary of findings table.
Results Literature search results The literature search strategy retrieved 3370 citations (Figure 1). After automated duplicate removal and screening on title and abstract of the retrieved studies, a total of 189 full-text articles were further assessed for inclusion. Only four studies met the pre-specified inclusion criteria (23, 23–26). Four otherwise eligible casecontrol studies were excluded because of the use of nonmigraineurs as controls (27–30). Reasons for exclusion of the remaining screened full papers are reported in Figure 1.
Characteristics of included studies A total of three nested case-control and one retrospective cohort study were included. Design, population, intervention and outcome characteristics are shown in Table 1. The included studies were all published between 2004 and 2007. For only two out of three case-control studies was information on sample size
XML Template (2014) [17.9.2014–12:49pm] //blrnas3.glyph.com/cenpro/ApplicationFiles/Journals/SAGE/3B2/CEPJ/Vol00000/140155/APPFile/SG-CEPJ140155.3d
(CEP)
[1–14] [PREPRINTER stage]
4
Cephalalgia 0(0)
PubMed: 1518 EMBASE: 1805 CDSR: 14 DARE: 26 HTA: 7
Duplicate references removed
Total records retrieved n = 3370
References screened through title/abstract n = 2929
Full-text articles assessed for eligibility n = 189
n = 441
Records excluded n = 2740
Full-text articles excluded n =185 Reasons for exclusion:
Studies included in qualitative synthesis n=4 One cohort study Three case-control studies
Studies included in quantitative synthesis (meta-analysis) n=3
Case reports: 24 Case series: 12 Case-control studies: -Abstract only (Ongoing study): 1 -No exposure of interest: 3 -No migraineur controls: 4 Cohort studies -No outcome of interest: 1 -No exposure of interest: 1 Comments/letters: 9 Descriptive studies: 25 Guidelines: 2 In vitro studies: 2 Interventional studies: 24 Meta-analysis of interventional studies: 1 Newsletters: 2 Review: 69 Treatment-related cost analysis with no data of interest: 5
Figure 1. Selection process for study inclusion in the systematic review. CDSR: Cochrane Database of Systematic Reviews; DARE: Database of Abstract of Review of Effects; HTA: Health Technology Assessment Database.
reported (about 800 individuals). As for the only included cohort study, sample size was 130,411. Included studies addressed the risk of CV outcomes associated with either the recency or the intensity (i.e. number of prescribed/dispensed doses) of exposure to the studied drugs. Extracted results were presented and pooled accordingly.
Studies estimating the risk of CV events related to the intensity of migraine-specific drug use Two out of three included case-control studies investigated the risk of ischemic complications and adverse cardiac events associated with the intensity of use of
migraine-specific drugs (Table 2(a) and (b)). However, one such study concerned triptans only. Within the cohort study from Velentgas et al. (23), the risk of stroke associated with the intensity of use of triptans and ergotamines was estimated. Concerning triptan treatment, none of the three available studies reported a statistically significant increased risk for the CV outcome under study when the highest frequency of exposure category was considered. However, statistically significant stroke risk emerged for the intermediate intensity category in the study by Velentgas et al. (Table 2(a)). As for ergotamines, only one of the two available studies reported an increased risk for ischemic complications in ‘‘overuser’’
France (outpatients)
UK (outpatients) General Practice Research Database (medical records)
Lugardon (2007)
Becker (2007)
1) First-time migraine diagnosis 2) Date of drug prescription
1) First prescription of triptans 2) Date of drug dispensing
Likely incident
Incident
Without
With and without
Without
1) Patients with >1 prescription of a triptan or ergotamine 2) Date of drug dispensing Prevalent
Without
Type of users
With or without CV diseases
1) Migraine diagnosis Prevalent (ICD9) and/or triptan dispensation 2) Date of drug dispensing
1) Cohort inclusion 2) Exposure assessment
Company funding
Cardiac eventsi
Age ( 1 years), Triptansh (126) sex, index date
Age, sex, general Triptans (not practice, and specified) calendar time
>18 years (50.1 years) 69%
8
>0 to
