Abstract
Aim
Calcitonin gene-related peptide (CGRP) monoclonal antibodies (mAbs) have expanded preventive treatment options for migraine, especially in patients with previous prophylactic failures. While their efficacy and tolerability are well established, reasons for treatment discontinuation remain insufficiently characterized. This study aimed to systematically evaluate discontinuation rates and underlying reasons for stopping CGRP mAb therapy in a real-world cohort of drug-resistant migraine patients.
Methods
This retrospective, single-center cohort study included adults with episodic or chronic migraine treated with erenumab, galcanezumab, or fremanezumab at the West German Headache Center. Baseline clinical characteristics, monthly headache and migraine days, and acute medication intake were extracted from routine clinical documentation. Treatment status was assessed via medical records and telephone verification. Reasons for discontinuation were self-reported and categorized as perceived lack of effectiveness, adverse events, or subjective remission.
Results
Of 252 included patients, 74.6% (n = 188) continued therapy, while 25.4% (n = 64) discontinued. Perceived ineffectiveness was the most common reason for discontinuation (67.2%, n = 43), and 76.7% (n = 33) of these patients stopped within six months. Only 9.3% (n = 4) reported secondary loss of effectiveness after an initial response. Patient-reported qualitative improvements—such as reduced attack intensity or duration—were frequent among patients who continued treatment, even when formal response criteria based on monthly headache days (MHDs) were not met. Prior responsiveness to triptans was associated with continuation of therapy.
Conclusion
In this real-world cohort of drug-resistant migraine, treatment persistence with CGRP mAbs was high. Most discontinuations occurred within six months and were driven by perceived lack of effectiveness. Qualitative improvements beyond reductions in MHDs were important factors contributing to continued therapy. These findings highlight the relevance of patient-reported improvements in clinical decision-making.
This is a visual representation of the abstract.
Introduction
Monoclonal antibodies (mAbs) targeting the calcitonin gene-related peptide (CGRP) or its receptor have emerged in recent years as a promising option for the prophylactic treatment of migraine. Due to their targeted mechanism of action, favorable tolerability, and ease of administration, mAbs are considered an important advancement in the management of this chronic condition. 1 Clinical trials and real-world data have demonstrated a significant reduction in headache days and an improvement in quality of life with CGRP mAb therapy, leading to its incorporated into European and German guidelines.2,3 Additionally, mAbs have shown improved adherence compared to traditional oral preventives. 4
Despite these positive outcomes, clinical practice shows that a subset of patients discontinues CGRP mAb therapy. Evidence on the underlying reasons for treatment discontinuation remains limited. Treatment termination is likely influenced by multiple, partly overlapping factors, including insufficient objective response, absence of subjective benefit, and adverse events. Beyond these aspects, individual expectations and the way patients perceive and integrate therapeutic effects into their daily lives may further shape discontinuation decisions.
Recent real-world studies have begun to characterize patients who discontinue anti-CGRP mAbs, their unmet needs, and subsequent management pathways, emphasizing that treatment discontinuation extends beyond predefined response thresholds. 5 Evidence that migraine frequency and disability often worsen rapidly after cessation, even following prolonged treatment, 6 underscores that discontinuation is not a clinically neutral event. Consequently, the timing and reasons for treatment discontinuation may provide important insights into patient decision-making and perceived treatment benefits or limitations, highlighting the clinical relevance of identifying potentially modifiable factors.
The aim of this study is to assess discontinuation rates of CGRP mAb therapy in a clinical cohort and to analyze the underlying reasons in detail. Particularly, it focusses on how patients define effectiveness, how qualitative changes in the migraine course influence treatment decisions, and when treatment discontinuations occur.
Methods
This single-center, retrospective analysis was conducted at the West German Headache Center, University Hospital Essen, Germany, and included adults with drug-resistant episodic (EM) and chronic migraine (CM) who initiated CGRP mAb therapy between March 2018 and May 2022. Baseline data were collected from questionnaires, headache diaries, and medical documentation, covering the three months prior to CGRP mAb initiation and the first three months of prophylaxis, including demographics, migraine characteristics, and treatment-related variables. The independent ethics committee of the University Hospital Essen approved the analysis (22-11051-BO). As this was an analysis of internal routine data, written informed consent was not required.
Patients were eligible if they were 18 years or older and had received at least three months of CGRP mAb treatment (erenumab, galcanezumab or fremanezumab). Exclusion criteria comprised missing data on monthly headache days (MHDs) or monthly migraine days (MMDs) before and during the first three months of treatment, as well as lacking baseline demographic information.
All patients had previously failed to respond to at least four approved nonspecific preventive medications for EM and at least five for CM, had discontinued these treatments due to adverse events, or were contraindicated for their use. Approved prophylactic classes comprised beta-blockers (metoprolol or propranolol), tricyclic antidepressants (amitriptyline), calcium channel blockers (flunarizine), anticonvulsants (topiramate), and, in CM, onabotulinumtoxinA. Consequently, all patients were classified as drug-resistant.
Although some patients switched between different CGRP mAbs, only MHD, MMD, and monthly days of acute medication intake (AMD) data from the first initiated CGRP mAb therapy were included in the analysis. However, to assess treatment duration and classify treatment status, the entire CGRP mAb therapy period—including any sequential mAb switches—was considered, as long as no treatment gap exceeded six months. Treatment duration was thus defined as the time from the initiation of the first mAb to the last recorded prescription of any CGRP mAb.
Due to the clinical setting, not all patients attended regular follow-up visits and instead received only prescriptions. Some patients were comanaged by external neurologists or general practitioners outside our center. To determine therapy status, we reviewed the dates of CGRP mAb prescriptions in digital documentation. If the last prescription was issued more than three months ago, therapy status was verified through telephone interviews. For patients who had discontinued therapy, self-reported primary reasons for discontinuation were recorded, including whether a perceived gradual loss of effectiveness contributed to the decision. Actual MHD and MMD data were also collected.
EM, CM, and medication overuse headache were defined according to International Classification of Headache Disorders, Third Edition criteria. Patients with MHD ≥ 15 and MMD between five and eight were classified as CM. Response definitions were based on MHD rather than MMD, as differentiation between headache and migraine days is often challenging in patients with CM in routine clinical practice, particularly in retrospective real-world data; this approach has also been used in previous real-world studies. 5 Objective response was defined as ≥ 50% reduction in MHD for EM and ≥ 30% reduction for CM after three months of CGRP mAb prophylaxis. Nonresponse (NR) was defined accordingly. Daily headache was defined as the presence of headaches every day for at least three months prior to CGRP mAb initiation. Subjective benefit was defined as patient-reported qualitative improvements in migraine characteristics, such as reduced attack intensity, duration, or functional impairment. Treatment effectiveness was defined as the patient's overall judgment integrating both aspects, resulting in treatment continuation or discontinuation.
Based on treatment status and self-reported reasons, patients were categorized as follows:
Descriptive statistics were primarily used to analyze baseline characteristics, continuation rates, and reasons for discontinuation, which are reported as absolute and relative numbers. Data distribution was assessed visually. As continuous variables were not normally distributed, they are reported as medians with interquartile ranges. Between-group comparisons of continuous variables were performed using the Mann–Whitney U test. Categorical variables were compared using Fisher's exact test or chi-square test, as appropriate. Due to small sample sizes, the DAE and DMR groups were analyzed descriptively without formal hypothesis testing.
To further investigate response patterns among nonresponders (NRs), analyses were conducted between patients classified as NR in the CT and DI groups, addressing multiple related questions on clinical response. Because several migraine characteristics after the first three months of mAb prophylaxis were compared, a Bonferroni correction was applied to account for multiple testing, resulting in an adjusted significance level of α = .008. Chi-square tests with continuity correction were used, and statistical analyses were conducted using R (version 4.4.2).
Results
Of 340 screened patients, 252 met inclusion criteria; 27 were excluded and 60 did not participate in follow-up (Figure 1). Eighteen patients fulfilled criteria of the extended classification for CM. Baseline characteristics are summarized in Table 1 and Supplemental Tables S1 and S2.
Flow chart depicting study population selection and exclusion criteria. * < 4 Monthly migraine days; ** Exclusion due to missing data occurred in cases of missing MHD and/or MMD information in the baseline form or the three-month form for the initial antibody.
Baseline characteristics and outcomes.
Note: CT: continued treatment; DAE: discontinuation due to adverse events; DI: discontinuation due to perceived ineffectiveness; DMR: discontinuation due to migraine remission; MOH: medication overuse headache; NOA: nonopioid analgesics (including NSAIDs and metamizole); Q1: first quartile; Q3: third quartile.
The overall column represents the total study population across all groups and is descriptive only. Percentages are calculated based on available (nonmissing) data. Δ values indicate change from baseline to three months. Q1 and Q3 denote the 25th and 75th percentiles, respectively. Between-group comparisons were restricted to the CT and DI groups.
Missing data occurred in the following variables: aura (CT n = 4, DI n = 3, overall n = 7); type of acute medication—NOA (CT n = 5, DI n = 1, overall n = 6); triptan efficacy (CT n = 2, overall n = 2); baseline MOH (DMR n = 1, overall n = 1); baseline AMD (DMR n = 1); loss of MOH (DMR n = 1, overall n = 1); development of MOH (DMR n = 1, overall n = 1).
Superscript numbers indicate the statistical test applied (CT vs. DI).
Mann–Whitney U test.
Fisher's exact test.
Chi-square test.
At follow-up, 188 patients (74.6%, n = 188/252) continued therapy (CT group), while 64 (25.4%, n = 64/252) had discontinued (DI, DMR, or DAE group). Perceived ineffectiveness accounted for the majority of discontinuations (DI group, 67.2%, n = 43/64), followed by remission (DMR group, 17.2%, n = 11/64) and adverse events (DAE group, 15.6%, n = 10/64).
DI patients were characterized by early discontinuation and lack of subjective benefit. Discontinuation occurred predominantly within the first six months of treatment (76.7%, 33/43), while 11.6% (5/43) discontinued between six and nine months. In contrast, most CT patients (76.1%, n = 143/188) remained on therapy for more than two years (Figure 2).
Duration of CGRP mAb therapy by discontinuation reason.
Switching patterns also differed. Most CT patients (64.4%, n = 121/188) received only one mAb, whereas a higher proportion of DI patients switched treatments, with 41.9% (n = 18/43) received two and 14.0% (n = 6/43) received three different mAbs (Figure 3).
Distribution of number of tried CGRP mAb agents by groups.
Further between-group differences were observed at baseline. Patients continuing treatment reported aura more often (CT: 46.2%, n = 85/184 vs. DI: 22.5%, n = 9/40; p = .008), showed a higher rate of triptan effectiveness (CT: 90.9%, n = 169/186 vs. DI: 67.4%, n = 29/43; p < .001), and were less likely to report daily headache (13.3%, n = 25/188 vs. 25.6%, n = 11/43; p = .061; Table 1).
Treatment response diverged early, with significantly smaller median reductions in MHD, MMD, and AMD in the DI group after three months (CT vs. DI, all p < .001). In the DI group, 30.2% (n = 13/43) showed an increase in MHD or MMD, compared with 13.8% in the CT group (n = 26/188; p = .014). Four DI patients (9.3%, n = 4/43) experienced a secondary loss of effectiveness over time after an initial response, including one after SARS-CoV-2 infection. Although classified as NR, 73.6% of all NR patients (n = 95/129) continued therapy. NR was significantly more frequent in the DI group than in the CT group (odds ratio 2.8, 95% CI [1.36–5.98]; p = .006).
To identify possible reasons, a subgroup analysis of NR-CT and NR-DI was conducted. It revealed distinct patterns in patient-reported changes of migraine characteristics. NR-CT patients more frequently reported subjective improvements in attack duration (70.0%, n = 63/90) and attack intensity (69.0%, n = 60/87), whereas NR-DI patients predominantly described unchanged or worsening symptoms (duration: 75.0%, n = 24/32; intensity: 59.4%, n = 19/32). NR-CT was significantly associated with improvement of duration (p = .003), intensity (p < .001), and need for rest (p = .007), but not of light/noise sensitivity (p = .115), nausea (p = .015), or vertigo (p = .040) (Figure 4). Detailed characteristics of the subgroups can be found in Supplemental Table S3.
Percentage distribution of development of migraine characteristics after three months of CGRP mAb prophylaxis among nonresponder subgroups. Nonresponse defined as a reduction after three months < 50% in MHD for EM and < 30% in MHD for CM. Tested with Chi2 with continuity correction. Adjusted α = .008 with Bonferroni correction.
Adverse events leading to discontinuation were heterogeneous (detailed listed in the Supplemental List S4) and most frequently occurred within the first year (60.0%, n = 6/10, Figure 2). In the DMR group, patients discontinued therapy due to sustained clinical improvement and perceived remission (4.4%, n = 11/252). All patients responded initially to prophylaxis and subsequently discontinued treatment after a treatment-free trial. Eight of 11 DMR patients discontinued within the first year (Figure 2). Some patients attributed remission to dietary changes, menopause, and testosterone injections, while others cited no specific causes.
Discussion
In this real-world cohort of drug-resistant migraine patients, we observed high long-term persistence with CGRP mAb therapy, alongside a clinically relevant proportion of treatment discontinuations (25.4%, 64/252). Most patients continued therapy over extended periods, indicating sustained perceived benefit in a drug-resistant population. At the same time, discontinuation occurred predominantly within the first months after treatment initiation and was mainly driven by perceived lack of effectiveness (67.2% 43/63) rather than adverse events (15.6%, 10/64) or clinical improvement (17.2%, 11/64). These findings complement evidence from open-label and extension studies demonstrating sustained efficacy and good tolerability of CGRP-targeted therapies, 7 align with discontinuation rates reported in other real-world cohorts (Supplemental Table S5),8–25 and provide further insights into how and when treatment success or failure is judged in routine clinical practice.
Objective NR was already apparent within the first three months of therapy and differed between CT and DI. Subsequent treatment discontinuation due to perceived ineffectiveness occurred predominantly within the first six months of treatment (76.7%, 33/43). Thus, only a minority of patients discontinued therapy beyond this time frame. Notably, despite fulfilling formal NR criteria based on MHD, a substantial proportion of patients continued CGRP mAb therapy (50.5%, 95/188). Treatment continuation among NRs was closely associated with subjective benefits. In the NR-CT subgroup, approximately two-thirds of patients reported reductions in attack intensity and duration, whereas most NR-DI patients described unchanged or worsening symptoms after three months of treatment. In contrast, nonheadache symptoms such as light/noise sensitivity, nausea, or vertigo did not differ meaningfully between groups. This dissociation underscores that numerical reductions in headache days alone do not fully capture perceived treatment effectiveness in routine clinical practice.
Survey data from the Greek Society of Migraine and Headache Patients indicate that patients—particularly those with CM—often hold high expectations toward preventive therapy, frequently considering reductions of ≥50–75% in headache days as necessary for treatment success, whereas smaller reductions may already be meaningful in EM. 26 In contrast to these expectations, more than half of patients in our cohort who were classified as NR according to MHD criteria continued CGRP mAb therapy, often over prolonged periods. Treatment continuation in this group was associated with subjective benefits, as patients were less likely to discontinue therapy when migraine intensity or duration improved, even in the absence of a formal MHD response. These findings complement recent real-world observations by Burgalassi et al., who highlighted that treatment discontinuation reflects a multifactorial decision process shaped by patient expectations, perceived benefit, and unmet needs rather than objective response thresholds alone. 5
Only a small subset of patients who discontinued therapy due to perceived ineffectiveness reported a secondary loss of effectiveness over time after an initial response. Relative to the total treated population, this phenomenon accounted for a very small proportion of discontinuations and therefore does not appear to represent a major driver of treatment cessation in routine care (9.3%, 4/43). This observation contrasts with reports from other real-world cohorts, such as the study by Kaltseis et al., in which secondary loss of effectiveness was described in a higher proportion of patients after several months of treatment. 23 Differences in study design, patient selection, follow-up structure, and clinical management, including switching strategies, may partly explain these discrepancies. Taken together, we suggest that while secondary loss of effectiveness can occur in individual cases, it is substantially less frequent than primary NR and should be interpreted cautiously when counseling patients or evaluating discontinuation decisions.
These findings underscore the importance of improvements in life quality in shaping treatment adherence. A study by López-Bravo et al. supports this interpretation, showing that patient satisfaction with CGRP mAb does not depend solely on the reduction of migraine days, but also on perceived tolerability and ease of use. 27 Especially in the context of CGRP mAb therapy—where effects on intensity, duration, and functional impairment may precede reductions in attack frequency—early counseling on response patterns and realistic timelines may help to avoid premature discontinuation. Patients’ expectations and the quality of physician–patient communication may play a crucial role in treatment adherence.
Given that subjective benefits may precede or occur independently of reductions in headache days, patients’ perception of treatment effectiveness in routine care is not exclusively defined by numerical response thresholds. Improvements in attack intensity, duration, functional impairment, or other nonheadache most bothersome symptoms may substantially shape individual treatment appraisal and continuation decisions, particularly in drug-resistant patients in whom even partial relief can represent meaningful clinical benefit. This highlights the importance of integrating qualitative symptom changes into clinical interpretation and patient counseling, while acknowledging that nonheadache most bothersome symptoms were not systematically assessed in the present study and therefore represent a limitation.
In our study, patients who reported a good or partial response to triptans continued CGRP mAb therapy more often, whereas poor or absent triptan response was overrepresented among patients discontinuing treatment due to perceived ineffectiveness. This finding is consistent with the concept of CGRP-sensitive versus CGRP-insensitive migraine phenotypes, in which responsiveness to acute CGRP-modulating therapies parallels response to preventive CGRP blockade. 28 Our results support this model in a real-world, drug-resistant population and align with meta-analytic data identifying triptan responsiveness as a predictor of CGRP mAb effectiveness. 29 While daily headache was more frequent among patients discontinuing treatment, no consistent pattern emerged regarding psychiatric comorbidities, possibly reflecting underreporting in retrospective routine documentation. Taken together, these observations suggest that readily available clinical markers, such as triptan response, may support early expectation management and individualized treatment planning.
Adverse events accounted for only 4.0% (10/252), highlighting the overall good tolerability of CGRP mAbs in this real-world cohort. The reported adverse events were heterogeneous and did not cluster around a specific symptom or system, which limits conclusion regarding typical side effect profiles (see Supplemental Table S1 and List S4).
Discontinuation due to sustained clinical improvement was observed in only a small proportion of patients (DMR, 4.4%; 11/252). Although all patients in this subgroup initially responded to CGRP mAb therapy, sustained remission cannot be attributed to pharmacological effect alone, as additional individual factors such as dietary changes, menopause, or testosterone supplementation may have contributed. This finding suggests that treatment discontinuation despite favorable clinical outcomes represents a heterogeneous phenomenon in real-world practice.
This study has several limitations. Its retrospective design may introduce recall bias and reduce data accuracy. Due to the number of patients, the specific effects of the individual mAb (erenumab, galcanezumab, fremanezumab) could not be analyzed. Particularly in the subgroup, the small sample size may impact the robustness and generalizability of our findings. The reliance on patient-reported adverse events and comorbidities may lead to underreporting or misclassification. In addition, although total treatment duration included possible therapy switches between different CGRP mAbs, outcome measures were analyzed exclusively for the first initiated mAb.
Furthermore, our sample consisted exclusively of drug-resistant patients, which provides insight into particularly challenging clinical population but limits generalizability of our findings to broader patient populations. Comparisons with other real-world studies are constrained by differences in sample size, study design, treatment duration, and outcome definitions, all of which should be considered when interpreting our findings. Moreover, a substantial proportion of screened patients did not participate in follow-up, which may have resulted in under- or overestimation of discontinuation rates and reasons, as some patients may have discontinued therapy without reporting their motives. This loss to follow-up could therefore bias estimates of treatment continuation. In addition, important aspects of migraine burden, including most bothersome symptoms, were not assessed, and validated outcome measures such as Migraine Disability Assessment Scale or HIT-6 were not used, limiting the ability to evaluate their influence on patients’ perception of treatment effectiveness.
Nevertheless, this real-world study provides insight into treatment behavior among severely affected, drug-resistant migraine patients and helps to better define expectations regarding CGRP mAb therapy. CGRP mAb treatment in this cohort was characterized by high long-term persistence and good tolerability. Treatment discontinuation occurred in a clinically relevant minority of patients and was predominantly driven by perceived lack of effectiveness, typically associated with objective NR and absence of subjective benefit. Prior response to triptans emerged as a potential clinical marker of continued benefit from CGRP mAbs. Importantly, patients with objective NR but meaningful subjective improvements predominantly continued treatment, underscoring that patients’ perception of treatment effectiveness in routine care extends beyond numerical response thresholds. These findings highlight the importance of integrating qualitative patient-reported benefits into clinical decision-making and aligning treatment expectations early in the course of therapy. Further prospective studies are needed to refine predictors of response and to better define patient-centered criteria for treatment continuation or discontinuation.
Clinical implications
Most discontinuations of CGRP mAb therapy occur in the first six months and are linked to perceived lack of objective response and subjective benefits.
Subjective benefits in intensity and duration strongly influence treatment continuation.
Triptan responsiveness may help identify patients likely to benefit from CGRP antibodies.
Patient counselling on response patterns may reduce premature discontinuation.
Supplemental Material
sj-docx-1-rep-10.1177_25158163261433966 - Supplemental material for Reasons for discontinuation of CGRP antibody therapy: Insights from a retrospective real-world cohort study
Supplemental material, sj-docx-1-rep-10.1177_25158163261433966 for Reasons for discontinuation of CGRP antibody therapy: Insights from a retrospective real-world cohort study by Lennart Menzel, Dominik Fiebelkorn, Vincent Breunung, Sarah Flossdorf, Alexander Wolfgang Becker, Christoph Kleinschnitz, Diana Lindner, Michael Nsaka, Dagny Holle and Armin Scheffler in Cephalalgia Reports
Footnotes
Acknowledgments
AS was part of the UMEA Clinician Scientist program, funded by Faculty of Medicine, University of Duisburg-Essen and German Research Fund (DFG). Following large language models were used to improve English quality: DeepL Wright, and ChatGPT 5.2. Graphical abstract created using Apple Keynote and Microsoft PowerPoint (Apple Inc., Cupertino, CA, USA; Microsoft, Redmond, WA, USA).
Ethical considerations
The independent ethics committee of the University Hospital Essen approved the analysis (22-11051-BO).
Consent to participate
As this was an analysis of internal routine data, written informed consent was not required.
Consent for publishing
Yes.
Author contributions
AS, DH and LM designed and conceptualized the study, interpreted the data and drafted the manuscript. SF corrected the manuscript regarding statistics and revised the manuscript for intellectual content. LM, DF, AWB had the role in the acquisition of data and revised the manuscript for intellectual content. CK, DL and MN revised the manuscript for intellectual content. All authors read and approved the final manuscript.
Funding
The authors received no financial support for the research, authorship, and/or publication of this article.
Declaration of conflicting interests
DH has received scientific support and/or honoraria from Biogen, Novartis, Eli Lilly, Sanofi Aventis, Teva, Allergan, and Hormosan. DL received honoraria from Novartis. CK has received honoraria, a consulting or advisory role to declare from Novartis and Teva. MN received travel fees from Licher MT. AS has received travel fees and/or honoraria from Teva and Novartis. AWB, VB, DF, SF and LM declare that there is no conflict of interest
Data availability statement
The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.
Open practices
Not applicable.
Supplemental material
Supplemental material for this article is available online.
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
