Neurology

Calcitonin gene-related peptide (CGRP) is a neuropeptide that plays a key role in migraine pathophysiology by promoting the dilation of cerebral blood vessels and transmitting pain signals.¹ CGRP has generated interest for the prevention and acute treatment of migraine. Since 2018, 8 novel CGRP-targeting therapies have been approved by the US Food and Drug Administration (FDA) for the management of migraines.^2,3 For migraine prevention, there are 4 injectable monoclonal antibodies (mAbs) directed against the CGRP receptor (erenumab) or the CGRP ligand (fremanezumab, galcanezumab, and eptinezumab). There are also 2 oral small-molecule CGRP receptor antagonists, termed gepants, that also are approved for migraine prevention (atogepant and rimegepant). Three gepants are approved for acute migraine treatment and are administered orally (rimegepant and ubrogepant) or intranasally (zavegepant) (Table 1).

CGRP-targeting therapies have received attention for their role in vasodilation within the cerebral, coronary, and renal vasculature.⁴ CGRP-mediated vasodilatory effects cause systemic regulation of blood pressure (BP) and play a protective role in hypertension.² Some studies, particularly with erenumab, have shown that the inhibitory role of the agent leads to an increase in BP, as well as gastrointestinal issues such as constipation.^2,5 The FDA recently updated monitoring recommendations for all CGRP-targeting therapies to include the potential for BP elevations and hypertension. Outside of this, there is no definitive evidence linking dual CGRP-targeted therapy to higher cardiovascular or gastrointestinal risks and prescribing information does not carry contraindications.⁶

In a 2021 consensus statement, the American Headache Society (AHS) recommended CGRP-targeting therapies for migraine prevention after inability to tolerate or inadequate response to an 8-week trial of ≥ 2 drug classes including antihypertensives, antiseizure medications, antidepressants, and onabotulinumtoxinA.⁷ For acute treatment, AHS recommended gepant use after contraindication to or inadequate response to ≥ 2 triptans. Guidance on combination CGRP-targeting therapies for both prevention and acute treatment was not provided.⁷ More recently, the AHS published a position statement noting substantial efficacy and safety data for CGRP-targeting therapies and suggested its consideration as a first-line option for migraine prevention, though use for acute treatment or combination CGRP-targeting therapies for both prevention and acute treatment were not addressed.⁸

The International Headache Society guidelines for the acute treatment of migraines recommend nonopioid analgesics as first-line therapy for mild migraine attacks. For moderate to severe attacks, triptans with or without a nonopioid analgesic were recommended as first-line therapy, prior to consideration of CGRP-targeted therapy.⁹ The increased use of this new drug class has also led to combination use of CGRP-targeting therapies for migraine prevention and acute treatment as seen in clinical practice and reflected by some case reports, case series, and small studies describing such use.^10-14 In light of the similar mechanism of action of these therapies and the physiologic role of CGRP, there have been calls for safety evaluation.¹⁵

To our knowledge, no studies have evaluated dual CGRP-targeting regimens for migraine in the veteran population. In 2023, the US Department of Veterans Affairs (VA) and US Department of Defense (DoD) updated their clinical practice guidelines for the management of headache.³ For migraine prevention, the VA/DoD guidelines include a strong recommendation for the use of erenumab, fremanezumab, and galcanezumab; a weak recommendation for the use of atogepant; and a recommendation neither for nor against the use of rimegepant. For acute treatment, the guidelines assign a weak recommendation for the use of rimegepant and ubrogepant. Combination use was not addressed.³

Prior to the VA/DoD guidelines, the Veterans Health Administration restricted the dual use of CGRP-targeting therapies for both preventive and acute migraine treatment. However, the VA Pharmacy Benefit Management Service removed the restriction in the Criteria for Use documents, allowing broader access to these medications for veterans.^16-22 This change permits the use of CGRP-targeting drugs for both acute and preventive migraine treatment after initial data reflecting real-world case reports and open-label studies suggested possible efficacy without a clear safety concern.^11,12 This study aims to fill the gap in the literature by evaluating the safety, efficacy, and overall outcomes of combination CGRP-targeting treatment for migraine prevention and acute treatment in a veteran population.

Methods

This single-center, retrospective, medication use evaluation at the Ralph H. Johnson VA Medical Center (RHJVAMC) was reviewed by the RHJVAMC Research and Development Committee and Quality Improvement Program Evaluation Self Certification Tool, which both determined that institutional review board approval was not required because it was considered part of routine care and quality improvement. Computerized Patient Record System (CPRS) data were reviewed between April 1, 2023 (after the Criteria for Use for CGRP-targeting therapies was updated), through January 31, 2025. Patients were included if they had a confirmed diagnosis of migraine using the International Classification of Headache Disorders, 3rd edition criteria and had concomitant active prescriptions for both a preventive and acute treatment CGRP-targeting agent during the project period.²³ Only patients receiving care from the RHJVAMC neurology department were included.

The primary objective was to assess the safety of dual CGRP-targeting therapies for migraine treatment. Key safety endpoints included effects on liver function, kidney function, and BP. Safety outcomes were graded using Common Terminology Criteria for Adverse Events.²⁴ Changes in liver function were categorized as grade 1, 2, or 3 elevations: grade 1 (aspartate aminotransferase [AST]/alanine aminotransferase [ALT] up to 3x the upper limit of normal [ULN] or bilirubin > 1.5 x ULN); grade 2 (AST/ALT 3-6 x ULN or bilirubin 1.5-3 x ULN); and grade 3 (AST/ALT 5-10 x ULN or bilirubin 3-10 x ULN). Kidney function changes were assessed by serum creatinine levels using a similar grading system: Grade 1 (≤ 1.5 x ULN); grade 2 (1.5-3 x baseline of normal); and grade 3 (3-6 x ULN or baseline). Changes in BP were monitored from baseline to the time of the first neurology follow-up. Elevations were grouped into 2 categories, defined as BP ≥ 140 mm Hg systolic and/or 90 mm Hg diastolic (category 1) and ≥ 160 mm Hg systolic and/or 100 mm Hg diastolic (category 2). Neurology documentation was also reviewed in CPRS for individual patient-reported adverse effects (AEs). Safety endpoints were tracked for any occurrence during the project period.

The secondary objective was to describe the patient-reported efficacy of adding a gepant for acute migraine treatment to existing CGRP-targeting therapies for migraine prevention, in those patients who were stable for ≥ 12 weeks on the preventive therapy. Neurology documentation of headache characteristics, including headache severity as rated on a numerical pain score from 0 (no pain) to 10 (worst pain), and duration of headaches (in hours) were recorded during the project period. Changes in headache characteristics were tracked from baseline (ie, the neurology visit when the gepant was first requested) to the first neurology follow-up within 6 months of initiating gepant for acute treatment. If ranges were provided within documentation, a mean was calculated and used for data collection. Neurology documentation was also reviewed for any patient report of overall effectiveness with the added gepant, and categorized as symptoms improved, worsened, or did not change based on subjective report. Descriptive statistics were used for data analysis. A 1-sample Wilcoxon signed rank test was performed as an exploratory analysis for change in headache characteristics from baseline to first neurology follow-up within 6 months. Each individual CGRP regimen was counted as a unique data point to adequately describe changes associated with each new medication and/or dose adjustment. Therefore, patients could be included more than once to account for each distinct treatment regimen.

Results

From April 1, 2023, to January 31, 2025, 96 patients were identified with active prescriptions for dual CGRP-targeting therapies. Of the 96 patients, 89 were included in the final analysis; 1 patient lacked a migraine diagnosis and 6 did not have a concomitant dual CGRP-targeted regimen and were excluded. The mean age of patients was 46.8 years and 54 (61%) were female. The most common migraine diagnosis was chronic migraine in 68 patients (76%). Triptans, ibuprofen, and acetaminophen were the most commonly used acute treatment medications (Table 2).

Safety Assessment

Many of the 89 unique patients trialed > 1 regimen. Thus, for the safety analysis, we analyzed 149 patients on unique dual CGRP-targeting regimens (Table 3). Ubrogepant was used by 126 patients (84.6%) for acute treatment. For preventive therapy, 63 patients (42.3%) used erenumab injections and 55 (36.9%) used fremanezumab injections. Seven patients (4.7%) reported AEs (Table 4). Five of the 7 AEs were noted in the package inserts.^25-32 One patient taking both atogepant and ubrogepant reported brain fog that resolved after a dose reduction of atogepant to every other day dosing. A patient taking fremanezumab and rimegepant reported myalgia/joint pain after the first fremanezumab injection, which resolved after a few days and did not recur during the study period.

Nine of 149 patient regimens (6.0%) were associated with changes in liver function tests or serum creatinine, though all but 1 were grade 1 (1 patient had a grade 2 ALT elevation). Twenty-five patients (16.8%) experienced changes in BP, most of which were category 1 elevations. Four patients had systolic or diastolic BP ≥ 160 mm Hg or 100 mm Hg, respectively (Table 5).

Efficacy Assessment

Of the 149 unique dual CGRP regimens, 59 were eligible for the exploratory efficacy analysis. Data were excluded from the efficacy analysis if patients had not been on a stable CGRP preventive migraine regimen for ≥ 12 weeks prior to the addition of a gepant. Fourteen regimens were excluded due to a lack of clear documentation on efficacy, leaving 45 analyzed regimens. Of the 45 regimens, 34 were from unique patients. There was no median change in migraine intensity or duration found in the efficacy analysis (0.0, P = .18, and 0.0, P = .92, respectively). Ten patients on dual CGRP therapy reported that the addition of a gepant for acute treatment improved their symptoms, 20 reported that their symptoms were unchanged and/or worsened, and 29 lacked documentation.

Discussion

This study aimed to describe the safety and efficacy of concomitant CGRP regimens for migraine prevention and acute treatment. To our knowledge, this was the first descriptive study of these agents in a veteran population. The potential for increased AEs with concomitant use of CGRP antagonists is due to the similarities in the mechanism of action between the agents, which both target the same receptor/ligand pathway. Given CGRP activity in both the gastrointestinal and cardiovascular systems, the potential for related AEs is speculative. Patient-reported AEs occurred in 7 of 149 unique treatment regimens reviewed for an incidence rate < 5%. All AEs were nonserious and self-limiting.

Our findings are consistent with available research. A 2024 retrospective, exploratory real-world study evaluating the safety and tolerability of combining CGRP-targeting mAbs with gepants reported findings consistent with our results. This analysis included adult patients treated with ≥ 1 previous anti-CGRP mAb and found that 234 of 516 patients included received a combination of a gepant in addition to a CGRP-targeting mAb. Of these 234 patients, 1.3% reported nonserious AEs.³³ Similarly, in a multicenter, open-label, long-term safety study in adults experiencing multiple monthly migraine attacks, a subgroup of 13 participants taking a stable dose of an anti-CGRP mAb also took rimegepant 75 mg as needed for acute treatment for 12 weeks. These patients experienced no serious AEs or any AEs leading to discontinuation.¹⁴ A study evaluating the drug-drug interaction, safety, and tolerability of dual therapy (atogepant 60 mg daily and ubrogepant 100 mg every 3 days) in 26 patients found no serious AEs, including no significant changes from baseline in laboratory results, vital signs, or safety-related 12-lead electrocardiogram parameters.¹⁵The TANDEM real-world, open-label, prospective study demonstrated similar results. It evaluated the safety and tolerability of concomitant use of ubrogepant and atogepant in patients with episodic migraines and found no increase in AEs when comparing atogepant alone with combination therapy. Twenty-six patients (9.9%) discontinued treatment due to AEs. The most common treatment-related AEs were constipation, nausea, decreased appetite, and fatigue. Efficacy data were also noted to be an exploratory endpoint in the TANDEM study; however, results have not been published.¹²

Within this safety analysis, new onset gastrointestinal AEs, specifically nausea, only occurred in 1 patient. Hypertension occurred in 25 treatment regimens (16.8%) for 21 unique patients (4 BP elevations occurred in 1 patient on 4 different regimens). However, the retrospective nature of reporting may limit accurate assessment. A closer analysis determined that elevated BP readings correlated with elevated pain scores at the time of the readings, which could have factored into the BP elevations. However, ongoing monitoring is needed due to an increased risk of hypertension, particularly given recent FDA labeling updates for CGRP-targeting therapies including gepants. In light of this, and the overall low incidence of hypertension reported, no new safety concerns were identified.

Limitations

Efficacy data in this project were exploratory. This evaluation did not show a significant difference in migraine intensity or duration after adding a gepant for acute treatment. The study was not powered to detect a significant difference. Limited data exist assessing efficacy outcomes with dual CGRP-targeting treatment regimens. The COURAGE study assessed the real-world effectiveness of ubrogepant and CGRP mAbs with or without the addition of onabotulinumtoxinA. The final analysis of the ubrogepant and CGRP mAb arm included 245 total patients and assessed meaningful migraine pain relief, restoration of normal function after a migraine, and treatment satisfaction. By hour 2, 61.6% of patients reported achieving migraine pain relief, rising to 80.4% by hour 4. Return to normal function occurred in 34.7% at hour 2 and 55.5% by hour 4.¹³ The long-term safety and efficacy of combining erenumab and rimegepant were described in a case series involving 2 patients. Both patients reported that the concomitant CGRP-targeted therapies were effective and reported no AEs.¹⁴

The retrospective design of this study meant that there was potential for limited documentation and introduction of bias into the results. Data were collected at a single VA health care system, and thus, results may not be generalizable to a broader population. However, the study population was consistent with the higher incidence of migraine expected in females in the general population. The sample size was limited, particularly in the exploratory efficacy endpoint assessment.

Limitations were observed due to inconsistent documentation regarding headache characteristics, making it challenging to draw meaningful conclusions from this data set. Additional confounding factors, including polypharmacy, nonadherence to medications, and comorbidities, may have skewed results. For example, while our study design required that the preventive CGRP-targeting medication be stable for 12 weeks for inclusion in further efficacy analysis, other medications commonly used for migraine prevention may have been adjusted (which was not accounted for in this analysis). Given this, more large-scale, placebo-controlled, randomized studies are needed to continue to assess the safety and efficacy of these combination treatment regimens.

Conclusions

Few AEs or safety events were reported with combination CGRP-targeting treatment for acute and preventive treatment of migraine. Those that were identified were considered mild. Efficacy data were limited, and further studies are needed to fully assess outcomes.

Calcitonin gene-related peptide (CGRP) is a neuropeptide that plays a key role in migraine pathophysiology by promoting the dilation of cerebral blood vessels and transmitting pain signals.¹ CGRP has generated interest for the prevention and acute treatment of migraine. Since 2018, 8 novel CGRP-targeting therapies have been approved by the US Food and Drug Administration (FDA) for the management of migraines.^2,3 For migraine prevention, there are 4 injectable monoclonal antibodies (mAbs) directed against the CGRP receptor (erenumab) or the CGRP ligand (fremanezumab, galcanezumab, and eptinezumab). There are also 2 oral small-molecule CGRP receptor antagonists, termed gepants, that also are approved for migraine prevention (atogepant and rimegepant). Three gepants are approved for acute migraine treatment and are administered orally (rimegepant and ubrogepant) or intranasally (zavegepant) (Table 1).

CGRP-targeting therapies have received attention for their role in vasodilation within the cerebral, coronary, and renal vasculature.⁴ CGRP-mediated vasodilatory effects cause systemic regulation of blood pressure (BP) and play a protective role in hypertension.² Some studies, particularly with erenumab, have shown that the inhibitory role of the agent leads to an increase in BP, as well as gastrointestinal issues such as constipation.^2,5 The FDA recently updated monitoring recommendations for all CGRP-targeting therapies to include the potential for BP elevations and hypertension. Outside of this, there is no definitive evidence linking dual CGRP-targeted therapy to higher cardiovascular or gastrointestinal risks and prescribing information does not carry contraindications.⁶

In a 2021 consensus statement, the American Headache Society (AHS) recommended CGRP-targeting therapies for migraine prevention after inability to tolerate or inadequate response to an 8-week trial of ≥ 2 drug classes including antihypertensives, antiseizure medications, antidepressants, and onabotulinumtoxinA.⁷ For acute treatment, AHS recommended gepant use after contraindication to or inadequate response to ≥ 2 triptans. Guidance on combination CGRP-targeting therapies for both prevention and acute treatment was not provided.⁷ More recently, the AHS published a position statement noting substantial efficacy and safety data for CGRP-targeting therapies and suggested its consideration as a first-line option for migraine prevention, though use for acute treatment or combination CGRP-targeting therapies for both prevention and acute treatment were not addressed.⁸

The International Headache Society guidelines for the acute treatment of migraines recommend nonopioid analgesics as first-line therapy for mild migraine attacks. For moderate to severe attacks, triptans with or without a nonopioid analgesic were recommended as first-line therapy, prior to consideration of CGRP-targeted therapy.⁹ The increased use of this new drug class has also led to combination use of CGRP-targeting therapies for migraine prevention and acute treatment as seen in clinical practice and reflected by some case reports, case series, and small studies describing such use.^10-14 In light of the similar mechanism of action of these therapies and the physiologic role of CGRP, there have been calls for safety evaluation.¹⁵

To our knowledge, no studies have evaluated dual CGRP-targeting regimens for migraine in the veteran population. In 2023, the US Department of Veterans Affairs (VA) and US Department of Defense (DoD) updated their clinical practice guidelines for the management of headache.³ For migraine prevention, the VA/DoD guidelines include a strong recommendation for the use of erenumab, fremanezumab, and galcanezumab; a weak recommendation for the use of atogepant; and a recommendation neither for nor against the use of rimegepant. For acute treatment, the guidelines assign a weak recommendation for the use of rimegepant and ubrogepant. Combination use was not addressed.³

Prior to the VA/DoD guidelines, the Veterans Health Administration restricted the dual use of CGRP-targeting therapies for both preventive and acute migraine treatment. However, the VA Pharmacy Benefit Management Service removed the restriction in the Criteria for Use documents, allowing broader access to these medications for veterans.^16-22 This change permits the use of CGRP-targeting drugs for both acute and preventive migraine treatment after initial data reflecting real-world case reports and open-label studies suggested possible efficacy without a clear safety concern.^11,12 This study aims to fill the gap in the literature by evaluating the safety, efficacy, and overall outcomes of combination CGRP-targeting treatment for migraine prevention and acute treatment in a veteran population.

Methods

This single-center, retrospective, medication use evaluation at the Ralph H. Johnson VA Medical Center (RHJVAMC) was reviewed by the RHJVAMC Research and Development Committee and Quality Improvement Program Evaluation Self Certification Tool, which both determined that institutional review board approval was not required because it was considered part of routine care and quality improvement. Computerized Patient Record System (CPRS) data were reviewed between April 1, 2023 (after the Criteria for Use for CGRP-targeting therapies was updated), through January 31, 2025. Patients were included if they had a confirmed diagnosis of migraine using the International Classification of Headache Disorders, 3rd edition criteria and had concomitant active prescriptions for both a preventive and acute treatment CGRP-targeting agent during the project period.²³ Only patients receiving care from the RHJVAMC neurology department were included.

The primary objective was to assess the safety of dual CGRP-targeting therapies for migraine treatment. Key safety endpoints included effects on liver function, kidney function, and BP. Safety outcomes were graded using Common Terminology Criteria for Adverse Events.²⁴ Changes in liver function were categorized as grade 1, 2, or 3 elevations: grade 1 (aspartate aminotransferase [AST]/alanine aminotransferase [ALT] up to 3x the upper limit of normal [ULN] or bilirubin > 1.5 x ULN); grade 2 (AST/ALT 3-6 x ULN or bilirubin 1.5-3 x ULN); and grade 3 (AST/ALT 5-10 x ULN or bilirubin 3-10 x ULN). Kidney function changes were assessed by serum creatinine levels using a similar grading system: Grade 1 (≤ 1.5 x ULN); grade 2 (1.5-3 x baseline of normal); and grade 3 (3-6 x ULN or baseline). Changes in BP were monitored from baseline to the time of the first neurology follow-up. Elevations were grouped into 2 categories, defined as BP ≥ 140 mm Hg systolic and/or 90 mm Hg diastolic (category 1) and ≥ 160 mm Hg systolic and/or 100 mm Hg diastolic (category 2). Neurology documentation was also reviewed in CPRS for individual patient-reported adverse effects (AEs). Safety endpoints were tracked for any occurrence during the project period.

The secondary objective was to describe the patient-reported efficacy of adding a gepant for acute migraine treatment to existing CGRP-targeting therapies for migraine prevention, in those patients who were stable for ≥ 12 weeks on the preventive therapy. Neurology documentation of headache characteristics, including headache severity as rated on a numerical pain score from 0 (no pain) to 10 (worst pain), and duration of headaches (in hours) were recorded during the project period. Changes in headache characteristics were tracked from baseline (ie, the neurology visit when the gepant was first requested) to the first neurology follow-up within 6 months of initiating gepant for acute treatment. If ranges were provided within documentation, a mean was calculated and used for data collection. Neurology documentation was also reviewed for any patient report of overall effectiveness with the added gepant, and categorized as symptoms improved, worsened, or did not change based on subjective report. Descriptive statistics were used for data analysis. A 1-sample Wilcoxon signed rank test was performed as an exploratory analysis for change in headache characteristics from baseline to first neurology follow-up within 6 months. Each individual CGRP regimen was counted as a unique data point to adequately describe changes associated with each new medication and/or dose adjustment. Therefore, patients could be included more than once to account for each distinct treatment regimen.

Results

From April 1, 2023, to January 31, 2025, 96 patients were identified with active prescriptions for dual CGRP-targeting therapies. Of the 96 patients, 89 were included in the final analysis; 1 patient lacked a migraine diagnosis and 6 did not have a concomitant dual CGRP-targeted regimen and were excluded. The mean age of patients was 46.8 years and 54 (61%) were female. The most common migraine diagnosis was chronic migraine in 68 patients (76%). Triptans, ibuprofen, and acetaminophen were the most commonly used acute treatment medications (Table 2).

Safety Assessment

Many of the 89 unique patients trialed > 1 regimen. Thus, for the safety analysis, we analyzed 149 patients on unique dual CGRP-targeting regimens (Table 3). Ubrogepant was used by 126 patients (84.6%) for acute treatment. For preventive therapy, 63 patients (42.3%) used erenumab injections and 55 (36.9%) used fremanezumab injections. Seven patients (4.7%) reported AEs (Table 4). Five of the 7 AEs were noted in the package inserts.^25-32 One patient taking both atogepant and ubrogepant reported brain fog that resolved after a dose reduction of atogepant to every other day dosing. A patient taking fremanezumab and rimegepant reported myalgia/joint pain after the first fremanezumab injection, which resolved after a few days and did not recur during the study period.

Nine of 149 patient regimens (6.0%) were associated with changes in liver function tests or serum creatinine, though all but 1 were grade 1 (1 patient had a grade 2 ALT elevation). Twenty-five patients (16.8%) experienced changes in BP, most of which were category 1 elevations. Four patients had systolic or diastolic BP ≥ 160 mm Hg or 100 mm Hg, respectively (Table 5).

Efficacy Assessment

Of the 149 unique dual CGRP regimens, 59 were eligible for the exploratory efficacy analysis. Data were excluded from the efficacy analysis if patients had not been on a stable CGRP preventive migraine regimen for ≥ 12 weeks prior to the addition of a gepant. Fourteen regimens were excluded due to a lack of clear documentation on efficacy, leaving 45 analyzed regimens. Of the 45 regimens, 34 were from unique patients. There was no median change in migraine intensity or duration found in the efficacy analysis (0.0, P = .18, and 0.0, P = .92, respectively). Ten patients on dual CGRP therapy reported that the addition of a gepant for acute treatment improved their symptoms, 20 reported that their symptoms were unchanged and/or worsened, and 29 lacked documentation.

Discussion

This study aimed to describe the safety and efficacy of concomitant CGRP regimens for migraine prevention and acute treatment. To our knowledge, this was the first descriptive study of these agents in a veteran population. The potential for increased AEs with concomitant use of CGRP antagonists is due to the similarities in the mechanism of action between the agents, which both target the same receptor/ligand pathway. Given CGRP activity in both the gastrointestinal and cardiovascular systems, the potential for related AEs is speculative. Patient-reported AEs occurred in 7 of 149 unique treatment regimens reviewed for an incidence rate < 5%. All AEs were nonserious and self-limiting.

Our findings are consistent with available research. A 2024 retrospective, exploratory real-world study evaluating the safety and tolerability of combining CGRP-targeting mAbs with gepants reported findings consistent with our results. This analysis included adult patients treated with ≥ 1 previous anti-CGRP mAb and found that 234 of 516 patients included received a combination of a gepant in addition to a CGRP-targeting mAb. Of these 234 patients, 1.3% reported nonserious AEs.³³ Similarly, in a multicenter, open-label, long-term safety study in adults experiencing multiple monthly migraine attacks, a subgroup of 13 participants taking a stable dose of an anti-CGRP mAb also took rimegepant 75 mg as needed for acute treatment for 12 weeks. These patients experienced no serious AEs or any AEs leading to discontinuation.¹⁴ A study evaluating the drug-drug interaction, safety, and tolerability of dual therapy (atogepant 60 mg daily and ubrogepant 100 mg every 3 days) in 26 patients found no serious AEs, including no significant changes from baseline in laboratory results, vital signs, or safety-related 12-lead electrocardiogram parameters.¹⁵The TANDEM real-world, open-label, prospective study demonstrated similar results. It evaluated the safety and tolerability of concomitant use of ubrogepant and atogepant in patients with episodic migraines and found no increase in AEs when comparing atogepant alone with combination therapy. Twenty-six patients (9.9%) discontinued treatment due to AEs. The most common treatment-related AEs were constipation, nausea, decreased appetite, and fatigue. Efficacy data were also noted to be an exploratory endpoint in the TANDEM study; however, results have not been published.¹²

Within this safety analysis, new onset gastrointestinal AEs, specifically nausea, only occurred in 1 patient. Hypertension occurred in 25 treatment regimens (16.8%) for 21 unique patients (4 BP elevations occurred in 1 patient on 4 different regimens). However, the retrospective nature of reporting may limit accurate assessment. A closer analysis determined that elevated BP readings correlated with elevated pain scores at the time of the readings, which could have factored into the BP elevations. However, ongoing monitoring is needed due to an increased risk of hypertension, particularly given recent FDA labeling updates for CGRP-targeting therapies including gepants. In light of this, and the overall low incidence of hypertension reported, no new safety concerns were identified.

Limitations

Efficacy data in this project were exploratory. This evaluation did not show a significant difference in migraine intensity or duration after adding a gepant for acute treatment. The study was not powered to detect a significant difference. Limited data exist assessing efficacy outcomes with dual CGRP-targeting treatment regimens. The COURAGE study assessed the real-world effectiveness of ubrogepant and CGRP mAbs with or without the addition of onabotulinumtoxinA. The final analysis of the ubrogepant and CGRP mAb arm included 245 total patients and assessed meaningful migraine pain relief, restoration of normal function after a migraine, and treatment satisfaction. By hour 2, 61.6% of patients reported achieving migraine pain relief, rising to 80.4% by hour 4. Return to normal function occurred in 34.7% at hour 2 and 55.5% by hour 4.¹³ The long-term safety and efficacy of combining erenumab and rimegepant were described in a case series involving 2 patients. Both patients reported that the concomitant CGRP-targeted therapies were effective and reported no AEs.¹⁴

The retrospective design of this study meant that there was potential for limited documentation and introduction of bias into the results. Data were collected at a single VA health care system, and thus, results may not be generalizable to a broader population. However, the study population was consistent with the higher incidence of migraine expected in females in the general population. The sample size was limited, particularly in the exploratory efficacy endpoint assessment.

Limitations were observed due to inconsistent documentation regarding headache characteristics, making it challenging to draw meaningful conclusions from this data set. Additional confounding factors, including polypharmacy, nonadherence to medications, and comorbidities, may have skewed results. For example, while our study design required that the preventive CGRP-targeting medication be stable for 12 weeks for inclusion in further efficacy analysis, other medications commonly used for migraine prevention may have been adjusted (which was not accounted for in this analysis). Given this, more large-scale, placebo-controlled, randomized studies are needed to continue to assess the safety and efficacy of these combination treatment regimens.

Conclusions

Few AEs or safety events were reported with combination CGRP-targeting treatment for acute and preventive treatment of migraine. Those that were identified were considered mild. Efficacy data were limited, and further studies are needed to fully assess outcomes.

Calcitonin gene-related peptide (CGRP) is a neuropeptide that plays a key role in migraine pathophysiology by promoting the dilation of cerebral blood vessels and transmitting pain signals.¹ CGRP has generated interest for the prevention and acute treatment of migraine. Since 2018, 8 novel CGRP-targeting therapies have been approved by the US Food and Drug Administration (FDA) for the management of migraines.^2,3 For migraine prevention, there are 4 injectable monoclonal antibodies (mAbs) directed against the CGRP receptor (erenumab) or the CGRP ligand (fremanezumab, galcanezumab, and eptinezumab). There are also 2 oral small-molecule CGRP receptor antagonists, termed gepants, that also are approved for migraine prevention (atogepant and rimegepant). Three gepants are approved for acute migraine treatment and are administered orally (rimegepant and ubrogepant) or intranasally (zavegepant) (Table 1).

CGRP-targeting therapies have received attention for their role in vasodilation within the cerebral, coronary, and renal vasculature.⁴ CGRP-mediated vasodilatory effects cause systemic regulation of blood pressure (BP) and play a protective role in hypertension.² Some studies, particularly with erenumab, have shown that the inhibitory role of the agent leads to an increase in BP, as well as gastrointestinal issues such as constipation.^2,5 The FDA recently updated monitoring recommendations for all CGRP-targeting therapies to include the potential for BP elevations and hypertension. Outside of this, there is no definitive evidence linking dual CGRP-targeted therapy to higher cardiovascular or gastrointestinal risks and prescribing information does not carry contraindications.⁶

In a 2021 consensus statement, the American Headache Society (AHS) recommended CGRP-targeting therapies for migraine prevention after inability to tolerate or inadequate response to an 8-week trial of ≥ 2 drug classes including antihypertensives, antiseizure medications, antidepressants, and onabotulinumtoxinA.⁷ For acute treatment, AHS recommended gepant use after contraindication to or inadequate response to ≥ 2 triptans. Guidance on combination CGRP-targeting therapies for both prevention and acute treatment was not provided.⁷ More recently, the AHS published a position statement noting substantial efficacy and safety data for CGRP-targeting therapies and suggested its consideration as a first-line option for migraine prevention, though use for acute treatment or combination CGRP-targeting therapies for both prevention and acute treatment were not addressed.⁸

The International Headache Society guidelines for the acute treatment of migraines recommend nonopioid analgesics as first-line therapy for mild migraine attacks. For moderate to severe attacks, triptans with or without a nonopioid analgesic were recommended as first-line therapy, prior to consideration of CGRP-targeted therapy.⁹ The increased use of this new drug class has also led to combination use of CGRP-targeting therapies for migraine prevention and acute treatment as seen in clinical practice and reflected by some case reports, case series, and small studies describing such use.^10-14 In light of the similar mechanism of action of these therapies and the physiologic role of CGRP, there have been calls for safety evaluation.¹⁵

To our knowledge, no studies have evaluated dual CGRP-targeting regimens for migraine in the veteran population. In 2023, the US Department of Veterans Affairs (VA) and US Department of Defense (DoD) updated their clinical practice guidelines for the management of headache.³ For migraine prevention, the VA/DoD guidelines include a strong recommendation for the use of erenumab, fremanezumab, and galcanezumab; a weak recommendation for the use of atogepant; and a recommendation neither for nor against the use of rimegepant. For acute treatment, the guidelines assign a weak recommendation for the use of rimegepant and ubrogepant. Combination use was not addressed.³

Prior to the VA/DoD guidelines, the Veterans Health Administration restricted the dual use of CGRP-targeting therapies for both preventive and acute migraine treatment. However, the VA Pharmacy Benefit Management Service removed the restriction in the Criteria for Use documents, allowing broader access to these medications for veterans.^16-22 This change permits the use of CGRP-targeting drugs for both acute and preventive migraine treatment after initial data reflecting real-world case reports and open-label studies suggested possible efficacy without a clear safety concern.^11,12 This study aims to fill the gap in the literature by evaluating the safety, efficacy, and overall outcomes of combination CGRP-targeting treatment for migraine prevention and acute treatment in a veteran population.

Methods

This single-center, retrospective, medication use evaluation at the Ralph H. Johnson VA Medical Center (RHJVAMC) was reviewed by the RHJVAMC Research and Development Committee and Quality Improvement Program Evaluation Self Certification Tool, which both determined that institutional review board approval was not required because it was considered part of routine care and quality improvement. Computerized Patient Record System (CPRS) data were reviewed between April 1, 2023 (after the Criteria for Use for CGRP-targeting therapies was updated), through January 31, 2025. Patients were included if they had a confirmed diagnosis of migraine using the International Classification of Headache Disorders, 3rd edition criteria and had concomitant active prescriptions for both a preventive and acute treatment CGRP-targeting agent during the project period.²³ Only patients receiving care from the RHJVAMC neurology department were included.

The primary objective was to assess the safety of dual CGRP-targeting therapies for migraine treatment. Key safety endpoints included effects on liver function, kidney function, and BP. Safety outcomes were graded using Common Terminology Criteria for Adverse Events.²⁴ Changes in liver function were categorized as grade 1, 2, or 3 elevations: grade 1 (aspartate aminotransferase [AST]/alanine aminotransferase [ALT] up to 3x the upper limit of normal [ULN] or bilirubin > 1.5 x ULN); grade 2 (AST/ALT 3-6 x ULN or bilirubin 1.5-3 x ULN); and grade 3 (AST/ALT 5-10 x ULN or bilirubin 3-10 x ULN). Kidney function changes were assessed by serum creatinine levels using a similar grading system: Grade 1 (≤ 1.5 x ULN); grade 2 (1.5-3 x baseline of normal); and grade 3 (3-6 x ULN or baseline). Changes in BP were monitored from baseline to the time of the first neurology follow-up. Elevations were grouped into 2 categories, defined as BP ≥ 140 mm Hg systolic and/or 90 mm Hg diastolic (category 1) and ≥ 160 mm Hg systolic and/or 100 mm Hg diastolic (category 2). Neurology documentation was also reviewed in CPRS for individual patient-reported adverse effects (AEs). Safety endpoints were tracked for any occurrence during the project period.

The secondary objective was to describe the patient-reported efficacy of adding a gepant for acute migraine treatment to existing CGRP-targeting therapies for migraine prevention, in those patients who were stable for ≥ 12 weeks on the preventive therapy. Neurology documentation of headache characteristics, including headache severity as rated on a numerical pain score from 0 (no pain) to 10 (worst pain), and duration of headaches (in hours) were recorded during the project period. Changes in headache characteristics were tracked from baseline (ie, the neurology visit when the gepant was first requested) to the first neurology follow-up within 6 months of initiating gepant for acute treatment. If ranges were provided within documentation, a mean was calculated and used for data collection. Neurology documentation was also reviewed for any patient report of overall effectiveness with the added gepant, and categorized as symptoms improved, worsened, or did not change based on subjective report. Descriptive statistics were used for data analysis. A 1-sample Wilcoxon signed rank test was performed as an exploratory analysis for change in headache characteristics from baseline to first neurology follow-up within 6 months. Each individual CGRP regimen was counted as a unique data point to adequately describe changes associated with each new medication and/or dose adjustment. Therefore, patients could be included more than once to account for each distinct treatment regimen.

Results

From April 1, 2023, to January 31, 2025, 96 patients were identified with active prescriptions for dual CGRP-targeting therapies. Of the 96 patients, 89 were included in the final analysis; 1 patient lacked a migraine diagnosis and 6 did not have a concomitant dual CGRP-targeted regimen and were excluded. The mean age of patients was 46.8 years and 54 (61%) were female. The most common migraine diagnosis was chronic migraine in 68 patients (76%). Triptans, ibuprofen, and acetaminophen were the most commonly used acute treatment medications (Table 2).

Safety Assessment

Many of the 89 unique patients trialed > 1 regimen. Thus, for the safety analysis, we analyzed 149 patients on unique dual CGRP-targeting regimens (Table 3). Ubrogepant was used by 126 patients (84.6%) for acute treatment. For preventive therapy, 63 patients (42.3%) used erenumab injections and 55 (36.9%) used fremanezumab injections. Seven patients (4.7%) reported AEs (Table 4). Five of the 7 AEs were noted in the package inserts.^25-32 One patient taking both atogepant and ubrogepant reported brain fog that resolved after a dose reduction of atogepant to every other day dosing. A patient taking fremanezumab and rimegepant reported myalgia/joint pain after the first fremanezumab injection, which resolved after a few days and did not recur during the study period.

Nine of 149 patient regimens (6.0%) were associated with changes in liver function tests or serum creatinine, though all but 1 were grade 1 (1 patient had a grade 2 ALT elevation). Twenty-five patients (16.8%) experienced changes in BP, most of which were category 1 elevations. Four patients had systolic or diastolic BP ≥ 160 mm Hg or 100 mm Hg, respectively (Table 5).

Efficacy Assessment

Of the 149 unique dual CGRP regimens, 59 were eligible for the exploratory efficacy analysis. Data were excluded from the efficacy analysis if patients had not been on a stable CGRP preventive migraine regimen for ≥ 12 weeks prior to the addition of a gepant. Fourteen regimens were excluded due to a lack of clear documentation on efficacy, leaving 45 analyzed regimens. Of the 45 regimens, 34 were from unique patients. There was no median change in migraine intensity or duration found in the efficacy analysis (0.0, P = .18, and 0.0, P = .92, respectively). Ten patients on dual CGRP therapy reported that the addition of a gepant for acute treatment improved their symptoms, 20 reported that their symptoms were unchanged and/or worsened, and 29 lacked documentation.

Discussion

This study aimed to describe the safety and efficacy of concomitant CGRP regimens for migraine prevention and acute treatment. To our knowledge, this was the first descriptive study of these agents in a veteran population. The potential for increased AEs with concomitant use of CGRP antagonists is due to the similarities in the mechanism of action between the agents, which both target the same receptor/ligand pathway. Given CGRP activity in both the gastrointestinal and cardiovascular systems, the potential for related AEs is speculative. Patient-reported AEs occurred in 7 of 149 unique treatment regimens reviewed for an incidence rate < 5%. All AEs were nonserious and self-limiting.

Our findings are consistent with available research. A 2024 retrospective, exploratory real-world study evaluating the safety and tolerability of combining CGRP-targeting mAbs with gepants reported findings consistent with our results. This analysis included adult patients treated with ≥ 1 previous anti-CGRP mAb and found that 234 of 516 patients included received a combination of a gepant in addition to a CGRP-targeting mAb. Of these 234 patients, 1.3% reported nonserious AEs.³³ Similarly, in a multicenter, open-label, long-term safety study in adults experiencing multiple monthly migraine attacks, a subgroup of 13 participants taking a stable dose of an anti-CGRP mAb also took rimegepant 75 mg as needed for acute treatment for 12 weeks. These patients experienced no serious AEs or any AEs leading to discontinuation.¹⁴ A study evaluating the drug-drug interaction, safety, and tolerability of dual therapy (atogepant 60 mg daily and ubrogepant 100 mg every 3 days) in 26 patients found no serious AEs, including no significant changes from baseline in laboratory results, vital signs, or safety-related 12-lead electrocardiogram parameters.¹⁵The TANDEM real-world, open-label, prospective study demonstrated similar results. It evaluated the safety and tolerability of concomitant use of ubrogepant and atogepant in patients with episodic migraines and found no increase in AEs when comparing atogepant alone with combination therapy. Twenty-six patients (9.9%) discontinued treatment due to AEs. The most common treatment-related AEs were constipation, nausea, decreased appetite, and fatigue. Efficacy data were also noted to be an exploratory endpoint in the TANDEM study; however, results have not been published.¹²

Within this safety analysis, new onset gastrointestinal AEs, specifically nausea, only occurred in 1 patient. Hypertension occurred in 25 treatment regimens (16.8%) for 21 unique patients (4 BP elevations occurred in 1 patient on 4 different regimens). However, the retrospective nature of reporting may limit accurate assessment. A closer analysis determined that elevated BP readings correlated with elevated pain scores at the time of the readings, which could have factored into the BP elevations. However, ongoing monitoring is needed due to an increased risk of hypertension, particularly given recent FDA labeling updates for CGRP-targeting therapies including gepants. In light of this, and the overall low incidence of hypertension reported, no new safety concerns were identified.

Limitations

Efficacy data in this project were exploratory. This evaluation did not show a significant difference in migraine intensity or duration after adding a gepant for acute treatment. The study was not powered to detect a significant difference. Limited data exist assessing efficacy outcomes with dual CGRP-targeting treatment regimens. The COURAGE study assessed the real-world effectiveness of ubrogepant and CGRP mAbs with or without the addition of onabotulinumtoxinA. The final analysis of the ubrogepant and CGRP mAb arm included 245 total patients and assessed meaningful migraine pain relief, restoration of normal function after a migraine, and treatment satisfaction. By hour 2, 61.6% of patients reported achieving migraine pain relief, rising to 80.4% by hour 4. Return to normal function occurred in 34.7% at hour 2 and 55.5% by hour 4.¹³ The long-term safety and efficacy of combining erenumab and rimegepant were described in a case series involving 2 patients. Both patients reported that the concomitant CGRP-targeted therapies were effective and reported no AEs.¹⁴

The retrospective design of this study meant that there was potential for limited documentation and introduction of bias into the results. Data were collected at a single VA health care system, and thus, results may not be generalizable to a broader population. However, the study population was consistent with the higher incidence of migraine expected in females in the general population. The sample size was limited, particularly in the exploratory efficacy endpoint assessment.

Limitations were observed due to inconsistent documentation regarding headache characteristics, making it challenging to draw meaningful conclusions from this data set. Additional confounding factors, including polypharmacy, nonadherence to medications, and comorbidities, may have skewed results. For example, while our study design required that the preventive CGRP-targeting medication be stable for 12 weeks for inclusion in further efficacy analysis, other medications commonly used for migraine prevention may have been adjusted (which was not accounted for in this analysis). Given this, more large-scale, placebo-controlled, randomized studies are needed to continue to assess the safety and efficacy of these combination treatment regimens.

Conclusions

Few AEs or safety events were reported with combination CGRP-targeting treatment for acute and preventive treatment of migraine. Those that were identified were considered mild. Efficacy data were limited, and further studies are needed to fully assess outcomes.

In leveraging existing, readily available evidence-based health care information (eg, systematic reviews, clinical practice guidelines), clinicians have historically made recommendations based on treatment responses of the average patient.¹ Recently, this approach has been expanded into data-driven, evidence-based precision medical care for individuals across a wide range of disciplines and care settings. These precision medicine approaches use information related to an individual’s genes, environment, and lifestyle to tailor recommendations regarding prevention, diagnosis, and treatment.

Applying precision medicine approaches to the unique exposures and experiences of service members and veterans—particularly those who served in combat environments—through the incorporation of biopsychosocial factors into medical decision-making may be even more pertinent. This sentiment is reflected in Section 305 of the Commander John Scott Hannon Veterans Mental Health Care Improvement Act of 2019, which outlines the Precision Medicine Initiative of the US Department of Veterans Affairs (VA) to identify and validate brain and mental health biomarkers.² Despite widespread consensus regarding the promise of precision medicine, large, rich datasets with elements pertaining to common military exposures such as traumatic brain injury (TBI) and posttraumatic stress disorder (PTSD) are limited.

Existing datasets, most of which are relatively small or focus on specific cohorts (eg, older veterans, transitioning veterans), continue to create barriers to advancing precision medicine. For example, in classically designed clinical trials, analyses are generally conducted in a manner that may obfuscate efficacy among subcohorts of individuals, thereby underscoring the need to explore alternative strategies to unify existing datasets capable of revealing such heterogeneity.³ The evidence base for precision medical care is limited, drawing from published trials with relatively small sample sizes and even larger cohort studies have limited biomarker data. Additionally, these models are often exploratory during development, and to avoid statistical overfitting of an exploratory model, validation in similar datasets is needed—an added burden when data sources are small or underpowered to begin with.

A promising approach is to combine and harmonize the largest, most deeply characterized data sources from similar samples. Although combining such datasets may appear to require minimal time and effort, harmonizing similar variables in an evidence-based and replicable manner requires time and expertise, even when participant characteristics and outcomes are similar.^4-7

Challenges related to harmonization are related to the wide range of strategies (eg, self-report questionnaires, clinical interviews, electronic health record review) used to measure common brain and mental health constructs, such as depression. Even when similar methods (eg, self-report measures) are implemented, challenges persist. For example, if a study used a depression measure that focused primarily on cognitive symptoms (eg, pessimism, self-dislike, suicidal ideation) and another study used a depression measure composed of items more heavily weighted towards somatic symptoms (eg, insomnia, loss of appetite, weight loss, decreased libido), combining their data could be challenging, particularly if researchers, clinicians, or administrators are interested in more than dichotomous outcomes (eg, depression vs no depression).^8,9

To address this knowledge gap and harmonize multimodal data from varied sources, well-planned and reproducible curation is needed. Longitudinal cohort studies of service members and veterans with military combat and training exposure histories provide researchers and other stakeholders access to extant biopsychosocial data shown to affect risk for adverse health outcomes; however, efforts to facilitate individually tailored treatment or other precision medicine approaches would benefit from the synthesis of such datasets.¹⁰

Members of the VA Total Brain Diagnostics (TBD) team are engaged in harmonizing variables from the Long-Term Impact of Military-Relevant Brain Injury Consortium–Chronic Effects of Neurotrauma Consortium (LIMBIC-CENC)¹¹ and the Translational Research Center for TBI and Stress Disorders (TRACTS).^12-21 While there is overlap across LIMBIC-CENC and TRACTS with respect to data domains, considerable data harmonization is needed to allow for future valid and meaningful analyses, particularly those involving multivariable predictors.

Data Sources

Both data sources for the TBD harmonization project, LIMBIC-CENC and TRACTS, include extensive, longitudinal data collected from relatively large cohorts of veterans and service members with combat exposure. Both studies collect detailed data related to potential brain injury history and include participants with and without a history of TBI. Similarly, both include extensive collection of fluid biomarkers and imaging data, as well as measures of biopsychosocial functioning.

Data collection sites for LIMBIC-CENC include 16 recruitment sites, 9 at VA medical centers (Richmond, Houston, Tampa, San Antonio, Portland, Minneapolis, Boston, Salisbury, San Diego) and 7 at military treatment sites (Alexandria, San Diego, Tampa, Tacoma, Columbia, Coronado, Hinesville), in addition to 11 assessment sites (Richmond, Houston, Tampa, San Antonio, Portland, Minneapolis, Boston, Salisbury, San Diego, Alexandria, Augusta). Data for TRACTS are collected at sites in Boston and Houston.

LIMBIC-CENC is a 12-year, 17-site cohort of service members and veteran participants with combat exposure who are well characterized at baseline and undergo annual reassessments. As of December 2025, > 3100 participants have been recruited, and nearly 90% remain in follow-up. Data collection includes > 6200 annual follow-up evaluations and > 1550 5-year re-evaluations, with 400 enrolled participants followed up annually.

TRACTS is a 16-year, 2-site cohort of veterans with combat exposure who complete comprehensive assessments at enrollment, undergo annual reassessments, and complete comprehensive reassessment every 5 years thereafter. As of December 2025, > 1075 participants have completed baseline (Time 1) assessments, > 600 have completed the 2-year re-evaluation (Time 2), > 175 have completed the 5-year re-evaluation (Time 3), and > 35 have completed 10-year evaluations (Time 4), with about 50 new participants added and 100 enrolled participants followed up annually. More data on participant characteristics are available for both LIMBIC-CENC and TRACTS in previous publications.^11,22These 2 ongoing, prospective, longitudinal cohorts of service members and veterans offer access to a wide range of potential risk factors that can affect response to care and outcomes, including demographics (eg, age, sex), injury characteristics (eg, pre-exposure factors, exposure factors), biomarkers (eg, serum, saliva, brain imaging, evoked potentials), and functional measures (eg, computerized posturography, computerized eye tracking, sensory testing, clinical examination, neuropsychological assessments, symptom questionnaires).

Harmonization Strategy

Pooling and harmonizing data from large studies evaluating similar participant cohorts and conditions involves numerous steps to appropriately handle a variety of measurements and disparate variable names. The TBD team adapted a model data harmonization system developed by O’Neil et al through initial work harmonizing the Federal Interagency Traumatic Brain Injury Research Informatics System (FITBIR).^4-7 This process was expanded and generalized by the research team to combine data from LIMBIC-CENC and TRACTS to create a single pooled dataset for analysis (Figure).

This approach was selected because it accommodates heterogeneous study designs (eg, cross-sectional, longitudinal, case-control), data collection methods (eg, clinical assessment, self-reported, objective blood, and imaging biomarkers), and various assessments of the same construct (ie, different measures of brain injury). While exact matches for data collection methods and measures may be easily harmonized, the timing of assessment, number of assessments, assessment tool version, and other factors must be considered. The goal was to harmonize data from LIMBIC-CENC and TRACTS to allow additional data sources to be harmonized and incorporated in the future.

Original data files from each study were reshaped to represent participant-level observations with 1 unique measurement per row. The measurement represents what information was collected and the value recorded represents the unique observation. These data are linked to metadata from the original study, which includes the study’s definition of each measurement, how it was collected, and any available information regarding when it was collected in reference to study enrollment or injury. Additional information on the file source, row, and column position of each data point was added to enable recreation of the original data as needed.

The resulting dataset was used to harmonize measurements from LIMBIC-CENC and TRACTS into a priori-defined schemas for brain- and mental health-relevant concepts, including TBI severity, PTSD, substance use, depression, suicidal ideation, and functioning (including cognitive, physical, and social functioning). This process was facilitated using natural language processing (NLP). Each study uniquely defines all measurements and provides written definitions with the data. Measurement definitions serve as records describing what was collected, how it was collected, and how the study may have uniquely defined information for its purposes. For example, definitions of exposure to brain injury and severity of brain injury may differ between studies, and the study-provided definition defines these differences.

Definitions were converted into numeric vectors through sentence embedding, a process that preserves the semantic meaning of the definition.²³ Cosine similarity was used as the primary metric to compare the semantic textual similarity between pairs of measurement definitions. Cosine similarity ranges from 0 to 1, where 0 indicates no meaningful similarity and 1 indicates they have identical meanings.²⁴ This approach leverages the relationship between the definitions of each measurement provided by a study and enables quick comparison of all pairwise combinations of measurement definitions between studies.

Subsets of similar measurements across studies were organized into a priori-defined schema. Clinical experts then reviewed each schema and further refined them into domains, (eg, mechanism of injury, clinical signs, acute symptoms) and subdomains (children), such as loss of consciousness, amnesia, and alteration of consciousness. This approach allows efficient handling of 2 specific cases that commonly occur when pooling and harmonizing datasets: (1) identifying the same measurement with differing names; and (2) identifying different measurements with definitions that each relate to the same domain.

The Table provides a general example of the schema for TBI severity. This was an iterative process in which clinical experts reviewed study-defined measurement definitions to develop general harmonized domains, and NLP techniques facilitated and accelerated identification and organization of measurements within these domains.

Expected Impact

Harmonization combining LIMBIC-CENC and TRACTS datasets is ongoing. Preliminary descriptive analyses of baseline cohort data indicate that harmonization across data sources is appropriate, given the lack of significant heterogeneity across sites and studies for most domains. Work by members of the TBD team is expected to lay the foundation for the use of existing and ongoing prospective, longitudinal datasets (eg, LIMBIC-CENC, TRACTS) and linked large datasets (eg, VA Informatics and Computing Infrastructure including electronic health records, VA Million Veteran Program, DaVINCI [US Department of Defense and VA Infrastructure for Clinical Intelligence]) to generate generalizable, clinically relevant information to advance precision brain and mental health care among service members and veterans.

By enhancing existing practice, this synthesized dataset has the potential to inform tailored and personalized medicine approaches designed to meet the needs of veterans and service members. These data will serve as the starting point for multivariable models examining the intersection of physiologic, behavioral, and environmental factors. The goal of this data harmonization effort is to better elucidate how clinicians and researchers can select optimal approaches for veterans and service members with TBI histories by accounting for a comprehensive set of physiologic, behavioral, and environmental factors in an individually tailored manner. These data may further extend existing clinical practice guideline approaches, inform shared decision-making, and enhance functional outcomes beyond those currently available.

Conclusions

Individuals who have served in the military have unique biopsychosocial exposures that are associated with brain and mental health disorders. To address these needs, the nationwide TBD team has initiated the creation of a unified, longitudinal dataset that includes harmonized measures from existing LIMBIC-CENC and TRACTS protocols. Initial data harmonization efforts are required to facilitate precision prognostics, diagnostics, and tailored interventions, with the goal of improving veterans’ brain and mental health and psychosocial functioning and enabling tailored and evidence-informed, individualized clinical care.

In leveraging existing, readily available evidence-based health care information (eg, systematic reviews, clinical practice guidelines), clinicians have historically made recommendations based on treatment responses of the average patient.¹ Recently, this approach has been expanded into data-driven, evidence-based precision medical care for individuals across a wide range of disciplines and care settings. These precision medicine approaches use information related to an individual’s genes, environment, and lifestyle to tailor recommendations regarding prevention, diagnosis, and treatment.

Applying precision medicine approaches to the unique exposures and experiences of service members and veterans—particularly those who served in combat environments—through the incorporation of biopsychosocial factors into medical decision-making may be even more pertinent. This sentiment is reflected in Section 305 of the Commander John Scott Hannon Veterans Mental Health Care Improvement Act of 2019, which outlines the Precision Medicine Initiative of the US Department of Veterans Affairs (VA) to identify and validate brain and mental health biomarkers.² Despite widespread consensus regarding the promise of precision medicine, large, rich datasets with elements pertaining to common military exposures such as traumatic brain injury (TBI) and posttraumatic stress disorder (PTSD) are limited.

Existing datasets, most of which are relatively small or focus on specific cohorts (eg, older veterans, transitioning veterans), continue to create barriers to advancing precision medicine. For example, in classically designed clinical trials, analyses are generally conducted in a manner that may obfuscate efficacy among subcohorts of individuals, thereby underscoring the need to explore alternative strategies to unify existing datasets capable of revealing such heterogeneity.³ The evidence base for precision medical care is limited, drawing from published trials with relatively small sample sizes and even larger cohort studies have limited biomarker data. Additionally, these models are often exploratory during development, and to avoid statistical overfitting of an exploratory model, validation in similar datasets is needed—an added burden when data sources are small or underpowered to begin with.

A promising approach is to combine and harmonize the largest, most deeply characterized data sources from similar samples. Although combining such datasets may appear to require minimal time and effort, harmonizing similar variables in an evidence-based and replicable manner requires time and expertise, even when participant characteristics and outcomes are similar.^4-7

Challenges related to harmonization are related to the wide range of strategies (eg, self-report questionnaires, clinical interviews, electronic health record review) used to measure common brain and mental health constructs, such as depression. Even when similar methods (eg, self-report measures) are implemented, challenges persist. For example, if a study used a depression measure that focused primarily on cognitive symptoms (eg, pessimism, self-dislike, suicidal ideation) and another study used a depression measure composed of items more heavily weighted towards somatic symptoms (eg, insomnia, loss of appetite, weight loss, decreased libido), combining their data could be challenging, particularly if researchers, clinicians, or administrators are interested in more than dichotomous outcomes (eg, depression vs no depression).^8,9

To address this knowledge gap and harmonize multimodal data from varied sources, well-planned and reproducible curation is needed. Longitudinal cohort studies of service members and veterans with military combat and training exposure histories provide researchers and other stakeholders access to extant biopsychosocial data shown to affect risk for adverse health outcomes; however, efforts to facilitate individually tailored treatment or other precision medicine approaches would benefit from the synthesis of such datasets.¹⁰

Members of the VA Total Brain Diagnostics (TBD) team are engaged in harmonizing variables from the Long-Term Impact of Military-Relevant Brain Injury Consortium–Chronic Effects of Neurotrauma Consortium (LIMBIC-CENC)¹¹ and the Translational Research Center for TBI and Stress Disorders (TRACTS).^12-21 While there is overlap across LIMBIC-CENC and TRACTS with respect to data domains, considerable data harmonization is needed to allow for future valid and meaningful analyses, particularly those involving multivariable predictors.

Data Sources

Both data sources for the TBD harmonization project, LIMBIC-CENC and TRACTS, include extensive, longitudinal data collected from relatively large cohorts of veterans and service members with combat exposure. Both studies collect detailed data related to potential brain injury history and include participants with and without a history of TBI. Similarly, both include extensive collection of fluid biomarkers and imaging data, as well as measures of biopsychosocial functioning.

Data collection sites for LIMBIC-CENC include 16 recruitment sites, 9 at VA medical centers (Richmond, Houston, Tampa, San Antonio, Portland, Minneapolis, Boston, Salisbury, San Diego) and 7 at military treatment sites (Alexandria, San Diego, Tampa, Tacoma, Columbia, Coronado, Hinesville), in addition to 11 assessment sites (Richmond, Houston, Tampa, San Antonio, Portland, Minneapolis, Boston, Salisbury, San Diego, Alexandria, Augusta). Data for TRACTS are collected at sites in Boston and Houston.

LIMBIC-CENC is a 12-year, 17-site cohort of service members and veteran participants with combat exposure who are well characterized at baseline and undergo annual reassessments. As of December 2025, > 3100 participants have been recruited, and nearly 90% remain in follow-up. Data collection includes > 6200 annual follow-up evaluations and > 1550 5-year re-evaluations, with 400 enrolled participants followed up annually.

TRACTS is a 16-year, 2-site cohort of veterans with combat exposure who complete comprehensive assessments at enrollment, undergo annual reassessments, and complete comprehensive reassessment every 5 years thereafter. As of December 2025, > 1075 participants have completed baseline (Time 1) assessments, > 600 have completed the 2-year re-evaluation (Time 2), > 175 have completed the 5-year re-evaluation (Time 3), and > 35 have completed 10-year evaluations (Time 4), with about 50 new participants added and 100 enrolled participants followed up annually. More data on participant characteristics are available for both LIMBIC-CENC and TRACTS in previous publications.^11,22These 2 ongoing, prospective, longitudinal cohorts of service members and veterans offer access to a wide range of potential risk factors that can affect response to care and outcomes, including demographics (eg, age, sex), injury characteristics (eg, pre-exposure factors, exposure factors), biomarkers (eg, serum, saliva, brain imaging, evoked potentials), and functional measures (eg, computerized posturography, computerized eye tracking, sensory testing, clinical examination, neuropsychological assessments, symptom questionnaires).

Harmonization Strategy

Pooling and harmonizing data from large studies evaluating similar participant cohorts and conditions involves numerous steps to appropriately handle a variety of measurements and disparate variable names. The TBD team adapted a model data harmonization system developed by O’Neil et al through initial work harmonizing the Federal Interagency Traumatic Brain Injury Research Informatics System (FITBIR).^4-7 This process was expanded and generalized by the research team to combine data from LIMBIC-CENC and TRACTS to create a single pooled dataset for analysis (Figure).

This approach was selected because it accommodates heterogeneous study designs (eg, cross-sectional, longitudinal, case-control), data collection methods (eg, clinical assessment, self-reported, objective blood, and imaging biomarkers), and various assessments of the same construct (ie, different measures of brain injury). While exact matches for data collection methods and measures may be easily harmonized, the timing of assessment, number of assessments, assessment tool version, and other factors must be considered. The goal was to harmonize data from LIMBIC-CENC and TRACTS to allow additional data sources to be harmonized and incorporated in the future.

Original data files from each study were reshaped to represent participant-level observations with 1 unique measurement per row. The measurement represents what information was collected and the value recorded represents the unique observation. These data are linked to metadata from the original study, which includes the study’s definition of each measurement, how it was collected, and any available information regarding when it was collected in reference to study enrollment or injury. Additional information on the file source, row, and column position of each data point was added to enable recreation of the original data as needed.

The resulting dataset was used to harmonize measurements from LIMBIC-CENC and TRACTS into a priori-defined schemas for brain- and mental health-relevant concepts, including TBI severity, PTSD, substance use, depression, suicidal ideation, and functioning (including cognitive, physical, and social functioning). This process was facilitated using natural language processing (NLP). Each study uniquely defines all measurements and provides written definitions with the data. Measurement definitions serve as records describing what was collected, how it was collected, and how the study may have uniquely defined information for its purposes. For example, definitions of exposure to brain injury and severity of brain injury may differ between studies, and the study-provided definition defines these differences.

Definitions were converted into numeric vectors through sentence embedding, a process that preserves the semantic meaning of the definition.²³ Cosine similarity was used as the primary metric to compare the semantic textual similarity between pairs of measurement definitions. Cosine similarity ranges from 0 to 1, where 0 indicates no meaningful similarity and 1 indicates they have identical meanings.²⁴ This approach leverages the relationship between the definitions of each measurement provided by a study and enables quick comparison of all pairwise combinations of measurement definitions between studies.

Subsets of similar measurements across studies were organized into a priori-defined schema. Clinical experts then reviewed each schema and further refined them into domains, (eg, mechanism of injury, clinical signs, acute symptoms) and subdomains (children), such as loss of consciousness, amnesia, and alteration of consciousness. This approach allows efficient handling of 2 specific cases that commonly occur when pooling and harmonizing datasets: (1) identifying the same measurement with differing names; and (2) identifying different measurements with definitions that each relate to the same domain.

The Table provides a general example of the schema for TBI severity. This was an iterative process in which clinical experts reviewed study-defined measurement definitions to develop general harmonized domains, and NLP techniques facilitated and accelerated identification and organization of measurements within these domains.

Expected Impact

Harmonization combining LIMBIC-CENC and TRACTS datasets is ongoing. Preliminary descriptive analyses of baseline cohort data indicate that harmonization across data sources is appropriate, given the lack of significant heterogeneity across sites and studies for most domains. Work by members of the TBD team is expected to lay the foundation for the use of existing and ongoing prospective, longitudinal datasets (eg, LIMBIC-CENC, TRACTS) and linked large datasets (eg, VA Informatics and Computing Infrastructure including electronic health records, VA Million Veteran Program, DaVINCI [US Department of Defense and VA Infrastructure for Clinical Intelligence]) to generate generalizable, clinically relevant information to advance precision brain and mental health care among service members and veterans.

By enhancing existing practice, this synthesized dataset has the potential to inform tailored and personalized medicine approaches designed to meet the needs of veterans and service members. These data will serve as the starting point for multivariable models examining the intersection of physiologic, behavioral, and environmental factors. The goal of this data harmonization effort is to better elucidate how clinicians and researchers can select optimal approaches for veterans and service members with TBI histories by accounting for a comprehensive set of physiologic, behavioral, and environmental factors in an individually tailored manner. These data may further extend existing clinical practice guideline approaches, inform shared decision-making, and enhance functional outcomes beyond those currently available.

Conclusions

Individuals who have served in the military have unique biopsychosocial exposures that are associated with brain and mental health disorders. To address these needs, the nationwide TBD team has initiated the creation of a unified, longitudinal dataset that includes harmonized measures from existing LIMBIC-CENC and TRACTS protocols. Initial data harmonization efforts are required to facilitate precision prognostics, diagnostics, and tailored interventions, with the goal of improving veterans’ brain and mental health and psychosocial functioning and enabling tailored and evidence-informed, individualized clinical care.

In leveraging existing, readily available evidence-based health care information (eg, systematic reviews, clinical practice guidelines), clinicians have historically made recommendations based on treatment responses of the average patient.¹ Recently, this approach has been expanded into data-driven, evidence-based precision medical care for individuals across a wide range of disciplines and care settings. These precision medicine approaches use information related to an individual’s genes, environment, and lifestyle to tailor recommendations regarding prevention, diagnosis, and treatment.

Applying precision medicine approaches to the unique exposures and experiences of service members and veterans—particularly those who served in combat environments—through the incorporation of biopsychosocial factors into medical decision-making may be even more pertinent. This sentiment is reflected in Section 305 of the Commander John Scott Hannon Veterans Mental Health Care Improvement Act of 2019, which outlines the Precision Medicine Initiative of the US Department of Veterans Affairs (VA) to identify and validate brain and mental health biomarkers.² Despite widespread consensus regarding the promise of precision medicine, large, rich datasets with elements pertaining to common military exposures such as traumatic brain injury (TBI) and posttraumatic stress disorder (PTSD) are limited.

Existing datasets, most of which are relatively small or focus on specific cohorts (eg, older veterans, transitioning veterans), continue to create barriers to advancing precision medicine. For example, in classically designed clinical trials, analyses are generally conducted in a manner that may obfuscate efficacy among subcohorts of individuals, thereby underscoring the need to explore alternative strategies to unify existing datasets capable of revealing such heterogeneity.³ The evidence base for precision medical care is limited, drawing from published trials with relatively small sample sizes and even larger cohort studies have limited biomarker data. Additionally, these models are often exploratory during development, and to avoid statistical overfitting of an exploratory model, validation in similar datasets is needed—an added burden when data sources are small or underpowered to begin with.

A promising approach is to combine and harmonize the largest, most deeply characterized data sources from similar samples. Although combining such datasets may appear to require minimal time and effort, harmonizing similar variables in an evidence-based and replicable manner requires time and expertise, even when participant characteristics and outcomes are similar.^4-7

Challenges related to harmonization are related to the wide range of strategies (eg, self-report questionnaires, clinical interviews, electronic health record review) used to measure common brain and mental health constructs, such as depression. Even when similar methods (eg, self-report measures) are implemented, challenges persist. For example, if a study used a depression measure that focused primarily on cognitive symptoms (eg, pessimism, self-dislike, suicidal ideation) and another study used a depression measure composed of items more heavily weighted towards somatic symptoms (eg, insomnia, loss of appetite, weight loss, decreased libido), combining their data could be challenging, particularly if researchers, clinicians, or administrators are interested in more than dichotomous outcomes (eg, depression vs no depression).^8,9

To address this knowledge gap and harmonize multimodal data from varied sources, well-planned and reproducible curation is needed. Longitudinal cohort studies of service members and veterans with military combat and training exposure histories provide researchers and other stakeholders access to extant biopsychosocial data shown to affect risk for adverse health outcomes; however, efforts to facilitate individually tailored treatment or other precision medicine approaches would benefit from the synthesis of such datasets.¹⁰

Members of the VA Total Brain Diagnostics (TBD) team are engaged in harmonizing variables from the Long-Term Impact of Military-Relevant Brain Injury Consortium–Chronic Effects of Neurotrauma Consortium (LIMBIC-CENC)¹¹ and the Translational Research Center for TBI and Stress Disorders (TRACTS).^12-21 While there is overlap across LIMBIC-CENC and TRACTS with respect to data domains, considerable data harmonization is needed to allow for future valid and meaningful analyses, particularly those involving multivariable predictors.

Data Sources

Both data sources for the TBD harmonization project, LIMBIC-CENC and TRACTS, include extensive, longitudinal data collected from relatively large cohorts of veterans and service members with combat exposure. Both studies collect detailed data related to potential brain injury history and include participants with and without a history of TBI. Similarly, both include extensive collection of fluid biomarkers and imaging data, as well as measures of biopsychosocial functioning.

Data collection sites for LIMBIC-CENC include 16 recruitment sites, 9 at VA medical centers (Richmond, Houston, Tampa, San Antonio, Portland, Minneapolis, Boston, Salisbury, San Diego) and 7 at military treatment sites (Alexandria, San Diego, Tampa, Tacoma, Columbia, Coronado, Hinesville), in addition to 11 assessment sites (Richmond, Houston, Tampa, San Antonio, Portland, Minneapolis, Boston, Salisbury, San Diego, Alexandria, Augusta). Data for TRACTS are collected at sites in Boston and Houston.

LIMBIC-CENC is a 12-year, 17-site cohort of service members and veteran participants with combat exposure who are well characterized at baseline and undergo annual reassessments. As of December 2025, > 3100 participants have been recruited, and nearly 90% remain in follow-up. Data collection includes > 6200 annual follow-up evaluations and > 1550 5-year re-evaluations, with 400 enrolled participants followed up annually.

TRACTS is a 16-year, 2-site cohort of veterans with combat exposure who complete comprehensive assessments at enrollment, undergo annual reassessments, and complete comprehensive reassessment every 5 years thereafter. As of December 2025, > 1075 participants have completed baseline (Time 1) assessments, > 600 have completed the 2-year re-evaluation (Time 2), > 175 have completed the 5-year re-evaluation (Time 3), and > 35 have completed 10-year evaluations (Time 4), with about 50 new participants added and 100 enrolled participants followed up annually. More data on participant characteristics are available for both LIMBIC-CENC and TRACTS in previous publications.^11,22These 2 ongoing, prospective, longitudinal cohorts of service members and veterans offer access to a wide range of potential risk factors that can affect response to care and outcomes, including demographics (eg, age, sex), injury characteristics (eg, pre-exposure factors, exposure factors), biomarkers (eg, serum, saliva, brain imaging, evoked potentials), and functional measures (eg, computerized posturography, computerized eye tracking, sensory testing, clinical examination, neuropsychological assessments, symptom questionnaires).

Harmonization Strategy

Pooling and harmonizing data from large studies evaluating similar participant cohorts and conditions involves numerous steps to appropriately handle a variety of measurements and disparate variable names. The TBD team adapted a model data harmonization system developed by O’Neil et al through initial work harmonizing the Federal Interagency Traumatic Brain Injury Research Informatics System (FITBIR).^4-7 This process was expanded and generalized by the research team to combine data from LIMBIC-CENC and TRACTS to create a single pooled dataset for analysis (Figure).

This approach was selected because it accommodates heterogeneous study designs (eg, cross-sectional, longitudinal, case-control), data collection methods (eg, clinical assessment, self-reported, objective blood, and imaging biomarkers), and various assessments of the same construct (ie, different measures of brain injury). While exact matches for data collection methods and measures may be easily harmonized, the timing of assessment, number of assessments, assessment tool version, and other factors must be considered. The goal was to harmonize data from LIMBIC-CENC and TRACTS to allow additional data sources to be harmonized and incorporated in the future.

Original data files from each study were reshaped to represent participant-level observations with 1 unique measurement per row. The measurement represents what information was collected and the value recorded represents the unique observation. These data are linked to metadata from the original study, which includes the study’s definition of each measurement, how it was collected, and any available information regarding when it was collected in reference to study enrollment or injury. Additional information on the file source, row, and column position of each data point was added to enable recreation of the original data as needed.

The resulting dataset was used to harmonize measurements from LIMBIC-CENC and TRACTS into a priori-defined schemas for brain- and mental health-relevant concepts, including TBI severity, PTSD, substance use, depression, suicidal ideation, and functioning (including cognitive, physical, and social functioning). This process was facilitated using natural language processing (NLP). Each study uniquely defines all measurements and provides written definitions with the data. Measurement definitions serve as records describing what was collected, how it was collected, and how the study may have uniquely defined information for its purposes. For example, definitions of exposure to brain injury and severity of brain injury may differ between studies, and the study-provided definition defines these differences.

Definitions were converted into numeric vectors through sentence embedding, a process that preserves the semantic meaning of the definition.²³ Cosine similarity was used as the primary metric to compare the semantic textual similarity between pairs of measurement definitions. Cosine similarity ranges from 0 to 1, where 0 indicates no meaningful similarity and 1 indicates they have identical meanings.²⁴ This approach leverages the relationship between the definitions of each measurement provided by a study and enables quick comparison of all pairwise combinations of measurement definitions between studies.

Subsets of similar measurements across studies were organized into a priori-defined schema. Clinical experts then reviewed each schema and further refined them into domains, (eg, mechanism of injury, clinical signs, acute symptoms) and subdomains (children), such as loss of consciousness, amnesia, and alteration of consciousness. This approach allows efficient handling of 2 specific cases that commonly occur when pooling and harmonizing datasets: (1) identifying the same measurement with differing names; and (2) identifying different measurements with definitions that each relate to the same domain.

The Table provides a general example of the schema for TBI severity. This was an iterative process in which clinical experts reviewed study-defined measurement definitions to develop general harmonized domains, and NLP techniques facilitated and accelerated identification and organization of measurements within these domains.

Expected Impact

Harmonization combining LIMBIC-CENC and TRACTS datasets is ongoing. Preliminary descriptive analyses of baseline cohort data indicate that harmonization across data sources is appropriate, given the lack of significant heterogeneity across sites and studies for most domains. Work by members of the TBD team is expected to lay the foundation for the use of existing and ongoing prospective, longitudinal datasets (eg, LIMBIC-CENC, TRACTS) and linked large datasets (eg, VA Informatics and Computing Infrastructure including electronic health records, VA Million Veteran Program, DaVINCI [US Department of Defense and VA Infrastructure for Clinical Intelligence]) to generate generalizable, clinically relevant information to advance precision brain and mental health care among service members and veterans.

By enhancing existing practice, this synthesized dataset has the potential to inform tailored and personalized medicine approaches designed to meet the needs of veterans and service members. These data will serve as the starting point for multivariable models examining the intersection of physiologic, behavioral, and environmental factors. The goal of this data harmonization effort is to better elucidate how clinicians and researchers can select optimal approaches for veterans and service members with TBI histories by accounting for a comprehensive set of physiologic, behavioral, and environmental factors in an individually tailored manner. These data may further extend existing clinical practice guideline approaches, inform shared decision-making, and enhance functional outcomes beyond those currently available.

Conclusions

Individuals who have served in the military have unique biopsychosocial exposures that are associated with brain and mental health disorders. To address these needs, the nationwide TBD team has initiated the creation of a unified, longitudinal dataset that includes harmonized measures from existing LIMBIC-CENC and TRACTS protocols. Initial data harmonization efforts are required to facilitate precision prognostics, diagnostics, and tailored interventions, with the goal of improving veterans’ brain and mental health and psychosocial functioning and enabling tailored and evidence-informed, individualized clinical care.

The Veterans Health Administration (VHA) is the largest US integrated health care system, providing health care to > 9 million veterans annually. Dementia affects > 7.2 million Americans, and an estimated 450,000 veterans live with Alzheimer disease (AD).^1,2 Compared with the general population, veterans have a higher burden of chronic medical conditions and are disproportionately affected by AD due to exposure to military-related risk factors (eg, traumatic brain injury and posttraumatic stress disorder) and the high prevalence of nonmilitary risk factors, such as cardiovascular disease. The VHA is a pioneer in dementia care, having established a Dementia System of Care to provide primary and specialty care to veterans with dementia. The VHA also is leading the way in implementing the Institute for Healthcare Improvement Age-Friendly Health Systems (AFHS) framework for providing goal-concordant care in > 100 VHA medical centers. The VHA aims to be the largest AFHS in the country.

AD profoundly affects individuals and their families. The progressive nature of the most common form of dementia diminishes the quality of life for patients as well as their care partners in an ongoing fashion, often leading to emotional, physical, and financial strain. Costs for health and long-term care for people living with AD and other dementias were projected at $360 billion in 2024, largely due to the need for nursing home care.¹ Although several oral medications are available, their capacity to effectively mitigate the negative effects of AD is limited. Cholinesterase inhibitors and memantine may offer temporary symptomatic relief, but they do not alter disease progression.³ The use of these agents is relatively low, with about one-third of patients diagnosed with AD receiving these medications.⁴

Amyloid-Targeting Therapies

Recent advancements in biologics, particularly amyloid-targeting therapies, such as lecanemab and donanemab, offer new hope for managing AD. Older adults treated with these medications show less decline on measures of cognition and function than those receiving a placebo at 18 months.^5,6 However, accessing and using these medications is challenging.

Use of amyloid-targeting therapies poses challenges. The medications are expensive, potentially placing a financial burden on patients, families, and health care systems.⁷ Determining initial eligibility for treatment requires a battery of cognitive assessments, laboratory tests, advanced radiologic studies (eg, magnetic resonance imaging [MRI] of the brain and amyloid positron emission tomography [PET] scans), and possible cerebrospinal fluid (CSF) testing. Frequent ongoing assessments are necessary to monitor safety and efficacy. These treatments carry substantial risks, particularly amyloid-related imaging abnormalities (ARIA) such as cerebral edema, microhemorrhages, and superficial siderosis. Therefore, follow-up assessments typically occur around months 2, 3, 4, and 7, depending on which medication is selected. Finally, at present, both agents must be intravenous (IV)-administered in a monitored clinical setting, which requires additional coordination, transportation, and cost.

Ongoing evaluations and in-person administration particularly affect patients and care partners with limitations regarding transportation, time off work, and navigating complex health care systems.⁸ VHA clinicians at sites that have implemented or are interested in implementing amyloid-targeting therapy programs endorse similar challenges when implementing these therapies in their US Department of Veterans Affairs (VA) medical centers (VAMCs).⁹

The VHA was one of the first health care systems to use amyloid-targeting therapies, covering the cost of lecanemab and donanemab, in addition to costs associated with concomitant evaluation and testing. However, given the safety concerns with this novel class of medications, the VHA National Formulary Committee developed criteria for use and recommended the VA Center for Medication Safety (VAMedSAFE) conduct a mandatory real-time medication use evaluation (MUE). VAMedSAFE developed the MUE to monitor the safe and appropriate use of amyloid-targeting therapy for AD. Two authors (AJO, SMH) partnered with VAMedSAFE through the VA Pittsburgh Healthcare System Technology Enhancing Cognition and Health–Geriatric Research, Education, and Clinical Center (TECH-GRECC) to provide clinical expertise, substantive feedback for the development of the MUE, and guidance for VHA sites starting amyloid targeting-therapy programs. We started a VHA Amyloid-Targeting Therapy for AD SharePoint collaborative platform and VHA AD Therapeutics Community of Practice (CoP) for shared learning (Figure). The private SharePoint platform houses an array of implementation materials for VAMCs starting programs: key documents and links; educational materials; sample guidelines; note templates; and electronic health record screenshots. The CoP allows VHAs to share best practices and discuss challenges.

Even with these advantages, we found that ensuring the safe and appropriate use of amyloid-targeting therapies did not overcome the barriers associated with their complexity. This was especially true for veterans living in rural areas. Only 4 VAMCs had administered amyloid-targeting therapies in the first year they were available. Preliminary data demonstrated that 27 (84%) of 32 veterans who initiated lecanemab in the VHA between October 2023 and September 2024 resided in urban areas.¹⁰ To address the underutilization of amyloid-targeting therapy, we propose leveraging the strengths of VHA telehealth to facilitate expansion of access to these medications for veterans with early AD. Telehealth may substantially increase access to evaluation for veterans with early dementia and, when medically appropriate, to receive amyloid-targeting therapies by reducing transportation needs and mitigating costs while ensuring appropriate monitoring through ongoing clinical assessments.

Using Telehealth

The VHA is a pioneer in telehealth, with programs dating back to 2003.¹¹ Between October 1, 2018, and September 30, 2019, the VHA served > 900,000 veterans through the provision of > 2.6 million episodes of care via telehealth.¹² The COVID-19 pandemic further cemented the role of telemedicine as an essential component of health care. Telehealth has demonstrated success in the assessment and management of individuals living with dementia. At the VHA, the GRECC-Connect Project is a partnership between 9 urban GRECC sites that seek to provide consultative geriatric and dementia care to rural veterans through telehealth.¹³ Additional evidence supports the potential to leverage telehealth to effectively communicate results of amyloid PET scans.¹⁴

This approach is not without limitations such as the digital divide, or the gap that separates technology-enabled individuals and those unprepared to adopt technology due to limited digital literacy levels or access to needed hardware, software, and connectivity. The VHA has taken steps to address these digital divide barriers by broadly providing tools—such as tablets and broadband connectivity—to veterans. Specifically, the VHA has instituted digital divide consults to determine whether telehealth could be a potential solution for appropriate veterans and to provide an iPad (if eligible) to connect with VA clinicians. Complementary to the digital divide consult, a VHA-specific telehealth preparedness assessment tool is under development and being tested by 2 authors (JF, SMH). This telehealth preparedness assessment tool is designed to aid in the seamless integration of telehealth services with the support of tailored education materials specific to gaps in digital literacy that a veteran might experience.

Building on these initiatives, there is an opportunity to expand access to amyloid-targeting therapies, regardless of distance to large VAMCs, by leveraging telehealth as an alternative method of connecting patients with specialty care. Specifically, a hybrid approach could be used to accomplish the myriad initial and follow-up tasks involved in the provision of amyloid-targeting therapies (Table). Not all VHA facilities possess the specialty expertise to prescribe these medications, and local clinicians may not have sufficient knowledge and clinical support to prescribe and monitor these therapies.

The first step is identifying local and regional subject matter experts, followed by the development and expansion of these networks. The National TeleNeurology Program is a good example of a national telehealth program that leverages technology to bring specialty services to rural areas with limited access to care. Although amyloid-targeting therapies often require more complex logistics, such as laboratory tests and imaging, these initial hurdles can be overcome through localized services and collaboration between VAMCs.

While treatment and imaging will most likely need to occur at a VAMC, most basic laboratory studies can be performed at community-based outpatient clinics (CBOCs). Some CBOCs may not be able to process more specialized laboratory tests such as apolipoprotein E genetic testing. Samples for these tests can be collected and processed at VAMCs, which usually have contracts with outside laboratories capable of performing these studies. Most, although not all, VAMCs offer advanced imaging, including MRI of the brain and amyloid PETs. VAMCs without those modalities may need to coordinate with other regional VAMCs. Additionally, a pilot program is already underway whereby VAMCs without the ability to quantify the amount of amyloid on PETs are able to leverage technology and collaborations with other VAMCs to obtain these data.

Once the initial phases of evaluation and care are completed, telemedicine can be leveraged for follow-up and ongoing management. Interdisciplinary teams can help facilitate care related to amyloid-targeting therapies, including the close monitoring of veterans for development of ARIA.¹⁵ To achieve this monitoring, specialty clinic teams prescribing amyloid-targeting therapies, which may be geographically distant, need to coordinate with local primary care clinical teams and emergency clinicians. All of these health care team members, along with neurologists and neurosurgeons, should be involved in the development and implementation of protocols in the event that patients present to their local primary or specialty care clinics or emergency department with ARIA symptoms.

If amyloid-targeting therapies are to be provided along with other emerging treatments for rural veterans, telehealth must be part of the solution. There is a pressing need to explore innovative evaluation and delivery models for these therapies, particularly as we expect additional diagnostics and therapeutics to be available in the future. With the advent of commercially available blood tests (ie, blood biomarkers) for AD, there is hope for a transition away from PETs and CSF testing given their cost, limited access, and invasiveness for diagnosis and monitoring of AD. These advances will increase the utility of telehealth to help rural veterans access amyloid-targeting therapies.

Additionally, administering the drug at home or at local clinics, supported by a dedicated health care team or home health agency, could further improve accessibility. Telehealth can be leveraged in this scenario, allowing specialty clinics and specialists to connect with patients and clinicians based out of local clinics or even home health agencies. In this scenario, specialists can provide hands-on care guidance and oversight even though they may be geographically distant from care recipients. Transitioning from IV administration to subcutaneous formulations would further enhance convenience and reduce barriers; these formulations may be available soon.¹⁶ Addressing logistical challenges to care and access through technology-based solutions will require coordinated efforts and continued VHA investment.

Conclusions

The VHA has a large population of veterans with dementia, and the costs to care for these veterans will only increase. While the current benefits of amyloid-targeting therapies are modest, now is the time to establish care processes that will support future innovations in amyloid-targeting therapies and other treatments and diagnostics. We are developing better ways to detect AD using clinical decision support tools, improving care pathways and the management of AD, and leveraging telehealth to improve access. The VA is conducting research to investigate whether a cognitive screening and laboratory evaluation that includes a telehealth preparedness assessment will be feasible and effective for improving the detection of AD and access to treatment, and we plan to publish the results.

The lessons learned can be extended to non-VHA care settings to help achieve potential benefits for other patients with early AD. Emerging therapies have the potential to improve the quality of life for both patients and care partners, adding life to years and not just years to life. Policymakers and payors must prioritize research funding to evaluate the safety and efficacy of these approaches to the delivery of health services, ensuring that emerging therapies are accessible for all individuals affected by AD.

The Veterans Health Administration (VHA) is the largest US integrated health care system, providing health care to > 9 million veterans annually. Dementia affects > 7.2 million Americans, and an estimated 450,000 veterans live with Alzheimer disease (AD).^1,2 Compared with the general population, veterans have a higher burden of chronic medical conditions and are disproportionately affected by AD due to exposure to military-related risk factors (eg, traumatic brain injury and posttraumatic stress disorder) and the high prevalence of nonmilitary risk factors, such as cardiovascular disease. The VHA is a pioneer in dementia care, having established a Dementia System of Care to provide primary and specialty care to veterans with dementia. The VHA also is leading the way in implementing the Institute for Healthcare Improvement Age-Friendly Health Systems (AFHS) framework for providing goal-concordant care in > 100 VHA medical centers. The VHA aims to be the largest AFHS in the country.

AD profoundly affects individuals and their families. The progressive nature of the most common form of dementia diminishes the quality of life for patients as well as their care partners in an ongoing fashion, often leading to emotional, physical, and financial strain. Costs for health and long-term care for people living with AD and other dementias were projected at $360 billion in 2024, largely due to the need for nursing home care.¹ Although several oral medications are available, their capacity to effectively mitigate the negative effects of AD is limited. Cholinesterase inhibitors and memantine may offer temporary symptomatic relief, but they do not alter disease progression.³ The use of these agents is relatively low, with about one-third of patients diagnosed with AD receiving these medications.⁴

Amyloid-Targeting Therapies

Recent advancements in biologics, particularly amyloid-targeting therapies, such as lecanemab and donanemab, offer new hope for managing AD. Older adults treated with these medications show less decline on measures of cognition and function than those receiving a placebo at 18 months.^5,6 However, accessing and using these medications is challenging.

Use of amyloid-targeting therapies poses challenges. The medications are expensive, potentially placing a financial burden on patients, families, and health care systems.⁷ Determining initial eligibility for treatment requires a battery of cognitive assessments, laboratory tests, advanced radiologic studies (eg, magnetic resonance imaging [MRI] of the brain and amyloid positron emission tomography [PET] scans), and possible cerebrospinal fluid (CSF) testing. Frequent ongoing assessments are necessary to monitor safety and efficacy. These treatments carry substantial risks, particularly amyloid-related imaging abnormalities (ARIA) such as cerebral edema, microhemorrhages, and superficial siderosis. Therefore, follow-up assessments typically occur around months 2, 3, 4, and 7, depending on which medication is selected. Finally, at present, both agents must be intravenous (IV)-administered in a monitored clinical setting, which requires additional coordination, transportation, and cost.

Ongoing evaluations and in-person administration particularly affect patients and care partners with limitations regarding transportation, time off work, and navigating complex health care systems.⁸ VHA clinicians at sites that have implemented or are interested in implementing amyloid-targeting therapy programs endorse similar challenges when implementing these therapies in their US Department of Veterans Affairs (VA) medical centers (VAMCs).⁹

The VHA was one of the first health care systems to use amyloid-targeting therapies, covering the cost of lecanemab and donanemab, in addition to costs associated with concomitant evaluation and testing. However, given the safety concerns with this novel class of medications, the VHA National Formulary Committee developed criteria for use and recommended the VA Center for Medication Safety (VAMedSAFE) conduct a mandatory real-time medication use evaluation (MUE). VAMedSAFE developed the MUE to monitor the safe and appropriate use of amyloid-targeting therapy for AD. Two authors (AJO, SMH) partnered with VAMedSAFE through the VA Pittsburgh Healthcare System Technology Enhancing Cognition and Health–Geriatric Research, Education, and Clinical Center (TECH-GRECC) to provide clinical expertise, substantive feedback for the development of the MUE, and guidance for VHA sites starting amyloid targeting-therapy programs. We started a VHA Amyloid-Targeting Therapy for AD SharePoint collaborative platform and VHA AD Therapeutics Community of Practice (CoP) for shared learning (Figure). The private SharePoint platform houses an array of implementation materials for VAMCs starting programs: key documents and links; educational materials; sample guidelines; note templates; and electronic health record screenshots. The CoP allows VHAs to share best practices and discuss challenges.

Even with these advantages, we found that ensuring the safe and appropriate use of amyloid-targeting therapies did not overcome the barriers associated with their complexity. This was especially true for veterans living in rural areas. Only 4 VAMCs had administered amyloid-targeting therapies in the first year they were available. Preliminary data demonstrated that 27 (84%) of 32 veterans who initiated lecanemab in the VHA between October 2023 and September 2024 resided in urban areas.¹⁰ To address the underutilization of amyloid-targeting therapy, we propose leveraging the strengths of VHA telehealth to facilitate expansion of access to these medications for veterans with early AD. Telehealth may substantially increase access to evaluation for veterans with early dementia and, when medically appropriate, to receive amyloid-targeting therapies by reducing transportation needs and mitigating costs while ensuring appropriate monitoring through ongoing clinical assessments.

Using Telehealth

The VHA is a pioneer in telehealth, with programs dating back to 2003.¹¹ Between October 1, 2018, and September 30, 2019, the VHA served > 900,000 veterans through the provision of > 2.6 million episodes of care via telehealth.¹² The COVID-19 pandemic further cemented the role of telemedicine as an essential component of health care. Telehealth has demonstrated success in the assessment and management of individuals living with dementia. At the VHA, the GRECC-Connect Project is a partnership between 9 urban GRECC sites that seek to provide consultative geriatric and dementia care to rural veterans through telehealth.¹³ Additional evidence supports the potential to leverage telehealth to effectively communicate results of amyloid PET scans.¹⁴

This approach is not without limitations such as the digital divide, or the gap that separates technology-enabled individuals and those unprepared to adopt technology due to limited digital literacy levels or access to needed hardware, software, and connectivity. The VHA has taken steps to address these digital divide barriers by broadly providing tools—such as tablets and broadband connectivity—to veterans. Specifically, the VHA has instituted digital divide consults to determine whether telehealth could be a potential solution for appropriate veterans and to provide an iPad (if eligible) to connect with VA clinicians. Complementary to the digital divide consult, a VHA-specific telehealth preparedness assessment tool is under development and being tested by 2 authors (JF, SMH). This telehealth preparedness assessment tool is designed to aid in the seamless integration of telehealth services with the support of tailored education materials specific to gaps in digital literacy that a veteran might experience.

Building on these initiatives, there is an opportunity to expand access to amyloid-targeting therapies, regardless of distance to large VAMCs, by leveraging telehealth as an alternative method of connecting patients with specialty care. Specifically, a hybrid approach could be used to accomplish the myriad initial and follow-up tasks involved in the provision of amyloid-targeting therapies (Table). Not all VHA facilities possess the specialty expertise to prescribe these medications, and local clinicians may not have sufficient knowledge and clinical support to prescribe and monitor these therapies.

The first step is identifying local and regional subject matter experts, followed by the development and expansion of these networks. The National TeleNeurology Program is a good example of a national telehealth program that leverages technology to bring specialty services to rural areas with limited access to care. Although amyloid-targeting therapies often require more complex logistics, such as laboratory tests and imaging, these initial hurdles can be overcome through localized services and collaboration between VAMCs.

While treatment and imaging will most likely need to occur at a VAMC, most basic laboratory studies can be performed at community-based outpatient clinics (CBOCs). Some CBOCs may not be able to process more specialized laboratory tests such as apolipoprotein E genetic testing. Samples for these tests can be collected and processed at VAMCs, which usually have contracts with outside laboratories capable of performing these studies. Most, although not all, VAMCs offer advanced imaging, including MRI of the brain and amyloid PETs. VAMCs without those modalities may need to coordinate with other regional VAMCs. Additionally, a pilot program is already underway whereby VAMCs without the ability to quantify the amount of amyloid on PETs are able to leverage technology and collaborations with other VAMCs to obtain these data.

Once the initial phases of evaluation and care are completed, telemedicine can be leveraged for follow-up and ongoing management. Interdisciplinary teams can help facilitate care related to amyloid-targeting therapies, including the close monitoring of veterans for development of ARIA.¹⁵ To achieve this monitoring, specialty clinic teams prescribing amyloid-targeting therapies, which may be geographically distant, need to coordinate with local primary care clinical teams and emergency clinicians. All of these health care team members, along with neurologists and neurosurgeons, should be involved in the development and implementation of protocols in the event that patients present to their local primary or specialty care clinics or emergency department with ARIA symptoms.

If amyloid-targeting therapies are to be provided along with other emerging treatments for rural veterans, telehealth must be part of the solution. There is a pressing need to explore innovative evaluation and delivery models for these therapies, particularly as we expect additional diagnostics and therapeutics to be available in the future. With the advent of commercially available blood tests (ie, blood biomarkers) for AD, there is hope for a transition away from PETs and CSF testing given their cost, limited access, and invasiveness for diagnosis and monitoring of AD. These advances will increase the utility of telehealth to help rural veterans access amyloid-targeting therapies.

Additionally, administering the drug at home or at local clinics, supported by a dedicated health care team or home health agency, could further improve accessibility. Telehealth can be leveraged in this scenario, allowing specialty clinics and specialists to connect with patients and clinicians based out of local clinics or even home health agencies. In this scenario, specialists can provide hands-on care guidance and oversight even though they may be geographically distant from care recipients. Transitioning from IV administration to subcutaneous formulations would further enhance convenience and reduce barriers; these formulations may be available soon.¹⁶ Addressing logistical challenges to care and access through technology-based solutions will require coordinated efforts and continued VHA investment.

Conclusions

The VHA has a large population of veterans with dementia, and the costs to care for these veterans will only increase. While the current benefits of amyloid-targeting therapies are modest, now is the time to establish care processes that will support future innovations in amyloid-targeting therapies and other treatments and diagnostics. We are developing better ways to detect AD using clinical decision support tools, improving care pathways and the management of AD, and leveraging telehealth to improve access. The VA is conducting research to investigate whether a cognitive screening and laboratory evaluation that includes a telehealth preparedness assessment will be feasible and effective for improving the detection of AD and access to treatment, and we plan to publish the results.

The lessons learned can be extended to non-VHA care settings to help achieve potential benefits for other patients with early AD. Emerging therapies have the potential to improve the quality of life for both patients and care partners, adding life to years and not just years to life. Policymakers and payors must prioritize research funding to evaluate the safety and efficacy of these approaches to the delivery of health services, ensuring that emerging therapies are accessible for all individuals affected by AD.

The Veterans Health Administration (VHA) is the largest US integrated health care system, providing health care to > 9 million veterans annually. Dementia affects > 7.2 million Americans, and an estimated 450,000 veterans live with Alzheimer disease (AD).^1,2 Compared with the general population, veterans have a higher burden of chronic medical conditions and are disproportionately affected by AD due to exposure to military-related risk factors (eg, traumatic brain injury and posttraumatic stress disorder) and the high prevalence of nonmilitary risk factors, such as cardiovascular disease. The VHA is a pioneer in dementia care, having established a Dementia System of Care to provide primary and specialty care to veterans with dementia. The VHA also is leading the way in implementing the Institute for Healthcare Improvement Age-Friendly Health Systems (AFHS) framework for providing goal-concordant care in > 100 VHA medical centers. The VHA aims to be the largest AFHS in the country.

AD profoundly affects individuals and their families. The progressive nature of the most common form of dementia diminishes the quality of life for patients as well as their care partners in an ongoing fashion, often leading to emotional, physical, and financial strain. Costs for health and long-term care for people living with AD and other dementias were projected at $360 billion in 2024, largely due to the need for nursing home care.¹ Although several oral medications are available, their capacity to effectively mitigate the negative effects of AD is limited. Cholinesterase inhibitors and memantine may offer temporary symptomatic relief, but they do not alter disease progression.³ The use of these agents is relatively low, with about one-third of patients diagnosed with AD receiving these medications.⁴

Amyloid-Targeting Therapies

Recent advancements in biologics, particularly amyloid-targeting therapies, such as lecanemab and donanemab, offer new hope for managing AD. Older adults treated with these medications show less decline on measures of cognition and function than those receiving a placebo at 18 months.^5,6 However, accessing and using these medications is challenging.

Use of amyloid-targeting therapies poses challenges. The medications are expensive, potentially placing a financial burden on patients, families, and health care systems.⁷ Determining initial eligibility for treatment requires a battery of cognitive assessments, laboratory tests, advanced radiologic studies (eg, magnetic resonance imaging [MRI] of the brain and amyloid positron emission tomography [PET] scans), and possible cerebrospinal fluid (CSF) testing. Frequent ongoing assessments are necessary to monitor safety and efficacy. These treatments carry substantial risks, particularly amyloid-related imaging abnormalities (ARIA) such as cerebral edema, microhemorrhages, and superficial siderosis. Therefore, follow-up assessments typically occur around months 2, 3, 4, and 7, depending on which medication is selected. Finally, at present, both agents must be intravenous (IV)-administered in a monitored clinical setting, which requires additional coordination, transportation, and cost.

Ongoing evaluations and in-person administration particularly affect patients and care partners with limitations regarding transportation, time off work, and navigating complex health care systems.⁸ VHA clinicians at sites that have implemented or are interested in implementing amyloid-targeting therapy programs endorse similar challenges when implementing these therapies in their US Department of Veterans Affairs (VA) medical centers (VAMCs).⁹

The VHA was one of the first health care systems to use amyloid-targeting therapies, covering the cost of lecanemab and donanemab, in addition to costs associated with concomitant evaluation and testing. However, given the safety concerns with this novel class of medications, the VHA National Formulary Committee developed criteria for use and recommended the VA Center for Medication Safety (VAMedSAFE) conduct a mandatory real-time medication use evaluation (MUE). VAMedSAFE developed the MUE to monitor the safe and appropriate use of amyloid-targeting therapy for AD. Two authors (AJO, SMH) partnered with VAMedSAFE through the VA Pittsburgh Healthcare System Technology Enhancing Cognition and Health–Geriatric Research, Education, and Clinical Center (TECH-GRECC) to provide clinical expertise, substantive feedback for the development of the MUE, and guidance for VHA sites starting amyloid targeting-therapy programs. We started a VHA Amyloid-Targeting Therapy for AD SharePoint collaborative platform and VHA AD Therapeutics Community of Practice (CoP) for shared learning (Figure). The private SharePoint platform houses an array of implementation materials for VAMCs starting programs: key documents and links; educational materials; sample guidelines; note templates; and electronic health record screenshots. The CoP allows VHAs to share best practices and discuss challenges.

Even with these advantages, we found that ensuring the safe and appropriate use of amyloid-targeting therapies did not overcome the barriers associated with their complexity. This was especially true for veterans living in rural areas. Only 4 VAMCs had administered amyloid-targeting therapies in the first year they were available. Preliminary data demonstrated that 27 (84%) of 32 veterans who initiated lecanemab in the VHA between October 2023 and September 2024 resided in urban areas.¹⁰ To address the underutilization of amyloid-targeting therapy, we propose leveraging the strengths of VHA telehealth to facilitate expansion of access to these medications for veterans with early AD. Telehealth may substantially increase access to evaluation for veterans with early dementia and, when medically appropriate, to receive amyloid-targeting therapies by reducing transportation needs and mitigating costs while ensuring appropriate monitoring through ongoing clinical assessments.

Using Telehealth

The VHA is a pioneer in telehealth, with programs dating back to 2003.¹¹ Between October 1, 2018, and September 30, 2019, the VHA served > 900,000 veterans through the provision of > 2.6 million episodes of care via telehealth.¹² The COVID-19 pandemic further cemented the role of telemedicine as an essential component of health care. Telehealth has demonstrated success in the assessment and management of individuals living with dementia. At the VHA, the GRECC-Connect Project is a partnership between 9 urban GRECC sites that seek to provide consultative geriatric and dementia care to rural veterans through telehealth.¹³ Additional evidence supports the potential to leverage telehealth to effectively communicate results of amyloid PET scans.¹⁴

This approach is not without limitations such as the digital divide, or the gap that separates technology-enabled individuals and those unprepared to adopt technology due to limited digital literacy levels or access to needed hardware, software, and connectivity. The VHA has taken steps to address these digital divide barriers by broadly providing tools—such as tablets and broadband connectivity—to veterans. Specifically, the VHA has instituted digital divide consults to determine whether telehealth could be a potential solution for appropriate veterans and to provide an iPad (if eligible) to connect with VA clinicians. Complementary to the digital divide consult, a VHA-specific telehealth preparedness assessment tool is under development and being tested by 2 authors (JF, SMH). This telehealth preparedness assessment tool is designed to aid in the seamless integration of telehealth services with the support of tailored education materials specific to gaps in digital literacy that a veteran might experience.

Building on these initiatives, there is an opportunity to expand access to amyloid-targeting therapies, regardless of distance to large VAMCs, by leveraging telehealth as an alternative method of connecting patients with specialty care. Specifically, a hybrid approach could be used to accomplish the myriad initial and follow-up tasks involved in the provision of amyloid-targeting therapies (Table). Not all VHA facilities possess the specialty expertise to prescribe these medications, and local clinicians may not have sufficient knowledge and clinical support to prescribe and monitor these therapies.

The first step is identifying local and regional subject matter experts, followed by the development and expansion of these networks. The National TeleNeurology Program is a good example of a national telehealth program that leverages technology to bring specialty services to rural areas with limited access to care. Although amyloid-targeting therapies often require more complex logistics, such as laboratory tests and imaging, these initial hurdles can be overcome through localized services and collaboration between VAMCs.

While treatment and imaging will most likely need to occur at a VAMC, most basic laboratory studies can be performed at community-based outpatient clinics (CBOCs). Some CBOCs may not be able to process more specialized laboratory tests such as apolipoprotein E genetic testing. Samples for these tests can be collected and processed at VAMCs, which usually have contracts with outside laboratories capable of performing these studies. Most, although not all, VAMCs offer advanced imaging, including MRI of the brain and amyloid PETs. VAMCs without those modalities may need to coordinate with other regional VAMCs. Additionally, a pilot program is already underway whereby VAMCs without the ability to quantify the amount of amyloid on PETs are able to leverage technology and collaborations with other VAMCs to obtain these data.

Once the initial phases of evaluation and care are completed, telemedicine can be leveraged for follow-up and ongoing management. Interdisciplinary teams can help facilitate care related to amyloid-targeting therapies, including the close monitoring of veterans for development of ARIA.¹⁵ To achieve this monitoring, specialty clinic teams prescribing amyloid-targeting therapies, which may be geographically distant, need to coordinate with local primary care clinical teams and emergency clinicians. All of these health care team members, along with neurologists and neurosurgeons, should be involved in the development and implementation of protocols in the event that patients present to their local primary or specialty care clinics or emergency department with ARIA symptoms.

If amyloid-targeting therapies are to be provided along with other emerging treatments for rural veterans, telehealth must be part of the solution. There is a pressing need to explore innovative evaluation and delivery models for these therapies, particularly as we expect additional diagnostics and therapeutics to be available in the future. With the advent of commercially available blood tests (ie, blood biomarkers) for AD, there is hope for a transition away from PETs and CSF testing given their cost, limited access, and invasiveness for diagnosis and monitoring of AD. These advances will increase the utility of telehealth to help rural veterans access amyloid-targeting therapies.

Additionally, administering the drug at home or at local clinics, supported by a dedicated health care team or home health agency, could further improve accessibility. Telehealth can be leveraged in this scenario, allowing specialty clinics and specialists to connect with patients and clinicians based out of local clinics or even home health agencies. In this scenario, specialists can provide hands-on care guidance and oversight even though they may be geographically distant from care recipients. Transitioning from IV administration to subcutaneous formulations would further enhance convenience and reduce barriers; these formulations may be available soon.¹⁶ Addressing logistical challenges to care and access through technology-based solutions will require coordinated efforts and continued VHA investment.

Conclusions

The VHA has a large population of veterans with dementia, and the costs to care for these veterans will only increase. While the current benefits of amyloid-targeting therapies are modest, now is the time to establish care processes that will support future innovations in amyloid-targeting therapies and other treatments and diagnostics. We are developing better ways to detect AD using clinical decision support tools, improving care pathways and the management of AD, and leveraging telehealth to improve access. The VA is conducting research to investigate whether a cognitive screening and laboratory evaluation that includes a telehealth preparedness assessment will be feasible and effective for improving the detection of AD and access to treatment, and we plan to publish the results.

The lessons learned can be extended to non-VHA care settings to help achieve potential benefits for other patients with early AD. Emerging therapies have the potential to improve the quality of life for both patients and care partners, adding life to years and not just years to life. Policymakers and payors must prioritize research funding to evaluate the safety and efficacy of these approaches to the delivery of health services, ensuring that emerging therapies are accessible for all individuals affected by AD.