|Home | About | Journals | Submit | Contact Us | Français|
There are no Food and Drug Administration (FDA)-approved medications indicated for the treatment of frontotemporal dementia (FTD). We sought to determine the most commonly used drugs used to treat behavioral variant FTD (bvFTD) in specialized dementia clinics.
Medication and demographic data from the Alzheimer’s Disease Research Centers of California (ARCC) and a multicenter FTD natural history study (NHS) data set were compared in bvFTD and Alzheimer’s disease (AD), and effects of demographic variables were assessed using logistic regression.
Overall, the percentage of patients taking one or more FDA-approved AD or psychiatric medications was similar in bvFTD and AD; however, after controlling for demographic variables, acetylcholinesterase inhibitor (AChI) use was less common in bvFTD, whereas memantine use remained similar in the 2 groups.
Despite lack of evidence for efficacy, the use of AChIs and memantine is common in bvFTD. Clinical trials should be pursued to determine the optimal therapeutic interventions for bvFTD.
Alzheimer’s disease (AD) and frontotemporal dementia (FTD) are common causes of dementia. Frontotemporal dementia is pathologically different from AD, and patients present with dissimilar clinical symptoms, commonly including changes in personality, behavior, and language. Treatment with acetylcholinesterase inhibitors (AChI; donepezil, galantamine, and rivastigmine) or the N-methyl-d-aspartate (NMDA)-receptor antagonist memantine is approved by the Food and Drug Administration (FDA) for use in AD, but there are no medications indicated for use in FTD.1 However, clinical experience suggests that many patients with FTD are treated off-label with AD medications.
Acetylcholinesterase inhibitors and memantine are well tolerated and have modest effects on the symptoms of AD, with a variety of randomized, placebo-controlled clinical trials demonstrating better cognitive, functional, and global outcomes when compared to placebo.2 Moreover, combinations of an AChI plus memantine may be more effective than treatment with donezepil alone.3 In addition to AChI and memantine, antipsychotics, antidepressants, and anxiolytics are often used in patients with AD to treat symptoms of agitation, hallucinations, delusions, and other behavioral dysfunctions with varying degrees of success.4,5
A number of medications that affect the function of specific neurotransmitter systems have been studied and may offer modest benefits for the behavioral and cognitive sequelae of behavioral variant (bv) FTD. There is selective vulnerability of serotonergic neurons in FTD, and a meta-analysis of SSRI treatment studies suggests that these medications offer modest benefits for improving the behavioral symptoms of bvFTD.6 A randomized, placebo-controlled trial demonstrated that trazodone improves behavior as measured by the neuropsychiatric inventory (NPI) in bvFTD.7 Although commonly used to control behavior in bvFTD, antipsychotic agents may be of limited use due to enhanced sensitivity to extrapyramidal side effects in these patients.8
Less success has been reported in treating the cognitive symptoms of bvFTD. Unlike AD, cholinergic neurons are relatively preserved in bvFTD, and studies using AChIs have produced mixed results. One open-label study suggested that rivastigmine was well tolerated in bvFTD,9 and a recent placebo-controlled study demonstrated a potential benefit of galantamine in logopenic variants of primary progressive aphasia (who may have underlying AD pathology10); however, experience from a number of FTD-specialty clinics suggests that AChIs frequently worsen behavior in bvFTD patients with more severe behavioral impairments.11,12 Three open-label studies of memantine in FTD have been reported, all suggesting that the FDA-approved dose for moderate–severe AD is well tolerated; however, no clear efficacy signals could be identified.13–15
Little is known about actual medication use in bvFTD. We compared the use of anti-AD medications, antipsychotics, mood stabilizers, antidepressants, and anxiolytics by patients with bvFTD to patients with AD in a combined cohort derived from the Alzheimer’s Research Centers of California (ARCC) database and a multicenter, natural history study (NHS) of FTD.16
Medication and demographic data were gathered from the Alzheimer’s Disease Research Centers of California (ARCC) Minimum Uniform Data Set (MUDS) and a multicenter FTLD NHS data set.16 Patients with dementia due to NINCDSADRDA (National Institute of Neurological and Communicative Diseases and Stroke-Alzheimer’s Disease and Related Disorders Association) probable AD17 or bvFTD18 were included in this study. Both AD and bvFTD cases were included from the ARCC data set. Progressive aphasia cases were not included because there were too few to analyze. Only patients with bvFTD from the NHS data set were included for analysis because these data set contained no patients with AD. Because the University of California, San Francisco (UCSF), and University of California, Los Angeles (UCLA), contributed participants to both the ARCC and the NHS data sets, dates of birth were analyzed to be certain that none of the bvFTD cases were duplicated in the final data set. Because memantine was approved for use in AD in October 2003, patients with a visit date prior to January 2004 were excluded from analysis. For the ARCC data, disease duration was calculated by subtracting the age at onset of symptoms from the age at the date of visit. The disease duration was already included as a variable in the FTLD NHS data set.
The study cohort consisted of 616 participants. Of these patients, 569 patients were from the ARCC data set, who had primary diagnoses of probable AD (n = 528) or bvFTD (n = 41). Forty-seven patients with bvFTD were from the FTLD NHS data set. Patient visits occurred between 2004 and 2008, and data came from 13 centers: 93 from UCSD, 67 from USC/Rancho Los Amigos, 45 from UC Davis/Sacramento, 44 from UC Davis/Martinez, 57 from UCSF, 67 from Stanford, 66 from UCSF/Fresno, 39 from UC Irvine, 19 from UCLA, 96 from USC/Los Angeles, 10 from Mayo Rochester, 9 from Mayo Jacksonville, and 4 from Mayo Scottsdale.
Variables that were analyzed included original data set, education level, age of onset, number of years symptoms present, date of visit, date of birth, gender, race, Mini-Mental State Examination (MMSE) score, anti-AD medication, and all other medication use were recorded. Because memantine was approved for use in AD in October 2003, patients with a visit date prior to January 2004 were excluded from analysis. Race was coded as Caucasian, African American, Asian/Pacific Islander, Native American, Hispanic/Latino, other, and unknown for the baseline characteristics analysis. However, for some analyses (eg, logistic regression), the patient race was coded as either non-Hispanic white or non-white to increase sample size. Age was calculated using the patients’ date of birth and the date of the visit.
Use of 5 different anti-AD medications and 15 other psychoactive medications (either over-the-counter or prescription) were recorded based on Multum drug codes. Alzheimer’s disease medications analyzed were memantine or AChIs donepezil, galantamine, and rivastigmine. Psychiatric medications were categorized as antipsychotic, mood stabilizer, antidepressant, and anxiolytic medications. Use of the following antipsychotic agents was recorded: promethazine, risperidone, olanzapine, quetiapine, qutiapine, haloperidol, clozapine, prochlorperazine, aripiprazole, and perphenazine. Mood stabilizers included carbamazepine, valproic acid, or lithium. Antidepressants included citalopram, sertraline, mirtazapine, trazodone, paroxetine, venlafaxine, maprotiline, amitriptyline, fluoxetine, dulozetine, nor-triptyline, doxepin, clomipramine, fluvoxamine, escitalopram, and bupropion. Anxiolytic medications consisted of lorazepam, clonazepam, alprazolam, temazepam, zolpidem, eszopiclone, doxylamine, flurazepam, midazolam, phenobarbital, triazolam, chlordiazepoxide, oxazepam, buspirone, diphenhydramine, hydroxyzine, meprobamate, and ramelteon.
Because 2 different databases were the source of our bvFTD sample, we conducted descriptive analyses on the variables that would be used in the larger analysis to determine whether the ARCC and NHS bvFTD samples could be combined. To assess whether there were differences between the bvFTD groups, a multivariate analysis of variance (MANOVA) was used to examine group differences between age, years of education and MMSE, and χ2 tests to analyze bivariate and categorical variables such as the number of anti-AD medications, gender, race, and AD and psychiatric medication use.
Results of the bvFTD group comparison are shown in supplementary table (available at http://aja.sagepub.com/supplemental). The ARCC bvFTD participants had lower MMSE scores than the NHS bvFTD participants (21.4 ± 1.22 vs 25.1 ± 0.6, P = .005), however, there were no significant differences in medication-related variables. For this reason, we elected to combine the ARCC and the NHS data sets to increase sample size but included original data set as an additional covariate in subsequent logistic regression models comparing medication use in the AD versus bvFTD groups.
After the 2 bvFTD samples were combined, similar analyses, MANOVA, t tests, and χ2 analyses were conducted comparing the AD and bvFTD groups on the demographic and descriptive data.
Logistic regression models (enter method) were constructed to investigate possible roles of gender, age, MMSE score, years of education, race, and original data set (ARCC or NHS) in medication use among the 2 diagnostic groups (AD vs FTD). However, due to small sample sizes within the non-white race/ethnicity categories, race/ethnicity was coded as either non-Hispanic white or non-white. The Homer and Lemeshow test19 and regression diagnostics measured how well the model fit the data; if P < .05, then the model was rejected and the variable that was least significant was removed until logistic regression model best fit the data.
Demographics for the combined bvFTD group as compared to the AD group are shown in Table 1. The patients with bvFTD were younger (P < .001) and were more highly educated (P < .001) than the patients with AD. The AD participants had lower MMSE scores than the patients with bvFTD (P < .001) and a higher proportion of AD participants were female than the bvFTD group (P < .001). The mean total number of all medications was higher in the patients with AD than in patients with bvFTD (P = .023).
The percentage of patients taking any AD medication (donepezil, galantamine, rivastigmine, or memantine) was comparable between the bvFTD and the AD groups, and there was no difference in the mean number of FDA-approved AD medications taken (P = .075) between the bvFTD and the AD groups (Table 1). The most frequently used AD medication was donepezil, followed by memantine. Logistic regression analysis (Table 2) suggested that differences in all AD medication use were explained by differences in MMSE score between the 2 groups. Mini-Mental State Examination was inversely associated with the use of any anti-AD medication; that is, adjusted for diagnosis, patients were less likely to take any anti-AD medication if they had a higher MMSE score (odds ratio [OR]: 0.94, 95% confidence interval [CI]: 0.91–0.97, P < .001). Patients treated with AD medications had a mean MMSE score of 18.6 ± 0.4 as compared to 20.4 ± 0.4 for those who were not (P < .001).
Acetylcholinesterase inhibitor use, driven by the use of donepezil, was higher in AD (P = .010 for AChI and P = .007 for donepezil; Table 1) than in bvFTD. Galantamine, rivastigmine, and memantine use was similar between the 2 groups. After adjusting for demographic variables (Table 2), the odds of AChI use (donepezil, galantamine, and rivastigmine) were higher in patients with AD than in patients with bvFTD even when other variables such as age, gender, race, education level, and MMSE scores were included in the model (OR for bvFTD: 0.45, 95%CI: 0.25–0.82). The most commonly used AChI was donepezil, and the logistic regression for donepezil analysis showed similar results.
There was no difference in the odds of memantine use between patients with AD and bvFTD (OR: 1.12, 95% CI: 0.58–2.16, P = 0.727). In all participants combined, memantine use was associated with more education (OR: 1.08, 95% CI: 1.02–1.14; P = .008) and lower MMSE scores (OR: 0.93, 95% CI: 0.90–0.96; P < .001). Individuals who took memantine had a mean of 13.9 ± 0.3 years of education as compared to 13.0 ± 0.2 years in those who did not take memantine and had mean MMSE scores of 17.9 ± 0.5 as compared to 19.9 ± 0.3 in those who did not.
Psychiatric medication use was comparable between patients with AD and bvFTD even when controlling for age, gender, race, education level, and MMSE score (Table 3). Adjusted for diagnosis, men were less likely to use psychiatric medications than women (OR: 0.65, 95% CI: 0.45–0.95; Table 3). There was a nonsignificant trend for younger patients to be more likely to take psychiatric medications (P = .014). Those who were taking any type of psychiatric medication had a mean age of 75.0 ± 0.7 as compared to those who were not, who had a mean age of 77.0 ± 0.6.
Antidepressant use was higher in patients with bvFTD than AD (P = .014; Table 1). However, after controlling for demographic variables, diagnosis was not associated with antidepressant use (OR: 1.29, 95% CI: 0.70–2.38; Table 3). In all participants combined, treatment with antidepressants was less likely if the patient was older (OR: 0.98, 95% CI: 0.96–1.0 P = .012) or male (OR: 0.55, 95% CI: 0.37–0.83; P = .004). The average age of individuals taking an antidepressant was 74.2 ± 0.8 as compared to 77.0 ± 0.5 individuals not taking antidepressants. Males made up only 31.6% of the population taking antidepressants.
Antipsychotic medication use was similar between the diagnostic groups (P = .213); 10.0% of patients with AD and 4.5% of patients with bvFTD took at least 1 antipsychotic. In the logistic regression, diagnosis did not predict antipsychotic use (P = .236). Individuals with higher MMSE scores were less likely use an antipsychotic (OR: 0.9, 95% CI: 0.86–0.95; P < .001). Patients taking an antipsychotic had mean MMSE score of 150.58 ± 1.02 and those who did not had mean MMSE score of 19.71 ± 0.27. Mood stabilizer and anxiolytic medication use was similar between the bvFTD and the AD groups.
On average, patients with AD took a greater number of medications than patients with bvFTD (P = .023; Table 1). The most common non-AD/psychiatric medications used in patients with AD were aspirin (31.4% of participants), followed by multivitamins (25.8%), calcium carbonate (15.2%), and levothyroxine (13.4%). The most frequently used non-AD/psychiatric medications in bvFTD were aspirin (39.8%), multivitamin (31.8%), vitamin C (17.0%), and vitamin E (15.9%).
Although FDA-approved AD medications are not indicated for use in bvFTD, we found that overall the use of such medications is as common in bvFTD as in AD in specialized dementia clinical and research centers. Moreover, despite concerns regarding efficacy and worsening of FTD-related behaviors by AChIs,1,12 40.9% of patients with bvFTD took an AChI. Although AChI use, particularly donepezil, was more common in AD than bvFTD, memantine use was nearly identical in the 2 groups. Similarly, prescription psychiatric medication use was similar between the AD and bvFTD groups, although antidepressant use was higher in the bvFTD than AD groups. This difference may have been accounted for by the younger age in the bvFTD group, because younger age was more strongly associated than diagnosis with antidepressant use when data from both groups were combined (Table 3).
We identified a number of interesting trends in medication use across all participants, irrespective of diagnosis. A lower MMSE score was associated with higher odds of use of antipsychotics, AD medications (as a class), an AChI, or memantine, regardless of diagnosis. We speculate that the association of MMSE scores with medication use may reflect greater certainty about a dementia diagnosis for AD medications and worsening behavior with increasing disease severity for antipsychotics. However, medication use was not associated with disease duration in either group. This may reflect the difficulty in accurately determining disease onset, heterogeneous rates of decline, or a inconsistent relationship between disease duration and severity in each of the groups.
The odds of using memantine were increased with higher education levels, and there was a nonsignificant trend for greater odds of any AD medication use in individuals with higher levels of education (P = .117). Memantine was first made available in January 2004, and all data in this study reflect visits from 2004 to 2008. Patients with a higher education level might have been more likely to be prescribed this medication, due to a greater knowledge of newly available therapies, or possibly due to a higher socioeconomic status, leading to a greater ability to afford new medications.
There were important limitations in conducting this study. First, the number of patients with bvFTD in our analysis was small in comparison to the number of patients with probable AD. This may have limited our ability to detect differences in the pattern of medication use.19 Overall, the small number of patients taking antipsychotics, mood stabilizers, and anxiolytic medications also limited our ability to detect disease-specific differences in the use of these medications. Moreover, because dosage information was not available for analysis, differences in AD medication dose between groups may have been missed. In addition, because the medication data were derived from 2 different data sets with few common neuropsychological or functional assessments, we were forced to rely on MMSE as a measure of disease severity, which may have underestimated the relative disease severity of the patients with bvFTD.20 Finally, the retrospective nature of the data may have introduced a variety of unknown biases that were not accounted for in our analyses.
It is not clear from our study why such a large proportion of patients with bvFTD are treated with FDA-approved AD medications. The treatment of bvFTD patients with AD medications might reflect uncertainties about underlying diagnosis, patient or caregiver preference, or a general bias to medicate demented individuals. Because the bvFTD cases were not autopsy-confirmed, it is possible that some may have had underlying AD pathology with some clinical features of AD that contributed to diagnostic uncertainty. Although AChI use was higher in AD than in bvFTD, memantine use was similar in the 2 groups. This may reflect conflicting published reports on effects of AChIs in bvFTD, which include both “positive” open-label studies, suggesting improved behavior9 and “negative” studies documenting adverse effects of cholinesterase inhibitors.12 In contrast, open-label studies of memantine have suggested either possible mild benefits13,15 or no effects.14 The higher percentage of patients with bvFTD taking antidepressants is likely to reflect the published literature on bvFTD treatment, including a placebo-controlled trial of trazodone 7 and other open-label studies, which suggest benefits of SSRIs in bvFTD.1,6
Our analysis shows that off-label use of antidementia and psychiatric medications is common in bvFTD, despite little evidence to support such treatment. Larger and more comprehensive studies are needed to understand the reasons for current treatment practices. More importantly, randomized, placebo-controlled clinical trials of antidementia medications should be pursued in bvFTD to allow for evidence-based decision making in determining the treatment of patients with this common form of dementia.
The authors disclosed receipt of the following financial support for the research and/or authorship of this article: K23-NS48855 (ALB), R01-AG031278 (ALB), P01-AG019724 (BLM), P50-AG0300601 (BLM), John Douglas French Foundation (ALB), L. Hillblom Foundation (BLM), the State of California, R01-AG023195 (DK) and P50-AG 16574 (Mayo Alzheimer’s Disease Research Center), P30-AG19610 (Arizona ADC) and P50-AG016570 (UCLA ADRC).
Declaration of Conflicting Interest
The authors declared a potential conflict of interest as follows: ALB has received research support from Elan, Forest, Genentech, Medivation, Myriad, Novartis, Pfizer and Wyeth for conducting clinical trials and has served as a consultant for Satoris and Bristol Myers Squibb. DK has received research support from Elan and Forest. BB has received research support from Myriad. BLM has received research support from Novartis.