Histograms of the number of patients within each etiologic group stratified by age (A) and the percentage of patients that each etiologic group contributed stratified by age (B). In B, the age range is truncated at age 43 years to account for the mean 2-year delay from onset to presentation, which likely explains the smaller number of patients identified as having onset at age 44 or 45 years.
Kelley BJ, Boeve BF, Josephs KA. Young-Onset DementiaDemographic and Etiologic Characteristics of 235 Patients. Arch Neurol. 2008;65(11):1502-1508. doi:10.1001/archneur.65.11.1502
Copyright 2008 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.2008
Onset of dementia before age 45 years presents a difficult clinical circumstance, having a broad differential diagnosis and numerous psychosocial implications for the patient and their family. Few data exist regarding the demographics characterizing this population or the etiologic diagnoses among those affected.
To characterize the demographic characteristics and the etiologic causes of dementia with age at onset younger than 45 years.
Observational, retrospective, single-cohort study.
Multispecialty group academic medical center.
We searched the Mayo Clinic Rochester electronic Medical Record Linkage System to identify individuals who were seen for evaluation of progressive cognitive decline between the ages of 17 and 45 years from January 1996 through December 2006. This search identified 235 individuals who met the established inclusion and exclusion criteria.
Main Outcome Measures
All available clinical, laboratory, magnetic resonance imaging, and pathological data were reviewed.
Causes varied, with neurodegenerative etiologies accounting for 31.1% of the cohort; Alzheimer disease was uncommon. Autoimmune or inflammatory causes accounted for 21.3%. At last follow-up, 44 patients (18.7%) had an unknown etiology, despite exhaustive evaluation. Cause varied with age, with inborn errors of metabolism being more common before age 30 years and with neurodegenerative etiologies being more common after age 35 years.
Young-onset dementia (age at onset, <45 years) includes a broad variety of etiologies, with few patients having a potentially treatable disorder. The etiologic spectrum and the relative percentages of patients within etiologic groups differed in important ways from existing reports of early-onset dementia (ie, age at onset, <65 years).
The differential diagnosis of young-onset dementia (YOD) is extensive, including sporadic hereditary etiologies, neurodegenerative diseases, adult presentations of inborn errors of metabolism, other metabolic or storage diseases, and other syndromic diagnoses.1,2 In some individuals, the etiology remains indeterminate even after brain biopsy.3 Although investigations of the etiologies and prevalence of early-onset dementia (EOD) have been performed,4- 7 the available literature regarding YOD exists primarily as individual case reports or as small case series focused on a single etiology.
To better characterize this population, we reviewed the experience of our department of neurology with YOD to more clearly delineate the demographic and etiologic characteristics of this population. We define YOD as dementia with onset between the ages of 17 and 45 years. This term is used to distinguish this population from studies of EOD, typically defined as dementia occurring before age 65 years.4,5,8- 10
This protocol was approved by the Mayo Foundation Institutional Review Board. We searched the Medical Record Linkage System of patients evaluated at Mayo Clinic Rochester from January 1996 through December 2006, using the relevant Hospital International Classification of Diseases Adapted codes for dementia and for neurodegenerative disorders that may manifest with dementia between the ages of 17 and 45 years. All available medical records were reviewed, and relevant clinical data were extracted. The age at onset was considered to be the age at which cognitive changes were apparent to the patient, family, or close associates.
Included patients with onset between the ages of 17 and 45 years with progressive cognitive decline that impaired their ability to function independently. We required documented evidence of previously normal cognitive functioning until the age of 17 years. For most patients, completion of high school without requiring special educational assistance was taken as sufficient documentation. For the few individuals not completing high school, report of sustained normal cognitive functioning and gainful employment before developing progressive cognitive decline was accepted as evidence of previously normal baseline cognitive function.
Patients manifesting long-standing static encephalopathy or having diagnoses of mental retardation were excluded. Patients requiring special education throughout their academic career were excluded. Those with cognitive deficits referable to an isolated event (eg, stroke, traumatic brain injury [TBI], and subarachnoid hemorrhage) were excluded.
Statistical analyses were performed using commercially available software (JMP Software, version 6.0.0; SAS Institute Inc, Cary, North Carolina), with α = .05. Sex ratios and binomial data such as the presence or absence of family history were compared across groups using the χ2 test. The Kruskal-Wallace test was used to compare all continuous data such as age at onset across all groups.
Application of the inclusion and exclusion criteria to the search results identified 235 patients appropriate for this study. Summary demographics of the cohort are given in Table 1. The female-to-male ratio was 1:1, and the cohort was 90.2% white. The mean number of years of education was 13.9. A positive family history was present in 25.5%. Onset was characterized as acute (significant decline during <1 week), subacute (significant decline occurring ≤6 weeks of onset), or gradual (decline occurring for ≥7 weeks). Clinical course was characterized as being gradually progressive, relapsing progressive (progressive cognitive decline punctuated by more rapid episodes of decline from which recovery may or may not have occurred), or relapsing remitting (sustained periods of worsened cognitive impairment from which substantial [but usually incomplete] recovery typically occurred). Rapidly progressive disease was defined as progression to severe dementia or death within 18 months of onset.
The final diagnoses of 235 patients were grouped into several etiologic categories similar to the groupings typically used in formulating differential diagnoses. Table 2 summarizes these grouped data and identifies many of the specific diagnoses identified within each etiologic group. The most common etiologic groups were neurodegenerative (31.1%), autoimmune or inflammatory (21.3%), and unknown (18.7%). Metabolic etiologies were identified in 10.6%. Figure, A, shows the number of individuals within each etiologic group stratified by age, and Figure, B, shows the relative percentages that each etiologic group contributed to the specified age ranges. Table 3 gives the demographic characteristics of each etiologic group.
Age at onset ranged from 17 to 45 years, as defined by our inclusion criteria. There was a significant difference in age at onset across the etiologic groups (P < .01). Educational attainment ranged from 8 to 20 years. No significant difference in the number of years of education was identified among etiologic groups (P = .26). Overall, the male-to-female ratio was 1:1 for our population. No significant difference was noted in sex distribution among etiologic groups (P = .43). There was no significant difference in etiology between white and nonwhite patients (P = .87).
Among 60 patients having a family history of EOD or neurologic presentation similar to that of a proband, genetic or hereditary causes were identified in 31 individuals (51.7% of those reporting a family history). Identified types of disorders were the following: 12 patients with Huntington disease; 3 patients with prion disease; 2 patients each with mitochondrial, adrenoleukodystrophy, frontotemporal dementia (FTD) with parkinsonism linked to chromosome 17, and hereditary neuropathy, deafness, and dementia; and 1 patient each with Wilson disease, tuberous sclerosis, GM2 gangliosidosis, hexosaminidase A deficiency, spinocerebellar atrophy type 3, metachromatic leukodystrophy, Alzheimer disease (AD) associated with mutation in presenilin 1 (PSEN1), and adult-onset autosomal dominant leukodystrophy with neuroaxonal spheroids and periodic acid–Schiff—positive macrophages. There was a significant difference in family history among etiologic groups (P < .001).
A significant difference in disease duration before presentation was identified among etiologic groups (P < .001). As summarized in Table 3, patients having infectious, vascular, and autoimmune or inflammatory etiologies had shorter times between onset and presentation compared with those having neurodegenerative or unknown etiologies.
Our retrospective review identified 235 patients with age at onset of progressive dementia younger than 45 years. The referral bias toward atypical presentations inherent to performing this study at a tertiary care facility precludes our ability to reliably infer frequency analysis. Although an argument could be made that all cases of YOD are atypical simply by virtue of age at onset, the validity of generalizations from this cohort to the more general population of all patients with YOD should be confirmed in a prospectively studied cohort. Furthermore, our search spanned 11 years, during which the available diagnostic testing has undergone significant changes.
Nevertheless, several important points can be made. Our study describes a large cohort of patients who experienced progressive cognitive decline before age 45 years. One study4 investigated the prevalence and etiologies of EOD in 185 patients with onset of dementia before age 65 years, among whom 15 individuals had onset of dementia between the ages of 30 and 44 years. Harvey et al estimated the prevalence of EOD (age range, 30-65 years) at 54 cases per 100 000. Although no prevalence estimates were explicitly calculated for the YOD age range, data in that study indicate that the prevalence rate for ages 30 to 44 years would be estimated to be 12.1 cases per 100 000. This underscores the importance of our cohort to provide a resource examining a large population of patients having this rare neurologic presentation.
Exclusion of nonprogressive cognitive impairment referable to a specific incident removed patients having TBI or acute cerebrovascular events from our population. Although TBI may be similar to YOD in its personal effect, societal costs, and therapeutic challenges, TBI differs in many important ways from progressive dementias. Furthermore, the readily identifiable antecedent cause does not present the diagnostic challenges and dilemmas faced by clinicians evaluating the patients described in our cohort.
In analysis of presenile AD, family history has been shown to be a strong risk factor.11 The generalization that family history of EOD is common in YOD would seem to be partly valid because only 25.5% of our cohort reported a positive family history of EOD or neurologic presentation similar to that of the study patient. Statistical analysis of our cohort grouped by etiology demonstrated a significant difference in family history between etiologic groups. Family history was more common (36.0%-43.80%) among those having neurodegenerative, metabolic, or unknown causes and less common (0%-14.0%) among those in the autoimmune or inflammatory, vascular, or other etiologic categories. The presence of a positive family history in 9.1% of those having infectious causes is likely a spurious result of 1 individual (among 11 in this category) reporting a positive family history.
Ultimately, diagnoses having a genetic basis were confirmed in 50.0% of those reporting a positive family history. Although it is tempting to conclude that the presence of a positive family history should be used to guide genetic testing in this age range, there is likely an inherent bias in our cohort toward performing genetic testing in individuals with a family history, and the finding may simply be an artifact of this bias. Supporting this conjecture, 21 individuals without positive family histories had diagnoses that have a genetic basis.
A significant difference in age at onset was found among the etiologic groups, likely driven by the younger mean age at onset among patients having metabolic etiologies. Figure, B, illustrates that the percentage of individuals having a metabolic disorder decreased dramatically at about age 29 years. With that said, it should also be noted that we identified several individuals who manifested disease in their 40s with these etiologies as well. Figure, A, shows the substantial variability in the age at onset within the metabolic group.
Table 4 gives etiologic data from several studies4,5,8- 10 of EOD. It also summarizes data from the present study, highlighting the important ways in which this cohort differs from previous studies.
Together, neurodegenerative etiologies accounted for 31.1% of patients in our cohort. The number of individuals having a neurodegenerative cause increased with older age (Figure). A neurodegenerative etiology was uncommon before age 29 years, and the rate increased to account for roughly 30% of patients having onset after age 33 years. Because of the longer mean duration of symptoms before presentation, neurodegenerative etiologies may be slightly underrepresented at the older end of our age range relative to autoimmune or inflammatory etiologies. However, the overall delay from onset to presentation is likely to have a larger effect.
Previous investigations have focused on EOD or presenile dementia, typically defined as having onset before age 65 years. A prospective analysis of 112 patients determined to have dementia found that 19% had frontotemporal lobar degeneration and 12% were diagnosed as having AD.9 In another EOD population comprising 185 subjects, 34% had AD and 12% had FTD.4 These populations differed geographically as well as in the exclusion of patients with TBI in the study by Harvey et al,4 whereas the study by Panegyres and Frencham9 included the TBI. In our EOD series, we identified only 4 patients with AD, suggesting that AD as a cause of dementia between the ages of 17 and 45 years is uncommon, but it will become more important as a diagnostic entity after age 45 years. We also found that FTD accounted for 13.2% of our cohort, in good agreement with a well-constructed study4 of EOD from the United Kingdom that also required progressive cognitive decline and excluded TBI among the study cohort. Among other studies of EOD, FTD accounted for 2.5%5 to 38.4%9 of the defined populations (Table 4), likely related to methodological differences. The youngest individual in our cohort diagnosed as having FTD developed symptoms at age 20 years and was found to have a mutation in the gene encoding the microtubule-associated protein tau (MAPT). Another individual developed FTD symptoms at age 23 years unassociated with a mutation in MAPT or in progranulin (PGRN).
It was expected that few individuals would manifest AD in this age range, particularly because individuals with Down syndrome were excluded owing to preexisting mental retardation. However, it is striking that only 4 individuals had AD, 1 of whom had a known familial mutation in PSEN1. A recent literature review demonstrated that most individuals reported as having very young onset of AD (<35 years) had PSEN1 mutations.12 Although the significance of this finding is unclear because cases unidentified as having a mutation may not have been reported, this underscores that the diagnostic algorithm for YOD must differ in important ways from that of later-onset dementia.
Only 1 patient in our cohort had Lewy body dementia (LBD), similar to the finding that LBD is an uncommon cause for EOD, particularly in comparison with late-onset dementia cohorts.9,10 These data are somewhat at odds with the finding that 7% of individuals in an EOD cohort had LBD.4 The reasons for this difference are unclear.
Autoimmune or inflammatory causes were an important origin of dementia in this population, with multiple sclerosis (MS) accounting for 11.1% of patients in our cohort. This contrasts with the estimate that MS accounted for 3% of cases in a North American EOD series.5 That series included a large number of patients having TBI as a cause for their cognitive impairment. Excluding those having TBI, MS would have accounted for 4% of the resulting North American cohort,5 in good agreement with 4.3% reported in a London, England, EOD cohort.4 The agreement between these studies suggests that MS accounts for a substantially larger percentage of YOD cases than EOD.
Other inflammatory disorders such as neuropsychiatric lupus and autoimmune encephalopathy were also important causes of dementia in our cohort. To our knowledge, these etiologies have not been reported in series investigating EOD.4,5,8- 10,13 Autoimmune or inflammatory causes remained an important etiologic group throughout the age range of our sample numerically and as a percentage of patients manifesting disease at a given age (Figure). The autoimmune and inflammatory etiologic group are particularly important because they represent rare treatable and potentially reversible groups of patients with disease that can manifest as YOD or as EOD.14,15
Adult presentations of many inborn errors of metabolism have been described as case reports or as case series but have not been previously reported in investigations of cohorts of patients with EOD or YOD, to our knowledge. We specifically searched several of these diagnoses, assuming that some clinicians might simply code (for example) Niemann-Pick type C disease as a diagnosis and not separately code the individual's cognitive impairment. Adding these search terms yielded the conclusion that these diagnoses remain an important and identifiable cause for progressive dementia between the ages of 17 and 45 years.
Although relevant diagnoses were specifically searched in an attempt to identify patients, many individuals having these diseases did not have evidence of progressive cognitive decline at the time of evaluation and were excluded from the study. Therefore, although vascular etiologies are an important cause of cognitive impairment, the presentation in most is distinct from that in the cohort described herein. Vascular etiologies have previously been reported to be an important cause of EOD,5 although the definition of vascular dementia used in that study was not clearly delineated. Our study investigated a younger age range and explicitly required progressive cognitive impairment. These differences in inclusion and exclusion criteria likely account for the differing rates of vascular dementia between these cohorts.
It is striking that, despite extensive evaluation, no etiologic diagnosis was able to be established in 18.7% of the cohort. Indeed, even brain biopsy did not elucidate a diagnosis in 8 of these patients. This finding is not unique to YOD, and a recent study3 investigating diagnostic brain biopsy findings in the evaluation of dementia found that the most common biopsy finding for the series as a whole (37%) was nonspecific gliosis and that 43% of biopsy specimens in that series were nondiagnostic.
During the 11-year span during which our cohort was evaluated, diagnostic imaging, laboratory testing, pathological examination techniques, and genetic testing have undergone substantial changes. A prospectively designed single-center study would likely have had similar limitations because of the length of time necessary to collect a sufficient number of patients for analysis. A multicenter design might overcome this limitation.
The personal effect and societal costs associated with the development of dementia in this age range are enormous.16,17 Caregivers of patients with EOD report significantly higher caregiver burden compared with caregivers of patients with late-onset dementia.18 This higher burden may be due to less availability of community or social support groups, few service settings appropriate to the needs of patients with EOD or YOD, influence on family life, and financial burdens associated with the effect of the patient's disease on employment, savings, and other resources.16,18
A study19 investigating survival in an EOD cohort (age at onset, 45-65 years) reported that earlier onset did not predict a significantly different rate of decline and that the median survival of 6.08 years for EOD was comparable to that of late-onset AD. The EOD cohort consisted of individuals having AD, mixed dementia, vascular dementia, or dementia not otherwise specified. The median survival was similar to that reported for an early-onset AD cohort.6 These results suggest that individuals with EOD have dramatically reduced longevity. Given the differences in causative etiologies, and particularly the substantial number of individuals having autoimmune or inflammatory etiologies (some of which may be treatable), it is unclear whether individuals having YOD would have a similarly foreshortened lifespan.
Data obtained from this retrospective cohort do not enable us to perform a frequency analysis for various diagnoses or to estimate the prevalence or incidence of YOD. To accurately ascertain these figures, a population-based study would be required; however, the rarity of YOD would make establishing such a cohort impractical. A regional or national registry might provide a resource by which this population could be better defined and characterized. It could also better characterize the societal burden of such diseases and provide valuable insights into how the evaluation of YOD might be optimized. Furthermore, the existence of a registry could provide families of such individuals with a valuable support network to address the innumerable personal, social, spiritual, financial, and other implications of the development of progressive dementia at a young age.
Young-onset dementia (age at onset, <45 years) includes different etiologies, with few patients having a potentially treatable disorder. The etiologic spectrum and the relative percentages of patients within etiologic groups differed in important ways from existing reports of EOD (ie, age at onset, <65 years).
Correspondence: Keith A. Josephs, MST, MD, Department of Neurology, Mayo Clinic, 200 First St SW, Rochester, MN 55905 (firstname.lastname@example.org).
Accepted for Publication: April 6, 2008.
Author Contributions: All authors had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Kelley and Josephs. Acquisition of data: Kelley. Analysis and interpretation of data: Kelley, Boeve, and Josephs. Drafting of the manuscript: Kelley. Critical revision of the manuscript for important intellectual content: Kelley, Boeve, and Josephs. Statistical analysis: Kelley and Josephs. Administrative, technical, and material support: Kelley and Boeve. Study supervision: Josephs.
Financial Disclosure: None reported.