Assessment of executive function declines in presymptomatic and mildly symptomatic familial frontotemporal dementia: NIH-EXAMINER as a potential clinical trial endpoint.

INTRODUCTION: Identifying clinical measures that track disease in the earliest stages of frontotemporal lobar degeneration (FTLD) is important for clinical trials. Familial FTLD provides a unique paradigm to study early FTLD. Executive dysfunction is a clinically relevant hallmark of FTLD and may be a marker of disease progression.
METHODS: Ninety-three mutation carriers with no symptoms or minimal/questionable symptoms (MAPT, n = 31; GRN, n = 28; C9orf72, n = 34; Clinical Dementia Rating scale plus NACC FTLD Module < 1) and 78 noncarriers enrolled through Advancing Research and Treatment in Frontotemporal Lobar Degeneration/Longitudinal Evaluation of Familial Frontotemporal Dementia Subjects studies completed the Executive Abilities: Measures and Instruments for Neurobehavioral Evaluation and Research (NIH-EXAMINER) and the UDS neuropsychological battery. Linear mixed-effects models were used to identify group differences in cognition at baseline and longitudinally. We examined associations between cognition, clinical functioning, and magnetic resonance imaging volumes.
RESULTS: NIH-EXAMINER scores detected baseline and differences in slopes between carriers and noncarriers, even in carriers with a baseline Clinical Dementia Rating scale plus NACC FTLD Module = 0. NIH-EXAMINER declines were associated with worsening clinical symptoms and brain volume loss. DISCUSSION: The NIH-EXAMINER is sensitive to cognitive changes in presymptomatic familial FTLD and is a promising surrogate endpoint.

Background

Neurodegenerative diseases, such as Alzheimer’s disease (AD) and frontotemporal lobar degeneration (FTLD), are major public health concerns in our growing aging population. A concerted effort is underway to develop disease-modifying treatments, but recent trials have failed to demonstrate efficacy in the symptomatic phases [1,2]. Recently, pharmaceutical trials have shifted their focus to patients in the presymptomatic stages of disease [3], based on the lack of efficacy in the symptomatic stages and promising work in preclinical animal models [4]. Patients with autosomal dominant mutations that cause neurodegeneration offer the opportunity to study treatments in the earliest phases of illness.

About 30% of FTLD cases are familial (f-FTLD), most often due to mutations in the C9orf72, GRN, or MAPT genes [5]. Detailed knowledge of the molecular pathophysiology of these mutations has resulted in more refined therapeutic targets [6]. Thus, trials in the presymptomatic or questionably symptomatic phases of f-FTLD are expected in the near future [6]. In anticipation of such studies, we are faced with the challenge of validating potential endpoints in presymptomatic and mildly symptomatic mutation carriers, in whom symptoms are particularly subtle and difficult to identify and monitor.

Tracking cognition in f-FTLD is further complicated by the diversity of phenotypes that manifest; these include behavioral variant of frontotemporal dementia (bvFTD), primary progressive aphasia (PPA), motor neuron disease, and parkinsonism. In symptomatic FTLD, neuropsychological changes can be appreciated in several cognitive domains. Although episodic memory deficits are appreciated in bvFTD [7,8] and language deficits are required to diagnose PPA [9], executive function deficits are common across clinical syndromes [10-13]. Relatively less is known about the presymptomatic phases of f-FTLD, but emerging literature suggests that early executive deficits may be a primary cognitive feature [14-16].

Executive dysfunction can be challenging to evaluate with a single measure because this umbrella term comprises several subdomains, such as working memory, set-shifting, and inhibition. The Executive Abilities: Measures and Instruments for Neurobehavioral Evaluation and Research (NIH-EXAMINER) is a psychometrically validated, computerized battery developed to quantify many facets of executive functions [17]. The core battery includes six subtests combined using item response theory (IRT) to calculate an overall Executive Composite score and three factor scores. The NIH-EXAMINER is sensitive to executive dysfunction in symptomatic, sporadic FTLD [12]. Given its comprehensive and IRT-based approach to measuring executive functions, we hypothesized that the NIH-EXAMINER will be more sensitive to cognitive changes in early f-FTLD than a standard paper-and-pencil test, Trail Making Test (TMT), Part B.

The goal of the present study was to assess the ability of the NIH-EXAMINER to detect cognitive changes in the earliest stages of f-FTLD, using the Longitudinal Evaluation of Familial Frontotemporal Dementia Subjects (LEFFTDS; U01AG045390) and Advancing Research and Treatment in Frontotemporal Lobar Degeneration (ARTFL; U54NS092089) cohorts. These cohorts have been established by a consortium of 18 centers in the US and Canada, which follows f-FTLD family members, both mutation carriers and noncarriers, longitudinally. The present work investigates differences in baseline and longitudinal executive function between early-stage f-FTLD mutation carriers and a cohort of noncarrier family members. We examined the NIH-EXAMINER Executive Composite score and three factor scores, as well as a traditional paper-and-pencil measure of executive functions and processing speed. As episodic memory is known to be a common feature in bvFTD, we also investigated visual and verbal memory measures as potential endpoints. The results of this work will help inform endpoint selection for future clinical trials that enroll patients in the earliest stages of f-FTLD.

Methods

Study participants

Participants were 169 members of families affected by genetic forms of FTLD who were enrolled in the LEFFTDS and ARTFL studies, which include annual longitudinal evaluations. Participants for this study were enrolled if at least one first-degree relative had a mutation in the MAPT, GRN, or C9orf72 genes. In the present study, carrier inclusion was based on the functional rating from the CDR® Dementia Staging Instrument plus Behavior and Language domains from the NACC FTLD Module (CDR® plus NACC FTLD; Olney et al., this issue) scale. This measure of functional impairment includes ratings of six domains of cognition and daily function that are included in the traditional CDR domains, plus two additional domains assessing core features of FTLD: language and behavior. Mutation carriers were included if they were deemed to be asymptomatic (CDR® plus NACC FTLD = 0) or mildly/questionably symptomatic (CDR® plus NACC FTLD = 0.5) at baseline. Noncarrier family members who met study inclusion/exclusion criteria were included as controls given their similar early environment, genetic background, and demographics. Participants provided informed consent before study procedures, and the study was conducted in accordance with Internal Review Board approval.

Measures

Neuropsychological evaluation

All patients enrolled in the ARTFL/LEFFTDS protocol were administered the neuropsychological battery from the Uniform Data Set (UDSNB), version 3 [18]. This battery includes the Montreal Cognitive Assessment (MoCA), a screen of general cognitive functioning. Participants also received the TMT [19]. In part A of this timed test, participants connect letters in order as quickly as possible. In part B, participants alternate between connecting letters and numbers in sequential order. Part A measures processing speed, whereas part B requires set-shifting, an element of executive functioning. The UDSNB includes two measures of phonemic fluency and two measures of category fluency, which were summed together to create a UDSNB Verbal Fluency score to determine whether a standard paper-and-pencil measure of verbal fluency performed similar to the NIH-EXAMINER, IRT-derived fluency composite. Memory measures included the California Verbal Learning Test, Short Form [20], a nine-item list learning task. The outcome was free recall after a 10-minute delay. Visual memory was quantified as the number of correct items recalled 10 minutes after copying the Benson Figure [18].

NIH-EXAMINER

The NIH-EXAMINER is a computerized battery developed to be a clinical trial endpoint. Six subtests were administered to ARTFL/LEFFTDS participants and combined to form an Executive Composite score using the IRT, as well as three factor scores: Working Memory, Cognitive Control, and Fluency factors. The NIH-EXAMINER scoring program computes a standard error of measurement for each composite score for each individual. Consistent with recommendations from the normative study [17], we removed an individual’s data for any composite score if that score’s standard error was greater than 0.75 (Executive Composite, n = 1; Working Memory, n = 26). For each individual, all other composites scores below the standard error threshold were retained. Descriptions of each subtest are presented in the supplemental materials and initial publication [17].

FTLD-specific Clinical Dementia Rating scale

Disease severity was defined using the CDR® plus NACC FTLD [21]. The eight domain scores were summarized to create a global score using a recently developed algorithm (Olney et al., this issue). In addition, each individual domain was scored on a scale from 0 to 3, and the raw values from each domain were summed to create the CDR® plus NACC FTLD Sum of Boxes (CDR® plus NACC FTLD-SB), an integer (0–24) measure of symptom severity.

Genetic testing

All participants had genetic testing at the same laboratory at the University of California, Los Angeles using published methods (Ramos, this issue); a brief description is also included as a supplement in this article.

Neuroimaging

Participants were scanned at 3 tesla on magnetic resonance imaging scanners from one of three vendors: Philips Medical Systems, Siemens, or General Electric Medical Systems. A standard imaging protocol [22] was used by all centers and managed and reviewed for quality by a core group at the Mayo Clinic, Rochester. To create four lobar regions of interest, we summed all modulated gray matter atlas regions of interest bilaterally within each lobe.

Statistical analysis

We fitted linear mixed-effects models allowing for random intercepts and slopes in modeling longitudinal trajectories. Our primary outcomes of interest were the NIH-EXAMINER Executive Composite and TMT-B, evaluated in separate models. The primary predictors were group, time (in years, as a continuous variable), and the interaction between group (carriers vs. noncarriers) and time, allowing for assessment of group differences at baseline and in rates of change. The models also covaried for baseline age, education, and gender, as well as the interaction of each of these terms with time. We assessed baseline differences via the group effect at time zero (baseline) and differential rates of decline between carriers and noncarriers by examining the group by time interaction. Secondary analyses also examined each of the three subcomponents of the NIH-EXAMINER composite (i.e., Working Memory, Cognitive Control, and Fluency) to explore which components might be driving observed effects in the primary EXAMINER score. We also secondarily examined standard neuropsychological measures of memory (California Verbal Learning Test and Benson Recall), processing speed (TMT-A), and verbal fluency (UDS Verbal Fluency score), as well as the CDR® plus NACC FTLD-SB in separate models. We observed some non-normality of TMT-B residuals; log transformation led to residuals that were approximately normal. We ran a sensitivity analysis using the log-transformed variable, and the same pattern of results remain. We report nontransformed data to enhance interpretability of the parameter estimates.

Effect sizes for each measure were evaluated by calculating sample sizes (per arm) required to detect 25% and 40% reductions in decline [23], using 10,000-fold bootstrapping as described in Supplementary Methods.

In follow-up analyses, we assessed the associations between the NIH-EXAMINER Executive Composite score, CDR® plus NACC FTLD-SB, and the four lobar volumes (see Supplementary Methods for details). We also analyzed the association between the Executive Composite and lobar volumes in the noncarrier controls to further validate the association between this measure and neural tissue.

To mimic the conditions of clinical trial enrolling presymptomatic carriers, we conducted follow-up analyses only including carriers with a baseline CDR® plus NACC FTLD-SB = 0. As most trials use two measurements (i.e., pre- and post-test) acquired over a ~1-year time frame, we also restricted our sample to those whose second visit was acquired within 1.5 years of baseline. These two conditions were applied to create a restricted data set that was used for the sensitivity analyses.

Results

Demographics, genetic, and clinical characteristics

Participants were 93 mutation carriers (31 MAPT, 28 GRN, and 34 C9orf72), 66 of whom had follow-up data. Seventy-eight family members without a mutation (noncarriers) were included as a comparison group, 43 with follow-up data. Demographics are presented in Table 1.

Table 1

Participant demographics

Measure	Mutation carriers	Noncarriers
N	93	78
n with longitudinal data	66	49
Total observations	182	137
Visits per person (range)	1.9 (1–3)	1.6 (1–3)
Age (SD)	46.0 (13.9)	48.8 (13.4)
Education (SD)	15.8 (2.3)	15.3 (2.7)
Male [n (%)]	49 (52.7)	37 (47.4)
Baseline cognition
MoCA	26.71 (2.74)	26.95 (2.45)
CDR® plus NACC FTLD = 0 [n(%)]	64 (68.8)	64 (83.1)
CDR® plus NACC FTLD = 0.5 [n(%)]	29 (31.18)	13 (16.9)

NOTE. Parenthetic values are standard deviations (SDs) unless otherwise noted.

Abbreviations: FTLD, frontotemporal lobar degeneration; CDR® plus NACC FTLD, Clinical Dementia Rating scale plus National Alzheimer Coordinating Center FTLD Module; MoCA, Montreal Cognitive Assessment.

Group differences in neuropsychological performances

Group differences in baseline cognition and longitudinal cognitive slopes are presented in Table 2. Please see Supplementary Table 1 for a description of baseline performances and rates of change for carriers and noncarriers separately. At baseline, mutation carriers performed worse than noncarriers on the Executive Composite, and there was a statistically significant group by time interaction, indicating that carriers had a faster declining longitudinal change (steeper slope) on the Executive Composite (Fig. 1A). A preliminary exploration of Executive Composite decline by mutation type (i.e., C9orf72, MAPT, PGN) did not find a statistically significant difference in their slopes (b = .1, P = .086, 95% confidence interval [CI] [−.02, .24]). For TMT-B, although neither baseline nor longitudinal differences reached statistical significance, both were in the expected direction. The same pattern of results was observed for Trails A, with both baseline and longitudinal differences reaching statistical significance. Similar to the Executive Composite, carriers performed worse than noncarriers at baseline and showed greater increases in functional impairments (CDR® plus NACC FTLD-SB) over time.

Table 2

Baseline and longitudinal differences in neuropsychological performances between carriers and noncarriers

	Baseline differences		Differences in slope
	b (95% CI)	p	b (95% CI)	p
NIH-EXAMINER
Executive Composite	0.22 (−0.4, −0.04)	.016	−0.18 (−0.32, −0.05)	.008
Factor scores
Working Memory	−0.12 (−0.33, 0.09)	.252	0.03 (−0.15, 0.21)	.747
Cognitive Control	−0.3 (20.49, −0.12)	.001	−0.11 (−0.23, 0.02)	.086
Fluency factor	−.07 (−0.27, 0.14)	.51	20.25 (−0.41, −0.08)	.003
CDR® plus NACC FTLD-SB	0.21 (0.04, 0.37)	.014	0.43 (0.07, 0.79)	.018
UDSNB measures
Trails A	2.91 (0.67, 5.14)	.011	2.29 (0.01, 4.56)	.049
Trails B	8.24 (20.08, 16.56)	.052	9.79 (−0.31, 19.89)	.057
Verbal Fluency	−0.91 (−5.44, 3.62)	.694	23.37 (−6.15, −0.58)	.018
CVLT-SF 10′ Delay	−0.39 (−0.98, 0.2)	.196	20.28 (−0.75, 0.19)	.238
Benson Delayed Recall	−0.55 (−1.31, 0.2)	.151	20.32 (−0.98, 0.34)	.338

NOTE. b is the unstandardized parameter estimate, with noncarriers = 0 and carriers = 1.

Abbreviations: CVLT-SF, California Verbal Learning Test, Short Form; FTLD, frontotemporal lobar degeneration; CDR® plus NACC FTLD-SB, Clinical Dementia Rating scale plus National Alzheimer Coordinating Center FTLD Module Sum of Boxes; NIH-EXAMINER, NIH–Executive Abilities: Measures and Instruments for Neurobehavioral Evaluation and Research; UDSNB, neuropsychological battery from the Uniform Data Set.

Fig. 1.

Baseline differences and longitudinal executive function declines are detectable in presymptomatic and mildly/questionably symptomatic familial FTLD using the NIH-EXAMINER. NOTE. These figures display fitted regression lines of each group’s mean trajectory estimated by the fixed, carrier status by time interaction term in the linear mixed-effects model. Error bars represent the 95% confidence intervals. * indicates baseline differences (P = .016). *** indicates longitudinal differences (P<.009). (A) This sample includes 93 mutation carriers with a global CDR® plus NACC FTLD = 0 or 0.5 at their baseline visit. Mutation carriers are compared with 78 noncarrier controls using linear mixed-effects models. This figure displays the fitted results of the mutation status by time interaction from a linear mixed-effects model, showing mutation carriers had a significantly more negative slope on the Executive Composite than noncarriers and significantly poorer performance at baseline. EXAMINER Executive Composite scores are displayed on the y-axis in z-score units. The arrow indicates that lower scores are associated with poorer performance. (B) This sample includes 66 mutation carriers with a global CDR® plus NACC FTLD = 0 at their baseline visit, compared with 64 noncarrier controls. This figure displays the fitted results of the mutation status by time interaction from a linear mixed-effects model. Mutation carriers showed a significantly more negative slope on the Executive Composite than noncarriers. Baseline performance did not differ significantly. The y-axis is in z-score units; the arrow signifies that lower scores on this composite indicate poorer performance. Abbreviations: FTLD, frontotemporal lobar degeneration; CDR® plus NACC FTLD, Clinical Dementia Rating scale plus National Alzheimer Coordinating Center FTLD Module; NIH-EXAMINER, NIH–Executive Abilities: Measures and Instruments for Neurobehavioral Evaluation and Research.

In addition, we investigated whether any of the three subtests comprising the Executive Composite differed between carriers and noncarriers. For the Fluency factor score, there was a statistically significant difference in the longitudinal trajectories between groups, in the expected direction. We also looked at a summary score of UDSNB Verbal Fluency measures. Similar to the EXAMINER Fluency factor score, there was a statistically significant difference in the longitudinal trajectories between groups, in the expected direction. Baseline differences in UDSNB Verbal Fluency were also in the expected direction but did not reach statistical significance. For the Control factor, carriers exhibited worse baseline performance than noncarriers, and estimated group differences in rate of decline were in the expected direction, although not statistically significant. For the Working Memory factor, both baseline and longitudinally, results were in the expected direction but did not reach statistical significance. For measures of visual and verbal episodic memory, baseline and longitudinal differences on both measures were in the expected direction not statistically significant.

Sample size estimates

We calculated the sample size (per arm) required to detect a moderate (40%) reduction in decline in a 1-year clinical trial (power = .80, alpha = .05), assuming 20% attrition. Sample size estimates (Table 3) were generally lower when the sample was restricted to those with a baseline CDR® plus NACC FTLD = 0.5, although the differences did not reach statistical significance based on the 95% CI. NIH-EXAMINER composite and factor scores and the CDR® plus NACC FTLD-SB were among the largest effect sizes regardless of the inclusion criteria.

Table 3

Estimated sample sizes to detect therapeutic effects in f-FTLD for trials with two different enrollment criteria based on baseline CDR® plus NACC FTLD

	Baseline CDR® plus NACC FTLD 0 or 0.5		Baseline CDR® plus NACC FTLD = 0.5
Cognitive measure	n	Sample size (95% confidence interval)	n	Sample size (95% confidence interval)
NIH-EXAMINER
Executive Composite	66	2336 (565, >100,000)	19	539 (142, 85,401)
Fluency factor	63	898 (330, 11,085)	17	245 (89, 3324)
Cognitive Control	60	15,271 (1411, >100,000)	18	>100,000 (>100,000, NE)
Working Memory	55	1507 (379, >100,000)	17	4208 (223, >100,000)
CDR® plus NACC FTLD-SB	61	1018 (559, 2586)	18	514 (290, 1374)
UDS measures
TMT Part A	64	966 (343, 16,290)	17	1162 (251, >100,000)
TMT Part B	63	2135 (677, >100,000)	16	1118 (153, >100,000)
UDSNB Verbal Fluency	63	1721 (470, >100,000)	17	362 (126, 11,563)

NOTE. This table presents sample size estimates for each arm of a clinical trial to detect a moderate (40%) reduction in slope, assuming power = 0.8, alpha = .05, and 20% attrition rate. 95% Confidence intervals were calculated using a 10,000-fold bootstrap procedure. NE denotes not estimated by the bootstrapping procedure.

Abbreviations: FTLD, frontotemporal lobar degeneration; CDR® plus NACC FTLD, Clinical Dementia Rating scale plus National Alzheimer Coordinating Center FTLD Module; CDR® plus NACC FTLD-SB, Clinical Dementia Rating scale plus National Alzheimer Coordinating Center FTLD Module Sum of Boxes; NIH-EXAMINER, NIH–Executive Abilities: Measures and Instruments for Neurobehavioral Evaluation and Research; TMT, Trail Making Test; UDSNB, neuropsychological battery from the Uniform Data Set.

Follow-up analyses

NIH-EXAMINER performance was analyzed in carriers (n = 66) with a baseline CDR® plus NACC FTLD = 0 and a follow-up assessment within 1.5 years of baseline (Fig. 1B), compared with 64 noncarriers. Mutation carriers continued to show a significantly more negative slope on the Executive Composite than noncarriers (b = −0.23, P = .006, 95% CI [−0.39, −0.06]). Baseline performance did not differ significantly (b = −0.10, P = 0.368, 95% CI [−0.31, 0.11]). For TMT-B, although these carriers performed worse at baseline than noncarriers (b = 8.36, P = .015, 95% CI [1.61, 15.12]), longitudinal trajectories did not statistically differ (b = 8.37, P = .11, 95% CI [−1.9, 18.64]). The model for the CDR® plus NACC FTLD-SB did not converge possibly because only 5 of 43 asymptomatic carriers showed any change between baseline and follow-up.

To further support the clinical significance of the NIH-EXAMINER, we evaluated its association with the CDR® plus NACC FTLD-SB and brain volumes (Table 4). In mutation carriers, those with higher mean Composite scores showed higher SB scores (b = −1.13, P <.001, 95% CI [−1.49, −1.72]), and those with greater longitudinal executive declines showed steeper increases in CDR® plus NACC FTLD-SB (more functional loss), (b = −2.33, P <.001, 95% CI [−2.93, −1.72]). Within-person declines in the Executive Composite were associated with significantly greater volume loss over time in frontal and parietal lobes; relationships were also in the expected direction for occipital and temporal lobes but were not statistically significant. Between-person estimates showed that greater volume (averaged across timepoints) in all lobes was associated with better executive performance. In noncarrier controls (Table 4), greater volume in all lobes was associated with Executive Composite scores (between-person), although longitudinal, intraindividual associations between the composite and volume loss did not reach statistical significance.

Table 4

Associations of the NIH-EXAMINER Executive Composite with lobar volumes

	b coefficient	P value	95% CI
Mutation carriers
Frontal
Within-person*	2386.65	.001	996.91, 3776.39
Between-person*	2616.41	.03	260.51, 4972.32
Parietal
Within-person*	1118.90	<.001	499.43, 1738.39
Between-person*	1325.34	.011	306.57, 2344.11
Temporal
Within-person	581.14	.056	−15.07, 1177.36
Between-person*	1234.14	.032	106.83, 2361.44
Occipital
Within-person	232.96	.08	−28.27, 494.20
Between-person*	597.59	.011	134.17, 1061.02
Noncarriers
Frontal
Within-person	−691.05	.463	−2536.44, 1154.35
Between-person*	6181.13	.002	2309.18, 10,053.07
Parietal
Within-person	−313.87	.588	−1450.82, 823.09
Between-person*	2365.52	.004	775.49, 3955.55
Temporal
Within-person	−933.83	.082	−1985.83, 118.17
Between-person*	2596.59	.002	951.58, 4241.59
Occipital
Within-person	290.31	.715	−574.63, 394.00
Between-person*	1256.67	.002	470.06, 2043.28

NOTE. Within-person results (unstandardized b) indicate the change in brain volume (mm3) associated with a 1 z-score loss of NIH-EXAMINER Composite performance over time. Between-person results (unstandardized b) indicate overall relationships, across visits, among lobar volumes (mm3) and NIH-EXAMINER Composite performance.

Abbreviations: NIH-EXAMINER, NIH–Executive Abilities: Measures and Instruments for Neurobehavioral Evaluation and Research.

P <.05.

Discussion

This study evaluated longitudinal changes in executive functions, processing speed, and memory in the earliest stages of f-FTLD using the NIH-EXAMINER and paper-and-pencil measures. At baseline, mutation carriers performed worse on the NIH-EXAMINER Executive Composite than did noncarrier family members. Nonmutation carriers appeared to improve over time, possibly due to practice effects. Carriers showed significantly steeper, negative longitudinal trajectories on this composite compared with nonmutation carriers. This finding remained statistically significant when the sample was restricted to the first two visits of carriers without any observable symptoms at baseline (CDR® plus NACC FTLD = 0). Moreover, in a clinical trial using the NIH-EXAMINER, the sample size required to detect a treatment effect would be over 50% less than a study using TMT-B as the endpoint. We further assessed the clinical relevance of the NIH-EXAMINER by analyzing its association with the CDR® plus NACC FTLD-SB. Consistent with expectations, those with greater functional impairment performed worse on the composite and greater within-person NIH-EXAMINER change tracked with longitudinal functional decline. The association of the NIH-EXAMINER with atrophy rates provided another measure of the validity of the Executive Composite, suggesting an association with the underlying disease neurobiology. Taken together, these results provide evidence that mutation carriers show detectable, early declines in executive functions and suggest that the NIH-EXAMINER has potential as a sensitive, surrogate endpoint for clinical trials of presymptomatic and mildly symptomatic f-FTLD. Although the CDR® plus NACC FTLD-SB was also able to detect longitudinal changes in the entire sample, when limiting the sample to those at the very earliest stages (global CDR® plus NACC FTLD = 0), very few people showed any change. This suggests that the CDR® plus NACC FTLD-SB may perform well in trials including carriers at the mild cognitive impairment stage, whereas it may not be sufficiently sensitive in trials enrolling asymptomatic carriers.

The NIH-EXAMINER was designed as a clinical trial endpoint to provide a more psychometrically robust composite score, compared with standard neuropsychological assessments [17]. We demonstrate that the NIH-EXAMINER is sensitive to detecting early cognitive changes in presymptomatic and mildly/questionably symptomatic mutation carriers, consistent with the earlier demonstration of its utility in symptomatic FTLD [12]. Convergent validity is demonstrated by showing longitudinal associations with functional independence and atrophy. This adds to prior research indicating that the NIH-EXAMINER Executive Composite correlates with real-world executive behavior and dorsolateral prefrontal volumes [24]. The advantage of the NIH-EXAMINER may stem in part from the wide range of executive constructs that are captured. This multidomain assessment is also structured for modularity, such that individual subtests can be administered in isolation. For example, if validation studies suggest the Verbal Fluency composite produces the largest effect sizes in a certain genetic variant, that composite could be used as a trial endpoint, reducing trial and participant burden. Moreover, the NIH-EXAMINER uses IRT methods [25,26] to generate composite scores; these methods allow for composite construction even in the absence of certain subtests, making it particularly useful for longitudinal studies such as clinical trials. All NIH-EXAMINER components, in English and Spanish, are in the public domain and freely available to qualified users at http://memory.ucsf.edu/resources/examiner.

This study contributes to a growing body of literature characterizing cognitive trajectories in the initial phases of f-FTLD. Jiskoot et al. [27] published two longitudinal studies of a cohort of 46 presymptomatic MAPT and GRN mutation carriers with 2 years and 4 years of follow-up data [28]. At 4-year follow-up, although declines were observed in memory, language, and social cognition scores, no differences between carriers and controls were seen in the in longitudinal trajectories of scores derived from paper-and-pencil executive function measures. However, longitudinal decline in executive functions was observed in those GRN and MAPT mutation carriers who later developed symptoms. An earlier cross-sectional study (78 asymptomatic carriers) showed executive function differences between noncarriers and carriers (MAPT and GRN) in the presymptomatic stage [14], replicating the results of several smaller studies in carriers of GRN (n’s = 8 and 13) [29,30] and MAPT (n’s = 4 and 10) [15,16] mutations, respectively. Our work buttresses findings that declines in processing speed (TMT-A) and executive functions are detectable in the early stages of f-FTLD and builds on these works by showing these declivities are associated with longitudinal changes in clinical status and brain volumes. Furthermore, our study includes carriers of the C9orf72 expansion, whereas many prior studies did not. In contrast to the study by Jiskoot et al. described previously [28], we did not find statistically significant longitudinal differences in episodic memory, although the coefficients were in the expected direction. Follow-up analyses suggested that the verbal fluency domain of executive functions appears to be a particularly affected in early stages of f-FTLD, and the NIH-EXAMINER Fluency factor produces the most encouraging effect sizes for clinical trials. Given that the Fluency factor and TMT-A both rely on processing speed, follow-up studies of processing speed measures in this cohort may be warranted.

This study has several limitations. First, we group the three major classes of f-FTLD mutations in a single analysis to maximize power. As sample sizes continue to grow in this and other cohorts, these findings should be replicated in larger groups of each genetic variant to confirm the generalizability of the NIH-EXAMINER for trials enrolling carriers of particular mutation types. A second limitation is the large sample size required to detect a treatment effect in the early phases of f-FTLD. It is important to note that this calculation was based on a convenience sample, as we did not limit inclusion to those most likely to exhibit clinical change during the study period. The mean age of our cohort is 46 years, which is younger than the mean age of onset for f-FTLD mutations (50 to early 60s; Olney, this issue) [31]. Our recent work [32] suggests that brain atrophy may improve the ability to predict which mutation carriers are most likely to convert to dementia, thereby improving power and reducing the number of participants required to detect an effect. Further work will be needed to validate the NIH-EXAMINER in the symptomatic phases of FTLD; we refer readers to our study on other potential clinical and neuroimaging endpoints in symptomatic FTLD [33].

Despite its limitations, the present study adds to our knowledge of cognition in presymptomatic and mildly/questionably symptomatic f-FTLD and provides indications that the NIH-EXAMINER is well suited to detect early changes in a heterogeneous group of f-FTLD mutation carriers and is associated with the neurobiology of f-FTLD. Given the relatively large sample sizes required for a trial in mutations carriers with CDR® plus NACC FTLD = 0 and 0.5, however, improved enrichment strategies for enrolling patients closest to developing unequivocal dementia, major refinements, new scales, or different methodologic approaches may be necessary to measure clinical changes in presymptomatic to very early symptomatic trial participants. Regardless, the results are encouraging and suggest that further study of the NIH-EXAMINER as a potential surrogate endpoint for f-FTLD is warranted.