Huntington’s disease (HD) is an autosomal-dominant neurodegenerative disease that is characterized by a triad of progressive motor, cognitive, and behavioral abnormalities. The diagnosis of manifest HD is established by unequivocal motor signs, however, progressive cognitive impairment typically begins much earlier (1–4). Capturing cognitive deficits in the pre-motor manifest stage (preHD) remains critically important for detecting the earliest signs of decline and improving the precision of disease staging systems (5, 6).

The cognitive profile of HD has been extensively studied and is characterized by marked declines in executive functioning and processing speed (6), which affects multiple other domains (1–3, 6) and performance on many neuropsychological tests, including memory-based tasks (5, 7–10). For example, early manifest HD patients often exhibit more pronounced deficits during the retrieval rather than encoding phase on free-recall memory paradigms (11–15). In preHD, these deficits are more subtle and, depending on the cognitive assessment, may go undetected. For example, while the Wechsler Memory Scale failed to differentiate preHD from controls (16), other paradigms, such as verbal learning tests, were sensitive to these early changes (8, 17).

A pilot study by Sierra et al. (17) showed that the Loewenstein–Acevedo Scales for Semantic Interference and Learning (LASSI-L) (18)–a verbal learning paradigm specifically designed to elicit interference effects and intrusions during memory retrieval–captured robust differences between preHD and controls. Notably, the LASSI-L also appeared to outperform several commonly used tests in preHD–including Symbol Digit Modalities Test (SDMT) (19), Stroop Color Word Test (SWRT) (20), Trail Making Test A/B (TMT) (21), and semantic fluency (Animals) (17, 22). These preliminary results suggested that the LASSI-L could detect early cognitive changes in preHD. However, replication and validation studies had yet to be performed.

In this study, we sought to replicate and extend the findings previously reported by Sierra et al. (17) and to further analyze the sensitivity, validity and utility of the LASSI-L in preHD. Our approach was guided by the recommendations for the use and validation of cognitive scales in HD outlined by the International Parkinson and Movement Disorders Society (MDS) (10). Specifically, we administered the LASSI-L to larger preHD and control cohorts, which allowed for specificity, sensitivity and receiver operating curve (ROC) analyzes. We then compared preHD performance at Beth Israel Deaconess Medical Center (BIDMC) to an age and education-matched preHD cohort at University of California, San Diego, School of Medicine (UCSD) to determine whether the findings could be replicated across sites. We analyzed longitudinal changes in performance on the LASSI-L to investigate test–retest reliability and practice effects. We directly compared individual performance on LASSI-L to the California Verbal Learning Test (CVLT-II) (14), a similar paradigm previously used in HD. We then repeated these group comparisons using a smaller preHD cohort restricted to participants predicted to be early in the disease course (Far subgroup). Finally, to establish construct validity, we correlated LASSI-L performance with established neuroimaging biomarkers in preHD: volumetric decline in caudate, putamen, globus pallidus, and nucleus accumbens (23).

Fifty participants (25 preHD and 25 HC) were enrolled in the study at BIDMC. Age range for this cohort was 25–65; mean age was 40.68 (SD = 11.12) for preHD and 36.84 (SD = 12.02) for HC. Mean education was 15.76 years (SD = 2.67) for preHD and 16.40 (SD = 2.31) for HC. 72% of the preHD group and 52% of the HC group were female. Mean CAG repeat length was 42.46 (SD = 2.28). Mean CAPs score was 0.79 (SD = 0.23). Mean TMS for the preHD group was 2.44 (SD = 2.68) and no participant had a TMS > 8 (Table 1). SWRT, MMSE, Semantic Fluency, and TMT-A group differences were statistically significant between preHD and HC (Table 1). In a subgroup analysis of the 45/50 participants who completed the Q-Motor speeded tapping (22 preHD and 23 HC), there were no group differences on tap duration variability (non-dominant hand) or mean inter-tap interval (non-dominant hand). Groups differences above (demographics, cognitive testing) remained unchanged when restricting the analysis to the 45 participants who completed Q-Motor.

BIDMC preHD and HC group differences on the LASSI-L can be seen on Table 2. Additionally, PIE was statistically significant for both PSI Intrusions (p = 0.004; d = 0.307), and frPSI intrusions (p = 0.025; d = 0.056). Using ANCOVA, PSI group differences remained significant after controlling for age [F (1,30) = 4.60, p = 0.046] and education [F (1,4) = 28.47, p < 0.0001]. AUROCs were: PSI = 0.855, PIE = 0.718, Stroop Word Reading = 0.748, and SDMT = 0.638 (Figure 1). AUROC comparisons were: PSI vs. PIE (p = 0.04); PSI vs. SDMT (p = 0.003); and PSI vs. Word Reading (p = 0.10). The correct words on the final Delayed recall section originated from both lists in balanced fashion (53% list A, 47% list B).

Performance on LASSI-L was compared to CVLT-II for all participants who completed both tests (n = 33, 16 preHD and 17 HC). Significant differences between preHD and HC were observed for the following sections of CVLT-II: List A Trial 1–5 (total), List A Short Delay, List A Short Delay Cued Recall Total, and Trial B Recall; however, only Trial B Recall survived FDR < 0.05 (Table 3). A subsequent comparison of LASSI-L to CVLT-II was repeated using only the preHD participants in the Far subgroup, using CAPs score cutoff <0.85 (> 7.59 years from diagnosis). PreHD Far subgroup (n = 13) characteristics were: age 37.7 years (SD = 11.6), education 15.23 years (SD =3.11), TMS 1.69 (SD = 2.06) and CAPs 0.66 (SD = 0.15) The only significant group differences between Far subgroup and HC for any cognitive test included in our battery were for LASSI-L PSI (p = 0.003; d = 1.174) and MMSE (p = 0.013; d = 0.950); however, only PSI survived FDR < 0.05 (Table 4).

The UCSD preHD comparison cohort had the following characteristics: mean age 42.28 years (SD = 11.95), mean education 15.56 years (SD = 1.92), 56% female, CAG repeat length 42.16 (SD = 2.25), CAPs 0.80 (SD = 0.22). The BIDMC and UCSD cohorts were matched with respect to age, education, sex, CAG repeats, CAPs and TMS (Table 1). There were no group differences on SWRT, MMSE, and SDMT (Table 1). There were no group differences between preHD cohorts on any section of the LASSI-L (Table 2).

As of 3/15/2023, 17 participants (9 preHD and 8 HC) have completed the final 12-month visit. For preHD subjects in this subgroup, there were no significant changes in TMS score (p = 0.766; Table 5). For the entire subgroup (preHD and HC), there were no group differences in LASSI-L performance between timepoints; scores on several sections of the LASSI-L were highly correlated: PSI (r = 0.88; p = < 0.00001), frPSI (r = 0.80; p = 0.0001), RSI (r = 0.69; p = 0.002), and Delayed Recall (r = 0.68; p = 0.002; Table 5). Similarly, Q-Motor measures were highly correlated between timepoints (Table 5).

The theoretical performance threshold for PSI, (1.5 SD below HC mean = 10.65), was 7.20, indicating that 15/25 preHD (60%) and 3/25 (12%) controls were impaired on this test (Supplemental Figure S3).

In this study, we showed that the LASSI-L captured early cognitive deficits in preHD and outperformed several established assessments in HD, thereby extending the primary findings reported in 2023 by Sierra et al. (17) These results were reliable across different sites and examiners and in a longitudinal analysis, the LASSI-L did not show significant practice effect after a 12-month interval. The PSI section demonstrated an excellent AUROC that was significantly better than SDMT, indicating that it could effectively discriminate preHD from healthy controls. Finally, a significant correlation between the number of intrusions on PSI and an established neuroimaging biomarker in preHD - caudate atrophy-further confirmed the construct validity of this assessment.

The current study confirms that two sections of the LASSI-L are remarkably sensitive to early changes in HD: PSI (p = 0.0001; d = 1.548) and delayed recall (p = 0.0001; d = 1.262) (17). PSI measures the inference of list A on the initial cued recall of list B. In multiple preHD cohorts, PSI has repeatedly exhibited large effect sizes indicating that a susceptibility to interference on memory tasks may be one of the earliest and most reliable deficits in this population. The sensitivity of the PSI is likely secondary to the simultaneous demands this section exerts on cognitive domains that are central to early dysfunction in preHD (6): executive function is challenged by list interference, the need to inhibit cued intrusions, and the need to retrieve information; processing speed is challenged by short timed sections; memory is challenged by the need to encode two supra-span lists.

The final delayed recall section exhibited the next largest effect size (Cohen’s d = 1.262). Sierra et al. (17) had previously interpreted this finding to be the result of combined proactive and retroactive interference (i.e., list A on list B and vice versa). In the preHD group, correct responses on the final delayed recall section originated from both lists in equal fashion (53% list A, 47% list B), lending support to this interpretation. In the current study, significant group differences were now also observed on many other sections: A1 Cued Recall, A2 Cued Recall, frPSI, RSI, and Intrusions for both PSI and frPSI. The larger replication study was likely adequately powered to uncover small but significant effects on these sections as well.

The LASSI-L detected changes in a preHD cohort that was at an early stage in the disease process, which was confirmed by a low mean UHDRS™-TMS score (2.2), a low mean CAPs score (0.79) and Q-Motor scores that were similar to healthy controls. In a subgroup analysis restricted to preHD subjects predicted to be far from diagnosis, the LASSI-L was, after controlling for multiple comparisons, the only cognitive assessment that detected significant differences with healthy controls (Table 4). The LASSI-L’s ability to detect early changes in this Far subgroup further highlights its potential utility as a marker of early change.

The Huntington’s Disease Integrated Staging System (HD-ISS) online calculator (see footnote 2) revealed that the majority of Stage 0 participants were already showing signs of significant cognitive impairment on PSI and that substituting LASSI-L for SDMT could also have shifted 50% of Stage 1 participants to Stage 2. These results further suggest that deficits on PSI occur early in the disease process. Incorporating the LASSI-L as a landmark assessment into future iterations of such staging systems could help delineate substages or provide an additional reliable endpoint for clinical trials targeting these early periods.

An important consideration when using cognitive tests as endpoints is the potential for practice effects upon repeated exposure. Tests associated with significant practice effects can obscure clinical deterioration, which can increase sample size requirements or the time required to observe the effect of an intervention (36). In a small subgroup analysis of participants who were able to complete the 12 month study, several sections of the LASSI-L, most notably PSI, appeared to demonstrate good initial test–retest reliability over 12 months (r = 0.88) without clear practice effects (Table 5). Importantly, these results mirrored the lack of significant clinical decline, as assessed by TMS and Q-Motor (Table 5). Although this finding is consistent with prior reports of LASSI-L reliability (37), larger samples will be required to firmly confirm this trend.

The LASSI-L shares certain features with verbal learning tests commonly used in HD. To investigate the relative sensitivity of each paradigm without creating confounds, we substituted the CVLT-II for the LASSI-L at the 6 month visit. In our study, sections of both tests demonstrated an ability to capture interference effects in the preHD group (LASSI-L: PSI, frPSI, RSI and CVLT-II: Trial B, Tables 2, 3), suggesting that verbal learning paradigms that elicit strong interference effects (proactive and retroactive) may be more likely to detect the earliest cognitive changes in preHD.

Finally, it is useful to note that the PSI and associated errors (B1 intrusions) are recorded within the first 8 min of the test, indicating that future iterations of the LASSI-L could be significantly shortened without compromising test integrity. This feature may be particularly valuable for clinical trials that seek to limit participant fatigue.

Existing literature extensively documents memory impairments associated with premanifest or early-stage Huntington’s disease (HD) (6, 38–42). These impairments primarily stem from executive function deficits that manifest in various abnormalities, including utilization of passive learning strategies (43), and difficulties with source memory (38). Other studies have also indicated that individuals with HD are more susceptible to interference effects on the temporal order of closely presented stimuli (44) and exhibit perseveration during paired-associate learning (45) and extra-dimensional shift learning (46). These deficits may provide insights into why the LASSI-L PSI section, which requires disregarding or inhibiting recently encoded items from an initial list within a brief time frame, emerges as the most sensitive section for this population.

Considerable evidence supports the effectiveness of verbal learning tasks in detecting early changes in individuals with HD. Earlier studies conducted before genetic testing for HD was available yielded mixed results in the preHD population. However, a notable study utilizing the PREDICT-HD dataset (47) demonstrated significant differences in HVLT-R performance between preHD individuals and controls as early as DCL1 (minimal motor abnormalities). While their study did not find differences in the DCL0 group (corresponding to the preHD Far group in our study), the authors suggested that this may be due to ceiling effects in the selected measures and recommended the use of more challenging memory assessment tools.

The combination of short timed sections, cueing (which optimizes initial encoding of list A but induces interference/intrusions on list B), and repeated list switching in the LASSI-L provides a comprehensive approach for evaluating various cognitive domains, including psychomotor speed, interference, perseveration, and source retrieval, which have been identified as weaknesses in preHD. This straightforward paradigm offers a practical means to evaluate these cognitive aspects simultaneously.

The LASSI-L is a sensitive, reliable and efficient tool for the early detection of cognitive impairment in preHD. It is simple to administer, outperforms other cognitive tests, is not associated with clear practice effects and demonstrates evidence of construct validity. With further longitudinal validation, it could be a useful addition to current research and clinical trial protocols aiming to reliably capture the earliest cognitive changes in pre-motor manifest HD.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving human participants were reviewed and approved by Beth Israel Deaconess Medical Center Institutional Review Board. The patients/participants provided their written informed consent to participate in this study.

All authors listed have made a substantial, direct, and intellectual contribution to the work and approved it for publication.

This work was supported by the Huntington’s disease Society of America HDSA Human Biology Project.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.