Atypical multisensory integration in Niemann-Pick type C disease – towards potential biomarkers

Niemann-Pick type C (NPC) disease is a rare progressive lysosomal storage disorder caused by mutations in either the NPC1 or NPC2 gene, with about 95% of cases attributable to the former [1],[2]. Individuals with NPC cannot properly metabolize cholesterol and other lipids which accumulate in the brain and in visceral organs (e.g. liver and spleen), ultimately causing cell dysfunction and organ system failure. Although NPC1 and NPC2 proteins are expressed ubiquitously, brain tissue is the most severely affected, resulting in widespread intraneuronal storage of cholesterol and glycosphingolipids that ultimately results in massive neurodegeneration [3]-[6]. While appearing relatively typical during the early stages of the disease, over time NPC children develop vertical gaze palsy, motor system impairment, learning difficulties and clumsiness, as well as seizures [7]-[9]. Documented changes in brain include ectopic dendrite growth, altered synaptic connectivity affecting cortical pyramidal neurons, axonal degeneration, myelin loss, gliosis and the formation of neurofibrillary tangles similar to Alzheimer's disease [10],[11]. Neuronal death is prominent in some brain regions such as the cerebellum where Purkinje cells selectively die, undoubtedly contributing to the clinically-evident motor system dysfunction [5],[10],[12]. Effective treatments are limited, although promising clinical trials are underway based on results in animal models of NPC [11],[13],[14].

Key to advancing new treatments for this and related lysosomal diseases with neural involvement is the development of objective biomarkers of neurological function against which the efficacy of new drugs can be tested in human patients. Our work and that of others has demonstrated the essential role that multisensory integration (MSI) plays in typical perception and cognition [15]-[24]. Because inputs from the various senses (e.g., the auditory, visual and somatosensory systems) initially arrive into widely separated regions of the neocortex, MSI must involve ongoing communication between relatively far-flung cortical regions, although it may well be initiated even earlier in the hierarchy within nuclei of the thalamus [25]. In this sense, probing multisensory functioning provides an excellent assay of inter-regional communication, and the fidelity of the multisensory system must at least in part be a function of the integrity of long-range neural connectivity. For this reason we expected measures of MSI to provide a sensitive metric of neural dysfunction in NPC disease. What's more, MSI processes show a prolonged period of neuroplasticity, with continued development of these abilities seen into the late teenage years [22],[26]. As such, measures of MSI may provide useful biomarkers against which to test the impact of treatment on brain function.

A straightforward way to measure multisensory integration is to compare reaction times (RT) to unisensory and multisensory events during a simple speeded response task. It has been firmly established that adults react more quickly to multisensory than unisensory inputs [21],[27]-[30]. For such behavioral facilitation to be unequivocally attributed to multisensory integration, this speeding up must exceed what is predicted due to the mere presence of a redundant signal (i.e. two inputs). That is, when two stimulus copies are presented simultaneously, even if both were to be processed entirely independently in the brain, one would still expect to see a speeding up of responses since there is increased likelihood that either of the two stimuli will yield a fast reaction-time relative to just one input. This is often referred to as the Redundant Signals Effect (RSE), and its presence does not, of itself, necessarily point to integration effects. The so-called “race model” is applied to test for the presence of true multisensory effects, by assessing whether responses to multisensory inputs are faster than the fastest possible responses produced by the unisensory conditions [31]-[33]. This is achieved by comparing the probabilities of making fast responses during multisensory events to those during unisensory events. The race model is said to be violated whenever the cumulative probability (CP) of a response at a given latency for the multisensory condition is greater than the sum of the CPs from each of the unisensory conditions. When the race model is violated, it is taken to be a strong indication that the inputs from the two different senses are interacting (in a non-additive way) to produce the speeding of the responses. Work from our laboratory suggests that this metric of MSI RT-speeding follows a developmental trajectory, with little evidence for behavioral enhancement before age 9, but that near full maturity is reached by age 16 [26],[34]. Moreover, in these developmental studies, behavioral performance was shown to benefit from MSI at the single participant level for 95% of neurotypical participants aged 11-16, and 100% of participants aged 13-16. This relatively protracted developmental trajectory of MSI behavioral facilitation is consistently seen across laboratories [35],[36]. Here we used this behavioral approach to assay multisensory function in three boys with NPC -- two adolescents (14 years, 7 months & 14 years, 5 months old) and one younger boy (11 years, 1 month) -- comparing their performance to that of 16 neurotypical adolescent boys aged 13-15, and 19 neurotypical boys aged 10-13, respectively.

To our knowledge, this is the first study to examine multisensory processes in NPC. The observed lack of race model violation in NPC suggests compromised connectivity between auditory and visual areas of the brain, possibly at both sub-cortical and cortical levels. It is likely that these inter-sensory connections develop very early in life, strengthen across childhood, and stabilize during adolescence [26],[34],[47],[48].Understanding when exactly during the progression of NPC that MSI becomes compromised will require further investigation and will be crucial to maximizing the clinical usefulness of this measure in the NPC population. Two possible scenarios are that; 1) MSI-induced behavioral facilitation never quite reaches “healthy” levels in these individuals or 2) that like many of the other symptoms exhibited in this population, NPC patients experience a degradation of MSI function with progression of the disease state. In either case, this metric of MSI presents a behavioral marker against which to measure improved neurocognitive function due to experimental treatment interventions.

In terms of everyday functioning, an obvious question is what impact deficits in multisensory processing will have on the abilities of NPC children to effectively navigate their environment. For example, effective MSI leads to improved speech perception when a listener has the benefit of watching the facial articulations of a speaker, especially if the fidelity of the auditory input is affected by noisy background environmental conditions [17],[22],[23],[49],[50]. Thus, one implication is that these children may find communication more difficult in challenging multi-speaker scenarios, not uncommon in classrooms or other social settings. MSI is also vital to more basic functions, such as maintaining balance through visuo-vestibular and visual-somatosensory integration [15] and in speeded orienting to reliable multisensory events, whether it be for object identification or cueing initiation of approach/avoidance behaviors [16],[18]-[21],[24]. A more comprehensive understanding of the multisensory integration abilities of these children is clearly called for, and it will be of significant interest to assess the underlying neurophysiology in turn [51],[52].

Another obvious outcome of the current study is that the NPC children show basic motor deficits. While it is true that there are neurotypical participants who are as slow to respond to unisensory inputs, and others who show similarly high variance in RTs, no neurotypical children show the poor response rates we see in the NPC children. Simply put, the NPC children are slow, variable and inaccurate and this triumvirate of issues clearly points to fundamental sensory-motor issues. That said, we do not believe that the MSI deficits observed here are primarily due to these issues, since these issues apply equally to all the experimental conditions (both unisensory and multisensory; also see Additional file 1). As the race model analysis is conducted at the individual participant level, where the cumulative probability distributions are calculated for each participant and within-subject RTs are compared to determine the multisensory benefit, general motor delays are accounted for. It could reasonably be asked, though, whether simple tests of motor speed, variance and accuracy might not prove equally useful biomarkers for NPC. However, it bears re-emphasizing that while the NPC children do show these issues, their performance levels do not fall completely outside the normal distribution for these measures, whereas for the measures of multisensory integration, they clearly do.

It is worth pointing out that these children with NPC are, at some basic level, benefitting from multisensory stimulation, even if not in an integrative manner. The fact that mean RTs and hits are improved in some cases, even in the absence of significant multisensory integration, when patients are exposed to stimulation in two sensory streams is promising, especially in terms of sensory training. This may have implications for the development of assistive technologies used for communication, particularly during the more progressed phases of the disease.

A natural question that arises is whether the multisensory deficit we observe in NPC can be meaningfully impacted through intervention. The landscape is actually quite promising in this regard since several studies now point to multisensory and unisensory gain with repeated training. These studies show that training can lead to improvement in MSI-dependent tasks such as speech-perception [53], that training can narrow the time window during which two sensory inputs are seen as “synchronous” and thus integrated [54], and that MSI networks can be engaged and enhanced in training activities where abstract stimuli are paired, such as specific sounds with abstract shapes, or musical tones with symbols [55],[56]. Work in animal models also supports the notion that sensory integration abilities can be impacted through practice with training-induced multisensory enhancement noted in both behavior and activity patterns at the single cell level in the superior colliculus, in both juvenile [57] and adult cats [58].

An obvious limitation of the current work is the relatively small cohort of three patients with NPC that we were able to test. Ideally, one would like to have greater numbers. However, the disease prevalence rate for NPC is estimated at 1-in-120,000 [6],[8],[59], so recruitment of larger populations is extremely challenging. It is worth emphasizing that the atypical multisensory integration pattern noted here is highly consistent across the 3 NPC patients in our sample and the findings are strengthened by comparison of these 3 patients to large existing datasets of neurotypical age-matched children. In all 3 cases, the performance metrics of the NPC patients fall completely outside the “normative” curve for MSI development.

This study uncovered clear multisensory deficits in three patients with NPC. The simple-to-acquire measures of multisensory response speed described here may prove to be useful endpoints against which to track disease progression and to assess the efficacy of therapeutic interventions. Specific environmental accommodations should be considered to address the potential impact of deteriorating multisensory mechanisms in these children.

GA, SM, SUW and JJF conceived the study. GA and ABB coordinated data collection. JSB and GA conducted the primary data analyses. GA wrote the initial draft of the paper and all authors provided multiple rounds of input during the editorial process. The senior author, JJF, attests that all authors had full access to the data and that each author saw and approved the final submitted version of this manuscript.