Our main goal was to examine the neurophysiological characteristics of auditory processing in girls with Rett syndrome as well as their modulation by the rate of stimulus presentation. We showed that early stages of auditory processing are relatively preserved at least in the subgroup of RS patients, while the later stages, reflected in the P2 and N2 components of ERP, are largely affected being both delayed and attenuated. At the same time, N1 and P2 components of ERP, demonstrated preserve modulation by the stimulation rate in Rett syndrome showing enlargement with the prolongation of stimulus onset asynchrony. Developmental stagnation of some neurophysiological characteristics was also observed in RS. Below we consider these findings in detail.

Preserved modulation of ERP components by the rate of presentation in RS

In line with the previous studies [7,8,9] on TD, we also showed that ERP components (N1P1 and P2N1) become more pronounced with increasing stimulus onset asynchronies. The novel result is that ERPs enlargement was mostly between the SOA 900 ms and SOA 1800 ms conditions, while further prolongation in SOA up to 3600 ms did not influence ERPs. This is a rather surprising result as previous studies reported that ERP continues to increase for the SOA of at least up to 12 s, however, they were all conducted in adults [11, 12]. For the children population the studies of the effects of SOA on ERP are quite limited and none of the studies included in their design the SOA longer than 2400 ms [36, 41]. The absence of further enlargement in the ERP components after 1800 ms SOA may be related to the fact that in children the response recovery period is faster, and, as a result, sensory memory is shorter. In line with this result, it has been shown that in children, previous experience has less influence on perception [42]. However, this result needs further investigation.

An important finding is that both N1P1 and P2N1 modulation by SOA, that are largely independent, is preserved in the evident-ERP subgroup of patients with RS, even in spite of the largely altered (delayed and attenuated) P2N1 component. This result points to the typical functional meaning of these components and preservation of basic learning ability, reflected in neurophysiological adaptation in RS. As ERP components typically enlarge with the SOA prolongation, we can differentiate major ERP components in RS much better with longer SOA, suggesting that auditory information processing is less affected when tones are presented at a slow rate (e.g. with SOA 1800 and 3600 ms). However, clear between group differences is evident even with these longer SOAs.

The effect of presentation rate on auditory processing in Rett syndrome has been demonstrated previously in the oddball paradigm by means of mismatch negativity (MMN)—neurophysiological correlate of change detection assessed from ERP difference wave (standard, frequent ERP minus rare deviant ERP). MMN in response to frequency deviation has been registered in girls with RS only with a short (450 ms) interval between stimuli, but not with a longer stimulus onset asynchrony (900 ms and 1800 ms) [29]. This might indicate that the stimulus representation in Rett syndrome persists for a shorter period of time and vanishes with increasing SOA. In the present study, we have found that as SOA increases from 900 to 1800 ms, N1P2 and P2N1 amplitudes in RS become larger, indicating typical release from adaptation, which at first glance is inconsistent with these MMN data. However, the neuronal mechanisms that underlie MMN and N1/P2 components most likely are different with the MMN linked to predictive error processing and the N1/P2 associated with stimulus-specific adaptation (SSA) [43,44,45]. Thus, the results obtained in the present and Brima's work may indicate changes in different subprocesses of auditory processing and features of different neuronal populations’ activation in RS.

Thus, choice of the most appropriate SOA for future auditory ERP studies should be driven mainly by the process of interest, with longer SOA more appropriate for tracing change detection deficits and shorter SOA being more sensitive for alternation of basic auditory components. While speculative, such neurophysiological abnormalities together with clearly prolonged timing of information processing result in a very limited time window within which the RS brain processes information more or less adequately. Clinical field might potentially benefit from further work in collaboration with practitioners aiming to develop the training/information presentation protocols for RS that allow: (1) information that needs to be linked together being demonstrated within a short time frame and at the same time (2) provides enough time for the information processing that is clearly prolonged in RS.

We believe that other SOA effects, revealed in our study, do not represent the meaningful insight into the core deficits in RS, as they can be explained by contamination of ERP components, as enlargement in one component can lead to latency shift and reduction in the neighboring components of opposite polarity. In particular, a heightening of the N1 amplitude with SOA prolongation could lead to (1) delay in the next components (P2), which appeared as an increase in its latency with increasing SOA as well as (2) shortening of the previous component latency, which appears as P1 latency decrease with SOA prolongation. Similarly, an enlargement in the amplitude of P2 leads to a prolongation in the latency of the following N2 component.

Delayed and reduced ERP components in RS

In our study significant reduction in the amplitude was shown for P2 (P2N1) and N2 (N2P2) components in RS. Also, the P2 and N1 components were significantly delayed in the RS group compared to TD.

Previous studies describing auditory evoked potentials in patients with Rett syndrome often highlight their attenuation and delay in comparison to typically developing children [16, 18, 29]. In particular, Sysoeva's study demonstrated a reduction in the amplitude of the P2 and N2 components of auditory ERP in response to simple (tones) and more complex (phonemes) types of stimuli [15], with the former being confirmed in our current study. An atypical reduction of P2 measured as P2N1 amplitude was also reported in a multisite study of Saby and colleagues, however, unlike our results their RS patients showed also a reduction in N1 amplitude (N1P1 peak-to-peak) [16]. This discrepancy may be due to the slightly different experimental conditions (e.g. varied SOA from 0.6 to 2.0 s) or the wider age range of their patients (2–37 years old), since SOA variability as well as age was shown to increase N1 amplitude [35].

At the cognitive level these effects can be linked to a disturbance of information processing in the late stages. The P2 component is associated with the consolidation in the auditory memory [21]. Meanwhile, N2 is associated with the inhibition of irrelevant information [26] and categorization [22,23,24,25]. Thus, a reduction in both of these components in Rett syndrome may be related to the alternation in these abilities.

However, the deficits start at the level of the N1 component, which is significantly delayed in Rett syndrome. Such latency shift suggests the increased processing time needed for auditory stimuli identification. For the first time, changes in the auditory ERPs’ latencies in Rett syndrome have been described by Bader in 1989 who reported some patients with RS demonstrating the delay of Pa, N1 and P2 components at the individual level [46]. Delay in P2 component has been also confirmed in subsequent studies [15], including our current work. The MMN, that coincides in latency with the N1 component (circa 120 ms), has been also shown to be delayed in Rett syndrome as compared to the control group [18]. While ERP’s latencies shift in RS in our and previous studies is only about 20–40 ms, information processing in the auditory system is extremely fast, making even small temporal delays critical, especially for complex stimuli processing such as speech.

Noteworthy, the ERP abnormalities similar to what we found in patients with RS is also observed in RS animal models: Mecp2-deficient mice and rats demonstrated delayed and reduced analogous auditory cortex response [47,48,49]. Moreover, ERP component delay is evident in RS not only in auditory modality, but also observed in response to visual stimuli [50,51,52]. So this pattern is quite consistent for Rett syndrome and related Mecp2 damages.

Developmental course of ERP components and their relationship with Rett syndrome severity

We have found atypical age dynamics (stagnation) for several neurophysiological characteristics in patients with RS. TD group in our study characterized by developmental decrease in the amplitude and latency of P1, an increase in the amplitude of N1P1 and a shortening of the latency of N1, as well as an increase in the amplitude of P2N1, which corresponds to the previous literature [7, 33,34,35,36]. Some age-effects were more pronounced for particular SOA conditions, generally corresponding to longer SOAs, where components are larger. Modulation by presentation rate of N1P1 and P2N1 amplitudes also show increase with age, corresponding with previous results [35]. Among these developmental changes only N1P1 amplitude followed typical developmental increase in RS. All other components showed atypical age dynamics, e.g. full developmental stagnation irrespective to SOAs for P2N1, or absence of age-related decrease in P1 latency specific only to 900-ms SOA condition with preserved P1 latency decreased for longer SOAs. Thus, this atypical age dynamics of neurophysiological components corresponds well with the view on the Rett syndrome as a disorder of stagnated development also seen at the behavioral level.

The absence of a significant association with age in Rett syndrome, in contrast to the TD group, has been described previously by Sysoeva for 900-ms SOA condition. This work demonstrated the age-dependent increase in N1 and decrease in P1 and N2 components in the TD group, but not in the RS group, but this result was not adequately powered to detect between-group differences in the developmental trajectory [15]. Differences in age dynamics between the groups have been described by Saby et al., but in this work Rett syndrome group demonstrated a decline in N1P1 peak to peak amplitude with age which was not observed in the TD group [16]. It was suggested that this negative dynamic could represent a progressive aspect of the disease process. Despite the fact that we found the typical increase of these components (P1, N1) peak-to-peak values in RS group in our sample, which could be caused by differences in the experimental procedure and a narrow age range, our results also indicate altered auditory ERP development, especially for P2N1, which was reduced in the RS group.

Multiple papers [15, 16, 52] demonstrated correlation of ERP amplitudes or latency to Rett syndrome severity. In Saby’s work Clinical Severity Score (CSS [53, 54]) and Motor-Behavioral Assessment (MBA [54]) were used [16]. Both of these scales showed significant correlation with N1 amplitude and N1P1 peak-to-peak amplitude. Sysoeva et al. used the same RSSS as in our study and observed significant correlation with P2 amplitude [15]. In our study, no significant correlations between ERP components and RSSS were found. The absence of significant correlations may be due to the coarseness of the chosen scale or to some difficulties in adapting it for the Russian populations.

Approaches to no-evident ERP group

It is a continuous struggle for researchers, especially those who work with a challenging population, to include as much data as possible preferentially without compromising on signal to noise ratio (SNR). However, quite often a lot of data needed to be excluded from the final sample due to various, sometimes subjective, reasons. For example, if you exclude participants without evident ERP components—that seems reasonable as it might represent some problems with EEG recording—you might exclude the patients that indeed have very small and not very pronounced ERP and this absence of evident ERP might be important characteristics of the population considered. Our sample contained about half of the RS group without evident ERPs, and we selected for the main analysis only those who have evident ERPs with measurable ERP components. Abnormalities that are revealed in the evident-ERP group cannot be attributed to the poor SNR or other technical problems, and considered in our study as the genuine neurophysiological atypicalities that characterize RS. Moreover, in this group we can examine not only the amplitude but also latency of the ERP components providing a more in depth view on the origin of the observed changes.

The causes of the fact that only half of the participants with Rett syndrome had evident components are currently unclear, but the phenomenon itself corresponds with previous studies in RS [16, 18, 29, 52]. This specific pattern was not due to age and the severity of the symptoms of the disease. One of the possible causes of this could be abnormal background EEG, which has always been considered one of the features of this syndrome [55, 56]. The epileptiform activity in Rett syndrome is expressed by the appearance of spikes and sharp waves in the central temporal regions and a slowing of the theta-band background EEG in the frontal-central regions [17, 57,58,59]. These features are unrelated to seizures and occur even in patients without a history of epilepsy [57, 58, 60, 61]. Thus, against the background of an epileptiform activity or just increased low-frequency oscillations, the detection of evoked potentials could be problematic. Whether this “ERP absence” is a true characteristic of some RS patients or a result of technical problems in detection of ERPs in the presence of atypical background EEG activity is an important direction for future research.