Fetal cortical activation to sound at 33 weeks of gestation: A functional MRI study

Abstract

Hearing already functions before birth, but little is known about the neural basis of fetal life experiences. Recent imaging studies have validated the use of functional magnetic resonance imaging (fMRI) in pregnant women at 38-weeks of gestation. The aim of the present study was to examine fetal brain activation to sound, using fMRI at the beginning of the third trimester of pregnancy. 6 pregnant women between 28- and 34-weeks of gestation were scanned using a magnetic strength of 1.5 T, with an auditory stimulus applied to their abdomen. 3 fetuses with a gestational age of 33 weeks, showed significant activation to sound in the left temporal lobe, measured using a new data-driven approach (Independent Component Analysis for fMRI time series). Only 2 of these fetuses showed left temporal activation, when the standard voxel-wise analysis method was used (p = 0.007; p = 0.001). Moreover, motion parameters added as predictors of the General Linear Model confirmed that motion cannot account for the signal variance in the fetal temporal cortex (p = 0.01). Comparison between the statistical maps obtained from MRI scans of the fetuses with those obtained from adults, made it possible to confirm our hypothesis, that there is brain activation in the primary auditory cortex in response to sound. Measurement of the fetal hemodynamic response revealed an average fMRI signal change of + 3.5%. This study shows that it is possible to use fMRI to detect early fetal brain function, but also confirms that sound processing occurs beyond the reflexive sub-cortical level, at the beginning of the third trimester of pregnancy.

Introduction

In the last few decades, an increasing interest in fetal perception has developed, both in the scientific community, and in all the other sectors of society. It is now well-known that functioning of the different sensory systems in the mammalian fetus, varies according to the degree of maturity (Lecanuet et al., 1995). In utero, the somesthetic system develops first, followed by the chemosensory, vestibular, auditory and visual systems. During human fetal brain development, the ‘early preterm phase’, between 24 and 32 weeks of gestation (WG), is also characterized by the rapid growth of thalamo-cortical fibers (Kostović and Jovanov-Milošević, 2006), to form sensory-driven circuitry (Penn and Shatz, 1999). Hearing certainly remains the area of fetal human perception, which has been most extensively described. On the one hand, histological studies provide evidence for functionality of the cochlear neuroreceptors from 28 WG, although they do not show whether the site for processing sounds is cortical or sub-cortical. Moreover myelination of the auditory nerve and the acoustic radiations continue until the age of 1 or 2 years old (Werner, 1998) and a strict causal relationship with function cannot be drawn from this maturational conductance improvement. On the other hand, care should be taken when looking at studies of the psychophysics of fetal hearing, because the most acute sounds (frequencies > 5000 Hz) are attenuated by the successive barriers (tissue and liquid) between the external air and the fetal cochlea. Thus, the mother's sound environment cannot be completely heard by the fetus. Nevertheless, behavioral research on fetal sensoriality has shown fetal reactivity to external sound stimuli, by measuring the fetal heart rate and active fetal movements (Lecanuet et al., 1995). This fetal auditory capacity can be observed from the 28th WG for external sounds with an amplitude of 90 to 110 dB SPL (decibel sound pressure level). The fetus can also habituate to repeated sound stimuli from 32 WG (Morokuma et al., 2004). This phenomenon is different from sensory or motor fatigue, since initial fetal reactivity recovers when the stimulus is modified. Near term, fetuses are also sensitive to complex stimuli, like variations in music (Kisilevsky et al., 2004), transposition of syllables (“ba” and “bi”) or of a pair of syllables (“babi” and “biba”) (Lecanuet et al., 1995). Finally, newborns between 2 and 4 days old only, prefer the voice of their mother, to an unfamiliar voice (Decasper and Fifer, 1980).

These fetal-response measurements are however subject to many endogenous and exogenous influences, which produce considerable bias in the interpretation of the results. All the same, observation of the fetal heart rate and motor reactions implies that the sounds have been received and processed at least as far as the sub-cortical level. However, the exact level in the central structures, which is involved in these processes of sensory integration, is subject to debate. For some authors, limited processing at the midbrain level is sufficient for orientation or preference behavior to occur, in response to external stimuli (Joseph, 2000). Moreover, this type of processing seems to function in anencephalic human neonates (Tuber et al., 1980). Further investigation is therefore necessary to identify precisely the specific neural location of these fetal responses. Functional imaging studies, measuring auditory evoked potentials (AEPs) in premature neonates at 33 WG (Morlet et al., 1995), have shown early cortical activity with almost mature biomechanical function of the cochlear signal. The lack of electrical activity in the olivo-cochlear system, in premature neonates before 32 WG, actually indicates that before that stage, the immature auditory pathways cannot relay the information from the periphery to the cortex. The different components of late AEPs, are only detectable from 33 WG (Pasman et al., 1991). Taken together, these findings suggest that there is a change in processing of complex sounds at around 32–33 WG. In vivo studies, evaluating fetal audition during the third trimester of gestation, have also been performed. A fetal cerebral auditory evoked response with a mean latency of 200 ms, has been shown in around 50% of near-term fetuses tested, using the magnetoencephalography method (Preissl et al., 2004). More recently, a MEG-research conducted in 18 fetuses every two weeks beginning from 27 WG until term has shown a continuous decrease in latency with maturational age (Holst et al., 2005). However, this non-invasive method provides information with good temporal resolution in fetuses, but without good source-localizing information. Functional Magnetic Resonance Imaging (fMRI) studies, using a 0.5 T. scanner, showed unilateral temporal activation, in response to stimulation by pure tones (p < 0.05), in around 45% of fetuses tested between 37 and 41 WG (Moore et al., 2001, Hykin et al., 1999). However, the localization capacity of these fMRI studies was limited, since there was no coregistration of the functional and anatomical images, and there were numerous motion artifacts.

Although fetal behavioral responses can be measured in response to sounds at the beginning of the third trimester, currently-available imaging data do not allow simultaneous fetal cortical activity to be identified at this stage of gestation. Previous fetal fMRI studies were performed near to term (Moore et al., 2001, Hykin et al., 1999), and referred to the standard voxel-wise hypothesis-driven method for analysis of the fMRI data-sets (HDA), which means data processing with reference to a hypothetical model for presentation of the stimulus (Kiebel and Holmes, 2003). Although this has been proved to be an effective approach, there are nevertheless a number of limitations when using it in fetal imaging. Firstly, since HDA is a univariate analysis method, the problem of multiple comparisons of the data should be taken into account, by correcting the statistical thresholds, in order to avoid a large number of false positives (Genovese et al., 2002). These corrections may cause the activation cluster-size to be reduced, or the smallest activations to be deleted. Secondly, HDA is sensitive to motion artifacts, which constitute a major problem in fetal imaging (Gowland and Fulford, 2004). Finally, HDA is particularly sensitive to unpredictable events, for example a decrease in the performance of the participants in the study during testing (Quigley et al., 2002), which are common during fetal life, due to rapid waking–sleeping cycles. For all of these reasons, we propose the use of recent data-driven multivariate analysis methods for fMRI time series, such as Independent Component Analysis (ICA — McKeown et al., 2003). ICA allows the multiple signal sources in the raw data to be separated out blindly, and to take into account data, for which the temporal structure is difficult to model a-priori. This analysis method seems to be particularly adapted for studying fetal cerebral activity. The main objective of this study was, in fact, to demonstrate, using fMRI, early cortical processing of sound heard by a fetus at the beginning of the third trimester of gestation. Risk studies performed on animal and human molecules and cells, did not find any cytotoxic or mutagenic effect of clinical or superclinical doses of magnetic exposure (e.g. Myers et al., 1998). Moreover, recent studies in human fetuses have demonstrated that the fetal heart rate shows no significant changes (Michel et al., 2003) before, during or after the examination, and that there is no auditory deficit at 3 years old, as a result of exposure to the noise of an MRI scanner during the fetal period (Baker et al., 1994, Glover et al., 1995). The use of a static magnetic field of 1.5 T., together with new data-driven method for fMRI analysis, will therefore make it possible to localize fetal neural activity at the beginning of the 3rd trimester of pregnancy more accurately than previously. Finally, interpretation of the results will be made easier by anatomic and functional coregistration, and comparison of the results obtained in the fetuses with those from an adult control group, under the same experimental conditions. This study is to our knowledge, the first fetal brain fMRI study to be performed at 33 WG.