Main

Olfaction is essential for neonatal behavioral adaptation in many mammals, including humans (1). Olfactory signals help the newborn baby localize and attach to the nipple at the first sucking bout (2). Birth and the first hours of life are crucial for olfactory learning. During this period the smell of the mother and that of the newborn interact with each other, and animal studies have shown that changes in the processing of the olfactory signals occur on both sides (3, 4). The cortical brain structures involved in these processes are incompletely known, and seem to have a broad distribution (5). In adults, olfactory stimuli are processed in the lateral and anterior orbito-frontal gyri of the frontal lobe (5–7), but differences might be related to the type of the odorant, and whether this is pleasant or unpleasant (8, 9).

Studies on primates have shown that the olfactory bulbs send direct information to the primary olfactory cortex, i.e. the piriform and entorhinal cortices of the temporal lobes, and also to the thalamus, hypothalamus, and amygdala. After relaying in these three regions, signals are sent to the olfactory associative cortex in the orbito-frontal region. Several other regions of the brain, which influence mood and behavior, are involved in smell perception (5–7). In general, pleasant odors elicit approach behavior, unpleasant odors elicit avoidance behavior (10, 11).

Near-infrared spectroscopy (NIRS) is a bedside, noninvasive methodology, able to detect concentration changes of natural chromophores like oxygenated Hb [Hb O2] and deoxygenated Hb [Hb H] (12). In recent years, NIRS has been used to study functional activation of various areas of the brain. This is based on the assumption that an increase in recorded [Hb O2] concentration represents an increase in blood flow, which in turn reflects neuronal activation. Most studies of cortical function with NIRS have been performed on adult subjects (13, 14). In newborn infants, activity of the visual cortex was studied after repeated light stimulation using NIRS (15).

The aim of the present study was to use NIRS to monitor the activity of the olfactory cortex as mirrored by the hemodynamic response when newborns were exposed to a biologically meaningful stimulus such as the smell of colostrum. Vanilla was used as a positive control and distilled water as a negative one.

METHODS

Study group.

Twenty-three healthy, full-term newborn infants were included in the study. They were all delivered vaginally by healthy, nonsmoking mothers after an uneventful pregnancy (mean duration 39.4 wk, minimum 37, maximum 41, SD 1.3). Babies whose mothers had used scented soaps, perfumes, or deodorants during the 3 d before the experiment were not eligible.

The babies were studied at a postnatal age varying between 6 h and 192 h (median 48.0 h, SEM 10.1 h), when lying in the bed in the supine position and being in a quiet, awake state (16). All babies were exclusively breast-fed and at least 30 min had elapsed since the last feeding. In the original study group of 30 babies, 7 fell asleep during the test and were excluded, leaving 23 babies with a male/female ratio of 12/11 in the study group. Parents were present during the test, but asked not to speak to or to touch the baby during the experiment. The room temperature was kept around 22–24°C, the light was dim, and the noise level was reduced as much as possible. The local Ethics Committee approved the study and parental informed consent was obtained.

Recruiting procedure.

Every weekday morning the files of new delivered babies were scrutinized and the parents of those newborns who fulfilled the admission criteria were given an informative letter about smell perception in the newborn and NIRS. Thereafter, two of the authors (M.B. and L.L.B.) met the parents and explained the aim of the research and answered questions. Approximately 50% of the parents agreed to participate in the study. The most common reason for refusing was related to the use of the near-infrared instrumentation.

Odor preparation.

As odorant sources we used (i) the own mother's colostrum, manually expressed into a plastic cup just before the test; (ii) vanilla essence (4-hydroxy-3-methoxybenzaldehyde) dissolved in an oily solution, commercially available (Body Shop International PLC, West Sussex, England); and (iii) distilled water as a negative control. The substance to be tested was soaked into a cotton bud sized about 1.5 × 0.5 cm attached to the end of a 20-cm long stick. The cotton buds were prepared in a room separated from the testing room. The vanilla smell was judged as very strong and the colostrum smell as very weak by the adult nose. As in many other studies on olfaction, vanilla was chosen as a positive control because of its properties to activate mainly the primary olfactory system, with very little effect on the trigeminal system (17, 18).

NIRS settings.

A NIRO 300 (Hamamatsu Photonics, Hamamatsu, Japan) device was used. Changes in the concentration of oxyhemoglobin [Hb O2] and deoxyhemoglobin [Hb H] were monitored, starting from an arbitrary zero point. The changes in the total hemoglobin ([Hb tot] = [Hb O2] + [Hb H]), reflecting changes in cerebral blood volume (CBV) (19) were also calculated. The basic principles of NIRS and its reliability in studying newborn infants have been described in detail (12, 20, 21).

Briefly, NIRO 300 uses four pulsed laser diodes, which produce light at wavelengths of 775, 810, 850, 910 nm. The pulse frequency of each diode is approximately 2 kHz, each pulse having a duration of 100 ns. NIRO 300 is provided with an emission probe [8 mm (Ø) × 4 mm (H)], and a detection probe [20 mm (Ø) × 8 mm (H)], called optodes. The average output power from the emission optode was about 1 mW. The optodes were placed in a special dark, semirigid, rubber holder, which kept an interoptode distance of 4 cm constant during the recording. They were positioned over left orbito-frontal gyrus of the frontal lobe. The emitting optode was placed 2 cm above the midpoint of the line connecting the external angle of the left eye to the homolateral tragus. We established a differential path-length factor of 4.2 (22). Sampling rate was performed every 0.5 s.

Smell stimulus.

After a 90-s starting period, followed by a 30-s period of NIRS baseline definition, each infant received the stimuli in the following order: (1) control, (2) colostrum, and (3) vanilla (Fig. 1). The smell stimuli were administered by moving the cotton bud slowly from one nostril to the other at a distance of approximately 1–2 cm. Particular attention was paid to preventing the tip of the cotton bud from touching the infant's nose. Each exposure lasted 30 s. A trigger signal was marked on the NIRO 300 at the beginning and end of the stimulus. There was a 2-min interval between each exposure. Before administering a new stimulus, NIRS baseline was redefined.

Figure 1
figure 1

Time schedule of the experimental procedures. Fastening of the optodes was followed by 90 s of adaptation. Each test epoch consisted of 30 s of baseline definition followed by 30 s of smell exposure. The order of exposure was the same in all experiments.

Full size image

Data storing.

All NIRS data were stored offline through an RS232 board on a personal computer. The start of each 30-s smell exposure period was marked. Changes in [Hb O2] and [Hb H] were analyzed, and [Hb tot] values were calculated. Each recording lasted at least 9 min and 30 s as shown in Figure 1. Sampling rate was established at 2 Hz, and for each 30-s period of smell exposure a 60-value epoch was created for further analysis.

Data analysis and statistics.

Three 60-value epochs, one for each smell exposure, were used for statistical analysis. Two different approaches to the values were used. First, the mean values of each subject's recording during the three different smell conditions (control, colostrum, and vanilla) were compared. Second, for all three stimuli an averaged value for each single measurement out of the 60 total was calculated for all 23 babies. In such a way it was possible to represent the average changes in [Hb O2] for each smell exposure. To compare these curves, and to calculate the latency interval from the onset of the stimulus to recorded changes of [Hb O2] during which the values were likely similar, a time series stepwise analysis was carried out. For each sampling point (every 0.5 s) after the onset of odor exposure, the mean values obtained from the 23 babies were compared with the cutoff point at which the three averaged curves became significantly different.

ANOVA (two-way ANOVA for repeated measurements) and, subsequently, Student–Newman-Keuls posttesting, were used to compare the changes of [Hb O2] in response to the three different stimuli. The Newman-Keuls test used for post hoc comparisons sorts the means into ascending order. For each pair of means, the program then assesses the probability under the null hypothesis (no differences between means in the population) of obtaining differences between means of this (or greater) magnitude, given the respective number of samples. Thus, it actually tests the significance of ranges, given the respective number of samples. Also, an ANOVA for repeated measurement has been performed for each second's mean values of the averaged curve.

To represent better the averaged curves an “aggregated index” was used. This specified the number of consecutive observations from which the mean was calculated. Spearman rank correlation was used to examine whether postnatal age or gestational age were related to the magnitude of changes of [Hb O2]. The Mann-Whitney U test was applied to evaluate whether there were any sex differences in the magnitude of the changes of [Hb O2]. Statistical analysis calculations were completed by means of the program Statistica® (version 1998, StatSoft, Inc., Tulsa, OK, U.S.A.).

RESULTS

During exposure to vanilla as the positive control, each infant showed a marked increase in [Hb O2] over the left orbito-frontal region, slowly tapering after the end of the exposure. In contrast, during exposure to distilled water as the negative control, only small fluctuations around the baseline were recorded. The response to the smell of colostrum was variable: in some babies there was an increase in [Hb O2] similar to that elicited by vanilla, in others only minor or no changes were recorded. Positive responses were seen to colostrum and to vanilla but not to water (Fig. 2). As also seen in this figure, the response to the smell stimuli far outlasted the period of exposure. Such postexposure [Hb O2] changes could last from 30 to 40 s (for colostrum) and 60 to 70 s (for vanilla) before returning to the baseline. Systematic studies could not be undertaken because in most infants registration artifacts caused by limb movements interfered with the recordings. During exposure to colostrum and to vanilla but not to water we could observe sniffing movements of the alae nasi.

Figure 2
figure 2

Changes in [Hb O2] during exposure to colostrum, vanilla, and water in one baby 7 h postpartum. Similar changes were observed in all babies during exposure to vanilla, but only in babies 24 h old or younger during exposure to colostrum. A line for each smell test indicate the 30-s exposure onset and offset.

Full size image

For each type of stimulus we averaged all [Hb O2] values collected every 0.5 s, i.e. 60 registrations during each exposure (Fig. 3). The comparison of the changes during exposure to water, colostrum, and vanilla showed a significant difference (ANOVA for repeated measurements, p < 0.00001). The post hoc comparison showed also a difference between groups (control versus colostrum, p = 0.0004; control versus vanilla, p = 0.0001; colostrum versus vanilla, p = 0.0001).

Figure 3
figure 3

Averaged curves (means ± SEM) of [Hb O2] changes during smell exposures. The means were derived from 23 values at each 0.5-s recording. ANOVA for repeated measurements showed a significant difference between the three responses (p < 0.0001). Newman-Keuls post hoc comparison evidenced statistically significant differences as follows: control vs colostrum (p = 0.0004), control vs vanilla (p = 0.0001), colostrum vs vanilla (p = 0.0001). Statistical analysis of [Hb O2] differences after exposure offset was impossible due to the frequent occurrence of movement artifacts. The panels at the right show magnifications of the first 10 s of recording to visualize onset of [Hb O2] rise. The [Hb O2] values became statistically different from the baseline at 6–8 s for colostrum and at 5 s for vanilla (Newman-Keuls post hoc comparison, *p = 0.002 and **p = 0.03, respectively).

Full size image

The averaged curve for [Hb O2] changes during colostrum exposure showed an increase from the baseline starting approximately 6–8 s after onset (p = 0.002). The difference was significant at 6 and 8 s, but not at 7 s. After vanilla exposure a significant difference from the baseline was seen after 5 s (p = 0.03) (Fig. 3).

A preliminary look at the individual [Hb O2] curves obtained during exposure to colostrum smell gave the impression that an increase in [Hb O2] was seen mainly in the youngest individuals. This was confirmed by demonstration of a statistically significant negative correlation between changes in [Hb O2] and postnatal age (r = −0.64, p = 0.001 with 95% confidence interval) (Fig. 4). Those babies showing the greatest increase in [Hb O2] were between 6 and 24 h old at testing. During exposure to vanilla smell the [Hb O2] changes do not correlate with postnatal age (r = 0.40, p = 0.9).

Figure 4
figure 4

Changes in [Hb O2]vs postnatal age during colostrum exposure. Changes in [Hb O2] were inversely correlated to postnatal age (r = −0.64, p = 0.001 with 95% confidence interval). An increase in [Hb O2] was seen only in infants being 24 h or less at exposure. For vanilla there were no similar age-related differences.

Full size image

In the nine infants who were 24 h or younger at testing, the [Hb O2] response to colostrum, vanilla, and water were compared using the same type of calculation as described above when examining the whole material. There was a significant difference between all three test conditions (ANOVA for repeated measurements, p = 0.0001). At the post hoc comparison with Newman-Keuls test, the changes between each smell test were statistically different (control versus colostrum, p = 0.009; control versus vanilla, p = 0.0002; colostrum versus vanilla, p = 0.01). In the 14 babies older than 24 h there was no significant difference between the changes in [Hb O2] during control and colostrum exposure (Newman-Keuls post hoc comparison, p = 0.8).

For all cases and exposures examined, changes in [Hb tot] were essentially the same as for [Hb O2] (data not shown). No sex differences in the changes of [Hb O2] were observed during the test (Mann-Whitney U test, p = 0.6).

DISCUSSION

The main finding of this study was that the NIRS technique can be used in the neonatal period to record activity in the orbito-frontal cortex—as mirrored by changes in blood circulation—during exposure to biologically meaningful as well as artificial odors, colostrum, and vanilla, respectively. The NIRS technique has several advantages: it is performed at bedside, noninvasive, not painful, and confers no exposure to harmful illumination. One of the limitations is that this technique does not give an image of the brain activation that is as anatomically detailed as that obtained with functional magnetic resonance imaging. Studies with functional magnetic resonance imaging in adults have shown that olfactory stimulation activates a number of different regions of the brain (5, 6, 10). On the basis of these studies, we chose to place the emitting and receiving optodes over the left orbito-frontal region (belonging to the secondary olfactory cortex). In so doing we did not explore directly other regions possibly activated, such as the entorhinal cortex or the temporal lobe.

The changes in blood flow caused by vanilla smell (perceived by adults as very strong) were greater than those caused by colostrum smell (perceived by adults as very faint or nonsmelling). It is tempting to explain this in terms of quantitative differences in emitted molecules and hence in the number of odor receptors engaged or in intensity of their activation. However, we cannot exclude that a biologically meaningful odor such as milk, which together with breast odor, is a key factor in the newborn's ability to locate and grasp the nipple, also has cortical targets different from vanilla that were not in the region of near-infrared illumination. Moreover, a different number of receptors involved in perceiving the smell, likely related to the molecule concentration of the odorant, might elicit different cortical responses.

We also noted a 6–8 s delay for colostrum and a 5-s delay for vanilla, before [Hb O2] rose in the orbito-frontal region, suggesting a corresponding delay until the stimulus activated this part of the cortex. This differs from visual stimuli, which elicit a nearly immediate response in [Hb O2] (15).

It may be argued that the observed changes in [Hb O2] mirrors a general arousal response rather than a specific activation of the secondary olfactory cortex. Although this interpretation cannot be excluded—the fact that breast milk and breast odor as well as other familiar odors elicit a target-oriented approach in the newborn (1)—a specific effect seems more probable. A biologically interesting finding was that the olfactory response to colostrum, but not to vanilla or water, was inversely correlated to postnatal age. This was unexpected inasmuch as the smell of milk, as well as that of the mother's breast, elicits an approach behavior in the newborn well beyond the first postpartum day (23). We can only speculate about the reasons. The older the baby, the more experience it has of taste and smell of colostrum, and the response might have been successively modified in strength or localization by coupling to recognized food. A second possibility is that a putative odorous compound is more concentrated in the first few drops of colostrum than a few days later, when milk secretion has increased, as is seen for other compounds such as antibodies (24). Still, another possibility is that the very early colostrum smells different from the late colostrum. Thus, Singh (25) found in rats that circulating oxytocin triggered the release of some unidentified substance that was attractive for the pups. If a corresponding mechanism exists in humans, the surge in circulating oxytocin during delivery might have influenced the smell of early colostrum.

Catecholamines seem to play an important role in olfactory learning (26) and a recent study by Kawai et al. (27) suggests that adrenergic pathways might influence signaling by the olfactory epithelium. This could be possible via a double action on the olfactory receptor neuron: an increase in the threshold of the spike train of action potentials and an augmented firing rate in response to the stimuli. Thus, a high concentration of adrenaline, which modulates signal encoding of the olfactory epithelial adrenergic neurons, might enhance the high olfactory response of newborns around birth. The hypothesis that such a mechanism influenced the different responses among the youngest and the somewhat older infants does not fit adequately to our studied population. First, the surge in catecholamines after vaginal delivery decreases markedly within approximately 1 h after birth, stabilizing at a blood concentration, which was probably in the same range in all the babies we have studied. Second, such a mechanism should have influenced both colostrum and vanilla responses.

In conclusion, this study shows that brain cortical activation after odor stimulation in the newborn baby can be recorded by means of a bedside, noninvasive method such as NIRS. The olfactory testing paradigm might be used as a diagnostic tool to monitor brain cortical competence.