Sucking versus swallowing coordination, integration, and performance in preterm and term infants

Abstract

Mammalian infants must be able to integrate the acquisition, transport, and swallowing of food to effectively feed. Understanding how these processes are coordinated is critical, as they have differences in neural control and sensitivity to perturbation. Despite this, most studies of infant feeding focus on isolated processes, resulting in a limited understanding of the role of sensorimotor integration in the different processes involved in infant feeding. This is especially problematic in the context of preterm infants, as they are considered to have pathophysiological brain development and often experience feeding difficulties. Here, we use an animal model to study how the different properties of food acquisition, transport, and swallowing differ between term and preterm infants longitudinally through infancy to understand which processes are sensitive to variation in the bolus being swallowed. We found that term infants are better able to acquire milk than preterm infants, and that properties of acquisition are strongly correlated with the size of the bolus being swallowed. In contrast, behaviors occurring during the pharyngeal swallow, such as hyoid and soft palate movements, show little to no correlation with bolus size. These results highlight the pathophysiological nature of the preterm brain and also demonstrate that behaviors occurring during oral transport are much more likely to respond to sensory intervention than those occurring during the “pharyngeal phase.”

NEW & NOTEWORTHY Physiological maturation of infant feeding is clinically and developmentally significant, but seldom examined as an integrated function. Using longitudinal high-speed videofluoroscopic data, we found that properties of sucking, such as the length of the suck, are more sensitive to swallow physiology than those associated with the pharyngeal swallow itself, such as hyoid excursion. Prematurity impacted the function and maturation of the feeding system, resulting in a physiology that fundamentally differs from term infants by weaning.

INTRODUCTION

Infant feeding in mammals is complex and requires the successful integration of suction generation (milk acquisition), intraoral transport, and swallowing (30, 31). These behaviors further involve the coordination of multiple oropharyngeal structures, cranial nerves, and over 25 different paired muscles (40, 51). Despite the high level of integration necessary for successful feeding, most research focuses on the dynamics of isolated processes and structures, yet still makes conclusions about the overall system. For example, the biomechanics of milk extraction and tongue function to generate suction have been extensively studied (9, 15), but the downstream effects of differences in milk acquisition are not often measured. Similarly, an extensive body of work on the neural control of swallowing exists (51, 63, 65), and clinical research has a series of measures for documenting performance from the perspective of aspiration (33, 60). However, swallowing research has focused primarily on the swallow itself, despite the well-known link between acquisition, transport, and swallowing. This focus on the individual processes of feeding fails to address how these processes, which occur together in infant feeding, affect one another in both normal and pathological situations.

Understanding the connections between food acquisition, transport, and the pharyngeal swallow is especially critical given their differences in neural control, and their sensitivity to perturbation (35, 51). Furthermore, the sensory information that is gleaned during the oral stages of feeding before the onset of the pharyngeal swallow is sensitive to sensorimotor feedback (19, 31, 62) and plays a critical role in regulating the excursion and timing of the hyolaryngeal structures involved in swallowing (18, 36, 63). In contrast, the pharyngeal swallow is considered to be reflexive and the behaviors following its initiation show little sensitivity to sensory feedback (34, 36). These differences in neural control are further demonstrated by the fact that in decerebrate animal models, a pharyngeal swallow can be elicited that is no different from the swallow occurring in intact animals, but suckling behaviors do not occur with any regularity (29, 66, 68). The movements of the structures associated with the pharyngeal swallow are thus governed by sensory inputs gleaned during the oral phase of feeding (18, 36, 63). Due to this, most successful interventions for managing swallowing problems in human adults focus on modifying sensorimotor patterns before the initiation of the pharyngeal swallow, such as by thickening liquids or providing foods that are sour (20, 32, 44). A further limitation of existing research is that our knowledge of both sensorimotor integration as well as the links between oral processing and pharyngeal function is largely from experiments in adults. Whether these links and integration differ in infants and change as a function of neurological development is largely unknown.

Preterm infants are especially prone to experience feeding difficulties and are considered to be a neurologically compromised population (2, 7, 17, 38, 54, 56). They are thought to have decreased abilities to generate suction and acquire food (41, 42), and in some cases are not able to successfully produce rhythmic sucking patterns and must instead be fed through gavage feeding, resulting in reduced feeding experience (38, 39). Similarly, preterm infants often exhibit challenges associated with swallowing, as they have underdeveloped pharyngeal kinetic rhythmicity (57), less efficient upper esophageal sphincter kinetics (59), and decreased coordination between movements of the hyoid and thyroid, which are thought to be important in airway protection (46). However, most work studying infant feeding has focused on individual behaviors in isolation, and how the developmental deficits in these distinct processes interact in preterm infants remains poorly understood. As a result, we lack a comprehensive integrated characterization of the mechanisms of swallowing dysfunction in preterm infants. Examining infant feeding from an integrated perspective is especially critical, because evidence from animal models suggests that many of the issues that preterm infants face in feeding physiology are grounded in coordination among multiple behaviors, rather than poor performance in any one behavior (10, 13, 46, 47). Furthermore, without intervention, poor feeding performance in preterm infants most likely remains throughout infancy (39, 46–49). Only by understanding what parts of feeding physiology in infants are sensitive to sensorimotor inputs, and how those can be linked to feeding performance from an integrated perspective, can we develop targeted interventions toward improving feeding outcomes in infants with feeding difficulties, such as those born prematurely.

Here, we use a validated animal model, pigs (24), to examine the relationships between movements of the oropharyngeal structures responsible for both milk acquisition and for swallowing in term and preterm infants. We analyzed oral measures of milk acquisition and pharyngeal measures of swallow function to test how they are related to performance as a function of age and prematurity. To study how milk acquisition occurs, we focused on properties of the interaction of the tongue with the palate to create suction by measuring suction amplitude and length. The analyses of swallow physiology discussed in this paper include the excursion of the soft palate and the hyoid, which are active in both propelling a bolus from the oropharynx into the esophagus, as well as protecting the airway by flipping the epiglottis. Animal models represent a powerful tool in swallowing physiology for a number of reasons. First, they allow for data to be collected at much higher temporal resolution [100 fps vs. 15 fps in human infants (43)]. The temporal resolution of data collection is critical because an entire swallow can occur in less than 0.75 s, four frames in clinical work, which is inadequate for accurate quantification of most measures of swallow physiology. Second, they allow for longitudinal collection of data to test the impact of postnatal maturation on swallow physiology, which is difficult to accomplish in clinical care, resulting in most studies being cross sectional, rather than longitudinal (24). Third, an animal model allows for spatial and temporal analyses of the structures involved in infant feeding at a resolution that cannot be accomplished in clinical settings, as many structures, like the hyoid bone, are not visible in videofluoroscopy until infants are at least 9 mo of age (58).

We have three primary questions in this work. 1) Do different properties of suction generation differ between term and preterm infants? 2) Which structures show significant changes in movement as a function of changes in bolus size? And 3) How does postnatal maturation impact performance, and is there an interaction between gestational age at birth and postnatal maturation? We use bolus size as a metric of overall swallow performance because it is both the strongest predictor of the likelihood of aspiration in infants (49) and because it represents how efficient infants are feeding, with larger boluses indicating more efficient feeding (48).

METHODS

Animal housing and care.

Animal housing and care followed those outlined in Refs. 46 and 49, and were approved by the Northeast Ohio Medical University (NEOMED) Institutional Animal Care and Use Committee (No. 17–04–071). Infant pigs (Yorkshire/Landrace sows, Soup Farms, Wooster, OH) were delivered via cesarean section at term (n = 1 litter, 8 infants, 114 days gestation) or preterm [n = 1 litter, 16 infants, 108 days gestation; equivalent to 30–32-wk human gestation (14)], following published protocols (1, 47). Infants used in this study were part of a larger study on the effect of preterm birth and RLN lesion on infant feeding performance, and include only nonlesioned individuals [n = 4 term, n = 5 preterm (46, 49)].

For the first 24 h of life, infants were fed colostrum (CL-Sow Replacer, Cuprem Inc., Kenesaw, NE), and transitioned to formula for the rest of the experiment (Solustart Pig Milk Replacement, Land o’ Lakes, Arden Mills, MN). For the duration of the experiment, infant pigs were fed via a bottle and modified standard (not crosscut) nipple (NASCO Farm & Ranch, Fort Atkinson, WI). Care followed standard protocols for infant pigs (26, 27, 66).

Surgical procedures.

Surgical procedures are as mentioned previously in Ref. 46. All infant pigs underwent two separate procedures before data collection.

Tantalum markers (0.8 mm) were placed into the subdermal space dorsal to the snout, in the hard palate, tongue midline (anterior, middle, and posterior locations), soft palate, and palatopharyngeal arches while pigs were under isoflurane anesthesia between 5 and 6 days of age using a custom bead injector needle.

In a separate sterile surgery occurring between 5 and 6 days of age, beads adjacent to the hyoid bone and thyroid cartilage were sutured in place. Hyoid markers were sutured to the location of the insertion of the sternohyoid at the midline. Thyroid markers were sutured to the fascia over the thyroid eminence.

Data collection.

Data collection procedures are outlined in detail in Mayerl et al. (48, 49), but in short, we collected videofluoroscopic data (GE 9400C-Arm, 75–85 kV, 4–5 MA) with high-speed video (100 fps, XC1M, XCitex, Cambridge, MA) using standard X-ray Reconstruction of Moving Morphology [XROMM (6)] techniques when pigs were 7 days postnatal (2–3-mo human postnatal development) and 17 days postnatal [6–9-mo human equivalent (14)]. At 7 days postnatal, infant pigs can sustain their body temperatures to be transported out of the temperature-controlled Comparative Medicine Unit, and at 17 days, pigs begin to consume a larger proportion of solid food (64). During recording, infants were fed infant formula mixed with barium (E-Z Paque Barium Sulfate, EZ EM Inc., NY). Pigs fed ad libitum during this time, and we selected at least 20 swallows per pig per age for analysis.

Data processing.

Swallows were identified by determining the frame where the bolus was accumulated in the supraglottic space before passing the epiglottis following published procedures (47, 49). We used the swallows identified in our previous work on hyoid and soft palate excursion on the same sets of infant pigs (13, 46). We combined kinematic data of movements of the hyoid and soft palate with data on bolus size (49) for 307 swallows at day 7 (n = 135 term, 172 preterm) and 277 swallows at day 17 (n = 138 term, 139 preterm). Details on the measurement of hyoid and soft palate excursion can be found in Refs. 13 and 46, and a description of bolus size measurement for each swallow can be found in Ref. 49.

We identified each suck in the feeding sequence by determining the frame where the middle tongue made contact with the hard palate in the lateral x-ray view. Suck amplitude was measured as the maximum dorsoventral distance between the hard palate and tongue during a suck cycle (Fig. 1). Suck length was calculated as the distance from the nipple to the location of the tongue-soft palate seal at the frame of maximum suck amplitude using ImageJ (61), which corresponds to the moment of maximum suction generation (22). Seal length was measured as the distance from the anterior end of the hard palate to the tip of the nipple (Fig. 1). For each video, we scaled the image to millimeters using an object of known size to control for differences in the pig position and orientation relative to the X-ray unit. For comparisons of suck dynamics with swallow properties, we averaged the suck measure for each suck preceding the swallow.

Fig. 1.

Measurements of suck amplitude (dotted orange line), suck length (solid orange line), and seal length (white line).

We calculated total pharyngeal transit time by subtracting the time of swallow initiation from the time that the bolus passed over the epiglottis and the epiglottis began protracting back to its resting state. To calculate bolus velocity during pharyngeal transit, we measured the distance between the most posterior hard palate marker and the most anterior point of the bolus at the start of the swallow and at the end of pharyngeal transit. The difference between these distances represented total bolus excursion, which we then divided by total pharyngeal transit time to find bolus velocity.

Statistical analyses.

All statistical analyses were performed in R (v 3.5.0, http://www.r-project.org). To evaluate the relationship between the number of sucks per swallow and the length of each suck, we used a multinomial logistic regression with suck length, gestational age at birth, and postnatal age as the main effects. We performed linear mixed effects models using the R package lme4 (4) to test for the relationship between suck length and gestational age at birth, postnatal age, and their interaction, with individual as a random effect. We performed similar analyses in our previous work (reported here for continuity) for hyoid excursion and soft palate excursion work (13, 46).

We performed linear mixed-effects models to test for the impact of gestational age at birth, postnatal maturation, bolus size, and their interactions on mean suck length per swallow (the average suck length of the sucks preceding each swallow), total pharyngeal transit time, bolus velocity, soft palate excursion, and hyoid excursion, with individual as a random effect [variable ∼gestational age at birth × postnatal age × bolus size + (1/individual) (28)]. Where interactions were significant, we performed linear regression analyses for each group individually to test the impact of bolus size on performance. Data and code used for statistical analyses are available upon request.

RESULTS

Although preterm infants weighed less than term infants at birth (preterm = 0.61 ± 1.14 kg, term = 1.41 ± 0.17 kg, Tukey’s post hoc t = −4.2, P = 0.002), by the time infants reached 7 days postnatal, these differences were not significant (preterm = 1.5 ± 0.40 kg, term = 1.83 ± 0.25 kg, Tukey’s post hoc t = −1.5, P = 0.64). At 17 days postnatal, infants were also similar in body mass (preterm = 2.7 ± 0.8 kg, term = 2.7 ± 0.32 kg, Tukey’s post hoc t = −0.1, P = 1). Because there was no difference in size, we did not normalize data for size in subsequent analyses.

Suck dynamics.

There were no differences in suck amplitude between term and preterm infants at either age, although suck amplitude did increase with postnatal age in preterm (Table 1, t < 0.001), but not term (Table 1, t = 0.50) infants, indicating an interaction between age and gestational age at birth (Fig. 2A, χ² = 11.42, P = 0.001). In contrast, we found substantive differences in suck length between term and preterm infants at both ages. Preterm infants had lower suck lengths than term infants at days 7 and 17 (Table 1, χ² = 7.29, P = 0.007), and postnatal age resulted in increased suck length for term and preterm infants (χ² = 231.17, P < 0.001). We found no interaction between gestational age at birth and postnatal age (Fig. 2B, χ² = 0.002, P = 0.96). There were differences in nipple placement within the mouth, as preterm infants tended to position the nipple deeper in the oral cavity than term infants (preterm seal distance mean ∼28 mm, term seal distance ∼19 mm). As suck length showed greater variation than suck amplitude, we focused subsequent analyses on how it impacted other performance metrics.

Table 1.

Means (±SD) values for preterm and term infants at 7 and 17 days of age

	Preterm 7	Term 7	Preterm 17	Term 17
Sucks per swallow	2.82 ± 0.95	2.14 ± 0.85	2.63 ± 0.97	1.98 ± 0.37
Suck amplitude, mm	7.43 ± 1.75	8.68 ± 0.98	7.89 ± 1.15	8.89 ± 0.89
Suck length, mm	15.14 ± 4.56	27.61 ± 5.62	20.71 ± 8.14	31.60 ± 2.76
Total Pharyngeal Transit (TPT), s	0.099 ± 0.02	0.117 ± 0.02	0.098 ± 0.02	0.180 ± 0.05
Bolus velocity, mm/s	183.02 ± 57.97	205.05 ± 36.13	202.50 ± 62.50	193.13 ± 31.07
Bolus size, mm2	87.44 ± 35.95	141.82 ± 50.90	121.63 ± 54.93	177.87 ± 35.32
Soft palate excursion, mm	1.53 ± 0.64	3.85 ± 1.02	1.59 ± 0.57	3.53 ± 0.62
Hyoid excursion, mm	3.02 ± 0.68	2.94 ± 0.76	3.75 ± 0.63	4.30 ± 0.60

Fig. 2.

Suck amplitude (A) and suck length (B) in preterm (red) and term (blue) infants at 7 and 17 days postnatal. Boxplots indicate mean and interquartile range, width of violin plot indicates distribution of the data along the y-axis, small dots indicate outliers, large black circles indicate means, and lines between plots indicate statistically significant differences between groups.

Sucks per swallow and the relationship to suck length.

Preterm infant pigs had more sucks per swallow than term infant pigs at both ages (P = 0.003, Table 1), although there was no effect of postnatal age nor an interaction between postnatal age and gestational age at birth. Suck length was the only predictor of the number of sucks taken per swallow, with larger suck lengths predicting fewer sucks per swallow (Table 2, Fig. 3).

Table 2.

Multinomial regression results

Sucks per Swallow	Suck Length Odds, P	Birth Status Odds, P	Age Odds, P
2	−0.16 ± 0.06 (<0.001)	0.21 ± 0.70 (0.76)	1.44 ± 0.62 (0.21)
3	−0.27 ± 0.06 (<0.001)	−0.20 ± 0.82 (0.80)	1.26 ± 0.68 (0.06)
4	−0.25 ± 0.07 (<0.001)	−1.36 ± 1.07 (0.20)	0.93 ± 0.74 (0.21)
5	−0.20 ± 0.08 (0.002)	−0.38 ± 1.22 (0.76)	0.98 ± 0.95 (0.30)
6	−0.24 ± 0.11 (0.003)	Not Applicable	1.95 ± 1.41 (0.17)

Fig. 3.

The relationship between the number of sucks per swallow and the average suck length. A larger distance is more likely to have fewer sucks per swallow than a smaller distance.

Suck length related to bolus size.

There was an interaction between bolus size, postnatal age, and gestational age at birth for average suck length per suck (χ² = 52.5, P < 0.001). Thus, we performed linear regression analyses on the relationship between suck length and bolus size for each of the four groups independently. We found a statistically significant relationship between bolus size and suck length for all groups except 17-day-old term infants (Table 3, Supplemental Table S1; all Supplemental material is available at https://doi.org/10.6084/m9.figshare.12760097.v1). Preterm infants and younger term infants had a positive correlation between bolus size and suck length (Fig. 4A).

Table 3.

Statistics from regression results, indicating t values, P values, and (R²) values

	Preterm 7	Term 7	Preterm 17	Term 17
Seal distance	6.6, <0.001 (0.32)	5.9, <0.001 (0.38)	12.7, <0.001 (0.65)	−1.0, 0.31 (0.001)
TPT	3.9, <0.001 (0.14)	7.8, <0.001 (0.52)	13.5, <0.001 (0.67)	6.5, <0.001 (0.45)
Bolus velocity	4.5, < 0.001 (0.18)	2.9, 0.004 (0.12)	16.1, <0.001 (0.75)	−1.8, 0.08 (0.04)
Soft palate	2.6, 0.01 (0.09)	2.0, 0.05 (0.06)	9.5, <0.001 (0.65)	0.4, 0.71 (−0.02)
Hyoid	0.13, 0.90 (0.01)	1.7, 0.09 (0.04)	2.3, 0.02 (0.09)	−1.1, 0.30 (0.04)

Bold values are statistically significant and explain at least 30% of the variation.

Fig. 4.

The relationship between bolus area and suck length (A), soft palate excursion (B), and hyoid excursion (C) in infant pigs. At day 17, term pigs (red) exhibit different behaviors than younger term pigs (orange) and preterm pigs at both ages (blue and teal). We found a strong relationship between suck length and bolus size, a weak but significant effect of soft palate excursion on bolus size in all groups but 17-day old term infants (red), and no relationship between hyoid excursion and bolus size in any group.

The summed suck length for all sucks per swallow did not differ between term and preterm infants (χ² = 1.75, P = 0.19), although older pigs did have a larger summed distance than younger pigs, regardless of gestational age at birth (χ² = 9.13, P = 0.003).

Soft palate excursion relative to bolus size.

Our previous work identified an interaction between gestational age at birth and postnatal age on soft palate excursion (χ² = 9.48, P = 0.002), with term pigs exhibiting higher soft palate excursion than preterm pigs, regardless of age [Table 1, day 7, t = 14.99, P < 0.001; day 17, t = 13.02, P < 0.001 (13)]. Adding bolus size to the model resulted in a significant interaction between gestational age at birth and postnatal age (χ² = 6.58, P = 0.01), with no interactions with bolus size.

Due to the interaction between gestational age at birth and postnatal age, we performed separate analyses of soft palate excursion relative to bolus size in each of the four groups. We found a significant relationship between soft palate excursion and bolus size in all groups except older term infants whereby a larger soft palate excursion was correlated with a larger bolus size (Fig. 4B), although R² values were only substantial in older preterm infants (Table 3, Supplemental Table S1).

Hyoid excursion relative to bolus size.

Our previous work identified an interaction between gestational age at birth and postnatal age on hyoid excursion (χ² = 21.99, P < 0.001). We found an effect of postnatal age on hyoid excursion in preterm (t = −7.14, P < 0.001) and term infant pigs (t = 13.26, P < 0.001), with older infants exhibiting greater hyoid excursion (Table 1). No effect of gestational age at birth was found for either age [P > 0.3 (46)]. The addition of bolus size to the model resulted in a significant interaction between gestational age at birth and postnatal age (χ² = 7.19, P = 0.007), with no interactions with bolus size.

Due to the interaction between gestational age at birth and postnatal age, we performed separate analyses of hyoid excursion relative to bolus size in each of the four groups. Regression results of hyoid excursion relative to bolus size identified no relationship within any of the four groups (Fig. 4C, Table 3, Supplemental Table S1).

Total pharyngeal transit time related to bolus size.

We found a significant interaction between bolus size, gestational age at birth, and postnatal age on pharyngeal transit time (χ² = 12.43, P < 0.001). Thus, we performed linear regression analyses on the relationship between suck length and bolus size on each of the four groups independently. There was a significant positive relationship between bolus size and pharyngeal transit time in all four groups, with larger boluses resulting in longer transit times for all four groups (Fig. 5A, Table 3, Supplemental Table S1), although this was most evident in older term infants (Fig. 5A).

Fig. 5.

The relationship between bolus area with pharyngeal transit time (A) and bolus velocity (B) in infant pigs. In all groups, increased bolus area is associated with increased pharyngeal transit time, and in all groups but older term infants (red), a larger bolus area is associated with increased bolus velocity.

Bolus velocity related to bolus size.

There was a significant interaction between bolus size, gestational age at birth and postnatal age on bolus velocity (χ² = 11.14, P = 0.001). As such, we performed linear regression analyses on the relationship between suck length and bolus size on each of the four groups independently. There was a significant positive relationship between bolus size and bolus velocity, with larger boluses resulting in faster velocities for all groups except older term infants (Fig. 5B, Table 3, Supplemental Table S1). However, R² values were only substantial for older preterm infants (Table 3).

DISCUSSION

Our results highlight the pathophysiological neural control of feeding function in preterm infants (2, 3) and show that the problems that preterm infants face in their feeding physiology are complex and multifaceted. Preterm infants had decreased abilities to generate suction as exhibited by decreased suck lengths and increased sucks per swallow, lower excursion of the soft palate, and did not mature in their feeding physiology when compared with term infants. Preterm infants thus not only experience challenges with acquiring and transporting milk (41, 42), with the swallow itself (13, 39, 59), and with coordinating behaviors within the swallow (46, 48), but also most likely experience difficulties due to the spatiotemporal links between food acquisition, transport, and swallowing. By studying infant feeding from an integrated perspective, we can elucidate which aspects of infant feeding physiology are sensitive to sensorimotor feedback, providing insight into the neural control of feeding, as well as potential targets for interventions to improve feeding performance.

The primary determinant of increased suction generation appears to be the length of the suck, rather than the dorsoventral pumping of the tongue, as younger infants and preterm infants had decreased suck lengths, but similar suck amplitudes, compared with older infants and term infants. Previous research on suction performance in infants has focused on performance using pressure transducers (25, 41, 42, 50), but has not examined how the location of the seal between tongue and palate impacts the role of producing negative pressure to generate suction (67). Suck length is likely to be critical to overall feeding performance, as it was tightly correlated with bolus size, and appears to be the mechanism driving the number of sucks taken per swallow, with smaller suck lengths resulting in more sucks being necessary to acquire an adequate amount of milk reach a threshold volume to stimulate the internal branch of the superior laryngeal nerve and trigger a swallow. Future work should explicitly examine the relationship between kinematic parameters of the suck (suck length and amplitude) and changes in intraoral pressure.

Our previous work comparing preterm and term infant swallow performance identified differences between term and preterm infant bolus sizes (49), as well as in the coordination of the hyoid bone and thyroid cartilage (46), and in palatopharyngeal arch and soft palate function (13). Our current analyses focused on the relationship between the movement of these structures and bolus size, and aids in understanding their sensitivity to sensory input regarding the swallow. We found very little relationship between variation in swallow properties (bolus size) and behaviors associated with the pharyngeal stage of the swallow (hyoid and soft palate movement). However, we found a strong correlation between suck length, occurring in the oral stage of the swallow, and bolus size, and that decreased suck length in preterm infants reduces milk acquisition, which they compensate for partially by increasing the number of sucks per swallow to enable them to reach an adequate volume to trigger a swallow. The strong relationship between properties of the “oral stage” of the swallow (suck length and sucks per swallow) and bolus size, and the weak relationships between properties of the “pharyngeal stage” of the swallow (hyoid and soft palate excursion) highlight the differences in the neural control of feeding in infants. The sensory information obtained during infant suckling before the onset of the swallow is sensitive to sensorimotor feedback (19, 31, 62) and has the potential to modulate properties of the swallow itself (as exhibited by changes in bolus size). In contrast, movements of structures during the pharyngeal stage are likely to be closer to being reflexive, and similar to work on adults, show little sensitivity to sensorimotor feedback (34, 36). This suggests that research that focuses on sucking or swallowing in isolation is likely to provide an incomplete picture of feeding performance in infants, from both clinical and physiological perspectives.

Clinical implications.

Clinically, a poor understanding of the mechanisms governing performance results in a limitation in the design of evidence-based interventions. By understanding the impact of acquisition and transport on swallow performance, we can design interventions to improve preterm swallow performance by altering their suck lengths. For example, a nipple designed to ensure a latch where the nipple was closer to the anterior margin of the mouth might allow infants to have a greater space to generate pressure to acquire milk and would enable preterm infants to improve their ability to generate suction. So long as caution was taken to ensure nipple design did not alter flow rates, this would not only improve their ability to acquire and transport food, which they struggle to do (38, 39, 41, 42), but also likely improve their swallow performance, and result in increased pharyngeal kinetic rhythmicity and more efficient esophageal sphincter kinetics (57, 59).

The possibility that altering properties of suction generation would improve swallow performance is further supported by evidence from research on nonnutritive sucking, which has demonstrated that altering properties of pacifiers in human infants changes infant behavior and performance (53, 71, 72). By targeting behaviors that are sensitive to sensorimotor feedback, instead of focusing only on end products of feeding performance, interventions might decrease preterm infant feeding difficulties and improve their ability to coordinate among structures during feeding (10, 13, 46, 47). The interaction between infant biomechanics and milk acquisition is complex, and more work is needed to determine how altering properties of the nipple impact performance (21, 23).

Implications for physiology.

Physiologically, studying components of feeding in isolation limits the discovery of mechanisms driving performance, from both neural and biomechanical perspectives. Infant feeding requires integrated control of multiple behaviors (suction generation, intraoral transport, and swallowing) and multiple oropharyngeal structures controlled by over 25 paired muscles (30, 51). Most of these structures, including the tongue, hyoid, and soft palate, are innervated by multiple muscles, and require a coordinated sequence of activity to move successfully. Furthermore, many of these muscles (especially those of the tongue and suprahyoid muscles) are involved in both the oral and pharyngeal phases of feeding. In contrast, the action of a single nerve, the interior superior laryngeal nerve (iSLN) is all that is required to elicit a swallow (12, 51, 63). We found that term and preterm infants require similar amounts of total suction to reach a threshold level for the iSLN to trigger, but because preterm infants had decreased abilities to generate suction, they required more sucks to reach the threshold volume to trigger the swallow. This suggests that the sensitivity of the iSLN may be similar between term and preterm infants, although preterm infants still exhibit smaller bolus sizes than term infants at any age (48, 49). However, this is likely due to the fact that the behaviors leading up to, and following the trigger of the iSLN have the potential to exhibit differences in their sensitivity to sensory feedback (34, 36), and thus differences in the mechanisms giving rise to dysfunction.

Finally, preterm birth appears to fundamentally alter the postnatal neural maturation of the brain, as by day 17, term infants appear to exhibit a different swallow physiology than preterm infants and younger term infants, which is similar to recent research on the development of nonnutritive sucking in term infants (45). Instead, older term infants appear to have reached a point where their swallowing may be prepared for weaning, which normally begins at ∼17 days of age (5, 64). Older term infants showed a much weaker relationship between bolus size and suck length, bolus velocity, and total pharyngeal transit time than younger infants and older preterm infants. This likely reflects the overall larger boluses of term infants and suggests that the control of infant feeding changes over ontogeny as infants approach weaning (8, 64). In contrast, older preterm infants showed very little of this neural maturation, and likely struggle to develop their swallow physiology as they approach a weaning age. Our previous work has shown that increases in bolus size in preterm pigs as they age is entirely accounted for by increases in body size, whereas in term pigs it increases faster than body size, resulting from greater filling of the valleculae (48). The changes in neural and motor control that allow term pigs to fill the valleculae more may result in different strategies for dealing with large boluses as animals get older. Such changes do not happen in preterm pigs. Preterm infant humans also experience challenges associated with weaning, likely a reflection of the profound and long-lasting impact of preterm birth on brain development (2, 17, 56). Our results highlight potential neural and biomechanical mechanisms limiting their development, as their reduced suction generation and subsequent inability to increase bolus size substantially may result in their feeding physiology not getting the experience needed to prepare for swallowing solid foods.

Limitations and future directions.

Although the general relationships we have demonstrated here are likely to apply to all term and preterm mammals, there are a number of limitations relative to our specific conclusions. First, it is well known that nipple properties impact feeding performance in human infants (11, 16, 55, 69), and it is possible that our specific results may differ if infants fed on a nipple with different material properties, although the patterns observed here would likely remain similar. Our results are also most generalizable to infants fed on bottles, as breastfeeding fundamentally differs from bottle feeding in terms of both muscle function and pressure generation (23, 37, 52), and an effective latch depends not just on the infant, but also on the caregiver during breastfeeding. Future work should examine the impact of nipple properties and the differences between breast and bottle in the integrated framework presented here. Finally, although we have examined feeding along the suck-swallow axis, feeding in infants must also be coordinated with respiration. Preterm infants generally exhibit poor swallow-breathe coordination (47, 70), and it is possible that smaller boluses in these infants reflect a mechanism to reduce the frequency of aspiration during feeding (48, 49). Future research should aim at testing potential interventions targeted at enhancing suck-swallow-breathe coordination in these populations, and the potential that altering parameters of sucking might improve swallow-breathe coordination.

CONCLUSIONS

This work and others together demonstrate that the problems that preterm infants face are substantial, multifaceted, and rooted in a neural control of feeding that is pathophysiological (2, 47). Furthermore, their similar levels of performance at single tasks, but reduced performance in behaviors requiring coordination suggest that these problems arise due to poor neural connections required to coordinate between behaviors (46, 47, 49). By studying infant feeding from an integrated perspective, we show that behaviors occurring in the oral stage of feeding are more likely to show sensitivity to sensory inputs than those occurring after the swallow has been initiated. In particular, we found that suck length was better correlated with bolus size, and thus swallow performance, than either hyoid or soft palate excursion. Prematurity affects both function and maturation of the components of the feeding system, resulting in a complete feeding sequence that is fundamentally unlike that of term infants by the time of weaning.

Data Availability

Data used in statistical analyses are available upon request.

GRANTS

This project was funded by National Institutes of Health R01HD088561 to R. Z. German.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

C.J.M. and R.Z.G. conceived and designed research; C.J.M., C.E.E., E.A.C., A.M.M., F.D.H.G., L.E.B., B.M.S., and R.Z.G. performed experiments; C.J.M., C.E.E., E.A.C., and A.M.M. analyzed data; C.J.M. interpreted results of experiments; C.J.M. prepared figures; C.J.M. drafted manuscript; C.J.M., C.E.E., E.A.C., A.M.M., F.D.H.G., L.E.B., B.M.S., and R.Z.G. edited and revised manuscript; C.J.M., C.E.E., E.A.C., A.M.M., F.D.H.G., L.E.B., B.M.S., and R.Z.G. approved final version of manuscript.