The study's hypotheses, specific aims, common protocol, enrollment, shipping, compliance, and specimen donation have been described in detail (34), as well as the approach to autopsy consent in socioeconomically disadvantaged populations (47). In brief, the Safe Passage Study was an international prospective, multicenter longitudinal cohort study with data collection conducted between August 2007 and October 2016. Clinical sites were selected based upon known high rates of maternal drinking and smoking during pregnancy and known high rates of SIDS in the population; however, all women from the catchment areas presenting for care at these sites were eligible to participate. These predominately include pregnant (1) American Indian and Caucasian women from the Northern Plains and (2) mixed ancestry women of the Western Cape, South Africa. Screening and enrollment occurred at pre-natal clinics affiliated with each clinical site between 6 weeks gestation up to, but not including, delivery. The maternal and fetal/infant dyads were followed during pregnancy and from birth until infants were 1 year of age, i.e., the risk period of SIDS. Detailed information regarding quantity, frequency, and timing of substance use was self-reported up to 4 times during pregnancy (recruitment, 20–24, 28–32, and 34+ gestational weeks) and at 1 month post-delivery. At sites in South Africa, a medico-legal autopsy was performed upon demise. Consent for research was sought from the family as soon as possible after death (47). At sites within the United States, an autopsy was regularly ordered by the coroner/medical examiner after which the family was approached for consent to the donation of tissue for research purposes. If consent was given, brain portions were frozen and shipped on dry ice to the Developmental Brain and Pathology Center (DBPC), Department of Pathology, Boston Children's Hospital, the centralized laboratory for research analysis (48). The study, including the use of brain tissue, was approved by the Institutional Review Boards (IRBs) of the local hospitals at which the infants were autopsied and Boston Children's Hospital. When designed, the Safe Passage Study for brain analysis anticipated and was powered on 37 SIDS cases and 37 non-SIDS controls, which would allow detection of a difference in receptor binding levels as small as 0.67 standard deviation.

SIDS was defined using a study definition that included the sudden unexpected death of an infant, <1 year of age, whose cause of death remained unexplained after review of all available information, including performance of a complete autopsy, examination or report of the death scene, and review of the clinical history (1). In addition, SIDS included deaths that might otherwise have been classified as undetermined including infants dying in unsafe sleep conditions but without evidence of mechanical asphyxia or suffocation by overlay. Known cause of death (KCOD) controls were defined as infants whose cause of death was identified after review of all available information (33). All demises were adjudicated by a team of pediatric pathologists, neuropathologists, forensic pathologists, a neonatologist, geneticist, obstetrician, and developmental psychologist who were blinded to pre-natal exposures. Case reviews did include toxicology, genetic testing (when appropriate and available), and metabolic testing (when appropriate and available). The postmortem interval (PMI) was calculated as the time when the infant was last seen alive to proclamation of time of discovery, as in previous studies by us (16, 49). Prospective collection of demise cases in the Safe Passage Study included 28 SIDS and 38 control cases that died after discharge from the hospital (post-discharge) (33). Of these cases, 12 SIDS and 10 post-discharge KCOD (post-KCOD) controls were available for neurochemical studies. There were 16 SIDS and 28 post-KCOD cases that were accrued in the Safe Passage Study but not available for neurochemistry either due to a lack of consent for autopsy research or due to technical issues related to quality of tissue. There were 45 KCOD controls that died prior to leaving the hospital after delivery [pre-discharge KCOD (pre-KCOD)] and 10 of these were analyzed for ¹²⁵I-epibatidine binding to provide baseline developmental data.

At autopsy, the brain was removed, weighed fresh, and examined for gross developmental and acquired abnormalities. The entire brainstem was removed from the level of the midbrain at the mammillary bodies to the cervicomedullary junction in 1.5-cm samples, sectioned on a Leica motorized cryostat at 20 μm, mounted on glass microscopic slides, and stored at −80°C until used for tissue receptor autoradiography. Frozen blocks were stored at −80°C in air-tight plastic containers.

To examine nicotinic receptors, we used receptor ligand autoradiography. Unlike homogenate radioligand binding, autoradiography allows the visualization of spatial anatomy and provides details of regional expression patterns (50). It is quantitative in nature and thus provides benefits over other techniques like immunohistochemistry. Radioligand ¹²⁵I-epibatidine was used for autoradiographic analysis. Epibatidine is a nAChR agonist with high affinity to α4β2 nAChRs, one of the major subtypes of nAChRs in the brain, and with lower affinity to α7 nAChRs (51). The autoradiography procedures were performed according to the detailed methodology previously reported from our laboratory for ³H-epibatidine (52), with modification for the iodinated ligand. All steps were performed at room temperature. In brief, total ¹²⁵I-epibatidine binding was determined by incubation of the frozen, unfixed sections with 0.5 nM ¹²⁵I-epibatidine (2200 Ci/mmol, PerkinElmer, Waltham, MA, USA) in binding buffer consisting of 50 nM Tris–HCl, 120 mM NaCl, 5 mM KCl, 2.5 mM CaCl₂, and 1 mM MgCl₂, pH 7.4 for 60 min, an incubation time sufficient for equilibrium to be reached in epibatidine binding experiments (53). Non-specific binding was determined in adjacent sections by addition of 0.5 nM ¹²⁵I-epibatidine and 300 μm of L-nicotine bitartrate. To remove unbound ligand, the sections were washed in a series of buffer changes (5 min each) followed by 3 dips in distilled water. Sections were then left overnight for drying, after which they were placed in cassettes and exposed to a BAS_TR2025 phosphoimaging plate (GE Healthcare Life Sciences, Marlborough, MA) for 20 h, with a set of ¹²⁵I standards (American Radiolabeled Chemicals, Inc, MO, USA) calibrated by the manufacture in terms of radioactivity per unit weight. The standards allowed for the conversion of relative optical density to fentamoles per milligram (fmol/mg) of tissue to determine binding levels. A BAS-500 Bioimaging Analyzer (Fuji-Film) with Image Reader version 1.8 software (FujiFilm) was used to generate digital autoradiographic images from phosphoimaging plates. Quantitative densitometry of autoradiograms was performed using a MCID 5+ imaging system (Imaging Research). Specific receptor binding was determined by subtracting non-specific binding from total binding in individual tissue sections. The same sections incubated for autoradiography were subsequently stained with hematoxylin and eosin for anatomical assessment.

For relevance, we provide detail of the neuroanatomical functions and connectivity of the medullary and pontine nuclei studied here (see Supplemental Material). The human brainstem sites measured (Figure 1) were defined with reference to Olszewski and Baxter brainstem atlas (54) and confirmed with Paxinos and Huang brainstem atlas (55). For each case, ¹²⁵I-epibatdine binding in the brainstem was measured at 3 levels; mid medulla at the level of nucleus of Roller, which included the nucleus of the solitary tract (NTS) (all visceral sensory inputs of the autonomic nervous system and sympathetic autonomic system integration), the hypoglossal nucleus (HG) (airway patency, especially during sleep), dorsal motor nucleus of the vagus (DMX) (preganglionic vagal outflow of the parasympathetic autonomic nervous system), centralis (CEN), principal inferior olive (PIO), and medial accessory olive (MAO) (the cerebellar network); rostral medulla at the level of the nucleus pre-positus, which included the raphe obscurus (RO), gigantocellularis (GC), paragigantocellularis lateralis (PGCL), core nuclei of serotonergic homeostatic medullary network, and the dorsal accessory olive (DAO); and rostral pons at the level of nucleus parabrachialis lateralis, which included the locus coeruleus (LC) (major source neurons of the noradrenergic ascending arousal network), nucleus pontis oralis (PoO) (part of source neurons of the cholinergic ascending arousal system), griseum pontis (GRPo) (pre-cerebellar nucleus, part of the pontocerebellar network), and median raphe (MR) (part of the rostral 5-HT ascending arousal network) (Figure 1). With the exception of the midline nuclei (RO and MR), ¹²⁵I-epibatidine binding was measured from both sides (left and right) of the section and the means calculated to determine final value in fmol/mg.

This study used a modified timeline follow-back method to collect exposure related to maternal smoking/drinking during pregnancy (56). At each visit during and after pregnancy, the participant was asked about the last date of use (separately for alcohol and smoking). For data relating to alcohol, they were asked about consumption for ±15 days around last menstrual period, as well as 30 days prior to the last drinking day since their last research appointment. For smoking, they were asked about the frequency of smoking and number of cigarettes on a typical day for the 30 days prior to the last date of use since their last research appointment. To estimate the total number of drinks consumed during pregnancy, each drink consumed was first standardized where 1 drink is defined as 14 g of ethanol. The study design did not allow consumption data to be collected on every single day of pregnancy, so missing values were imputed using the k-nearest neighbor (kNN) method. Methods for alcohol imputation are cited elsewhere (57). Since frequency of cigarette use was collected more sparsely, average cigarettes smoked per week during pregnancy was used. The number of cigarettes smoked each week during pregnancy was calculated, and missing weeks was imputed in a similar way to the alcohol imputation. After imputation, an average number of cigarettes during pregnancy was calculated.

Descriptive analysis was conducted to analyze differences between cause of death (SIDS vs. KCOD controls) and demographics, maternal substance use during pregnancy, infant sleep practices, and relevant autopsy and clinical findings. Analyses used included Student's t-test or Mann–Whitney U for continuous variables and chi-square testing with Fisher's exact test for categorical variables. Maternal demographics assessed included age, education, housing type, history of loss by SIDS, and delivery type. Infant demographics assessed included birth weight and length, gestational age at birth, post-natal age at death, gender, and race. Autopsy findings assessed included postmortem interval, body weight at autopsy, and brain weight at autopsy. Maternal use of alcohol and smoking during pregnancy were assessed as binary values (used during pregnancy or not) and as continuous values (total number of drinks during pregnancy and average number of cigarettes per week). Maternal use of alcohol and smoking by trimester was assessed as continuous values (number of drinks by trimester and average cigarettes per week). Infant sleep practices assessed included sleep position last placed, sleep position found, and whether or not the infant was covered by bedding or blankets.

Multivariate linear regression models were built to analyze differences in mean ¹²⁵I-epibatidine binding values by case diagnosis and exposure. Post-conceptional age (PCA) was controlled for in all models, as it is significantly different by case diagnosis and associated with ¹²⁵I-epibatidine binding. A subanalysis assessed the effect of development (PCA) on ¹²⁵I-epibatidine binding in the 10 pre-KCOD- and 10 post-KCOD-control infants. There was no effect of PMI on binding; therefore, PMI was not controlled for in the analyses. p < 0.05 were considered statistically significant. Analyses were conducted using SAS 9.4.

The demise cohort for ¹²⁵I-epibatidine analyses include SIDS (n = 12), post-KCOD controls (n = 10), and pre-KCOD controls (n = 10) (see above). The causes of death for the control groups are given in Table 1, with demises separated as pre- or post-discharge, based on whether the infant died in the hospital without being discharged after birth or at home after discharge. Selected demographic data including incidence of exposure are summarized in Table 2. Pre-KCOD cases were included only for the purpose of looking at developmental changes in receptor binding. Pre-KCOD cases (n = 10) ranged from 25.8 to 41.9 postconceptional weeks (mean = 32.8 weeks) (Table 2). Sixty percent of pre-KCOD cases were male (n = 6) and 80% (n = 8) were South African mixed race with the other 20% being American Indian (n = 1) or Caucasian (n = 1). The SIDS cohort was statistically analyzed relative to the post-KCOD cases only. In the comparison between SIDS and post-KCOD, there was no significant difference in mean gestational age (GA), PCA, PMI, birth weight, sex, pre-mature birth, autopsy body, or brain weight (Table 2). The majority of cases was from the South Africa clinical site (Table 2), accounting for the predominately South African mixed-race assignment in SIDS (92%) and post-KCOD (80%) [p = non-significant (ns)]. We assessed several maternal characteristics and found a significant difference only in education status (p = 0.03) with the majority of post-KCOD mothers completing high school (60%) and the majority of SIDS mothers reporting some high school education (67%) (Table 2). Other demographic measures were not significantly different between SIDS and post-KCOD controls. These include crowding index (>1 person/room), employed (yes/no), marital status, and housing type (council housing, informal shack/squatter, apartment/house, other) (data not shown).

The presence of infection in KCOD controls, including peripheral and central infection, is noted in Table 1. Evidence of mild infection was noted at autopsy in 7 out of 12 SIDS cases including group B Streptococcus in the lungs (n = 1), inflammatory changes in the larynx (n = 1), inflammatory changes in the pharynx (n = 1), inflammatory cells in the lamina propria (n = 1), mild pneumonitis (n = 2), and positive Clostridioides difficile in the stool (n = 1).

Information regarding maternal smoking and drinking during pregnancy was available for all cases in the SIDS (n = 12) and post-KCOD cases (n = 10). The incidence of maternal smoking during pregnancy (yes or no) was 100% (12/12) in the SIDS group (Table 2) and ranged from an average of 0.1 cigarettes per week to 62.3 (median of 20.6) (Table 3A). This was not statistically different from post-KCOD cases whose incidence of smoking was 90% (Table 2) and ranged from an average of 0 cigarettes per week to 58.6 (median of 28.7) (Table 3A). Maternal smoking was neither statistically different between SIDS and controls in any one trimester nor was it statistically different across trimesters in SIDS or in controls. The incidence of maternal drinking during pregnancy (yes or no) was 50% (6/12) in the SIDS group (Table 2) and ranged from 0 drinks during pregnancy to 210.75 drinks in pregnancy (median of 2.6) (Table 3B). This was not statistically different from post-KCOD cases whose incidence of drinking was 70% (Table 2) and ranged from 0 drinks during pregnancy to 102.5 drinks in pregnancy (median of 8.3) (Table 3). There was no statistical difference between SIDS and post-KCOD cases when exposure was analyzed by trimesters (Tables 3A,B).

Information regarding sleep-related risk factors was available for all cases within the SIDS group but only 2 of the 10 post-KCOD controls (data not shown). Within the SIDS group, the prevalence of infants last placed supine was 8% (1/12), on their side (lateral) was 33% (4/12), and prone was 58% (7/12). The prevalence of SIDS infants in the demised state on their side (lateral) was 64% (7/11) and in the prone position was 36% (4/11). In the post-KCOD group, the position last being placed supine was two of two; one of two was found dead in the supine position, and the other was found prone.

Information regarding post-natal exposure to cigarette smoke was available in 11 out of 22 (50%) of post-discharge infants. Of these 11, only 5 (2 SIDS and 3 post-KCOD controls) reported the infant being exposed to cigarettes at home. Information regarding other illicit drugs (i.e., marijuana and methamphetamine) was available on 24 of 32 total KCOD controls and SIDS. In total, only 4 cases (one pre-KCOD control, two post-KCOD controls, and one SIDS) reported any illicit drug use.

We assessed ¹²⁵I-epibatidine binding in the medulla and the pons from 26 postconceptional (PC) weeks (midgestation) to 64 PC weeks (~6 post-natal months) in the pre-KCOD (n = 10) and post-KCOD (n = 10) cases for developmental information. The inclusion of pre-KCOD controls allowed us to analyze a greater range of ages and provide increased information on development through gestation. By midgestation binding was differentially localized to nuclei of interest in a relatively fixed distribution (Figure 1). Medullary binding patterns across all controls combined showed the highest binding in the PIO and lowest binding in the RO (Table 4, Figure 1). In the rostral pons, binding in nuclei of the pontine tegmentum (MR, LC, and PoO) were relatively high compared to measurements in the basis pontis (GRPO) (Table 4, Figure 1). There was a significant developmental decrease in binding with increasing postconceptional age (p < 0.05) in the RO, GC, PGCL, and DAO (beta values from −0.39 to −0.25) of the rostral medulla and in the CEN of the mid medulla (beta = −0.22) (Table 4). There was a marginally significant developmental decrease binding in in the LC and PoO of the rostral pons (p = 0.07; beta = −0.22 and p = 0.05; beta = −0.24, respectively). To illustrate this effect, a decrease in ¹²⁵I-epibatidine binding with increasing PC age is shown for the RO, GC, and PGCL of the rostral medulla (Figure 2). Of note, three rostral medullary sites found to have decreased binding with age (RO, GC, and PGCL) were also found to be significantly affected by nicotine exposure when SIDS and post-KCOD controls were combined (see Table 6). The effects of age on these sites remained significant even after adjusting for the effect of smoking (see Table 6).

Using analysis of covariance (ACOVA) controlling for the effect of PCA, there was a significant effect of diagnosis only in rostral pons PoO (p = 0.02) with post-KCOD controls having a higher binding than SIDS (12.1 ± 0.9 fmol/mg and 9.1 ± 0.8 fmol/mg, respectively) (Table 5). There was a marginally significant effect in the LC of the rostral pons (p = 0.08) with post-KCOD controls having a higher binding than SIDS (Table 5). The demise cohort within PASS has a relatively high prevalence of pre-term birth (58%, SIDS; 50% KCOD) (Table 2). Thus, we analyzed ¹²⁵I-epibatidine binding in the PASS cohort based on pre-term (<37 gestational weeks) or term birth (>37 gestational weeks). ¹²⁵I-Epibatidine binding from pre-term SIDS (n = 7) was compared to pre-term post-KCOD controls (n = 5), adjusting for PCA. Similarly, ¹²⁵I-epibatidine binding from term SIDS (n = 5) was compared to term post-KCOD controls (n = 5), adjusting for PCA. We found no significant effect of diagnosis in any nuclei in either pre-term or term cases (data not shown).

We assessed the effect of the amount of maternal drinking (number of drinks consumed) on ¹²⁵I-epibatidine binding in all SIDS (n = 12) and post-KCOD controls (n = 10) combined, adjusting for PCA. There were no significant effects of amount of drinking on ¹²⁵I-epibatidine binding in any nuclei (Table 6). Given that 50% of SIDS mothers and 70% of post-KCOD mothers reported drinking during pregnancy, we repeated the analyses with exposed-only, SIDS (n = 6), and post-KCOD controls (n = 7) combined. Similar to the analyses in Table 5, the analysis in alcohol exposed-only cases showed no effect of amount of drinking on ¹²⁵I-epibatidine binding in any nuclei (data not shown). There was no effect of maternal drinking on pre-KCOD controls (data not shown). Similarly, we assessed the effects of amount of smoking on SIDS and post-KCOD controls combined, adjusting for PCA. A significant positive association between average cigarettes per week and ¹²⁵I-epibatidine binding was found in the following rostral medullary nuclei: RO, GC, and PGCL (p = 0.01, 0.02, and 0.002, respectively) (Table 6). Given the effect of smoking on both SIDS and post-KCOD controls combined, we analyzed each group separately to determine if cigarette smoking had a differential effect on one compared to the other. In post-KCOD controls, we found a significant positive association between ¹²⁵I-epibatidine binding and cigarettes smoked per week in the RO (p = 0.047) and the PGCL (p = 0.03) but little effect in any nuclei in SIDS (Table 7). Of note, there was a significant effect of PCA in the RO, GC, and PGCL of post-KCOD controls consistent with the developmental data of Table 4. Likewise, there was a significant effect of PCA in these same nuclei in SIDS cases. Given the effect of smoking on ¹²⁵I-epibatidine binding in sites of the rostral medulla, we reanalyzed the SIDS vs. post-KCOD control data to examine the effect of diagnosis controlled for smoking and PCA. The significant effect of diagnosis remained in the PoO (p = 0.04). A marginal significance of diagnosis was seen in the GC (p = 0.07) and DAO (p = 0.07) (data not shown).

The baseline (control) expression pattern of the nicotinic receptors in the brainstem of humans and other species has been studied extensively at the mRNA, protein, and receptor level via the methods of in situ hybridization, immunohistochemistry, and receptor-ligand binding autoradiography, respectively [reviewed in Vivekanandarajah et al. (36)]. The relative distribution of ¹²⁵I-epibatidine binding in selected human infant brainstem sites of this study is similar in pattern to previous observations with ³H-epibatidine binding (40), ³H-nicotine binding (19, 21, 59), and immunohistochemistry (24, 60). In this study, we observed a significant decrease in ¹²⁵I-epibatidine binding with increasing PCA in all controls combined (pre- and post-discharge) in the rostral medullary nuclei RO, GC, PGCL, and DAO, mid-medullary nucleus CEN, and a marginally significant effect of age in the PoO of the rostral pons. Given that our KCOD controls ranged from 26 to 64 PC weeks, these data provide information on nAChRs and epibatidine binding prenatally through the second half of gestation and postnatally into the first 6 months of life. The decrease in binding with age suggests that this developmental window is a dynamic period of cholinergic influence on brain development and function. While a decrease in the DAO, PoO, and CEN with age has previously been reported using either ³H-epibatidine (DAO) (40) or ³H-nicotine (CEN, PoO) (19), a developmental decrease in the RO, GC, and PGCL using ¹²⁵I-epibatidine as a ligand has not been shown. The RO, GC, and PGCL nuclei of the medullary 5-HT system contain 5-HT cells and project diffusely to other regions of the brainstem and the spinal cord. These nuclei form part of the homeostatic network of the rostral medulla, critically involved in cardiorespiratory integration and arousal (11). Serotonergic neurons within the nuclei mediate these homeostatic responses (11, 61, 62). Given that serotonergic neurons in these regions express nicotinic receptors (40) and that these serotonergic neurons are likely undergoing developmental changes in 5-HT neurotransmission (as determined by developmental changes in ligand binding to general 5-HT receptor agonist ³H-LSD), particularly from midgestation to infancy (63), developmental changes in ¹²⁵I-epibatidine binding support a dynamic and complex relationship between the neurotransmitter systems—a relationship likely vulnerable to dysregulation during development due to pre-natal exposure to nicotine and inappropriate nicotinic receptor binding at these sites. Evidence supporting the effects of nicotine on 5-HT function is detailed below. Studies of additional neurotransmitter systems (including 5-HT) in the PASS cohort are warranted to address potential interrelationships between receptor systems.

In our analysis between SIDS and post-KCOD controls, we found no difference in ¹²⁵I-epibatidine binding in the SIDS brainstem in 14 out of 15 brainstem sites. This lack of difference is consistent with the findings of Duncan et al. (21) and Nachmanoff et al. (19), both of which used nicotine as a ligand as opposed to epibatidine and both of which showed no difference in any nuclei when SIDS were compared to controls. We did, however, find a significant decrease in SIDS compared to post-KCOD controls in the PoO and a marginal decrease in the LC, both nuclei of the rostral pons. The LC and PoO are components of the extrathalamic and thalamic arousal system, respectively, and a decrease in cholinergic neurotransmission at these sites in SIDS infants potentially reflect an impairment of arousal responses. With regard to sleeping/waking, the functional significance of a nAChR deficiency in the PoO and LC (marginally decreased) of the SIDS cases compared to controls is unknown. However, given a postulated role of the PoO in the generation of rapid eye movement (REM) sleep (64), we speculate that this lesion may interfere with REM–non-REM (NREM) sleep regulation and the generation of REM sleep in SIDS cases with associated dysfunction in breathing and/or heart rate. Similarly, the LC's involvement in sleep to wake transitions and arousal (65) further support a role for dysfunctional sleep regulation as part of the pathogenesis of SIDS. While the current finding of decreased binding in the PoO and LC in SIDS differs from previous studies that found no change at these sites, these pontine sites were previously identified as affected by maternal smoking (19), thus supporting their inherent vulnerability to alterations in cholinergic development and potentially function.

A major finding in this study is that pre-natal smoking is associated with an increase in ¹²⁵I-epibatidine binding in SIDS and controls combined in 3 sites of the rostral medulla after controlling for PCA—the GC, RO, and PGCL. Additional analyses show that this significant effect is being driven mainly by post-KCOD controls, particularly in the RO and PGCL where smoking significantly increases ¹²⁵I-epibatidine binding in the controls but not in SIDS. Our data of increased receptor binding in response to pre-natal smoking are consistent with the literature (see below) and suggests a compensatory upregulation of ¹²⁵I-epibatidine-labeled nAChRs, possibly in response to a decrease in cholinergic activity at these sites (66, 67). Further studies on other cholinergic markers in this dataset are necessary to address the compensatory relationship between cholinergic markers and nAChRs in our dataset. It is of interest that the significant increase in ¹²⁵I-epibatidine binding, seen predominately in controls, occurred in three of the five medullary nuclei that are significantly decreasing with age. This represents complex and dynamic neuroplastic changes associated with both development and exposure. The fact that the effect of exposure in the SIDS infants was not statistically significant at these sites suggests a lack of compensatory neuroplasticity within these sites in SIDS. Whether the normal mechanisms shown to be involved in upregulation of nAChRs [post-translational receptor assembly (68), trafficking (68), cell surface expression (68), degradation (68), and affinity alterations (69); post-transcriptional modification by microRNAs (70)] are deficient in SIDS is unknown; however, the potential consequences of incomplete compensation are noteworthy.

Given that acetylcholine modulates serotonergic activity in regulation of cardiorespiratory and homeostatic functions (71–74), an imbalance of acetylcholine transmission (via abnormal nicotinic and/or muscarinic receptors) in SIDS infants likely puts them at risk for sudden death in response to homeostatic challenges. As noted, embedded within the GC, PGCL, and RO are 5-HT neurons. Our data showing an effect of pre-natal smoking in these nuclei support literature reporting an effect of pre-natal nicotine on 5-HT neuron neurotransmission in general (67, 75–77) and directly on medullary 5-HT neurons (78, 79). Numerous experimental and human studies [reviewed in Vivekanandarajah et al. (36)] have shown that maternal cigarette smoke/nicotine exposure adversely affects ventilation (80), breathing drive (81), respiration rate (82–84), ventilatory drive (85, 86), respiratory rhythm pattern generation (87), and arousal responses to adverse stimuli such as hypoxia and hypercapnia (88–91)—responses mediated in part by medullary 5-HT neurons within the RO, GC, and PGCL. It is important to note that maternal smoking exposes a fetus to more than just nicotine, including toxins that have been shown to have unique neurodevelopmental effects and/or combine with nicotine to exacerbate nicotine's effect (66, 67). Given this, our findings are likely not attributed to nicotine alone.

Our data on the amount of pre-natal smoke exposure positively associated with post-natal nicotinic receptor binding are in agreement with the experimental animal models studying the brainstem. Animal models of chronic pre-natal exposure have generally shown increased nAChR expression in the infant brainstem (92–96). Maternal nicotine exposure during pregnancy have resulted in increased nAChR binding in mouse (97) and rhesus monkey (93), increased nicotinic receptor mRNA in the rat (95), and increased nAChR protein expression in the mouse (98) at various brainstem sites. Slotkin et al. also reported an overall increase in ¹²⁵I α-bungarotoxin binding in the brainstem that emerged in the early post-natal period (99). Studies of human infant pre-natal exposures demonstrate conflicting results. Our laboratory previously reported an increase in ³H-nicotine binding in mesopontine nuclei related to cardiorespiration (nucleus parabrachialis) and arousal (LC and PoO) (19) and a decrease in binding in the LC nucleus, periaqueductal gray, and the raphe dorsalis, part of the ascending 5-HT arousal system (21). Other groups have reported decreased α7, β2 nAChR protein expression in the arcuate nucleus, XII, and NTS (22), decreased α4 nAChR fiber staining in the cribriform nucleus (24) in smoke-exposed infants, and increased α7 nAChR protein expression in major brainstem sites associated with pre-natal smoke exposure (100). The differences in the direction of change may be attributable to the composition of the control groups across the datasets with differing cohorts of infants who die from varying causes, the patterns of smoking, and/or the experimental measures, including the use of different ligands and antibodies. Irrespective of direction (up- or downregulation due to exposure), the specificity of change only in KCOD controls is consistent with our other studies showing alterations in controls but not in SIDS. Furthermore, these differences provide an example of the complexity of human autopsy studies where, without measurable cotinine levels, it is unclear whether the differences are due to acute changes in circulating nicotine close to the time of death or due to early or sustained developmental effects of nicotine exposure in utero.

Interestingly, there was no effect of the amount of maternal alcohol use on ¹²⁵I-epibatidine binding in the brainstem. This differs from the Aberdeen Area Infant Mortality Study (AAIMS) showing a significant reduction in nicotinic receptor binding with an increase in the average number of drinks per month during pregnancy (21). It also differs from experimental studies, both in vivo (101–103) and in vitro (104), that have shown changes in nicotinic receptor expression following alcohol exposure. It is possible that our continuous measure of drinks per pregnancy is not discerning enough to detect small effects or effects that are trimester specific. It is also possible that the size of the cohort does not have the statistical power to detect effects of alcohol on ¹²⁵I-epibatidine. With these caveats, however, our data do support that, in this cohort, pre-natal smoke exposure plays a greater role in altering nicotinic receptor binding than pre-natal alcohol exposure. Although the results of the present study show that amount of pre-natal alcohol exposure does not alter brainstem nicotinic receptors in infants, further studies are warranted to investigate the effect on other neurotransmitter systems, such as the serotonergic system.

A major limitation of this study is the relatively small sample size. The validity of results is supported, however, by their general consistency with previously reported data in other cohorts and animal models. A second limitation of this study is the fact that all SIDS, nearly all post-KCOD controls (90%), and all pre-KCOD controls were exposed to pre-natal smoking to some degree, thus making it difficult to detect differences in binding between SIDS and post-KCOD controls controlling for exposure. Despite this, the continuous values measuring pre-natal exposure to smoking enabled us to determine the association between the amount of exposure and nicotinic receptor binding and the developmental trajectory of nicotinic receptor binding with age. A third limitation is that the autopsied control infants were not representative of living controls. This limitation is not specific to the Safe Passage Study but is inherent in all autopsy case/control studies. Another limitation of the study is that there was no experimental verification of the biological concentration of pre-natal exposure in the infant's system. The pre-natal exposures related to smoking and alcohol was based on the mother's self-report of her consumption without verification with biomarkers such as cotinine measurements for smoke exposure and ethanol metabolites for alcohol exposure in infant blood. Finally, the selected radioligand ¹²⁵I-epibatidine binds with most, but not optimally, or with all nicotinic receptor subtypes. Additional analysis of and comparison between ¹²⁵I-bungarotoxin and ³H-nicotine would allow for a more comprehensive analysis of nAChRs in the brainstem (114, 115).