Multiple Serotonergic Brainstem Abnormalities in Sudden Infant Death Syndrome

Context The serotonergic (5-hydroxytryptamine [5-HT]) neurons in the medulla oblongata project extensively to autonomic and respiratory nuclei in the brainstem and spinal cord and help regulate homeostatic function. Previously, abnormalities in 5-HT receptor binding in the medullae of infants dying from sudden infant death syndrome (SIDS) were identified, suggesting that medullary 5-HT dysfunction may be responsible for a subset of SIDS cases.

Objective To investigate cellular defects associated with altered 5-HT receptor binding in the 5-HT pathways of the medulla in SIDS cases.

Design, Setting, and Participants Frozen medullae from infants dying from SIDS (cases) or from causes other than SIDS (controls) were obtained from the San Diego Medical Examiner's office between 1997 and 2005. Markers of 5-HT function were compared between SIDS cases and controls, adjusted for postconceptional age and postmortem interval. The number of samples available for each analysis ranged from 16 to 31 for SIDS cases and 6 to 10 for controls. An exploratory analysis of the correlation between markers and 6 recognized risk factors for SIDS was performed.

Main Outcome Measures 5-HT neuron count and density, 5-HT_1A receptor binding density, and 5-HT transporter (5-HTT) binding density in the medullary 5-HT system; correlation between these markers and 6 recognized risk factors for SIDS.

Results Compared with controls, SIDS cases had a significantly higher 5-HT neuron count (mean [SD], 148.04 [51.96] vs 72.56 [52.36] cells, respectively; P<.001) and 5-HT neuron density (P<.001), as well as a significantly lower density of 5-HT_1A receptor binding sites (P≤.01 for all 9 nuclei) in regions of the medulla involved in homeostatic function. The ratio of 5-HTT binding density to 5-HT neuron count in the medulla was significantly lower in SIDS cases compared with controls (mean [SD], 0.70 [0.33] vs 1.93 [1.25] fmol/mg, respectively; P = .001). Male SIDS cases had significantly lower 5-HT_1A binding density in the raphé obscurus compared with female cases (mean [SD], 16.2 [2.0] vs 29.6 [16.5] fmol/mg, respectively; P = .04) or with male and female controls combined (mean [SD], 53.9 [19.8] fmol/mg; P = .005). No association was found between 5-HT neuron count or density, 5-HT_1A receptor binding density, or 5-HTT receptor binding density and other risk factors.

Conclusions Medullary 5-HT pathology in SIDS is more extensive than previously delineated, potentially including abnormal 5-HT neuron firing, synthesis, release, and clearance. This study also provides preliminary neurochemical evidence that may help explain the increased vulnerability of boys to SIDS.

Sudden infant death syndrome (SIDS) is the leading cause of postneonatal infant mortality in the United States, with an overall incidence of 0.67/1000 live births.^1-3 Despite intensive research, the causes of SIDS remain unknown. Moreover, controversies abound about the role of certain practices, eg, bed sharing^4,5 or use of pacifiers,^5-7 in SIDS, in large part due to the lack of understanding of the basic biological mechanisms. We have proposed the triple risk model,⁸ which suggests that sudden death results when 3 factors impinge on the infant simultaneously: (1) an underlying vulnerability; (2) an exogenous stressor (eg, prone sleep position, bed sharing); and (3) the critical developmental period, ie, the first 6 months of postnatal life, when the infant is at greatest risk for SIDS.⁸

The serotonergic (5-hydroxytryptamine [5-HT]) system of the medulla oblongata consists of 5-HT neurons located in the midline raphé, lateral extraraphé, and ventral surface and helps regulate autonomic and respiratory function.¹ These medullary nuclei are interconnected⁹ and project extensively to nuclei in the brainstem and spinal cord that influence respiratory drive,¹⁰ blood pressure regulation,¹¹ thermoregulation,¹² upper airway reflexes, and arousal.^13-15 Medullary 5-HT neurons have also been proposed to be central respiratory chemosensors.^16,17 Moreover, they are involved in the induction of long-term facilitation of respiration in response to episodic hypoxia¹⁸ and play a critical role in the generation of respiratory rhythm in vivo.^19,20

Previously, we identified altered 5-HT receptor binding density in components of the medullary 5-HT system in SIDS cases in 2 independent data sets using a nonselective radioligand that binds to 5-HT_1A-1D and 5-HT₂ receptors.^21,22 Additionally, we reported a case of an infant with SIDS who displayed altered autonomic and respiratory function at birth and 5-HT receptor binding abnormalities at autopsy 2 weeks later.^1,23 Recently, polymorphisms in the promoter region (5HTTLPR)^24,25 and in intron 2²⁶ of SLC6A4, the gene for the 5-HT transporter (5-HTT), were reported in higher frequencies in populations with SIDS compared with controls. Taken together, these observations support the idea that medullary 5-HT dysfunction results in a failure of autonomic and respiratory responses to hypoxia or hypercapnia and in sudden death for at least a subset of SIDS cases.

The extent, nature, and pathogenesis of this dysfunction, however, remain to be determined. The subtype(s) (potentially up to 7) of 5-HT receptor affected and their cellular localization are unknown. It is unknown if 5-HT neuron count and the expression of 5-HTT, important markers of 5-HT function, are also altered. Moreover, the level of available 5-HT and its relationship to alterations in 5-HT receptor binding observed in SIDS is unknown. Determination of the expression and distribution of these markers of 5-HT function are necessary to fully characterize the nature and pathogenesis of 5-HT dysfunction in SIDS cases.

In this study, we determined 5-HT neuron count and density, 5-HT_1A receptor binding density, and 5-HTT binding density in SIDS cases compared with controls in an effort to examine in greater detail the cellular components of the 5-HT pathway and to gain greater insight into the extent and pathogenesis of the 5-HT abnormalities. We analyzed the 5-HT_1A receptor because it is found in high densities in regions in which binding was most severely reduced in previous studies, is recognized as a somatodendritic autoreceptor that controls 5-HT neuron firing, and plays important roles in cardiorespiratory control^12,27-30 and neural development.^31,32 We analyzed 5-HTT because it regulates synaptic 5-HT concentration³³ and because of the identification of SLC6A4 gene polymorphisms as risk factors for SIDS.^24-26 In addition, all SIDS cases and controls in this study were genotyped for the 5HTTLPR polymorphism to allow correlation of genotype with data on neuron count and binding density.³⁴ We also explored the potential effects of the recognized SIDS risk factors of sex, sleep position, bed sharing, history of illness within 1 week of death, and prematurity on 5-HT neuron count, 5-HT_1A receptor binding density, and 5-HTT binding density to determine their potential role in the pathogenesis of SIDS.

Frozen medullae from a total of 31 infants dying from SIDS (cases) and 10 infants with acute death from causes other than SIDS (controls) were obtained from the Office of the Chief Medical Examiner, San Diego, Calif, between 1997 and 2005, representing a new data set that has not previously been published. These infants represent all infant autopsies with a postmortem interval less than 27 hours for whom a study technician was available and the brainstem collected and, for controls, for whom the death was not under investigation. All SIDS cases were diagnosed by one of the authors (H.F.K.), an expert in SIDS pathology, in conjunction with the San Diego Medical Examiner's office. The cause of death of the controls was determined at autopsy by the San Diego Medical Examiner's office, and included acute deaths without hospitalization (drowning, n = 2; asphyxia secondary to a plastic bag, n = 1; pneumonia with acute respiratory distress, n = 2; group B streptococcal sepsis, n = 1; unsuspected congenital heart disease, n = 1) and 3 hospitalized deaths (suspected inborn error of metabolism, n = 1; congenital heart disease, n = 2).

The number of recognized SIDS risk factors in each SIDS case were divided into 2 different categories: (1) “abnormality” risks, ie, factors that might increase the probability of an infant having an underlying vulnerability (ie, the medullary 5-HT abnormality) as a result of an underlying genetic predisposition or adverse prenatal exposure; and (2) “stressors,” ie, environmental or physical factors that impinge on the vulnerable infant during the critical postnatal period, possibly challenging homeostatic function.⁵ The number of risk factors in each category and the total number of risk factors was determined for each SIDS case by review of the autopsy report. The race/ethnicity (defined by the parents) of each infant in the study was recorded because the SIDS rate differs significantly between different racial/ethnic populations and may influence observations.

Appropriate medullary tissue from all infants was not available for all portions of the study. For the 5-HT neuron counting portion, 16 SIDS cases and 7 controls were analyzed; for 5-HT_1A receptor binding, 16 SIDS cases and 6 controls were analyzed; and for 5-HTT binding, 30 SIDS cases and 7 controls were analyzed. All 31 SIDS cases and 10 controls were genotyped for the 5HTTLPR polymorphism due to the availability of nonmedullary tissue from which DNA could be extracted. All analyses were performed in a blinded fashion, with the investigator unaware of the case or control status of each infant.

This study was approved by the Committee on Clinical Investigation at Children's Hospital Boston. All tissue was obtained under the auspices of the San Diego Medical Examiner system in accordance with California law.³⁵ Under this statute, it is not required to obtain informed consent of individual parents for research of sudden and unexpected infant death.

Immunocytochemical testing was performed for tryptophan hydroxylase on 20-μm frozen sections of medulla using the PH8 antibody (Chemicon International, Temecula, Calif) according to a previously described protocol.³⁶ Sections were postfixed in 4% paraformaldehyde before being incubated in PH8 antibody (1:8000) overnight at 4°C. Staining was then developed by addition of di-amino benzamide substrate (Dako, Glostrup, Denmark) to the section before coverslipping. Tryptophan hydroxylase neurons were counted at 2 standardized levels of the mid- and rostral medulla by 1 examiner using computer-based methods with Neurolucida version 6.02.2 (Microbrightfield Inc, Williston, Vt). Medullary levels were determined by reference to the brainstem atlas of Olszewski and Baxter.³⁷ The mid-medulla level corresponds to Plate XII, and the rostral medulla level corresponds to Plate XIV in the atlas.³⁷ The perimeter of each section was traced (× 2) and the distribution of immunoreactive cells within the 3 medullary regions marked using different graphic symbols and colors (× 10). Immunolabeled cell bodies were counted only if they were morphologically identifiable as neurons. Immunopositive cells were identified as belonging to 1 of 5 morphological cell types: granular, fusiform, pyramidal, multipolar, or undetermined. All sections were counted twice and the mean value used for analysis.

The autoradiography procedures for determination of ³H-8-OH-DPAT (³H8-hydroxy-2-[di-N-propylamino]-tetralin) binding to 5-HT_1A receptors and ¹²⁵I-RTI-55 (3 beta-[4-iodophenyl]tropan-2 beta-carboxylic acid ester labeled with sodium iodide I 125) binding to 5-HTT were performed according to previously described protocols³⁸ on 20-μm sections of frozen medulla adjacent to those stained for tryptophan hydroxylase immunoreactivity. Total 5-HT_1A receptor binding was determined by incubation of tissue sections in 4-nM ³H-8-OH-DPAT (PerkinElmer Inc, Wellesley, Mass) for 60 minutes at room temperature. Nonspecific binding was determined by addition of 10-μM serotonin to the solution. Total 5-HTT binding was determined by incubation of tissue sections in 0.15-nM ¹²⁵I-RTI-55 (PerkinElmer) for 90 minutes at room temperature. Nonspecific binding was determined by the addition of 100-nM citalopram hydrobromide (Tocris, St Louis, Mo). Sections were then placed in cassettes and exposed to ³H-sensitive film (Kodak Biomax MR; Eastman Kodak Co, Rochester, NY) for 12 weeks or to a BAS-TR2025 phosphoimaging plate (FujiFilm Corp, Tokyo, Japan) for 4 weeks, with a set of ³H standards (Amersham, Buckinghamshire, England) for determination of ³H-8-OH-DPAT binding, or to film for 5 hours with a set of ¹²⁵I standards (Amersham) for determination of ¹²⁵I-RTI-55 binding. Film autoradiograms were generated according to standard laboratory procedure for development of light-sensitive film. A BAS-5000 Bioimaging Analyzer (FujiFilm) with Image Reader version 1.8 software (FujiFilm) was used to generate digital autoradiographic images from phosphoimaging plates. For each specimen, receptor or transporter binding density was analyzed in 9 medullary nuclei (all nuclei were not available in all cases) at a defined level of the brainstem (2 autoradiograms for each nucleus) according to previously published methods.^9,22,38 Quantitative densitometry of autoradiograms was performed using an MCID 5+ imaging system (Imaging Research Inc, St Catharines, Ontario).

DNA was isolated from 50 to 100 mg of brain tissue using standard methods and a Puregene reagent kit (Gentra Systems, Minneapolis, Minn) according to the manufacturer's instructions. DNA samples were saved in Tris-EDTA hydration buffer at −20°C prior to genotyping. Polymerase chain reaction (PCR) amplification was conducted in a final 20-μL volume consisting of 80 to 100 ng of genomic DNA, using an AGS Gold PCR kit (Thermo-Hybaid, Franklin, Mass) following the protocol described by Narita et al.²⁴ PCR products were visualized by 1% agarose gel electrophoresis with ethidium bromide staining. DNA bands were identified under UV light, and genotype assignment for each case was determined by correlation with a standard 50–base pair DNA ladder (Invitrogen Corp, Carlsbad, Calif).

We compared 5-HT neuron count, 5-HT_1A receptor binding density, and 5-HTT binding density in the raphé obscurus between the SIDS infants who were (1) male (n = 15) or female (n = 16); (2) born prematurely (<37 gestational weeks at birth) (n = 11) or at term (n = 20); (3) found dead in the prone (n = 15), side (n = 5), or supine (n = 5) position; (4) found dead lying face down (n = 9) or face up (n = 10); (5) bed sharing (n = 7) or not bed sharing (n = 24) on the night of death; and (6) sick with minor illness within 1 week of death (n = 13) or not ill (n = 18). This analysis was performed to determine if different subsets of SIDS cases are characterized by differences in 5-HT neuron count, 5-HT_1A receptor binding, and/or 5-HTT binding sites. Sufficient clinical data were not available for similar analysis concerning maternal socioeconomic class or cigarette smoking, drinking, or use of illicit drugs during pregnancy. Similarly, small sample size prohibited analysis of race.

t Tests were used to compare the number and density of neurons in the raphé, extraraphé, and ventral regions, as well as the combined cell count by level in SIDS cases and controls. t Tests were also used to compare the total section area and the proportion of each type of cell, and paired t tests were used to compare cell counts between levels. The density of 5-HT_1A receptor and 5-HTT binding sites in medullary nuclei in SIDS cases and controls were compared using the Wilcoxon rank sum test. Analysis of covariance was used in all 3 marker studies to control for the effects of postconceptional age and postmortem interval on neuron counts and binding density, as well as to consider the ratio of 5-HTT binding density to 5-HT neuron count in the raphé. A t test was used to compare postconceptional age between SIDS cases and controls. The effects of SIDS risk factors on neuron count and binding in SIDS cases were analyzed using t tests. Adjustment for multiple testing was not performed due to small sample size. Dose-response regressions of genotype (indicator variables for at least one “l” allele and the “ll” genotype) on neuron count or binding density were performed to test for potential effects of 5HTTLPR genotype on 5-HT neuron count, 5-HT_1A receptor binding, and 5-HTT binding. In all analyses, P<.05 was considered statistically significant.

Available clinicopathologic data for the 31 SIDS cases and 10 controls are presented in Table 1. The age for the SIDS cases ranged from 4 to 36 postnatal weeks (mean [SD], 16.4 [10.6] weeks), and in controls from 1 day to 52 postnatal weeks (12.2 [17.0] weeks) (P = .48). The postmortem interval was less than 27 hours in all infants and was significantly longer in SIDS cases compared with controls (mean [SD], 19.2 [5.1] hours vs 13.3 [7.5] hours, respectively; P = .007). Postmortem interval did not significantly affect 5-HT_1A receptor or 5-HTT binding density, but 5-HT neuron count decreased significantly with postmortem interval. All of the data were corrected for the effects of postmortem interval. There was no statistically significant difference between the acute and hospitalized controls in 5-HT neuron count, 5-HT_1A receptor binding density, or 5-HTT binding density in the raphé obscurus (P>.20 for all), and thus the values from acute and hospitalized controls were combined for comparison with the SIDS cases. There was no histological evidence of brainstem pathology in any of the controls.

In the rostral medulla, the number of 5-HT neurons in the midline raphé, lateral extraraphé, and at the ventral surface was significantly higher in SIDS cases compared with controls. In the mid-medulla, the 5-HT neuron count combined for all subregions and the 5-HT neuron count in the midline raphé were significantly higher in SIDS cases compared with controls (Figure 1 and Table 2). The mean (SD) cell density of 5-HT neurons in the raphé, extraraphé, and ventral surface regions combined were significantly higher in SIDS cases vs controls in the rostral medulla (0.81 [0.40] vs 0.54 [0.26] cells/mm², respectively; P<.001) and in the mid-medulla (0.55 [0.19] vs 0.41 [0.24] cells/mm², respectively; P = .003).

Granular neurons represented a significantly greater proportion of all 5-HT cells in SIDS cases compared with controls in the rostral medulla (mean [SD] ratio of granular neurons to combined 5-HT neurons, 0.26 [0.05] vs 0.19 [0.07], respectively; P = .04). Multipolar cells represented a significantly smaller proportion of all 5-HT neurons counted in SIDS cases compared with controls in the mid-medulla (mean [SD] ratio of multipolar neurons to combined 5-HT neurons, 0.02 [0.01] vs 0.04 [0.02]; P = .03).

5-HT_1A receptor binding density was significantly reduced in SIDS cases compared with controls in all nuclei analyzed, with the exception of the principal inferior olivary nucleus (Figure 2 and Table 2).

No significant differences in absolute 5-HTT binding between SIDS cases and controls were observed in any of the nuclei analyzed (Table 2). The ratio of 5-HTT binding density to 5-HT neuron count in the raphé obscurus, however, was significantly lower in SIDS cases compared with controls (Table 2), suggesting a relative reduction of 5-HTT expression per 5-HT neuron count.

The 31 SIDS cases in the data set comprised 15 male (48%) and 16 female (52%) infants of different races and ethnicities. Eleven (35%) were born prematurely. Fifteen SIDS cases (48%) were found in the prone sleeping position, 9 (29%) were found face down, 7 (23%) were bed sharing at the time of death, and 13 (42%) had a history of illness in the week preceding death. All 31 SIDS cases were exposed to at least 1 abnormality risk factor or stressor. Thirty (97%) were subject to at least 1 abnormality risk factor, 28 (90%) were exposed to at least 1 stressor at the time of death, and 27 (87%) were subject to at least 1 factor from both categories. The mean (SD) 5-HT_1A receptor binding density in the raphé obscurus was significantly lower in male (n = 6) compared with female (n = 10) SIDS cases (16.2 [4.8] vs 29.6 [16.5] fmol/mg, respectively; P = .04) and significantly higher in controls (53.9 [19.8] fmol/mg; P = .005) compared with male SIDS cases (P = .02) and female SIDS cases (P = .05). No significant difference in 5-HT neuron count or 5-HTT binding density, however, was observed between male and female SIDS cases. Similarly, no associations were observed for other risk factors.

Genotyping of the 31 SIDS cases revealed 6 of SS genotype, 18 of SL genotype, and 7 of LL genotype. Of the 10 controls, none were of the SS genotype, 7 were of the SL genotype, and 3 were of the LL genotype. The frequency distribution of these different genotypes was not significantly different between the SIDS cases and controls (P = .40). Regression analysis revealed no statistically significant relationship or consistent trends between 5HTTLPR genotype and 5-HT neuron count, 5-HT_1A receptor binding density, or 5-HTT binding density in the raphé obscurus in SIDS cases, or in the combined data set of both SIDS cases and controls.

We found that the medullary 5-HT abnormalities in SIDS are more extensive than previously suggested and that they involve multiple elements of 5-HT function, including 5-HT neuron count, 5-HT_1A receptor expression, and relative 5-HTT binding in the same cases. This study strengthens the hypothesis that medullary 5-HT dysfunction is associated with SIDS and may lead to death by a failure of respiratory and autonomic responses to homeostatic stressors during sleep. This study also found, in an explanatory analysis, reduced 5-HT_1A receptor binding density in male compared with female SIDS cases, an observation that may help explain why males are more vulnerable to SIDS.⁵ These 5-HT abnormalities were documented during the era of stringent public messages on risk reduction, including that for supine sleeping position. The majority (65%) of the SIDS cases in this data set, however, were sleeping prone or on their side at the time of death, indicating the need for continued public health messages on safe sleeping practices.

The increased number of 5-HT neurons in medullary sites in SIDS cases, coupled with the reduction in 5-HT_1A receptor binding and relative reduction in 5-HTT binding in these sites, suggest that the synthesis and availability of 5-HT (and by extrapolation, neuron firing) is altered within 5-HT pathways. It is unclear, however, specifically how these functions are altered. It is possible that an increased number of 5-HT neurons may lead to an excess of extracellular 5-HT and a compensatory down-regulation of 5-HT_1A receptors. Alternatively, 5-HT synthesis, release, or both may be dysfunctional in the 5-HT neurons (which are overabundant in compensation), resulting in a deficiency of extracellular 5-HT. The level of available 5-HT in the medullae of SIDS cases has yet to be determined, and the molecular and cellular regulatory mechanisms between neuron count and receptor and transporter expression are incompletely understood. It is difficult, therefore, to predict with any certainty which, if either, of these proposed conditions exist in SIDS. Determination of medullary 5-HT level in SIDS and the elucidation of the pathways and mechanisms regulating the expression of the 5-HT markers analyzed in this study will therefore be necessary before the nature and pathogenesis of the medullary 5-HT dysfunction in SIDS can be determined. The increased number of 5-HT neurons, the greater proportion of neurons of “simple” (ie, granular) morphology, and the decreased number of neurons of “complex” (ie, multipolar) morphology in SIDS cases compared with controls, however, supports the concept of an underlying developmental disorder involving abnormal regulation of 5-HT neuron count and delayed neuronal differentiation and maturation. This idea is supported by the relative reduction in 5-HTT expression observed in the medullae of SIDS cases. The 5-HTT is expressed predominantly in peri-synaptic sites on 5-HT neuron terminals. Reduced 5-HTT expression may be due to reduced expression of 5-HTT protein at 5-HT neuron terminals, a reduced number of 5-HT terminals and synapses, or both, consistent with abnormal or delayed 5-HT synapse formation and neuron development. Of note, 5-HT neuron migration appears to be relatively unaffected in SIDS cases, as we observed 5-HT neurons in the same anatomical positions in the component nuclei of the medullary 5-HT system as in the controls.

In this study, we described known risk factors for SIDS either as “abnormality” risk factors or as “stressors” in an effort to distinguish between the 2 types of risks we propose are involved in the pathogenesis of SIDS. Eighty-seven percent of the SIDS cases in this study were observed to be both at risk from having medullary 5-HT abnormalities and exposed to an exogenous stressor at the time of death. These data support the triple risk model.⁸ The prone or side sleeping position, face-down sleeping, and bed sharing are recognized as important risk factors or stressors for SIDS.^5,39 We identified that 77% of the SIDS cases in this study slept prone or on their side, shared a bed, or both, indicating that these sleep practices remain major risk factors for SIDS. We found that abnormalities of 5-HT markers, specifically in 5-HT_1A receptor binding, occurred in the medullae of SIDS cases regardless of sleeping position or whether they were bed sharing at the time of death. These observations suggest that the SIDS cases within this study shared a common underlying vulnerability, ie, an intrinsic medullary 5-HT abnormality. The increased risk of SIDS in the prone or face-down position may reflect the infants' inability to respond to the asphyxial or hypercarbic challenge in the face-down position, due to the abnormalities in the medullary 5-HT system that compromise protective reflexes, including arousal and head turning.

The identification of a significantly lower density of 5-HT_1A receptor binding in male compared with female SIDS cases in this study provides neurochemical evidence that may help explain the increased risk of SIDS in males (with a 2:1 ratio) compared with females.⁵ A recent study involving chemical ablation of approximately 60% of medullary 5-HT neurons in neonatal piglets reported that males exhibited a blunted response to inspired carbon dioxide during sleep, whereas females responded normally.⁴⁰ Of note, this carbon dioxide response is modulated in part by 5-HT_1A receptors in the medullary raphé.¹⁷ This experimental finding raises the possibility that in humans male infants may similarly be less sensitive to carbon dioxide than female infants, and that loss of medullary 5-HT_1A receptors, as observed in the SIDS cases in this study, may attenuate respiratory responses to hypercapnia to a greater extent in male compared with female infants, thus placing them at greater risk for SIDS.

Interestingly, experimental evidence indicates that 5-HT_1A receptor expression in the forebrain is normally significantly lower in the human male brain compared with the human female brain.^41-43 Deficits in postnatal brain levels of 5-HT_1A receptor expression following in utero cocaine exposure persist for a greater length of time in male compared with female rats,⁴⁴ suggesting that the neonatal male infant brain is less resilient to exposure to at least some pharmacologically active toxins affecting 5-HT function in the maternal circulation than the neonatal female brain. Taking these observations together, intrinsic differences between male and female infants in baseline brain 5-HT_1A receptor expression, 5-HT neuronal plasticity, and carbon dioxide sensitivity during sleep (which may be modulated by medullary 5-HT_1A receptors in humans) provide evidence that may explain, at least in part, the greater risk of SIDS in male infants. Given that this observation was based on a sample size of 6 males and 10 females with a P value of .04 in an exploratory multiple testing environment, it is imperative to repeat it in an independent data set.

We found no significant difference in 5HTTLPR genotype frequency between SIDS cases and controls and no relationship between genotype frequency and 5-HT neuron count, 5-HT_1A receptor binding density, or 5-HTT binding density in SIDS cases or controls. We caution, however, that the sample size of SIDS cases and controls was small and may be insufficient to establish a link between SIDS, genotype, and 5-HT brainstem abnormalities. Interpretation of these data is also complicated by the multiple race/ethnicity categories in the data set, because the frequency of the 5HTTLPR genotype varies between racial and ethnic populations. It is also noteworthy that none of the controls were of the “SS” genotype. This is most likely a consequence of the small number of controls, but it is also possible that it may reflect a genetic bias in the control population in this study.

A potential limitation of this study is the relatively small number of SIDS cases and controls in the data set due to the difficulty in accruing specimens over a reasonable time frame. Controls are particularly difficult to obtain due to rarity of acute deaths in the age group from causes other than SIDS. Due to the tissue limitations, we were not able to examine all three 5-HT markers in all cases. Despite the relatively small sample size, however, we found highly significant differences, both qualitative and quantitative, between the SIDS cases and controls in more than one 5-HT marker. Moreover, the finding of reduced 5-HT_1A receptor binding replicates the finding of 5-HT receptor defects in SIDS cases in the previous data sets.^21-23 The small sample size is especially problematic for the subset (risk-factor) analysis, in which multiple statistical comparisons were performed. Thus, while we observed a significant difference in 5-HT_1A receptor binding between male and female SIDS cases, this observation should be regarded as preliminary and will need to be confirmed in a larger data set. Other limiting factors that should be taken into consideration include the (statistically nonsignificant) difference in age between SIDS cases and controls; race/ethnicity; and exposure of the fetus to potentially harmful substances (eg, nicotine, alcohol) during gestation. However, we found that age was not a significant influence in any of our analyses, while the small sample size and the unavailability of data on maternal smoking and alcohol use for the majority of cases prevented us from assessing the effect of race/ethnicity and prenatal exposures, respectively, on our data. Future analysis of a larger data set, with complete information on all potential risk factors, is therefore needed.

Multiple, interrelated markers of 5-HT function are abnormally expressed in the medulla in the same SIDS cases, suggesting that basic elements of 5-HT neurotransmission, including neuron firing, synthesis, release, and clearance of 5-HT, are dysfunctional in SIDS. The finding that 5-HT neuron count is increased in the SIDS cases associated with increased morphologic immaturity suggests the possibility that the 5-HT abnormalities are developmental in origin. This study provides biological plausibility for certain risk-reduction strategies in SIDS, as well as for the triple risk model for SIDS. Moreover, it generates new hypotheses for testing about 5-HT–related brainstem pathology underlying sudden death in early life in future SIDS autopsies and in experimental mechanistic models.

Corresponding Author: David S. Paterson, PhD, 300 Longwood Ave, Enders Bldg Room 1109, Boston, MA 02115 (david.paterson@childrens.harvard.edu).

Author Contributions: Dr Paterson had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Paterson, Belliveau, Beggs, Kinney.

Acquisition of data: Paterson, Thompson, Belliveau, Darnall, Chadwick, Krous, Kinney.

Analysis and interpretation of data: Paterson, Trachtenberg, Thompson, Belliveau, Darnall, Kinney.

Drafting of the manuscript: Paterson, Kinney.

Critical revision of the manuscript for important intellectual content: Paterson, Trachtenberg, Thompson, Belliveau, Beggs, Darnall, Chadwick, Krous, Kinney.

Statistical analysis: Trachtenberg.

Obtained funding: Paterson, Kinney.

Administrative, technical, or material support: Thompson, Belliveau, Darnall, Chadwick, Kinney.

Study supervision: Paterson, Krous, Kinney.

Funding/Support: This work was supported by grants from the Scottish Cot Death Trust, Evelyn Deborah Barrett Fellowship, CJ Foundation for SIDS, First Candle/SIDS Alliance, CJ Murphy Foundation, National Institute of Child Health and Human Development (R37-HD20991 and PO1-HD36379), and Children's Hospital Mental Retardation Core Grant (P30-HD18655).

Role of the Sponsors: None of funding organizations had any role in the design and conduct of the study; the collection, management, analysis, and interpretation of the data; or the preparation, review, and approval of the manuscript.

Acknowledgment: We thank the San Diego Medical Examiner system for help in the accrual of tissue specimens.