Main

SIDS deaths occur more often during wintertime than during the summer(1–3), and the frequency is also higher during periods with epidemics of upper respiratory infection(4). Approximately half of SIDS cases have had signs of an upper respiratory infection before death(1,5). Many of them also show mucosal immune stimulation in the lungs, trachea(6,7), duodenum(7), and salivary glands(8). However, to our knowledge, the larynx and epiglottis have not been systematically studied in SIDS.

We have previously shown that SIDS cases have elevated levels of IL-6 in CSF(9) and postulated a possible relationship between such central immune response indicated by CSF IL-6 levels and peripheral immune stimulation(9).

In the present study the larynx was chosen because it is an organ that is important with regard to obstructive apneas and the laryngeal reflex(10). Moreover, a respiratory tract infection is quite common in SIDS babies. Several studies have shown that laryngeal stimulation affects respiration(11,12), and Lindgren and Grøgaard(13) have shown that laryngeal stimulation in RSV-infected infants results in disturbances in the regulation of breathing.

Our hypothesis was therefore that immune reactions in the laryngeal mucosa may induce a central immune response with elevated levels of IL-6 in SIDS.

We thus wanted to examine whether there is a relationship between central immune response in SIDS, indicated by increased CSF IL-6 levels, and peripheral immune stimulation as expressed by the presence of IgA, IgG, and IgM immunocytes and T lymphocytes (CD3) in the laryngeal mucosa and HLA-DR expression in the surface epithelium and in the glands of the wall.

METHODS

Subjects. Seventeen SIDS cases with low CSF IL-6 levels (≤5 pg/mL) and 20 SIDS cases with high CSF IL-6 levels (≥30 pg/mL) were studied. IL-6 concentrations in CSF had been measured by ELISA technique (Biotrak Amersham). Fewer than half of the results of these measurements have been published elsewhere(9,14). All SIDS cases originated from the Institute of Forensic Medicine in the period July 1992 through April 1997. In this period, 70 SIDS cases were examined. However, 13 cases had to be excluded because of a postmortal time >48 h, and in 11 of the cases appropriate tissue samples or CSF was not available. The rationale for selecting the SIDS cases exclusively on the basis of CSF IL-6 levels was to compare the immune response in cases with obvious IL-6 elevation with cases with low normal values to detect any differences between these two groups. Five girls and four boys with a similar age range as those investigated were thus excluded. Forty-four percent of all the SIDS cases measured had clearly elevated IL-6 levels >30 pg/mL, whereas 37% had low normal levels <5 pg/mL.

There were 12 girls and 25 boys in the present material. The median age was 3.5 mo (range 1 wk to 19 mo). Further clinicopathologic information about the subjects is given in Tables 1 and 2. All information was obtained from medical records and police reports.

Table 1 Clinicopathologic information on SIDS cases with CSF IL-6 ≥30 pg/mL
Full size table
Table 2 Clinicopathologic information on SIDS cases with CSF IL-6 ≤5 pg/mL
Full size table

Microbiologic examination. In all the cases but one, a nasopharyngeal aspiration was performed during autopsy, and smears were made. After air-drying and fixation in acetone for 10 min, the slides were kept at -70° before staining. After applying FITC-conjugated MAb to influenza A and B viruses, a mixture of three MAb specific to parainfluenza virus types 1, 2, and 3, and MAb to RSV and adenovirus, the specimens were examined by a blind trial for respiratory viruses by a direct immunofluorescence technique. All antibodies were supplied by Dako, Cambridgeshire, United Kingdom. Blood, CSF, and tissue from lung, liver, spleen, and kidney were examined for bacterial growth. Culture specimens from bronchial tree secretions were obtained when present.

Tissue preparation and immunohistochemical staining. A longitudinal section from the larynx including the epiglottis was obtained by autopsy, which in all cases was performed within 48 h after death. The sections were approximately 2 mm thick and were either fixed directly in 96% cold alcohol(15) or extracted in cold PBS for 24 h to remove diffusable extracellular immunoglobulins before alcohol fixation and paraffin embedding(16). Serial sections from the PBS-extracted tissue samples were cut at 6 µm and evaluated by paired immunohistochemical staining with fluorochrome-conjugated polyclonal rabbit anti-human antisera to the immunoglobulin isotypes IgA, IgM, and IgG (Dako A/S, Copenhagen, Denmark). The directly fixed tissue samples were also cut into serial sections and either stained with fluorescein-conjugated rabbit anti-human pan T (CD3) cell (Dako A/S) or with a mouse monoclonal anti-human antibody to the nonpolymorphic site of the HLA-DR determinant (Becton Dickinson, Sunnyvale, CA). The latter two antigens were visualized by a three-step immunofluorescence method applying biotinylated goat anti-rabbit IgG (Dako A/S) and biotinylated goat anti-mouse IgG2a (Southern Biotechnology Associates, Birmingham, AL), respectively, and fluorescein-labeled avidin (Vector Laboratories, Burlingame, CA) (Table 3).

Table 3 Antibody information
Full size table

Quantitation of IgA, IgG, and IgM immunocytes, semiquantitative scoring of T lymphocyte density in the mucosa of the epiglottis and larynx, and semiquantitative evaluation of epithelial HLA-DR expression were performed with a Leitz Aristoplan fluorescence microscope (Leica Microscopie and Systeme GmbH, Wetzlar, Germany) equipped with a Ploem-type epilluminator for narrow-band excitation and selective filtration of green and red emission colors. Cell counting of immunocytes density was based on the counting of 30 grid fields. Scoring of the T cell density was based on dividing the mean cell counts into three categories: in the larynx, 0-15 cells per grid was given a score of 1, 16-29 cells per grid was given a score of 2, and ≥30 cells per grid was given a score of 3. For the epiglottis the scores were as follows: 0-9 cells per grid scored 1, 10-14 cells per grid scored 2, whereas ≥15 cells per grid corresponded to a score of 3. In most cases stable means for immunocyte counts and T cell scores were obtained after evaluation of 25 grid fields. HLA-DR expression was scored in the following way: intense positive staining in surface epithelium (squamous and columnar) was given a maximum score of 2, no staining in the surface epithelium was scored 0, intermediate staining was scored 1, no staining in the glandular epithelium was scored 0, staining of 50% of the glands was scored 2, and intense staining of most of the glands (≥75%) was scored 4. Intermediate staining patterns received scores of 1 or 3.

Reproducibility of cell counting. Interindividual reproducibility of cell counts was tested by counting 30 grids for the three immunocyte isotypes independently and another 30 grids for the immunocyte isotype IgA and IgG by two observers (Å.V. and T.O.R.). Intraindividual reproducibility of staining and morphometric evaluation was tested by an evaluation of IgA-positive immunocytes in the epiglottis and larynx in adjacent sections in 22 cases, by blind trial.

Statistical analyses. For comparison between groups, the Mann-Whitney U test was applied and for reproducibility testing, the κ test was used. To evaluate whether there were differences in symptoms before death according to sleeping position when found dead, and whether there were more infants with symptoms in the group with high CSF IL-6 levels than in the group with low CSF IL-6 levels, the χ2 test was applied.

RESULTS

Clinical symptoms. In total, there were 18 infants with symptoms of a mild upper respiratory infection (a common cold with rhinitis and/or cough and/or fever) and 19 without such symptoms. Among the infants with high CSF IL-6, 12 had signs of an infection, whereas the corresponding number in the group with low CSF IL-6 was six.

Nineteen of the babies were found dead in a prone position. Significantly more prone sleeping babies had symptoms before death than the babies sleeping on their side or back (p = 0.02). Babies found prone also had a higher number of IgA immunocytes in the larynx (p = 0.02). Furthermore, infants with symptoms before death had a higher number of IgA immunocytes in the laryngeal mucosa (p = 0.01).

Microbiologic examination. In three of the cases with high CSF IL-6 levels, respiratory viruses were found, whereas no such viruses were found in the low CSF IL-6 group. In one of the cases with respiratory viruses a potentially pathogenic bacteria was also found in the lung tissue. Furthermore, bacterial growth that might be of significance was found in six additional cases (Table 4). However, five of the cases with low IL-6 levels also showed bacterial growth with potential significance (Table 5).

Table 4 Microbiologic and immunologic examinations in SIDS cases with IL-6 ≥30 pg/mL
Full size table
Table 5 Microbiologic and immunologic examinations in SIDS cases with IL-6 ≤5 pg/mL
Full size table

Reproducibility. Cell counts by two different observers showed a high degree of reproducibility (κ = 0.89). Reproducibility of the staining procedure and scoring system applied by the same observer (Å. V. I and Å. V. II) was also substantial (κ = 0.72).

IgA, IgG, and IgM immunocytes. SIDS cases with IL-6 levels ≥30 pg/mL had a significantly higher number of IgA immunocytes in the epiglottis (p = 0.03) (Fig. 1a) and in the laryngeal mucosa (p = 0.007) (Fig. 1b) than cases with IL-6 levels ≤5 pg/mL. No differences were found for IgG and IgM immunocytes. Sections from larynx with IgA and IgG immunocytes, and IgA and IgM immunocytes, respectively, are shown in Figure 2, a and b.

Figure 1
figure 1

Mean immunocyte counts in the epiglottis (a) and the laryngeal mucosa (b) in SIDS cases with high CSF IL-6 levels (≥30 pg/mL) and low CSF IL-6 levels (≤5 pg/mL). SIDS cases with high IL-6 levels had significantly higher IgA cell numbers per tissue unit (mm2) in the epiglottis (p = 0.03), as well as in the larynx (p = 0.007), than those with low IL-6 levels.

Full size image
Figure 2
figure 2

Microscopic sections from laryngeal mucosa in a case with high CSF IL-6 values. a, Immunocytes stained green for IgA and red for IgG. b, Immunocytes stained green for IgA and red for IgM (magnification ×200). c, Routine stained section (hematoxylin and eosin) of laryngeal glands. d, Adjacent section stained green for HLA-DR (magnification ×200).

Full size image

HLA-DR expression and T cells. Sections from the larynx with laryngeal glands stained green for HLA-DR are shown in Figure 2, c and d. HLA-DR expression in laryngeal glandular epithelium was more extensive in SIDS cases with IL-6 levels ≥30 pg/mL than in those with levels ≤5 pg/mL (p = 0.05) (Fig. 3).

Figure 3
figure 3

HLA-DR scores in laryngeal epithelial cells in SIDS cases with high CSF IL-6 levels (≥30 pg/mL) and low CSF IL-6 levels (≤5 pg/mL). SIDS cases with high IL-6 levels had significantly higher HLA-DR staining scores than those with low CSF IL-6 levels (p = 0.05).

Full size image

No differences were found for T cells. When combining the scores for HLA-DR and CD3 in the larynx, however, the scores showed a more pronounced stimulation in the group with high levels of IL-6 in the CSF, the p value being close to significance (p = 0.05).

DISCUSSION

To our knowledge, the present paper demonstrates for the first time a link between an immune response in the laryngeal wall and high IL-6 levels in the CNS in SIDS. The normal upper limit of IL-6 in CSF is reported to be 5 pg/mL(17). This is in accordance with our previous studies, in which we have found the CSF IL-6 levels in acute, violent deaths to be <5 pg/mL(9,14). Steinmetz et al.(18) claim that serum IL-6 levels >30 pg/mL should be considered clearly elevated. In the absence of a specific reference value for CSF, this cutoff value was also chosen for CSF in the present study. Several studies of the prevalence of cytokines in CSF in cases of various types of meningitis and septic shock or bacteremia report the IL-6 levels to be very high(19,20), in most cases >100 pg/mL, with a median of several thousand picograms per milliliter. If we only include cases with values >100 pg/mL in the present material, the link between SIDS cases with high IL-6 and an immune stimulation in the larynx still shows the same trend.

The coexistence of signs of infection before death and an increased number of IgA immunocytes in the laryngeal mucosa in many SIDS cases pinpoints the significance of the larynx as an important anatomic site for possibly dangerous immune reactions in the pathogenesis of SIDS. Moreover it has been demonstrated that respiratory tract infections may interfere with respiratory control mechanisms(13). Thus, vagal reflexes may induce bradycardia and apnea(21). Lindgren and Grøgaard(13) recently showed that laryngeal stimulation in RSV-infected infants resulted in a reinforced reflex apnea response compared with noninfected infants. It is feasible that other infectious agents may act in the same way. Experiments in animals have shown that laryngeal stimulation may induce prolonged and even fatal apneas(11,12). In addition, stimulation of chemoreceptors and free nerve endings in the upper airway induces reflex apnea in infants(21,22).

Three of the SIDS cases in the group with high IL-6 levels are ≥12 months. They were included, however, because it has been shown that there is no relationship between age or sex of patients and the ability to produce IL-6(14,23,24). Furthermore, both mature and immature neonates may produce IL-6 as a response to life-threatening infection(24).

The fact that babies who were found dead in a prone position were more likely to have had symptoms of infection before death and to have increased laryngeal IgA response fits with the observation of a marked increase in the risk of dying from SIDS when factors such as prone sleeping position, upper airway infection, and heat stress, caused by either overwrapping or excessive environmental temperature, are all present(25,26).

Several studies have shown that levels of IL-6 in CSF are elevated in meningitis(19,20) and that IL-6 is an endogenous pyrogen(27). The link between peripheral immunologic organs and the CNS may be blood-borne mediators(28–30) or be caused by retrograde axonal transport(31). Thus, Maehlen et al.(31) have shown that peripheral stimulation may alter the immune response locally within the CNS through retrograde axonal transport. It is therefore possible that laryngeal stimulation may stimulate centers in the brain stem and thereby provoke a production of IL-6 in the CNS. Both microglia and astrocytes(32), as well as endothelial cells(33), are capable of producing IL-6. It is also possible that IL-6 may be produced locally and then, through retrograde axonal transport, reach the CNS. One might also speculate whether IL-6 in the cerebrospinal fluid could be a result of leakage from autolytic cells after death. In a previous study(14), CSF was sampled at two different times after death in 26 subjects. The median IL-6 level increased from 73 to 116.5 pg/mL in 18 h. In half of the subjects a postmortem increase was seen, whereas one third showed a decrease. Only one of the cases would have been transferred from the high IL-6 group to the low IL-6 group, whereas two cases would have been included if the second sample had been chosen for the study. Exclusion of the first case would not have influenced the results of the study.

To rule out the possibility of passive leakage-especially of IgG-into immunoglobulin-producing cells, paired immunohistochemical staining was applied, with double staining for IgA and IgM. No double staining was seen, except in two cases, in which a few scattered cells showed mixed fluorescence. These cells were excluded.

Noah et al.(34) found markedly elevated levels of the cytokines IL-1β, IL-1γ, IL-6, and TNF-α in nasal lavage fluid in infants and children with acute upper respiratory disease. They furthermore state that increased cytokine production in nasal secretions is a general phenomenon in viral upper respiratory infection and that it is not specific to certain pathogens. IL-6 may be produced by a wide variety of cells, including epithelial cells. Infection by several different viruses and bacteria may also enhance the IL-6 production in fibroblasts and monocytes. A production of IL-6 in laryngeal mucosa could therefore be a possible source of the observed IL-6 levels in CSF in our cases, transmitted either by retrograde axonal transport(31) or by blood. However, Waage et al.(19) and Chavanet et al.(20) have shown that the systemic circulation and the CNS represent separate compartments, and that cells normally present in the CNS may release the cytokines.

In three of our cases with high CSF IL-6 levels who were found prone and in whom there had been light infectious symptoms before death, viruses were detected in the nasopharyngeal secretions (RSV, adenovirus, and parainfluenza virus). It has also been shown that both lambs and infants will respond to heat stress with increased respiratory rate(35). In infants irregular breathing was almost always noted when the body temperature reached 37-37.1°C, and this was seen before any other sign of thermoregulation. There was also an increased frequency of apneic episodes.

Mistchenko et al.(36) show that in adenovirus infection in children, the serum IL-6 concentration is related to the severity of the disease. In moderately ill patients, there was no IL-6 production. We have previously shown that in a group of children dying of violent deaths there were significantly lower CSF IL-6 levels than in both the SIDS group and the group with infectious deaths(9,14). In the SIDS babies, the cytokines produced may give hyperpyrexia, which, when added to the increased temperature caused by known risk factors such as prone position and overwrapping, may induce irregular breathing(35) and thus hypoxia and finally coma and death. Laryngeal reflexes in the presence of hypoxemia have cardioinhibitory effects(37). The present study is in accordance with these findings.

We have previously looked at the development of immune response markers in the tracheal wall in fetal life and the first year after birth.(38). No IgA immunocytes were present until 2 wk after birth, from which time the number increased. A few IgG-, IgM-, and CD3-positive cells were seen throughout the period studied, with a significant increase of CD3- and IgM-positive cells after birth. The number of IgA-positive cells seemed stable after the fourth postnatal week. In one of our cases the infant was 1 wk old and did not have any IgA-positive cells, which is in accordance with these findings. All other subjects were 3 wk or older and should thus not be subject to variations owing to age with regard to IgA-positive cells. Gleeson et al.(39) have presented a paper that favors the hypothesis of an immunologic overreaction. They demonstrated an infant who ultimately died of SIDS and who had a more prolonged and intense immune response compared with control infants with the same upper respiratory infection.

Several researchers have reported evidence of similarities between SIDS and infectious deaths(7,14,40,41). We believe that the rapid development of mucosal immunity in the first months after birth(38) may cause a vulnerability for an immunologic overreaction; i.e. the babies may react in an inappropriately strong manner to minor immunologic stimuli, such as slight upper respiratory infections(42). This could be the reason for the striking discrepancy between the high IL-6 levels in the CSF in many SIDS cases and the lack of correspondingly severe clinical symptoms. A recent report in animal experiments(43) suggests a two-stage scenario: an initial sympathoexcitation with profuse sweating and tachycardia followed by sympathoinhibition, and enhanced vagal outflow inducing bradycardia and hypotension.

High doses of IL-6 have been given to humans, inducing fever and headache, but not hypotension(44). However, Minghini et al.(45) have recently reported that IL-6 and IL-1 are potent dilating agents for skeletal muscle resistance vessels under in vivo conditions, acting together with parenchymal or intravascular factors. Parents often report their babies found dead in SIDS to be extremely warm and sweaty(14). The fever-inducing effects of IL-6 could induce apnea and irregular breathing, which eventually could lead to hypoxic apnea and death. A possible effect of IL-6 on the vasculature and thus the blood pressure is a question that apparently is still not solved.