Objective. Vertebral artery compression causing brainstem ischemia has been suggested to underlie the sudden infant death syndrome. Vertebral artery distortion from neck movements has been demonstrated by angiography in infants, but direct evidence for arterial compression is lacking. In an attempt to demonstrate vertebral artery compression from head movement, we examined at postmortem the vertebral arteries of infants after neck extension or rotation.
Methods. The C1–C7 spinal column, together with a 2-cm rim of skull base, was removed from 20 infants dying from sudden infant death syndrome or other causes. In 5 cases the neck was extended, in 9 cases it was rotated 90° to the right, and in 6 cases the neck was held in the neutral position. The neck was maintained in these positions during formalin fixation, and serial sections of selected blocks were examined microscopically.
Results. In 3 of 5 extended cases, bilateral vertebral artery compression was seen between the occipital bone and C1. In 3 of 9 rotated cases, the left vertebral artery was compressed adjacent to C1 before the artery entered the transverse foramen. No vertebral artery compression was seen in the necks held in the neutral position.
Conclusions. The vertebral arteries of some infants can be compressed by neck movement. This could induce lethal brainstem ischemia in infants with inadequate collateral blood flow or with poor compensatory arterial dilatation, and may underlie some cases of sudden infant death syndrome.
It has been suggested that head movements causing compression of the vertebral arteries in the neck, with subsequent brainstem ischemia, could underlie some cases of the sudden infant death syndrome (SIDS).1–3 Angiographic studies have shown that neck rotation in normal adults can compress one vertebral artery in the atlantoaxial region,4,,5 and postmortem angiographic1 and vascular flow6 studies have suggested that neck extension and/or rotation can compress infant vertebral arteries. Anatomic differences in infants at the base of the brain may predispose to vertebral artery compression.1 In particular, the lateral mass of C1, which provides a buttress against vertebral artery compression, is small in infants, and an unstable atlantooccipital joint and large foramen magnum allow C1 to invert into the foramen magnum on neck extension and compress the vertebral arteries.1 The suspicion that neck movements are related to SIDS has been strengthened by the finding of a strong link between SIDS and the prone sleeping position.7 This is because an infant placed prone must rotate or extend its neck to clear its nose from the bedding, and if this causes vertebral artery compression, fatal acute brainstem ischemia could result.
Although postmortem angiographic1 and vascular flow6 studies have suggested vertebral artery compression in infants, no arterial compression on infant neck movement has been demonstrated directly. We therefore looked for both macroscopic and microscopic evidence of vertebral artery compression in the extended or rotated necks of infants. Because most neck extension occurs at the occipito-C1 level and most rotation between C1 and C2, we examined these regions in particular for vertebral artery compression.
Infants studied (Table 1) were those who had postmortem examinations at the New South Wales Institute of Forensic Medicine in Sydney. Of the 20 infants, 9 were diagnosed based on clinical history, death scene investigation, and postmortem findings as dying from SIDS, and 11 infants were judged to have died from other causes (aspiration, 3 infants; undetermined, 3 infants; congenital heart disease, 2 infants; pneumonia, 2 infants; and sepsis, 1 infant). Postmortem delay was <24 hours for all cases. The project was approved by the University of Sydney human ethics committee.
Spinal Column Dissection and Manipulation
At the time of the postmortem examination the cervical spinal column was removed together with a 2-cm rim of skull base around the foramen magnum. Care was taken to ensure that all spinal and occipitospinal ligaments were preserved. The skull + column was replaced for cosmetic purposes by plastic tubing. Skull + columns were fixed in one of the following three positions: 1) Five were extended by tying a silk suture to the posterior margin of the rim of skull base and the posterior margin of C7 and tightening the suture without undue force (Fig 1). 2) Nine were rotated 90° to the right and the rotation maintained in a set of purpose-built clamps. 3) Six were left in the neutral position. All specimens were fixed in 10% neutral buffered formalin for 1 week before dissection. After fixation, the specimens retained their positions after the suture was cut (extended cases) or after removal from the clamps (rotated cases).
Microtomy and Staining
Skull + columns were decalcified in RDO (Phoenix) and blocked in two ways. (1) In the extended necks, the skull + C1 + C2 was separated from the rest of the column by a transverse incision between C2 and C3. The skull + C1 + C2 segment was then divided into 5-mm sagittal blocks and the C3–C7 segment into 5-mm transverse blocks. (2) The rotated skull + columns were divided transversely into 5-mm blocks. The neutral position skull + columns were blocked as either the extended (two cases) or rotated (four cases) specimens to act as controls. All blocks were processed routinely and embedded in paraffin. In the blocks encompassing the extended skull + C1 + C2 and rotated C1 + 2 regions, 10-μm gapless serial sections were cut (see Fig 1) and each 20th section stained with hematoxylin and eosin. Single 10-μm sections were cut from other blocks.
Assessment of Arterial Compression
Because arteries can collapse artifactually after death, compression was judged to be present only when the artery could be seen to be narrowed by surrounding tissue and when the nearby vertebral vein was also narrowed. The method was not therefore able to detect if an artery had been narrowed by longitudinal distortion, which would stretch the artery within a normal-sized perivascular space.
To counteract passive arterial collapse, the basilar artery of a single case was tied off and a fluid substance, Vinylite, was injected into the vertebral arteries, which were then also tied off before the neck was extended. Vinylite has the property of being a fluid until it contacts water, when it solidifies into a rubbery mass. The vertebral arteries were therefore kept open by the pressure of the injected fluid but remained compressible during the initial neck extension, giving almost perfectly round lumens in noncompressed regions. An image analysis program (National Institutes of Health Image) was used to measure the cross-sectional luminal area of these arteries. The percentage of decrease in area was estimated by dividing the image-measured area by the theoretical maximum area, calculated as if the luminal circumference was a perfect circle (area = circumference2/4π).
Each vertebral artery ran from lateral to medial between the occipital bone and C1 before entering the foramen magnum (Figs 2 and 3). The artery lay on the superior margin of C1, with the posterior margin of the lateral mass of C1 lying anterior to the artery (see Figs 2 and3). The artery did not lie in a groove on the surface of C1 (as in adults) but ran freely over the superior surface (see Figs 2 and 3). In young infants the lateral mass consisted of a small outcrop of fibrous connective tissue with a thin covering of cartilage, whereas in older infants the lateral mass was larger and consisted mostly of bone. The lateral mass ran from anteromedial to posterolateral, so medially the vertebral artery was not protected by it. The posterior atlantooccipital membrane arched closely over the top of the vertebral artery (see Figs 2 and 3). The remainder of the space between the occipital bone and C1 was taken up by loose connective tissue.
In all five cases neck extension narrowed the soft-tissue space between the occipital bone and C1, and in three infants this led to compression of both vertebral arteries (see Table 1). In each case the vertebral artery was compressed between the posterior atlantooccipital membrane and the superior margin of C1. Vertebral artery compression was found in the following two situations: 1) where the posterior atlantooccipital membrane was particularly thick, even when the posterior margin of the C1 lateral mass was well formed (Fig 4); on the other hand, when this membrane was thin, extension did not compress the vertebral artery, even when the lateral mass was small (see Fig 3); and 2) medially on the surface of C1, where the posterior margin of the lateral mass did not protect the artery (Fig 5B). Over the lateral surface of C1, the vertebral artery was tucked behind the posterior margin of the lateral mass, so compression was not usually seen laterally, or it was only moderate (see Fig 5A).
In the case in which Vinylite was injected into the vertebral arteries, a reduction of cross-sectional area of 63% was found between the skull and C1 in the vertebral artery on one side (Fig 5A), and a 50% reduction on the other side (see Fig 5B). This represents a considerable decrease in blood flow in the vertebrobasilar system, because blood flow is related to the fourth power of the vessel diameter.8
In six of the nine cases in which the head was rotated to the right, no compression of the vertebral artery was seen between C1 and C2 (see Table 1). At most levels from the superior margin of C1 to the inferior margin of C2, the vertebral artery was surrounded by a generous periarterial space occupied by veins and nerves, especially when protected by the bony margins of the transverse foramen (Fig 6). In the remaining three cases, the vertebral artery on the left was compressed between soft tissue (muscle and nerve) and the lateral margin of C1 before the artery entered the transverse foramen of C1 (Figs 7 and8). This compression was very localized, with the artery only a few millimeters away on either side returning to its normal diameter.
Neutral Neck Position
No vertebral artery compression was seen in the four cases where C1 + C2 blocks were cut transversely or in the two cases in which the skull + C1 + C2 blocks were cut sagittally.
Variation in Size of Vertebral Arteries
Side-to-side size variations in the vertebral artery were seen on transverse sections of some columns, the right-left differences varying from minor to marked (Fig 9).
Anatomy predisposing to vertebral artery compression in extension
In a series of meticulous dissections, Gilles et al1demonstrated why the vertebral arteries of infants are particularly predisposed to compression on neck extension. Major findings were that in infants (1) the lateral mass on the superior surface of C1 is small and is therefore not a good barrier against downward compression on the vertebral artery, (2) the arch of the atlas is smaller than the foramen magnum, allowing the arch to invert up into the foramen during neck extension and so compress the vertebral arteries, (3) the ligaments joining the occiput and C1 are poorly formed, allowing greater mobility than in adults, and (4) the vertebral artery lies exposed on the surface of C1 and not in a groove in the bone as in adults. Furthermore, we have found that the posterior atlantooccipital membrane, which lies immediately above the vertebral artery, is often thick and compresses the vertebral artery during extension. The anatomy of the tissues surrounding the vertebral artery in infants therefore allows for more vertebral artery compression on neck extension than in adults.
Our results indicate that vertebral arteries can be compressed either after neck extension or rotation. The degree of neck movement applied was relatively modest, given the flexibility of the normal infant neck. Although we looked at the effects of extension and rotation separately, a study from Germany suggests that a combination of neck movement may produce even more vertebral artery compression than extension or rotation alone.6 When the flow of fluid under pressure within neck arteries was examined postmortem in these infants, a decrease in vertebral flow could be demonstrated in various neck positions, particularly in extension combined with rotation.6 This study was, however, unable to demonstrate where the arterial compression was taking place, and what structures were responsible for the compression. Another study combined with a technique such as magnetic resonance imaging could yield useful information, because the site of arterial compression would then be able to be examined in a number of different neck positions in a single infant. Unfortunately, we had insufficient numbers of cases to look for histologic evidence of compression in these combined movements.
Risk Factors and the Hypothesis
Any pathologic finding that purports to be able to explain SIDS must be consistent with the known risk factors for the condition, ie, prone sleeping, the 1- to 6-month age range, overheating, and maternal smoking. The link between SIDS and these risk factors could be postulated to be decreased vertebral artery flow on the following grounds (Fig 10):
The Prone Sleeping Position
The prone sleeping position is now accepted as the major risk factor for SIDS.7 An infant sleeping face down, especially in soft bedding,9 is likely to either extend or rotate its head (or both) to clear its nose from the bedding. Because these are the neck movements implicated in causing vertebral artery compression, this compression with subsequent brainstem ischemia provides a pathogenetic link between the prone sleeping position and SIDS.
Age Range for SIDS
Most infants who die from SIDS do so between 1 and 6 months of age.10 As an infant grows, the adverse anatomic factors that predispose to vertebral artery compression from head movement (a small C1 lateral mass, no grooving on C1, and atlantooccipital instability) begin to resolve, leading to a greatly reduced chance of artery compression in later life.1 Furthermore, the infant brainstem appears particularly susceptible to ischemia, because in the first few months of life, blood flow to the brainstem lags behind the rapidly growing brainstem.2 On the other hand, infants <1 month of age may not have the strength or coordination to rotate or extend their heads to the degree needed to get major compressions of the vertebral arteries, so these very young infants would be less susceptible to brainstem ischemia. Hence, both the lower and upper age limits for most SIDS deaths can be explained by vertebral artery compression caused by neck movement.
The risk factor for SIDS of overheating11 is compatible with a hypothesis of reduced vertebral artery flow causing the sudden death. Heating can increase blood flow to the upper limbs four or five times above normal.12 The vertebrobasilar circulation is in competition with the vascular bed of the upper limbs for blood pressure and flow, because the vertebral arteries arise from the subclavian arteries. Any increase in subclavian artery flow from limb heating can divert blood from the vertebral arteries, so flow to the posterior fossa must be kept constant by vasodilatation in the vertebrobasilar system.13 If this finely regulated system is upset (see below), heating an infant could lead to slowing or reversal of vertebral artery flow.
Infants of mothers who smoke during pregnancy or postnatally have a high risk for SIDS.14 Of interest, passive smoking has been shown to impair arterial dilatation in healthy young adults.15 If a similar mechanism operates in smoke-exposed infants, their vertebral arteries may not be able to dilate efficiently in response to the subclavian steal induced by overheating.
Although these links are necessarily speculative at this time, it would be possible to test some of these factors individually (in particular, head position and body heating) on live infants by using ultrasound techniques to measure cerebrovascular blood flow.16
Recent quantitative studies have indicated that no significant neuronal loss or gliosis is found in the brainstems of most SIDS infants.17,,18 These negative pathologic findings in SIDS brainstems are what would be expected after an acute brainstem infarct, because some hours of survival are needed before microscopic evidence for nervous system infarction becomes evident.19
Why Should Only Some Get SIDS?
In our study it was evident that vertebral arteries can be compressed by head movement in a percentage of infants dying both from SIDS and from other causes. Why then would only some infants suffer sudden fatal brainstem ischemia when their vertebral arteries are compressed? First, a fatal outcome could ensue because of architectural variations in the vessels at the base of the brain. If one vertebral artery is compressed, for example by head rotation, the contralateral vertebral artery should in most cases be able to supply sufficient blood to the brainstem. In many infants, however, the sizes of the vertebral arteries differ, and in up to 40% of infants, flow in one vertebral artery is less than half that of the other.2Therefore, occlusion of the larger vertebral artery could put the brainstem in jeopardy. Even with both vertebral arteries compressed, for example by head extension, flow through the posterior communicating arteries from the carotid arteries should theoretically be able to save the brainstem from ischemia. However, flow in the very small posterior communicating arteries of infants is able to supply on average only 13% of basilar artery flow,2 so the infant brainstem is at risk of ischemia with bilateral vertebral artery compression. Second, a combination of two or more risk factors (eg, the age range at which anatomic features predispose to vertebral artery compression, the prone sleeping position with subsequent extreme neck movement, overheating with a subclavian steal, and insufficient vertebrobasilar dilatation caused by passive smoking) may need to be present to result in fatal brainstem ischemia. Those infants unlucky enough to have lethal combinations of risk factors would succumb to SIDS, whereas those with only one risk factor at any one time would survive.
In conclusion, this histopathologic study provides direct evidence that both neck extension and rotation in infants can compress the vertebral arteries. In infants with architectural anomalies in arteries at the base of the brain, or with risk factors that reduce vertebrobasilar flow, this compression could cause acute brainstem ischemia and death. Because an infant tends to rotate or extend its head in the prone sleeping position to clear its nose from the bedding, vertebral artery compression caused by these head movements could explain why the prone position is a major risk factor for SIDS.
This study was supported by a grant from the National SIDS Council of Australia.
We thank Dr Denise Crute for assistance with the dissections.
- Received April 21, 1998.
- Accepted August 6, 1998.
Reprint requests to (R.P.) Department of Pathology DO6, University of Sydney, Sydney NSW 2006, Australia.
- SIDS =
- sudden infant death syndrome
- Pamphlett R,
- Murray N
- Tatlow WFT,
- Bammer HG
- Saternus KS,
- Adam G
- ↵Guyton AC. Textbook of Medical Physiology. 8th ed. Philadelphia, PA: WB Saunders; 1991:150–158
- Ponsonby AL,
- Dwyer T,
- Gibbons LE,
- et al.
- Hellon RF
- Toole JF
- Mitchell EA,
- Ford RP,
- Stewart AW,
- et al.
- Wong WS,
- Tsuruda JS,
- Liberman RL,
- et al.
- ↵Graham D. Hypoxia and vascular disorders. In: Adams JH, Duchen LW eds. Greenfield's Neuropathology. 5th ed. London, England: Edward Arnold; 1992:153–268
- Copyright © 1999 American Academy of Pediatrics