Objective. The purpose of this study was to use proton magnetic resonance spectroscopy (MRS) as a metabolic assay to describe biochemical changes during the evolution of neuronal injury in infants after shaken baby syndrome (SBS), that explain the disparity between apparent physical injury and the neurological deficit after SBS.
Methodology. Three infants [6 months (A), 5 weeks (B), 7 months (C)] with SBS were examined repeatedly using localized quantitative proton MRS. Examinations were performed on days 7 and 13 (A), on days 1, 3, 5, and 12 (B), and on days 7 and 19 (C) posttrauma. Long-term follow-up examinations were performed 5 months posttrauma (A) and 4.6 months posttrauma (B). Data were compared to control data from 52 neurologically normal infants presented in a previous study.
Results. Spectra from parietal white matter obtained at approximately the same time after injury (5 to 7 days) showed markedly different patterns of abnormality. Infant A shows near normal levels of the neuronal marker N-acetyl aspartate, creatine, and phosphocreatine, although infant C shows absent N-acetyl aspartate, almost absent creatine and phosphocreatine, and a great excess of lactate/lipid and lipid. Analysis of the time course in infant B appears to connect these variations as markers of the severity of head injury suffered in the abuse, indicating a progression of biochemical abnormality. The principal cerebral metabolites detected by MRS that remain normal up to 24 hours fall precipitately to ∼40% of normal within 5 to 12 days, with lactate/lipid and lipid levels more than doubling concentration between days 5 and 12.
Conclusions. A strong impression is gained of MRS as a prognostic marker because infant A recovered although infants B and C remained in a state consistent with compromised neurological capacity. Loss of integrity of the proton MR spectrum appears to signal irreversible neurological damage and occurs at a time when clinical and neurological status gives no indication of long-term outcome. These results suggest the value of sequential MRS in the management of SBS.
- shaken baby syndrome
- traumatic brain injury
- neuronal injury
- magnetic resonance imaging
- magnetic resonance spectroscopy
Nonaccidental trauma accounts for a high proportion of head injury in infants less than 1 year.1 In 1971, it was postulated by Guthkelch2 that whiplash forces resulting in acceleration/deceleration may be the cause of the high frequency of subdural hematomas seen in battered children when often there was no external sign of injury to the head. One year later, Caffey3 coined the term whiplash shaken baby syndrome in describing a variety of clinical complications in abused infants, and stressed that the practice of shaking had the potential to cause the residual effects of brain damage and mental retardation. In a later study, Duhaime et al4 concluded that shaking alone was not enough to cause severe injury in a healthy infant and that such injury requires impact to occur. Contrary findings were presented by Hadley et al,5 who studied a younger population of infants and concluded that in young infants sustaining severe whiplash injuries, direct trauma was not a necessary part of the injury mechanism. However, regardless of the exact mechanism of injury, shaken baby syndrome (SBS) results in major and often fatal head injury, and results in neurological deficits that are out of proportion to both the degree of physical trauma, and changes observed on MRI or computed tomography. We have previously reported distinct patterns of metabolic abnormalities observed by proton magnetic resonance spectroscopy (1H MRS) in apparently normal MRI regions of cerebral cortex in older children6 and in adults,7 after closed head trauma. At first sight closed head injury is not a metabolic insult, but secondary neuronal injury is likely to be a determinant of long-term outcome in both children and adults. In infants subject to shaking, a particularly severe form of posttraumatic brain injury occurs, the biochemistry of which has not directly been studied. In this report, we used the noninvasive metabolic assay of1H MRS to follow the evolution of neuronal injury in three infants admitted to the hospital after severe injury from shaking. In one infant studied in more detail we describe major biochemical changes, which were progressive over days, leading to severe neurological injury, which was found in two of the three infants exposed to shaking. These neurochemical changes were absent in the third infant, who made a complete recovery. Due to the similarity of these events observed in vivo by MRS, and those occurring in vitro after Ca2+-associated phospholipase A2activation8 and lysozomal damage,9 we postulate that a similar metabolic cascade of membrane damage accounts for the disproportionately severe neuronal injury resulting from an apparently trivial mechanical trauma.4,10 Interest in identifying such mechanisms of a metabolic cascade has increased as specific metabolic inhibitors have been identified that might mitigate neuronal injury and hence improve outcome.8
In three infants (aged 5 months [A], 5 weeks [B], and 7 months [C]) the diagnosis of SBS was established clinically, and in particular, severe bilateral retinal hemorrhages were identified.
Infant A (5-month-old female) was admitted with apnea after a seizure that was reported initially as the result of a fall from a table (about 3 feet). However, one parent allegedly observed the other parent severely shaking the infant. There was no prior history of abuse, and no fractures were noted on skeletal survey.
Infant B (5-week-old female) was admitted within 4 hours of the injury when the mother brought the infant to the hospital after returning home to find her behaving “abnormally” in the care of the father. Police records were strongly suggestive of shaking without blows, but there was no other history of abuse. Skeletal survey was negative and no evidence of chronic abuse was detected. A clinical examination 2 weeks postnatal was normal.
Infant C (7-month-old female) was admitted comatose after falling down three stairsteps. There was never a clear history of abuse and therefore the interval postinjury is more difficult to assign. Initially, SBS was not suspected, but on skeletal survey a fractured occipital condyl was noted. Evidence for prior injury was limited to a possible chronic subdural hematoma on MRI.
The clinical and neurological status of each infant at time of admission and on the days MRS was performed is shown in Table1. Clinical status and outcome after long-term follow-up are also presented in Table 1.
The infants were examined repeatedly using localized quantitative proton MRS (STEAM TR/TE/TM = 1500/30/13.7 millisecond), as described in.11,12 All measurements were performed on a GE medical systems Signa MR scanner using version 5.4 software, operating at 1.5 T. Examinations were performed on days 7 and 13 (A), on days 1, 3, 5, and 12 (B), and on days 7 and 19 (C) posttrauma. Localization was done on conventional T1-weighted gradient echo or T2-weighted fast spin echo images. Spectra were obtained from parietal white and occipital gray matter, avoiding any focal lesions that were obvious on the MRI. Spectra were also obtained from the frontal lobes of infant B on days 3 and 12 and from a large occipital voxel in infant C on day 7. Longer-term follow-up examinations were performed 5 months posttrauma in infant A and 4.6 months posttrauma in infant C. In all, 29 quantitative MRS examinations were performed on the 3 infants.
MRS performed on 52 neurologically normal infants 0 to 12 years shows significant age-related changes.13 Data presented in this report consist of absolute metabolic concentrations obtained in each examination expressed as a fraction of the metabolite content of normal aged-matched subjects from graphical data presented by Kreis et al,13 and expanded to more easily show the expected age-related variations for the three infants in our study. A representative spectrum from a neurologically normal infant of appropriate age is also presented.
MRI performed early or late in two of the infants (A and C) showed similar results. In infant A, an MRI performed (3/14/95) 7 days postinjury showed an extradural and subdural collection over the right posterior-lateral parietal cortex with edema or effacement of gyri in the left hemisphere. By day 12 (3/20/95), the left hemisphere was significantly smaller than the right. (Long-term follow-up at 8 to 16 weeks showed complete resolution of the hematoma and reexpansion of the left hemisphere with a noticeable but minor dilation of the underlying left lateral ventricle, secondary to atrophy.)
In infant C, the only finding on the first MRI performed 7 days after injury was bilateral frontal subdural hematoma. Thirteen days later major changes in the brain of both hemispheres were clearly identifiable, and ventricles were dilated bilaterally.
In infant B there was a small area of focal hemorrhage in the right posterioparietal cortex on day 1; otherwise, MRI was essentially normal 1 day postinjury. Then 2 and 4 days later MRI showed little change but by day 11 extensive cortical changes were present throughout both hemispheres. By day 26, little normal-appearing brain could be identified, and MRI appearances were bilaterally similar to those in patient C.
MRIs of all infants 5 to 7 days posttrauma are presented in Fig1 and show subdural bleeding and the volumes of interest used for the MRS measurement at sites remote from hemorrhage. Little difference can be seen between the MRIs. However, spectra from parietal white matter obtained at approximately the same time after injury (5 to 7 days) showed markedly different patterns (Fig2). The spectrum from infant A is normal appearing and shows levels of the principle cerebral metabolites detected by MRS, [the neuronal marker N-acetyl aspartate (NAA), creatine and phosphocreatine (Cr), choline-containing compounds (Cho), and myoinositol (mI)], to be normal for an infant of that age. Greatly reduced NAA and Cr and appearing lactate/lipid resonances in the region of 1.07 to 1.46 ppm (Lac/Lip) and lipid resonances in the region from 0.69 to 1.07 ppm (Lip) are seen in infant B, although almost absent NAA, mI, and greatly reduced Cr and the appearance of dominant Lac/Lip and Lip resonances can be seen in infant C.
The extremes of MRS findings (and neurological outcome) after SBS are illustrated in two infants. The spectrum obtained from the parietal white matter of infant A at 7 days posttrauma appears normal for an infant of that age with the spectrum from 13 days posttrauma showing only a slight decrease in the Cr resonance (Fig 3A). Levels of neuronal marker NAA, Cho, and mI were within the normal range on both days. The levels of Lac/Lip and Lip appear within the normal range for both examinations. Infant A had a good outcome and is normal neurologically. In contrast (Fig 3B), the spectrum from the parietal white matter of infant C at 7 days posttrauma shows almost absent NAA, Cr, and mI, with a dominant peak appearing in the Lac/Lip region. At 19 days posttrauma there has been an increase in the Cho resonance, the Lac/Lip resonance is still the dominant peak, while the Lip resonance has doubled in size from day 7. Infant C suffered severe neurological deficit.
The detailed analysis of the time course in patient B appears to connect the variations observed by MRS between infants A and C as markers of the severity of head injury suffered in the abuse, and indicate a progression of biochemical abnormality (Fig4A). All the principle metabolite concentrations observable by MRS (NAA, Cr, Cho, mI) remain near normal levels up to 24 hours posttrauma. A spectrum from a neurologically normal age-matched control is shown in Fig 4B for comparison. Sequential MR images from patient B on day 1 through 4.6 months show relatively few abnormalities until day 5 (Fig 4C). By day 3 it is obvious that a metabolic insult has occurred with the level of neuronal marker NAA falling drastically. The levels of Cr, Cho, and mI have also decreased. The other noticeable change to the spectrum is the appearance of Lac/Lip in the region from 1.07 to 1.46 ppm and unidentified peaks in the region between 0.69 and 1.07 ppm (Lip). The spectrum at day 5 shows little change except for further decreases in Cr and mI. By day 12 the Cho peak has shown a slight recovery; however, the peaks from Lac/Lip and Lip have doubled in size and now dominate the spectrum. Infant B also suffered severe neurological deficit. Curves for normal development covering the entire pediatric age range from the parietal white matter of presumed neurologically normal children are presented in Fig 4D.
Figure 5A combines results from all three infants with the data from infant B illustrating an apparent progression of loss of neuronal marker NAA (t 1/2 = 2.0 days) with levels falling to 20% of normal 12 days posttrauma. The loss of Cr and mI, two other principle metabolites, can also be observed in infant B. The Cr levels are at 78% of normal 1 day posttrauma and fall to 40% of normal by day 5 and remain at that level. The levels of mI in infant B fall in parallel to Cr over this time course. The striking feature of this Figure is that by day 5 in infant B and day 7 in infant C (the earliest day an examination could be performed) the levels of these three metabolites have fallen precipitately below normal, ending up at or less than 40% of normal by day 12 in infant B and day 19 in infant C. This contrasts with the levels from infant A who remained at or near normal with Cr only decreasing to 80% of normal by day 13 posttrauma. This detailed analysis of the time course of infant B appears to indicate that the loss of these metabolites is a progression of biochemical abnormality.
In contrast Fig 5B shows the level of Cho never falling lower than 50% in infants A and B, and although Cho is less than 40% of normal at day 7 in infant C (similar to the levels of the other metabolites for this infant at this time) it recovers to 80% of normal by day 19. A modest recovery is also observed for Cho levels in infant B between days 5 and 12. A slight decrease in Cho was observed in infant A on day 13.
The time course of the appearance of resonances from Lac/Lip (1.07 to 1.46 ppm) and Lip (0.69 to 1.07 ppm) after injury is shown in Fig 5C. Peak areas were normalized to values obtained from age-matched controls with a value of 1.00 (institutional units) considered normal. The levels of Lac/Lip and Lip in infant A remain at or near normal levels with a slight decrease observed over time. For infant B the Lac/Lip level was already increased 50% by day 1 with a marked increase occurring between day 5 and 12 up to 170% of normal. The Lip level is initially within normal range with the most dramatic increase occurring between day 5 and 12 paralleling the Lac/Lip increase. The level of Lac/Lip in infant C had reached 120% more than normal by the time of first examination on day 7 and remained at this elevated level. However, the Lip resonance was in the normal range on day 7 with a dramatic increase up to 120% above normal occurring by day 19. The most dramatic increases in these resonances appeared to occur after days 5 to 7 posttrauma which was later than the initial loss of other metabolites which was complete by days 5 to 7 in infants B and C (Fig5A).
The difference spectrum (day 1 − day 12) for infant B is shown in Fig 6 and emphasizes the coordinated loss of NAA, Cr, and mI, and to a considerably lesser extent Cho that has occurred over this 12-day period. In addition, the major increases in the Lac/Lip and Lip resonances are easily observed and are now seen to be accompanied by the appearance of an unidentified resonance at 2.77 ppm. Although this is likely to be a lipid, from the comparison with a model emulsion (not shown), it is not a long chain fatty acid or triglyceride. The peaks in this region are not present in the same appropriate amounts.
MRI and spectra from long-term follow-up examinations are shown in Fig7 from infant A (5 months posttrauma) and from infant B (4.6 months posttrauma) and are representative of the extremes of neurological outcome. The MRI of infant A shows slight atrophy and the spectrum from the parietal white matter is normal appearing with the levels of the principle metabolites within normal range. Infant B had a grossly abnormal MRI at 4.6 months posttrauma showing severe atrophy and global degeneration. The spectrum from the parietal white matter is abnormal with no recovery in the levels of NAA and a prominent resonance in the Lac/Lip region. Cr and Cho appear normal with mI elevated possibly reflecting a hyperosmolar state.14 The apparent recovery of some of the metabolites may be due to consolidation, ie, in initial examinations the volumes of interest contained a large amount of metabolically insulted and dying cells although in the long-term examination the volumes of interest contained only the viable tissue remaining. We were unable to obtain parental agreement for a long-term follow-up examination on infant C.
Three infants with SBS were examined repeatedly with the noninvasive assay of 1H MRS to follow biochemical changes occurring in the brain in response to neuronal injury. The MRI results for each infant showed similar findings with the major abnormality in each case being the presence of subdural hematoma. During the course of the initial examinations (up to 19 days posttrauma) there was little or no change detected by MRI that could indicate the severity of neuronal injury sustained and provide a guide as a prognostic indicator for long-term outcome. In addition, the clinical and neurological state of each infant at the time of MRS examination gave no indication concerning the long-term prognosis for each infant. However, major changes were observed in the MRS results from two of the three infants at this time. The MRS examination of infant A at 7 days posttrauma appeared normal although the examination of infant C at the same time interval revealed major abnormalities indicating that a severe metabolic insult had occurred (Fig 3, A and B). From the time course of biochemical events occurring in the brain of infant B (Figs 4 and 5A), the likely sequence of the neurochemical disorder of SBS may be concluded for all three patients. The biochemical explanations for the variable (and poor) outcome after shaking are clearly seen from these results to be the almost complete disappearance of neuronal marker, NAA, along with severe loss of Cr and mI. The loss of these metabolites occurs rapidly during the first 5 days posttrauma indicating a massive disruption of neuronal function at the intracellular level. The MRI findings (near-normal in two-thirds) and clinical status (poor in all three) were not sensitive to the varied and severe changes we observed with MRS.
There is also an unusual metabolic component of the response to SBS which is likely to be a lipid. It is a metabolic marker that arises progressively, after a quiescent or lag period of 5 to 7 days posttrauma with the most dramatic increases occurring after the initial loss of the other cerebral metabolites (Fig 5C). The Lip resonance between 0.69 to 1.07 ppm is initially at or slightly above normal for all three infants but doubles in amount between days 5 to 7 and days 12 to 19 in infants B and C, respectively. The Lac/Lip resonance between 1.07 and 1.46 ppm is initially increased in infant B with the most dramatic change paralleling the Lip increase after day 5. In infant C this resonance was already double the normal levels by day 7 which possibly indicates that this infant had suffered a greater degree of injury in the abuse. The higher initial values of this resonance in infants B and C probably reflect a measure of hypoxia or ischemia as this resonance has a contribution from lactate.
In parallel with these dramatic biochemical changes was the severe clinical neurological impairment in two of the three infants. Although a larger series is required to confirm our observations, a strong impression is gained of MRS as a prognostic marker. Only infant A recovered significant neurological function. In contrast, infants B and C remained in a severely compromised neurological state.
Neuronal damage is the result of the biomechanical forces involved in the mechanism of injury. However, the severity of secondary effects, resulting from ischemia due to perturbed autoregulation or raised intracranial pressure from brain swelling,15 is a major determinant of long-term outcome in the days after trauma. It has been postulated that SBS results from acceleration/deceleration forces sufficient to cause rupture of the bridging veins running from the cerebral cortex to the venous sinuses resulting in subdural hematoma.2 In our study all the MRIs showed the presence of subdural hematoma. Duhaime et al4 reported that in fatal cases of SBS studied all had signs of blunt impact to the head although more than half of these findings were only noted at autopsy. These findings included skull fractures and scalp contusions mostly in the occipital or parietooccipital region. However, our results show that despite similar MRI findings, in some cases metabolic events are triggered that lead to irreversible neuronal damage and resultant poor neurological outcomes.
We believe that the loss of the metabolites detected by MRS and the appearance of the Lac/Lip and Lip resonances may signal that neuron membrane damage and solubilization of phospholipids, thus rendering the lipids MRS visible, is the first stage in a cascade of metabolic events resulting in neuronal injury. Although detailed understanding requires further studies in vitro, this is the first noninvasive demonstration that lipolysis may be an important prerequisite for one form of neuronal injury in man.
Lysosomal damage with the release of acid hydrolases,9 and perhaps most importantly, phospholipase A28 is recognized as a cause or at least an accompaniment of neuronal damage. It is possible that deleterious effects on neurons are more marked in those immature neurons in which final axonal connections are incomplete. The speed with which the loss of neuronal marker NAA occurs (t 1/2 ∼ 2 days) in the only infant from whom accurate data are available (B), and from the earliest data point available in infant C, suggests an accelerated process triggered by an initial injury (shaking). Auto catalytic enzymatic processes are one of a number of potential mechanisms to describe such an event. This cascade may explain why the degree of neurological damage suffered is greater than the amount of trauma apparently involved. We postulate that trauma itself, although not the sole cause of neuronal injury, is the initiating event that may release enzymes that cause the continued development of neurological damage after trauma has ceased.
Monitoring these events over time in vivo is the first step in the search for pharmacological interventions which may interrupt the devastating neurological consequence of this metabolic cascade after brain trauma in infants.
Numerous MRS studies are now available to document the metabolic changes in infant brain subject to a variety of injuries. To our knowledge, we are the first to indicate the nature and severity of the biochemical response to traumatic injury, but Auld et al16describe other patterns in a variety of nontraumatic brain injuries. Hypoxic injury17,18 also results in loss of NAA and accumulation of lactate. The pattern seen here in SBS appears somewhat different from hypoxic injury alone.19 A larger study is required to establish the specificity, if any, of any of the differences seen in our small series, to SBS.
THE AVAILABILITY, COST, AND CLINICAL PLACE OF1H MRS IN SBS
It is no longer unusual to find 1H MRS facilities within university or community hospitals. With automation and rapid scanning techniques, 1H MRS can be performed without specialized personnel or equipment within 6 to 8 minutes.
We suggest that 1H MRS may provide important insights into SBS and, by providing prognostic information, has the potential to modify current patient care. The added cost of MRS is trivial compared with computed tomography or MRI, so that it should be considered in SBS in cases where the neurological picture is out of scale with the standard brain image.
This work was supported by the Jameson Foundation; the Whittier Family Foundation, South Pasadena, CA; and the Schulte Research Institute, Santa Barbara, CA.
We are grateful to Drs William Caton and Kathleen Egan for permission to examine patients under their care, and to Kathy Harris, RN, for sustained clinical supervision.
- Received December 7, 1995.
- Accepted March 4, 1996.
Reprint requests to (B.D.R.) Huntington Medical Research Institutes, 660 S Fair Oaks Ave, Pasadena, California.
- SBS =
- shaken baby syndrome •
- MRI =
- magnetic resonance imaging •
- MRS =
- magnetic resonance spectroscopy •
- FSE =
- fast spin echo •
- NAA =
- N-acetyl aspartate •
- Cr =
- creatine and phosphocreatine •
- Cho =
- choline-containing compounds •
- mI =
- myoinositol •
- Lac/Lip =
- lactate/lipid •
- Lip =
- Krugman RD
- Guthkelch AN
- ↵Haseler LJ, Ernst T, Caton W, Moats RA, Shonk T, Ross BD. Clinical outcome after closed head injury: 1H MRS in children. In: Proceedings 3rd Society of Magnetic Resonance. Vol 1, Nice, France: 1995:381
- ↵Ernst T, Ross BD, Shonk T, Kreis R, Caton W, Clark C. Quantitative proton MRS for prognosis after closed head injury. In: Proceedings 12th Society of Magnetic Resonance in Medicine. Vol 1. New York, NY: 1993:323
- Verity MA
- Nixon RA,
- Cataldo AM
- Leadbeatter S,
- James R,
- Caydon S,
- Knight B
- ↵Ross BD, Ernst T, and Kreis R. Proton magnetic resonance spectroscopy in hypoxic ischemic disorders. In: Bax M, Faerber EN, eds. MRI of the Central Nervous System in Infants and Children. London: MacKeith Press; 1996, in press
- Copyright © 1997 American Academy of Pediatrics