Head Injury Research at the University of Cincinnati
Christopher D. Smith, M.D.
Head injury causes significant morbidity and mortality, especially in young people, and the treatment of cerebral trauma is a subject of considerable importance in clinical medicine. In this regard, I have been asked to comment on the validity and clinical relevance of a series of experiments conducted over the last 12 years at the University of Cincinnati. I should explain, at the outset, that my orientation is more toward "clinically useful", as opposed to what might be termed "background" or basic research. Nevertheless, I question whether this research is valid from anyone's point of view.
The Tornheim Protocol
The experiments involved inflicting head injuries on domestic cats according to the following protocol:
1. Cats were obtained from unspecified sources and maintained in the laboratory under unspecified conditions until the start of the experiment.
2. Each cat received ketamine, an anesthetic with a duration of action of approximately 1-2 hours. Ketamine is a dissociative anesthetic, in the same class as phencyclidine (PCP). Ketamine increases cerebral blood flow and intracranial pressure, and for this reason is not used in neurosurgery. The use of this drug in experiments designed to study the effects of blunt trauma on the brain was inappropriate, since the drug could bias the results in unknown ways.
3. The head was "scalped", i.e. the skin and soft tissues of the head were cut and reflected backward in order to expose the skull.
4. The head was then placed under a "Remington Humane Stunner," a device which used a .22 caliber cartridge to propel a metal plate into the skull. In order to use this device, the skull was positioned under the metal plate, and the cat's chin was placed on a wadded mass of aluminum foil. Prior to 1983, 16 square feet of aluminum foil were wadded into the chin cushion. After 1983 only six square feet were used, in order to allow the head to react more violently. Also in 1983, six gauze pads "simulating the soft tissues reflected during dissection" were placed between the cat's head and the impacting plate. No reason was given for dissecting the scalp to expose the skull in the first place. Certainly, this procedure detracted from the claim that this protocol "closely simulates the clinical situation."
5. The "Remington Humane Stunner" was fired, resulting in head trauma. The outcome of this trauma was highly variable. Sometimes there was a skull fracture. Sometimes the fracture was depressed. Sometimes the underlying brain was contused. Sometimes there was subarachnoid hemorrhage, subdural hemorrhage, or epidural hemorrhage. The cat usually stopped breathing. If spontaneous breathing did not resume, some cats were placed on a respirator, and some were discarded.
6. None of the studies discussed below adequately specified ventilation parameters. Any study which purports to examine the effects of head injury should specify ventilation parameters, since otherwise it is impossible to separate the effects of ventilation from other variables.
7. Each cat was maintained for a variable period of time (up to two days) in the laboratory. Apparently, the cats were not fed and anesthesia was often withheld. The research papers provided insufficient information to evaluate whether the cat was conscious during the experiment. When anesthesia was given, "the dose was reduced by 50% to 75% following preparation, allowing light anesthesia for the remainder of the experiment". In clinical medicine, "light" anesthesia means that the subject responds to pain.
8. After this period of time, some cats were again "anesthetized" with ketamine and their heads frozen in liquid nitrogen. The heads were then cut off and sliced into 5 mm sections. Alternatively, some cats were killed by perfusion of the brain with buffered formalin.
9. Sections of the head were examined under a low-power (dissecting) microscope. Standard pathologic sections and stains were not obtained.
The Tornheim "Model"
The stated purpose of this bizarre protocol was to "develop an experimental model suitable for testing pharmacologic agents which may modify cerebral edema." The experimenters recognized that "...ideally, the model should resemble clinical head injury" and they maintained that their method appeared to fulfill this criterion.
In order to discuss these claims, it is necessary to consider the general subject of animal models of diseases. An "animal model" of a disease reproduces, in the laboratory, certain features of a human disease. For example, M. leprae reliably produces leprosy in armadillos; drugs for leprosy can be investigated in armadillos without resorting to experiments on human beings. One necessary attribute of an "animal model" is reproducibility in a controlled environment. In the classic scientific method, one controls as many variables as possible in order to investigate the effect of a single variable.
In the Tornheim experiments, according to the investigators themselves, there was not a reproducible model of head trauma. Specifically, they acknowledged "a lack of precise morphologic reproducibility". For example, when 19 cats were subjected to a "standardized", uniform blow from the "Remington Humane Stunner":
4 Died within 48 hours.
1 Developed bilateral hemispheric contusions.
8 Developed unilateral gray and white matter contusions.
4 Developed unilateral gray matter contusions.
2 Showed no contusions.
The experimenters noted that "the thickness and contour of the cranial bones are quite variable, undoubtedly providing the basis for much of the variation in response to impact seen with our animals."
Not only did the Tornheim protocol fail to create reproducible head
injury in cats, it also failed to model human head injury. The mass of
the brain determines, in part, the forces transmitted to the brain during
trauma. The human brain weighs 100 times that of cats. Other anatomical
differences include the angle that the brain stem makes with the cerebral
hemispheres and with the cervical cord, and the shape and structure of
tentorium, sphenoidal wings, and faix. Species differences have largely defeated all attempts at modeling human head injury. For example, an obvious difference between the Tornheim "model" and human head injury is the absence, in the former, of contracoup injuries, which are generally felt to be more significant than coup injuries in human head trauma.
In their classic monograph on head injury, Jennett and Teasdale1 warn the "unwary" of "embarking on experiments of their own without due attention to the hard road already trodden by those before them." It seems that the Tornheim experimenters may have failed to heed this warning. The principles of traumatic cerebral edema were elucidated by Cannon in 19022 and Ommaya in 1974.3 The Tornheim experiments, however, neglected basic determinants of intracranial pressure.
The Tornheim "Model" of Cerebral Edema
Faced with the extreme variability of the results produced by the "Remington Humane Stunner," the Tornheim group disregarded the results of all the head trauma experiments except those which caused unilateral contusions to the brain and which were not quickly fatal. In other words, they abandoned the notion that the "Remington Humane Stunner" gave reproducible results, and elected instead to study a subset of cats in which the "Remington Humane Stunner" produced a visible lesion. Obviously, one cannot claim to be studying "head trauma" in general when in fact one selects only a subset of defined morphologic lesions, the mechanisms of which are unknown. Rather than subject the cats to the "Remington Humane Stunner," one might as well hit them on the head with a hammer (as Cannon did) -- the results would be the same. At best, the Tornheim experiments studied a subset of cats which sustained grossly similar non-fatal unilateral cerebral contusions. These injuries did not resemble any of the classical human head injuries (concussion, coup and contracoup cerebral contusion, intracerebral hematoma, epidural and subdural hematoma.) For a practicing physician, such "experiments" are clinically irrelevant.
If the Tornheim experiments did not provide a useful model of human head injury, did they provide a useful experiment in basic science? Could the Tornheim protocol have provided a defined group of cats with traumatic cerebral edema, with which one might explore the relative advantages and disadvantages of various drugs? Could the results be useful in guiding research for therapeutic interventions in humans?
In my opinion, the experiments provided no insight into the physiology of head injury. The basic problem was a lack of control of the fundamental parameters.
In 1783, Monro4 recognized that the rigidity of the skull implies that the total intracranial volume is always the same. Consequently, an increase in intracranial pressure (e.g. from cerebral edema) depends upon a balance achieved between any process tending to add to the intracranial volume and compensatory processes resulting in a reduction in cerebral contents.
Few of these processes are recognized or controlled in the Tornheim experiments. Posture (head up or head down, a determinant of cerebral perfusion pressure) was neglected entirely, mechanical hyperventilation (the most important clinical determinant of intracranial pressure) was largely neglected, arterial and pulse pressures were seldom measured, ketamine was used throughout, the time courses seemed to be arbitrary (sometimes 15 minutes, sometimes 6 hours, sometimes 48 hours), and no coherent account of the physiology of head trauma in cats (not to mention humans) was presented.
The experimenters changed their protocol, making comparison with previous work difficult. Prior to 1982, cats were struck in the midline. After 1982, they were struck on one side of the head. A "firm" chin rest (16 square feet of crumpled aluminum foil) was changed in 1983 to a "soft" chin rest (6 square feet of crumpled aluminum foil) in order to increase rotational forces. Experiments by Denny-Brown and Russell in 19415 showed that rotational injuries were more important than direct blows in producing brain injury. Despite the attempt to produce shear forces experimentally, only local contusions were produced. Data collected under the "revised" protocol still failed to control major determinants of intracranial pressure.
Notes on Specific Publications
1. "Effect of Dexamethasone on Cerebral Edema From Cranial Impact
in the Cat" (1978)6
The experimenters were unable to demonstrate any effect of dexamethasone on head-injured cats. This meant that either (a) dexamethasone had no effect, or (b) dexamethasone had an effect, but the experimental design failed to demonstrate it. Many clinicians believe that steroids are beneficial in the treatment of head trauma.
The investigators claimed that "we used an experimental model designed to simulate the trauma of impact seen clinically," and then they used an experimental protocol in cats which appears to have few similarities with clinical head trauma. For example, dexamethasone was administered intraperitoneally and intramuscularly in extremely high doses, whereas in clinical medicine it is given IV in lower doses.
The experimenters noted that the "pathological results of cranial impact in these cats were not highly reproducible" and proceeded to select only those animals suffering unilateral contusion for study. As noted above, this invalidated any application of the results of this study to clinical medicine.
2. "Effect of Furosemide on Experimental Traumatic Cerebral Edema" (1979)7
The experimenters claimed to study "experimental conditions designed to mimic the clinical situation as closely as possible." For example, "treatment was begun 1 hour after head trauma to comply with the time gap required for transportation and evaluation of a patient." In keeping with their "clinical" orientation, the experimenters did not provide anesthesia for any of the cats during the 48 hours following surgical exposure of the dorsal surface of the skull and head injury with the "Remington Humane Stunner." As usual, few of the variables known to influence intracranial pressure were controlled.
One group of cats was allowed "ad lib" access to water for two days. These cats drank an average of 7.7 ml/lb/day, i.e. about 2 ounces per day for a 10-lb cat. A normal cat needs about 38 ml/lb/day of water (12 ounces for a 10-lb cat).8 Thus, these cats were too sick to drink adequate amounts of water, and indeed they lost nearly 8% of their body weight over two days.
In the "experimental" group, the researchers provided 10 ml/lb/day of saline by clysis for two days. This grossly inadequate fluid intake resulted in weight loss averaging 8% of total body weight. Some of the animals were given intramuscular furosemide during this period. These animals lost an average of 11% of total body weight.
This protocol is very different from a "patient" who is transported and evaluated following head injury. In a modern ICU, fluid balance is of critical importance in management; fluids are given IV, not by clysis; drugs are given IV, not IM, and patients are appropriately sedated.
Since no data were given, it is impossible to comment on the validity of the statistical analysis. If we accept it at face value, however, the results seemed to indicate that furosemide reduces intracranial pressure. However, this effect of furosemide was known prior to this experiment, and thus there appear to be no new insights from the project.
3. "Acute changes in regional brain water content following experimental closed head injury" (1981)9
In this study, the cat's head was positioned eccentrically beneath the "Remington Humane Stunner" for unspecified reasons. As usual, many animals were excluded due to several criteria, including prolonged apnea, early death, "poor respiratory patterns following injury," and results of gross pathology. The authors concluded that head injury causes edema. Although head injury probably does cause edema, this conclusion was not supported by their data. The authors measured specific gravity of tissue from selected cat brains, and they included only animals with hemorrhages in the study. Thus, the correct conclusion was that traumatic hemorrhage in cat brains is associated with edema. This was not a surprising finding; indeed it was already known.
4. "Acute responses to blunt head trauma" (1983)10
This was the first of the "revised method of injury" papers produced by the Tornheim group. The experimenters placed the cat's chin on a softer pedestal and struck the skull eccentrically, in order to produce the rotational forces which are known to determine the amount of brain damage sustained in human subjects. It is by no means clear that these rotational forces are important in the cat.
The results indicated significant differences between the cat and human. All cats showed subarachnoid hemorrhages (this is unusual in humans). Skull fractures, contusions and subdural hematomas were common in cats (in humans, skull fractures and contusions are common, and subdural hematomas are rare). Epidural hemorrhages were 2/3 as common as subdural hemorrhages in cats. (They are extremely rare in humans). Bilateral contusions were rare in cats. (They are the rule in humans).
The authors concluded that "the experimental head injury model described in this study is clinically relevant and causes a combination of primary intracranial events that are seen frequently in human head trauma." From my knowledge of the literature and from clinical experience, this is factually incorrect. If anything, the experiment showed the differences between cats and humans in their response to head trauma.
5. "Acute responses to experimental blunt head trauma" (1984)11
In another "revised trauma method" paper, the contralateral hemisphere (which was expected to be injured by rotational forces) was used as a control. They selected only cats with left cerebral contusions involving the ectosylvian gyrus, and they excluded cats with prolonged post-trauma apnea, early death, and extracerebral hematomas. This selection process invalidated any claim that the study related to clinical head trauma. The mechanisms of injury were irrelevant here, since the experimenters selected a subset of cats based on post-hoc gross anatomic criteria.
Cat brains were examined at 15 minutes and 6 hours after injury for evidence of cerebral edema. After the authors acknowledged a number of problems with the methodology, they suggested that there might be brain edema following head trauma. This, of course, was already known. I find no new or useful data in this paper, which highlights the problems involved in using the Tornheim model to study cerebral trauma, and I disagree with their final conclusion that "further investigations focused on mechanisms of pathology are warranted."
6. "Acute changes in regional cerebral metabolite values following experimental blunt head trauma" (1985)12
In this paper, the experimenters acknowledged that "possible exacerbation of the injury by systemic events such as hypoxemia and hypotension, which of themselves can affect cerebral metabolism, confound interpretation of these alterations in evaluating the pathophysiology of human head trauma." Variables such as body temperature, arterial pO2 and pCO2, arterial blood pressure, and intracranial pressure were measured.
In the introduction to the paper, the authors discussed "concussive trauma" as though it were the subject of these experiments. Concussive head injury has never been reproduced in experimental animal models (Jennett and Teasdale, 1982), and it was not reproduced here. The report concerned cerebral contusion, not concussion.
As usual, the results of experimental head trauma with the "Remington Humane Stunner" were too variable to permit easy analysis. The authors selected only slightly more than half of the cats for inclusion in their study. So-called "Grade 0" cat brains (a total of 5), which had no visible brain tissue hemorrhage, were compared to so-called "Grade 2" cat brains (a total of 8), which had subarachnoid hemorrhage plus other trauma. Some had skull fractures, some had epidural hematomas, some had subdural hematomas, and some had intracranial mass effects. In other words, "Grade 2" cats were a heterogeneous group of animals who had suffered severe head trauma. The authors reported that in these cats, cerebral metabolism was impaired at 15 minutes. This was expected. The authors "failed to find a correlation between the size of the contusion, the presence of cranial fracture, or the systemic responses to trauma as compared with the degrees of cortical metabolite alterations in individual cats."
The paper contained several pages of discussion on the difficulties of conducting controlled experiments in this setting. It concluded, as usual, that "further studies are warranted".
I do not think that further studies by the Tornheim group are warranted. A review of their work reveals that it has been unproductive and wasteful of public funds. For the past 12 years, this group has produced a series of non-results, clinically irrelevant observations, and imprecise experiments using a bizarre protocol. Many of the pertinent physiological parameters were uncontrolled in these experiments, the consequences of head trauma were highly variable, and much of the data was excluded. As a practicing Emergency physician, I find the results irrelevant to patient management.
1. Jennett B, Teasdale G: Management of Head Injuries. Philadelphia, F.A. Davis, 1981.
2. Cannon WB: Cerebral Pressure Following Trauma. Am J Physiol 1902;6:91-121.
3. Ommaya, AK, Gennarelli TA: Cerebral concussion and traumatic unconsciousness: correlations of experimental and clinical observations on blunt head injuries. Brain 1974;97:633-654.
4. Monro A: Observations on the Structure and Function of the Nervous System. Edinberg, Creech and Johnson, 1783.
5. Denny-Brown D, Russell WR: Experimental Cerebral Concussion. Brain 1941;64:93-164.
6. Tornheim PA. McLaurin RL: Effect of dexamethasone on cerebral edema from cranial impact in the cat. J Neurosurg 1978;48:220-227.
7. Tornheim PA, McLaurin RL, Sawaya R: Effect of furosemide on experimental traumatic cerebral edema. Neurosurgery 1979;4:48-52.
8. Catcott EJ: Feline Medicine and Surgery. American Veterinary Publications, 1975.
9. Tornheim PA. McL.aurin RL: Acute changes in regional brain water content following experimental closed head injury. J Neurosurg 1981;55:407-413.
10. Tornheim PA, Boleslaw HL, Hirsch CS, Brown DL, McLaurin RL: Acute responses to blunt head trauma. J Neurosurg 1983;59:431-438.
11. Tornheim PA. Prioleau GR, McLaurin RL: Acute responses to experimental blunt head trauma. J Neurosurg 1984;60:473-480.
12. Wagner KR, Tornheim PA, Eichhold MK: Acute changes in regional cerebral metabolite values following experimental blunt head trauma. J Neurosurg 1985;63:88-96.