Maxillofacial fractures and craniocerebral injuries – stress propagation from face to neurocranium in a finite element analysis

Generally it is very difficult to perform studies concerning traumatic scenarios in
a realistic and valid set-up because of technical and ethical reasons. The well-known
and fundamental studies published by Le Fort in 1901 16],17] would not be practicable because of absence of cadavers for studies. Moreover, Le
Fort’s study design supposably would hardly pass an institutional review board today.
The validity of cadaver studies is limited, as the specimen will have undergone postmortal
alterations and in most cases will be destroyed by impacting trauma so that results
are not really reproducable. Experiments in laboratory rhesus monkeys concerning cerebral
concussion after sagittal plane angular acceleration deliver information about the
viscoelastic behaviour of bridging veins 18], but are only partially transferable to biomechanics of the human skull and impacting
forces in a real trauma. The use of animal models can be discussed controversially
in general.

About thirty years ago first steps were made towards finite element analysis, which
was limited by computing capacity. This led to rather simple 2D-models 19], data derived from cadaver data in most finite element models 20],21].

Approaches have been made to analyse brain-skull interaction during and after trauma.
Here it is highly desirable to have a very fine finite element model of the skull,
where all tissues and their mechanical behaviour could be incorporated.

Unfortunately such a model does not exist by now. It is very difficult to gain valid
information concerning biomechanical properties of involved tissues, which probably
differ depending on age, gender and ethnicity. Even if these questions were answered
and all material parameters were available, such a precise model would require an
enormous computing capacity. Another important point is the origin of the data used
for generating a finite element model. Still most models presented in literature derive
from cadaver data. Postmortal alterations and the fact that the age of cadavers mostly
is not the typical age of patients suffering from facial skull fractures is often
neglected.

Zong et al. 22] presented a three-dimensional finite element model of the human head, which consisted
of brain, inner and outer layer of the skull, diploë, cerebrospinal fluid and cervical
elements. Even if this model might be sophisticated concerning different tissues,
it consists only of a small number of elements and is rather simplified. Nevertheless,
it delivers valuable information, as it showed dynamics by using a vector quantity
to display power flow in magnitude and direction. Moreover they showed that a power
flow exists in frontal impacts in three directions, namely to the skull base, along
the cranial vault and in direction to the brain. These findings correspond well with
our own findings, where stresses propagate from facial skull to viscerocranium.

A sophisticated FE model of the skull, containing scalp, outer table, spongious bone,
inner table, cerebrospinal fluid and brain was presented by Hamel et al. 23] in a forensic study about skull fractures caused by falls in 2013. This model consisted
of 497,000 elements and data derived from the CT-scan of a 30 years old male, which
is in line with our own finite element model. Unfortunately, their study reports only
about skull fracture but not about accompanying brain injuries.

In 2013 Mao et al. reported on another very detailed and high quality finite element
human head, which was integrated into the Global Human Body Models 24]. This consisted of skull, brain, falx, tentorium, cerebrospinal fluid spaces and
even bridging vein meshes. Data derived from the CT-data of an average American. The
model had 270,552 elements in total. Even in this highly validated FE model the authors
point out that mechanical characteristics of skull-brain interface structures are
not fully understood by now, especially how they interact under in vivo conditions.

There is no controversy that brain damage may not only result from direct trauma to
the brain tissue like in open brain or in missile injuries, but also from indirect
trauma. In most cases intracranial haemorrhage will be found causative for brain damage.
According to pathological data published by Crooks 25] extradural haematomas count for five to fifteen percent, subdural haematomas for
26 to 63% and intracerebral haematomas for fifteen percent in severe head injuries.
The most prevalent haemorrhage derives from tearing of bridging veins. Supposably
similar rates will apply to non-pathological cases.

A threshold for rupture of vessels cannot really be defined, but it is known that
the risk of vessel rupture increases with angular acceleration. Concerning time factor
it could be shown that an acceleration pulse greater than 5 msec would also result
in failure of the visco-elastic behaviour of bridging veins 18],25].

Another reason for brain damage could be axonal injury. Unfortunately it is even more
difficult to characterize a threshold for this. Attempts have been made in using small
26] and large animal models. Miller et al. analyzed the relationship between lesion patterns
as a sign for diffuse axonal injury and loading conditions in a minipig model 27], but here the pigs underwent repeated tangential acceleration. It seems doubtful
that such a study design really correlates with direction and extent of forces applied
to human beings in assaults or even car accidents. Bain et al. tried to characterize
thresholds for traumatic axonal damage in a guinea pig model 28]. They analyzed the dynamic optic nerve elongation and concluded that axonal thresholds
deriving from this analysis can be directly applied to human head injury. Obviously
it is a noble goal to gain threshold properties for several tissues to have the possibility
to integrate them into FE models, but we have certain doubts, whether tearing on a
guinea pig’s eye will deliver the desired information. So one always will get to the
basic problem that it is extremely difficult to simulate traumatic scenarios in a
valid and reproducable manner.

The finite element model presented in this study derived from the CT-data of a 34 years
old man and consisted of nearly 740,000 tetrahedrons. This represents a very high
resolution. As individual bone parameters according to grey scale values had been
attributed and a transient mode of simulation had been chosen, this resulted in a
model, which is supposed to give valid and reliable information concerning stress
propagation in the human skull.

Our results show stress propagation from facial skull towards skull base in impacts
that would cause fractures of the infraorbital rim, the orbital floor and in the nasoorbitoethmoid
region. That corresponds to types of fractures which are frequently encountered in
maxillofacial surgery. Stresses reach about 150 megapascals in the occipital bone
and about 100 megapascals in the skull base.

150 megapascals are seen as a threshold for facial bone to fail and fracture 14]. As occipital bone is by far thicker than facial bone and forces required for fractures
of the occipital bone are tenfold to forces required for fractures of facial bone
29], this does not mean that bone will also fracture in the occiput, but it is a strong
hint that there are considerable stresses, which might cause brain damage and laceration
or disrupture of bridging veins. Stresses of 150 megapascals correspond to a typical
single fisticuff. A further interpretation of these results is difficult. Brain injuries
are expected, when the peak brain pressure is higher than 173 kPa 30]. But there is no possibility to conclude from stress in bone to brain pressure directly,
as little is known about the mechanical behaviour of skull brain interface. Less is
known about the clinically relevant behaviour of bridging veins. Even in more recent
studies, which use highly developed FE models with simulation of all tissues, no conclusions
about propagation of stresses from bone to bridging vessels and brain are possible
31]. Moreover most studies deal with direct impact to the skull and the brain beneath,
whereas the subject of our study focusses on stresses propagated from facial impact
to skull base. Further studies to deal with these questions are required.

Concerning impacts to the supraorbital arch we have seen that there is nearly no propagation
of stresses to the skull base in comparison to the infraorbital rim and the nasoorbitoethmoid
region, although impact was identical and produced fractures at the side of impact
in all three scenarios. Stresses seem to be absorbed in the supraorbital region. So
the supraorbital arch is a structure, which is able to carry loads from impacts and
to protect skull and brain. This phenomenon has been reported in an earlier investigation
32].

The FE model used in this study has its limitations, as it is a model consisting only
of midfacial and skull bone, but not of the brain and other head tissues. Nevertheless,
it delivers valid information about stress propagation within the skull. Its informative
value is supported by many studies about prevalence of craniocerbral injuries and
even deaths in patients with facial fractures. Thorén et al. reported on associated
injuries in patients with facial factures and found that a quarter of these patients
had associated injuries, of which 11% were brain injuries 33]. Kaiser published a case report on death in an assault victim presenting with a fracture
of the orbital wall and lacerations of the chin because of extensive basal subarachnoidal
bleeding 34].

There is discussion whether the midface has the function of a cushion and might absorb
forces to the facial skull to protect the brain, or whether forces will propagate
in direction to skull base and brain 35],36]. Keenan et al. presented a case–control study of 3849 injured bicyclists and five
scene deaths and found an odds ratio of 9.9 for the risk of intracranial injury in
association with facial fractures. They interpreted facial fractures as signs for
increased risk of brain injury 36]. This is in accordance with our own findings concerning stress propagation.

Adamec et al. concluded in their study about the injury risk of a headbutt that a
headbutt, which might be comparable to a fisticuff, will unlikely cause lethal injuries,
but they also point out that under certain cirumstances, e.g. support of the victim’s
head when standing against a wall or lying on the floor, life-threatening injuries
could occur 29]. As they used volunteers, who were obliged to perform a headbutt, for their biomechanical
studies in our opinion occurring forces might have been smaller than in real assault
situations. According to our own finite element studies and also according to clinical
experience headbutts and fisticuffs definitely may lead to fractures and brain contusion.
Salentijn et al. also saw a clear association of facial trauma with traumatic brain
injury 6],7].