Back to the future: Estimating pre-injury brain volume in patients with traumatic brain injury

Highlights

• A method was developed to estimate brain volume just before the time of injury.

• TBI led to decrease in cerebral white matter volume and increase in other regions.

• The magnitude and pattern of changes was possibly unique among brain disorders.

• Regression analysis showed perfect prediction of group membership (TBI vs. normal).

• This approach may lead to a method for diagnosing mild or moderate TBI using MRI.

Abstract

Introduction

A recent meta-analysis by Hedman et al. allows for accurate estimation of brain volume changes throughout the life span. Additionally, Tate et al. showed that intracranial volume at a later point in life can be used to estimate reliably brain volume at an earlier point in life. These advancements were combined to create a model which allowed the estimation of brain volume just prior to injury in a group of patients with mild or moderate traumatic brain injury (TBI). This volume estimation model was used in combination with actual measurements of brain volume to test hypotheses about progressive brain volume changes in the patients.

Methods

Twenty six patients with mild or moderate TBI were compared to 20 normal control subjects. NeuroQuant® was used to measure brain MRI volume. Brain volume after the injury (from MRI scans performed at t1 and t2) was compared to brain volume just before the injury (volume estimation at t0) using longitudinal designs. Groups were compared with respect to volume changes in whole brain parenchyma (WBP) and its 3 major subdivisions: cortical gray matter (GM), cerebral white matter (CWM) and subcortical nuclei + infratentorial regions (SCN + IFT).

Results

Using the normal control data, the volume estimation model was tested by comparing measured brain volume to estimated brain volume; reliability ranged from good to excellent. During the initial phase after injury (t0–t1), the TBI patients had abnormally rapid atrophy of WBP and CWM, and abnormally rapid enlargement of SCN + IFT. Rates of volume change during t0–t1 correlated with cross-sectional measures of volume change at t1, supporting the internal reliability of the volume estimation model. A logistic regression analysis using the volume change data produced a function which perfectly predicted group membership (TBI patients vs. normal control subjects).

Conclusions

During the first few months after injury, patients with mild or moderate TBI have rapid atrophy of WBP and CWM, and rapid enlargement of SCN + IFT. The magnitude and pattern of the changes in volume may allow for the eventual development of diagnostic tools based on the volume estimation approach.

Introduction

Decades of research have shown that traumatic brain injury (TBI) causes brain atrophy (Bigler, 2005, Bigler, 2011). Despite this impressive body of work, brain structural studies before and after injury are rare. To our knowledge, there have been only two studies published using quantitative structural brain imaging before and after injury. In total, the 2 studies examined 2 patients with severe TBI who showed progressive brain atrophy and 4 patients with mild TBI who did not (Bigler and Snyder, 1995, Gale et al., 1995). The small number of patients and limited (by today's standards) volumetric methods may have decreased the ability to detect abnormalities in the patient group. In contrast, more recent longitudinal studies of mild or moderate TBI patients, which examined brain structure at two points after injury, have consistently found abnormalities (Hofman et al., 2001, MacKenzie et al., 2002, Ross et al., 2012a, Ross et al., 2012b, Ross et al., 2012c, Zhou et al., 2013).

The lack of studies before and after injury is understandable for several reasons. Since it is not known when an accident will occur, usually an MRI cannot be obtained just before the accident. Also it would be impractical to get baseline MRI scans on large groups of normal subjects and then study the small percentage who would have a TBI afterwards. However, it would be possible to overcome these challenges, at least in part, if it were possible to reliably estimate brain volume just before injury.

For many years, it has been known that normally the brain and skull change volume with a characteristic pattern throughout the life span (Courchesne et al., 2000). The brain and skull reach maximal volume around age 13, with the skull growing to be just big enough to cover the brain. Whole brain volume then changes relatively little overall until about age 35, when it begins to decrease. During later adulthood, the rate of atrophy progressively increases. In contrast, intracranial volume does not change during adult life. Based on these observations, Tate et al. (2011)—following the lead of Blatter et al. (1995)—showed that brain volume at an earlier point in life can be estimated reliably from intracranial volume measured later in life.

Further progress toward building a volume estimation model was achieved by Hedman et al., who conducted a meta-analysis of 56 studies (which included 2211 normal control subjects) of longitudinal change in MRI brain volume throughout the life span. Using curve-fitting regression techniques, they produced growth/atrophy curves for whole brain parenchyma (WBP), cortical gray matter (GM) and cerebral white matter (CWM). Thus, they created models which allow for accurate estimation of brain volume changes throughout the life span.

By considering combining the work of these researchers, it seemed possible to create a volume estimation method which could be used to test hypotheses about patients with traumatic brain injury (TBI). Intracranial volume could be measured in TBI patients (after the accident) and used to estimate brain volume for each patient just before the accident. The reliability of this method could be tested in a group of normal control subjects.

Accordingly, the aims of the current study were as follows: (1) develop methods for estimating brain volume throughout the life span; (2) use total intracranial volume (ICV), in combination with the Hedman growth/atrophy curves, to predict brain volume just before injury (t0); (3) compare TBI patients to normal control subjects, using longitudinal changes in brain volume (from t0 to t1), to test the hypothesis that patients have more rapid volume changes than normal control subjects; and (4) explore the relationship between longitudinal changes (t0–t1) and traditional brain volume measures (t1 cross sectional measures, and t1–t2 longitudinal measures).