Could the Fornix Be Used To Predict Cognitive Impairment In the Elderly?

KEYWORDS:

• biomarkers
• cognitive impairment
• fornix
• memory
• normal aging
• white matter tracts

The answer is “yes, it‘s conceivable”. The question suggests an attractive possibility that could become feasible with much more research. In addition, perhaps even more importantly, it implies a shift in early-stage indicators of possible future cognitive impairment, from brain gray matter pathology to early, subtle changes in the white matter. It beckons to a deeper understanding of how the brain maintains its communication networks and how these networks begin to fail. Such a shift could yield important pieces in the developing picture of aging brain processes, including trajectories leading to cognitive impairment. It could also provide clues for lifestyle and intervention strategies to minimize such an outcome.

But to realize these goals – to use brain structures as clinically relevant predictors and to extend our understanding of brain dynamics in early cognitive decline – current questions about the technology to acquire and analyze brain white matter imaging will have to be resolved.

The fornix is a small, wishbone-shaped brain structure providing a main efferent pathway from the hippocampi to the rest of the brain [1]. It is well known that the hippocampi are involved in creating spatial and episodic memory. The fornix consists mainly of myelinated axons from neuron cell bodies in specific areas of the hippocampi [2], connecting them to the mammillary bodies, which, in turn, connect to the thalamic nuclei as part of the Papez circuit [1]. Linking hippocampal output to the rest of the brain, the fornix is crucial in the formation of episodic memory (recall of specific events or things) but interestingly, not so much for familiarity, which seems to be implemented by another circuit (think of the experience of seeing a face that looks familiar, but being unable to recall the person‘s name. While common in healthy people, this may be accentuated in those with fornix damage) [3,4].

As the fornix is involved in memory, it should not be surprising that it is also involved in cognitive impairment. Recent articles have shown that among subjects who have mild cognitive impairment (MCI), the state of the fornix may be a predictor of whether they will later progress to full Alzheimer‘s disease (AD) [5,6]. The fornix of subjects with AD differs significantly from that in normal healthy subjects [7]. An article from our laboratory has now extended the predictive sensitivity of the fornix to earlier in the trajectory of cognitive aging [8]. We demonstrated that among cognitively normal elderly subjects, the fornix can statistically predict future conversion (clinical rediagnosis) to MCI or AD.

However, our paper was not the first to mention the fornix in conversion from normal to MCI. Research from the University of New South Wales (Australia) in 2012 found white matter changes in the fornix, as well as in the parahippocampal cingulum and precuneus that predicted future decline from normal cognition [9]. A more recent result from this group unambiguously stated that white matter changes, rather than hippocampal atrophy, characterized subjects who had early amnestic MCI (MCI with memory impairment) [10]. Both of their results used diffusion tensor imaging (DTI) to measure microstructural white matter integrity, whereas ours focused primarily on measuring fornix white matter volume and only secondarily on fornix DTI measures. We also used different statistical models to compute the strength of factors predicting conversion. This highlights an issue of methodology – the technology for assessing white matter integrity – which I will return to later on.

The overlap of our and their research projects raises vital questions for both imaging methodology and brain biology. First, there is a strong unified message that brain white matter structures may be key players in early cognitive decline. This contrasts with the well-documented involvement of gray matter atrophy during later stages of impairment. What could be the reasons for this? There are at least two possible hypotheses. One is the ‘Wallerian degeneration‘ hypothesis that loss of neuron cell bodies (in the gray matter) leads to axonal deterioration (in the white matter), but due to limitations of current imaging, we can detect the latter before we see the former. In other words, gray matter atrophy really is driving losses in the white matter too, but we cannot see the gray matter atrophy at early stages. A second possibility is that another process independent of the gray matter is directly affecting the white matter. Recent mouse studies have shown that elevated levels of β-amyloid – the protein also implicated in the development of gray matter atrophy in AD – can directly interfere with the action of oligodendrocytes in maintaining myelin, the insulating sheath of axonal bundles [11]. Degeneration of myelin could degrade signal transmission without actually disconnecting brain regions.

The New South Wales group has argued against the Wallerian degeneration hypothesis [9,10,12], since they found no statistically significant gray matter changes associated with cognitive conversion among normal subjects. Previous research in our laboratory did find an association with hippocampal atrophy and fornix degeneration in subjects at later stages of cognitive impairment, MCI and AD [13]. But this does not necessarily show causation. Perhaps the hippocampi and fornix were both responding to elevated levels of β-amyloid according to the second hypothesis, and in those subjects already cognitively impaired, the brain changes were sufficiently marked that we could detect atrophy of both gray and white matter structures. In summary, the question of the pathological processes underlying the very earliest brain changes remains a compelling open question.

A related biological question arises from the specific white matter structures found to predict loss of cognition [8,9]. What is ‘special‘ about the fornix, the precuneus or parahippocampal cingulum? Are they particularly vulnerable, and if so, why? Are there other relevant structures that have not yet been identified? The fornix, precuneus and parahippocampal cingulum are all associated with memory [9]. A recently proposed hypothesis for amnestic MCI subjects is that these individuals begin with memory as the sole affected function and progress to impairment in other domains, such as executive control [14]. This progression is paralleled in the brain, starting with a common area of white matter demyelination – in the splenium of the corpus callosum and neighboring areas of the posterior corona radiate – and spreading to other areas in the frontal and temporal lobes in predictable patterns. In brief, progression from amnestic MCI to AD “starts with a single source and follows a common scenario” [14]. This is the framework I would like to see applied to the entire trajectory from normal cognition to AD. Does there exist a predictable sequence of white matter deterioration, beginning before any clinical manifestations, in predictable location(s), such as the fornix and other memory-related structures, and then progressing in a typical pattern to other brain structures?

The focus on incipient change in clinically normal subjects is a start towards filling in this framework [8,9]. However, brain changes in the preclinical stages can be subtle and hard to detect, which explains why there are many more papers studying the MCI and AD stages in which brain changes are more pronounced. This highlights questions regarding the current methodology for assessing white matter. DTI, the most popular approach, is so-named because it measures the preferred directions of Brownian motion or diffusion of water molecules in the brain. Along highly structured pathways, such as myelinated axonal bundles, water diffusion should be much stronger parallel to the bundle than perpendicular to it. If this is the case, a measure called fractional anisotropy is high. But if DTI picks up a significant component of diffusion perpendicular to the bundle, this might suggest deterioration of the insulating myelin, like a defective hose creating puddles from water leaking sideways [15]. Somewhat paradoxically, if the parallel diffusion component is especially strong without fractional anisotropy also being high, this may indicate that the axon itself has broken down inside the myelin, leaving a clearer pathway along which motion may occur – as if any obstructing gravel inside the hose has been cleared out [16].

Therefore, DTI would seem to hold promise for distinguishing between two microscopic causes of white matter degeneration – loss of myelin and loss of axons. However, there have been contradictory results, showing both increases and decreases of parallel diffusivity among MCI and AD subjects [17,18]. There also remains controversy about DTI image acquisition and the meaning and most appropriate uses of the DTI indices. In addition, DTI gives typically lower-resolution images, which require sophisticated techniques (themselves not uncontroversial) to enhance statistical power.

In summary, the fornix and other white matter tracts are sensitive indicators of incipient cognitive decline. We cannot yet use the fornix as a clinical predictor of cognitive loss in normal subjects. We don‘t yet fully understand the fascinating questions implied by such a possibility: why the fornix and other white matter tracts may potentially be such markers; how they are maintained; how they fit into a broader communication network that sustains cognition; and why they may be vulnerable. With time and matured technology, however, we will, and then we can.

Financial & competing interests disclosure

The author has no relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties.

No writing assistance was utilized in the production of this manuscript.