Alzheimer's disease (AD), the most common form of dementia in the elderly, is thought to be caused by an imbalance between amyloid-β (Aβ) production and clearance leading to Aβ accumulation in the central nervous system (CNS), which causes neuronal damage and death, manifesting as progressive clinical dementia [1–3]. It has been shown that patients with AD have 30% slower clearance of Aβ [1]. One of the possible etiologies of decreased Aβ clearance is impaired cerebrospinal fluid (CSF) flow [1, 4]. When venous hypertension occurs in the superior sagittal sinus, CSF absorption is also impaired, leading to altered CSF outflow [5–7]. Jugular venous reflux (JVR) characterized by a retrograde flow in internal jugular veins (IJVs) during Valsalva-like manoeuvres (VM) or spontaneously, is found more frequently in the elderly [8, 9]. Studies have shown that JVR can retrogradely transmit hypertension into the cerebral venous system and that it is associated with white matter (WM) abnormalities in the elderly [10, 11]. Consequently, there is reason to believe that JVR might alter CSF absorption, and that this in turn might influence the clearance of Aβ. Given this, we hypothesized that JVR might be associated with mild cognitive impairment (MCI) and AD, and that this association might manifest itself in structural changes in the brain parenchyma. To this end, we undertook a case-controlled study to explore the issue by evaluating the relationship between JVR and global and tissue specific brain parenchyma volumetric parameters. Volumetric analysis of brain parenchyma structures measured on serial magnetic resonance imaging (MRI) scans has been shown to provide an objective and quantitative method for examining neuropathological changes associated with AD [12–18].

The results of the study do not support the hypothesis that JVR is specifically associated with MCI and AD. The incidence of JVR was very similar in both the control and MCI groups, and was actually lower in the AD group. Having said this, the results suggest that JVR is associated with a rather unexpected phenomenon. JVR appears to be associated with structural changes in the brain parenchyma of patients with MCI and AD that were not observed in the control group. This is highlighted in the results presented in Table 3, which revealed a marked difference in response to JVR between the controls and the other two groups. Overall, the subjects with JVR had greater WB and GM volumes, which was accompanied by decreased CSF volume, compared with those without JVR (Table 2). This effect was particularly marked in the MCI and AD groups, whereas it was absent in the controls (Table 3). As such, the observation that AD patients with JVR exhibit larger WB volumes is a surprising finding, as AD is normally characterized by advanced brain atrophy. This can be clearly seen if one analyzes the JVR-negative and positive groups separately. In the JVR-negative subjects there was a statistically significant reduction in global and tissue specific brain parenchymal volumes in the AD and MCI groups compared with the controls - a finding that is consistent with the observations of many other researchers [12–15, 24]. However, no corresponding reduction was observed in the JVR-positive group, implying that in some way JVR inhibited brain volumetric loss in the AD and MCI subjects. Compared with insufficient cerebral arterial supply (arterial ischemia), cerebral venous drainage impairment with venous hypertension causes more severe vasogenic edema and brain–blood barrier damage [5–7]. Previous studies of JVR provide evidence that retrograde-transmitted venous hypertension from JVR can reach the cerebral venous system [10, 11]. It is therefore possible that JVR retrogradely transmits venous hypertension into the brain, leading to increased permeability of the blood–brain barrier (BBB), resulting in vasogenic edema, causing the brain tissue to swell. Disruption of the BBB will allow plasma molecules to pass into the brain, with the result that an osmotic pressure gradient is established which will contribute to edema formation. While it is not known if this mechanism is at work, it is noticeable that JVR was associated with a marked reduction in CSF volume in the MCI and AD subjects, something that would be consistent with an influx of CSF into the parenchymal tissue. Alternatively, JVR might promote the retention of blood in the cerebral veins [25] – something that might increase the volume of the brain parenchyma.

The results of the ROC analysis demonstrate that it is possible to discriminate between the JVR-positive and negative MCI and AD subjects with reasonable accuracy using just the MRI variables, normalized WB volume and normalized GM volume, whereas this was not the case in the control group. As such, this finding appears to corroborate those of the univariate analysis. Furthermore, the ROC analysis suggests that a progressive effect may be occurring, which is stronger in the AD group (area under curve (AUC) = 0.794; p = 0.007) than in the MCI group (AUC = 0.713; p = 0.012). While this finding is difficult to interpret, it is known that MCI is frequently a precursor to AD [26].

While the exact physiological mechanisms behind our intriguing observations are unclear, it is known that JVR can induce hypertension in the dural sinuses [10, 11] and that this can alter intracranial CSF dynamics [27]. Therefore, it may be that retrograde-transmitted venous pressure associated with JVR inhibits CSF absorption into the superior sagittal sinus [28]. Absorption of CSF into the dural venous sinuses requires a pressure gradient of about 5–7 mmHg [29, 30]. Therefore, an increase in venous pressure of few mmHg due to occlusion of the venous drainage pathways [31], or reflux, will tend to inhibit the bulk flow of CSF, as observed by Zamboni et al. [32]. If CSF flow is inhibited, then this might result in increased biochemical concentrations in the CSF. Overproduction of Aβ is thought to damage WM in AD [33]. Given that patients with AD have been shown to exhibit 30% slower Aβ clearance [1], it has been postulated [4] that accumulation of Aβ in the CSF, arising from venous hypertension, may precipitate the onset of AD. However, our results do not support this conclusion. While JVR may be associated with accumulation of Aβ in the CSF, our results do not indicate that JVR precipitates the onset of AD. Indeed, the majority of AD patients in our study were JVR-negative. Having said this, the results suggest that JVR is having an effect on the brain parenchyma of the MCI and AD patients, and the possibility that this might involve CSF accumulation of Aβ cannot be excluded. Further investigations are therefore needed to elucidate the underlying neuropathological mechanisms associated with our observations.

In this study we focused solely on JVR and ignored other phenomena associated with constricted cerebral venous outflow [34]. However, it may be that restricted venous outflow, such as that associated with chronic cerebrospinal venous insufficiency [35, 36], might also be influential. Studies of cerebral arteriovenous malformation have shown that the elevated venous pressure and its insults to intracranial structures are more severe when combined with obstruction in other venous outflow tracts [37, 38]. Therefore it may be the case in JVR, that retrogradely-transmitted venous pressure into cerebral circulation needs additionally an obstruction of contralateral venous outflow pathway to cause significant venous hypertension and consequently intracranial insults. It is important to remember that a diagnosis of ‘no-JVR’ does not preclude the possibility that constricted cerebral venous outflow might be present. Furthermore, engorged veins are frequently observed upstream of stenotic lesions [39] and it may be that muscular compression of these veins also contributes to JVR.

Although it yielded novel and interesting findings, it should be noted that our study was limited in its scope, having a relatively small sample size. In particular, the AD group contained fewer JVR-positive individuals compared with the other two groups. Also, because of the limited numbers involved, we restricted ourselves to a simple positive/negative JVR classification and did not distinguish between bilateral, left and right-sided JVR. It is therefore not known the extent to which left and right sidedness in JVR influences brain atrophy and further work will be required to evaluate this. Furthermore, there are other venous abnormalities associated with aging and other neurological disorders which we were not able to assess [9, 27, 40]. Nevertheless, our findings are novel and suggest that cerebral venous drainage impairment may influence the neuropsychology of AD. Further studies, particularly longitudinal studies, are therefore needed to build on our initial findings.

JVR is associated with intracranial structural changes in MCI and AD patients, which result in increased WB and GM volumes. Although the neuropathology associated with this unexpected and counterintuitive finding requires further investigation, it may be that JVR retrogradely transmits venous hypertension into the brain, and that this leads to the brain tissues swelling due to vasogenic edema.