Syndromes of progressive neurological disturbances, including psychomotor retardation, gait unsteadiness, and urinary incontinence associated with ventricular dilation, in the setting of normal cerebrospinal fluid (CSF) pressure and the absence of papilledema, have been termed as “normal pressure hydrocephalus” (NPH) in Adams et al. (1965). Patients without known precipitating factors are diagnosed with idiopathic NPH (iNPH), the mechanism of which remains largely unknown. However, the steep increase in the incidence of iNPH in individuals who are 60 years of age or older suggests an association with aging (Jaraj et al., 2014; Martín-Láez et al., 2016). Some recent studies have emphasized on the primary role of abnormal water/blood drainage (Williams, 2008; Bateman and Siddique, 2014) or viscoelasticity changes in the brain parenchyma (Owler et al., 2004; Sack et al., 2009) as the likely mechanisms underlying age-related development of the disease (Keong et al., 2016).

Nevertheless, since the initial reports (Hakim and Adams, 1965), the immediate improvement in symptoms following removal of CSF through a lumbar tap has not only been useful for clinical purposes, but has also suggested abnormal perfusion as the direct cause of clinical manifestations (Wikkelsø et al., 1982). Despite the body of evidence demonstrating changes in blood flow following the “tap test” (TT), there are no established diagnostic criteria based on blood flow imaging using single-photon emission computed tomography (SPECT) or positron emission tomography (PET) (Relkin et al., 2005; Keong et al., 2016). It is critical that iNPH be diagnosed sufficiently early to enable CSF diversion using a ventriculoperitoneal or lumboperitoneal shunt (Gallia et al., 2006) where appropriate to prevent irreversible damage. Thus, there is a need for novel, non-invasive techniques to assess this condition in the elderly population.

Recently, mapping the low-frequency phase in a blood oxygenation level-dependent (BOLD) signal time-series has been proposed as a clinically useful biomarker in cerebrovascular diseases (Amemiya et al., 2013; Lv et al., 2013; Christen et al., 2015; Ni et al., 2017). This method is based on a low-frequency oscillation of systemic origin (sLFO), which can be similarly detected using near-infrared spectroscopy from a fingertip and BOLD signals from the brain. There is a constant phase difference across body parts, indicating that this fluctuation of hemoglobin content travels throughout the vasculature (Tong and Frederick, 2012, 2014). The method, termed “lag mapping,” has recently been confirmed to reflect cerebral blood flow (CBF) in general, with high sensitivity to the venous part of the circulation (Tong et al., 2017). Several reports have already indicated that this information can be reliably extracted using conventional functional magnetic resonance imaging settings at a repetition time of 2–3 s, and a duration of 3 min (Anderson et al., 2011; Amemiya et al., 2013). Importantly, because sLFO has its origin in the systemic circulation, this information primarily reflects vascular structure and not neurovascular coupling (Anderson et al., 2011; Taylor Webb et al., 2013; Erdoǧan et al., 2016; Aso et al., 2017a). Moreover, because it is a blood-tracking technique, changes in venous drainage patterns should be reflected in the results. Due to the lack of valves in the cerebral veins, flow direction is known to be reversible across collaterals (Gisolf et al., 2004; Andeweg, 1996).

In the present study, we acquired resting-state BOLD magnetic resonance imaging (MRI) scans before and after a spinal TT, and compared the BOLD lag maps to evaluate the effect of treatment on brain perfusion. Taking advantage of the high sensitivity of the technique in the venous side of the vasculature, changes in drainage time in each part of the brain were measured. Furthermore, we compared the results from patients and healthy control subjects. Our study is the first to apply lag mapping to detect divergence in cerebral blood circulation between the pre- and post-TT states in subjects with iNPH, and to compare it with processes in normal aging, thereby providing new insights into the mechanisms underlying the etiology of iNPH.

Using a BOLD-based blood-tracking method in subjects with iNPH, we observed an abnormal phase in the periventricular region where the deep veins converge. Under healthy conditions, the phase or relative drainage time in this region consistently exhibited a late venous phase, as reported earlier (Lv et al., 2013; Tong and Frederick, 2014). This abnormally long drainage or “wash-out” time in iNPH was normalized by TT, while the global mean of the phase remained stable. Collectively, these results permit an interpretation that a part of the deep venous system is drained by collaterals in iNPH instead of the normal route via the ICV. The broad change after TT may reflect the normalization of this state, involving a change in the drainage pattern (Andeweg, 1996). Altered venous drainage has been observed in chronic NPH (Kuriyama et al., 2008; Bateman and Siddique, 2014) and the periventricular area may be one of the commonly affected sites of this venous inefficiency.

This interpretation is consistent with a variety of earlier observations. Greitz (1969) reported an increase in ICV size following a shunt operation in three of seven patients, thus suggesting venous collapse before treatment. A similar abnormality in the ICV has also been associated with some types of hydrocephalus (Boller et al., 1970). In relation to hemodynamic dysfunction, Momjian et al. (2004) used PET and reported abnormal CBF with a maximal reduction adjacent to the ventricles. Our results are also in line with those of several SPECT studies demonstrating an increase in CBF after lumbar puncture (Waldemar et al., 1993; Mataró et al., 2003). A recent study investigating the acute effect of TT demonstrated a relationship between clinical improvement and increases in CBF in a similar region using arterial spin labeling (ASL), an alternative, highly reproducible method (Virhammar et al., 2014). Interestingly, Walter et al. (2005) reported a higher time-to-peak than the arrival time following TT in the periventricular area, which was considered unique in this condition. As lag mapping is sensitive to the capillary and venous parts of circulation, combination with ASL may facilitate understanding of these TT-induced changes in different segments of the vasculature. It is also notable that the effect of TT would be merely a partial reversal of the iNPH pathology, which may involve a complex sequence of events (Keong et al., 2016). For example, there should have been a dynamic interplay between the intracranial pressure change by TT and recovery of blood flow, both of which have mutual interaction with impaired brain compliance in iNPH (Eide and Sorteberg, 2010; Goffin et al., 2016). The present finding suggests that the periventricular venous insufficiency, possibly causing local CBF reduction, may be one of the key factors proximal to the clinical manifestation of iNPH (Andeweg, 1996).

The prominent age-related changes in the lag map were unexpected, but may have implications in iNPH pathophysiology, primarily due to the spatial coincidence of the changes. This effect of aging cannot be fully accounted for by cerebral atrophy for two reasons. First, simple dilation of the ventricles in the elderly should result in extension of the periventricular regions, which contradicts the present observation. Erosion of the clusters into the ventricles may be due to magnetic field inhomogeneity in the perivascular CSF space, causing T2^∗ signal cancelation (Cheng and Haacke, 2001). Second, the dorsal part of the SSS and its tributaries exhibited a completely opposite change, despite the similarly enlarged CSF space due to cortical atrophy. This finding may, therefore, reflect a change in the cerebral venous systems during the normal course of aging.

A short (or negative) drainage time of a region, relative to the sLFO phase of the global mean signal, indicates its position in the venous part of the cerebral vasculature (Aso et al., 2017a). The present results, therefore, suggest that in the elderly, the superficial system moves downstream to serve as the venous outlet because the deep system is not as efficient as it was in the earlier stages of life. Moreover, the superficial clusters in Figure 5A, bottom coincide with the location of the anastomotic veins of Trolard, which connect the transverse sinus and the SSS, suggesting involvement of distant collateral pathways. In contrast to the observation in iNPH, this is a new finding and lacks explanation; however, there are lines of evidence regarding age-related stenosis or occlusion of the periventricular vasculature that can be considered (Moody et al., 1995; Farkas et al., 2006).

The fact that both normal aging and abnormalities in iNPH (which is corrected by TT) involve deep venous insufficiency may have etiological implications, as this suggests altered venous drainage in the absence of pathological ventricular dilation. Accordingly, for example, a causal relationship between hydrocephalus and periventricular edema may be questioned (Keong et al., 2016). It can also imply an initiating role of venous congestion in brain compliance reduction which develops during both pathological and aging processes (Eide and Sorteberg, 2010; Beggs, 2013). Although the concept of venous inefficiency as the cause of hydrocephalus is not new (Kuriyama et al., 2008; Williams, 2008), it has not been linked to aging. Although the role of CSF in the mechanism cannot be inferred from the present data, it is interesting that both SPMs in Figure 5A encompass regions related to CSF turnover. The lateral lacunae of the SSS, with rich arachnoid granulations, are mainly distributed in the dorsal part, although the role of arachnoid granulations in CSF turnover remains a matter of debate (Grzybowski et al., 2007). Thus, the full picture of underlying mechanisms is yet to be determined for both of these findings in iNPH and normal aging. Nevertheless, the difference in drainage capacity between the superficial and deep systems is of interest because it may possibly create a type of “transmantle pressure” in the brain parenchyma in elderly individuals, which may be a driving force in the disease process.

While we clearly demonstrated a within-subject hierarchical shift in parts of the superficial and deep venous systems, there was, at the same time, a large inter-individual variation in drainage time (Figure 5B), suggesting non-uniform drainage pathways. This is consistent with the inter-individual variability of the lag map itself, primarily arising from variations in the vascular structure. Unlike arterial arrival time, imaging of venous drainage time has rarely been performed due to a lack of optimized techniques. Therefore, our findings remain to be confirmed by other methods. It is, nevertheless, a novel advantage of lag mapping that timing can be measured in both directions using the sLFO phase as a virtual tracer directly injected into the brain vasculature (Aso et al., 2017b; Tong et al., 2017). Hence, another clinical implication of the current results is that lag mapping may provide a biomarker for diagnosing, monitoring, and even predicting the development of iNPH. Given the substantial individual variations in the lag map, an ROI analysis or within-individual, annual checkups may be necessary for screening. Our findings warrant a longitudinal study to assess the diagnostic or predictive performance of this novel marker—taking advantage of its non-invasive nature—in the management of this treatable condition in elderly individuals. Limitations of the current study include its relatively small patient cohort and lack of definitive diagnosis of iNPH in most of the cases. Although most subjects experienced clinical improvement after TT, only three underwent ventriculoperitoneal shunting (one of these patients was excluded due to high head movement). In fact, many of our patients and their caregivers finally decided not to undergo the shunt operation due to limited improvements (Iseki et al., 2014). This relatively poor clinical outcome of TT may be due to old age and advanced stage of the disease, both of which are potential sources of bias. Further research involving a larger cohort with a wider range of disease severity should be pursued.

This study was carried out in accordance with the recommendations of Ethics Committee of Kyoto University Graduate School and Faculty of Medicine and Nagahama city hospital with written informed consent from all subjects. All subjects gave written informed consent in accordance with the Declaration of Helsinki. The protocol was approved by both ethics committees.

All authors approved the manuscript version to be published. TS and TA contributed to the conception and design of this research, data analysis and interpretation, and drafting of the manuscript. SN contributed to data acquisition, data analysis, and interpretation. TKo, TU, YN, MO, NO, KY, TKi, TKu, KU, TM, and SM were involved in data acquisition and manuscript revision. HF contributed to the conception and design of this research, as well as data interpretation.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.