The role of diffusion tensor imaging in idiopathic normal pressure hydrocephalus: A literature review

Abstract

Idiopathic normal pressure hydrocephalus (iNPH) is a syndrome that comprises a triad of gait disturbance, dementia and urinary incontinence, associated with ventriculomegaly in the absence of elevated intraventricular cerebrospinal fluid (CSF) pressure. It is important to identify patients with iNPH because some of its clinical features may be reversed by the insertion of a CSF shunt. The diagnosis is based on clinical history, physical examination and brain imaging, especially magnetic resonance imaging (MRI). Recently, some papers have investigated the role of diffusion tensor imaging (DTI) in evaluating white matter alterations in patients with iNPH. DTI analysis in specific anatomical regions seems to be a promising MR biomarker of iNPH and could also be used in the differential diagnosis from other dementias. However, there is a substantial lack of structured reviews on this topic. Thus, we performed a literature search and analyzed the most recent and pivotal articles that investigated the role of DTI in iNPH in order to provide an up-to-date overview of the application of DTI in this setting. We reviewed studies published between January 2000 and June 2020. Thirty-eight studies and four reviews were included. Despite heterogeneity in analysis approaches, the majority of studies reported significant correlations between DTI and clinical symptoms in iNPH patients, as well as different DTI patterns in patients with iNPH compared to those with Alzheimer or Parkinson diseases. It remains to be determined whether DTI could predict the success after CSF shunting.

Introduction

Idiopathic normal pressure hydrocephalus (iNPH) is a clinical syndrome that was first described in 1965 by Hakim and Adams.¹ Its prevalence in the general population is still unclear, ranging from 0.2% to 3.7% among people aged 65 and older, and more common in people aged ≥80 years (from 5.9% to 8.9%).^2–4

iNPH classically presents with gait and balance disturbance, cognitive decline and urinary incontinence associated with ventriculomegaly in the absence of persistently elevated intraventricular cerebrospinal fluid (CSF) pressures.⁵ It is important to identify patients with iNPH early because many of its symptoms may be reversed by the prompt insertion of a CSF shunt. However, iNPH can manifest as an insidiously progressive, chronic disorder, and can be easily misdiagnosed as Alzheimer disease (AD), Parkinson disease or vascular dementia, especially in early phase of the diseases.

Diagnosis of iNPH is based on clinical and neurological examination, supported by imaging findings on computed tomography (CT) and magnetic resonance imaging (MRI).⁶ Documentation of ventricular enlargement at CT or MRI is necessary but not sufficient in itself to establish a diagnosis of iNPH. It is complicated by the variability that exists in its clinical presentation and course. Accordingly, iNPH has been classified into ‘probable,’ ‘possible’ and ‘unlikely’ categories by the International and the Japanese guidelines, according to which participants in all iNPH studies since 2005 are selected.^6–8 Another important factor that makes studies of iNPH challenging is the high rate of co-existence of other pathologies. Espay et al. revealed that the initial diagnosis of iNPH was revised later in as many as 25% of the patients (AD, dementia with Lewy bodies and progressive supranuclear palsy).⁹

Classically accepted imaging criteria for iNPH include⁶: (a) ventricular enlargement not entirely attributable to cerebral atrophy or congenital enlargement (Evan’s index 0.3 or more); (b) no macroscopic obstruction to CSF flow; and (c) at least one of the following supportive features:

1. enlargement of the temporal horns of the lateral ventricles not entirely attributable to hippocampus atrophy;

2. callosal angle of 40 degrees or more;

3. evidence of altered brain water content, including periventricular signal changes on CT and MRI not attributable to microvascular ischaemic changes or demyelination;

4. an aqueductal or fourth ventricular flow void on MRI.

In particular, the callosal angle, the diameter of the temporal horns and the presence of disproportionately enlarged subarachnoid space hydrocephalus (DESH) were proposed as predictors of a positive outcome after shunting in patients with iNPH and recommended as non-invasive tools that may aid in the selection of shunt candidates.¹⁰ Ishii et al. measured the callosal angle on the coronal MR images and reported that, when using the threshold of the mean −2SD value of the normal control group (below 90 degrees), an accuracy of 93%, sensitivity of 97% and specificity of 88% were observed for differentiation of iNPH from AD patients.¹¹ DESH is considered useful in distinguishing hydrocephalus from cerebral atrophy. However, there are some iNPH patients without a DESH appearance and vice versa.¹² The cingulate sulcus sign has also been proposed as an MRI feature of iNPH; it denotes the posterior part of the cingulate sulcus being narrower than the anterior part.¹³ Recently, the iNPH Radscale has also become a valuable diagnostic tool, standardized evaluation in the workup in patients with suspected iNPH. It is applicable to both CT and MRI; a high iNPH Radscale score together with clinical symptoms should raise suspicion of iNPH.^14–15

Diffusion tensor imaging (DTI) was originally proposed for use in MRI by Peter Basser et al.^16–17 in 1994, and recently, many papers investigated its role in analyzing white matter (WM) alterations in patients with iNPH.¹²,^18–54 DTI is a non-invasive MR technique that provides information about the orientation and anisotropy of the WM tracts, and may delineate microstructural changes in cerebral WM; high-resolution, multitensor imaging (more than six diffusion directions) is currently combined with various other advanced MR techniques to study connectivity within the brain. In fact, illnesses that disrupt the normal organization or integrity of cerebral WM have a qualitative and quantitative impact on DTI measures.⁵⁵ The main quantitative parameters derived from DTI data are: fractional anisotropy (FA), which reflects the directionality of molecular displacement by diffusion and varies between 0 (isotropic diffusion) and 1 (infinite anisotropic diffusion); mean diffusivity (MD), which reflects the average magnitude of molecular displacement by diffusion; axial diffusivity, representing the largest eigenvalue; and radial diffusivity, representing the average of the two smaller eigenvalues. Classically, axial diffusivity and radial diffusivity have been interpreted as the mean magnitude of water diffusion parallel/perpendicular to the axon^26,56,57; however, axons within a given voxel are not perfectly aligned but have orientational dispersion and fibre crossings. Fibre tractography is a 3D reconstruction technique that allows to delineate non-invasively the WM fibre pathways of the brain, and its fibre tracking can be performed using data collected by DTI.^58–59 Tract-specific analysis uses tractography as an anatomical landmark, and values are measured along the tractography, which is drawn by defining seed and target regions of interest (ROIs). Tractography may be used to reconstruct a tract of interest to obtain qualitative anatomical information (visual evidence of disruption of the tract), extract quantitative measures (volumetric and diffusion metrics) or make inferences about connectivity.⁶⁰ Tract-based spatial statistic (TBSS) is a voxelwise statistical analysis developed for DTI data.⁶¹

As DTI allows an early and in vivo evaluation of microstructural changes in WM fibres, it has the rationale to have an important role in the management of iNPH patients. In fact, DTI analysis of cerebral WM in specific anatomical regions seems to be a promising MRI biomarker of iNPH and could also be used in the differential diagnosis from other neurodegenerative disorders, such as AD, Parkinson disease and vascular dementia.²⁶ As iNPH and AD are both associated with cognitive decline and ventriculomegaly, many efforts have been made in order to analyze microstructural changes in WM integrity in elderly patients, and to use them as potential imaging markers for early differentiation between iNPH and AD.

Moreover, DTI has been proposed as a non-invasive predictive test to determine the likelihood of shunt responsiveness. In the literature, the only effective treatment for iNPH is a CSF shunt, usually configured between the lateral ventricle and the abdomen (ventriculoperitoneal (VP) shunt). Between 60% and 80% of patients could improve following shunt surgery.⁶² The CSF Tap Test, also known as the large-volume lumbar puncture, is a predictive test that can easily be performed at most neurological centres.^63–64 The rationale is that if the patient has a significant response to CSF removal, shunt surgery should help. However, it is an invasive procedure and the absence of response to a Tap Test does not exclude shunt responsiveness because it is specific rather than sensitive (range of 50% to 80%).⁶⁵ Thus, there is an urgent need for a better prediction of positive shunt responsiveness, and consequently an increasing interest in studying WM changes in iNPH by using DTI in Shunt Responders (SR) and non-responders.

However, after the literature review performed by Siasios et al. and published in 2016,⁶⁶ there is a substantial lack of structured reviews on the recent and different applications of DTI in iNPH patients. Thus, we performed an extensive literature search and analyzed the most recent and pivotal original articles and systematic reviews that investigated the role of DTI in iNPH, in order to provide a literature review, as well as an up-to-date overview of the application of DTI in this setting.

Materials and methods

We reviewed studies that applied DTI analysis in patients with iNPH. We searched MEDLINE (Medical Literature Analysis and Retrieval System Online; through the PubMed interface), Web of Science and Google, using the following keywords: ‘Diffusion Tensor Imaging’ AND ‘Normal Pressure Hydrocephalus’. A complementary search was made using the first two keywords combinations and ‘tractography’ or ‘FA’.

Studies were selected according to the following inclusion criteria:

• evaluation of patients with iNPH, according to the current clinical and radiological definitions⁶;

• publication from January 2000 until June 30, 2020;

• original articles and reviews;

• collection of MRI data for DTI metrics;

• evaluation of at least one DTI metric (FA, AD, RD or MD);

• post-processing of images performed with ROI or other strategies in order to obtain quantitative values.

The search was limited to articles published in English. Studies that performed tractography but did not report DTI quantitative measurements, as well as case reports, conference abstracts, notes and letters to the editor, were excluded.

All retrieved articles were separately reviewed by two of the authors (I.G. and G.C.). In addition, the reference lists of all the retrieved articles were meticulously reviewed to identify any additional pertinent articles. In the review of the articles identified, particular attention was paid to the study design, its methodological characteristics, number of patients, MRI scanner characteristics (magnet field strength, b-values, number of directions used), the technical characteristics of the FA and DTI protocols used, and the potential clinical significance of the DTI findings.

Results

A total of 59 manuscripts were retrieved, and 16 were excluded based on the title and/or abstract. Out of 16, 3 analyzed DTI in children with obstructive hydrocephalus, 2 applied DTI in hydrocephalus following stroke and intracranial haemorrhage, 3 did not perform DTI examination, 2 performed tractography without quantitative parameters, 1 analyzed communicating hydrocephalus in a rodent model, 4 were letters to the editor and 1 was a case report. One manuscript was also excluded as the full text was written in Polish. Finally, 42 manuscripts were included in the current review. The flow diagram of the literature review process, according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) Guidelines for systematic reviews⁶⁷, is shown in Figure 1. Excluded manuscripts are listed in the supplementary file.

Figure 1.

Flow diagram of the literature review process, according to the PRISMA [67].

These retrieved studies were separated in reviews (n = 4)^66,68–70 and original articles (n = 38)^12,18–54 (Table 1). The following information was retrieved from the manuscripts: number of subjects; methods of analysis; analyzed DTI parameters and cerebral regions, neurological tests for diagnosis and correlation to DTI quantitative data. Moreover, we tried to classify: (1) studies attempting to define DTI characteristics in iNPH patients by comparison with the corresponding characteristics in age-matched healthy controls; (2) studies examining the role of DTI in differentiating iNPH from other neurodegenerative diseases, i.e. AD and Parkinson disease; (3) studies correlating DTI to neurological tests and CSF flowmetry data; and (4) studies evaluating the potential prognostic role of DTI for iNPH patients undergoing CSF drainage (Tap Test or surgical shunt placement).

Table 1.

Synopsis of the study populations, imaging protocols and study parameters in the clinical original studies identified in the present review.

Authors & year	Population	MRI scan	b-values (s/mm2)	Directions	CSFflowmetry	Analysis method	Studied DTI parameters	Analyzed cerebral regions	Neurological tests
Ades-Aron et al. 201818	23NPH/10AlzD/11HCs	3T	0, 1000, 2000	60	No	Kurtosis & Tract-specific analysis	MD, AD, RD, FA, MK, AK, RK	CST	/
Atasoy et al. 202019	37NPH	1.5T	0, 1000	60	Yes	ROI	FA	PVWM	/
Chen et al. 201620	28NPH/17HCs	1.5T	0, 1000	15	No	ROI	FA, AD, MD	Anterior thalamic nuclei and mammillo–thalamo–cingulate projection	RCFT, WAIS-III, WMS-WL, &COWAT
Daouk et al. 201421	12NPH/18AlzD	3T	0, 1000	12	Yes	ROI	FA, ADC, AD, RD	IC	MMSE
Demura et al. 201222	36NPH	1.5T	0, 860	15	No	ROI	FA, ADC	Frontal PVWM, PLIC, corona radiata & centrum semiovale	MMSE, FAB, TUG, iNPHGS
Eleftheriou et al. 202023	13NPH/9HCs	1.5T	0, 1000	15	No	ROI	ADC, FA	CC (genu and splenium), IC, centrum semiovale, frontal WM & thalamus	TUG, MMSE, 10m walk/steps, Romberg, TMT-A
Grazzini et al. 202024	15NPH/15HCs	1.5T	0, 1000	12	Yes	ROI & atlas-based tract–mapping	FA, MD, AD, RD	Genu and splenium of the CC, forceps minor, pre-rolandic motor cortex, post–rolandic sensitive cortex, ATR, IC, cerebral peduncles and optic radiations	MMSE, FAB
Hattingen et al. 201025	11NPH/10HCs	3T	0, 700	12	No	ROI & TBSS	FA, MD	CST & CC	10-m walking test, MMSE, Dem Tect
Hattori et al. 201126	18NPH/11AlzD/11PD/19HCs	1.5T	0, 1000	13	No	Tract-specific analysis	FA, AD	CST	MMSE
Hattori et al. 201227	22NPH/20AlzD/20HCs	1.5T	0, 1000	13	No	Tract-specific analysis	FA	Fornix	MMSE
Hattori et al. 201228	20NPH/20HCs	1.5T	0, 1000	13	No	ROI, Tract-specific analysis, & TBSS	FA, ADC, RD, AD	CC (genu & splenium), CST, IC, PVWM, uncinate fasciculus & cingulum	MMSE
Hong et al. 201029	13NPH/15AlzD/15HCs	1.5T	0, 1000	25	No	ROI	FA, MD	Hippocampus	MMSE, CDR, CDR-SOB
Hořínek et al. 201630	17NPH/14AlzD/17HCs	3T	0, 1000	30	No	TBSS	FA, MD, AD, RD	CC, PVWM, fronto-parietal subcortical WM, corona radiata	TUG, 10-m gait test
Ivkovic et al. 201531	25NPH	3T	0, 1000	33	No	Ball-and-stick model	FA, MD, AD, RD	IC, PVWM	MMSE, 10-m gait test
Jurcoane et al. 201432	26NPH/10HCs	3T	0, 1000	60	No	TBSS	FA, AD, RD, MD, & MTR	CST & SLF	10-m walk/step test, DemTect, MMSE, TMT
Kamiya et al. 201633	29NPH/13HCs	3T	0, 500, 1000, 1500, 2000, 2500	32	No	TBSS	DKI, FA, AD, RD	IC, corona radiata, SLF, inferior fronto-occipital fasciculus, CC, subcortical fronto-parietal WM, posterior cingulum	MMSE, FAB, TMT-A
Kamiya et al. 201734	10NPH/14HCs	3T	0, 500, 1000, 1500, 2000, 2500	32	No	Tract-specific analysis, NODDI & WMTI	FA, MD, AD, RD	CST	/
Kang et al. 201635	54NPH	3T	0, 600	45	No	ROI & atlas-based tract–mapping	FA, MD, AD, RD	CST, ATR, cingulum–hippocampus, IFO, ILF	MMSE, CDR, TMT-A, TUG, 10-m walk test, UPDRS, iNPHGS
Kang et al. 201636	28NPH/28AlzD/20HCs	3T	0, 600	45	No	VOI & TBSS	FA, MD	Anterior corona radiate, CC (genu & splenium), SLF, PTR, external capsule & middle cerebellar peduncle	MMSE, TUG, iNPHGS, CDR, FAB, TMT-A, 10-m walk test, gait test
Kanno et al. 201137	20NPH/20AlzD/20PD	1.5T	0, 1000	13	No	Voxel-based analysis	FA, MD	PVWM, SCWM of the temporal, parietal & occipital lobes (SLF, sagittal stratum) & ALIC	TUG, MMSE, FAB, iNPHGS
Kanno et al. 201738	50NPH/10HCs	1.5T	0, 1000	13	No	TBSS	FA, MD	Subcortical WM	TUG, MMSE, FAB, iNPHGS
Keong et al. 201739	16NPH/9HCs	3T	0, 350, 650, 1000, 1300, 1600	12	No	ROI	FA, MD, AD, RD	Genu and body of the CC, ILF, ATR), the junction of the IFO/UNC & PLIC	MMSE, HVLT-R, COWAT
Kim et al. 201140	16NPH/10AlzD/10PD/10HCs	3T	0, 600	45	No	ROI	FA, MD	PVWM, ALIC, PLIC, CC (genu & SCC), SLF & ILF	MMSE
Koyama et al. 201241	10NPH/10HCs	3T	0, 1000	12	No	ROI	FA	Forceps minor and CST	modified Rankin Scale, TUG, MMSE, FAB
Koyama et al. 201342	24NPH/21HCs	3T	0, 1000	12	No	ROI & TBSS	FA	CC, PLIC & cerebral peduncle	TUG
Lenfeldt et al. 201143	18NPH/10HCs	1.5T	0, 1000	6	No	ROI	FA	CC (genu, splenium, truncus), IC, frontal & lat PVWM, centrum semiovale	MMSE, TMT-A, Gait video recording scale, gait velocity
Lock et al. 201944	12 complexNPH/16 classicNPH/5HCs	3T	0, 1000	20	No	Tract-specific analysis	FA, MD, AD, RD	DTI profiles	MMSE, 10-m walking test, Tinetti gait examination
Marumoto et al. 201245	10NPH/10HC/18PD	3T	0, 1000	12	No	ROI	FA	ATR, minor & major forceps, SLF & CST	TUG, MMSE, FAB
Nakanishi et al. 201346	11NPH/6HCs	3T	0, 500, 1000, 1500, 2000, 2500	32	No	Tract-specific analysis	FA, RD, AD	CST	/
Nicot et al. 201447	10NPH (6iNPH & 4secondary NPH)	3T	0, 1000	12	No	ROI conventional & delineation method	FA, ADC, RD, AD	IC & body of CC	/
Radovnický et al. 201612	27NPH (15DESH)/24HCs	1.5T	0, 1000	20	No	ROI	FA	ALIC, PLIC & CC	10-m walk test
Reiss-Zimmermann et al. 201448	15NPH/14HCs	3T	0, 1000	20	No	ROI & TBSS	FA, MD, AD, RD	CST, SCWM, PVWM, PLIC, midbrain, CC (genu, body, & splenium) & caudate nucleus head	MMSE, modified Rankin Scale
Saito et al. 202049	12NPH/8HCs	3T	–	–	No	Tract-specific analyses & TBSS	FA, MD, AD, RD	ATR	MMSE, FAB, TMT-A
Scheel et al. 201250	13NPH(7iNPH)/13HCs	1.5T	0, 1000	25	No	ROI & TBSS	FA, MD, AD, RD	CST (pre-central gyrus, PVWM, PLIC, pons, midbrain), caudate heads & CC	/
Tan et al. 201851	21NPH/21HCs	3T	0, 800	32	No	ROI & TBSS	FA, MD, AD, RD	CC& IC	HDI, HOQ
Tsai et al. 201852	22NPH/16HCs	1.5T	0, 1000	15	Yes	ROI	FA, MD, AD, RD	Thalamic nuclei & CST	Clinical gait scale
Yokota et al. 201953	24NPH/12HCs	3T	0, 1200 – 0, 1000	12, 64, 128	No	Atlas-based ALPS index & ROI	FA, MD, AD, RD	PVWM (projection and association fibres)	/
Younes et al. 201954	9NPH/13AlzD/20HCs	3T	0, 1000	48-64	No	ROI & Tract-specific analysis	FA, MD, AD, RD	STR, CST & dentatorubrothalamic tract	MOCA

AlzD: Alzheimer disease; ALPS: analysis along the perivascular space; AD: axial diffusivity; AK: axial kurtosis; ALIC: anterior limb of IC; ATR: anterior thalamic radiation; CC: corpus callosum; CDR: Clinical Dementia Rating Scale; CDR-SOB: Clinical Dementia Rating Scale Sum of Boxes; COWAT: Rey controlled oral word association test; CSF: cerebrospinal fluid; DESH: disproportionately enlarged subarachnoid space hydrocephalus; FAB: frontal assessment battery; HCs: healthy controls; HDI: Headache Disability Inventory; HOQ: Hydrocephalus Outcome Questionnaire; HVLT-R: Hopkins Verbal Learning Task-Revised; IC: internal capsule; IFO/UNC: inferior fronto-occipital/uncinate fasciculi; ILF: inferior longitudinal fasciculus; iNPHGS: iNPH Grading Scale; MD: mean diffusivity; MK: mean kurtosis; MMSE: mini mental score examination; MOCA: Montreal Cognitive Assessment; NODDI: neurite orientation dispersion and density imaging; PD: Parkinson disease; PLIC: posterior limb of IC; PVWM: periventricular WM; RCFR: Rey complex figure test; RD: radial diffusivity; RK: radial kurtosis; ROI: region of interest; SCWM: subcortical white matter; SLF: superior longitudinal fasciculus; STR: superior thalamic radiation; TMT-A: Trail Making A test; TUG: Timed Up and Go test; UPDRS: Unified Parkinson Disease Rating Scale; WAIS-III: Wechsler Adult Intelligence Scale III; WMS-WL: Wechsler Memory Scale III; WMTI: white matter tract integrity.

All the studies were performed after 2010, with a constant trend (search results by year are shown in Figure 2). To note, from 2010 to 2014 the research topic focused on iNPH diagnosis and differential from other neurodegenerative diseases, while after 2014 our focus was shifted to an increasing interest in the potential prognostic role of DTI in iNPH, in order to predict patients who could benefit from CSF shunting.

Figure 2.

Search results by year. The graph shows that all the studies were published from 2010 to 2020 with a constant trend.

A total of 771 subjects were included in the selected studies. All of the studies reported at least one DTI quantitative parameter: FA was measured in 100% of the studies. MRI scans were performed on 1.5T (n = 15, 39.5%) or 3T (n = 23, 60.5%) scanners; the number of directions during diffusion acquisitions ranged from 12 to 128, and the number of b-values from 2 to 6. Thirty-two studies (84.2%) used two b-values (specifically, 68.4% used b-values 0 and 1000 s/mm²). Detailed scan parameters for each study are reported in Table 1.

The majority of studies measured DTI metrics according to ROI-based methods (n = 25, 65.8%), alone or associated with tract-specific analysis and/or whole-brain processing in TBSS. The most commonly studied regions were the corpus callosum (CC), internal capsule (IC), thalamic radiation, cortico-spinal tract (CST), subcortical and periventricular fronto-parietal WM, fornix and cerebral peduncles.

Despite great heterogeneity in methods of collecting and analyzing the data, as reported in Table 1, the majority of studies reported statistically significant differences between DTI parameters in iNPH patients and healthy controls, as well as between iNPH and Alzheimer’s disease or Parkinson disease patients.

Some studies reported clinical evaluation and psychometric tests, and correlated them to DTI values.

Many measures were used as assessments of cognitive function (Mini Mental State Examination (MMSE),^{21–29,31,32,35–37,43–45,48,49} trail making test A (TMT-A),^{23,32,33,35,36,43,49} DemTect test,²⁵,³² Montreal Cognitive Assessment (MOCA)⁵⁴; learning and memory (Hopkins Verbal Learning Task-Revised (HVLT-R)³⁹); verbal fluency and executive function (COWAT),²⁰,³⁹ Wechsler Memory Scale III (WMS-WL),²⁰ Rey complex figure test (RCFT),²⁰ according to Deckersbach et al.⁷¹); conceptualization, mental flexibility, motor programming, sensitivity to interference, inhibitory control and environmental autonomy due to frontal lobe dysfunction (frontal assessment battery (FAB)^{22,24,33,36,37,41,45,49}); grade of dementia (Clinical Dementia Rating Scale (CDR),^29,35,36, Clinical Dementia Rating Scale Sum of Boxes (CDR-SOB)²⁹); motor impairment and gait (gait velocity,⁴³ Timed Up and Go (TUG) test,^{22,23,30,35–38,41,42,45} 10-m walk steps¹²,²³,²⁵,^30–32,³⁶,⁴⁴ according to Tsakanikas et al.,⁷² Romberg test,²³ Tinetti gait examination⁴⁴); degree of overall disability (modified Rankin score⁴¹,⁴⁸). Tan et al.⁵¹ included quantitative headache scores: Headache Disability Inventory (HDI) and Hydrocephalus Outcome Questionnaire (HOQ). In particular, the more frequently used psychometric tests were MMSE (n = 24), TUG (n = 10) and FAB (n = 8).

Sixteen studies^{12,22,23,31,32,34,35,38–40,43,47–51} analyzed DTI indices before and after a CSF drainage (n = 6)^{22,31,32,35,43,48} or a shunt surgery (n = 10),¹²,²³,³⁴,^38–40,⁴⁷,^49–51 and correlated them with clinical data.

Discussion

Technical considerations

A wide variety of DTI protocols and strategies for examining various areas of the brain in iNPH patients have been described in the literature. Thus, as results are dependent on the adoption of good practices regarding acquisition parameters, pre- and post-processing, it is crucial to improve repeatability and reproducibility when applying DTI analysis to iNPH patients.

First of all, significant inter-scanner differences can be found in DTI measures. Types of inter-scanner differences include variability in receiver coils, reconstruction algorithms, magnetic fields and acquisition parameters.⁷³ In particular, the b-values and the number of diffusion encoding directions are of fundamental importance, because DTI model results in a b-value-dependent parameterization of the diffusion profile.⁷⁴ A minimum of six non-collinear diffusion encoding directions are required to measure the full diffusion tensor in the brain imaging.⁷⁵

The types of analyses more frequently applied in DTI studies in iNPH included manual ROI placement, tract-specific analysis, voxel-based analysis and TBSS. All of these analysis techniques have their pros and cons, and present specific challenges. Although the Intraclass Correlation Coefficient (ICC) for intrarater and interrater reliability of DTI measurement by ROIs has been reported >0.80, constituting excellent agreement,⁴⁰ the placement of individual ROIs in ROI-Based Analysis is known to be operator-depending biased, time-consuming, limited in feasibility and generalizability. Furthermore, the low resolution of DTI images can hinder delineation of the ROI. Registration of the DTI dataset to anatomical T1/T2 images can improve spatial resolution and facilitate ROI drawing. However, misregistration/misalignment can occur. Furthermore, the best choice for ROI placement remains an open question. While some authors identified the CST as the main region to be analyzed,¹⁸,²⁶,³⁴,⁴⁶ other studies demonstrated more significant results in the forceps minor, IC and frontal areas, with correlation to clinical indexes.²⁴,⁴¹,⁴⁵ Just as the ROI approach, tract-specific analysis is a simple method that can be used directly to derive values for structures that are anatomically the same among individuals. However, it has limitations because it is sensitive only to changes in those few parts of the brain that they can accurately measure. When applying TBSS in iNPH patients, particular attention should be paid to the crossing fibres. In fact, the interpretation of DTI data is further complicated in regions of crossing WM tracts.⁷⁶ This is unfortunately unavoidable as many areas of the brain have considerable fibre crossing (e.g. CST, the centrum semiovale, uncinate fasciculi and transpontine fibres). Consequently, changes in the angle and relative volume fractions of crossing WM fibre groups within a voxel can result in significant anisotropy changes without any WM abnormalities. Hattori et al.²⁸ reported some misregistered areas around the thalamus or fornix in the lateral ventricle or in the posterior body of the lateral ventricle. However, the results of the TBSS analysis have been considered consistent with those of the ROI analysis and the tract-specific analysis of the CST. The cost function masking and enantiomorphic normalization can be used as alternatives to overcome lesion deformations.⁷⁷ The variability in FA can be reduced considerably by focusing on WM tracts in specific anatomical regions, particularly those with fewer WM crossings, e.g., the CC. Consequently, in the absence of other information, FA is a highly sensitive but nonspecific biomarker of neuropathology. This imposes challenges on the interpretation of DTI measurements for diagnostic and therapeutic applications.

Beyond DTI-based tractography, other advanced techniques and models were reported in the literature. In particular, biophysical models aiming at specific insights into white matter microstructure, such as neurite orientation dispersion and density imaging (NODDI) and WM tract integrity (WMTI), have also been applied to iNPH, to disentangle the effects of stretching and axon density.³⁴ NODDI is a diffusion MR technique for estimating the microstructural complexity of dendrites and axons in vivo on clinical MRI scanners. It may provide neurite density and orientation dispersion estimates, thereby disentangling two key contributing factors to FA and enabling the analysis of each factor individually.⁷⁸ The WMTI model relates diffusional kurtosis imaging metrics to features of WM microstructure, and it is regarded as a good approximation for fibre dispersion of up to 30° in clinically relevant scans.⁷⁹ Diffusional kurtosis imaging (DKI) is considered an extension of DTI and allows the diffusional kurtosis to be estimated with clinical scanners using standard diffusion-weighted pulse sequences.⁸⁰ In our review, three studies applied kurtosis to iNPH patients.¹⁸,³³,⁴⁶

Diagnosis of iNPH: patients versus healthy controls

While the use of DTI in iNPH is relatively new, evidence from the currently available studies suggests that this technique would be a useful and sensitive tool in detecting the WM microstructural disease in iNPH. The retrieved studies analyzed different cerebral regions for ROIs’ positioning and extraction of main DTI parameters. In particular, regions are mostly chosen based on previous neuropathological studies on iNPH patients, which demonstrated degeneration, axonal loss and chronic ischaemic changes in these areas. In fact, functional and neuropathological studies of chronic hydrocephalus have suggested the presence of WM changes in multiple sites, presumably resulting from mechanical pressure due to ventricular enlargement and metabolic changes.^81–83 Hence, the interest in DTI has increased in the last few years, in order to detect and quantify these WM alterations in patients with iNPH.

Although the use of different technical protocols (scanner strength, b-values and number of directions, as reported in Table 1), most of the studies demonstrated that iNPH patients exhibit significantly lower FA values and higher MD and AD values in the frontal area and in the forceps minor in comparison to the healthy control subjects,²⁴,³²,⁴¹,⁴⁵ as well as in the genu and splenium of the CC.²⁴,²⁵,²⁸,³³,³⁷,⁴² The much lower FA values in the forceps minor and in the CC were associated with more severe clinical symptoms such as gait disturbance.^41–42

The pathogenesis of iNPH may underlie the decreased FA values in these regions. First, the mechanical pressure resulting from ventricular enlargement stretches and compresses the periventricular WM and the CC, causing axonal loss.^81–83 Second, there may be interstitial oedema due to CSF suffusion. On the contrary, RD in the periventricular WM has been reported to be greater in patients with iNPH than in the controls, supporting the presence of free water and/or interstitial oedema in the WM of patients with iNPH.²⁸ Eleftheriou et al. postulated an ongoing pathological process also in the thalamus and in the frontal WM supporting the hypothesis of a shunt-reversible thalamocortical circuit dysfunction in iNPH.²³

However, some of the identified studies reported discrepant results concerning FA metrics in CST of iNPH patients. Some authors reported higher CST FA values in the iNPH patients compared with the healthy controls, assessing that DTI parameters of the CST correlated with the severity of gait disturbances.^{25,26,28,46,53} Thus, Hattori et al. assessed that the CST FA values had a sensitivity of 94% and specificity of 80% at a cutoff value of 0.59 in discriminating patients with iNPH from other subjects, suggesting that patients with iNPH have altered microstructures in the CST, and that quantitative CST evaluation by using DTI may be useful for differentiating patients with iNPH from healthy subjects.²⁶ On the contrary, Koyama et al. and Marumoto et al. found a significantly decreased FA in the same areas.⁴¹,⁴⁵ Higher FA values in iNPH than in controls were found also in the posterior limb of the IC.⁴⁰ Thus, CST seems to be a more challenging region to investigate, also with TBSS analysis, as fibre crossing can cause contradictory results.

Anterior thalamic nuclei and radiation are controversial sites too. Chen et al. revealed FA elevations within the anterior thalamic nuclei in early iNPH patients compared with controls,²⁰ while Saito et al. reported lower FA.⁴⁹

A few studies evaluated the microstructural integrity of the hippocampus and the fornix in iNPH patients by using ROI and tract-specific analysis.²⁷,²⁹ FA values of the fornix were reported to be significantly lower in patients with iNPH than in healthy controls.²⁷ This finding supports the hypothesis that in patients with iNPH the fornix may be damaged.

Differential diagnosis between iNPH and other dementias

According to a recent review, DTI analysis of the CC, IC, hippocampus and fornix, combined with measurement of Evans index and volumetric brain analysis, is a promising MRI biomarker of iNPH and could be used in the differential diagnosis from AD and other dementias.⁶⁸

About the ventricular size, Daouk et al. assessed that both AD and iNPH involve CSF disorders and showed that iNPH and AD patients have different DTI findings in the IC²¹; they found a correlation between FA in the IC and ventricular volume only in the iNPH population. On the contrary, Hořínek et al. observed that the ventricular volumes were correlated with DTI parameters in the area next to the ventricles in both AD and iNPH patients.³⁰ Combining increased MD of the superior thalamic radiation with ventricular volume, Younes et al. found a separation of iNPH from the AD subjects that both present with ventriculomegaly. Additionally, ventricular to sulcal CSF ratio was greater in the iNPH patients compared to AD and healthy controls.⁵⁴

Kang et al. showed lower FA and higher MD in the anterior corona radiate, CC, superior longitudinal fasciculus, thalamic radiation, external capsule and middle cerebellar peduncle in iNPH patients in comparison with AD.³⁶ Ades-Aron et al. found that AD and axial kurtosis differ between iNPH and AD in an area close to the superior IC and corona radiate, reporting greater AD and lower axial kurtosis in the iNPH group in comparison with AD with a p < 0.05.¹⁸

Hattori et al. focused on the CST and suggested that the CST FA values are useful for differentiating patients with iNPH from those with AD or Parkinson disease.²⁶

Hong et al. measured FA and MD values in the hippocampal head, body and tail on both sides and reported that FA values were the lowest in AD patients, iNPH patients and the healthy controls in this order. MD values were the highest in the same order. Hippocampal volume was not different between patients with iNPH and AD, so they concluded that microstructural alterations of the hippocampus are more sensitive than the volumetric changes in discriminating AD from iNPH.²⁹

Hattori et al. showed that FA values, volume and mean cross-sectional area of the fornix were significantly lower and the length of the fornix was significantly higher in iNPH patients than in those with AD.²⁷ It could be due to elevation and stretching of the corpus callosum in the anteroposterior direction by the mechanical pressure from the enlarged lateral ventricles in iNPH, thereby determining involvement and deformation of the fornix itself. FA values have been reported to be lower in patients with AD than in healthy controls at the level of the fornix²⁹,^84–87: as in AD, the hippocampus is the primary site of damage, and the fornix can undergo secondary degeneration and microstructural changes due to hippocampal atrophy.⁸⁴

In addition to dementia and ventriculomegaly, another key feature of iNPH is gait. Gait abnormalities in the elderly, characterized by short steps and frozen gait, can be caused by several diseases, including mainly iNPH and PD.

Hattori et al. demonstrated that alterations in CST microstructure can discriminate iNPH from AD and Parkinson disease with a sensitivity of 94% and specificity of 80%.²⁶ Kanno et al. reported that the MD values were significantly higher in the periventricular WM of iNPH patients (including IC, CC, corona radiata and subcortical orbitofrontal WM) than in patients with Parkinson disease. In contrast, the MD values of the subcortical WM of the left superior frontal gyrus were significantly lower in iNPH patients than in the Parkinson disease group.³⁷ Marumoto et al. found that iNPH patients had significantly lower FA in particular for anterior thalamic radiation and forceps minor as compared to the Parkinson disease group, and that the gait capability, evaluated with TUG time, correlated with anterior thalamic radiation FA in both groups.⁴⁵

DTI parameters and neurological tests

In literature, the more commonly used psychometric tests for iNPH patients’ evaluation were MMSE, FAB and TUG. MMSE is a simplified scored form of the cognitive mental status examination.⁸⁸ The FAB is reported to be sensitive to frontal lobe dysfunction, and it is routinely used as it has been reported that executive dysfunction is a characteristic feature of the cognitive impairment in patients with iNPH.^38,89–91 The TUG test is a reliable and valid test for quantifying functional mobility that may also be useful in following clinical change over time. The patient is observed and timed while he rises from an armchair, walks 3 m, turns, walks back and sits down again. The test is quick, requires no special equipment or training, and is easily included as part of the routine medical examination.⁹²

These tests are useful not only for clinical patients’ selection, but they also demonstrated a correlation to DTI findings in many studies. Marumoto et al. reported that in the iNPH group, the TUG time correlated negatively with FA for anterior thalamic radiation (p < 0.05) and positively with FA for forceps minor (p < 0.05), while neither TUG number of steps nor MMSE had a significant correlation with FA.⁴⁵ Kamiya et al. reported that FA showed positive correlations with MMSE, FAB and TMT-A in the superior longitudinal fasciculus, fronto-occipital fasciculus, inferior longitudinal fasciculus, corona radiata (anterior, superior and posterior), the IC, the CC and the cingulum; AD and RD analyses showed no significant correlations with MMSE and FAB, while negative correlation with TMT-A was observed in the limited portion of the frontal deep WM.³³ In our previous study, we did not find significant correlations between DTI parameters and MMSE, while we observed a significant positive correlation between forceps minor RD, AD, FA and FAB in the iNPH group (p = 0.01).²⁴ Similarly, Kanno et al. showed that the total scores on the FAB were correlated with the FA in the frontal and parietal subcortical WM, and clinical indexes of gait disturbance were correlated with the FA in the anterior limb of the left IC and under the left supplementary motor area.³⁷ In addition, Saito et al. reported that the changes of FA and AD in anterior thalamic radiation were positively correlated with the improvement of FAB scores.⁴⁹ Koyama et al. previously described that lower FA values in the CC tended to be associated with more severe clinical symptoms for both cognitive impairment (p = 0.057) and urinary incontinence (p = 0.052).⁴² Moreover, the total scores of the MMSE and the FAB, and the TUG in the SR groups have been reported to be improved after shunt placement compared with those at baseline.³⁸ Thus, regions with low FA seem to be related to motor and cognitive dysfunction in iNPH, and the FAB appears to be the best neurological test to represent WM microstructural changes in iNPH, and to achieve a prognostic index in patients who may benefit from shunt surgery.²⁴

DTI and CSF flowmetry parameters

MR imaging of the brain for possible iNPH usually includes phase contrast (PC)-MRI CSF flow study, as cine phase contrast sequence can demonstrate CSF pulsatile flow throughout the cardiac cycle.⁹³ However, a very few studies analyzed the correlation between DTI parameters and CSF flowmetry, such as aqueductal CSF flow peak velocity and stroke volume. In particular, in our previous study we did not find significant correlations between DTI parameters and CSF flow peak velocity and stroke volume in all our investigated cerebral regions, while FA, MD, RD and AD were correlated to the FAB.²⁴ Atasoy et al. reported that patients with iNPH had increased CSF stroke volume in comparison to the control group, and they found increasing CSF stroke volume values with increasing severity of clinical iNPH symptoms; however, they did not demonstrate correlation between DTI and CSF flowmetry parameters.¹⁹ In Daouk et al., FA was correlated with ventricular volume in the iNPH population, but not with the aqueductal CSF stroke volume.²¹ These preliminary data seem to suggest that DTI could be more sensitive than CSF flowmetry in identifying clinical subgroups of iNPH patients, or at least it may provide complementary information to flowmetry. Further studies are required to better analyze these correlations, and to define whether DTI could add information to CSF flowmetry in iNPH workup.

DTI parameters after CSF drainage and shunting surgery

It is known that symptoms of iNPH may be reversible with a CSF shunt; consequently, predictive tests to determine the likelihood of shunt responsiveness are recommended. However, the relative importance of individual DTI measures with respect to the interpretation of responsiveness to CSF drainage currently remains a matter for debate. In the last 5 years, many studies analyzed pre- and post-shunt DTI findings in iNPH patients. However, in the literature, as for diagnosis of iNPH, we can find contradictory results and there is no consensus according to the best regions to investigate.

Reiss-Zimmermann et al., Scheel et al. and Eleftheriou et al. reported that, following shunt surgery, all DTI parameters showed a trend towards normalization in iNPH patients, yet differences to healthy control subjects remained.²³,⁴⁸,⁵⁰ According to Kim et al., iNPH SRs could be distinguished from other conditions on the basis of higher FA in the posterior limb of the IC.⁴⁰ On the other hand, Jurcoane et al. affirmed that the specificity to discriminate SRs from SNRs was low for all DTI values alone (max. 69% sensitivity for FA), although a moderate sensitivity.³² Similarly, Lenfeldt et al. assessed that FA values could not separate SRs from SNRs.⁴³

With a didactic purpose, we tried to summarize the different findings of the retrieved studies comparing pre- and post-shunt DTI data in iNPH patients: thus, we divided the results focusing on FA, MD and AD changes after surgery, and on the analyzed cerebral regions.

FA changes after surgery

In many of the studies, FA in the CST tended to decrease after shunt surgery.³²,³⁴,⁴⁰,⁵⁰ Jurcoane et al. also described higher FA in the CST of drainage responders compared to controls.³²

Demura et al. determined FA in several WM regions in 36 iNPH patients before and 24 hours after a CSF Tap Test. A significant increase in FA after the CSF Tap Test in the body of the CC was observed in both responders and non-responders (p < 0.05).²²

Recently, Eleftheriou et al. found a decrease in FA in the CC and IC and an increase in the frontal WM after surgery. Specifically, they reported a significant FA decrease in the splenium of the CC (p = 0.02) and in the IC (p = 0.02), as well as a negative correlation between FA changes versus gait results in the frontal WM (r = −0.7, p = 0.008).²³

Kanno et al. observed that, compared with the healthy controls group, FA in the CC and in the subcortical WM of the convexity and the occipital cortex was lower in SR at baseline; the post-operative FA was decreased in the corona radiata in the SR group.³⁸ Compared with the pre-operative images, the post-operative FA was only decreased in the corona radiata and only in the SR group. There were no significant regions in which DTI indices were altered after shunt placement in the SNR group.³⁸

Lenfeldt et al. reported that FA did not normalize after CSF drainage in any of their analyzed cerebral regions.⁴³ Nicot et al. found a significant decrease in FA in IC and CC after surgery (p < 0.05)⁴⁷; Reiss-Zimmermann reported a decrease in FA in the splenium of CC, too.⁴⁸ Saito et al. assessed that, when comparing pre-and post-operative iNPH patients, significantly decreased FA values were found in the periventricular WM including the posterior limb of the IC, external capsule, CC and corona radiata.⁴⁹

Kang et al. described that CSF Tap Test non-responders, when compared to responders, exhibited lower FA in the left anterior thalamic radiation, left cingulum–hippocampus and left inferior fronto-occipital fasciculus.³⁵ Thus, FA values in iNPH in critical areas may reflect the severity of WM involvement. In general, it has been known that not all clinical symptoms of iNPH were reversed by the shunt operation. This suggests that the cause of iNPH included reversible (e.g. compressed WM according to ‘predominant stretch/compression’ theory) and irreversible (e.g. destruction of WM) mechanisms.⁴⁵ Therefore, in case of severe axonal loss, CSF drainage should not improve the symptoms. Kamiya et al. reported that the pathologically high orientational coherence within the CST in iNPH tends to normalize after shunt surgery,³⁴ which can be interpreted as recovery from axon stretching, while the axon density remained unchanged after the surgery, as index of chronic and irreversible neuronal damage. These data suggest that DTI can distinguish between reversible and irreversible changes in iNPH, bringing us closer to a quantitative marker that can predict treatment outcome.

MD changes after surgery

Ivkovic et al. described greater changes in MD in external lumbar drainage responders (8.2% ± 3.1%) compared to non-responders (2.1% ± 3%). They also noticed a slight increase in MD within the IC of the responders.³¹ Kang et al. revealed that CSF Tap Test non-responders, when compared to responders, exhibited higher MD in the left cingulum–hippocampus and left inferior longitudinal fasciculus.³⁵ Kanno et al. discovered that, in comparison with the healthy controls group, MD in the periventricular and peri-Sylvian WM was higher in the SR group.³⁸ Reiss-Zimmermann reported an increase of MD in the CC after the CSF Tap Test.⁴⁸

AD changes after surgery

Keong et al. demonstrated a significant change in AD in the posterior limb of the IC at 2 weeks following surgery.³⁹ Jurcoane et al. found that for AD upon CSF drainage, a decrease of >0.7% discriminated drainage responders from drainage non-responders with 82% sensitivity, and a decrease of >1% predicted overall improvement after shunting with 87.5% sensitivity and 75% specificity.³²

Limitations

Our review has some limitations. First, for the purpose of the review, we excluded studies not reporting DTI metrics, such as myelin quantification, apparent diffusion coefficients, WM volume or qualitative tractography-based information. We also excluded studies not based on the tensor. All of them may convey complementary, critical information about the underlying WM alterations in iNPH patients. Second, the choice of keywords may have led to non-inclusion of studies that addressed the aims of this review.

Conclusions

In the last years, DTI has increased its application in the brain, especially in the field of WM analysis. Evidence from currently available studies suggests that DTI may play a role in the diagnosis and differentiation of iNPH from other disorders with similar clinical and imaging features. Moreover, DTI could contribute to the selection of iNPH patients for surgery and predict outcome post CSF shunt. In fact, previous DTI studies found interesting results before and after the shunt surgery, but we believe that the clinicians and the radiologists should focus more on cerebral regions such as IC, frontal WM, forceps minor and thalamus, and correlate the DTI results with the clinical data, to evaluate if there could be a prognostic ROI placement before shunt surgery, in order to identify responders, and of course to differentiate iNPH from other diseases. However, the number of published studies on DTI for iNPH is still small, and it is noteworthy that all of them had relatively low numbers of participants. Moreover, the papers used different technical protocols and analyzed various anatomical areas, with currently no consensus about the better cerebral regions to explore, and arrived at sometimes contrasting results. Large-scale, multicenter studies and the adoption of widely accepted imaging protocols are required to increase the reproducibility and diagnostic accuracy of DTI measurements. At the same time, there is the need to define the normal range of DTI parameters and to standardize them for different age groups. Despite these limitations, in light of these current findings, we believe that DTI offers a significant non-invasive diagnostic tool for iNPH patients, as it also allows a good correlation to clinical status, probably better than CSF flowmetry. Thus, it could be useful to include it in the routine MR protocol in iNPH patients.