Abstract
INTRODUCTION
Fifty years after the initial description of idiopathic normal pressure hydrocephalus (iNPH) by Hakim and Adams [1], the differentiation of iNPH from similar neurological conditions (i.e., iNPH mimics) represents a challenge for clinicians since the triad of gait disturbance, urinary incontinence, and cognitive impairment is unspecific, and the subcortical atrophy commonly misinterpreted as ventriculomegaly [2]. If the comparison of quantitative motor [3] and non-motor parameters [4], including, respectively, dual-tasking and neuropsychological tests before and after cerebrospinal fluid (CSF) tapping has been established to identify iNPH from iNPH mimics, these approaches are time-consuming, expensive, and required invasive procedures, such as CSF tapping. Since parkinsonism is frequently noticed in patients with iNPH as well as in iNPH mimics [5], we wanted to assess its prevalence and its discriminating value in a clinical sample of patients addressed for suspected iNPH, and therefore including both populations (iNPH and iNPH mimics).
To address these knowledge gaps, we conducted a retrospective study in patients with a suspicion of iNPH (including iNPH and its mimics) to compare the prevalence of parkinsonism and the clinical symptoms of the triad (i.e., gait disturbance, urinary incontinence, and cognitive impairment) between iNPH and iNPH mimics and their respective discriminating values, while taking into account the influence of white matter changes (as the contribution of white matter changes to parkinsonism is matter of debate [6, 7]). Since it has been demonstrated that the presence of parkinsonism is frequent in iNPH mimics [5] and that symptoms of the triad are similar between iNPH and its mimics [2], we hypothesized that parkinsonism would represent an appropriate clinical feature to identify iNPH from iNPH mimics with a greater prevalence in iNPH mimics. Establishing new clinical marker to better distinguish iNPH from its mimics has direct clinical implication by improving the selection of older candidates for invasive shunt surgery.
MATERIALS AND METHODS
Participants
A total of 189 patients were referred to the Department of Neurology of the Geneva University Hospitals between January 2007 and September 2015 for a suspicion of iNPH based on gait and/or cognitive impairments with ventricular enlargement on brain imaging (CT or MRI) defined by an Evans ratio >0.30 without macroscopic obstruction of the CSF flow, as previously detailed [3, 8] (See Figure 1). A neurological assessment including a comprehensive neurological examination with a detailed gait evaluation and a neuropsychological assessment were performed. Inclusion criteria for this study were patients with a suspicion of iNPH with (i) a comprehensive neurological examination and (ii) a specific assessment of parkinsonism. Exclusion criteria were presence of an acute medical illness in the past three months and a diagnosis of secondary NPH. Forty-eight patients were excluded: thirty-six for an acute medical condition and twelve for a secondary NPH. Following exclusion, 141 patients were included in this study (75.7±7.1 years; 31.2% women): seventy-nine (70 with possible and 9 with probable iNPH) fulfilled the iNPH consensus guideline criteria [9]; the remaining sixty-two patients were classified as iNPH mimics. Thirty-eight iNPH patients (48.1%) accepted the ventriculoperitoneal shunt procedure (at the time of the analysis); and 84.2% of them reported a good clinical outcome. The iNPH mimics included patients with mixed dementia (13), vascular dementia (10), Parkinson’s disease (6), primary progressive freezing of gait (5), supranuclear palsy (5), drug induced (5), Alzheimer’s disease (4), alcohol-related dementia (4), dementia with Lewy bodies (3), frontotemporal dementia (3), depression (3), and multisystem atrophy (1). A diagnosis (iNPH versus iNPH mimics) was assigned before shunt placement and after reviewing all available clinical and neuropsychological data, as well as brain imaging (either CT or MRI) and blood/CSF laboratory results at consensus case conferences involving behavioral neurologists and neuropsychologists blinded for the presence of parkinsonism; the subscores at iNPH grading scale [10]; and the burden of white matter disease measured with the age-related white matter changes scale (ARWMC) [11]. Parkinsonism was defined by the same board-certified neurologist based on the United Kingdom Parkinson’s disease society brain bank clinical diagnostic criteria (step one: presence of bradykinesia and at least one of the following sign: muscular rigidity, rest tremor, or postural instability) [12]. The presence or absence of rigidity, rest tremor, and postural instability associated with bradykinesia was indicated for each patient with parkinsonism. We used the validated iNPH grading scale [10] for describing gait disorders, cognitive impairment and urinary incontinence. This study protocol was approved by the ethical committee of Geneva University Hospitals.
White matter lesions
White matter lesions burden was assessed by a valid scale with moderate to good interrater reliability, the ARWMC [11] on every neuroimaging scanner at the time of the clinical assessment (23 CT and 116 MRI). Total score (range: 0–30) and subscores (range: 0–6) were computed on the five regions combining the left and right hemispheres: frontal, temporal, parieto-occipital, basal ganglia and infratentorial. All brain scans were rated by the same neurologist (GA) blinded for diagnosis.
Covariates
Comorbidities were documented by the Global health status score (GHS; range 0–10), based on the presence of diabetes, chronic heart failure, arthritis, hypertension, depression, stroke, Parkinson’s disease, chronic obstructive pulmonary disease, angina, and myocardial infarction [13]. A vascular risk factor score (range 0–5) was computed on the presence of diabetes, hypertension, hypercholesterolemia, body mass index >30 or smoking; a cardiovascular risk factor score (range 0–4) on the presence of myocardial infarction, angina, arrhythmia or chronic heart failure; a cerebrovascular score (range 0–2) on the presence of stroke or transient ischemic attacks [14].
Statistics
Descriptive statistics were calculated and we compared iNPH and mimics based on two-sample t-test or Fischer exact test as appropriate. Univariable and multivariable logistic regression models were used to compute odds ratio with 95% confidence intervals to assess the association between diagnostic status (iNPH versus mimics; dependent variable); and parkinsonism, or iNPH grading scale subscores (for gait, cognitive, and urinary disturbances) (independent variables). Age, gender, GHS (comorbidities) and the ARWMC total score were used as covariates in the multivariable models. In an additional analysis, the ARWMC subregion’s scores were also used as covariates. All analyses were conducted using SPSS version 22 (SPSS Inc., Chicago, IL, USA).
RESULTS
Characteristics of the patients with iNPH and mimics are provided in Table 1. The prevalence of parkinsonism is almost double in mimics than in patients with iNPH (40.3% versus 20.3%; p-value: 0.015), whereas the symptoms of the triad and the repartition of parkinsonian signs are similar between both groups: for the cognitive impairment, the mean iNPH grading scale for the combined groups was 2.17±0.67 (a score of 2 refers to existence of amnesia or inattention but no disorientation in time and place); for gait disorders, the mean iNPH grading scale was 2.03±0.45 (a score of 2 refers to unstable but independent gait); for urinary disturbance, the mean iNPH grading scale was 1.16±1.07 (a score of 1 refers to pollakiuria or urinary urgency). The association of rigidity and postural instability with bradykinesia was the most frequent parkinsonian signs in patients with parkinsonism in both groups. In patients with parkinsonism, rigidity was the most prevalent associated parkinsonian sign in patients with iNPH mimics (80%), whereas it was postural instability in iNPH (81%). Individual associated parkinsonian signs were similar between iNPH and iNPH mimics in patients with parkinsonism. The global and the regional white matter changes were also similar between both groups.
The presence of parkinsonism was associated with the diagnosis of mimics (adjusted odds ratio: 2.28; 95% CI: 1.06–4.93) even after adjustment for global white matter lesions, whereas gait, cognitive and urinary symptoms were not associated with diagnostic group (Table 2). Substituting total ARWMC score by regional ARWMC scores in any regions (and particularly in the basal ganglia) does not modify the adjusted association between parkinsonism and iNPH mimics. In order to take into account the neuroimaging modality (CT versus MRI) for the ARWMC rating, we found similar results in the subgroup of the 116 patients with MRI.
DISCUSSION
Our study reported that the symptoms of the triad of iNPH are similar between both groups but parkinsonism is more frequent in mimics than in iNPH patients. This association between parkinsonism and iNPH mimics was not influenced by white matter changes.
The prevalence of parkinsonism ranges from 2.4 to 9.6% in community-dwelling older adults, leading by Parkinson’s disease, drugs induced parkinsonism or vascular dementia [15]— diagnoses also frequently observed in the group of iNPH mimics. Only a low prevalence of patients with iNPH (7.2%) contributes to parkinsonism in older adults [15], although the prevalence of parkinsonian signs can reach 86% in iNPH [16], suggesting that iNPH may cause a generalized dysfunction of the motor system. In the present study focusing on patients with suspected iNPH, the prevalence rate of parkinsonism was twice as high for iNPH mimics than that for iNPH (40.3% versus 20.3%). At the individual level, the presence of parkinsonism in a patient with suspected iNPH should alert the physician to a potential diagnosis of iNPH mimics. Interestingly, the current association between parkinsonism and iNPH mimics was not explained by the location (even in the basal ganglia or the frontal lobe) or the burden of the white matter changes, strongly suggesting a non-vascular mechanism (i.e., neurodegenerative).
Following our prediction, symptoms of the triad of iNPH were similar between iNPH and iNPH mimics, like the repartition of parkinsonian signs between iNPH and iNPH mimics. Interestingly, postural instability was the most prevalent parkinsonian sign in patients with iNPH and was previously identified as a bad clinical predictor for shunt response [17]. If expensive and time consuming quantitative assessments of motor and non-motor symptoms of the triad of iNPH can distinguish iNPH from its mimics [3, 8], the routine clinical assessment of these symptoms is unspecific for the correct identification of iNPH.
It is important to note that our iNPH group included a majority of patients with possible iNPH (88.6%), reflecting an outpatient neurology clinic for patient with suspicion of iNPH and not a biaised iNPH neurosurgery clinic. Interestingly, in a neuropathological study of iNPH patients selected for shunt surgery, 89% of these patients presented concurrent brain histopathology (mainly cerebrovascular disease or Alzheimer’s disease changes) on brain biopsy [18].
Comparing parkinsonism, iNPH symptoms and ARWMC between iNPH patients and its mimics in such a large cohort of patients with suspected iNPH represents the main strength of this study. Further studies should include a quantification of the parkinsonism (i.e., UPDRS total score) or the location of the parkinsonism (axial versus peripheral) in this comparison between iNPH and its mimics. The generalization of the study findings should be limited to patients that are able to walk, because none of our included patients presented a iNPH grading scale of 4 for the subitem of gait disturbance (walking not possible). Finally, the absence of pathological confirmation represents a main limitation of this study.
In conclusion, the presence of parkinsonism is more prevalent in iNPH mimics than in iNPH and could represent a contributive clinical feature to discriminate between these both conditions. Furthermore, this association between parkinsonism and iNPH mimics is not explained by white matter disease and suggests a neurodegenerative process.
