Introduction

Idiopathic normal pressure hydrocephalus (iNPH) is a syndrome affecting the elderly population and characterized by gait and balance disturbances, urinary incontinence, and cognitive impairment in the setting of ventriculomegaly and cerebrospinal fluid (CSF) opening pressure within the normal range [1, 2].

Ventriculoperitoneal shunt (VPS) is the current standard treatment for iNPH patients [3], leading to long-lasting clinical improvement [4]. While VPS is the most common surgical approach in North America and Europe [3], lumboperitoneal shunt (LPS) is increasingly performed in Japan [5,6,7]. Despite shorter surgery time and lower risks of intracranial complications (e.g. hemorrhages, seizures, and infections) compared to VPS [8], LPS is not widely performed and is not considered as the first-choice treatment for iNPH [5,6,7, 9, 10]. This may be due to associated higher failure rates and the possibility of symptomatic overdrainage, especially in the past [8].

Several studies reported the safety and non-inferiority of LPS compared to VPS in improving symptoms in iNPH patients [5,6,7, 9,10,11]. However, prospective, controlled trials are few and incomplete, as—for instance—they lack a standardized outcome assessment. Furthermore, to date, there are no studies evaluating preoperative magnetic resonance imaging (MRI) predictors of outcome after LPS as well as neuroimaging correlates of clinical improvement.

In this prospective, controlled, monocentric study, we first aimed to describe the clinical and neuroimaging 12-month follow-up of two parallel cohorts of iNPH patients, who did or did not undergo LPS as first-line surgical treatment. Second, we investigated preoperative MRI predictors of outcome after LPS.

Methods

We consecutively recruited 78 patients with probable iNPH, diagnosed according to the iNPH International Guidelines [1]. Subjects were enrolled at the “Parkinson's Disease and Movement Disorders Unit” of the Mondino Foundation in Pavia (Italy) from January 2016 to June 2018. At baseline, all patients underwent clinical and neuropsychological assessments, 3 T MRI scan, and tap test (i.e. large volume lumbar puncture). Gait, cognitive functions, balance, and urinary continence were evaluated with the iNPH Rating Scale (iNPHRS), where a value of 100 represents normal function and 0 the most severe level of dysfunction [12]. Gait was also assessed measuring the number of seconds and steps needed for ten-meter walking at free pace along a straight line, turning around, and walking back to the starting position, with a walking aid if usually adopted. The mean values of two consecutive video-recorded trials were used for the ten-meter seconds and steps. The neuropsychological evaluation also included the education-adjusted scores of the Mini Mental State Examination (MMSE) [13] and the Montreal Cognitive Assessment (MoCA) [14].

MRI scans (3 T Skyra, Siemens, Erlangen, Germany) were assessed for calculating the Evans’ index [15] on axial T1-weighted images and for investigating the following supportive features [1, 2]: small callosal angle [16] on coronal T1-weighted images at the level of the posterior commissure; aqueductal or fourth ventricular flow void [17] on sagittal T2-weighted images; disproportionately enlarged subarachnoid space hydrocephalus (DESH) [18], i.e. obliteration of the high-convexity sulci on axial T1-weighted images and dilation of the Sylvian fissures on coronal T1-weighted images. Periventricular (PWM) and deep white matter (DWM) hyperintensities on axial FLAIR images were assessed with the Fazekas scale [19].

A positive response to tap test [20], defined as at least 10% improvement of ten-meter seconds or steps, was evaluated in all subjects.

A surgical CSF diversion was offered to all patients but 34 subjects declined in favor of a “wait and see” approach given only mild impairment of activities of daily living and less often fear of surgery. These patients could reconsider shunt anytime but nobody opted for surgery throughout the follow-up period. Among the remaining 44 subjects, all patients with a negative response to tap test had a positive response to external lumbar drainage [21]. The 44 subjects accepting surgery underwent lumboperitoneal shunt (LPS) within 1 month after baseline at the “Neurosurgery Unit” of the San Matteo Foundation in Pavia, after exclusion of spinal canal stenosis or previous lumbar spine arthrodesis. In all cases, an adjustable pressure valve (Medtronic PS Medical® Strata® NSC LPS) was used. A one-step procedure was employed by placing the patient in the lateral position with valve housing in a lumbar pocket.

All patients who did (LPS group) or did not (control group) undergo surgery were then followed up for 12 months through scheduled clinical and neuropsychological evaluations every 6 months. 3 T MRI scan was repeated at 12-month follow-up.

Statistical analyses

Between-group comparisons were performed with Student’s t test or Wilcoxon rank-sum test for continuous variables when appropriate, while Pearson’s chi-square test was used for categorical variables. Spearman’s rank correlation coefficient was applied for bivariate analyses. Within-group follow-up differences were obtained with Wilcoxon signed-rank test for continuous variables and McNemar’s test for categorical variables. We also calculated the mean percentage rates of change at follow-up evaluations compared to the baseline as follows: [(b − a)/a] x 100, where ‘a’ is the mean baseline score and ‘b’ is the mean follow-up score of a clinical rating scale. In particular, responders to LPS experienced at least 10% improvement at follow-up evaluations compared to the baseline. Nominal univariate logistic regression analyses were performed to identify preoperative MRI parameters predicting outcome after LPS through log-likelihood chi-square test. With this regard, a positive outcome was defined as at least 10% improvement of iNPHRS total score after LPS at 12-month follow-up. A value of p < 0.05 was considered significant. Statistical analyses were performed using JMP Pro 14.0 (SAS Institute Inc., USA).

Results

Demographic, clinical, and instrumental findings of iNPH patients at baseline are listed in Table 1. Compared to control subjects, before surgery the LPS cohort showed more severe gait and balance impairment, more frequent and more pronounced urinary urgency/incontinence. Furthermore, LPS patients more often had a positive response to tap test and higher Fazekas scores (PWM and total).

Table 1 Demographic, clinical, and instrumental findings of iNPH patients at baseline
Full size table

Significant correlations between clinical and MRI features of iNPH patients at baseline are shown in Table 2. Of relevance, gait and balance impairment positively correlated with PWM hyperintensities. Urinary incontinence was instead positively associated with DWM hyperintensities. Conversely, no relationship between cognitive dysfunction and MRI findings was detected.

Table 2 Significant correlations between clinical and MRI features of iNPH patients at baseline
Full size table

In the immediate post-surgery phase, four (9.1%) subjects complained of headache, which disappeared increasing the valve opening pressure. At the hospital discharge, the valve opening pressure was set to 1.5 (90–110 mmH2O) in 39 and 2.0 (145–165 mmH2O) in five patients. As a result of poor or absent improvement of motor performances within 1 month after surgery, the valve opening pressure was reduced in eight subjects, i.e. from 1.5 to 1.0 (35–55 mmH2O) in six and from 2.0 to 1.5 in two patients.

We reported no case of subdural hygroma or hematoma, shunt tube migration or obstruction, bowel perforation, meningitis or other infectious complications.

Within-group and between-group clinical follow-up differences are shown in Fig. 1 and Table 3, respectively. Responders to LPS are indicated in Table 3.

Fig. 1
figure 1

Clinical rating scales in iNPH patients who did or did not undergo LPS. Mean scores are shown. Within-group follow-up differences compared to the baseline were calculated (*p < 0.05, **p < 0.01, ***p < 0.001). Vertical error bars represent standard errors

Full size image
Table 3 Responders to LPS and clinical follow-up rates of change of iNPH patients undergoing LPS compared to controls
Full size table

6-Month clinical follow-up

Improvement of gait and balance was showed in the LPS group. To a lesser extent, the urinary continence also improved in the LPS cohort. An increase of iNPHRS total score with LPS was thus noticed. Instead, no significant cognitive changes were recognized in the LPS group. None of the patients developed shunt-related complications.

No significant variations were noted in the control group.

12-Month clinical follow-up

Improvement of gait and balance persisted in the LPS group. However, a trend to slightly decreased effectiveness was noted in this cohort if compared to the 6-month follow-up. No significant cognitive changes with LPS were recognized similarly to the previous evaluation. Change of urinary continence was no longer significant in the within-group analysis of the LPS cohort. However, improvement of urinary continence with LPS persisted in the between-group comparison. Accordingly, a reduction of iNPHRS total score was noted in the LPS cohort compared to the 6-month follow-up (p = 0.0047). Nevertheless, an increase of this score persisted in LPS patients both at within-group and between-group analyses compared to the baseline. None of the subjects had shunt-related complications.

On the other hand, control patients showed worsening of gait and urinary continence. In particular, the decrease of iNPHRS continence score in the control cohort was also significant if compared to the 6-month follow-up (p = 0.0120). In contrast, balance and cognitive functions were not significantly different in the untreated cohort. Hence, a decrease of iNPHRS total score was showed in the control group compared to both the baseline and the 6-month follow-up (p = 0.0230).

12-Month MRI follow-up

Compared to the baseline, small callosal angle and DESH were less frequently detected in the LPS group, whereas they did not significantly change in control patients (Fig. 2). Of note, Fazekas score for PWM and total score were reduced in the LPS group and increased in control subjects (Fig. 2). Conversely, Evans’ index, flow void sign, and Fazekas score for DWM did not significantly change at follow-up in any of the cohorts (Fig. 2).

Fig. 2
figure 2

MRI parameters in iNPH patients who did or did not undergo LPS. Mean scores for continuous variables and frequency distribution for categorical variables are shown. Within-group follow-up differences compared to the baseline were calculated (*p < 0.05, **p < 0.01, ***p < 0.001). Vertical error bars represent standard errors

Full size image

Interestingly, a decrease of Fazekas score for PWM and total score was identified in responders to LPS (Fig. 3). No other MRI parameters could significantly distinguish responders from not responders to LPS (Fig. 3).

Fig. 3
figure 3

MRI parameters in iNPH patients who did (Resp) or did not (No resp) respond to LPS. Mean scores for continuous variables and frequency distribution for categorical variables are shown. Within-group follow-up differences compared to the baseline were calculated (*p < 0.05, **p < 0.01, ***p < 0.001). Vertical error bars represent standard errors

Full size image

Outcome prediction after lumboperitoneal shunt

No preoperative MRI parameter was significant outcome predictor after LPS.

Discussion

In our study, we showed the safety and effectiveness of LPS in improving gait, balance, and to a lesser extent urinary continence without any significant influence on cognitive functions in iNPH patients. The efficacy of LPS occurred within 6 months after surgery and persisted with a slight reduction at 12-month follow-up.

In keeping with previous studies [3, 11], gait and balance demonstrated the highest mean improvement rates. However, we noted a lower average improvement for urinary continence and cognitive functions. In this regard, controversy still exists on whether urinary and neuropsychological effectiveness can be achieved after surgery, especially in terms of reversibility of cognitive impairment [1, 3, 22].

A direct comparison between our findings and few previous reports concerning the efficacy of LPS in iNPH is not reliable given the different methods used to determine and quantify the outcome. To date, three retrospective, uncontrolled studies [5, 9, 10] and two prospective, controlled trials [6, 7] reported the effectiveness of LPS in iNPH. Among the retrospective investigations, only one study assessed each typical clinical domain through rating scales [5]. Both prospective trials instead evaluated iNPH patients with objective outcome measures [6, 7]. However, several studies used the modified Rankin Scale, which was developed for subjects with stroke providing a general disability measure, and ordinal or nominal rating scales for assessment of symptoms severity in iNPH [5,6,7, 10]. Quantitative outcome measures, such as the iNPHRS, should be employed to maximize accuracy and validity of results [12]. Unfortunately, there is still no agreement on a standardized, unbiased, and practical system of measuring outcome for research or clinical use in iNPH [2, 23]. Of note, we standardized mean rates of change and provided a consistent definition of responsivity or positive outcome for each clinical rating scale.

Among previous prospective trials, one study was controlled for the first three months (after which also the control group underwent LPS) [6], whereas a multicentric investigation used a previously conducted VPS cohort study as historical control [7]. In our study, we considered iNPH patients who declined surgery as control cohort, further describing their natural history of progressive worsening of gait and urinary continence at 12-month follow-up. This partially agrees with a Swedish study, in which shunt was inadvertently delayed in a cohort of iNPH subjects, who showed worsening of gait, balance, and cognitive functions in the meantime [24]. Other authors also stated the progression of symptoms in iNPH patients while waiting for surgery [25].

Regardless of the subsequent response, among patients who underwent LPS, we observed a decreased frequency of small callosal angle and DESH at 12-month follow-up compared to the baseline, whereas Evans’ index and flow void sign did not significantly change. Our results are in keeping with reports showing trend to normalization of callosal angle [26, 27], disappearance of DESH [18, 28], and no remarkable reduction of Evans’ index [29] after shunt. Conversely, few studies noticed a significant postoperative decrease of Evans’ index [26, 27], which is, however, commonly deemed as sign of overdrainage.

In our study, no baseline MRI parameter was significant outcome predictor. The lack of correlation between preoperative morphologic MRI markers and post-surgery outcome corroborates some studies [26, 29,30,31] and argues against other investigations that reported associations between postoperative improvement and baseline presence of small callosal angle [32], DESH [33], flow void sign [17], and white matter hyperintensities [28, 34].

Of interest, in responders to LPS, we found postoperative reduction of PWM hyperintensities without significant changes of DWM hyperintensities. This may be related to post-shunt improvement of gait and balance disturbances, which were positively associated with PWM hyperintensities at baseline. These findings may also partially explain the less noticeable improvement of continence and cognitive functions, which demonstrated no correlation with PWM hyperintensities at baseline. Control subjects instead showed an increase of PWM hyperintensities at 12-month follow-up, further corroborating the relationship between motor impairment and PWM involvement. In agreement with our results, several studies in iNPH reported reversible white matter lesions, mostly in periventricular regions, and their possible relationship with positive outcome after surgery [30, 34, 35]. Interestingly, acetazolamide was also able to reduce PWM hyperintensities and improve gait in iNPH, with a mechanism probably similar to the effect of shunt [36]. PWM hyperintensities can reflect transependymal resorption secondary to reverse flow of CSF or compression resulting from raised intraventricular pressure [37, 38]. Both these mechanisms can lead to increased diffusivity [38] and decreased perfusion in periventricular regions [39, 40]. Accordingly, reduced diffusivity [38, 40], increased perfusion [39, 40], and improvement of the axonal and myelin damage [41] may play a role in PWM changes after surgery. In particular, it is considered that the decreased compression of periventricular regions may be associated with gait improvement after shunt [42].

In our study, we acknowledge several limitations. First, our trial was not randomized. However, randomized controlled trials and sham surgery may be unethical in iNPH as a delayed surgery has significant long-term effects on functional outcome [24]. On the other hand, adopting patients unwilling to undergo shunt as control cohort is also biased as these subjects had milder impairment compared to LPS patients. Still, this notion strengthens our findings as we observed an improvement despite more severe impairment of gait, balance, and urinary continence in the LPS cohort at baseline. In this regard, a greater impairment might lead iNPH patients to choose surgery. Second, this study was not blinded, thus meaning that part of the reported outcome may be due to a placebo effect. Interestingly, however, the reversibility of an MRI parameter (i.e. PWM hyperintensities) and its correlation with the clinical outcome substantiate a real effect beyond placebo. Finally, our study has a rather short follow-up, especially given the recent controversy around the real nosological value of iNPH [23]. Nevertheless, our study duration is in keeping with most trials published so far [5,6,7, 9, 10].

In conclusion, our study suggests that LPS is safe and effective in improving symptoms, mainly gait disturbances and balance impairment in iNPH. Hence, this technique would deserve to be considered among the first-line treatments for iNPH, particularly for fragile patients unable to undergo intracranial procedures. An early shunt is desirable to prevent clinical worsening. Of note, post-surgery decrease of PWM hyperintensities may be a useful MRI marker surrogate for clinical effectiveness of LPS. Future studies with blinded evaluators, longer follow-ups, and possible direct comparisons with VPS are certainly warranted.