Browse the corpus

A Randomized Trial of Shunting for Idiopathic Normal-Pressure Hydrocephalus. BACKGROUND: Idiopathic normal-pressure hydrocephalus is a neurologic disorder characterized by impaired gait, balance, cognition, and bladder control in older adults. The disorder is treated with shunt surgery, but the effectiveness of shunting is unclear. METHODS: We conducted a double-blind, randomized, placebo-controlled trial involving participants selected for shunt surgery on the basis of gait-velocity improvement with cerebrospinal fluid (CSF) drainage. Participants were randomly assigned to an open-shunt valve setting (opening pressure, 110 mm of water) or a placebo valve setting (opening pressure, >400 mm of water) of a noninvasively adjustable shunt. The primary outcome was the change in gait velocity 3 months after surgery. Secondary outcomes were the change at 3 months in the Tinetti scale total score (range, 0 to 28; lower scores indicate worse gait and balance), Montreal Cognitive Assessment (MoCA) score (range, 0 to 30; lower scores indicate worse cognition), and Overactive Bladder Questionnaire score (range, 0 to 100; higher scores indicate worse urinary incontinence). RESULTS: A total of 99 participants underwent randomization and received the assigned intervention. At 3 months, gait velocity had increased in the open-shunt group (mean [±SD] change, 0.23±0.23 m per second; assessed in 49 participants) and was unchanged in the placebo group (mean change, 0.03±0.23 m per second; assessed in 49 participants), resulting in a treatment difference of 0.21 m per second (95% confidence interval, 0.12 to 0.31; P<0.001). A significantly greater improvement in the open-shunt group than the placebo group was seen for the Tinetti scale score (mean change, 2.9 points vs. 0.5 points; P = 0.003) but not the MoCA score (1.3 points vs. 0.3 points) or the Overactive Bladder Questionnaire score (-3.3 points vs. -1.5 points). The results regarding adverse events were mixed, with more participants in the placebo group reporting falls (46% vs. 24%), an equal percentage having cerebral bleeding (2% in both groups), and more participants in the open-shunt group having subdural bleeding (12% vs. 2%) and positional headaches (59% vs. 28%). CONCLUSIONS: Among participants with idiopathic normal-pressure hydrocephalus who had a response to temporary CSF drainage, shunting resulted in significant improvements at 3 months in gait velocity and a measure of gait and balance but not in measures of cognition or incontinence. (Funded by the National Institute of Neurological Disorders and Stroke and the Trial Innovation Network; PENS ClinicalTrials.gov number, NCT05081128.).

Idiopathic normal pressure hydrocephalus (iNPH) is a disorder of neurologic gait and balance impairment, urinary urgency and incontinence, cognitive impairment, and cerebral ventriculomegaly in individuals over age 60 first described in 1965.1 The prevalence of iNPH is estimated as high as 1.5% (1,500/100,000) for ages 65-70 years, and increases with age, reaching as much as 7.7% for ages ≥86 years.2-4 Untreated, iNPH is associated with neurological impairment and mortality.5-7 While no medical treatment exists, the symptoms of iNPH have long been reported to be reversible with surgical implantation of a cerebrospinal fluid (CSF) shunt, most commonly ventriculoperitoneal (VP) or lumboperitoneal shunts.8,9 The 2005 International iNPH Guidelines and the Japanese Guidelines standardized patient selection for shunt surgery 10,11, resulting in studies where 70-80% of patients selected on the basis of response to temporary CSF drainage or CSF infusion testing improved. 8,9,12-14

loperitoneal (VP) or lumboperitoneal shunts.8,9 The 2005 International iNPH Guidelines and the Japanese Guidelines standardized patient selection for shunt surgery 10,11, resulting in studies where 70-80% of patients selected on the basis of response to temporary CSF drainage or CSF infusion testing improved. 8,9,12-14 Nonetheless, the clinical effectiveness of shunt surgery for iNPH has been subject to skepticism to the point of a call for a moratorium on shunt surgery because of variability of study results, questionable durability of benefit, the risks of surgery, and the potential for a strong placebo effect.15,16 Three small-sample randomized placebo-controlled trials suggested that shunt surgery was effective, but a 2024 Cochrane review noted, “there is a need for similar studies to increase the certainty of the findings presented”.17-20 Therefore, we designed and conducted the Placebo-Controlled Effectiveness in iNPH Shunting (PENS) Trial.

This is an international, multi-center, prospective, double-blind, randomized, placebo-controlled trial. The study intervention was the initial setting of a commercially available noninvasively adjustable shunt valve (Codman Certas Plus with SiphonGuard®, Integra LifeSciences, Princeton, NJ) randomized either to an Open Shunt setting or a Placebo setting. The investigators designed the study (JM, MAW, AE, MGL, MGH, HLK, NAD, AM, JH, HJ, RH) and wrote the manuscript, vouch for the data, and agreed to publish the paper (MGL, MAW, MGH, HLK, NAD, AM, JH, HJ, DFH, RH). The Johns Hopkins University Clinical Coordinating and Administrative Center gathered the data, and the University of Utah Data Coordinating Center analyzed the data. Study sites were selected based on their experience in managing iNPH. For U.S. sites, the central Institutional Review Board (IRB) was at Johns Hopkins Medicine (IRB00305245), with local IRBs at the Canadian and Swedish sites. Additional details are in the Supplementary Appendix.

Coordinating Center analyzed the data. Study sites were selected based on their experience in managing iNPH. For U.S. sites, the central Institutional Review Board (IRB) was at Johns Hopkins Medicine (IRB00305245), with local IRBs at the Canadian and Swedish sites. Additional details are in the Supplementary Appendix. Following the International iNPH Guidelines, patients selected for VP shunt surgery based on their gait velocity response to temporary CSF drainage were eligible. At baseline, patients who could not walk 10 meters or who required assistance from a person to walk were excluded, as were those with gait velocity ≥1.0 m/sec unless velocity improved ≥30% with CSF drainage. Comorbidities that did not interfere with participation in study assessments were allowed. Patients with secondary hydrocephalus, previous cranial neurosurgical procedures or who required chronic anticoagulation were excluded. Placebo was assigned setting 8 (opening pressure >400 mm H2O).21 Open Shunt was assigned setting 4 (110 mm H2O); however, settings 5 or 6 (145 or 180 mm H2O) were allowed at the surgeon’s discretion for patient safety.21 Randomization occurred immediately before surgery in a 1:1 ratio. Treatment assignments were concealed from study participants and personnel, except for investigators responsible for shunt adjustments, and were disclosed only for patient safety reasons. After completion of 3-month endpoint evaluations, shunts in both groups were adjusted to an open setting while concealing the initial setting. Participants were informed that the shunt would remain open.

Placebo was assigned setting 8 (opening pressure >400 mm H2O).21 Open Shunt was assigned setting 4 (110 mm H2O); however, settings 5 or 6 (145 or 180 mm H2O) were allowed at the surgeon’s discretion for patient safety.21 Randomization occurred immediately before surgery in a 1:1 ratio. Treatment assignments were concealed from study participants and personnel, except for investigators responsible for shunt adjustments, and were disclosed only for patient safety reasons. After completion of 3-month endpoint evaluations, shunts in both groups were adjusted to an open setting while concealing the initial setting. Participants were informed that the shunt would remain open. Because eligibility was based on response to CSF drainage, an interval of ≥4 weeks was required to allow symptoms to return before performing baseline study assessments. The primary endpoint was 3 months after surgery. Outcomes to 12 months of open shunt condition are being collected, including detailed neuropsychology testing, and CSF and imaging biomarkers not included in this report. The primary outcome measure was the change from baseline in maximum gait velocity (m/sec) over three 10-meter trials. Participants were instructed to walk as fast as they felt safe without running to a goal 2 meters beyond the finish line. Time was recorded from a standing start (toes at starting line) until one foot crossed the finish line. Participants were permitted assistive devices, but not assistance from another person.

trials. Participants were instructed to walk as fast as they felt safe without running to a goal 2 meters beyond the finish line. Time was recorded from a standing start (toes at starting line) until one foot crossed the finish line. Participants were permitted assistive devices, but not assistance from another person. Secondary outcome measures were the Performance-Oriented Mobility Assessment (Tinetti Scale),22 the Montreal Cognitive Assessment version 8.1 (MoCA),23 and the Overactive Bladder Questionnaire, Short Form (OABQsf).24 Tertiary outcome measures were the Symbol Digit Modalities Test (SDMT) 25; the NIH Toolbox Flanker Inhibitory Control and Attention (NIH-Flanker) and Dimensional Change Card Sort (NIH-DCCS) subtests 26; the Beck Depression Inventory (BDI-II) 27; the Lawton Activities of Daily Living/Instrumental Activities of Daily Living questionnaire (Lawton ADL/IADL) 28; the modified Rankin Scale (mRS),29,30 and the EQ-5D-5L Visual Analog Scale (EQ-VAS) for quality of life.31 MRI evaluation of lateral ventricular volume and the Evans ratio was performed on de-identified T1 images using segmentation from spatially localized atlas network tiles and a deep-learning “brain extraction tool”.32,33

Tertiary outcome measures were the Symbol Digit Modalities Test (SDMT) 25; the NIH Toolbox Flanker Inhibitory Control and Attention (NIH-Flanker) and Dimensional Change Card Sort (NIH-DCCS) subtests 26; the Beck Depression Inventory (BDI-II) 27; the Lawton Activities of Daily Living/Instrumental Activities of Daily Living questionnaire (Lawton ADL/IADL) 28; the modified Rankin Scale (mRS),29,30 and the EQ-5D-5L Visual Analog Scale (EQ-VAS) for quality of life.31 MRI evaluation of lateral ventricular volume and the Evans ratio was performed on de-identified T1 images using segmentation from spatially localized atlas network tiles and a deep-learning “brain extraction tool”.32,33 Research coordinators were trained to administer the gait and neuropsychology assessments. All gait assessments were video recorded for blinded review, quality control and feedback. Recordings were not used for data acquisition. Research coordinators were approved by a study neuropsychologist. MoCA certification was required. For quality control, discrepancies in administration and scoring were reviewed, and training was provided when needed. Predefined safety endpoints included overall falls, clinically significant falls, surgical complications, and complications related to CSF over-drainage, including subdural hygromas and hematomas. Pre-defined thresholds for stopping the study were regularly reviewed with the Data Safety Monitoring Board (DSMB). Shunt settings were checked at each visit and after MRI scans, and if needed, adjusted to the appropriate setting while concealing the initial shunt setting.

ge, including subdural hygromas and hematomas. Pre-defined thresholds for stopping the study were regularly reviewed with the Data Safety Monitoring Board (DSMB). Shunt settings were checked at each visit and after MRI scans, and if needed, adjusted to the appropriate setting while concealing the initial shunt setting. A sample size of 100 participants was specified, providing >90% power to detect a between group gait velocity difference of 0.2 m/s (SD=0.29 m/s), accounting for interim analyses and attrition. 19 Two conservative interim efficacy analyses for 3-month outcomes were prespecified upon completion of approximately one-third and two-thirds of the target population. 34 Reported analyses implemented multiple imputation, which assumes that differences between missing and available 3-month outcomes can be accounted for by available patient baseline and follow-up information. Three-month outcomes were analyzed using linear regression with treatment arm as the predictor, adjusting for baseline values and prespecified covariates: age for all outcomes, and educational attainment for neuropsychological measures. Separate covariances were modeled by treatment arm. Robustness assessments included: replicating trial results using the average of the 3 gait trials; excluding patients with off-protocol valve adjustments (“per-protocol”); using the treatment implemented as of the 3-month evaluation as a predictor (“as-treated”); and modeling using imputed missing outcome data (Tables S8 - S11).

assessments included: replicating trial results using the average of the 3 gait trials; excluding patients with off-protocol valve adjustments (“per-protocol”); using the treatment implemented as of the 3-month evaluation as a predictor (“as-treated”); and modeling using imputed missing outcome data (Tables S8 - S11). Significance testing was 2-sided, with α=0.05 for the primary outcome. Overall Type 1 error for secondary outcomes was controlled using a Bonferroni-Holm procedure.35 Tertiary outcome analyses were not controlled for multiplicity; results are shown as point estimates with 95% confidence intervals without formal hypothesis testing. A post-hoc logistic regression analysis of dichotomized 3-month mRS (values grouped as 0-2 and 3-6) was also performed controlling for treatment arm as the predictors, adjusting for baseline mRS. The proportion of participants with complications was compared between treatment arms using two-sided Fisher’s exact testing with mid-p-value correction.36

Following the International iNPH Guidelines, patients selected for VP shunt surgery based on their gait velocity response to temporary CSF drainage were eligible. At baseline, patients who could not walk 10 meters or who required assistance from a person to walk were excluded, as were those with gait velocity ≥1.0 m/sec unless velocity improved ≥30% with CSF drainage. Comorbidities that did not interfere with participation in study assessments were allowed. Patients with secondary hydrocephalus, previous cranial neurosurgical procedures or who required chronic anticoagulation were excluded.

Placebo was assigned setting 8 (opening pressure >400 mm H2O).21 Open Shunt was assigned setting 4 (110 mm H2O); however, settings 5 or 6 (145 or 180 mm H2O) were allowed at the surgeon’s discretion for patient safety.21 Randomization occurred immediately before surgery in a 1:1 ratio. Treatment assignments were concealed from study participants and personnel, except for investigators responsible for shunt adjustments, and were disclosed only for patient safety reasons. After completion of 3-month endpoint evaluations, shunts in both groups were adjusted to an open setting while concealing the initial setting. Participants were informed that the shunt would remain open.

Because eligibility was based on response to CSF drainage, an interval of ≥4 weeks was required to allow symptoms to return before performing baseline study assessments. The primary endpoint was 3 months after surgery. Outcomes to 12 months of open shunt condition are being collected, including detailed neuropsychology testing, and CSF and imaging biomarkers not included in this report. The primary outcome measure was the change from baseline in maximum gait velocity (m/sec) over three 10-meter trials. Participants were instructed to walk as fast as they felt safe without running to a goal 2 meters beyond the finish line. Time was recorded from a standing start (toes at starting line) until one foot crossed the finish line. Participants were permitted assistive devices, but not assistance from another person. Secondary outcome measures were the Performance-Oriented Mobility Assessment (Tinetti Scale),22 the Montreal Cognitive Assessment version 8.1 (MoCA),23 and the Overactive Bladder Questionnaire, Short Form (OABQsf).24 Tertiary outcome measures were the Symbol Digit Modalities Test (SDMT) 25; the NIH Toolbox Flanker Inhibitory Control and Attention (NIH-Flanker) and Dimensional Change Card Sort (NIH-DCCS) subtests 26; the Beck Depression Inventory (BDI-II) 27; the Lawton Activities of Daily Living/Instrumental Activities of Daily Living questionnaire (Lawton ADL/IADL) 28; the modified Rankin Scale (mRS),29,30 and the EQ-5D-5L Visual Analog Scale (EQ-VAS) for quality of life.31

and Dimensional Change Card Sort (NIH-DCCS) subtests 26; the Beck Depression Inventory (BDI-II) 27; the Lawton Activities of Daily Living/Instrumental Activities of Daily Living questionnaire (Lawton ADL/IADL) 28; the modified Rankin Scale (mRS),29,30 and the EQ-5D-5L Visual Analog Scale (EQ-VAS) for quality of life.31 MRI evaluation of lateral ventricular volume and the Evans ratio was performed on de-identified T1 images using segmentation from spatially localized atlas network tiles and a deep-learning “brain extraction tool”.32,33

Research coordinators were trained to administer the gait and neuropsychology assessments. All gait assessments were video recorded for blinded review, quality control and feedback. Recordings were not used for data acquisition. Research coordinators were approved by a study neuropsychologist. MoCA certification was required. For quality control, discrepancies in administration and scoring were reviewed, and training was provided when needed. Predefined safety endpoints included overall falls, clinically significant falls, surgical complications, and complications related to CSF over-drainage, including subdural hygromas and hematomas. Pre-defined thresholds for stopping the study were regularly reviewed with the Data Safety Monitoring Board (DSMB). Shunt settings were checked at each visit and after MRI scans, and if needed, adjusted to the appropriate setting while concealing the initial shunt setting.

Predefined safety endpoints included overall falls, clinically significant falls, surgical complications, and complications related to CSF over-drainage, including subdural hygromas and hematomas. Pre-defined thresholds for stopping the study were regularly reviewed with the Data Safety Monitoring Board (DSMB). Shunt settings were checked at each visit and after MRI scans, and if needed, adjusted to the appropriate setting while concealing the initial shunt setting.

A sample size of 100 participants was specified, providing >90% power to detect a between group gait velocity difference of 0.2 m/s (SD=0.29 m/s), accounting for interim analyses and attrition. 19 Two conservative interim efficacy analyses for 3-month outcomes were prespecified upon completion of approximately one-third and two-thirds of the target population. 34 Reported analyses implemented multiple imputation, which assumes that differences between missing and available 3-month outcomes can be accounted for by available patient baseline and follow-up information. Three-month outcomes were analyzed using linear regression with treatment arm as the predictor, adjusting for baseline values and prespecified covariates: age for all outcomes, and educational attainment for neuropsychological measures. Separate covariances were modeled by treatment arm. Robustness assessments included: replicating trial results using the average of the 3 gait trials; excluding patients with off-protocol valve adjustments (“per-protocol”); using the treatment implemented as of the 3-month evaluation as a predictor (“as-treated”); and modeling using imputed missing outcome data (Tables S8 - S11).

Based on inclusion/exclusion criteria, 300 of 1424 patients undergoing CSF drainage between April 25, 2022 and January 29, 2025 were eligible. Of 237 approached for the study, 125 (53%) provided consent and 99 were randomized, 49 to Open Shunt and 50 to Placebo. In the Placebo group, 3 participants withdrew from the study and one died before the primary endpoint (Figure S1, Tables S1-S5). Participants were 51.5% male and 48.5% female, with mean age 75.0 ± 5.7 years, and 63% with a Bachelor’s degree or higher. The 2 study groups were similar in age, sex, educational attainment, and comorbidities (Tables 1 and 2, Table S6). After the first planned interim analysis on February 5, 2024, the DSMB continued the study. After the second interim analysis (74 of 99 randomized participants with 3-month outcomes) on January 29, 2025 showed a significant treatment effect favoring Open Shunt (observed P<0.001; monitoring boundary for stopping was p<0.024), the DSMB stopped enrollment while allowing the study to continue to completion at 12 months of open shunt condition. After consulting with the IRB, the shunts for two Placebo participants who had not reached the 3-month endpoint were changed to an open setting; they are included in the study analyses.

ing was p<0.024), the DSMB stopped enrollment while allowing the study to continue to completion at 12 months of open shunt condition. After consulting with the IRB, the shunts for two Placebo participants who had not reached the 3-month endpoint were changed to an open setting; they are included in the study analyses. Gait velocity increased for Open Shunt (0.23 ± 0.23 m/s) and was unchanged for Placebo (0.03 ± 0.23 m/s), for a treatment difference of 0.21 m/s (P<0.001; 95% confidence interval 0.12 to 0.31) (Table 3). Figure 1 shows the distribution of gait velocity change, excluding imputed missing data. For 80% (37/46) of Open Shunt and 24% (10/42) of Placebo, gait velocity change was ≥0.1 m/s, exceeding the “substantial meaningful change” of 0.10 m/s in the elderly.37 A significant treatment difference favoring Open Shunt vs Placebo was seen for the Tinetti (2.9 vs 0.5, P = 0.003), but not the MoCA (1.3 vs 0.3) or OABQsf (−3.3 vs −1.5) (Table 3). A difference favoring Open Shunt appeared consistent for all measures, except the BDI-II (Table 3). The EQ-VAS appeared increased for Open Shunt and decreased for Placebo (10.9 vs −3.6), for a difference of 12.4 (95% confidence interval 4.4 to 20.4). The proportion with mRS 0-2 appeared to increase from 56% to 78% for Open Shunt and decrease from 45% to 43% for Placebo (Figure S2). The lateral ventricular volume appeared to decrease more for Open Shunt than Placebo (−27.4 vs −4.4), for a difference of −21.0 mL (95% confidence interval −33.9 to −8.1) (Table 3).

A difference favoring Open Shunt appeared consistent for all measures, except the BDI-II (Table 3). The EQ-VAS appeared increased for Open Shunt and decreased for Placebo (10.9 vs −3.6), for a difference of 12.4 (95% confidence interval 4.4 to 20.4). The proportion with mRS 0-2 appeared to increase from 56% to 78% for Open Shunt and decrease from 45% to 43% for Placebo (Figure S2). The lateral ventricular volume appeared to decrease more for Open Shunt than Placebo (−27.4 vs −4.4), for a difference of −21.0 mL (95% confidence interval −33.9 to −8.1) (Table 3). Safety monitoring endpoints and serious adverse events (SAE) are shown in Tables 4 and S7. More participants reported falls for Placebo (23/50; 46%) than Open Shunt (12/49; 24.5%) (P=0.03). The number with clinically significant falls was low and not different between groups (Placebo=3; Open Shunt=2).

The lateral ventricular volume appeared to decrease more for Open Shunt than Placebo (−27.4 vs −4.4), for a difference of −21.0 mL (95% confidence interval −33.9 to −8.1) (Table 3). Safety monitoring endpoints and serious adverse events (SAE) are shown in Tables 4 and S7. More participants reported falls for Placebo (23/50; 46%) than Open Shunt (12/49; 24.5%) (P=0.03). The number with clinically significant falls was low and not different between groups (Placebo=3; Open Shunt=2). Thirteen Open Shunt and 9 Placebo participants had SAEs. One death occurred in the Placebo group following a parenchymal hemorrhage during surgery and subsequent complications that were unrelated to the treatment assignment. There were more subdural hematomas (SDH) for Open Shunt (n=6; 12%) than Placebo (n=1; 2.0%) (P=0.04). Three Open Shunt participants required surgical drainage, middle meningeal artery embolization, or both. Of these, one resolved by 3 months, one required shunt revision surgery, and one required shunt removal due to wound dehiscence. Four small SDH were addressed with either shunt setting adjustment (n=3), of which 2 resolved by 3 months; or no treatment (n=1), which was not resolved by 3 months. There were 3 subdural hygromas for Open Shunt and 0 for Placebo (P=0.06); all were treated with shunt setting adjustment.

removal due to wound dehiscence. Four small SDH were addressed with either shunt setting adjustment (n=3), of which 2 resolved by 3 months; or no treatment (n=1), which was not resolved by 3 months. There were 3 subdural hygromas for Open Shunt and 0 for Placebo (P=0.06); all were treated with shunt setting adjustment. More participants reported positional headaches suggesting low CSF pressure for Open Shunt (29/49; 59.2%) than Placebo (14/50; 28.0%) (P=0.002). More initial shunt setting changes before the primary endpoint (i.e., off-protocol) occurred for Open Shunt (n=13; 26.5%) than Placebo (n=5; 10.0%) (P=0.03). For Open Shunt, the setting was raised for low pressure symptoms in 13/13. For Placebo, the setting was lowered, i.e., changed to an open setting, for under-drainage symptoms in 3, and following the DSMB decision in 2.

Gait velocity increased for Open Shunt (0.23 ± 0.23 m/s) and was unchanged for Placebo (0.03 ± 0.23 m/s), for a treatment difference of 0.21 m/s (P<0.001; 95% confidence interval 0.12 to 0.31) (Table 3). Figure 1 shows the distribution of gait velocity change, excluding imputed missing data. For 80% (37/46) of Open Shunt and 24% (10/42) of Placebo, gait velocity change was ≥0.1 m/s, exceeding the “substantial meaningful change” of 0.10 m/s in the elderly.37 A significant treatment difference favoring Open Shunt vs Placebo was seen for the Tinetti (2.9 vs 0.5, P = 0.003), but not the MoCA (1.3 vs 0.3) or OABQsf (−3.3 vs −1.5) (Table 3). A difference favoring Open Shunt appeared consistent for all measures, except the BDI-II (Table 3). The EQ-VAS appeared increased for Open Shunt and decreased for Placebo (10.9 vs −3.6), for a difference of 12.4 (95% confidence interval 4.4 to 20.4). The proportion with mRS 0-2 appeared to increase from 56% to 78% for Open Shunt and decrease from 45% to 43% for Placebo (Figure S2). The lateral ventricular volume appeared to decrease more for Open Shunt than Placebo (−27.4 vs −4.4), for a difference of −21.0 mL (95% confidence interval −33.9 to −8.1) (Table 3).

Gait velocity increased for Open Shunt (0.23 ± 0.23 m/s) and was unchanged for Placebo (0.03 ± 0.23 m/s), for a treatment difference of 0.21 m/s (P<0.001; 95% confidence interval 0.12 to 0.31) (Table 3). Figure 1 shows the distribution of gait velocity change, excluding imputed missing data. For 80% (37/46) of Open Shunt and 24% (10/42) of Placebo, gait velocity change was ≥0.1 m/s, exceeding the “substantial meaningful change” of 0.10 m/s in the elderly.37

A difference favoring Open Shunt appeared consistent for all measures, except the BDI-II (Table 3). The EQ-VAS appeared increased for Open Shunt and decreased for Placebo (10.9 vs −3.6), for a difference of 12.4 (95% confidence interval 4.4 to 20.4). The proportion with mRS 0-2 appeared to increase from 56% to 78% for Open Shunt and decrease from 45% to 43% for Placebo (Figure S2).

Safety monitoring endpoints and serious adverse events (SAE) are shown in Tables 4 and S7. More participants reported falls for Placebo (23/50; 46%) than Open Shunt (12/49; 24.5%) (P=0.03). The number with clinically significant falls was low and not different between groups (Placebo=3; Open Shunt=2). Thirteen Open Shunt and 9 Placebo participants had SAEs. One death occurred in the Placebo group following a parenchymal hemorrhage during surgery and subsequent complications that were unrelated to the treatment assignment. There were more subdural hematomas (SDH) for Open Shunt (n=6; 12%) than Placebo (n=1; 2.0%) (P=0.04). Three Open Shunt participants required surgical drainage, middle meningeal artery embolization, or both. Of these, one resolved by 3 months, one required shunt revision surgery, and one required shunt removal due to wound dehiscence. Four small SDH were addressed with either shunt setting adjustment (n=3), of which 2 resolved by 3 months; or no treatment (n=1), which was not resolved by 3 months. There were 3 subdural hygromas for Open Shunt and 0 for Placebo (P=0.06); all were treated with shunt setting adjustment.

This study provides evidence that VP shunt surgery is an effective treatment for improving gait velocity in patients with iNPH selected based on response to temporary CSF drainage in accordance with the International iNPH Guidelines. Within 3 months of surgery, participants with an Open Shunt setting experienced a clinically meaningful improvement in the primary outcome measure of gait velocity. Small improvements in gait and balance on the Tinetti scale were also seen, but there was no significant change in screening measures of cognition or bladder symptoms. However, tertiary outcome measures for cognition, functional independence, and quality of life appeared improved. Open shunting appeared to reduce lateral ventricular volume, consistent with a functioning shunt. In contrast, for Placebo, gait velocity was unchanged, and nearly all secondary and tertiary outcome measures appeared unchanged or worse, including lateral ventricular volume. For 80% of Open Shunt the gait velocity change surpassed the 0.10 m/s “substantial meaningful change” for the elderly population.37 For this degree of improvement, the number needed to treat (NNT) for shunt surgery is 1.79. Further, the mean gait velocity change for Open Shunt (0.23 m/s) is over twice the substantial meaningful change in the elderly and exceeds the “large” minimum clinically important difference of 0.22 m/s in Parkinsonism.38

on.37 For this degree of improvement, the number needed to treat (NNT) for shunt surgery is 1.79. Further, the mean gait velocity change for Open Shunt (0.23 m/s) is over twice the substantial meaningful change in the elderly and exceeds the “large” minimum clinically important difference of 0.22 m/s in Parkinsonism.38 Gait velocity, and change in gait velocity, are recognized markers of health and functional capacity in the elderly population. Low gait velocity is associated with a substantially higher risk of falling. Healthy elderly adults with gait velocity <0.7 m/s have >5 times the risk of falling than those with velocity >1.1 m/s.39 The combined effect of improved gait velocity and a lower rate of falls in the Open Shunt group suggests that shunt surgery may have a broad beneficial clinical impact on the health of elderly patients with iNPH. Importantly, the results of this study are consistent with many studies and clinical series over the last 25 years.8,9,12-14,40 The gait velocity improvement for Open Shunt at 3 months is similar to that in a prospective registry series of 193 patients at 4 months (0.25 m/s) and a pilot blinded, placebo-controlled study at 4 months (0.28 m/s).12,19

is study are consistent with many studies and clinical series over the last 25 years.8,9,12-14,40 The gait velocity improvement for Open Shunt at 3 months is similar to that in a prospective registry series of 193 patients at 4 months (0.25 m/s) and a pilot blinded, placebo-controlled study at 4 months (0.28 m/s).12,19 While complications in the first 3 months of iNPH treatment occurred in this study, the rates are similar to those in previous iNPH clinical trials and lower than those found in national surveys. 8,41,42 The most serious was a parenchymal hemorrhage at the time of surgery, which ultimately resulted in a death, and 3 SDH that required intervention. However, 4 SDH and 3 hygromas were treated with noninvasive shunt setting adjustment. Similarly, low pressure headaches were treated by shunt setting adjustment. The major strengths of the study include its blinded design, statistical power, rigorous training of clinical assessors, and quality control procedures. The generalizability of the findings is supported by the selection of patients for shunt surgery across multiple centers in accordance with the widely used International Guidelines. Because the primary outcome was gait velocity, patients who could not walk 10 meters or whose velocity was too fast were not eligible, and the study results may not be applicable for such patients. Patients requiring chronic anticoagulation were excluded to avoid bleeding complications, although prior studies suggest that careful management can mitigate these risks. 43,44

s who could not walk 10 meters or whose velocity was too fast were not eligible, and the study results may not be applicable for such patients. Patients requiring chronic anticoagulation were excluded to avoid bleeding complications, although prior studies suggest that careful management can mitigate these risks. 43,44 A potential limitation is the 3-month primary timepoint, which is standard for iNPH research. This study does not yet address the long-term outcomes. However, the outcome measures and complications reported here are being collected for all participants through 12 months, and will be supplemented by a comprehensive neuropsychological battery in approximately 2/3 of participants, and imaging and CSF biomarkers in all that may yield more insight into the extent, pace, and timing of outcomes, as well as the neuronal or glial biological mechanisms responsible for symptom development and recovery in iNPH. While using a specific shunt valve might raise concern about generalizability, the function of all differential pressure valve mechanisms is essentially the same and previous studies using different shunt valves found comparable improvements.8,12,14 There is no indication that the valve in this study has advantages or disadvantages compared to other valves with similar characteristics and adjustability.

The major strengths of the study include its blinded design, statistical power, rigorous training of clinical assessors, and quality control procedures. The generalizability of the findings is supported by the selection of patients for shunt surgery across multiple centers in accordance with the widely used International Guidelines. Because the primary outcome was gait velocity, patients who could not walk 10 meters or whose velocity was too fast were not eligible, and the study results may not be applicable for such patients. Patients requiring chronic anticoagulation were excluded to avoid bleeding complications, although prior studies suggest that careful management can mitigate these risks. 43,44 A potential limitation is the 3-month primary timepoint, which is standard for iNPH research. This study does not yet address the long-term outcomes. However, the outcome measures and complications reported here are being collected for all participants through 12 months, and will be supplemented by a comprehensive neuropsychological battery in approximately 2/3 of participants, and imaging and CSF biomarkers in all that may yield more insight into the extent, pace, and timing of outcomes, as well as the neuronal or glial biological mechanisms responsible for symptom development and recovery in iNPH.

pplemented by a comprehensive neuropsychological battery in approximately 2/3 of participants, and imaging and CSF biomarkers in all that may yield more insight into the extent, pace, and timing of outcomes, as well as the neuronal or glial biological mechanisms responsible for symptom development and recovery in iNPH. While using a specific shunt valve might raise concern about generalizability, the function of all differential pressure valve mechanisms is essentially the same and previous studies using different shunt valves found comparable improvements.8,12,14 There is no indication that the valve in this study has advantages or disadvantages compared to other valves with similar characteristics and adjustability.

The PENS Trial provides evidence that shunt surgery is effective for improving gait velocity at 3 months in patients with idiopathic normal pressure hydrocephalus selected for shunt surgery in accordance with the International iNPH Guidelines. The results of this study support the potential benefits of evaluating patients who have symptoms, neurological findings, and brain imaging consistent with iNPH and treating with shunt surgery when indicated.