Introduction

Patients with idiopathic normal pressure hydrocephalus (iNPH) often present with gait disturbance, dementia, and urinary incontinence, and have ventriculomegaly on CT and MR imaging [14]. A ventriculoperitoneal (VP) shunt can be inserted to relieve these symptoms [5, 13]. Infusion tests are performed to improve selection of patients who will benefit from shunt surgery [8, 9, 25]. Based on the infusion test, it is possible to estimate the absorption of cerebrospinal fluid (CSF) measured as CSF resistance to outflow (Rout), as deficit in the absorption of CSF is assumed to be part of the pathology of iNPH [7]. Because of a positive predictive value as high as 86% for Rout > 12 mmHg/ml/min regarding effect of VP shunt treatment [25], infusion tests are widely accepted in the diagnosis of iNPH [6, 16, 19]. There are different methods for performing infusion tests; the most reliable being outflow resistance obtained with steady state, e.g., constant rate infusion [24]. Infusion test can be performed both via lumbar puncture and via an intraventricular drain [4]. In contrast to the lumbar test, which is performed within minutes after needle insertion, an intraventricular infusion test can be performed sooner or later after drain insertion. This raises the question if timing of the infusion test influences the measurement.

The present study was conducted to investigate if infusion tests performed shortly after surgery provide the same Rout as tests performed after 24 h in the same patient.

Methods

The study was approved by the Danish Data Protection Agency (ID 2012-58-0004) and the National Committee on Health Research Ethics for the Capital Region of Denmark (ID H-15016032). Written informed consent was obtained from all patients included in the study.

Study population

We included 10 patients scheduled for an intraventricular infusion test at the Department of Neurosurgery, Rigshospitalet, between January 2016 and September 2017. All patients had hydrocephalus on CT/MRI, symptoms of normal pressure hydrocephalus (NPH), and were scheduled for intraventricular infusion test at our department as part of their clinical evaluation. Patients were not included if the hydrocephalus nurse was not available for both infusion tests. Our hydrocephalus service includes a clinical nurse specialist allocated specifically to patients with hydrocephalus and other CSF disorders.

Infusion test protocol

Practical aspects of infusion tests are center-specific with no published guideline for the timing of the infusion test after external ventricular drain (EVD) insertion. In our center, we have routinely used a 24-hour time gap between EVD placement and the infusion test due to a concern that intracranial pressure (ICP) is affected by surgery in the immediate postoperative period, and that CSF may egress along the drain until brain tissue glues to the drain yielding falsely low Rout values.

Infusion tests were performed with constant rate of 60 ml/hour intraventricular infusion of artificial cerebrospinal fluid (Ringer Lactate) through an EVD while measuring ICP with the patient in a supine position. The Raumedic Neurovent was inserted as EVD. The device has a tip transducer integrated into the drain and can be used both to measure ICP and perform CSF drainage. The first infusion test was performed 1 h after insertion of the EVD, and repeated with the same set-up 24 h after EVD insertion.

Baseline ICP was measured for 10 min before the infusion was started. When ICP reached a plateau (determined visually by the hydrocephalus nurse), infusion was stopped [10, 12]. After the test was completed, ICP monitoring was continued until it reached baseline level. Infusion was discontinued if the patient experienced headache or nausea, or if ICP increased above 40 mmHg.

ICP was monitored and analyzed with the software package ICM+ (Cambridge Enterprises Ltd, Cambridge, UK) [21].

The resistance to CSF outflow (Rout) was calculated by the formula:

$$ \mathrm{Rout}=\frac{\mathrm{plateau}\ \mathrm{ICP}\hbox{--} \mathrm{baseline}\ \mathrm{ICP}}{\mathrm{infusion}\ \mathrm{rate}} $$

All infusion tests were performed by a specialized hydrocephalus nurse (authors DC and SDJ), which ensured consistency in practical procedure routine.

Study variables

For all patients, we collected baseline data on age, sex, symptomology (gait disturbance, incontinence, and signs of dementia), and duration of symptoms. Results extracted from infusion tests were: Rout, plateau ICP, baseline ICP, duration of test, and discomfort during test. Baseline and plateau ICP was assessed through blinded evaluation of ICP curves (by consensus among authors THA, ALC, AVH, and MJ) (Fig. 1). We also noted treatment consequences of the infusion test as well as clinical status at 3-month follow-up. As the predictive value of Rout is not the focus of this investigation, clinical outcome after surgical treatment was assessed retrospectively by reviewing medical records for the 3-month follow-up with self-reported clinical status.

Fig. 1
figure 1

Example of an intracranial pressure (ICP) curve from infusion test (patient 4, 1 h after external ventricular drain insertion). ICP intracranial pressure, AMP pulse wave amplitude of ICP signal

Full size image

Statistics

The main study endpoint was the reproducibility of Rout and Rout values from infusion tests after 1 and 24 h and was compared using Wilcoxon signed rank test and a Bland-Altman plot. A P value < 0.05 was considered the statistical level of significance. The statistical software package “R” version 3.2.0 (The R Foundation for Statistical Computing, Vienna, Austria) was used for analyzing data and visualization of the collected data. Sample size calculation was performed for paired sample significance testing with power 0.9 and alpha 0.05, and the results from a previous publication by Tans et al. [24] reporting a standard deviation of 1.2 mmHg/ml/min between repeated measurements of Rout.

Results

Patients

Ten patients (6 men and 4 women) scheduled for an intraventricular infusion test at the Department of Neurosurgery at Copenhagen University Hospital, Rigshospitalet, were included in this study. Median and mean age was 75 years (range 64–81). All patients had signs of communicating hydrocephalus on head CT scan, and had symptoms consistent with NPH (Table 1). Eight of the ten patients had memory difficulties, all ten patients had gait and/or balance disturbance, and all ten patients had urinary incontinence. Of the 10 patients included in the investigation, 8 patients completed both infusion tests. Patients 4 and 5 were the two patients not completing the infusion test planned 24 h after EVD insertion (Table 1). One patient was excluded as an ICP plateau was not clearly visible during the infusion test performed 1 h after EVD placement. The final study population thus comprised 7 patients.

Table 1 Patient characteristics of the 10 patients suspected for normal pressure hydrocephalus (NPH)
Full size table

Infusion tests

Median Rout one hour after EVD insertion was 12.9 mmHg/ml/min (range 7.0–22.0), and after 24 h median Rout was 11.3 mmHg/ml/min (range 7.8–18.1). This difference in Rout was not statistically significant (P = 0.83). In 4 out of the 7 patients, the difference in Rout was < 1 mmHg/ml/min. The remaining three patients had the following difference in Rout values between the test after 1 and 24 h + 2.4, − 4.7, and + 5.2 mmHg/ml/min (Table 2). Differences in Rout between repeated infusion tests can be illustrated by a Bland-Altman plot (Fig. 2).

Table 2 Resistance to CSF outflow (Rout) 1 vs. 24 h after external ventricular drain (EVD) insertion (mmHg/ml/min)
Full size table
Fig. 2
figure 2

Bland-Altman plot for comparison of Rout from infusion tests performed 1 and 24 h after EVD insertion. The mean values of Rout are plotted on the x-axis. Differences between Rout values are plotted on the y-axis

Full size image

Median baseline ICP one hour after surgery was 11.0 mmHg (range 3.7–16.6), and 24 h after surgery it was 12.6 mmHg (range 4.2–29.8) mmHg (P = 0.70). Baseline ICP changed considerably between the two tests in three patients (patient 4, 5, and 7, see Table 3). In patient 7, this change in baseline ICP from 9.5 to 16.0 mmHg did not translate to a large difference in Rout values, since the plateau ICP changed in parallel from 18.7 to 24.7 mmHg. Patients 4 and 5 had higher baseline ICP during the test performed after 24 h, and subsequent infusion tests were not performed.

Table 3 Baseline intracranial pressure (ICP) 1 vs. 24 h after external ventricular drain (EVD) insertion (mmHg)
Full size table

The median plateau ICP 1 h after surgery was 25.7 mmHg (range 17.2–36.1), and 24 h after surgery the median plateau ICP was 23.5 mmHg (range 12.3–31.1, P = 0.81, Table 4).

Table 4 Plateau intracranial pressure (ICP) 1 vs. 24 h after external ventricular drain (EVD) insertion (mmHg)
Full size table

Clinical consequences of infusion tests

Using a cutoff at Rout > 12 mmHg/ml/min as the threshold for pathological Rout, five patients had increased Rout in both their infusion tests and were selected for surgical treatment. Four patients underwent shunt surgery and one patient with an infusion test indicating obstructive hydrocephalus and symptoms of NPH underwent ETV (endoscopic third ventriculostomy). The remaining five patients were not considered eligible for surgery due to Rout below the threshold.

Three patients reported clinical improvement at the 3-month follow-up visit after shunt surgery. One of these reported possible effect on memory but no effect on urinary incontinence or gait, and two reported effect on memory and cognition. The fourth patient with a VP-shunt had no clinical effect after shunt insertion and head CT indicated overdrainage through the shunt system. The patient had no symptoms of overdrainage, and a subsequent head CT showed regression of overdrainage. The patient who underwent ETV experienced short-term effect on gait, and the patient was seen again after 6 months and reported effect on all NPH symptoms (gait, memory, and urinary incontinence). One patient had a complication of ventriculitis and was treated with intrathecal antibiotics.

Discussion

The diagnosis of iNPH relies on clinical presentation, head CT/MRI scans, and additional invasive diagnostic tests, e.g., infusion tests and tap tests [12]. In the present investigation, we tested the hypothesis that Rout differs according to time after EVD insertion and found no significant difference in Rout calculated from ICP measurements during infusion tests performed 1 and 24 h after EVD insertion. For 4 out of 7 patients, the difference in Rout was < 1 mmHg/ml/min between infusion tests performed 1 and 24 h after EVD insertion. The remaining 3 patients had a difference in Rout of more than 2 mmHg/ml/min.

Reproducibility of Rout

We routinely do not perform intraventricular infusion tests earlier than 24 h after EVD insertion due to concerns that surgery might affect testing by introducing biological changes to the CSF system. Theoretically, a CSF spill during the EVD insertion, CSF leakage along the catheter, or ICP changes induced by recent surgery might affect testing. In our experience, the CSF spill is minimal, since the surgeon plugs the EVD as soon as the ventricular system has been reached. In lumbar infusion tests, a falsely high increase in pressure can be measured if there is blockage of tissue at the orifice of the needle [12]. Our study does not support these theoretical concerns. With a reliable result 1 h after EVD insertion, the infusion test can be performed earlier than we do in our current clinical practice. The EVD can then be explanted a few hours after surgery and the patient discharged on the same day. This will shorten length of stay for each patient and reduce infection risk. This study therefore offers a faster expedition of diagnosis, and presumably a shorter waiting period until shunt surgery, which improves the result of shunt surgery for patients with iNPH [3].

Other investigations of repeated infusion tests have also found a high reproducibility of Rout [1, 2, 22,23,24]. Studies are mainly on lumbar infusion tests or test both lumbar and intraventricular but does not perform subanalysis regarding location [23, 24]. Reproducibility of Rout was high in studies repeating infusion tests only 5 min apart [2] and up to 102 days between repeated tests [23]. Tans et al. [24] performed both intraventricular and lumbar infusion tests and performed repeated test in 10 patients. They found a mean difference in Rout of 1.7 ± 1.2 mmHg/ml/min, but the location (intraventricular or lumbar) for the 10 patients and the time gap between repeated infusion tests were not provided. Juniewich et al. [15] support the notion that the infusion test introduces changes to the CSF system. They measured resting ICP before and after infusion, performing both lumbar and intraventricular constant rate infusions tests, and found the differences between post and pre infusion resting ICP (∆ICP) higher than 1 mmHg in 20 infusion tests out of 27. This was considered significant. The mean ∆ICP for the 20 infusion tests was 3.0 ± 0.7 mmHg. However, decreasing ICP was only recorded for about 10 min after infusion and not necessarily until ICP returned to resting ICP. In our study, 3 patients exhibited an increase in baseline ICP before the second infusion test, and 2 exhibited decrease in baseline ICP, but the difference in baseline ICP for the entire group was not statistically significant. We also found no significant difference for plateau ICP between infusions tests. We can however not exclude the possibility that the infusion tests performed 24 h after EVD insertion are affected by the previous infusion test. Therefore, we assessed the agreement between replicate measurements by a Bland-Altman plot (Fig. 2). For patients 1 and 2, the agreement between paired infusion tests was poor, though the plot illustrates good agreement for most of the patients. Our data thus support that the infusion test does not need to be delayed beyond 1 h after drain insertion.

Treatment and pathological Rout

The diagnosis of iNPH can be a difficult, making infusion tests an important tool in selecting patients who would benefit from shunt surgery. Malm et al. [18] performed lumbar infusion tests in normal individuals aged 60–82 years, and found a mean Rout of 11 mmHg/ml/min. In a study with intraventricular infusion tests, a Rout below 10 mmHg min/ml was considered physiological, Rout 10–13 mmHg min/ml borderline and Rout above 13 mmHg/min/ml pathological [20].

Eide et al. [11] performed constant rate lumbar infusion tests while measuring ICP in both the lumbar cerebrospinal space and in the brain parenchyma. They reported a higher lumbar Rout than ventricular Rout with a mean difference of 2.7 mmHg/ml/min. This might be explained by differences in compliance of the two compartments. However, it suggests that differentiated thresholds for abnormal Rout should be adapted for intraventricular and lumbar infusion tests.

There is no complete consensus regarding the treatment threshold of Rout. Previous publications have concluded that shunt surgery based on Rout values above 18 mmHg/ml/min improves outcome [6, 7]. However, other investigators recommend a lower threshold at 12 or 13 mmHg/ml/min [11, 15, 17, 20]. Setting the limit at 18 mmHg/ml/min yields a lower sensitivity, possibly excluding patients who would benefit from a shunt [17]. The infusion tests for patients 1 and 2 had a difference in Rout of + 5.7 and − 4.7 mmHg/ml/min, respectively. This difference seems large enough to potentially change the clinical consequence drawn from the tests.

We performed power analysis to detect a difference of 2 mmHg/ml/min, as we believe a significant difference should be of at least 2–3 mmHg/ml/min. We have not chosen a specific value for a significant difference since a difference in Rout matters more when in the area of 15–18 mmHg/ml/min where a pathological Rout is unsure. A difference in Rout above 18 mmHg/ml/min or below 12–15 mmHg/ml/min is less essential since the clinical conclusion on pathological or not does change. For the patients with a large difference in Rout , it can certainly be debated that though the statistical value is not influenced, the clinical value might be. Swallow et al. [23] commented that the parameters estimated with infusion testing have to be considered as a range in which the actual value of the patient lies rather than as a fixed value. We concur with this statement, since analysis of the ICP curve partly depends on clinical assessment, and the plateau is not always easily identified. The risk of finding different results seems to be related to technical variability and inter-observer differences in analyzing the ICP curves.

Limitations

It was not possible to repeat infusion tests in all patients. Two patients had infusion tests with a pattern of obstructive hydrocephalus with a steep increase in ICP, and no ICP plateau. This study was intended for patients with normal pressure hydrocephalus only.

We believe that inter-observer variation was an important limitation in the analysis of ICP curves.

Conclusion

We compared Rout estimated through infusion tests performed 1 and 24 h after EVD insertion. For two of the seven patients, there was poor agreement between paired infusion tests, though overall there was no significant difference in Rout between paired tests. We therefore propose that infusion tests can be performed shortly after EVD insertion, though further studies seem needed to confirm this.