ABSTRACT
Background
Lumbar puncture (LP) is essential for detecting intracranial hypertension (IH) but is often non‐diagnostic and not without complications. A skull extensometer‐based non‐invasive intracranial pressure waveform assessment (NIWA) has shown promise in detecting IH in neurocritical settings. This study investigated its feasibility and accuracy among outpatient neurological individuals.
Methods
This cross‐sectional study enrolled adults referred to routine outpatient LP at a tertiary‐care hospital. Subjects underwent baseline supine NIWA, then were sequentially divided (2:1:1 ratio) to: LP then follow‐up supine NIWA; follow‐up NIWA while sitting then LP; or follow‐up NIWA during Valsalva maneuver then LP. Several waveform parameters and their variation between baseline and follow‐up assessments were compared with LP results. IH was defined as LP opening pressure > 25cmH2O.
Results
The study included 130 subjects, and 28 (21.5%) presented with IH. Dicrotic peak slope < 0.143 predicted IH with 89.5% sensitivity and 29.7% specificity (AUROC = 0.680, 95% CI 0.541–0.819). Equations derived from multivariate logistic regression models including mostly latency‐related parameters of the entire waveform and its percussion and dicrotic peaks predicted cerebrospinal fluid (CSF) chemocytology changes with AUROC = 0.661 (95% CI 0.542–0.780) and overall abnormal LP results (IH and/or altered CSF chemocytology) with 96.2% sensitivity and 24.1% specificity (AUROC = 0.696, 95% CI 0.572–0.820). Parameter changes after LP, during Valsalva maneuver, and sitting position did not improve diagnostic accuracy.
Conclusions
NIWA was feasible and detected IH and CSF changes with high sensitivity among real‐world neurological outpatients. While LP remains pivotal to many cases, pre‐LP screening with NIWA may enhance diagnostic yield and mitigate complications and healthcare‐related costs.
Keywords: cerebrospinal fluid, headache, intracranial hypertension, intracranial pressure, spinal puncture
In this cross‐sectional study, the morphology of the cardiac‐gated intracranial pressure waveform was assessed with a skull extensometer‐based non‐invasive method in 130 individuals referred to routine lumbar puncture (LP). Waveform parameters recorded in supine position were compared with LP results and were found to detect intracranial hypertension and/or chemocytological changes with 96.2% sensitivity. Waveform parameters were also sensitive to intracranial pressure variations induced by Valsalva maneuver, CSF extraction, and sitting position, but these changes did not improve the method's accuracy.
1. Introduction
Lumbar puncture (LP) is pivotal to the evaluation and monitorization of chronic intracranial hypertension (IH) in neurological outpatient practice [1, 2]. Nonetheless, it may be non‐diagnostic in a relevant proportion of patients. It is estimated that 40%–50% of those suspected of chronic hydrocephalus [3] and idiopathic IH [4] undergoing LP present normal results. Moreover, despite serious complications being rare, LP may result in post‐dural puncture headache in up to 40% [5, 6]. Although usually self‐remitting, this complication often requires bed rest, and sometimes hospitalization and blood patch treatment, amounting to over $17,000 per case when managed conservatively [5, 7]. Therefore, even though LP is a central diagnostic tool for many clinical conditions, strategies for pre‐LP screening, especially in cases of IH suspicion of low and medium pretest probabilities, have gained attention [4].
Meanwhile, there has been growing interest in the development of non‐invasive methods for estimating intracranial pressure (ICP) in neurocritical care settings [8]. Their implementation in daily practice has, however, been undermined by several factors, including low portability, high costs, limited accuracy, and unknown cutoff values [9]. Attempting to overcome these limitations, a new portable, low‐cost, and non‐invasive approach has been developed. By using an extensometer to measure microscale skull diameter variations over time [10], it was possible to accurately estimate ICP waveform morphology, both in animal and human models [11, 12]. In healthy conditions, these waveforms typically present with three peaks, which are named from first to last as: percussion, tidal, and dicrotic [13, 14]. When ICP is normal, the percussion peak (P1) is larger than the tidal (P2) and dicrotic peaks (P3) [14]. However, in cases of acute severe IH, this morphology changes, with P2 > P1 [13, 14].
This non‐invasive ICP waveform assessment (NIWA) technique has shown positive initial results among acute brain injury individuals when compared to traditional invasive ICP measurements, being able to detect IH with sensitivities ranging from 60% to 100% and specificities of 60% to 77% [15, 16, 17]. However, its diagnostic performance in the outpatient setting is yet to be characterized.
2. Methods
2.1. Study Design
This cross‐sectional exploratory study aimed to explore the accuracy and feasibility of a skull extensometer‐based NIWA approach in detecting IH in a real‐world outpatient neurological population. Participants were enrolled consecutively among individuals referenced to routine outpatient LP for investigation of different neurological conditions. Data collection took place at the CSF laboratory from the University of Sao Paulo Clinics Hospital in Sao Paulo, Brazil. All individuals underwent a structured interview and a brief physical examination, followed by the baseline NIWA in the supine position at 0°. They were then divided into three groups (Figure 1):
Group post‐LP (L): routine LP was performed, and then a follow‐up NIWA in the horizontal supine position was repeated.
Group Sitting (S): a follow‐up NIWA with the participant sitting on a chair with a 90° backrest was performed, followed by LP
Group Valsalva (V): a follow‐up NIWA during controlled Valsalva maneuver in the supine position at 0° was performed, followed by LP. Participants performed the Valsalva maneuver by blowing into a 10 mL syringe barrel connected to a standard manometer, and were requested to maintain its pointer between 20 and 30 mmHg for at least 20 s (Supporting Information p.4 and Figure S1).
FIGURE 1.
Study design. NIWA, non‐invasive intracranial pressure waveform assessment.
Group allocation was sequential with a 2:1:1 ratio, starting with group L until its planned sample size was completed, then moving consecutively to S and V. The primary analysis consisted of comparing baseline NIWA data with LP results for the whole sample. Additionally, it was explored whether changes induced by ICP‐lowering (groups L and S) or ‐raising (Group V) maneuvers could improve diagnostic performance by comparing the variations of waveform parameters between baseline and follow‐up NIWA with LP results, within each group. This research protocol has been approved by the Ethics Committee of the Clinics Hospital of the University of Sao Paulo Medical School (#3.408.380) and complies with the declaration of Helsinki. All participants have provided written informed consent before inclusion.
2.2. Eligibility Criteria
Subjects aged ≥ 18 years, able to provide informed consent and to comply with the study procedures were included. Exclusion criteria were: relevant skull bone defects which may interfere with the NIWA; in whom the opening pressure measurement was technically not feasible; and with contraindications to Valsalva maneuver (specifically for Group V) or to the LP (Supporting Information Data S1, p. 3).
2.3. Lumbar Puncture and Cerebrospinal Fluid Analyses
LP was performed in lateral horizontal decubitus, using a 22‐gauge Quincke tip spinal needle, placed at the L5‐S1 or L4‐L5 intervertebral space. The opening pressure was measured with an aneroid manometer. CSF volume was determined individually, according to clinically indicated laboratory analyses, to minimize interference in routine care protocols. The obtained CSF sample was analyzed for general chemocytological data, including cell count and protein levels.
2.4. Non‐Invasive Intracranial Pressure Assessment and Data Processing
NIWA was performed using the BcSs‐PICNI‐2000 sensor (Brain4care Corp., São Carlos, Brazil), according to the user's guide instructions [18]. The sensor was positioned on the left temporal region, 5–6 cm above the external auditory meatus, in close contact with the scalp with an adjustable tape; and the participants were instructed to remain calm and minimize voluntary movements. The raw waveform signal was recorded for 5 min and then processed to create an average waveform pulse (Figure 2) depicting the overall behavior of individual recorded waveforms. Individual pulses beyond 2 standard deviations from the average were considered outliers and excluded from this analysis. The average waveform pulse obtained from the NIWA during the Valsalva maneuver (Group V) was extracted from the raw signal recorded only throughout this maneuver.
FIGURE 2.
Average intracranial pressure waveform pulse obtained by a 5‐min monitoring session with a non‐invasive cranial extensometer sensor. Illustrates the final average waveform pulse resulting from the data processing of the raw individual waveform pulses recorded during a 5‐min monitoring session with the studied non‐invasive approach. The blue line is the mean morphology of the waveform pulse during the session. The following investigated parameters are also indicated in this figure: Amplitude of the percussion wave (P1), amplitude of the tidal wave (P2), amplitude of the dicrotic wave (P3), latency between the beginning of the waveform and P1 (L1), P2 (L2), and P3 (L3), and total latency from the beginning to the end of the average waveform (TL). The light blue region under the waveform represents its area.
2.5. Data Extraction and Waveform Parameters
The average waveform pulse was visually examined by a researcher, blinded to LP results (Figure 2). The amplitudes, latencies, and slopes of the percussion (P1, L1 and S1), tidal (P2, L2 and S2) and dicrotic peaks (P3, L3 and S3) were measured. Latencies were defined as the time spanning from the lowest point before the beginning of the waveform until the highest peak amplitude. The slope was characterized by the ratio between peak amplitude and latency. Additionally, the total latency of the waveform (TL), the total area under the waveform (A), and latency to its maximum amplitude (i.e., latency to peak, LTP) were recorded. Finally, the following were calculated: ratio between P2 and P1 (P2/P1), and between A and TL (A/TL).
2.6. Sample Size Estimation
Few previous studies have investigated the accuracy of this NIWA approach compared to standard invasive techniques and, to the best of our knowledge, none have done so among adult outpatients. By the time this study was designed, the largest of these included 72 participants with neurocritical care conditions, 13 (18%) of whom were suffering from IH; and found positive associations between invasive ICP measurements and NIWA parameters [15]. Therefore, this exploratory study aimed to reach a convenience sample of 130 participants. This was grounded on the results from this previous study and a supposition that one in four subjects referred for LP would present with IH, based on the CSF laboratory's usual patient profile.
2.7. Statistical Analyses
Participants who successfully completed both LP opening pressure measurement and baseline NIWA constituted the primary analyses set. Firstly, waveform parameters were compared between‐subjects according to LP opening pressure (IH vs. normal), CSF chemocytology (normal vs. abnormal), and overall LP results (normal vs. abnormal). IH was defined as a LP opening pressure > 25cmH2O [19]; abnormal CSF chemocytology as a cell count > 4/mL and/or a protein level > 40 mg/dL; and abnormal overall LP test as IH and/or an abnormal CSF chemocytology. Categorical variables were compared with chi‐square test, or Fisher's exact test when appropriate. Parametric continuous variables were analyzed with Student's t‐test; and the non‐parametric ones with the Mann–Whitney U test. The accuracies of significantly different waveform parameters in discriminating between groups were estimated with the area under the Receiver Operating Curve (AUROC), and classified as poor (AUROC 0.5–0.69), acceptable (AUROC 0.7–0.79), excellent (AUROC 0.8–0.89) or outstanding (AUROC ≥ 0.9) [20]. All parameters were also assessed through univariate logistic regression, and those providing significant models were then used to build multivariate logistic ones. Variables presenting with multicollinearity (defined as a tolerance < 0.1) were excluded from the latter. Scoring equations were produced based on regression coefficients of significant multivariate models.
Then, waveform parameters were compared within‐subjects before and after ICP‐lowering (post‐LP or sitting position) and raising (controlled Valsalva maneuver) procedures. For this, a secondary dataset was used, comprised of participants who successfully completed both baseline and follow‐up NIWA. In Group V, only those whose Valsalva maneuver lasted > 20 s were included. Categorical variables were compared with the McNemar's test, parametric continuous variables with the paired‐samples t‐test, and non‐parametric ones with either the Wilcoxon signed‐rank test or the sign test when appropriate. Additionally, the percentage changes between the baseline and follow‐up NIWA were calculated for each parameter. These changes were also compared between‐subjects according to LP results, as described above.
Sensitivity analyses were performed investigating the association between waveform parameters and IH within the subsample without chemocytological changes; and the association between these parameters and chemocytological findings among subjects without IH. Categorical variables were expressed as absolute frequencies and percentages, while continuous ones were reported as mean ± standard deviation. Statistical significance was set at p < 0.05. All analyses were conducted with the Statistical Package for the Social Sciences version 20.0.0 (SPSS Inc., Chicago, IL, USA).
3. Results
3.1. Participants and Lumbar Puncture Results
The primary analyses set was composed by 130 individuals (Figure 3) with a mean age of 47.3 ± 14.8 years‐old, and 67 (51.5%) of whom were females (Tables 1 and S1). The most frequent diagnostic hypotheses for LP referral were central nervous system infections (n = 35, 26.9%), inflammatory central nervous system diseases (n = 19, 14.6%), and idiopathic IH (n = 14, 10.8%), while other causes comprised 47.7% (n = 62, Table S1). Mean opening pressure was 20.0 ± 6.9cmH2O, and IH was found in 28 (21.5%) individuals. An abnormal CSF chemocytology was observed in 66 (50.8%), and an abnormal overall test in 83 (63.8%).
FIGURE 3.
Data collection flowchart. †The ICP‐raising maneuver was the controlled Valsalva (Group V), while the ICP‐lowering ones were sitting position (Group S) and LP (Group L). ‡Adequate waveform signal readings were obtained for 2 subjects, but Valsalva duration was shorter than 20 s, therefore these were considered inadequate and excluded. NIWA, non‐invasive intracranial pressure waveform assessment; ICP, intracranial pressure.
TABLE 1.
General demographics and cerebrospinal fluid data from study participants.
| Total sample (N = 130) | Group L (N = 71) | Group S (N = 31) | Group V (N = 28) | |
|---|---|---|---|---|
| Female | 67 (51.5) | 37 (52.1) | 13 (41.9) | 17 (60.7) |
| Age (years) a | 47.3 ± 14.8 (18–82) | 47.6 ± 15.7 (18–82) | 47.4 ± 14.4 (25–76) | 47.2 ± 13.0 (19–75) |
| Ethnicity | ||||
| White | 80 (61.5) | 43 (60.6) | 24 (77.4) | 13 (46.4) |
| Afro‐descendant | 34 (26.2) | 17 (23.9) | 5 (16.1) | 12 (42.9) |
| Other | 16 (12.3) | 11 (15.5) | 2 (6.5) | 3 (10.7) |
| BMI (kg/m2) a | 26.5 ± 4.8 (16.6–43.6) | 26.5 ± 5.0 (17.1–43.6) | 26.6 ± 5.2 (17.2–38.9) | 26.4 ± 3.8 (16.6–33.1) |
| Previous intracranial surgery | 8 (6.2) | 4 (5.6) | 2 (6.5) | 2 (7.1) |
| Previous spinal surgery | 3 (2.3) | 1 (1.4) | 2 (6.5) | 0 (0) |
| Ventriculoperitoneal shunt | 5 (3.8) | 3 (4.2) | 1 (3.2) | 1 (3.6) |
| Lumboperitoneal shunt | 1 (0.8) | 0 (0) | 1 (3.2) | 0 (0) |
| CSF opening pressure (cmH2O) a | 20.0 ± 6.9 (3–42) | 20.7 ± 6.9 (5–42) | 17.1 ± 6.8 (3–40) | 21.6 ± 6.5 (12–36) |
| CSF opening pressure > 25cmH2O a | 28 (21.5) | 15 (21.1) | 4 (12.9) | 9 (32.1) |
| Abnormal CSF analysis b | 66 (50.8) | 35 (49.3) | 18 (58.1) | 13 (46.4) |
| CSF cell count per mL a | 5.3 ± 8.8 (0–49) | 4.8 ± 8.6 (0–49) | 6.4 ± 8.7 (1–33) | 5.0 ± 9.3 (0–41) |
| CSF cell count > 4/mL | 30 (23.1) | 15 (21.1) | 9 (29) | 6 (21.4) |
| CSF protein (mg/dL) a | 42.9 ± 20.6 (20–139) | 44.1 ± 23.1 (20–139) | 47.0 ± 20.2 (22–108) | 35.4 ± 11.4 (21–63) |
| CSF protein > 40 mg/dL | 54 (41.5) | 29 (40.8) | 17 (54.8) | 8 (28.6) |
| CSF glucose (mg/dL) a | 63.4 ± 18.5 (38–173) | 62.2 ± 17.0 (38–163) | 64.1 ± 13.9 (40–97) | 65.6 ± 25.6 (46–173) |
| CSF volume (mL) a | 10.0 ± 11.58 (3–70) | 11.6 ± 14.2 (5–70) | 8.5 ± 7.2 (3–45) | 7.46 ± 6.5 (5–40) |
| Overall abnormal CSF test c | 83 (63.8) | 45 (63.4) | 20 (64.5) | 18 (64.3) |
Note: Data presented as n (%), unless otherwise specified.
Abbreviation: CSF, cerebral spinal fluid.
Data presented as mean ± standard deviation (minimum‐maximum).
Defined as a cell count > 4/mL and/or a protein level > 40 mg/dL.
Defined as a lumbar puncture opening pressure > 25 cmH2O and/or an abnormal cerebrospinal fluid chemocytology.
3.2. Relationship Between Non‐Invasive Waveform Assessment and Lumbar Puncture Results
3.2.1. Lumbar Puncture Opening Pressure
Individuals with IH presented with overall smaller percussion and dicrotic peak amplitudes and slopes, when compared to those without, but with similar P2/P1 ratio (p = 0.621). This was characterized by lower mean P1 (0.022 ± 0.021 mV vs. 0.031 ± 0.027 mV, p = 0.021), P3 (0.015 ± 0.014 mV vs. 0.021 ± 0.128 mV, p = 0.038), S1 (0.095 ± 0.090 mV vs. 0.136 ± 0.108 mV, p = 0.020) and S3 (0.030 ± 0.022 mV vs. 0.046 ± 0.029 mV, p = 0.018). The area‐related parameters were also smaller: A (0.011 ± 0.010 vs. 0.016 ± 0.015, p = 0.037), and A/TL (0.013 ± 0.012 vs. 0.018 ± 0.014, p = 0.028). Nonetheless, the accuracy of these parameters was poor (Table 2). When subjects with abnormal chemocytology were excluded, just the amplitude parameters P1 (p = 0.030) and P3 (p = 0.038), and the area parameters A (p = 0.025) and A/TL (p = 0.026) remained significant, and their accuracy became generally acceptable (Table S2). Univariate logistic regression analyses with waveform parameters found that only S3 was able to significantly predict IH (p = 0.043, Table S3), with values lower than 0.143 presenting with 89.5% sensitivity and 29.7% specificity (AUROC = 0.680, 95% CI 0.541–0.819).
TABLE 2.
Waveform parameters of study subjects according to the presence of intracranial hypertension.
| No ICH (n = 102) | ICH (n = 28) | p | AUROC | 95% CI | |
|---|---|---|---|---|---|
| Percussion wave present | 102 (100) | 28 (100) | — | — | — |
| P1 | 0.031 ± 0.027 (0.004–0.142) | 0.022 ± 0.021 (0.003–0.097) | 0.021* | 0.674 | 0.526–0.821 |
| L1 | 0.243 ± 0.083 (0.125–0.565) | 0.239 ± 0.734 (0.140–0.400) | 0.849 | — | — |
| S1 | 0.136 ± 0.108 (0.013–0.606) | 0.095 ± 0.090 (0.012–0.464) | 0.020* | 0.680 | 0.534–0.826 |
| minP1P2 present | 89 (87.3) | 27 (96.4) | 0.300 | — | — |
| minP1P2 | 0.029 ± 0.028 (0.001–0.0147) | 0.020 ± 0.021 (0.003–0.103) | 0.695 | — | — |
| Tidal wave present | 89 (87.3) | 27 (96.4) | 0.300 | — | — |
| P2 | 0.025 ± 0.017 (0.002–0.093) | 0.020 ± 0.016 (0.003–0.067) | 0.079 | — | — |
| L2 | 0.369 ± 0.092 (0.215–0.780) | 0.369 ± 0.083 (0.185–00.515) | 0.192 | — | — |
| S2 | 0.073 ± 0.049 (0.010–0.316) | 0.053 ± 0.039 (0.008–0.168) | 0.051 | — | — |
| Dicrotic present | 66 (64.7) | 19 (67.9) | 0.756 | — | — |
| P3 | 0.021 ± 0.128 (0.001–0.065) | 0.015 ± 0.014 (0.003–0.062) | 0.038* | 0.657 | 0.541–0.819 |
| L3 | 0.473 ± 0.097 (0.300–0.800) | 0.484 ± 0.096 (0.290–0.670) | 0.672 | — | — |
| S3 | 0.046 ± 0.029 (0.004–0.173) | 0.030 ± 0.022 (0.007–0.094) | 0.018* | 0.680 | 0.541–0.819 |
| P2/P1 | 0.964 ± 0.267 (0.319–2.068) | 1.03 ± 0.335 (0.582–1.927) | 0.621 | — | — |
| LTP | 0.257 ± 0.117 (0.044–0.645) | 0.292 ± 0.117 (0.059–0.606) | 0.386 | — | — |
| Waveform area | 0.016 ± 0.015 (0.001–0.115) | 0.011 ± 0.010 (0.001–0.044) | 0.037* | 0.655 | 0.509–0.802 |
| TL | 0.848 ± 0.177 (0.515–1.805) | 0.837 ± 0.130 (0.645–1.120) | 0.753 | — | — |
| Area/TL | 0.018 ± 0.014 (0.001–0.065) | 0.013 ± 0.012 (0.002–0.050) | 0.028* | 0.660 | 0.512–0.809 |
Note: Data presented as mean ± standard deviation (minimum‐maximum).
Abbreviations: 95% CI, 95% confidence interval; AUROC, area under the Receiver Operating Characteristic curve; ICH, intracranial hypertension, defined as a lumbar puncture opening pressure ≥ 25cmH2O; L1, latency to P1; L2, latency to P2; L3, latency to P3; LTP, latency to peak; minP1P2, amplitude of the nadir between P1 and P2; P1, Amplitude of the percussion wave; P2, Amplitude of the tidal wave; P2/P1, ratio between the amplitudes of P2 and P1; P3, Amplitude of the dicrotic wave; S1, slope of P1; S2, slope of P2; S3, slope of P3; TL, total latency of the waveform.
p < 0.05.
3.2.2. Cerebrospinal Fluid Chemocytology
When compared to participants with normal CSF chemocytology, those with chemocytological changes presented with waveforms of overall shorter latencies. This was characterized by shorter mean L1 (0.221 ± 0.070 s vs. 0.265 ± 0.088 s, p = 0.002), L3 (0.451 ± 0.082 s vs. 0.498 ± 0.107 s, p = 0.031), and TL (0.806 ± 0.133 s vs. 0.887 ± 0.194 s, p = 0.008); and a consequently larger S1 (0.125 ± 0.090 s vs. 0.092 ± 0.064 s, p = 0.008). P2/P1 was not associated with this outcome (p = 0.900). These parameters had an overall poor accuracy, and S1 did not present significant discrimination capacity (Table 3). When excluding individuals with IH, waveform latency parameters L1 (p < 0.001), L3 (p = 0.005) and TL (p = 0.004) remained significantly shorter. This also improved AUROC values from 0.644–0.656 (Table 3) to 0.669–0.749 (Table S4), elevating diagnostic performance to the acceptable discrimination range.
TABLE 3.
Waveform parameters of study subjects according to cerebrospinal fluid analysis results.
| Normal CSF chemocitology | Abnormal CSF chemocitology a (n = 66) | p | AUROC | 95% CI | |
|---|---|---|---|---|---|
| Percussion wave present | 61 (100) | 66 (100) | — | — | — |
| P1 | 0.020 ± 0.014 (0.004–0.063) | 0.023 ± 0.016 (0.003–0.097) | 0.296 | — | — |
| L1 | 0.265 ± 0.088 (0.140–0.565) | 0.221 ± 0.070 (0.125–0.390) | 0.002* | 0.654 | 0.535–0.773 |
| S1 | 0.092 ± 0.064 (0.013–0.301) | 0.125 ± 0.090 (0.012–0.557) | 0.008* | 0.538 | 0.410–0.666 |
| minP1P2 present | 53 (86.9) | 60 (90.9) | 0.469 | — | — |
| minP1P2 | 0.173 ± 0.131 (0.003–0.550) | 0.131 ± 0.095 (0.003–0.285) | 0.074 | — | — |
| Tidal wave present | 53 (86.9) | 60 (90.9) | 0.469 | — | — |
| P2 | 0.020 ± 0.013 (0.002–0.055) | 0.025 ± 0.017 (0.004–0.093) | 0.277 | — | — |
| L2 | 0.329 ± 0.083 (0.185–0.520) | 0.323 ± 0.056 (0.225–0.445) | 0.117 | — | — |
| S2 | 0.060 ± 0.041 (0.008–0.188) | 0.078 ± 0.054 (0.010–0.316) | 0.116 | — | — |
| Dicrotic present | 41 (37.2) | 41 (62.1) | 0.549 | — | — |
| P3 | 0.018 ± 0.013 (0.002–0.056) | 0.021 ± 0.013 (0.003–0.065) | 0.277 | — | — |
| L3 | 0.498 ± 0.107 (0.290–0.800) | 0.451 ± 0.082 (0.300–0.670) | 0.031* | 0.644 | 0.521–0.767 |
| S3 | 0.039 ± 0.032 (0.004–0.173) | 0.046 ± 0.025 (0.007–0.129) | 0.080 | — | — |
| P2/P1 | 1.016 ± 0.300 (0.533–2.068) | 1.090 ± 0.262 (0.584–1.738) | 0.900 | — | — |
| LTP | 0.282 ± 0.132 (0.045–0.645) | 0.256 ± 0.110 (0.045–0.420) | 0.172 | — | — |
| Waveform area | 0.012 ± 0.009 (0.001–0.034) | 0.013 ± 0.009 (0.002–0.044) | 0.530 | — | — |
| TL | 0.887 ± 0.194 (1.805–1.280) | 0.806 ± 0.133 (0.515–1.125) | 0.008* | 0.656 | 0.535–0.776 |
| Area/TL | 0.013 ± 0.009 (0.001–0.038) | 0.015 ± 0.010 (0.002–0.062) | 0.226 | — | — |
Note: Data presented as mean ± standard deviation (minimum‐maximum); data available for 127 subjects.
Abbreviations: 95% CI, 95% confidence interval; AUROC, area under the Receiver Operating Characteristic curve; CSF, cerebrospinal fluid; ICH, intracranial hypertension, defined as a lumbar puncture opening pressure ≥ 25cmH2O; L1, latency to P1; L2, latency to P2; L3, latency to P3; LTP, latency to peak; minP1P2, amplitude of the nadir between P1 and P2; P1, Amplitude of the percussion wave; P2, Amplitude of the tidal wave; P2/P1, ratio between the amplitudes of P2 and P1; P3, Amplitude of the dicrotic wave; S1, slope of P1; S2, slope of P2; S3, slope of P3; TL, total latency of the waveform.
Defined as: CSF cell count > 4/mm3, CSF protein > 40 mg/dL.
p < 0.05.
Univariate regression analyses revealed that L1 (p = 0.004), S1 (p = 0.032), minP1P2 (p = 0.046), L3 (p = 0.036), LTP (p = 0.037), and TL (p = 0.010) were able to significantly predict an abnormal CSF chemocytology. A multivariate model including all these variables was built and significantly explained 16.4% to 22.5% of the probability of an altered chemocytology (p = 0.024), with 88.5% sensitivity and 44.8% specificity (Table S5). Based on the respective B of included variables, the following scoring equation was created:
This equation was able to detect abnormal chemocytology with an AUROC of 0.661 (95% CI 0.542–0.780), and scores higher than −5.0971 presented 95% sensitivity and 37.5% specificity (Figure S2).
3.2.3. Overall Lumbar Puncture Results
When compared to individuals with normal results, those with an overall abnormal LP also presented with significantly shorter mean waveform latencies: L1 (0.225 ± 0.073 s vs. 0.273 ± 0.089 s, p = 0.001), L3 (0.457 ± 0.087 s vs. 0.511 ± 0.107 s, p = 0.017) and TL (0.814 ± 0.135 s vs. 0.902 ± 0.208 s, p = 0.006). P2/P1 However, these parameters presented with poor accuracy, and TL had no significant discrimination capacity. P2/P1 did not differ significantly between these groups (p = 0.482, Table 4).
TABLE 4.
Non‐invasive intracranial pressure monitoring data of all subjects according to results of cerebrospinal fluid analysis and opening pressure combined.
| Normal lumbar puncture test (n = 47) | Abnormal lumbar puncture test a (n = 81) | p | AUROC | 95% Cl | |
|---|---|---|---|---|---|
| Percussion wave present | 47 (100) | 81 (100) | — | — | — |
| P1 | 0.030 ± 0.026 (0.004–0.142) | 0.029 ± 0.026 (0.003–0.131) | 0.716 | — | — |
| L1 | 0.273 ± 0.089 (0.140–0.565) | 0.225 ± 0.073 (0.125–0.400) | 0.001* | 0.650 | 0.525–0.775 |
| S1 | 0.117 ± 0.091 (0.013–0.437) | 0.133 ± 0.115 (0.012–0.606) | 0.469 | — | — |
| minP1P2 presence | 40 (85.1) | 74 (91.4) | 0.275 | — | — |
| minP1P2 | 0.158 ± 0.123 (0.009–0.480) | 0.145 ± 0.112 (0.004–0.550) | 0.456 | — | — |
| Tidal wave present | 40 (85.1) | 74 (91.4) | 0.275 | — | — |
| P2 | 0;023 ± 0.015 (0.002–0.075) | 0.024 ± 0.018 (0.003–0.093) | 0.991 | — | — |
| L2 | 0.378 ± 0.083 (0.225–0.550) | 0.366 ± 0.093 (0.185–0.780) | 0.413 | — | — |
| S2 | 0.065 ± 0.043 (0.009–0.188) | 0.069 ± 0.053 (0.007–0.316) | 0.835 | — | — |
| Dicrotic present | 32 (68.1) | 51 (63.0) | 0.559 | — | — |
| P3 | 0.020 ± 0.013 (0.002–0.056) | 0.019 ± 0.013 (0.003–0.065) | 0.861 | — | — |
| L3 | 0.511 ± 0.107 (0.320–0.800) | 0.457 ± 0.087 (0.290–0.670) | 0.017* | 0.629 | 0.501–0.756 |
| S3 | 0.042 ± 0.034 (0.004–0.173) | 0.042 ± 0.026 (0.007–0.129) | 0.778 | — | — |
| P2/P1 | 0.932 ± 0.157 (0.533–1.280) | 1.003 ± 0.335 (0.320–2.068) | 0.482 | — | — |
| LTP | 0.272 ± 0.115 (0.045–0.645) | 0.261 ± 0.112 (0.045–0.607) | 0.782 | — | — |
| Waveform area | 0.016 ± 0.017 (0.001–0.116) | 0.014 ± 0.012 (0.002–0.068) | 0.522 | — | — |
| TL | 0.902 ± 0.208 (0.525–1.805) | 0.814 ± 0.135 (0.515–1.125) | 0.006* | 0.591 | 0.463–0.720 |
| Area/TL | 0.017 ± 0.013 (0.001–0.064) | 0.017 ± 0.014 (0.002–0.065) | 0.814 | — | — |
Note: Data presented as mean ± standard deviation (minimum‐maximum); Data available for 128 subjects.
Abbreviations: 95% CI, 95% confidence interval; AUROC, area under the Receiver Operating Characteristic curve; ICH, intracranial hypertension, defined as a lumbar puncture opening pressure ≥ 25cmH2O; L1, latency to P1; L2, latency to P2; L3, latency to P3; LTP, latency to peak; minP1P2, amplitude of the nadir between P1 and P2; P1, Amplitude of the percussion wave; P2, Amplitude of the tidal wave; P2/P1, ratio between the amplitudes of P2 and P1; P3, Amplitude of the dicrotic wave; S1, slope of P1; S2, slope of P2; S3, slope of P3; TL, total latency of the waveform.
An abnormal lumbar puncture test was defined as an opening pressure > 25cmH2O and/or a CSF cell count > 4/mL and/or a CSF protein level > 40 mg/dL.
p < 0.05.
L1 (p = 0.002), L3 (p = 0.022) and TL (p = 0.009) were able to significantly predict an abnormal test result in univariate regression analyses. A multivariate model including these variables significantly explained 12.6% to 17.2% of the probability of an abnormal LP result (p = 0.012), with 92.3% sensitivity and 34.5% specificity (Table S6). Using the absolute B of these variables, as all were negative, the scoring equation below was created:
This equation presented with an AUROC of 0.696 (95% CI 0.572–0.820) for detecting the outcome in question, with scores lower than 5.0973 having a sensitivity of 96.2% and specificity of 24.1% (Figure S3).
3.3. Relationship Between Waveform Changes Induced by Intracranial Pressure‐Lowering and ‐Raising Maneuvers and Lumbar Puncture Results
3.3.1. Intracranial Pressure‐Lowering Maneuvers
Post‐LP NIWA (Group L) presented with prolonged TL parameter (0.894 ± 0.139 s vs. 0.862 ± 0.175 s, p < 0.001, Table S7) compared to baseline NIWA. Percentage changes of these parameters were not associated with IH, even after excluding individuals with normal chemocytology (Tables S8 and S9). Likewise, there was no association between them and an abnormal chemocytology. When subjects with IH were excluded, mean L1 percentage change became significantly higher in those with abnormal chemocytology (−0.6 ± 36.6 vs. 14.6 ± 89.6, p = 0.041), but provided poor accuracy (AUROC 0.663, 95% CI 0.516–0.810, Tables S10 and S11). No association was found between waveform parameters percentage change and an overall abnormal LP result (Table S12).
On the other hand, recordings in the sitting position (Group S) reduced the frequency of visible tidal (64.5% vs. 90.3%, p = 0.027) and dicrotic (12.9% vs. 64.5%, p < 0.001) peaks when compared to the baseline NIWA. Also, several parameters became significantly reduced: S1 (0.077 ± 0.060 mV/s vs. 0.136 ± 0.121 mV/s, p = 0.027), S2 (0.037 ± 0.031 mV/s vs. 0.070 ± 0.046 mV/s, p = 0.010), A (0.008 ± 0.006 mV/s vs. 0.014 ± 0.014 mV/s, p = 0.009), and L2 (0.037 ± 0.031 mV/s vs. 0.330 ± 0.085 mV/s, p = 0.045, Table S13). There was no association between parameters percentage change and IH (Table S14). A sensitivity analysis without individuals with abnormal chemocytology was not possible due to the small number of participants with IH (n = 4) among them. Additionally, there was no association between these percentage changes and CSF chemocytology, even when subjects with IH were excluded, nor between them and an overall abnormal LP (Tables S15, S16 and S17).
3.3.2. Intracranial Pressure‐Raising Maneuver
When compared to the baseline NIWA, controlled Valsalva maneuver (Group V) led to significant changes to the tidal wave, which presented reduced mean P2 (0.018 ± 0.018 s vs. 0.024 ± 0.013 s, p = 0.009) and S2 (0.047 ± 0.046 mV/s vs. 0.066 ± 0.039 mV/s, p < 0.001), and longer mean L2 (0.424 ± 0.125 s vs. 0.369 ± 0.064 s, p = 0.019). This resulted in a reduced mean P2/P1 ratio (0.819 ± 0.284 vs. 1.03 ± 0.360, p = 0.041). Also, the following parameters decreased significantly: P3 (0.011 ± 0.012 mV vs. 0.019 ± 0.013 mV, p < 0.001), S3 (0.023 ± 0.022 mV/s vs. 0.047 ± 0.043 mV/s, p < 0.001), and LTP (0.191 ± 0.097 s vs. 0.267 ± 0.126 s, p = 0.014; Table S18). Mean parameter percentage changes were not associated with IH, even after removing subjects with chemocytological changes (Tables S19 and S20). Although mean P2 percentage change was significantly smaller in individuals with abnormal chemocytology (−5.6%± 46.0% vs. 48.0%± 29.0%, p = 0.021), presenting with an acceptable accuracy, this finding was not retained after excluding subjects with IH (Tables S21 and S22). No association was observed between parameters percentage change and an overall abnormal LP test result (Table S23).
4. Discussion
This exploratory study provided first‐hand data about the accuracy of a cranial extensometer‐based NIWA measurements in predicting relevant abnormalities in LP results. IH was associated with significantly smaller amplitudes and slopes for the percussion (P1 and S1) and dicrotic (P3 and S3) peaks, and smaller waveform area. Particularly, reduced S3 was able to predict IH with 89.5% sensitivity. Additionally, equations including predominantly latency‐related parameters (i.e., L1, L3 and TL) could significantly predict changes in CSF chemocytology and an overall abnormal LP test (i.e., IH and/or altered CSF chemocytology) with ≥ 95% sensitivity, supporting its usefulness as a screening method.
Despite being studied since the 19th century, the mechanisms behind the ICP waveform three‐peaked morphology remain only partially understood [13]. The mainstay hypothesis is that the percussion peak (P1) results from the pulsation of the choroid plexus and large intracranial vessels; the tidal peak (P2) from the rebound of the latter, being closely related to peak brain arterial blood volume and brain compliance; and the dicrotic peak (P3) is associated with cranial venous outflow [14]. Initial studies have demonstrated that in normal circumstances P1 > P2 > P3 [14, 21], while when ICP rises pathologically, and intracranial compliance becomes depleted, P2 becomes dominant [22]. However, these studies were primarily conducted in acute severe IH models [13, 14]. There is a notable lack of data on the behavior of waveform morphology in chronic conditions found in outpatient settings, in which both disease‐related intracranial compliance changes and long‐term compensations may occur. Idiopathic IH has been associated with reduced intracranial compliance, which interestingly was not always correlated with LP opening pressure [23]. Meanwhile, CSF dynamics modifications in spontaneous intracranial hypotension have been found to be more pronounced in the early phases of the disease, when orthostatic headache is more frequent, tending to normalize over time, suggesting the role of adaptive mechanisms [24]. Therefore, it may be possible that these complex bidirectional changes to the intracranial compliance and CSF dynamics may interfere in waveform characteristics, and its association with ICP.
In fact, waveform changes observed here in association with IH differ from those previously described within the neurocritical care context [15, 16, 17]. In these preceding studies, P2, and therefore P2/P1, have been found to rise, signaling the depletion of intracranial compliancy mechanisms [13, 14, 15, 16, 17]. In fact, an elevated P2/P1 was able to discriminate IH with significant accuracies [10]. This has not been replicated in an outpatient context, in which P2/P1 was not associated with any of the tested outcomes. This may have resulted from compliance mechanisms not being completely exhausted or to successful compensatory responses occurring in the outpatient setting, but not in patients in critical states. Nonetheless, decreased P1 and P3 were associated with IH. As CSF production is mostly dependent on the choroid plexus [25, 26], a compensatory reduction of its perfusion and pulsatility may explain the lower P1. Meanwhile, P3 (and consequently, S3) reductions may reflect lower pulsatility of the cranial venous outflow, which has been previously described in idiopathic IH and attributed to increased venous pressure requirements to decompress intracranial veins as ICP rises [27, 28]. Additionally, for the first time, ICP waveform morphology has been independently linked to CSF chemocytology changes. Interestingly, this was characterized by reductions in latency‐related variables, instead of the waveform amplitudes. The mechanisms underlying this finding are unclear, but higher CSF viscosity that accompanies these chemocytological changes may have resulted in faster waveform pulse propagation.
Irrespective of the biological underpinnings of these waveform morphology changes, NIWA presented with a high sensitivity (i.e., 96.2%) for predicting relevant abnormalities in LP results. This is clinically meaningful, as it supports its use as a pre‐LP screening method to rule out the need for this procedure, particularly in cases of low‐to‐moderate pretest probabilities of an abnormal result. It is estimated that nearly 400.000 LP are performed yearly by neurologists in the United States, amounting to US$ 146 million [29, 30]. Considering that LP presents with a positive yield of approximately 50% in common outpatient neurological conditions [3, 4], if systematic screening with NIWA was performed before LP, up to 50,000 outpatient non‐diagnostic tests could be prevented yearly.
Meanwhile, the low specificity presented by NIWA in detecting an abnormal LP result may be explained by the influence of unknown factors on ICP waveform morphology, affecting intracranial compliance or wave propagation itself, aside from IH and meningeal inflammation. For example, individuals suffering from long‐term systemic hypertension have been found to have higher mean P2/P1 ratio [31] when compared to what has been documented in healthy subjects' cohorts [32]. A higher P2/P1 ratio has also been reported among women, the elderly [31], and individuals with chronic kidney disease [33]. Notably, this parameter tended to normalize after hemodialysis [33, 34]. Additionally, these factors, combined with physiological fluctuations in waveform morphology driven by the respiratory cycle [35], may have contributed to the relatively high variability observed in the distribution of the investigated waveform parameters. Notably, the magnitude of this dispersion was comparable to that reported for NIWA parameters addressed in the existing literature [15, 36].
Finally, this study also examined NIWA parameter behavior during experimentally‐induced ICP increases (Valsalva maneuver) and reductions (CSF extraction and sitting position) to investigate whether parameter variations differed according to LP results as a marker of craniospinal compliance impairment. Group L received a larger subsample because LP was the only intervention whose ICP‐lowering effect could be directly monitored via closing pressure, and NIWA could be undergone with less susceptibility to positioning and muscle‐strain artifacts. As expected, NIWA demonstrated sensitivity to maneuver‐induced ICP changes; however, these maneuvers did not significantly improve the method's screening performance. It is possible that, in the studied outpatient population, in which compensatory mechanisms are not expected to be severely impaired, these ICP changes were insufficient to produce significantly distinct parameter variations.
This study has limitations that must be considered. The sample was heterogeneous, encompassing various underlying conditions. Considering that long‐term adaptations in CSF dynamics may be different across different etiologies [37], waveform behavior and NIWA accuracy may also vary across etiological subgroups. Although the sample size is compatible with existing literature, only four participants in group S had IH, potentially underpowering the analyses of the contribution of sitting position NIWA to the detection of IH. Extracted CSF volume was not standardized and varied significantly between subjects, leading to heterogeneous ICP reductions after LP and thereby may have limited the analysis of resulting NIWA waveform morphological changes within group L. Moreover, LP opening pressure, the gold standard for IH diagnosis in the outpatient setting, is not completely precise, as measurements can be influenced by the subject's anxiety and Valsalva maneuver during the procedure. Although deriving parameters from average waveforms computed over several minutes enabled characterization of its typical morphology for each individual and condition, and is consistent with the existing literature [10], it may have introduced morphological smoothing and does not permit assessment of beat‐to‐beat waveform variability. Finally, as this was a cross‐sectional study, the NIWA sensitivity in detecting ICP and CSF chemocytology changes throughout time was not addressed.
In conclusion, the skull extensometer‐based NIWA approach was found to be feasible, and able to detect not only IH, but also CSF chemocytology abnormalities or both, with high sensitivity in a real‐life neurological outpatient population. The observed differences in recorded waveform behavior compared to patterns documented in neurocritical settings may reflect long‐term disease‐related and compensatory mechanisms, warranting further investigation. Nevertheless, the diagnostic performance of this approach supports its potential role as a pre‐LP screening tool. While LP will remain essential for the evaluation of many cases, the use of screening strategies such as this may provide higher diagnostic yields, especially in cases of low‐to‐moderate pretest probability, while reducing procedural complications and associated healthcare costs.
Author Contributions
Gabriel Taricani Kubota: conceptualization, methodology, project administration, supervision, funding acquisition. Alessandra Santos da Fonseca: investigation, data curation, formal analysis, validation, writing – original draft, writing – review and editing. Cintya Yukie Hayashi: investigation, writing – review and editing. Rebecca Gabriela de Oliveira Pereira: investigation, data curation, writing – review and editing. Thauan Gonçalves: investigation, writing – review and editing. Helio Rodrigues Gomes: investigation, data curation, formal analysis, writing – review and editing, validation. Valquíria Aparecida da Silva: investigation, data curation, writing – review and editing. Ida Fortini: validation, formal analysis, writing – review and editing. Eberval Gadelha Figueiredo: writing – review and editing, formal analysis, validation. Lin Tchia Yeng: validation, formal analysis, writing – review and editing. Manoel Jacobsen Teixeira: validation, formal analysis, writing – review and editing. Daniel Ciampi de Andrade: conceptualization, methodology, supervision, funding acquisition, project administration, writing – review and editing, writing – original draft, formal analysis, validation.
Funding
The study received financial support from LIM62—Pain Center of the Department of Neurology at the University of São Paulo, to cover the costs of conducting the analyses and supporting the essential data management team. Brain4Care Brazil (São Carlos, Brazil) provided the studied BcSs‐PICNI‐2000 sensor, as well as a stipend to co‐fund AFS during the data collection phase, under an investigator‐initiated research project. Brain4Care Brazil had no role in the initiative to conduct the study, its design, data analyses, drafting of the manuscript, revision of the manuscript, or in the decision on where and when to publish it. Brain4Care Brazil never had access to the data of the study before its publication.
Conflicts of Interest
C.Y.H. and T.G. were both employed by Brain4Care Brazil, which has developed the studied device, during the recruitment phase of this research. They provided training in the use of this device to the research team, as well as significant support to data collection, but did not take any other part in this study. The authors report no other potential conflicts of interest.
Supporting information
Figure S1: Controlled Valsalva maneuver.
Figure S2: Receiver operating curve of Equation 1 for discriminating an abnormal cerebrospinal fluid chemocytology.
Figure S3: Receiver operating curve of Equation 2 for discriminating an overall abnormal lumbar puncture test.
Table S1: Additional demographic and cerebral spinal fluid data from study participants.
Table S2: Waveform parameters of subjects with normal cerebrospinal fluid chemocytology, according to the presence of intracranial hypertension.
Table S3: Univariate logistic regression of waveform parameters for predicting intracranial hypertension.
Table S4: Waveform parameters data of study subjects without intracranial hypertension, according to cerebrospinal fluid analysis results.
Table S5: Univariate and multivariate logistic regression models with waveform parameters for predicting abnormal cerebrospinal fluid chemocytology.
Table S6: Univariate and multivariate logistic regression models with waveform parameters for predicting an overall abnormal lumbar puncture result.
Table S7: Changes in waveform parameters after the lumbar puncture.
Table S8: Association between percentage changes of waveform parameters after the lumbar puncture and intracranial hypertension.
Table S9: Association between percentage changes of waveform parameters after the lumbar puncture and intracranial hypertension among subjects with normal cerebrospinal fluid chemocytology.
Table S10: Association between percentage changes of waveform parameters after the lumbar puncture and cerebrospinal fluid chemocytology.
Table S11: Association between percentage changes of waveform parameters after the lumbar puncture and cerebrospinal fluid chemocytology, among individuals without intracranial hypertension.
Table S12: Association between percentage changes of waveform parameters after the lumbar puncture and the overall lumbar puncture test result.
Table S13: Changes in waveform parameters between supine and sitting positions.
Table S14: Association between percentage changes of waveform parameters during sitting position and intracranial hypertension.
Table S15: Association between percentage changes of waveform parameters during sitting position and cerebrospinal fluid chemocytology.
Table S16: Association between percentage changes of waveform parameters during sitting position and cerebrospinal fluid chemocytology among subjects without intracranial hypertension.
Table S17: Association between percentage changes of waveform parameters during sitting position and the overall lumbar puncture test result.
Table S18: Changes in waveform parameters between resting and controlled Valsalva maneuver conditions.
Table S19: Association between percentage changes of waveform parameters during controlled Valsalva maneuver and intracranial hypertension.
Table S20: Association between percentage changes of waveform parameters during controlled Valsalva maneuver and intracranial hypertension among subjects with normal cerebrospinal fluid chemocytology.
Table S21: Association between percentage changes of waveform parameters during controlled Valsalva maneuver and cerebrospinal fluid chemocytology.
Table S22: Association between percentage changes of waveform parameters during controlled Valsalva maneuver and cerebrospinal fluid chemocytology among individuals without intracranial hypertension.
Table S23: Association between percentage changes of waveform parameters during controlled Valsalva maneuver and the overall lumbar puncture test result.
Acknowledgments
The authors have nothing to report.
Data Availability Statement
Anonymized data not published within this article will be made available upon reasonable request to the corresponding author from any qualified investigator.
References
- 1. Wright B. L. C., Lai J. T. F., and Sinclair A. J., “Cerebrospinal Fluid and Lumbar Puncture: A Practical Review,” Journal of Neurology 259, no. 8 (2012): 1530–1545. [DOI] [PubMed] [Google Scholar]
- 2. Engelborghs S., Niemantsverdriet E., Struyfs H., et al., “Consensus Guidelines for Lumbar Puncture in Patients With Neurological Diseases,” Alzheimer's & Dementia: Diagnosis, Assessment & Disease Monitoring 8 (2017): 111–126. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Williams M. A., Nagel S. J., Luciano M. G., et al., “The Clinical Spectrum of Hydrocephalus in Adults: Report of the First 517 Patients of the Adult Hydrocephalus Clinical Research Network Registry,” Journal of Neurosurgery 132, no. 6 (2020): 1773–1784. [DOI] [PubMed] [Google Scholar]
- 4. Raggi A., Bianchi Marzoli S., Ciasca P., et al., “The Pre‐Lumbar Puncture Intracranial Hypertension Scale (PLIHS): A Practical Scale to Identify Subjects With Normal Cerebrospinal Fluid Pressure in the Management of Idiopathic Intracranial Hypertension,” Journal of the Neurological Sciences 429 (2021): 118058. [DOI] [PubMed] [Google Scholar]
- 5. Uppal V., Russell R., Sondekoppam R. V., et al., “Evidence‐Based Clinical Practice Guidelines on Postdural Puncture Headache: A Consensus Report From a Multisociety International Working Group,” Regional Anesthesia and Pain Medicine 49, no. 7 (2023): 471–501. [DOI] [PubMed] [Google Scholar]
- 6. Monserrate A. E., Ryman D. C., Ma S., et al., “Factors Associated With the Onset and Persistence of Post–Lumbar Puncture Headache,” JAMA Neurology 72, no. 3 (2015): 325–332. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Charalambous L. T., Rajkumar S., Liu B., et al., “Treatment Patterns and Healthcare Resource Utilization of Iatrogenic Spinal Cerebrospinal Fluid Leaks in the United States,” Clinical Spine Surgery 35, no. 9 (2022): E725–E730. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. De Moraes F. M. and Silva G. S., “Noninvasive Intracranial Pressure Monitoring Methods: A Critical Review,” Arquivos de Neuro‐Psiquiatria 79, no. 5 (2021): 437–446. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Raboel P. H., Bartek J., Andresen M., Bellander B. M., and Romner B., “Intracranial Pressure Monitoring: Invasive Versus Non‐Invasive Methods‐A Review,” Critical Care Research and Practice 2012 (2012): 950393. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Brasil S., Godoy D. A., and Hawryluk G. W. J., “A Point‐Of‐Care Noninvasive Technique for Surrogate ICP Waveforms Application in Neurocritical Care,” Neurocritical Care 40, no. 1 (2024): 170–176. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Frigieri G., Andrade R. a. P., Wang C. C., et al., “Analysis of a Minimally Invasive Intracranial Pressure Signals During Infusion at the Subarachnoid Spinal Space of Pigs,” Acta Neurochirurgica Supplement 126 (2018): 75–77. [DOI] [PubMed] [Google Scholar]
- 12. Frigieri G., Andrade R. a. P., Dias C., et al., “Analysis of a Non‐Invasive Intracranial Pressure Monitoring Method in Patients With Traumatic Brain Injury,” Acta Neurochirurgica Supplementum 126 (2018): 107–110. [DOI] [PubMed] [Google Scholar]
- 13. Frigieri G., Hayashi C. Y., Rabelo N. N., et al., “Intracranial Pressure Waveform: History, Fundamentals and Applications in Brain Injuries,” in Advancement and New Understanding in Brain Injury (InTechOpen, 2020). [Google Scholar]
- 14. Czosnyka M. and Czosnyka Z., “Origin of Intracranial Pressure Pulse Waveform,” Acta Neurochirurgica 162, no. 8 (2020): 1815–1817. [DOI] [PubMed] [Google Scholar]
- 15. Brasil S., Frigieri G., Taccone F. S., et al., “Noninvasive Intracranial Pressure Waveforms for Estimation of Intracranial Hypertension and Outcome Prediction in Acute Brain‐Injured Patients,” Journal of Clinical Monitoring and Computing 37, no. 3 (2023): 753–760. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. de Moraes F. M., Brasil S., Frigieri G., Robba C., Paiva W., and Silva G. S., “ICP Wave Morphology as a Screening Test to Exclude Intracranial Hypertension in Brain‐Injured Patients: A Non‐Invasive Perspective,” Journal of Clinical Monitoring and Computing 38, no. 4 (2024): 773–782. [DOI] [PubMed] [Google Scholar]
- 17. Frigieri G., Gonçalves T. L., Ocamoto G. N., de AP Andrade R., de Padua B. C., and Cardim D., “Clinical Performance of the Brain4care System for Noninvasive Detection of Intracranial Hypertension,” Neurocritical Care 43 (2025): 628–635. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Braincare Desenvolvimento e Inovações Tecnológicas SA , Guia do Usuário BcSs‐PICNI‐2000 ‐ Sensor Braincare não invasivo para monitorização da pressão intracraniana (PIC).
- 19. Whiteley W., Al‐Shahi R., Warlow C. P., Zeidler M., and Lueck C. J., “CSF Opening Pressure: Reference Interval and the Effect of Body Mass Index,” Neurology 67, no. 9 (2006): 1690–1691. [DOI] [PubMed] [Google Scholar]
- 20. Hosmer D. W., Lemeshow S., and Sturdivant R. X., Applied Logistic Regression: Third Edition (Wiley‐Interscience Publication, 2013). [Google Scholar]
- 21. Scalzo F., Hamilton R., Hu X., Scalzo F., Hamilton R., and Hu X., “Real‐Time Analysis of Intracranial Pressure Waveform Morphology,” in Advanced Topics in Neurological Disorders (IntechOpen, 2012). [Google Scholar]
- 22. Germon K., “Interpretation of ICP Pulse Waves to Determine Intracerebral Compliance,” Journal of Neuroscience Nursing 20, no. 6 (1988): 344–351. [DOI] [PubMed] [Google Scholar]
- 23. Chisholm J. T., Sudhakar P., Alhajeri A. N., and Smith J. H., “Intracranial Elastance Is Increased in Idiopathic Intracranial Hypertension,” European Journal of Neurology 24, no. 12 (2017): 1457–1463. [DOI] [PubMed] [Google Scholar]
- 24. Häni L., Fung C., Jesse C. M., et al., “Insights Into the Natural History of Spontaneous Intracranial Hypotension From Infusion Testing,” Neurology 95, no. 3 (2020): E247–E255. [DOI] [PubMed] [Google Scholar]
- 25. Bothwell S. W., Janigro D., and Patabendige A., “Cerebrospinal Fluid Dynamics and Intracranial Pressure Elevation in Neurological Diseases,” Fluids and Barriers of the CNS 16, no. 1 (2019): 1–18. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Eisma J. J., McKnight C. D., Hett K., et al., “Choroid Plexus Perfusion and Bulk Cerebrospinal Fluid Flow Across the Adult Lifespan,” Journal of Cerebral Blood Flow and Metabolism 43, no. 2 (2023): 269–280. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27. Unnerbäck M., Ottesen J. T., and Reinstrup P., “Increased Intracranial Pressure Attenuates the Pulsating Component of Cerebral Venous Outflow,” Neurocritical Care 31, no. 2 (2019): 273–279. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. Bateman G. A., “Vascular Hydraulics Associated With Idiopathic and Secondary Intracranial Hypertension,” AJNR. American Journal of Neuroradiology 23, no. 7 (2002): 1180–1186. [PMC free article] [PubMed] [Google Scholar]
- 29. Tung C. E., So Y. T., and Lansberg M. G., “Cost Comparison Between the Atraumatic and Cutting Lumbar Puncture Needles,” Neurology 78, no. 2 (2012): 109–113. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30. Menger R. P., Peyton W. A., Hefner M., Nanda A., and Cuellar H., “Economic Outcomes of the Addition of Fluoroscopic Guidance to the Lumbar Puncture Procedure: A Call for Standardized Training,” Journal of Spine 6, no. 1 (2017): 1000359. [Google Scholar]
- 31. da Costa M. M., Sousa A. L. L., Correia M. C., et al., “Intracranial Pressure Waveform in Patients With Essential Hypertension,” Frontiers in Cardiovascular Medicine 10 (2023): 10. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32. Ocamoto G. N., da Silva L. N., da Silva Rocha Tomaz C., et al., “Characterization of Intracranial Compliance in Healthy Subjects Using a Noninvasive Method ‐ Results From a Multicenter Prospective Observational Study,” Journal of Clinical Monitoring and Computing 38, no. 6 (2024): 1249–1261. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33. Koch M. S., Drewnowski B., Rickli C., Dos Santos F. A., Baroni G., and Vellosa J. C. R., “The Influence of Hemodialysis on Intracranial Pressure Waveform in Patients With Chronic Kidney Disease: An Observational Study,” São Paulo Medical Journal 142, no. 3 (2024): e2023068. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34. Rickli C., Cosmoski L. D., dos Santos F. A., et al., “Use of Non‐Invasive Intracranial Pressure Pulse Waveform to Monitor Patients With End‐Stage Renal Disease (ESRD),” PLoS One 16 (2021): e0240570. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35. Kazimierska A., Manet R., Vallet A., et al., “Analysis of Intracranial Pressure Pulse Waveform in Studies on Cerebrospinal Compliance: A Narrative Review,” Physiological Measurement 44, no. 10 (2023): 10TR01. [DOI] [PubMed] [Google Scholar]
- 36. de Moraes F. M., Rocha E., Barros F. C. D., et al., “Waveform Morphology as a Surrogate for ICP Monitoring: A Comparison Between an Invasive and a Noninvasive Method,” Neurocritical Care 37, no. 1 (2022): 219–227. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37. Gasparian S. S., Serova N. K., Sherbakova E. Y., and BeIova T. N., “Compensatory Mechanisms in Patients With Benign Intracranial Hypertension Syndrome,” Acta Neurochirurgica. Supplement 81 (2002): 31–33. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Figure S1: Controlled Valsalva maneuver.
Figure S2: Receiver operating curve of Equation 1 for discriminating an abnormal cerebrospinal fluid chemocytology.
Figure S3: Receiver operating curve of Equation 2 for discriminating an overall abnormal lumbar puncture test.
Table S1: Additional demographic and cerebral spinal fluid data from study participants.
Table S2: Waveform parameters of subjects with normal cerebrospinal fluid chemocytology, according to the presence of intracranial hypertension.
Table S3: Univariate logistic regression of waveform parameters for predicting intracranial hypertension.
Table S4: Waveform parameters data of study subjects without intracranial hypertension, according to cerebrospinal fluid analysis results.
Table S5: Univariate and multivariate logistic regression models with waveform parameters for predicting abnormal cerebrospinal fluid chemocytology.
Table S6: Univariate and multivariate logistic regression models with waveform parameters for predicting an overall abnormal lumbar puncture result.
Table S7: Changes in waveform parameters after the lumbar puncture.
Table S8: Association between percentage changes of waveform parameters after the lumbar puncture and intracranial hypertension.
Table S9: Association between percentage changes of waveform parameters after the lumbar puncture and intracranial hypertension among subjects with normal cerebrospinal fluid chemocytology.
Table S10: Association between percentage changes of waveform parameters after the lumbar puncture and cerebrospinal fluid chemocytology.
Table S11: Association between percentage changes of waveform parameters after the lumbar puncture and cerebrospinal fluid chemocytology, among individuals without intracranial hypertension.
Table S12: Association between percentage changes of waveform parameters after the lumbar puncture and the overall lumbar puncture test result.
Table S13: Changes in waveform parameters between supine and sitting positions.
Table S14: Association between percentage changes of waveform parameters during sitting position and intracranial hypertension.
Table S15: Association between percentage changes of waveform parameters during sitting position and cerebrospinal fluid chemocytology.
Table S16: Association between percentage changes of waveform parameters during sitting position and cerebrospinal fluid chemocytology among subjects without intracranial hypertension.
Table S17: Association between percentage changes of waveform parameters during sitting position and the overall lumbar puncture test result.
Table S18: Changes in waveform parameters between resting and controlled Valsalva maneuver conditions.
Table S19: Association between percentage changes of waveform parameters during controlled Valsalva maneuver and intracranial hypertension.
Table S20: Association between percentage changes of waveform parameters during controlled Valsalva maneuver and intracranial hypertension among subjects with normal cerebrospinal fluid chemocytology.
Table S21: Association between percentage changes of waveform parameters during controlled Valsalva maneuver and cerebrospinal fluid chemocytology.
Table S22: Association between percentage changes of waveform parameters during controlled Valsalva maneuver and cerebrospinal fluid chemocytology among individuals without intracranial hypertension.
Table S23: Association between percentage changes of waveform parameters during controlled Valsalva maneuver and the overall lumbar puncture test result.
Data Availability Statement
Anonymized data not published within this article will be made available upon reasonable request to the corresponding author from any qualified investigator.