Abstract
The prognostic value of cerebrospinal fluid (CSF) protein in Guillain Barré Syndrome (GBS) is unclear. We aimed to explore the potential association between CSF protein level and mechanical ventilation in GBS. We undertook a retrospective study of GBS patients from January 2000 to November 2019 at the University of Michigan. 94 patients were ultimately included for evaluation. After adjusting for the Erasmus GBS Respiratory Insufficiency Scale (EGRIS), we did not find a significant difference in CSF protein between ventilated and non-ventilated patients. Elevated CSF protein level does not appear to portend an increased likelihood of mechanical ventilation in GBS patients.
INTRODUCTION
Guillain-Barré Syndrome (GBS) is an autoimmune, non-length-dependent radiculoneuropathy that can lead to quadriparesis and respiratory failure. Nearly 25%of cases require mechanical ventilation and the disease carries a 3–10%mortality rate [1].
The decision to mechanical ventilate a patient in GBS is of utmost importance in critical care management [2]. Validated predictors of respiratory insufficiency include serum albumin, forced vital capacity, negative inspiratory force and the Erasmus GBS Respiratory Insufficiency Scale (EGRIS), as well as rapid disease progression, inability to cough, stand, or lift one’s head, bulbar dysfunction, bilateralfacial weakness, lack of foot flexion (at the end of immune therapy), and dysautonomia [3–7]. EGRIS predicts the need for mechanical ventilation and incorporates (1) time from weakness onset to admission, (2) the presence of facial and/or bulbar weakness, and (3) Medical Research council summated score (MRCSS) at the time of admission. A retrospective study of 541 GBS patients from a referral hospital in China essentially validated the EGRIS predictive model in a population that epidemiologically has more axonal variants of GBS, which tend to be more severe [8].
Higher CSF protein in GBS is believed to reflect nerve root inflammation, antibody and complement deposition, and nerve dysfunction, which is thought to worsen disease severity and respiratory insufficiency [9]. CSF protein is also greatly influenced by timing of lumbar puncture relative to symptom onset, as demonstrated in two retrospective studies, the first a Dutch study of 494 patients [10], and the second an Asian study of 507 patients [11].
With regards to the prognostic value of cerebrospinal fluid (CSF) protein, it is generally mixed. One study showed no prognostic value, as measured by the Inflammatory Rasch-built Overall Disability Scale (I-RODS), Fatigue Severity Scale (FSS) or the Medical Research Council Summated Score (MRCSS [12]. A previous, single-center retrospective study that we conducted at The Ohio State University suggested a possible association between CSF protein and mechanical ventilation (not statistically significant), but did not adjust for EGRIS, due to small study size. Based on that initial observation, we aimed to explore the potential utility of CSF protein as a short term prognostic biomarker for respiratory insufficiency, adjusting for EGRIS.
MATERIALS AND METHODS
Study Design
The study was approved by the institutional review board of the University of Michigan and done in compliance with guidelines on human experimentation. We undertook a retrospective study of all patients diagnosed with GBS from January 2000 to November 2019 using International Classification of Disease (ICD)-9 and -10 coding criteria (Guillain Barré Syndrome, Acute Inflammatory Demyelinating Polyradiculoneuropathy, Acute Motor Axonal Neuropathy, Acute Motor Sensory Axonal Neuropathy). Patients needed to be 18 years old or older, to have fulfilled Brighton criteria, and to have undergone CSF testing within 2 weeks of weakness onset. Patients with Miller-Fisher syndrome were excluded. The primary outcome was the need for mechanical ventilation. Baseline demographic and clinical information for each patient was collected, including age, sex, diabetes status (yes/no), EGRIS score (as defined previously), MRCSS, treatment administered (intravenous immunoglobulin, plasma exchange, neither, or both), Brighton score, mechanical ventilation (yes/no), time from weakness onset to lumbar puncture, time from admission to lumbar puncture, preceding diarrhea (yes/no), preceding upper respiratory tract illness (yes/no), chronic pulmonary issues (yes/no), as well as CSF protein, glucose, RBCs, and WBCs.
Statistical analysis
Descriptive statistics were used to characterize patient demographic and clinical information, stratified by use of mechanical ventilation. Fisher’s exact tests were used to compare categorical covariates, two-sample t-tests were used to compare continuous covariates that were normally distributed, and Wilcoxon rank-sum tests were used to compare highly positive-skewed covariates between patients with and without mechanical ventilation. Pearson’s correlation was used to assess the relationship between CSF protein and EGRIS score. We used multivariable logistic regression to determine the association between CSF protein and mechanical ventilation, after adjusting for EGRIS. All analyses were completed using R.v.3.4.2.
RESULTS
142 GBS patients were identified. We excluded 48 total patients due to the following reasons: age <18y (n = 20); a diagnosis of Miller-Fisher syndrome (n = 9); CSF data being beyond two weeks of onset of weakness (n = 8); and missing EGRIS data (n = 11). 94 patients were ultimately included in the final analysis. There were 13 (13.8%) patients that received mechanical ventilation. The population had a mean (SD) age of 48.6 (16.2) years, was 64.9%male, the mean (SD) time from weakness onset to lumbar puncture was 5.5 (3.7) days, and the mean (SD) time from admission to lumbar puncture was 0.7 (1.6) days. Demographic, clinical and CSF information stratified by mechanical ventilation use are displayed in Table 1. Of note, CSF samples were almost always obtained before or at the time of IVIG initiation of therapy. There were no significant differences in nearly all baseline characteristic between patients with and without mechanical ventilation. Patients with mechanical ventilation had lower median MRCSS at the time of CSF evaluation compared to patients without mechanical ventilation (mechanical ventilation: 42.0 (30.0–50.0) vs. no mechanical ventilation: 54.0 (46.0–58.0), Wilcoxon Rank-Sum Test p < 0.01).
Baseline characteristics between ventilated and non-ventilated patients
*Represents result from Wilcoxon Rank-Sum Test; &Represents result from Fisher’s Exact Test; $Represents result from Two-Sample T-Test; Legend: IQR: inner-quartile range, SD: standard deviation, EGRIS: Erasmsus GBS Respiratory Insufficiency Score.
The median (IQR) CSF protein level was higher in the ventilated group as compared to the non-ventilated group (mechanical ventilation: 107.0mg/dL (86.0–170.0) vs. no mechanical ventilation: 80.0 mg/dL (47.0–104.0), Wilcoxon Rank-Sum Test p < 0.01). After adjusting for EGRIS, CSF protein was not significantly associated with use of mechanical ventilation (odds ratio: 1.16, 95%CI: 0.93–1.43, unit = 30 mg/dL). After adjusting for CSF protein, EGRIS was significantly associated with mechanical ventilation (odds ratio: 1.67, 95%CI: 1.08–2.58). There was no significant correlation between CSF protein and EGRIS (Pearson’s correlation coefficient: 0.12, P = 0.26).
DISCUSSION
Our study showed no association between CSF protein level and the need for mechanical ventilation in GBS patients. As seen in our previous smaller (n = 36) study, in which median CSF protein level in the ventilated group was nearly double that in the non-ventilated group (116 mg/dL vs 57 mg/dL), our current study found ventilated patients to have higher median CSF protein levels than non-ventilated patients [13]. However, our current study adjusted for the EGRIS score and subsequently found no association. Despite its lack of association with ventilator status, we believe that CSF protein should continue to be explored as a prognostic biomarker, potentially as an adjunctive biomarker to existing predictive tools, and using disease severity outcome measures other than mechanical ventilation.
CSF protein elevation in GBS is thought to be due to blood-nerve barrier disturbance and damage to the peripheral nervous system at the level of the nerve roots. In addition, the presence of high levels of neurofilament proteins in CSF has been associated with poor clinical and electrodiagnostic outcomes [14, 15]. These findings suggest that higher CSF protein levels could be associated with more prominent radicular demyelination and axonal loss, which in turn, could be associated with more disability.
Other studies have evaluated the prognostic value of CSF protein and have shown mixed results, as previously discussed. Many studies show no definite clinical utility to CSF protein as a short- or long-term prognosis marker in GBS [12, 17]. Our previous and current studies are the only ones to explore the association between CSF protein and mechanical ventilation. While our previous study suggested a potential association, our current study, adjusting for EGRIS, definitively showed no clear role for CSF protein in predicting respiratory insufficiency.
Our study faced clear limitations. The limited number of patients in our mechanical ventilation arm (n = 13) precluded our ability to control for confounders beyond the EGRIS score in our logistic regression model. We also believe this small size to be the reason that the EGRIS score did not differ at a statistically significant level between the ventilated and non-ventilated groups (although it met statistical significance when adjusting for CSF protein). The retrospective nature of the study also limited our ability to ascertain baseline GBS disease severity (0 = healthy, 6 = death), as levels 1, 2, and 3 could not be accurately discerned for many patients.
We did not control for potentially confounding variables such as timing of lumbar puncture relative to onset of weakness nor timing of IVIG. The median time of lumbar puncture relative to symptom onset in the ventilated group was higher than in the non-ventilated group (7.1 days vs 5.3 days), which could have contributed to the higher CSF protein observed. However, this difference was not statistically significant, nor were other baseline characteristics.
It is worth noting that our percentage of ventilated patients (14%) is low in comparison to general figures for GBS (∼25%) [1, 4]. One possible explanation is selection bias. Our institution is a tertiary care center that receives transfers of ventilated GBS patients from other centers in the southeast Michigan region. In many instances, ventilated GBS patients were excluded from analysis due to either lack of clear admission CSF and clinical data per outside hospital records, or due to an inability to ascertain the timing of lumbar puncture relative to symptom onset.
In conclusion, higher CSF protein level does not associate with the need for mechanical ventilation in GBS patients. In light of the ongoing International GBS Outcome Study (IGOS) [18], the role of CSF protein, using other potential outcome measures, may be worthy of further exploration.
ACKNOWLEDGMENTS INCLUDING SOURCES OF SUPPORT
ER is supported by NIH T32 (NS0007222). All other authors report no sources of support.
CONFLICT OF INTEREST
The authors have no conflict of interest to report.