A, Serum albumin levels during a follow-up of 26 weeks in patients with GBS treated with IVIG. Dotted lines represent reference range values of albumin (3.5-5.5 g/dL [to convert to grams per liter, multiply by 10]). B, For comparison, the serum IgG levels in the same cohort of patients are shown (reprinted with permission from Wiley6) (to convert to grams per liter, multiply by 0.01). Boxes indicate medians with interquartile ranges (IQRs); whiskers (according to the Tukey test), 1.5 × IQR; and dots outside whiskers, outliers. C, Scatterplot showing the correlation between the change in albumin and IgG levels. D, Scatterplot showing the correlation between the albumin and IgG levels at 2 weeks after treatment. Solid line indicates regression (ρ); dotted lines, 95% CI of the regression.
Kaplan-Meier analysis of patients with GBS regaining the capacity to walk unaided for more than 10 m (GBS disability score of 2) in relation to tertiles of serum albumin levels at 2 weeks after start of intravenous immunoglobulin treatment (A) and hypoalbuminemia (albumin, <3.5 g/dL) (n = 60) vs normoalbuminemia (albumin, ≥3.5 g/dL) (n = 114) (B) at the same time point. To convert albumin to grams per liter, multiply by 10.
eFigure 1. Recovery Based on Pretreatment Serum Albumin Levels
eFigure 2. Age Influences Serum Albumin Levels
Customize your JAMA Network experience by selecting one or more topics from the list below.
Fokkink WR, Walgaard C, Kuitwaard K, Tio-Gillen AP, van Doorn PA, Jacobs BC. Association of Albumin Levels With Outcome in Intravenous Immunoglobulin–Treated Guillain-Barré Syndrome. JAMA Neurol. 2017;74(2):189–196. doi:10.1001/jamaneurol.2016.4480
Do serum albumin levels correlate with the clinical course and outcome of Guillain-Barré syndrome in patients treated with intravenous immunoglobulin?
In this cohort study of 174 patients with Guillain-Barré syndrome, the serum albumin level at 2 weeks after treatment correlated with clinical recovery, independent of other clinical prognostic factors. More than one-third of the patients were hypoalbuminemic, but even low-normal albumin levels were associated with poor outcome.
Albumin is an easily accessible and strong prognostic biomarker for Guillain-Barré syndrome in patients treated with intravenous immunoglobulin.
There is an urgent need for biomarkers to monitor treatment efficacy and anticipate outcome in patients with Guillain-Barré syndrome (GBS).
To assess whether there is an association between serum albumin levels, a widely used and relatively easily measurable biomarker of health and inflammation, and the clinical course and outcome of GBS in patients treated with intravenous immunoglobulin (IVIG).
Design, Setting, and Participants
We used serum samples derived from a cohort of patients with GBS admitted to hospitals across the Netherlands participating in national GBS studies from May 5, 1986, through August 2, 2000. Serum albumin was measured from January 13 to 20, 2011. Analysis was performed from February 25, 2013, to September 6, 2016. All patients fulfilled the criteria for GBS and had severe disease (defined as not being able to walk unaided >10 m). Patients misdiagnosed as having GBS were retrospectively excluded from the study. Serum samples were obtained before and after IVIG treatment at 4 standardized time points from 174 patients. Albumin levels were determined by routine diagnostic turbidimetry and related to demographics and clinical course during a follow-up of 6 months.
Main Outcomes and Measures
Serum albumin concentration was determined before and after treatment with IVIG and related to clinical outcome: muscle weakness (measured by Medical Research Council sum score), respiratory failure (measured by requirement and duration of mechanical ventilation), and ability to walk (measured by GBS disability score).
Serum albumin levels were determined in 174 patients with GBS (mean [SD] age, 49.6 [20.1] years; 99 males [56.9%]). Before treatment, the median serum albumin level was 4.2 g/dL (interquartile range, 3.8-4.5 g/dL), with hypoalbuminemia (albumin, <3.5 g/dL) in 20 (12.8%) of 156 patients. Two weeks after commencing treatment with IVIG (2 g/kg), the median serum albumin level decreased to 3.7 g/dL (interquartile range, 3.2-4.1 g/dL) (P < .001), and the number with hypoalbuminemia increased to 60 (34.5%) of 174 (P < .001). Hypoalbuminemia was associated with an increased chance of respiratory failure before (16 [36.4%] of 44, P = .001) or after (29 [54.7%] of 53, P < .001) IVIG treatment, inability to walk unaided (21 [35.0%] of 60 vs 6 [5.3%] of 114, P < .001), and severe muscle weakness at 4 weeks (Medical Research Council sum score, 31.8 vs 52.9, P < .001) and 6 months (Medical Research Council sum score, 49.4 vs 58.4, P < .001).
Conclusions and Relevance
Patients with GBS may develop hypoalbuminemia after treatment with IVIG, which is related to a more severe clinical course and a poorer outcome. Further studies are required to confirm that serum albumin can be used as a biomarker to monitor disease activity and treatment response to IVIG in patients with GBS.
Guillain-Barré syndrome (GBS) is a polyradiculoneuropathy characterized by a rapidly progressive bilateral paresis of the limbs. Nadir is typically reached within a number of days or weeks, followed by a recovery that is generally much slower and often incomplete.1 For more than 2 decades, intravenous immunoglobulin (IVIG) has been the treatment of choice, and all patients still receive the same high dose of IVIG (2 g/kg over 5 days).2 Despite the proven efficacy of this high-dose regimen in GBS, recovery and outcome still vary greatly among patients.3 Some of the more severely affected patients might benefit from an additional IVIG course, which is currently being studied in an ongoing randomized clinical trial (Second IVIG Dose in GBS Patients With Poor Prognosis).3 Still, the reasons why some patients respond poorly to IVIG therapy are unknown, and there is an urgent need to find a biomarker that, preferably, can be determined within the first 2 weeks of onset. Such a biomarker would allow a more personalized approach to monitor treatment efficacy and anticipate outcome.4,5 Existing prognostic models are based on clinical features, including the extent of muscle weakness and demographic factors, but previous studies3 failed to identify a serologic biomarker to enhance these models.
To further improve prognosis and assessment of treatment response, biomarkers reflecting IVIG efficacy are needed. For purposes of clinical practice, the measurement of such biomarkers should preferably be straightforward and accurate and align with routine diagnostic procedures. Although the change in serum level of IgG after IVIG treatment has been identified as a candidate biomarker, no distinction could be made between exogenous, IVIG-derived IgG and endogenous IgG.6 Whereas the IgG level after IVIG logically increases, the serum albumin level is reduced after high-dose IVIG therapy in diseases other than GBS.7,8 Like IgG, albumin binds to the neonatal Fc receptor (FcRn), which transports it back into the circulation.9,10 Furthermore, serum albumin is identified as an independent factor associated with outcome in amyotrophic lateral sclerosis and failure of IVIG therapy in Kawasaki disease.11,12 Therefore, serum albumin is an interesting alternative to IgG as a biomarker for GBS, fitting the profile of a routinely measured protein already established as a prognostic marker in numerous pathologic conditions.13
In this study, we aimed to determine whether serum albumin levels can serve as a prognostic marker in patients with GBS treated with IVIG. We determined the variation in serum albumin levels over time and assessed the serum albumin levels in response to IVIG after treatment. Finally, we analyzed whether circulatory albumin levels were related to disease severity and clinical outcome.
The patients were included in 2 clinical trials previously conducted from May 5, 1986, through August 2, 2000, in which clinical data and serum samples were prospectively collected according to a predefined standard protocol.14,15 Serum albumin was measured from January 13 to 20, 2011. Analysis was performed from February 25, 2013, to September 6, 2016. All patients provided written informed consent after approval by the institutional review board of Erasmus University Medical Center Rotterdam. The database, including all participants, was deidentified before analyzing.
All 174 patients in the current study received the same dosage of IVIG (2 g/kg over 5 days of Gammagard or Gammagard S/D [Baxter International]) and were previously used to appraise the pharmacokinetics of IgG (Table 1).6 For 18 patients (10.3%), stored pretreatment serum was of insufficient quantity to determine the albumin level. In one of the previous trials, intravenous methylprednisolone (500 mg/d for 5 consecutive days) was administered in addition to IVIG (concerning 60 patients [34.5%] included in the current study). Samples were taken at standardized time points and were stored at −80°C until use. Clinical condition of the patients during the trials was monitored using the Medical Research Council (MRC) sum score, ranging from 0 (tetraparalysis) to 60 (normal strength), and by the GBS disability scale, ranging from 0 (healthy) to 6 (deceased). Not being able to walk 10 m independently (GBS disability score >2) at 6 months was regarded as a poor outcome, as defined previously.6,16-18 Serum was collected before treatment (at randomization) and after treatment at 2, 4, 14, and 24 weeks (end of follow-up).14,15
Serum albumin concentrations were determined by routine automated diagnostic turbidimetry using a clinical chemistry analyzer (Hitachi 917; Hitachi Ltd). Because the serum samples (−80°C) were stored for a long period, a number of samples wherein the albumin level was previously determined were reanalyzed to exclude the possible influence of long-term storage on the measurement.
To compare the albumin levels at different time points, the Friedman analysis of variance with the Dunn post hoc test was used. Spearman ρ was used to assess correlation between the change in IgG (ΔIgG) and the change in serum albumin (Δalbumin) 2 weeks after IVIG treatment. The ΔIgG and Δalbumin were calculated by subtracting the pretreatment level from the 2-week posttreatment level of the respective serum protein. To assess the possible influence of serum albumin on disease severity, 2 strategies were used. In the first analysis, patients were stratified in tertiles based on the serum albumin level before or 2 weeks after commencing IVIG therapy. In the second analysis, patients were divided into 2 groups: those with hypoalbuminemia (albumin, <3.5 g/dL [to convert to grams per liter, multiply by 10]) and those within the reference range of serum albumin levels (3.5-5.5 g/dL).19 Because the serum albumin level decreases with advanced age, analysis of covariance (ANCOVA) and Cox proportional hazards regression analyses were corrected for age.20 The groups were compared by ANCOVA with respect to various outcome measures, including the requirement and duration of mechanical ventilation and the clinical severity at nadir at 4 weeks after initiating treatment and at the end of follow-up of 6 months. Kaplan-Meier analysis, stratified for the GBS disability score at entry or stratified for age groups (Table 1), was used to assess the ability to walk unaided at the end of follow-up. Subsequent Cox proportional hazards regression analysis was used to correct for a possible age effect. Logistic regression analysis was performed to assess the effect on disease severity and clinical outcome. The syntax of previously developed prognostic models was used to assess the contribution of albumin. These models were correlated with the need of respiratory support in the first week (Erasmus GBS Respiratory Insufficiency Score [EGRIS]) and the outcome during follow-up and at 6 months (Erasmus GBS Outcome Score or modified Erasmus GBS Outcome Score [EGOS/mEGOS]).16-18 Statistical analysis was performed using GraphPad Prism, version 6.0 for Windows (GraphPad Software) and SPSS, version 21.0 for Windows (SPSS Inc). Statistical significance was defined as a 2-sided P < .05.
Serum albumin levels were determined in 174 patients with GBS (mean [SD] age, 49.6 [20.1] years; 99 males [56.9%]) at several standardized, predefined time points (Figure 1A). In the acute phase of GBS before treatment, circulatory albumin levels were within the reference range in most patients (median, 4.2 g/dL; interquartile range [IQR], 3.8-4.5 g/dL), and only 20 (12.8%) of 156 presented with hypoalbuminemia (albumin, <3.5 g/dL). Two weeks after treatment with IVIG was started, serum albumin levels decreased compared with levels before treatment (median, 3.7 g/dL; IQR, 3.2-4.1 g/dL; P < .001), and a larger number of patients were hypoalbuminemic (60 [34.5%] of 174; P < .001). After 3 months, serum albumin levels returned to reference range values in all patients except in 2 (2.0%) of 101 patients with persistent hypoalbuminemia and 2 (2.0%) of 101 patients with hyperalbuminemia. At 6 months after treatment, the albumin levels were higher when compared with the pretreatment level (median serum albumin, 4.5 g/dL; IQR, 43-48 g/dL; P < .001). There was no influence of concomitant methylprednisolone treatment on serum albumin levels (median albumin, 3.7 g/dL in the IVIG-only group vs 3.65 g/dL in the IVIG and methylprednisolone group; P = .82, Mann-Whitney test).
After IVIG, the IgG level increased,6 but the serum albumin level decreased (Figure 1A and B). This ΔIgG correlated, albeit weakly, with Δalbumin (Figure 1C). Rather than a negative correlation, in which increasing levels of IgG reduce the serum albumin concentration, a positive correlation was observed. This somewhat unexpected result also became apparent from the correlation between the IgG and albumin serum levels at 2 weeks (Figure 1D). Pretreatment levels of serum albumin or IgG do not seem to profoundly influence posttreatment levels (eg, pretreatment serum albumin correlated weakly with ΔIgG: ρ = 0.23, P = .003; pretreatment IgG with Δalbumin: ρ = −0.11, P = .18). Other immunoglobulins (IgA and IgM) in a subgroup of 46 patients did not change significantly 2 weeks after IVIG compared with levels before treatment (mean pretreatment and 2-week posttreatment IgA, 310 and 320 mg/dL, respectively; P = .17, paired-samples t test; mean pretreatment and 2-week posttreatment IgM, 230 and 250 mg/dL, respectively, P = .07, paired-samples t test [to convert IgA and IgM to grams per liter, multiply by 0.01]).
To assess the potential of albumin as a biomarker for disease severity and outcome, patients were grouped according to their serum albumin levels. Patients were allocated to 1 of 2 groups (clinical hypoalbuminemia present or absent) or divided into tertiles (of approximately equal number) based on the raw data to maximize the value of information (also within the reference range values of albumin). This categorization based on serum albumin levels was performed both before and 2 weeks after IVIG.
The groups based on pretreatment serum albumin levels significantly differed in respiratory failure and the MRC sum score at nadir (Table 2). Patients with low serum albumin levels after treatment also required respiratory support more frequently and had a poorer GBS disability score and MRC sum score throughout and at the end of follow-up (all P < .001). Dividing the patients based on hypoalbuminemia vs normoalbuminemia yielded comparable results (ie, a higher serum albumin level was associated with a better clinical course and reduced disease severity at the end of follow-up [defined by the same clinical grading factors given in Table 2]). Intensive care unit (ICU) admission and mechanical ventilation could potentially influence the serum albumin levels. Therefore, we also analyzed the MRC sum score and GBS disability score at the end of follow-up, including mechanical ventilation and GBS disability score at nadir as covariates (mean [SD] GBS disability scores, 2.01 [1.5] for the <3.5-g/dL group, 1.2 [1.1] for the 3.5- to 4.0-g/dL group, and 0.7 [0.6] for the >4.0-g/dL group, P = .004; mean [SD] MRC sum scores, 48.7 [13.2] for the <3.5-g/dL group, 56.5 [6.1] for the 3.5- to 4.0-g/dL group, and 58.3 [4.3] for the >4.0-g/dL group, P = .01, ANCOVA). In addition, patients not requiring ICU admission or mechanical ventilation also had poor outcomes on the basis of serum albumin levels 2 weeks after starting IVIG treatment (mean [SD] GBS disability scores, 2.1 [1.7] for the <3.5-g/dL group, 1.1 [1.1] for the 3.5- to 4.0-g/dL group, and 0.6 [0.6] for the >4.0-g/dL group, P < .001; mean [SD] MRC sum score, 51.0 [11.2] for the <3.5-g/dL group, 57.5 [5.6] for the 3.5- to 4.0-g/dL group, and 58.3 [4.4] for the >4.0-g/dL group, P = .002, ANCOVA). Survival analysis (Kaplan-Meier) with respect to the time required to improve 1 point on the GBS disability scale or the achievement of walking more than 10 m unaided during the trial follow-up (6 months) was not significantly different for patients based on their pretreatment albumin levels (eFigure 1 in the Supplement). Patients who maintained a serum albumin level within low-normal or high-normal range after treatment recuperated faster than hypoalbuminemic patients (6 of 114 patients [5.3%] vs 21 of 60 [35.0%] were unable to walk unaided; log-rank test for trend, P < .001) (Figure 2). Serum albumin levels decreased with age, as denoted by the negative correlation between posttreatment level and age (ρ = −0.406, P < .001) and the higher age in the hypoalbuminemic group (Table 2). Stratifying for age (groupwise as given in Table 1) revealed similar results, although the negative effect on improvement of a low serum albumin concentration is more pronounced in elderly populations (>60 years of age) (eFigure 2 in the Supplement). When adjusted for age (Cox proportional hazards regression), posttreatment serum albumin levels remained a highly significant factor associated with regaining the ability to walk unaided (P < .001). Multivariate analysis with clinical predictive factors (Table 2) in prognostic models for GBS identified pretreatment and posttreatment serum albumin as independent factors. The addition of serum albumin to the 3 models improved the predictive capability in this cohort of patients, as expressed in the area under the curve when compared with models without the incorporation of serum albumin levels. The area under the curve increased from 0.83 to 0.85 for the EGRIS, from 0.91 to 0.92 for the mEGOS at 4 weeks, and from 0.83 to 0.85 for the mEGOS at 6 months.
We correlated the serum levels of albumin in 174 patients with GBS before and after standard high-dose IVIG therapy to clinical recovery. Before, but more evidently 2 weeks after, the start of IVIG treatment, serum albumin levels were significantly decreased, and low serum albumin levels were associated with a poorer clinical outcome. Subsequent logistic regression analysis identified albumin as an independent factor associated with outcome.
Clinical prognostic models have been developed previously to estimate the chance of respiratory failure (EGRIS) and disability at 1, 3, and 6 months (EGOS/mEGOS).16-18 Given the acute onset of GBS, biomarkers should ideally give an indication of outcome as early as possible to provide optimal medical care. Low albumin levels in serum obtained before treatment revealed a limited but potentially relevant association with poor outcome, especially with respect to the chance of respiratory failure. The pretreatment albumin levels were associated with respiratory failure independent of the clinical factors in the EGRIS model. Two weeks after IVIG was started, the variability of the serum albumin levels was most pronounced, with one-third of the patients having developed hypoalbuminemia (albumin level, <3.5 g/dL). These patients had a poor clinical outcome compared with patients with normal albumin levels. In addition, in the normoalbuminemic group, patients with low-normal levels (defined as 3.5-4.0 g/dL) had worse outcomes compared with patients with high-normal values (defined as >4.0 g/dL). Multiple regression analysis revealed that serum albumin levels at 2 weeks after treatment were an independent factor associated with respiratory failure and inability to walk at 3 and 6 months, improving the capability of the EGRIS and mEGOS for anticipating these outcomes.
In a healthy individual, serum albumin levels are kept within a well-defined reference range. The main causes of a reduction in serum albumin are increased catabolism, decreased production, and extravasation attributable to increased capillary permeability in the setting of inflammation or severe disease.13,21,22 All 3 causes may contribute to the observed reduced serum albumin levels in GBS. In addition, high-dose IVIG treatment in disorders other than GBS is related to a reduction of the serum albumin level.7,8,23 This effect may be caused by exhaustion of the albumin and IgG recycling pathway via FcRn that binds both proteins. In the present study, however, no association was found between an increase in serum IgG levels after IVIG and a decrease in albumin levels. This finding may indicate that there is no direct competition for binding to the FcRn, and previous studies24-26 found that human FcRn has distinct binding sites for IgG and albumin. Even in the absence of direct competition, serum albumin levels might reflect an individual’s recycling capacity (eg, expression levels of FcRn) and be an indicator of IVIG pharmacokinetics, as has been speculated before.8,27,28
Aside from a potential association with IVIG pharmacokinetics, albumin has been explored as a marker for prognosis in numerous other diseases and is a well-known indicator of general health.13,29 Moreover, a low serum albumin level is a strong marker of poor outcome in the setting of acute illness.30,31 Studies32-34 focusing on ICU and critically ill patients identified serum albumin as a biomarker for survival and the need for mechanical ventilation. In a previous study11 on biomarkers in amyotrophic lateral sclerosis, patients’ survival increased with higher serum albumin levels, even within the reference range.
Serum albumin levels before treatment and 2 weeks after IVIG treatment may be an additional prognostic factor in GBS. Our findings should now be validated in prospective studies, preferably with greater numbers of patients. This study did not seek to compare the prognostic capabilities of serum albumin vs the previously identified ΔIgG.6 Nonetheless, determining the serum albumin concentration once is more efficient than calculating the change in IgG over time. We also assessed Δalbumin, which gave comparable results. Hence, only the pretreatment or posttreatment levels were required for final analysis. The latter limits, of course, the practical use in GBS given the often rapid disease development that calls for early intervention. In our analyses regarding the prognostic models, we did not address the potential problem of overfitting, and no independent cohort of patients was available to validate our findings. Admission to the ICU could be an important factor in the reduction of serum albumin levels (eg, fluid therapy or mechanical ventilation), but patients who have not been admitted to the ICU also had reduced serum albumin levels of prognostic value to clinical outcome. Finally, we cannot rule out potential bias caused by the recumbence of patients, which is known to expand the plasma volume and thereby lower serum protein levels. However, this effect seems unlikely because levels of other serum proteins (IgA and IgM) did not decrease.
This study found that serum albumin, often already in use as a routine diagnostic indicator of overall health (eg, comprehensive metabolic panel), is an independent factor associated with the short- and long-term prognosis of patients with GBS treated with IVIG. The most auspicious finding is the prognostic value of pretreatment levels for the need for mechanical ventilation. Prospective studies should verify these findings to confirm the benefit of serum albumin as a biomarker for prognosis of GBS in clinical practice.
Corresponding Author: Bart C. Jacobs, MD, PhD, Departments of Neurology and Immunology, Erasmus MC, University Medical Center Rotterdam, PO Box 2040, 3000 CA Rotterdam, the Netherlands (firstname.lastname@example.org).
Accepted for Publication: September 13, 2016.
Published Online: December 27, 2016. doi:10.1001/jamaneurol.2016.4480
Author Contributions: Dr Fokkink had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Fokkink, Jacobs.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: Fokkink, Jacobs.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Fokkink, Walgaard.
Obtained funding: van Doorn, Jacobs.
Administrative, technical, or material support: Fokkink, Tio-Gillen.
Study supervision: Fokkink, Kuitwaard, van Doorn, Jacobs.
Conflict of Interest Disclosures: Dr Kuitwaard reported receiving unrestricted research grants from Baxalta and Grifols. Dr Jacobs reported receiving grants from the GBS-CIDP Foundation International, Baxalta, CSL-Behring, and Grifols. Dr van Doorn reported receiving grants from Sanquin Blood Supply, Baxalta, and Grifols. No other disclosures were reported.
Funding/Support: This work was supported by grant W.OR11-27 from the Prinses Beatrix Spierfonds.
Role of the Funder/Sponsor: The funding source had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and the decision to submit the manuscript for publication.