Therapeutic plasma exchange in patients with sepsis: Secondary analysis of a cluster‐randomized controlled trial

Abstract

Introduction

Sepsis is life‐threatening organ dysfunction caused by infection‐related inflammatory response. Therapeutic plasma exchange (TPE) can remove inflammatory mediators and benefit patients in different disease settings. However, no solid evidence showed the efficacy and safety of TPE in sepsis.

Methods

This study was a secondary analysis of a randomized controlled trial. Critically ill patients with sepsis were divided into two groups according to whether treated with TPE. The primary outcome was the delta Sequential Organ Failure Assessment (SOFA) score from days 1 to 7. Secondary outcomes included new‐onset organ failure, intensive care unit (ICU)‐free and alive days to day 28, and 28‐day mortality. Propensity score‐matched (PSM) analysis was applied to control confounders. Analysis of covariance (ANCOVA) and logistic regression were used to assess the association between TPE and selected outcomes.

Results

Among the 2772 critically ill patients enrolled in the trial, 742 patients with sepsis were selected and 22 patients received TPE were matched with 22 control patients. No significant difference was found in the delta SOFA score and 28‐day mortality between TPE group and control group. The ICU‐free and alive days in the TPE group were significantly shorter than the control group.

Conclusions

TPE may be not associated with improvement of organ failure and mortality in critically ill patients with sepsis and may be associated with a prolonged ICU stay.

Abbreviations

APACHE II

Acute Physiology and Chronic Health Evaluation II score

ASFA

American Society for Apheresis

BMI

body mass index

CI

confidence interval

HR

hazard ratio

ICU

intensive care unit

PSM

propensity score‐matched

RCT

randomized controlled trial

SD

standard deviation

SMD

standardized mean difference

SOFA

Sequential Organ Failure Assessment score

TPE

therapeutic plasma exchange

1. INTRODUCTION

Sepsis is defined as life‐threatening organ dysfunction caused by a dysregulated host response to infection. ¹ It is a major cause of mortality and morbidity globally and affects approximately 50 million people per year. ² With an improved understanding of the pathophysiology of sepsis, it is realized a combination of pro‐inflammatory and anti‐inflammatory reactions played an important role in the evolution of sepsis. ³ However, specific interventions targeting the inflammatory reactions are still lacking.

Therapeutic plasma exchange (TPE) is an extracorporeal blood purification technique for removing large molecular weight substances from the plasma. TPE is supposed to improve organ function during sepsis by removing inflammatory mediators. ⁴ However, there was limited evidence regarding the effect of plasma exchange in adult patients with sepsis since most studies in the literature were case reports, pediatric studies and observational studies. ⁵ , ⁶ , ⁷ A meta‐analysis included randomized controlled trials (RCT) comparing plasma exchange with usual care in critically ill patients with sepsis or septic shock found that plasma exchange was associated with reduced mortality in adults. No outcomes about the intensive care unit (ICU) or hospital lengths of stay were reported. ⁸ An RCT conducted in adult patients was published in 2002, targeting severe sepsis and septic shock patients based on the SEPSIS 1.0 definitions. ⁹ After adjusting for age and site of infection, it demonstrated that TPE was not associated with a reduction in mortality (P = .07). Another RCT was conducted in patients with septic shock. The result found that TPE could lead to rapid hemodynamic improvement but could not improve mortality and organ function. ¹⁰ Given the paucity of evidence in the literature, the 2019 American Society for Apheresis (ASFA) guidelines stated that “the optimum role of apheresis therapy is not established; decision making should be individualized” and gave a weak recommendation. ⁴

In this study, we aimed to evaluate the impact of TPE on clinical outcomes in a cohort of critically ill sepsis patients collected from a large cluster‐randomized trial.

2. PATIENTS AND METHODS

2.1. Study design and patients

We conducted a secondary analysis of the Actively Implementing an Evidence‐based Feeding Guideline for Critically Ill Patients (NEED trial). ¹¹ The trial is a multicenter, cluster‐randomized, parallel‐controlled trial conducted in 97 intensive care units (ICU) in China between March 2018 and July 2019. It enrolled adult patients admitted to ICU less than 24 hours, had one or more organ system failures, were expected to stay in ICU for more than 7 days, and were judged not likely to resume oral diet within 3 days. Patients who received EN in the past 3 days, were receiving palliative treatment, were expected to die within 48 hours, were pregnant, had a long‐term history of steroid use or other immunosuppressive agents, or were receiving radiotherapy or chemotherapy due to malignant diseases were excluded from the trial. The primary outcome of NEED trial was all‐cause mortality to 28 days since enrollment, and the secondary outcomes included process measures of guideline uptake, organ failure‐related outcomes and corresponding therapies, ICU‐free and alive days to 28 days, and the incidence of new infections. The detailed results of the NEED trial were published recently. ¹¹

In the present study, we collected a cohort of patients with sepsis from the NEED trial. The diagnosis of sepsis was made when the patient (a) had at least one confirmed infection site, (b) had one or more organ system failures (Sequential Organ Failure Assessment [SOFA] score for any individual organ system ≥2). Selected patients were dichotomized depending on whether they received TPE during the first 72 hours of ICU admission.

2.2. Data collection

For each selected patient, data including baseline characteristics, daily organ failure scores, daily laboratory testing, and prognosis were extracted from the electronic database. The baseline characteristics included age, sex, body mass index (BMI), ICU type, Acute Physiology and Chronic Health Evaluation II (APACHE II) score, SOFA score at enrollment. Organ failure assessment was censored until 7 days after enrollment, ICU discharge, or death, whichever occurred first. Laboratory testing included arterial blood gas results like arterial partial pressure of oxygen and arterial partial pressure of carbon dioxide and inflammatory biomarkers like white blood cell and C‐reactive protein.

2.3. Study outcomes and definition

The primary outcome was the delta SOFA score from days 1 to 7. The secondary outcomes included new‐onset organ failure (including respiratory failure, cardiovascular failure, and renal failure) to day 7, 28‐day mortality, and ICU‐free and alive days to day 28. Delta SOFA score was defined as the SOFA score on day 7 after enrollment minus the baseline SOFA score. Alive and ICU discharge was counted as a SOFA score of 0, and death was counted as a maximum SOFA score of 24. ¹² The new‐onset organ failure was defined as organ failure that was not present at enrollment. ICU‐free and alive days to day 28 were defined as the number of days alive and not admitted to an ICU after the patient's final discharge from the ICU before day 28. If the patients were admitted to an ICU on day 28 or died prior to day 28, ICU‐free and alive days would be 0.

2.4. Statistical analysis

Descriptive statistics were used to characterize the study population. The normality of the data was tested using the Shapiro‐Wilk test. Continuous data were reported as means with SDs. Categorical data were reported as frequencies and percentages. Baseline comparison was conducted using Student t test or Wilcoxon rank‐sum test for continuous variables depending on their normality and chi‐square test or fisher exact test for categorical variables.

To create maximally comparable groups regarding baseline characteristics, we performed a propensity score‐matched (PSM) analysis. Patients who received TPE were matched 1:1 with patients who did not, using their propensity score. The propensity score was estimated based on age, sex, BMI, ICU type, and baseline APACHE II score. Subsequently, genetic matching without replacement was performed within callipers 0.15 SDs. We used standardized mean difference (SMD) to assess the balance of baseline covariates between treatment arms in the matched sample with that in the unmatched sample. An SMD of more than 0.1 and a two‐sided P‐value of less than .05 indicated a meaningful imbalance in the baseline covariate.

Analysis of covariance (ANCOVA) was used for continuous outcomes presented with mean difference. Logistic regression was used to estimate OR with 95% confidence intervals for binary outcomes, and logistic regression with penalized‐likelihood estimates was used for variables with one cell having zero frequency. We used treatment effect to represent mean difference of continuous outcomes and OR of binary outcomes. Crude treatment effect represented relative treatment effect, and adjusted treatment effect represented treatment effect adjusted for age, sex, BMI, ICU type, APACHE II score. In addition, the time to discharge alive from ICU was assessed in a competing risk model with early death treated as a competing factor and reported as subdistribution hazard ratios with 95% CIs estimated from a Fine‐Gray model.

Two‐sided P‐values of less than .05 were considered to indicate statistical significance. Data were analyzed using the R software version 4.1.0 and SAS software version 9.4.

3. RESULTS

3.1. Study population

Among 2772 patients enrolled in the NEED trial, 742 patients met the diagnostic criteria of SEPSIS 3.0. Six patients were excluded due to loss of follow‐up. In total, 22 patients received TPE during the first 72 hours of ICU stay, and 714 did not. After matching, 22 patients in the TPE group were matched with 22 patients in the control group (Figure 1).

FIGURE 1.

Flowchart of patient selection

3.2. Propensity score‐matched analysis

Baseline characteristics for the unmatched and matched sample are presented in Table 1. Before matching, age, baseline APACHE II score, and baseline cardiovascular failure were significantly different between groups, while the matched sample was well balanced for all baseline characteristics. And SMDs comparing baseline covariates in the unmatched and matched sample were reported in Table S1 and Figure 2.

TABLE 1.

Baseline characteristics before and after propensity score matching of patients

Characteristics	Before matching			After matching
	TPE (n = 22)	Control (n = 714)	P	TPE (n = 22)	Control (n = 22)	P
Age	45.27 (12.34)	61.07 (18.25)	.0000	45.27 (12.34)	45.41 (12.04)	.9706
BMI	24.12 (3.73)	22.65 (3.09)	.0810	24.12 (3.73)	24.09 (3.76)	.9780
Sex			.0616			1.0000
Male, n (%)	12 (54.55)	251 (35.15)		12 (54.55)	12 (54.55)
Female, n (%)	10 (45.45)	463 (64.85)		10 (45.45)	10 (45.45)
ICU type			.3966			1.0000
General, n (%)	20 (90.91)	586 (82.07)		20 (90.91)	20 (90.91)
Special, n (%)	2 (9.09)	128 (17.93)		2 (9.09)	2 (9.09)
Baseline
APACHE II	15.14 (6.95)	19.04 (7.18)	.0165	15.14 (6.95)	15.09 (6.54)	.9823
SOFA	7.50 (3.85)	8.09 (3.61)	.4829	7.50 (3.85)	7.36 (2.95)	.8958
Organ failure
Respiratory, n (%)	19 (86.36)	549 (76.89)	.4393	19 (86.36)	19 (86.36)	1.0000
Cardiovascular, n (%)	2 (9.09)	249 (34.87)	.0109	2 (9.09)	8 (36.36)	.0689
Renal, n (%)	7 (31.82)	194 (27.17)	.6299	7 (31.82)	4 (18.18)	.4876

Note: Continuous data are shown as mean (SD) and categorical data are shown as n (%). Comparisons were performed with the use of chi‐square test or fisher exact test for categorical variables and Student t test for continuous variables.

Abbreviations: APACHE, acute physiology and chronic health evaluation; BMI, body mass index; ICU, intensive care unit; SOFA, Sequential Organ Failure Assessment.

FIGURE 2.

Standardized mean differences in unmatched and matched cohorts

3.3. Outcomes

Clinical outcomes are presented in Table 2. SOFA score on days 1 and 7 were available in all the study patients. The mean delta SOFA score was‐1.8 (SD, 2.6) in the TPE group and −1.7 (SD, 3.8) in the control group (mean difference, −0.05 [95% CI, −1.91, 2.00], P = .963). For new‐onset organ failure to day 7, there was two (9.1%) patient developed new‐onset respiratory failure in the TPE group, compared with one (4.6%) patient in the control group (OR, 0.48 [95% CI, 0.04, 5.67], P = .557). There was no significant difference in 28‐day mortality (OR, 0.63 [95% CI, 0.10, 4.22], P = .637).

TABLE 2.

Clinical outcomes between TPE group and control group

Outcomes	TPE (n = 22)	Control (n = 22)	Treatment effect (95% CI)	P
Primary outcome
ΔSOFA	−1.8 (2.6)	−1.7 (3.8)	0.05 (−1.91, 2.00) a	.963
Secondary outcome
New‐onset organ failure to day 7
Respiratory, n (%)	2 (9.1)	1 (4.6)	0.48 (0.04–5.67) b	.557
Cardiovascular, n (%)	2 (9.1)	0	0.18 (0.01–4.31) b	.292
Renal, n (%)	5 (22.7)	2 (9.1)	0.34 (0.06–1.98) b	.230
ICU‐free days to day 28	8.1 (7.9)	12.9 (8.0)	4.81 (0.01–9.62) a	.049
Mortality, n (%)	3 (13.6)	2 (9.1)	0.63 (0.10–4.22) b	.637

Note: Continuous data are shown as mean (SD) and categorical data are shown as n (%). ΔSOFA: SOFA on day 7‐SOFA on day 1.

Abbreviations: APACHE, acute physiology and chronic health evaluation; CI, confidence interval; ICU, intensive care unit; SOFA, Sequential Organ Failure Assessment.

^a Treatment effect is mean difference (continuous outcomes).

^b Treatment effect is odd ratios (dichotomous outcomes).

A significant reduction of ICU‐free and alive days was observed in the TPE group (8.1 days [SD, 8.0] vs 12.9 days [SD, 8.0]) (mean difference, 4.81 [95% CI, 0.01, 9.62], P = .049). However, as shown in Figure 3, using the Fine‐Gray model, no difference was found in the time to discharge alive from ICU between the two groups (HR, 0.785[95% CI, 0.434, 1.419], P = .34).

FIGURE 3.

Cumulative incidence Estimates for Patients in the TPE and control groups

3.4. TPE characteristics and immediate effects of TPE on clinical parameters

The application of TPE was decided by the treating physician based on the clinical assessment. The 22 patients in the TPE group received a median (interquartile) of 2.5 (1‐4.25) sessions of TPE, among whom eight patients received only once and 14 patients received repeated sessions. All the sessions were membrane TPE (filtration), and plasma and albumin were used for replacement. However, the detailed volume was not available. Moreover, heparin (low molecular weight heparin included) was the primary choice for anticoagulation. We evaluate the immediate effects of each TPE session (Table 3). No significant differences were observed in each TPE session.

TABLE 3.

Changes in clinical parameters after all TPE treatment

Variable	All treatment (n = 79)
	Before	After	P
Arterial blood gas
PH	7.38 (0.07)	7.4 (0.06)	.245
PaO2 (mmHg)	87.8 (36.2)	99.6 (34.2)	.068
PaCO2 (mmHg)	41.4 (14.2)	41.5 (12.6)	.779
Inflammatory biomarkers
WBC	13.2 (6.3)	15.6 (7.6)	.663
CRP (mg/L)	88.5 (36.3)	76.3 (45.9)	.345

Note: Values are shown as mean (SD).

Abbreviations: CRP, C‐reactive protein; FiO₂, fraction of inspired oxygen; PaCO₂, arterial partial pressure of carbon dioxide; PaO₂, arterial partial pressure of oxygen.

3.5. Sensitivity analysis

Outcomes between groups before matching were reported in Table S2. After adjusting for age, sex, BMI, ICU type, APACHE II, the TPE group had a higher incidence of new‐onset renal failure (22.7% vs 6.4%, OR, 0.2 [95% CI, 0.1, 0.7], P = .009) and a reduced ICU‐free and alive days to day 28 (8.1 [7.8] vs 11.3 [8.6], mean difference, −3.2 [95% CI, −6.9, 0.5], P = .03). No significant differences were observed in other outcomes.

4. DISCUSSION

This secondary analysis found that TPE might not be associated with improvement of organ failure in a short term (7 days) and 28‐day mortality in patients with sepsis, and it might prolong ICU stay in critically ill patients with sepsis. However, in the time to discharge alive from ICU analysis, which took early death as a competing factor, no significant difference was found between groups.

The discrepancy we observed should be attributed to methodological issues. ICU‐free and alive days is a composite outcome where patients who died before 28 days would be counted as 0, while in the competing risk model, death would be counted as a competing factor in the time‐to‐event analysis. All free days endpoints have natural pros and cons. ¹³ It combines the duration of critical care for survivors with mortality, and smaller sample sizes are therefore required for intervention aiming to reduce ICU duration and mortality. However, assigning different weightings to the two components (extended ICU stay and death) of ICU‐free days is problematic. Thus the clinical implication of ICU‐free days should be cautiously interpreted. For competing risk model, the percentage of competing events might influence the outcomes of competing risk models. In general, absolute percentages of competing events of >10% might contribute to bias when treating competing events as censoring events. However, absolute percentages of competing events of <10% might lead to false positive or false negative. ¹⁴ , ¹⁵ Taken together, the impact of TPE on ICU stay seems marginal, and larger randomized controlled studies are needed. Besides, TPE was an invasive treatment requiring central venous access and had risks of adverse events such as deep veno. For organ failure, the results of this study showed that the incidence of new‐onset respiratory failure, new‐onset cardiovascular failure and new‐onset renal failure were similar between groups, which suggested that TPE may not impact the evolution of organ failure.

The potential clinical benefits TPE mainly rest on removing inflammation mediators and replenishing the immunoglobulins. ¹⁶ Regarding biomarkers of inflammation, a previous study showed significant reductions in plasma concentrations of C‐reactive protein, α1 antitrypsin, haptoglobin, and complement component C3 after TPE. ¹⁷ TPE was also reported to lower circulating levels of endotoxin and cytokines such as tumor necrosis factor α and interleukin 1β in 20 patients. ¹⁸ , ¹⁹ , ²⁰ , ²¹ , ²² Moreover, TPE was associated with alterations in the immune system, including (a) Changes in lymphocyte numbers and distribution; (b) Changes in natural killer cell numbers and activity; (c) Changes in T suppressor function; (d) Alterations in the Th1/Th2 ratio. ²³ However, no inflammatory biomarkers changed significantly after TPE treatment in this study, which might be due to the few markers we collected during the trial. Besides, the time of blood sample collection was on a daily basis instead of immediately before and after TPE.

TPE has been suggested as a possible treatment for sepsis with multi‐organ failure not responsive to standard therapy. ²⁴ However, the available data regarding the clinical value of TPE in sepsis are poor. So far, only three RCTs have assessed the effect of TPE in adult patients with sepsis. The first enrolled 30 patients, of whom eight are children, and found no significant difference in mortality between the TPE and control groups. ¹⁷ The second randomized 106 patients and found significantly improved mortality after TPE treatment (33.3% vs 53.8%), but the difference was no longer significant after adjusting for age and site of infection. ⁹ Moreover, given that the two RCTs were conducted in 1999 and 2002, the diagnostic criteria of sepsis had significantly evolved, which would inevitably limit the generalizability of the finding in the era of SEPSIS 3.0. ²⁵ The third enroll 40 patients with septic shock, and found improvement in hemodynamics. However, its sample size was small, and no difference was found in mortality or organ function between groups. ¹⁰

There were several limitations of this study. First, this is a retrospective analysis of data from a large RCT, thus confounders cannot be excluded. PSM was applied to take the confounders into account, but bias may still exist. Second, technical details regarding the TPE sessions were not fully collected during the trial, and technical differences may impact the efficacy of TPE. Third, SOFA score was only recorded for the first 7 days, making long‐term assessment impossible. Although a long‐term effect of TPE was unlikely, we could not exclude the possibility.

5. CONCLUSION

The results of this study suggest that TPE may not be associated with improvement of organ failure and mortality in critically ill patients with sepsis, and may prolong ICU stay. A well‐designed randomized control trial with large sample sizes is needed to confirm our findings.

AUTHOR CONTRIBUTIONS

Concept and design: Lingyan Fu. Acquisition, analysis, or interpretation of data: Xiang Luo and Xiaoling Li. Drafting of the manuscript: Xiang Luo and Xiaoyan Lai. Critical revision of the manuscript for important intellectual content: Lu Ke, Jing Zhou, Man Liu, and Longxiang Cao. Obtained funding: Xiang Luo. All authors read and approved the final manuscript.

CONFLICT OF INTEREST

The authors declare that they have no competing interests.

ETHICS STATEMENT

This study was a secondary analysis of a cluster‐RCT (ISRCTN registry: ISRCTN12233792), which was approved by the ethics committee of Jinling Hospital (22017NZKY‐019‐02).

Supporting information

Table S1. SMD of covariates between unmatched and matched groups.

Table S2.Clinical outcomes between TPE group and control group in unmatched cohorts.

Luo X, Li X, Lai X, et al. Therapeutic plasma exchange in patients with sepsis: Secondary analysis of a cluster‐randomized controlled trial. J Clin Apher. 2023;38(1):55‐62. doi: 10.1002/jca.22027

DATA AVAILABILITY STATEMENT

The data that support the findings of this study are available from the corresponding author upon reasonable request.