Improved Survival beyond 28 Days up to 1 Year after CytoSorb Treatment for Refractory Septic Shock: A Propensity-Weighted Retrospective Survival Analysis

Worldwide, approximately 45 million patients suffer from sepsis with an estimated 11 million deaths each year [1-4]. Sepsis and septic shock are the leading causes of mortality with an amplified incidence due to increased aging of populations, comorbidities, and greater awareness of sepsis, but also because of emerging pandemics [5, 6]. Septic shock arises from a derailed host response to infection, with consequently a condition what is known as a “cytokine storm” leading to shock [6]. Sepsis is generally treated with early resuscitation and antibiotics, which is strongly associated with an improved outcome [5]. More innovative treatment options have been evaluated, most with disappointing results. Continuous renal replacement therapy (CRRT) proved able to clear cytokines from serum and possibly improve the cytokine storm; however, a clinical benefit seems to be lacking [7-9]. Recently, we have shown that CytoSorb treatment may lead to an improved clinical outcome at 28 days from intensive care unit (ICU) admission for refractory septic shock [10]. CytoSorb has been developed and approved for use in patients with severe cytokinemia [11-13]. It is an adsorber which can be used in addition to CRRT and has been shown to be able to reduce circulating cytokines by >90% [13]. It is currently not known whether CytoSorb therapy for refractory septic shock is associated with an improved survival beyond the first month of ICU admission. Moreover, factors determining (un)favourable outcomes and the optimal timing of CytoSorb therapy are currently not known. Therefore, the aim of the current study was to investigate the long-term outcome of the septic shock patients we had previously investigated for 28-day survival who were treated with CytoSorb and/or CRRT and to study factors associated with outcome in CytoSorb-treated patients.

This is a long-term follow-up of a retrospective cohort study, in which patients were treated at the ICU for septic shock with CytoSorb and/or CRRT. Here, we further studied the cohort of patients we had previously studied for 28-day survival. The selection methods and inclusion and exclusion criteria are described elsewhere in detail [10]. In short, patients admitted to the ICU with septic shock who were treated with CytoSorb or CRRT in the period from January 1, 2014, to April 1, 2017, were eligible for inclusion in the study. These patients were treated at the discretion of the attending intensive care physician according to local protocol. Patients were excluded from the analysis in case the primary reason for therapy was not septic shock. In case of improvement of septic shock or renal function, CytoSorb or CRRT was discontinued. CytoSorb was used according to the manufacturer’s protocol. It was placed in a blood-pump circuit with an optimal ultrafiltration rate of 250–400 mL/min. The CytoSorb adsorber was changed after 24 hours of use.

Septic shock was defined as persisting hypotension requiring vasopressive medication to maintain mean arterial pressure (MAP) ≥65 mmHg and having a serum lactate level of >2 mmol/L despite adequate volume resuscitation [6]. Shock reversal was defined as a serum lactate level of ≤2 mmol/L and discontinuation of vasopressive medication [14].

The primary endpoint was the time to all-cause mortality compared for CytoSorb versus CRRT alone. Also, factors associated with all-cause mortality during long-term follow-up were analyzed separately. All-cause mortality was measured from ICU admission until the last follow-up date known (irrespective of ICU, in-hospital, or out-of-hospital mortality).

A survival analysis was employed with application of the inverse probability of treatment weights (IPTW), including a stabilizing method to avoid bias from extreme weights. Stabilized IPTW (sIPTW) is applied to overcome differences in baseline patient characteristics, to mimic a randomized controlled trial [15]. The sIPTW was previously constructed [10] and contains the following variables at the start of therapy: age, presence of any comorbidity, SOFA score, lactate level, noradrenalin level, type of sepsis, and medical or surgical treatment. Since mean arterial pressure (MAP) at the start of therapy was at imbalance between the 2 treatment groups and associated with outcome, it was additionally accounted for in the final analysis [10, 15]. Factors associated with all-cause mortality in CytoSorb-treated patients were analyzed as well without application of sIPTW. For this, factors with a p value of <0.2 in univariate Cox regression analysis were considered for multivariate analyses, where a full, forced and final model was constructed. Using the same variables from the full model, the final model was constructed using the back step and forward step likelihood ratio method, where the model with the highest χ² value was selected. Moreover, factors determining outcome were selected and inspected for optimal clinical use by analyzing tests’ characteristics with Harrell’s C-statistic and for laboratory values choosing the optimal cutoff using the highest sensitivity and specificity sum. Laboratory values and variables with a skewed normal distribution were log-transformed prior to the analysis. SAS Enterprise version 9.4 and SPSS version 22.0 (SPSS Inc., Chicago, IL, USA) were used to perform the statistical analyses. All statistical tests were two-sided and evaluated at the 0.05 level of significance.

In total, 67 patients with septic shock were treated with CytoSorb, versus 49 with CRRT. Table 1 shows the patient characteristics at the start of therapy [10]. Patients were followed for a median duration of 30 days (interquartile range [IQR]: 5–334) with a maximum of 1,059 days after ICU admission. For those patients followed beyond 28 days after ICU admission (n = 59), the median follow-up duration was 333 days (IQR: 170–583). There was no difference in median follow-up duration between the 2 treatment groups (p = 0.880). Median ICU and hospital admission were 9 and 17–22 days, respectively, for both groups (p = 0.735 and p = 0.991). The mean duration from ICU admission to the start of treatment was 2.1 (SE 0.36) versus 1.66 (SE 0.38) days (p = 0.416) for CRRT versus CytoSorb, respectively. In the CRRT group, CRRT was given for a mean duration of 4.96 (SE 0.63) days versus 4.97 (SE 0.55) days in the CytoSorb group (p = 0.990). The mean treatment duration for the CytoSorb group was 2.34 (SE 0.16) days.

For the total group, at 1 year after ICU admission, the unweighted survival rate was similar in both groups with CytoSorb-treated patients being 42% (standard error [SE] 6%) and CRRT-treated patients, 43% (SE 7%) (Fig. 1a), while CytoSorb-treated patients had a more severe shock as mentioned before. For those patients who survived beyond 28 days of ICU admission, survival for CytoSorb and CRRT at 1 year after ICU admission was comparable at 80% (SE 7%) and 87% (SE 7%), respectively (p = 0.853).

By Kaplan-Meier analysis applying sIPTW, the 1-year survival was 37% (SE 7%) versus 26% (SE 12%) for CytoSorb versus CRRT, respectively (log-rank p = 0.21). By Cox regression analysis adjusted for the MAP at baseline of therapy, CytoSorb treatment showed a significantly higher survival rate compared to CRRT (adjusted HR [aHR] 0.59, 95% Wald confidence interval [CI]: 0.37–0.93, p = 0.025; Fig. 1b). In this model, presence of chronic kidney disease (aHR 1.10, p = 0.736) or gender (aHR 1.02, p = 0.940) were not significantly associated with outcome.

Factors in univariable and multivariable analyses associated with outcome in the CytoSorb group can be found in Table 2. Noradrenaline levels did not significantly interfere with lactate levels (p = 0.496). However, there was evidence of collinearity for noradrenaline and lactate levels. When lactate was replaced by noradrenaline, the multivariate Cox regression model showed a slight weaker χ² and noradrenaline was not significantly associated with outcome (HR 2.86, 95% CI: 0.99–8.21, p = 0.051). The final model (Table 2) had a Harrell C-statistic of 0.80 (SE 0.04, p < 0.001) for mortality prediction in the long term, but also in the short term (28-day mortality C-statistic 0.81, SE 0.04, p < 0.001). For a simple and practical approach, the most important factor from this model was selected for mortality prediction: it was observed that a lactate level cutoff of 6.0 mmol/L (HR 3.1, SE 0.33, p < 0.001) had the highest specificity (79%) and sensitivity (58%) sum and positive predictive value (79%) for long-term mortality, Harrell C-statistic 0.67 (95% CI: 0.54–0.80, p = 0.013). Figure 2 shows the Kaplan-Meier curve for the CytoSorb cohort stratified by lactate below or above 6.0 mmol/L. This same cutoff of lactate level also had the highest sensitivity (63%) and specificity (77%) for 28-day mortality where the positive predictive value was slightly lower (71%).

In this long-term follow-up study, we have shown that long-term survival with either CytoSorb or CRRT beyond 28 days of ICU admission is achievable. In a weighted analysis, CytoSorb showed an improved long-term outcome compared to CRRT alone. In CytoSorb-treated patients, it was observed that age, lactate level at the start of therapy, presence of any comorbidity, and abdominal sepsis were factors associated with long-term and also short-term all-cause mortality. This is the first study to show that the benefit achieved with CytoSorb in the first month of ICU admission is enduring in the long term and that a lactate level above 6.0 mmol/L is an important determinant of outcome.

It is currently not known what the best timing for CytoSorb therapy in septic shock is. In the current study, we have shown that lower levels of lactate or noradrenaline were associated with an improved outcome. A lactate level above 6.0 mmol/L had a positive predictive value for long-term all-cause mortality of almost 80%. Indeed, this is a retrospective analysis and confounding by indication may have an impact on the current data. Nevertheless, it was observed that patients more often were treated with CytoSorb as a last-resort option when lactate and noradrenaline levels were already (too) high, as >50% of the CytoSorb-treated patients already had lactate levels above 6.0 mmol/L at the start of treatment. In line with our findings, a current report also showed that initiation of CytoSorb within 48 h of admission was associated with better outcome [16]. Given these lines of evidence, it seems reasonable to suggest that CytoSorb should be initiated earlier in the disease course rather than in case of refractory sepsis with high lactatemia. The timing should be further evaluated in future studies.

As was shown in our previous publication, before applying weights CytoSorb-treated patients were not comparable to those treated with CRRT alone at the start of therapy as they had worse hemodynamic characteristics including higher levels of lactate, need for vasopressors, lower MAP, and higher SOFA scores. Nonetheless, they still had a similar outcome to those treated with CRRT alone. Indeed patients treated with CRRT more often showed chronic kidney disease; however, this factor was not associated with long-term outcome in our analysis. When we applied IPTW adjusting for MAP, it was shown that CytoSorb-treated patients had an improved survival. Presence of chronic kidney disease, male gender, or other factors (peripheral artery disease or COPD) which were at imbalance did not show significant associations with outcome.

The current data show that, not unexpectedly, the first days of ICU admission were most important in the determination of clinical outcome. It is important to underline that the 28-day mortality rate showed to be a good surrogate for overall outcome as >80% of patients who survived up to 28 days after ICU admission also survived in the long term. There are some important limitations of the current study. Next to the retrospective nature of the study, we did not have inflammatory cytokines such as IL-6 or procalcitonin levels available. In addition, although this is currently the largest cohort study of septic shock patients treated with CytoSorb analyzed for mortality, the number of patients assessed is still relatively small and the results are partly obtained from a propensity score analysis. Moreover, we were not able to perform an analysis for fluid balances as this would entail a dynamic analysis and this falls outside the scope of the current article. The results obtained from this study should be further confirmed in randomized clinical trials.

In conclusion, we have shown that the survival achieved with both CRRT and CytoSorb for patients with septic shock leads to an improved outcome beyond 28 days of ICU admission and may be enhanced for CytoSorb treatment. Importantly, lactate levels above 6.0 mmol/L were predictive for mortality with a 79% specificity and 80% positive predictive value. Therefore, for patients with septic shock not responding to fluid therapy and with a lactate level below 6.0 mmol/L, CytoSorb therapy should be considered, especially for young patients with abdominal sepsis. Future randomized trials should take into account this lactate level since too high lactate levels are associated with poor treatment response.

All patients, their relatives, or legal representatives gave their informed consent for the admission to the ICU and to undergo specific interventions. Patients were not recontacted for the purpose of this study. Recontacting patients for informed consent for the current retrospective chart study would introduce a serious bias because a large proportion is out of follow-up or deceased. All procedures were carried out per standard of care per protocol. This retrospective cohort study was performed in accordance with the protocol, the principles of the Declaration of Helsinki 1964 as modified by the 64th WMA General Assembly, Fortaleza, Brazil, October 2013, with notes of clarification on paragraphs 29 and 30 added by the WMA General Assembly, and the local national laws governing the conduct of clinical research studies. This study was approved by the ethics committee of the Maasstad Hospital Rotterdam, the Netherlands (MZ0582-07-14 and L2017-030).

W.P.B. and S.D. have received speaker’s fee from CytoSorb. C.I. has received a grant from CytoSorb to commence a randomized controlled trial on the effect of the adsorber on the microcirculation of critically ill patients at the Department of Intensive Care of the Erasmus Medical Center, Rotterdam.

The authors did not receive any funding.

W.P.B.: study design, data collection, statistical methods and analysis, writing of the manuscript, and approval of the final version. S.D. and C.I.: study design, critical review of the manuscript, and approval of the final version. W.P.B., S.D., and C.I. had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.