Changes in Hemogram-Derived Ratios in Patients with Sepsis Undergoing Cytokine-Filtered Continuous Renal Replacement Therapy in the Intensive Care Unit: A Retrospective Single-Center Study

Introduction

Over the past two decades, hemogram parameters and derived ratios have attracted considerable interest in the literature, with numerous studies examining their utility for diagnosis, differential diagnosis, and prediction of disease severity, morbidity, and mortality [1–3]. The abundance of such studies has occasionally led to skepticism about their clinical value [4]. Nonetheless, research on hemogram parameters continues to be frequently published. Evidence remains limited concerning how these ratios change in patients with sepsis who undergo cytokine-filtered continuous renal replacement therapy (CRRT). Clarity regarding these changes is important for clinicians because they may affect the interpretation of laboratory results and guide therapeutic decisions in intensive care practice.

Hemodiafiltration is a CRRT modality widely used in the treatment of critically ill patients with renal injury, particularly those who are hemodynamically unstable [4,5]. When combined with CRRT, cytokine filtration is regarded as an effective strategy for reducing kidney injury and enhancing cytokine clearance, especially in patients with sepsis [6,7]. Recent evidence suggests that cytokine adsorption during CRRT influences not only cytokine elimination but also hematologic and inflammatory biomarkers in patients with sepsis [8,9]. Many studies have investigated the effectiveness of both CRRT and cytokine filtration; CRRT research has focused on hemofiltration efficiency, whereas cytokine filtration research has emphasized cytokine removal. Various aspects of CRRT have been explored, including initiation timing, treatment duration, anticoagulation strategies, solute clearance, and catheter placement [5,10,11].

Hemogram-derived ratios, calculated as proportions of routinely measured blood cell counts such as the neutrophil-to-lymphocyte ratio (NLR) and platelet-to-lymphocyte ratio (PLR), are increasingly recognized as simple, inexpensive, and informative markers of systemic inflammation [12]. However, these indices may also be influenced by factors such as obesity, smoking, and chronic inflammatory states, which can limit their specificity [13,14]. Alterations in hemogram parameters, particularly a decrease in platelet count during hemodiafiltration, have been reported and are well known by clinicians [15]. Moreover, not only absolute cell counts, percentages, and platelet indices obtained from complete blood counts but also derived ratios are widely used in disease diagnosis, treatment response evaluation, and mortality prediction, as previously mentioned [3,16,17]. To date, no study has specifically examined the effects of cytokine-filtered CRRT on these ratios in patients with sepsis or septic shock. This gap in the literature leaves uncertainty regarding whether observed changes in NLR or PLR reflect underlying disease severity, therapeutic interventions, or both. Therefore, we assumed that an investigation of CRRT with concomitant cytokine filtration on hemogram-derived ratios would be of clinical value. We hypothesized that cytokine-filtered CRRT alters these ratios, highlighting the importance of selecting appropriate time points for laboratory assessment.

The primary aim of this study was to evaluate changes in PLR from baseline through consecutive treatment days in patients with sepsis and septic shock undergoing cytokine-filtered CRRT. Secondary aims included the assessment of changes in other hemogram-derived ratios, particularly NLR, throughout the treatment period.

Material and Methods

ETHICS AND STUDY REGISTRATION:

Ethics approval was obtained from the Istinye University Clinical Research Ethics Committee (Date: 21.12.2022, Approval No: 3/2022-2.K-96). The requirement for patient consent was waived due to the retrospective nature of the study. Although retrospective, the study was registered at ClinicalTrials.gov (Identifier: NCT06948630; submission date: April 21, 2025) to ensure transparency and compliance with research standards. The study was conducted in accordance with the principles of the Declaration of Helsinki.

STUDY DESIGN AND SETTING:

Data were analyzed from patients admitted to the adult intensive care units (ICUs) of our hospital between January 2020 and December 2022 who underwent cytokine-filtered CRRT during follow-up and treatment.

PATIENT POPULATION:

Inclusion criteria were age ≥18 years, receipt of cytokine-filtered CRRT for sepsis or septic shock, and availability of both baseline and scheduled follow-up hemogram measurements. Exclusion criteria were death during the CRRT treatment period, premature discontinuation of CRRT for any reason, CRRT duration <72 hours (given that ≥3 consecutive daily measurements were required for temporal analysis), or incomplete or inaccessible clinical or laboratory data. Patients were also excluded if hemogram measurements were not performed at the specified time points; if they had active malignancy or hematologic disease; or if they were receiving pulse corticosteroid or other potent immunosuppressive therapies. Additionally, patients were excluded if they receive CRRT for non-septic indications.

CRRT PROCEDURE AND DATA COLLECTION:

All patients were treated with double cross-linked styrene divinylbenzene copolymer cytokine filters using the Jafron HA330 cartridge (Jafron Biomedical Co., Zhuhai, China) integrated into a Prismaflex system (Baxter International Inc., Deerfield, IL, USA). Hemogram analyses were performed with a Sysmex XN-1000 hematology analyzer (Sysmex Corporation, Kobe, Japan). The final hemogram prior to the procedure and those obtained on Days 1, 2, and 3 were evaluated. All blood samples collected on CRRT days were obtained post-session. Clinical and laboratory data were retrieved from the hospital’s electronic medical records (Comed^® Information System, Istanbul, Türkiye).

Demographic data (age, sex, comorbidities) and Acute Physiology and Chronic Health Evaluation (APACHE) II scores of consecutive ICU patients diagnosed with sepsis according to Sepsis-3 criteria were recorded. Laboratory examinations included measurement of C-reactive protein (CRP, mg/L), white blood cell (WBC) count (×10⁹/L), PLR, and NLR for each patient at admission (Day 0) and on the following three days. Temporal changes were calculated as the difference between values on Days 1, 2, and 3 compared with baseline.

OUTCOME MEASUREMENTS:

The primary outcome was the change in PLR from baseline (Day 0) to Day 3 in patients who underwent hemodiafiltration with a cytokine filter. Secondary outcomes included changes in ratios obtained from hemograms at other time points. PLR was selected as the primary outcome because of its established prognostic value in sepsis and systemic inflammation. NLR and other ratios were evaluated as secondary outcomes to provide a broader assessment of inflammatory dynamics.

STATISTICAL ANALYSIS:

Statistical analyses were performed using the Statistical Package for the Social Sciences (SPSS) Version 22.0 (IBM Corp., Armonk, NY, USA). Continuous variables were expressed as mean±standard deviation. Absolute differences and percentage changes were calculated for laboratory parameters. Given that repeated measurements were obtained on Days 0, 1, 2, and 3, repeated measures analysis of variance was used to assess changes over time. When assumptions for parametric testing were not met, the non-parametric Friedman test was utilized. Post hoc pairwise comparisons were adjusted for multiple testing via the Bonferroni method. P-values <0.05 were considered statistically significant. Patients with missing laboratory values at predefined time points were excluded; no imputation was performed. Potential confounding factors, including vasopressor or antibiotic use, were not adjusted for and remain a limitation of the study.

Results

PATIENT CHARACTERISTICS:

In total, 114 patients underwent CRRT with cytokine filters during the study period. Of these, seven were excluded due to missing laboratory data, 11 due to device failure within 72 hours, and three due to death during the CRRT treatment period. Ninety-three patients – 67 men (72%) and 26 women (28%) – were included in the final analysis. The study flow chart is shown in Figure 1. The mean age was 72.4 years (standard deviation: 13.2). The mean APACHE II score was 29.9 (standard deviation: 7.5), indicating high disease severity. The mean ICU stay was 15.6 days (standard deviation: 16.4), and the sepsis filter was used an average of 3.1 times. Demographic data are presented in Table 1.

CHANGES IN CONVENTIONAL LABORATORY PARAMETERS:

Laboratory parameters over time are summarized in Table 2. CRP values showed only minor fluctuations, without significant differences across days. WBC counts also remained stable, with a non-significant downward trend from baseline (17.6±8.6×103/μL) to Day 3 (15.3±9.2×103/μL).

CHANGES IN HEMOGRAM-DERIVED RATIOS:

In contrast, hemogram-derived ratios demonstrated more pronounced changes. The PLR decreased from 290.1±264.4 at baseline to 186.1±189.7 on Day 3, whereas the NLR declined from 23.0±25.8 to 16.3±17.0 during the same period.

STATISTICAL ANALYSES OF INFLAMMATORY MARKERS:

Serial analyses of inflammatory markers are presented in Table 3 and Figure 1. No significant temporal changes were observed in CRP or WBC levels. In contrast, hemogram-derived ratios showed significant reductions during cytokine-filtered CRRT. PLR progressively decreased from baseline to Day 3, with the most pronounced change between Day 0 and Day 3 (mean change −104.0±198.3, P<0.0001). Similarly, NLR declined over time, with a significant reduction from Day 0 to Day 3 (mean change −6.7±14.9, P<0.0001). Intermediate day-to-day comparisons revealed a consistent downward trend, although not all demonstrated statistical significance. These findings suggest that, whereas conventional inflammatory markers such as CRP and WBC remained stable, hemogram-derived ratios were more responsive to cytokine filtration throughout the treatment course.

Discussion

This study evaluated changes in CRP, WBC, NLR, and PLR in patients who underwent cytokine-filtered CRRT. CRP and early-phase hemogram-derived indices did not demonstrate statistically significant changes; PLR significantly decreased between Day 0 and Day 2 and between Day 0 and Day 3. Additionally, a significant reduction in NLR was observed between Day 0 and Day 3. These results indicate that PLR and NLR were affected by the cytokine filter used in CRRT, despite the absence of significant changes in CRP and WBC, and the magnitude of this effect was remarkable.

Hemogram-derived parameters such as NLR, PLR, and the systemic immune-inflammation index are widely used for diagnostic purposes, treatment response assessment, and prognostic evaluation in various diseases [1,2,17,18]. Their relevance extends to critically ill patients, in whom they are frequently utilized due to the involvement of multiple failing organ systems [16,19]. The present study aimed to extend the conventional understanding that these ratios primarily reflect the host response to disease. Our findings suggest that therapeutic interventions commonly applied in the ICU, such as CRRT, can significantly influence the temporal dynamics of these ratios. Thus, caution is warranted when interpreting changes in NLR, PLR, and related indices because they may reflect both treatment effects and disease progression.

Extracorporeal blood purification filters, including CytoSorb® and Jafron®, have been developed to remove septic mediators and cytokines from the circulation. Although their mechanisms and overall efficacy are well established, relatively few studies have directly examined their impact on inflammatory biomarkers. In a prospective observational study, Zuccari et al. [20] found that CytoSorb® hemoadsorption was associated with reductions in circulating interleukin levels, particularly interleukin-8. In the present study, interleukins were not measured; instead, hemogram-derived ratios were assessed. These ratios represent a distinct category of inflammatory indices that are more readily available in routine clinical practice. Importantly, they demonstrated significant changes during cytokine-filtered CRRT. Taken together, these findings emphasize the need for comprehensive investigations that compare different hemoadsorption devices while simultaneously evaluating both classical cytokines and hemogram-derived ratios.

CRRT often results in a decline in white blood cells and platelets. This reduction may result from damage, fragmentation, or dilution of blood elements caused by the filtration process and blood recirculation through the extracorporeal circuit [21–24]. In CRRT using a cytokine-adsorbing filter, this effect is likely at least as pronounced as in conventional CRRT, and possibly even greater [21]. A recent study similar to ours evaluated ICU patients undergoing cytokine-filtered CRRT during the first 12 hours and revealed significant decreases in serum albumin and platelet counts, findings consistent with those observed in conventional CRRT [19]. In contrast, our study did not primarily examine platelet counts but focused on derived ratios; no significant changes were observed in these ratios on the first day. Instead of directly measured parameters such as platelet count, our analysis emphasized ratios such as PLR and NLR, which integrate changes across multiple components. A minor increase in one variable and a slight decrease in another may not be clinically meaningful in isolation, but their ratio can reflect a substantial mathematical shift. We believe such ratios provide greater insight than evaluating each parameter independently.

Recent research suggests that ratios such as NLR can predict mortality in ICU patients, regardless of the primary diagnosis [25]. However, in our view, the timing of blood cell count measurement has equal importance to the values themselves, given their susceptibility to influence from extracorporeal membrane oxygenation, conventional or cytokine-filtered CRRT, and various pharmacologic interventions. Although our study provides preliminary data, it also raises the following important questions. Should ratios be calculated based on laboratory values obtained at ICU admission, or at the initiation of mechanical ventilation? Should patients who received renal replacement therapy prior to ICU admission be evaluated differently? During CRRT, including cytokine-filtered modalities, which treatment day should be used for calculation? Efforts to address these questions are essential for meaningful conclusions in future research.

In our view, particularly important contributions of this study include the emphasis on recognizing day-to-day changes in hemogram parameters and their derived ratios, as well as the understanding that these fluctuations may be directly influenced by CRRT modalities. This perspective can help clinicians to make more informed decisions. It is essential to evaluate patients holistically throughout the treatment process, rather than relying solely on static values. When ratios such as NLR and PLR are used to predict mortality, estimate morbidity, or guide treatment strategies, it is important to acknowledge that these values may shift depending on the therapies administered. Rather than creating confusion, this awareness can help guide clinicians and promote a more nuanced interpretation of dynamic clinical data.

This study has some limitations. The most important is its retrospective design, which inherently carries a risk of bias. Given the single-center retrospective nature of this research, the generalizability of the findings may also be limited. Another limitation is the heterogeneity of the study population in terms of underlying systemic conditions and ICU admission diagnoses. However, we note that the primary objective was not to evaluate clinical outcomes or treatment responses, but to assess the impact of CRRT on hemogram-derived ratios such as NLR and PLR. Additionally, potential confounding factors such as the use of inotropes, antibiotics, or other medications that may have influenced blood cell counts and derived indices were not considered. All catheters were inserted via the femoral route; possible differences related to laterality (right vs left) or alternative insertion sites were not evaluated, which may represent another limitation. Furthermore, microvascular alterations, which may play an important role in sepsis-related outcomes, were not evaluated. Examination of microcirculatory changes could have provided additional insights into pathophysiological mechanisms underlying the observed alterations in hemogram-derived ratios. Future studies incorporating such parameters may help to better contextualize our findings. Further research in more homogeneous patient populations, ideally comparing standard CRRT with cytokine-filtered CRRT and including patients with and without sepsis, is needed to determine whether the observed changes in hemogram-derived ratios are specific to cytokine adsorption or may also occur with conventional CRRT alone.

Conclusions

This study demonstrates that hemogram-derived ratios such as PLR and NLR may substantially change over time in patients receiving cytokine-filtered CRRT, independent of conventional inflammatory markers such as CRP and WBC. These findings underscore the importance of considering treatment effects when interpreting such indices. By recognizing the dynamic nature of hemogram-derived ratios, clinicians can make more informed decisions during the management of critically ill patients.