Background

Sepsis and septic shock are one of the most serious healthcare problems jeopardizing human life [1], with extremely high morbidity and mortality rates [2, 3]. The pathophysiological changes in macrocirculation and microcirculation are not completely synchronous during septic shock, and the microcirculation status may still be hardly corrected even if the indicators of macrocirculation are improved, resulting in impaired tissue and organ perfusion [4]. Moreover, the severity and duration of hypoperfusion induced by microcirculation disturbance are associated with the increased mortality rate [5]. In current clinical practices, lactic acid (Lac) [6], skin mottling score [7, 8], capillary refill time (CRT) [1, 9], venous-to-arterial carbon dioxide difference (Pv-aCO2) [10, 11] are often utilized to assess the microcirculation status, but there are limitations in the monitoring tools for these indicators. Therefore, it is necessary to develop the methods capable of evaluating the microcirculation status more accurately and quickly.

There are four blood glucose monitoring approaches in clinic: arterial blood glucose, peripheral blood glucose, interstitial fluid (ISF) glucose, and venous blood glucose [12]. Currently, there is a significant difference in the balance between ISF glucose and arterial blood glucose (that is, there is a certain difference in the blood glucose simultaneously measured at two different sites) [13, 14]. Physiologically, such a difference can reflect the process of glucose exchange between the capillary walls and the ISF [15]. Under normal physiological conditions, the concentration of arterial blood glucose is higher than that of ISF glucose [16]. For patients with septic shock, four different types of disturbances (asymmetric perfusion, hemodilution, blood stasis, and capillary leakage) are suggested to exist in the microcirculation even though the macrocirculation has been corrected [17, 18]. In the case of microcirculation disturbance mainly manifested as capillary leakage, glucose leaks to the ISF through the damaged vascular endothelium, the arterial blood glucose and ISF glucose difference (GA−I) is dramatically decreased, and ISF glucose even exceeds arterial blood glucose due to different degrees of glucose utilization disorder [19, 20]. The increased cellular glucose consumption or reduced perfusion may lead to reduced ISF glucose and GA−I is increased. Moreover, with hemodilution as the main manifestation of microcirculation disturbance, GA−I may decrease or remain unchanged due to the significantly reduced blood arterial glucose resulting from severe hemodilution.

Herein, it was hypothesized that GA−I can accurately assess the microcirculation status in patients with septic shock and effectively predict their clinical prognosis, which may be an easy measurement of tissue perfusion. This study aims to investigate whether GA−I can accurately evaluate the microcirculation status in patients with septic shock and efficiently predict their clinical prognosis.

Materials and methods

Study design and patients

Patients with septic shock admitted to the intensive care unit (ICU) in this observational study were included from the Second Affiliated Hospital of Soochow University, China between November 2022 and January 2024. This study was approved by the institutional review board of the Second Affiliated Hospital of Soochow University (Approval number: No.LK2023055). The informed consents were obtained from the patients’ relatives at admission.

The inclusion criteria were as follows: (1) patients diagnosed as septic shock according to Sepsis 3.0 [i.e., infection + sequential organ failure assessment (SOFA) score ≥ 2 points + adequate fluid resuscitation still requiring vasoactive agents to maintain mean arterial pressure (MAP) ≥ 65 mmHg] [21]; (2) blood Lac > 2.0 mmol/L, (3) age ≥ 18 years old; and (4) an expected ICU stay ≥ 24 h. Pregnant women, or patients with an infection at the site of the ambulatory glucose monitoring device were excluded. All the enrolled patients were treated according to the latest sepsis guideline [1, 21].

Data collection

Patients subjected to standardized bedside testing at the time of enrollment (H0), H2, H4, H6, and H8 according to predetermined schemes. Besides, the general characteristics of the patients were recorded, including demographic information, diagnosis, disease severity [e.g., SOFA score (A higher score within 24 h after enrollment indicated severer organ dysfunction in the patient [22]), and Acute Physiology and Chronic Health Evaluation (APACHE) II score (a higher APACHE II score signified severer disease, poorer prognosis, and higher mortality rate [23]), and microcirculation-related parameters [reflecting organ and tissue perfusion) (such as Lac, skin mottling score, capillary refill time (CRT), urine volume, Pv-aCO2, central venous oxygen saturation (ScvO2) and GA−I]. All parameters were measured as per the predetermined schemes, and all researchers were pre-trained. CRT referred to the time from a physician applying an appropriate pressure to the fingernail bed of the patients for 15 s to exactly remove the blood from the nail tip and cause a crescent (whitening) underneath the fingernail bed to the full recovery of the color of the fingernail bed [9]. Meanwhile, CRT was measured for three consecutive times by two physicians to calculate the mean value, so as to minimize the measurement error. The skin mottling score reflected the size of piebald areas on the knees and thighs of the patients in the supine position with the legs stretched out and flush with the heart, which ranged from 0 to 5 points [7]. A semi-quantitative assessment of skin mottling at the bedside was performed using the skin mottling score system. This system employs a scoring scheme ranging from 0 to 5, with a score of 0 indicating the absence of mottling, while scores 1 through 5 represent increasing levels of mottling severity. Specifically, a score of 1 indicated the presence of mottling only in the center of the knee, while a score of 2 indicated that it had extended to the edge of the kneecap. A score of 3 indicated mottling above the knee but not extending beyond the mid-thigh and calf, while a score of 4 indicated its spread to the ends of the thigh and calve. Finally, a score of 5 represented mottling that had spread to the groin and ankle. Pv-aCO2 indicated the CO2 difference calculated by subtracting the partial pressure of CO2 in the blood of the superior vena cava (PvCO2) from that in the arterial blood (PaCO2), which were obtained from the arterial and central venous samples collected simultaneously. A central venous blood sample was collected from the patient using a central venous catheter, and the ScvO2 in the blood sample was measured by a blood gas analyzer. The laboratory parameters were recorded on the day of septic shock, including red blood cell (RBC), hematocrit (HCT), hemoglobin (Hb), white blood cell (WBC), platelets (PLT), C-reactive protein (CRP), neutrophil (N), lymphocyte (L), procalcitonin (PCT), alanine transaminase (ALT), aspartate aminotransferase (AST), total bilirubin (TBIL), direct bilirubin (DBIL), urea nitrogen (BUN), creatinine (Cr), alkaline phosphatase (ALP), albumin (ALB), prealbumin (PAB), creatine kinase (CK), serum amylase (AMS), myohemoglobin (MYO), troponin T (cTNT), N terminal pro B type natriuretic peptide (NT-ProBNP), prothrombin time (PT), activated partial thromboplastin time (APTT), international normalized ratio (INR), thrombin time (TT), fibrinogen (FIB), and D-dimer.

ISF and GA−I assessment

All septic shock patients wore the instantaneous glucose meters (Abbott Diabetes Care Ltd., UK). Arterial blood glucose was measured at the same time as arterial blood gas analysis, and ISF glucose was collected using the instantaneous glucose meters at this time. The difference between the two subtractions was GA−I.

Outcomes

All patients were followed up until discharge or death, with ICU mortality as the outcome. Besides, Lac, skin mottling score, CRT and Pv-aCO2 were monitored and employed as baseline parameters for assessing microcirculation status in septic shock because of their accessibility and accuracy.

Statistical analysis

R 4.3.1 (R Foundation for Statistical Computing, Vienna, Austria) was utilized for statistical analysis of data. Enumeration data were expressed as frequency and percentage, and measurement data were subjected to Shapiro-Wilk test for normality analysis. Normally distributed quantitative data were expressed as mean ± standard deviation and compared between groups by the t-test, while quantitative data not distributed normally were described as median and quartile and compared between groups by the Mann-Whitney U-test, chi-square test, and Fisher’s exact test. Spearman analysis was adopted to figure out the correlations of GA−I with Lac, skin mottling score, CRT, urine volume Pv-aCO2, and ScvO2, and skin mottling score. P < 0.05 suggested a statistically significant difference. Additionally, unrestricted cubic spline and generalized additional model was utilized for fitting.

Results

Characteristics of patients with septic shock

A total of 43 septic shock patients were enrolled in this cohort, of whom 18 died. The characteristics of these patients were showed in Table 1. There were statistical differences in age, CRP, PCT, BUN, CK, TT, NE dose, respiratory system infection, SOFA score, APACHAE-II score, mechanical ventilation treatment, and CRRT in the survival and death groups (all P < 0.05).

Table 1 Characteristics of patients with septic shock
Full size table

Correlations of GA−I with microcirculation parameters

Figure 1 shows the correlation between GA−I and microcirculation parameters (Lac, skin mottling score, CRT and urine volume). The GA−I levels were negative correlation with CRT (r = -0.369, P < 0.001), Lac (r = -0.269, P < 0.001), skin mottling score (r = -0.223, P < 0.001), and were positively associated with urine volume (r = 0.135, P < 0.05). No correlations were observed between GA−I and Pv-aCO2, ScvO2, which may be related to the small sample size (Fig. 2).

Fig. 1
figure 1

Correlations of GA−I with Lac, skin mottling score, CRT and urine volume

Full size image
Fig. 2
figure 2

Correlations of Pv-aCO2 with GA−I, Lac, skin mottling score, CRT, urine volume, and ScvO2

Full size image

Characteristics of microcirculation parameter levels within the first 8 h between the two groups

The microcirculation parameter levels between the two groups were shown in Table 2. The statistical differences were found in the mean lactic acid (1.96 mmol/L vs. 2.74 mmol/L), mean urine volume (2.0 mL/kg*h vs. 0.9 mL/kg*h), and mean CRT (1.4 s vs. 2.5 s) levels between the survival group and the ICU death group.

Table 2 Comparison for the microcirculation parameters between the two groups
Full size table

Relationship between GA−I and ICU mortality

The mean GA−I in the first eight hours was subjected to interquartile interval partition, and it was found that the mortality rate was 83.3%, 20.0%, 10.0% and 45.5%, when the mean GA−I was less than 0.3 mmol/L, located in 0.30–0.98 mmol/L, located in 0.98–2.14 mmol/L and greater than 2.14 mmol/L (Fig. 3). When the mean GA−I was less than 0.30 mmol/L or more than 2.14 mmol/L in the first eight hours of septic shock, the mortality rate was higher than that when the mean GA−I was between 0.30 and 2.14 mmol/L [65.2% vs. 15.0%, OR = 10.625, 95%CI: 2.355–47.503), P < 0.01].

Fig. 3
figure 3

Quartiles of the mean GA−I in the first eight hours

Full size image

Discussion

The application value of GA−I in assessing the microcirculation status in patients with septic shock was proposed in this study for the first time. The observations in this study revealed that GA−I had good accuracy in assessing the severity of microcirculation status in septic shock, and it was associated with CRT, skin mottling score and urine volume.

Currently, there is a lack of standards for microcirculation disturbance in septic shock, so it is hard to discover and validate new monitoring indicators. Moreover, novel indicators for monitoring microcirculation status should meet the following three characteristics [24, 25]: (1) Novel indicators should have a specific physiological relationship with microcirculation disturbance (at least theoretically). (2) Novel indicators should be able to effectively reflect the risk of death in patients with microcirculation disturbance and have an association with traditional indicators reflecting microcirculation status. (3) Novel indicators should be different in patients with and without septic shock. GA−I used in this study meets the above three characteristics well and is effective in predicting the risk of death in patients with septic shock.

As a bridge between blood and cells, ISF accounts for 75% of extracellular fluid and 15–25% of body weight [26]. Similar to plasma, ISF contains various biomarkers [27, 28], which has been applied for tumor microenvironment analysis, dermatologic drug bioavailability determination, and ISF glucose measurement in clinical practice [29, 30]. Continuous glucose monitoring, one of the clinical applications of ISF glucose, has been widely recognized by general diabetic patients due to its effectiveness and accuracy [31], but it has not been applied in critically ill patients, which is related to the special hemodynamic conditions of such patients. D Ballesteros et al. found [32] that peripheral blood glucose and ISF glucose are significantly correlated (Pearson r = 0.71, P < 0.0001) in patients with distributive shock (n = 18, with septic shock in all cases) and the mean difference is about 0.22 mmol/L, which is significantly smaller than that in patients with stable hemodynamics, that is, the “level difference” between arterial blood glucose and ISF glucose is significantly narrowed compared with that in patients with stable hemodynamics, also suggesting the endothelial barrier damage and capillary leakag. This indicates severe microcirculation disturbance in patients (Fig. 4, A – B). Antje Gottschalk et al. discovered during ISF glucose monitoring of patients undergoing surgery with extracorporeal circulation that in some patients with severe hypoperfusion, ISF glucose monitoring device provides hypoglycemia alert messages instead of parameters [33]. This is relevant to the fact that the administration of vasoactive agents and severe peripheral vasoconstriction during severe shock result in impaired glucose transport through the capillary network and ISF, that is, the “level difference” between arterial blood glucose and ISF glucose increases, which is visually reflected by a significant increase in GA−I (Fig. 4, A-C). Our findings showed that the mean mortality rate was higher in patients with a GA−I, less than 0.30 mmol/L and greater than 2.14 mmol/L, than that in those with a GA−I of 0.30–2.14 mmol/L in the first eight hours after the onset of septic shock.

Fig. 4
figure 4

(A) Under physiological conditions, capillary permeability is normal, and glucose is passively transported from the blood vessel to the tissue interstitium according to the concentration gradient. (B) During infection and inflammatory storm, increased capillary permeability leads to the leakage of fluid rich in glucose from the blood vessel to the tissue interstitium, resulting in an increase in tissue fluid glucose and a significant decrease in GA−I. (C) Due to the use of high-dose vasoactive drugs, small artery spasm causes blood stasis in capillaries, during which the transport of glucose from the capillary network to the interstitial fluid is inhibited, resulting in a significant increase in GA−I [17, 35]

Full size image

CRT, skin mottling score, and urine volume are parameters relatively important for assessing microcirculation disturbance [1, 34]. In this study, CRT, skin mottling score, and urine volume were used as benchmarks to compare the accuracy of GA−I. It was found that GA−I was non-linearly correlated with CRT (Figs. 1 and 2). A marked reduction in GA−I indicated “a decrease in the level difference” between arterial blood glucose and ISF glucose, capillary endothelium damage-induced microcirculation disturbance, and a prolongation of CRT, whereas an evident elevation in GA−I suggested “an increase in the level difference” between arterial blood glucose and ISF glucose, stagnant microcirculation, and a prolongation of CRT. GA−I exhibited an association with skin, mottling score, and urine volume. Hence, GA−I can be considered as a novel method to assess microcirculation disturbance.

Microcirculation disturbance with different levels is found in all patients with septic shock. Our finding showed that GA−I may be a new predictive factor for death of patients with septic shock in the ICU, which is a minimally invasive, continuous and rapid assessment tool. The results also showed that in a certain period of time, the variation range of GA−I was associated with disease severity and prognosis. The GA−I may be employed to provide a reference before and after treatment changes or specific interventions, which may be a direction of future studies.

This study has some shortcomings. Firstly, it is a single-center study with a small sample size, so its generalizability may be limited. Secondly, multiple microcirculation indicators were used as benchmarks due to the lack of standards for microcirculation disturbance at present. Moreover, as a proof-of-concept study, joint prediction using more advanced and complex models was not conducted in this study. Finally, the role of GA−I in predicting the ICU mortality was not further prospectively validated. Other possibilities for monitoring the microcirculation, such as SDF method, HVM, were not considered in our study. Relevant studies will be considered in the following work. In addition, several confounding factors were not considered due to the limitation of the sample size. Subsequently, a multicenter study will be carried out to prospectively validate the role of GA−I in predicting the ICU mortality of patients with septic shock, after which a multicenter randomized controlled study will be conducted to validate the role of GA−I in assessing microcirculation status in septic shock again.

Conclusions

In the current study, GA−I was correlated with microcirculation parameters, and with differences in survival. Future studies are needed to further explore the potential impact of GA−I on microcirculation and clinical prognosis of septic shock, and the bedside monitoring of GA−I may be beneficial for clinicians to identify high-risk patients.