Patterns of central venous oxygen saturation, lactate and veno-arterial CO2 difference in patients with septic shock.

BACKGROUND AND AIMS: Tissue hypoperfusion is reflected by metabolic parameters such as lactate, central venous oxygen saturation (ScvO2) and the veno-arterial CO2 (vaCO2) difference. We studied the relation of these parameters over time and with outcome in patients with severe septic shock.
MATERIALS AND METHODS: In this single-center, prospective observational cohort study, adult patients (≥18 years) with circulatory shock were included. Echocardiography and simultaneous arterial and venous blood gases were done on enrolment (0 h) and at 24, 48 and 72 h. The partial pressure of CO2, lactate and ScvO2 were recorded from the central venous blood samples. The vaCO2 was calculated as the difference in CO2 between paired venous and arterial blood gas samples.
RESULTS: Of the 104 patients with circulatory shock, 79 patients (44 males) with septic shock aged 49.8 (standard deviation ± 14.6) years and with sequential organ failure assessment (SOFA) score of 11.0 ± 3.4 were included. 71 patients (89.9%) were ventilated (11.4 ± 12.3 ventilator-free days). The duration of hospitalization was 16.6 ± 12.8 days and hospital mortality 50.6%. Lactate significantly decreased over time with a greater decrement in survivors than nonsurvivors (-0.35 vs. -0.10, P < 0.001). For every l/min increase in cardiac output, vaCO2 decreased by 0.34 mmHg (P = 0.006). There was no association between ScvO2 and mortality (P = 0.930). 0 h SOFA and vaCO2 ≤6 mmHg were strongly associated (P = 0.005, P = 0.018, respectively) with higher odds of mortality. However, this association was evident only in those with ScvO2 >70% and not in ScvO2 ≤70%.
CONCLUSION: In septic shock, vaCO2 ≤6 mmHg is independently associated with mortality, particularly in those with normalized ScvO2 consistent with metabolic microcirculatory abnormalities in these patients.

Introduction

The cardiovascular system provides tissue perfusion essential for cellular metabolism. Inadequate tissue perfusion, due to reduced perfusion pressure or abnormal distribution of blood flow, as occurs in shock, results in impaired tissue oxygenation, anaerobic metabolism, and organ-system dysfunction.[1] Although macrocirculatory decreases in cardiac output (CO) often elicit increased sympathetic tone that sustains blood pressure and causes tachycardia,[2] the impact of these compensatory mechanisms and the primary pathologic process on tissue oxygen delivery is more difficult to assess, because regional oxygen delivery is regulated by the microcirculation.[2] Microcirculatory dysfunction in shock can lead to refractory circulatory failure. Moreover, deranged cellular energetics in sepsis is not only due to inadequate tissue perfusion but also impaired mitochondrial respiration; their combination can contribute to treatment failure.[3] Thus, hemodynamic assessment and support requires consideration of both global and regional perfusion as well as end-organ function.[2]

Presently, resuscitation endpoints may be considered upstream targets, including CO, mean arterial pressure and global oxygen delivery. It is unclear, however, how such global resuscitation targets are influenced by and alter metabolic downstream variables such as serum lactate, central venous oxygen saturation (ScvO2), and the veno-arterial CO2 (vaCO2) difference. These measures are markers of global tissue perfusion and resuscitation effectiveness.[4] Elevated serum lactate in circulatory shock usually reflects inadequate tissue perfusion relative to metabolic demand and an associated tissue hypoxia.[5] Lactate levels correlate with morbidity and mortality in patients in shock. However, lactate levels, particularly in the critically ill, may be influenced by factors other than global hypoperfusion.[5] ScvO2, as a surrogate for mixed venous O2 saturation, is another marker of the adequacy of the circulation.[67] Early optimization of ScvO2 within the first 6-h of resuscitation may improve outcome in septic shock.[8] However, given the perfusion heterogeneity in septic shock,[9] ScvO2 might be normal despite overt or occult hypoperfusion.[1] vaCO2 difference, another metabolic parameter that has been explored in shock, is dependent on CO2 and CO. A vaCO2 >6 mmHg, as a result of increased oxygen utilization and CO2 production, reflects a low output state with hypoperfusion.[10] A low vaCO2 of <6 mmHg, on the other hand, has been postulated to reflect impaired utilization of oxygen and low CO2 production due to mitochondrial dysfunction.[310] Although recent interest has focused on the evaluation of these parameters as endpoints of resuscitation,[1112] it is unclear as to the relative value of using these measures as early sepsis resuscitation goals.[13]

A few studies have examined these metabolic parameters over time in patients resuscitated from circulatory shock. As suggested in a recent editorial, several patterns of microcirculatory abnormalities may exist in shock.[12] We hypothesized that different patterns of metabolic dysfunction occurred in patients in shock that may define pathophysiologic prognostic categories useful in tailoring resuscitation strategies amongst a diverse critically ill population.

Materials and Methods

Setting and subjects

We performed a prospective observational cohort study over 8-month (November 2012 until June 2013) in the Medical and Surgical Intensive Care Unit (ICU) and high Dependency Units of a Tertiary Care Hospital in India. The study was approved by the Institutional Review Board. Adult patients (≥18 years) were considered for inclusion if they fulfilled two or more criteria for systemic inflammatory response syndrome[14] and had refractory hypotension, defined as a systolic blood pressure that either was < 90 mmHg or required vasopressor therapy to maintain 90 mmHg even after an intravenous fluid challenge (20–30 ml/kg over 30-min). Exclusion criteria were pregnancy, do not resuscitate status, readmission to ICU within a single hospital stay, absence of an internal jugular or subclavian central venous access, patients who denied consent or who survived <24 h. Patients with other etiologies of shock as discovered during their ICU stay (cardiogenic, hypovolemic, and obstructive shock) were also excluded. Patients were followed up until death or hospital discharge.

Study protocol

All included patients had arterial (radial or femoral) and central catheters (internal jugular or subclavian, inserted according to standard protocol, with the tip of the catheter at the upper part of the right atrium). Echocardiography and simultaneous arterial and venous blood gas analyses were carried out, and sequential organ failure assessment (SOFA) score[15] and vasoactive inotrope score (VIS)[16] were calculated on enrolment (0 h, T0) and at 24 (T24), 48 (T48), and 72 h (T72) postenrolment. All assessments were done within 2 h of the specified time. The partial pressure of carbon dioxide, lactate and ScvO2 were recorded from the central venous blood samples. The vaCO2 was calculated as the difference in CO2 between paired venous and arterial blood gas samples. Central venous PCO2 has been used as a surrogate for mixed venous PCO2 to identify inadequately resuscitated patients in septic shock.[1217] The cut-off values of 6 mmHg for vaCO2 and 70% for ScvO2 were chosen according to previous studies.[817] CO was measured by transthoracic echocardiography using the left ventricular outflow method[18] with the Sonosite® Turbo portable ultrasound machine, as recently validated for the assessment of cardiac hemodynamics in ICU.[19] All measurements were taken by a single physician trained in ICU ultrasonography, and an average of three values was taken to reduce intra-observer variability. We also assessed inferior vena cava (IVC) diameter variability in mechanically ventilated patients with no spontaneous breathing (heavily sedated or paralyzed)[20] and IVC collapsibility in spontaneously breathing patients.[21] IVC variability of >12%[20] and collapsibility of >50%[21] were considered as markers of volume responsiveness and an under-filled intravascular state, respectively. All blood gas samples were analyzed by a point-of-care blood gas analyzer (GEM® Premier 4000, GEM® Premier 3000, Radiometer ABL800 flex®, Copenhagen Denmark). Additional tests for the evaluation of shocks such as complete blood counts, electrolytes, liver and renal function tests, electrocardiogram, and appropriate cultures were done as the clinical situation warranted.

Patients were categorized as a septic, cardiogenic, hypovolemic or obstructive shock. Patients in septic shock were analyzed for the stated outcomes. The diagnosis of septic shock was based on an identifiable focus of infection with shock.[22]

Outcomes

The primary outcome assessed was in-hospital mortality. Other outcomes included need and duration of ventilation, ventilator-free days,[23] and length of hospital stay. Mortality in the hospital included patients who took discharge against medical advice in a moribund condition.

Statistical analysis

Each of the metabolic parameters (lactate, ScvO2 and vaCO2) was assessed for their relationship with hospital outcome, CO and trends over time. Analysis of categorical outcome data was done by Fisher's exact test. Linear mixed effects models were used for assessment of change in outcome over time. Logistic regression analysis was undertaken to see the independent association of the baseline parameters with mortality. Data were expressed as mean ± standard deviation (SD). P ≤ 0.05 was considered statistically significant. Statistical analysis was performed using R version 3.0.1. Graphics were created using the ggplot2 package within R.[24]

Results

During the study period, 175 patients were admitted to the ICUs with hypotension, refractory to fluid therapy. Of these, 71 patients were excluded [Figure 1]. Of the remaining 104 patients who were enrolled, 79 patients (44 males, 35 females) with septic shock of diverse etiology [Table 1] and aged 49.8 (SD ± 14.6) years formed the study cohort.

Figure 1

Flow chart depicting patients screened, and reasons for exclusion - of the 175 patients screened, 104 patients were included. 79 patients were categorized as septic shock, 13 as cardiogenic shock, 8 as hypovolemic shock and 4 as an obstructive shock. One patient with chlorpromazine poisoning who presented with distributive shock was excluded as the patient did not fit into any specific etiologic category. DAMA: Discharged against medical advice

Table 1

Etiology of shock

The T0 SOFA score of 11.0 ± 3.4 suggested a cohort of very sick patients with septic shock. T0 ScvO2, vaCO2 and lactate were 71.6% ± 15.2%, 7.1 ± 4 mmHg and 5.8 ± 4.9 mmol/dl, respectively. At the time of recruitment, more than half the patients (59.5%) were under-filled as assessed by IVC variability and collapsibility. 71 (89.9%) patients were ventilated; ventilator-free days being 11.4 ± 12.3. The duration of hospitalisation was 16.6 ± 12.8 days, and in-hospital mortality was 50.6% [Table 2].

Table 2

Outcome data overall and categorized by type of shock

Enrolment SOFA was significantly (P = 0.005) associated with increased odds of death in patients with septic shock (odds ratio [OR]: 1.23, 95% confidence interval [CI]: 1.07-1.45. SOFA score declined significantly over time (P < 0.001) in septic [Table 3] as well as all types of shock (P < 0.001, not presented).

Table 3

Mixed-effects models of change in each outcome over time in patients with septic shock

Veno-arterial CO2

In septic patients, enrolment vaCO2 ≤6 mmHg was significantly associated (P = 0.018) with increased mortality (OR: 3.06, 95% CI: 1.23–7.94). However, this association was restricted to those with ScvO2 >70% (P = 0.039) and not in patients with ScvO2 ≤ 70 (P = 0.450). There was no evidence of a change in vaCO2 (P = 0.747) values over time [Table 3]. For every l/min increase in CO, vaCO2 significantly decreased by 0.338 mmHg after adjusting for repeated measures over time [P = 0.006, Figure 2].

Figure 2

Relationship between veno-arterial CO2 difference and cardiac output over time in septic patients - when veno-arterial CO2 was examined against time and cardiac output, there was no significant interaction between cardiac output and time (P = 0.664), that is, the slope was constant over time. Removing the interaction, there was no significant effect of time (P = 0.849) but for every l/min increase in cardiac output, the veno-arterial CO2 decreased significantly by 0.338 mmHg (P = 0.006). The dotted lines indicate mean values and the shaded area the 95% confidence interval

Central venous oxygen saturation

In patients with septic shock, enrolment ScvO2 >70 was not associated (P = 0.930) with higher odds of mortality (OR: 1.00, 95% CI: 0.42–2.60). There was no evidence of a change in ScvO2 (P = 0.063) values over time [Table 3]. For every l/min increase in CO, ScvO2 increased by 1.11% after adjusting for repeated measures over time [P = 0.027, Figure 3].

Figure 3

Relationship between central venous oxygen saturation and cardiac output over time in septic patients - central venous oxygen saturation versus cardiac output plus time showed no significant interaction (P = 0.990), implying the same effect of cardiac output on central venous oxygen saturation at each time point. Removing the interaction, there was a non-significant effect of time (P = 0.054) and significant effect of cardiac output (P = 0.027) implying that for every l/min increase in cardiac output, the central venous oxygen saturation increased by 1.11%. The dotted lines indicate mean values and the shaded area the 95% confidence interval

Lactate

Initial lactate was not associated (P = 0.109) with higher odds of mortality (OR: 1.09, 95% CI: 0.99–1.22). Lactate significantly decreased over time [P < 0.001, Table 3] with the rate of decrease more pronounced in survivors than nonsurvivors (0.35 vs. 0.10, P < 0.001).

Other parameters

Patients in septic shock also showed a significant decrease in hemoglobin level (0.39 g/dl every 24 h, P < 0.001) and VIS (4.42 units every 24 h, P < 0.001), [Table 3]. For every 24 h time interval, the odds of having IVC variability were significantly lower (OR: 0.69 every 24 h, P = 0.006). A significant increase in serial mean arterial pressure (3.56 mmHg every 24 h, P < 0.001) was also seen.

Discussion

Several metabolic patterns may exist in patients with shock. In our study, in septic patients, baseline vaCO2 of ≤ 6 mmHg was associated with a higher mortality, but only in those patients with ScvO2 >70%. Enrolment SOFA scores were associated with mortality. There was no significant association between initial ScvO2 and lactate levels with mortality. The lactate levels however significantly decreased over time, and the rate of decrease was more pronounced in survivors than nonsurvivors. For every l/min increase in CO, vaCO2 decreased by 0.34 mmHg (P = 0.006) and ScvO2 increased (P = 0.027) by 1.11% after adjusting for repeated measures over time. These observations are consistent with the hypothesis that primary metabolic failure in septic shock may be an important contributor for increased mortality.

Hypovolemia, hypotension, and low CO, along with ineffective oxygen utilization may be seen in shock patients. Downstream tissue parameters such as vaCO2, ScvO2, and lactate are used as markers of tissue perfusion.[4] A decrease in ScvO2 may be explained by a decrease in oxygen delivery or increase in tissue oxygen consumption or both.[25] Thus, increasing oxygen delivery to reverse a low ScvO2 has been used in resuscitation protocols, particularly the early goal-directed therapy in septic shock.[8] ScvO2 after the first 48 h of admission rather than initial ScvO2 may predict better mortality.[26] Overzealous correction of ScvO2 to supranormal levels can also lead to increased mortality in patients with preexistent hypoperfusion-induced organ injury.[2527] In our study, enrolment ScvO2 was not associated with mortality.

Some patients after resuscitation “normalize” ScvO2 (>70%) but continue to manifest features of tissue hypoperfusion as evidenced by an increased vaCO2.[17] These patients with a vaCO2 “gap” of >6 mmHg may indicate a subset who continue to remain inadequately resuscitated.[17] On the other hand, impaired mitochondrial respiration in sepsis with nonutilization of oxygen by the cell may decrease CO2 production and result in a “narrow” vaCO2 gap, since anaerobic metabolism is less efficient in producing CO2.[310] These patients may reflect those with cytopathic dysoxia or regional microcirculatory abnormalities in sepsis.[12] In such patients (those with normalized CO [ScVO2 >70%] and narrow vaCO2 gap), clinical recovery may be more reliant on reversal of the underlying process causing microcirculatory dysfunction rather than further aggressive attempts at optimization of CO and oxygen delivery. Indeed in our cohort, septic patients with a narrow vaCO2 gap and normalized ScvO2 had a higher mortality compared with those with a high vaCO2 gap.

Lactate has traditionally been used as a marker of anaerobic metabolism and tissue hypoperfusion. Hyperlactatemia represents the imbalance between lactate production and clearance. Lactate levels, particularly in critically ill patients, can be influenced by factors other than global perfusion such as hepatic dysfunction, drugs (e.g., catecholamine, metformin, and anti-HIV drugs), and regional (e.g., bowel) ischemia.[5] Serial lactate measurements and lactate clearance over time have been shown to be better prognostic markers of mortality than single lactate concentration, which are themselves not sensitive or specific for tissue hypoperfusion.[28] Our results are consistent with these previous observations. Initial lactate levels were not associated with mortality. However, the reduction in lactate level over time was more pronounced in survivors than in nonsurvivors (P < 0.001). These observations suggest that the rapidity of reversal of the septic process (either because therapy is instituted promptly or because the pathophysiology is reversible) is more important than the absolute enrolment (extent of derangement) lactate value.

While lactate, ScvO2 and vaCO2 reflect downstream tissue markers, CO reflects an upstream endpoint of resuscitation more easily manipulated[4] and determines the global delivery of oxygen to tissues. In support of this assumption, we noted that higher CO was associated with a significant reduction in vaCO2 [Figure 2] and a significant increase in ScvO2 [Figure 3]. This observation is consistent with earlier studies.[629]

SOFA score has been used as a prognostic indicator during the first few days of ICU admission.[30] Consistent with other studies, the enrolment SOFA score in our study was associated with mortality.

Our study has some limitations. First, four discrete time points were chosen at 24-h intervals for the evaluation of the various parameters. Continuous monitoring of these parameters and measuring changes with specific interventions would have been more relevant for defining the impact of specific interventions. However, this was not feasible due to cost constraints. Second, we did not use a defined treatment protocol common to all patients, but rather titrated care individually as the intensivists deemed appropriate. However, our treatment approach is common across the institution, and the same intensivists cared for all patient groups independent of shock etiology. Third, although we show a clear separation of sepsis mortality in those patients with ScvO2 >70% and vaCO2 <6 mmHg we did not assess mitochondrial function or see any correlation between these differences and macrocirculatory parameters. Fourth, we did not account for other factors affecting CO2 production and elimination like diet and alveolar ventilation. Thus, our conclusion that impaired mitochondrial function in sepsis caused the increased mortality remains hypothesis generating and warranting further study.

Conclusion

In septic shock, low vaCO2 is a predictor of mortality in patients with normalized ScvO2. These findings are consistent with cytopathic dysoxia and microcirculatory dysfunction in septic shock.

Financial support and sponsorship

Nil.

Conflicts of interest

There are no conflicts of interest.