Assessing fluid responsiveness with central venous oxygen saturation: the complex relationship between oxygenation and perfusion

In the process of hemodynamic resuscitation, the aim of volume expansion (VE) is to increase cardiac output (CO) and, consequently, oxygen delivery (DO₂) to restore oxygen availability at the tissue level. Such intervention should ideally be performed only when tissue hypoxia is suspected. Otherwise, despite increasing CO, the intervention could potentially lead to harmful effects.

Since CO is not routinely monitored in daily practice, some authors have suggested that certain metabolic variables, such as central venous oxygen saturation (S_cvO₂), could serve as indirect markers for assessing fluid responsiveness [1]. Mallat and colleagues recently published an interesting study in Critical Care, further confirming the association between a positive CO response and S_cvO₂ as a result of VE [2]. The authors propose that S_cvO₂ can be used in the absence of CO measurements to define fluid responsiveness in critically ill patients. While there is extensive evidence supporting this association, the relationship between a flow variable (CO) and a metabolic variable (S_cvO₂) is more complex than it appears and warrants cautious consideration when integrated into bedside clinical decisions.

According to the proposed indirect approach to fluid responsiveness, a certain increase in S_cvO₂ following VE would indicate a positive CO response, whereas an unchanged or marginally increased S_cvO₂ would indicate a negative CO response. However, S_cvO₂ changes are influenced not only by CO but also by the relationship between DO₂ and oxygen consumption (VO₂). According to Fick's principle,

$${\text{VO}}_{{2}} = {\text{ CO }}\cdot \, \left( {{\text{C}}_{{\text{a}}} {\text{O}}_{{2}} - {\text{ C}}_{{{\text{mv}}}} {\text{O}}_{{2}} } \right) \, \cdot{ 1}0$$

where C_aO₂ and C_mvO₂ are the arterial and mixed venous oxygen contents, respectively. This can be further derived into:

$${\text{CO }} = {\text{ VO}}_{{2}} / \, \left( {{\text{C}}_{{\text{a}}} {\text{O}}_{{2}} - {\text{ C}}_{{{\text{mv}}}} {\text{O}}_{{2}} } \right) \, \cdot{ 1}0$$

$${\text{CO }} = {\text{ VO}}_{{2}} / \, \left[ {{13}.{9 }\cdot \, } \right[{\text{Hb}}\left] { \, \cdot \, \left( {{\text{S}}_{{\text{a}}} {\text{O}}_{{2}} - {\text{ S}}_{{{\text{mv}}}} {\text{O}}_{{2}} } \right)} \right]$$

$${\text{S}}_{{{\text{mv}}}} {\text{O}}_{{2}} = {\text{ S}}_{{\text{a}}} {\text{O}}_{{2}} - \, \left[ {{\text{VO}}_{{2}} / \, ({13}.{9 }\cdot \, } \right[{\text{Hb}}\left] { \, \cdot{\text{ CO}})} \right]$$

Therefore, mixed venous oxygen saturation (S_mvO₂) depends on S_aO₂, VO₂, hemoglobin concentration ([Hb]), and CO. Consequently, changes in S_mvO₂ after VE depend on the fluids’ effects on each of these parameters, not solely on CO. While the interchangeability of S_mvO₂ and S_cvO₂ has been debated [3], S_cvO₂ has become more prominent as a monitoring variable due to practical considerations. For the purposes of this discussion, we assume that S_cvO₂ changes after VE are influenced by the same factors as S_mvO2: S_aO₂, VO₂, [Hb], and CO.

Although numerous studies have examined the ability of S_cvO₂ changes to detect significant CO increases, few have factored in VO₂ [2, 4,5,6,7,8,9,10]. In a recent study of early septic shock patients receiving fluid boluses, we also observed that S_cvO₂ changes differed significantly between CO responders and non-responders (5 ± 5% vs. 0 ± 5%, p < 0.001) [4]. However, among CO responders, S_cvO₂ evolution varied significantly between those whose VO₂ increased and those whose VO₂ did not. Interestingly, smaller increases in S_cvO₂ were observed in patients with VO₂ increases (2 ± 4% vs. 7 ± 5%, p = 0.03). Indeed, in CO responders, smaller S_cvO₂ increases were better predictors of VO₂ increases after VE. Similar results were reported by Monnet et al., who studied the metabolic response to VE in mixed critically ill patients with acute circulatory failure [5]. Among CO responders whose VO₂ increased, S_cvO₂ remained unchanged (from 70 ± 15 to 71 ± 13%, p = 0.2). In contrast, among CO responders with unchanged VO₂, S_cvO₂ increased significantly (from 64 ± 4 to 71 ± 2%, p < 0.01). Other authors have similarly reported significant S_cvO₂ increases in CO responders without corresponding VO₂ changes after fluid boluses [8,9,10]. These findings suggest that baseline VO₂/DO₂ dependency was not always present, even in patients where VE was administered to address potential tissue hypoperfusion.

Collectively, data from studies reporting CO, S_cvO₂, and VO₂ before and after VE indicate that, even though a positive response in CO may provoque increases in S_cvO₂ and VO₂, we also need to take into account two additional and significant patterns of response (Fig. 1): (1) Increases in CO result in S_cvO₂ increases, while VO₂ remains unchanged; and (2) Increases in CO do not cause changes in S_cvO₂, while VO₂ increases.

Fig. 1

Potential metabolic responses to increasing CO after VE. Depending on the baseline relationship between VO₂ and DO₂, the final observed impact on S_cvO₂ may significantly vary. CO, Cardiac Output; VO₂, Oxygen consumption; DO₂, Oxygen delivery; O₂ER, Oxygen extraction ratio; S_cvO₂, central venous oxygen saturation

The first pattern reflects the shunting effect of increased oxygen availability in tissues without hypoxia. The second reveals VO₂/DO₂ dependency, where increases in DO₂ proportionally enhance VO₂. These findings suggest that S_cvO₂ responses are closely tied to tissue oxygenation status rather than fluid responsiveness alone. Paradoxically, an S_cvO₂ increase following VE, interpreted as a positive effect, might simply reflect the shunting effect of increased oxygen availability without metabolic benefit. Solely relying on S_cvO₂ to interpret VE responses may lead to two errors:

• (1) A lack of S_cvO₂ increase is misinterpreted as a negative CO response, despite a CO increase improving tissue hypoxia.

• (2) An S_cvO₂ increase is misinterpreted as beneficial, when it merely indicates shunting in tissues no longer VO₂/DO₂ dependent.

In summary, while VE-induced changes in S_cvO₂ may serve as indirect markers of fluid responsiveness in certain scenarios, their interpretation has limitations, particularly in conditions involving tissue hypoxia. In VO₂/DO₂ dependent situations, a positive perfusion response may not correspond to increased central venous oxygenation. Since the primary goal of fluid expansion is to address tissue VO₂/DO₂ dependency, S_cvO₂ alone may be insufficient for accurately assessing both fluid responsiveness and VO₂ response.