Perfusion index and fluid responsiveness: Are we asking the right question?

Hemodynamic monitoring of critically ill patients is the cornerstone of patient management for an intensivist. It represents the starting point for detecting any meaningful clinical deterioration and delivering the appropriate therapy without delay. From this perspective, invasive arterial monitoring with an indwelling arterial catheter is established as the historical and undeniable standard method of monitoring. It provides clinicians with the most commonly required values for proper patient management. Indeed, arterial blood pressure and its waveform components (systolic, diastolic, and mean arterial pressure) remain the most accessible and clinically intuitive variables for bedside interpretation, and form the basis for decision-making in daily management. Accordingly, major international guidelines in sepsis and trauma consistently endorse early hemodynamic assessment, with therapeutic strategies guided by thresholds derived from invasive arterial pressure monitoring.^[1,2]

Over the last few decades, numerous indices derived from the arterial signal have been validated for detecting preload dependency and guiding vasoactive drug therapy, among other applications.^[3] This highlights the high benchmark that any new hemodynamic monitoring technology must meet to be considered a new standard.

Noninvasive monitoring is the logical alternative to invasive monitoring. Various technologies were developed to reproduce the invasive arterial signal and deliver similar values. Among them, photoplethysmography (PPG) has emerged as a potential approach. PPG has been widely used for pulse oximetry monitoring in anesthesia, perioperative care, and intensive care. PPG is a non-invasive optical technique used to detect changes in blood volume in the microvascular bed of tissue. The method consists of a light source (red and infrared) generated by the oximeter probe (light-emitting diodes) that is detected by a sensor (photodetector) positioned either opposite to (transmission mode, e.g., fingertip, earlobe) or adjacent to (reflection mode, e.g., forehead, wrist) the light source (Figure 1). The variation in absorbed light is proportional to the amount of blood at that site. The PPG waveform, obtained through variations in infrared light absorption, contains two components. The pulsatile component (called the AC component) reflects changes in the microvascular blood volume during one cardiac cycle. The non-pulsatile component (called the DC component) is related to the light absorbed by the other tissues, such as connective tissue, bone, venous, and capillary blood. From this principle, the perfusion index (PI) was derived, which is the ratio of the pulsatile to the non-pulsatile components of the plethysmographic waveform, reflecting the quality of skin perfusion^[4] (Figure 1). Schematically, PI can be considered as the ratio of stroke volume to vascular tone. Assuming vasomotor tone remains constant, changes in PI should reflect changes in stroke volume and cardiac output. Its potential relevance for intensivists, therefore, appears evident.

Figure 1.

Photoplethysmography principles and perfusion index calculation.

LED: Light-emitting diode; SpO₂: Oxygen saturation.

In this journal, Aissaoui et al.^[5] explored this concept by investigating the utility of PI in assessing preload dependency. Based on the relationship between stroke volume and PI, the authors hypothesized that PI could serve as a reliable marker to detect fluid responsiveness. They included 50 critically ill patients with acute circulatory failure. The population study was heterogeneous but predominantly comprised patients with septic shock and postoperative care. At inclusion, the majority (60%) presented with a vasoplegic hemodynamic profile, followed by a hemorrhage profile in 22% of cases. A 500-mL crystalloid fluid challenge was administered. Baseline PI values were similar between responders and non-responders (respectively, median PI=0.83, interquartile range [IQR]: 0.36–1.41 vs. median PI = 0.65, IQR: 0.40–1.57; P = 0.905) to fluid challenge. In addition, after the fluid challenge, PI values did not differ significantly between the two groups. In contrast, the relative change in PI was significantly greater in responders compared to non-responders (respectively, median PI variation = 60 [IQR: 17% – 108%] vs. 1 [IQR: −28% to 31%]; P=0.001). The diagnostic performance of PI to detect fluid responsiveness was moderate as the area under the curve (AUC) was 0.778 (95% confidence interval [CI]: 0.638 to 0.883, P = 0.004). However, one-third of patients were misclassified when PI variation was assessed against stroke volume variation, as shown in the four-quadrant plot. Importantly, another one-third of patients fell within a “gray zone,” where fluid responsiveness could not be reliably determined. From a clinical perspective, this proportion remains a concern, especially given the well-known adverse effects of fluid overload. However, it should also be recognized that the test allowed reliable classification in about 70% of patients.

How do these results align with the prior findings in the literature? Beurton et al.^[6] investigated the ability of PI variation during a passive leg-raising (PLR) maneuver to predict preload dependency in critically ill intensive care unit (ICU) patients, predominantly under mechanical ventilation. An increase in PI >9% induced by PLR predicted an increase in cardiac output >10% induced by PLR with an AUC of 0.89 (95% CI: 0.80 to 0.95). Another study showed that the increase in PI induced by an end-expiratory occlusion test reliably detected a positive response to PLR with an AUC of 0.95±0.03 in mechanically ventilated ICU patients.^[7] Several factors may explain the discrepancies between these findings and those of Aissaoui et al.’s study. First, different devices were used to measure PI. Indeed, in Aissaoui et al.’s study, the plethysmographic module of the Mindray Beneview T6 monitor (Masimo Corp., Irvine, CA) was used, whereas the pulse oximeter Radical-7 device (Masimo Corp., Irvine, CA) was used in the other studies.^[6,7] These devices may have different signal processing characteristics, such as stability, averaging time, and refresh rate. Reported values of PI precision have varied. For instance, 9% with the Mindray Beneview T6 monitor and 1.2% with the Radical-7 device,^[6,7] although these findings do not stem from a direct head-to-head comparison. Differences in precision may also be influenced by patient or study characteristics. Second, in the Aissaoui et al. study, only a single real-time PI value was recorded without temporal averaging, whereas in the other studies, the real-time PI values were averaged over 12 s. ^[6,7] Relying on a single instantaneous measurement may have introduced error and affected the validity of the results. Third, different methods were used to assess fluid responsiveness.

However, discordant results were observed in other studies. Indeed, changes in PI were not associated with fluid responsiveness status after volume expansion in ICU patients.^[[8], [9], [10]^] This can be explained by the fact that the relationship between PI and stroke volume is not straightforward and can also be influenced by other factors such as vascular tone and venous blood. A reduction in venous blood flow may cause the accumulation of venous blood in the fingers, leading to an increase in the non-pulsatile component and ultimately a decrease in PI, independent of changes in arterial blood flow. Some studies have found no or weak correlations between changes in PI and changes in stroke volume after PLR or volume expansion.^[6,9] Several explanations may account for this. The key point lies in the assumption that macro- and microcirculation are perfectly coupled. In these studies, stroke volume changes were evaluated following dynamic maneuvers such as fluid challenge or PLR, with the expectation that microcirculatory parameters would vary proportionally. However, PI reflects the microcirculation at the site of measurement–hence its name, “perfusion index.” A dissociation between macro- and microcirculation is common in acute circulatory failure and becomes more pronounced in septic shock.^[11,12] Indeed, Jacquet-Lagrèze et al.^[13] found no significant correlation between the changes in capillary refill time (CRT) and changes in microcirculatory variables induced by volume expansion in patients with acute circulatory failure. Similarly, Fage et al.^[14] reported that, in septic shock patients, CRT improved after volume expansion in some patients without an increase in cardiac output, whereas it failed to decrease in others despite the rise in cardiac output induced by volume expansion. Additionally, CRT was not an independent predictor of CRT change.^[14] These findings suggest that in some cases, no improvement in tissue perfusion occurs despite a significant increase in cardiac output, which could be attributed to the dissociation between macro- and microcirculatory variables. The effect of fluid challenge on microcirculation appears to be even more complex when considering the early or late phases of septic shock.^[15] It is important to emphasize that an improvement in PI following a fluid-induced increase in cardiac output may indicate that the rise in cardiac output has translated into enhanced tissue perfusion, a relationship that is not always observed.

The ANDROMEDA-SHOCK trial provides a clear illustration: although macrocirculation was optimized, resuscitation guided by a microcirculatory variable (CRT) significantly improved mortality in patients with septic shock.^[16,17] In other words, a microcirculatory tool (PI) has been evaluated against a macrocirculatory standard (stroke volume) under the assumption that the two are always coupled. In conclusion, the dissociation between macro- and microcirculation may account for the moderate to poor diagnostic performance of PI in predicting or detecting fluid responsiveness reported in this and other studies. Of note, an improvement in PI after fluid administration without a concomitant increase in cardiac output should not be regarded as a false positive or a limitation, as it may still confer clinical benefit.

Acknowledgments

None.

Funding

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Ethics Statement

Not applicable.

CRediT authorship contribution statement

Osama Abou-Arab: Writing – original draft. Christopher Lai: Writing – review & editing. Jihad Mallat: Writing – review & editing, Conceptualization.

Conflict of Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Managing Editor: Jingling Bao/ Zhiyu Wang