Functional Phenotyping: A New Role for Electrical Impedance Tomography

Pulmonary postoperative complications (PPCs) are common after surgery, and their occurrence may be related to several factors, including the alterations determined by general anesthesia and mechanical ventilation (1). Patient characteristics, such as body mass index (BMI) and age, have been associated with an increased occurrence of PPCs, but the physiological link between predictors and outcome is, in most of the cases, missing (2). Conventional imaging techniques, such as chest X-ray or lung computed tomography scan, are used to confirm diagnosis of PPCs such as pneumonia, atelectasis, or pleural effusion but, to date, do not provide help in the prediction or stratification of the mechanisms of risk.

Electrical impedance tomography (EIT) has been increasingly used in the last two decades at the bedside to evaluate ventilation (3) and perfusion distribution (4, 5). This information, as any clinical data, can be potentially used for different purposes (6): to understand the pathophysiology of a condition, prevent a disease, increase the efficacy of the diagnosis, or modify the treatment (7). Nevertheless, if we consider the parameter derived from EIT as stand-alone “biomarker,” the possibility of use of this technology can be pushed forward to new possible applications.

In this issue of the Journal, Iwata and colleagues (pp. 1328–1337) used EIT as a functional biomarker to identify postoperative clusters of patients, named phenotypes, with different incidence of PPCs in a population of high-risk (Assess Respiratory Risk in Surgical Patients in Catalonia ⩾ 45) patients admitted routinely to the ICU after surgery (8).

Of the three phenotypes, patients with inhomogeneous ventilation, both mostly ventral (phenotype 1) or dorsal (phenotype 3), showed higher incidence of PPCs and delayed weaning from mechanical ventilation. Of note, no alterations were found in these patients on the chest X-ray done at ICU admission.

The study shows an interesting and novel approach to EIT—not used to guide a specific intervention during or after surgery but extended outside of the operating theater to identify groups of patients potentially at risk. Nevertheless, a central question raised by this paper is what the meaning of the phenotype is and its impact on clinical outcomes.

One possibility is that a specific phenotype represents a single trajectory of a lung disease determined by other coexistent factors. An example could be the higher BMI found in phenotype 1 (i.e., the group of patients with ventral ventilation and, therefore, dorsal collapse). If this hypothesis is confirmed, EIT may help in identifying a pathophysiological connection between risk factors and complications, showing the casual pathway between exposure (hypoventilation due to higher BMI) and outcome (atelectasis and pneumonia).

A second option could be using EIT to titrate ventilation to modify the phenotype, for example increasing positive end-expiratory pressure (PEEP) in phenotype 1. Nevertheless, the study by Iwata and colleagues does not demonstrate if, by addressing ventilatory inhomogeneity, it is possible to modify the incidence of PPCs. This aspect remains unclear and needs to be explored in future studies. A previous study (9) showed that using EIT to optimize PEEP based on the best compromise between lung collapse and overdistension reduced driving pressure and the occurrence of postoperative atelectasis, but the study did not explore longer-term PPCs.

The analysis of functional phenotypes opens new future perspectives, for example allowing us to identify patients who would benefit from respiratory support who may have failed to demonstrate clinical relevance in unselected populations. Iwata and colleagues evaluated phenotypes in the postoperative setting, but, potentially, earlier patient evaluation (e.g., before anesthesia induction) could further improve tailored ventilation.

Recently, Jonkman and colleagues (3) revealed in patients with acute respiratory distress syndrome how EIT can assess lung recruitability, which might go otherwise unrecognized by more commonly used means. Translating the recruitability concept to perioperative medicine, patients with phenotype 1 may be further classified as recruitable or not recruitable. This aspect, and the clinical management of the two subphenotypes, should be taken into account. Moreover, we do not know if these phenotypes can be confirmed when intraoperative ventilation is titrated using advanced respiratory monitoring (6) or in other surgical contexts, such as thoracic or bariatric surgery.

Another issue is how to manage patients with phenotype 3 (i.e., patients with ventilation predominantly distributed in the dorsal lung). In these patients, nondependent silent spaces (5) were higher, meaning possible overdistension of the nondependent regions. How to address patients with overdistension at low PEEP levels (median value, 5.4 cm H₂O) remains an issue. Although EIT can in this case represent an incentive to reduce PEEP (10), body positioning or chest-wall loading may be interesting options to redistribute ventilation while still applying a minimal amount of PEEP (11).

Finally, evaluating perfusion using EIT (12) could potentially increase the strength of phenotyping, also considering the impact of anesthesia and mechanical ventilation on ventilation/perfusion matching (13). According to these physiological endpoints, the perfusion measured using EIT could be an adjunctive bedside tool to assess the modification of the ventilation/perfusion mismatch after ventilatory adjustments (14).

In conclusion, in this elegant analysis, the authors revealed, through EIT, specific functional biomarkers, expanding the use of this advanced technology not only to diagnosis and treatment but also to potential prevention. This novel approach for EIT can pave the way to “functional” biomarker-guided therapies and strongly consolidates EIT as a reference technique for personalized medicine in mechanically ventilated patients.