Risk factors for pulmonary complications after hepatic resection: role of intraoperative hemodynamic instability and hepatic ischemia

Liver resection is an increasingly common surgery and remains the main metastatic colorectal disease and hepatocarcinoma treatment [1]. Despite technical advances and high experience of liver resection of specialized centres, it is still burdened by relatively high rates of postoperative morbidity and mortality [2]. Postoperative pulmonary complications (PPCs) were the most frequent complication in this series than in others, where the prevalence can reach up to 50% of the patients [3‐7].

Many risk factors for PPCs have already been identified, such as advanced age, smoking status, history of chronic respiratory obstructive disease, ASA classification and incision site [8, 9]. These factors have been established for general elective surgery or upper abdominal procedures. Many fewer studies have specifically analysed the risk factor for PPCs after hepatectomy for cancer [4].

The clamping of the liver is a technique permitting decreasing intraoperative blood loss and thus perhaps decreasing the postoperative complications. However, this technique is followed by hepatic ischemia and then possibly lung injury by ischemia reperfusion syndrome. Although the role of liver and gut ischemia reperfusion has been well described for more than 20 years in animals [10, 11], few studies have specifically analyzed this phenomenon in humans. Thus, the aim of this study is to determinate the pre- and intraoperative risk factors of PPCs after hepatic resection, and especially to determine the role of hepatic ischemia, pre-operative biologic blood tests and the necessity of intraoperative vasopressive drugs in PPCs.

Between January 2013 and December 2013, 94 consecutive patients underwent hepatectomy. In seven patients, laparoscopic surgery was performed: in four patients, minor hepatectomy and in 3 patients major hepatectomy. All patients presenting mean arterial pressure less than 65 mmHg for more than 5 min, despite adjusted volemia, and in absence of surgical venous obstruction received administration of Vasopressive drugs.

The ability to predict and treat early postoperative morbidity after elective liver resection would allow individualized postoperative management. In this prospective one-year study, we identified three independent risk factors for PCCs after hepatic resection: prolonged time of ischemia, the use of vasopressor and elevated preoperative GGT.

Firstly, it is shown that prolonged time of ischemia during a hepatectomy is a risk factor of PCC within 7 days of postoperative. In our study, the OR for time of ischemia corresponds to an increase in the risk with every extra minute of ischemia. Patients with PPC had undergone longer time of hepatic ischemia (Fig. 1) and the longer the time of hepatic ischemia the greater was the rate of PPC (Fig. 2). The mechanism explaining the link between hepatic ischemia and PPCs could be the massive release of reactive oxygen species factors and cytokine production. This process has been demonstrated in experimental studies with kidneys and extra renal organs such as the lung [19, 20]. Indeed, it has been shown that acute kidney injury leads to a pro inflammatory and pro apoptotic pathways activation which then leads to an inflammatory response in the lung, then to ARDS. This phenomenon has been also described for the liver in an experimental study: liver ischemia-reperfusion (I/R) increases pulmonary permeability in rats [10] and induced pathological changes in lung parenchyma (alveolar septal thickening, leukocyte infiltration, hemorrhage and pulmonary edema) in rabbits [21]. Because the lung function depends directly on the cell membrane integrity of the alveolar-capillary network of the lung, whose structure is particularly sensitive to factors related to I/R, as the decrease in cellular energy levels and the action of the reactive oxygen species, liver ischemia reperfusion can infer pulmonary edema and contribute to acute respiratory distress syndrome (ARDS).

Moreover, hepatic clamping results in venous inflow occlusion from the gut and therefore leads to gut ischemia reperfusion. Collange et al. [22] showed that gut ischemia increases lactate, cytokine IL10 IL6 in systemic blood. IL-10 was mainly found in the lung, suggesting the interaction between lung inflammation and gut ischemia-reperfusion. Karsten Bartels et al. [23] also show the effects of gut ischemia to the cellular scale. Moreover Ben-Abraham et al. [11] demonstrated acute lung injury induced by mesenteric artery clamping/unclamping. What’s more, experimental studies have demonstrated that a chemical medication could prevent lung injury in hepatic ischemia reperfusion: Uchiyama et al. [24] proved that Edavarone, a potent free-radical scavenge, could decrease lung injury in rats by attenuating oxidative stress response. This could maybe open the possibility of preventing PPCs after elective liver surgery.

Secondly, the use of vasopressor has been shown as a risk factor of PPC within 7 days of postoperative. The association between vasopressor and PPC could have two origins: first, the cause of administration of these drugs by the anaesthesiologist, hemodynamic instability, could be the cause of PPC: hemodynamic instability, whatever is its cause, could lead to inadequate lung perfusion in the intraoperative time and then to PPCs [25]. Second, vasopressor could have per se a deleterious effect. In a recent survey, authors [26] described the intraoperative fluid and pharmacologic management during liver transplantation in the USA. One of the conclusions was that the use of vasopressors was even less since the liver transplantation program was efficient.

In a recent large study analyzing pre- and intraoperative risk factors for ARDS after liver transplantation, authors showed that large intraoperative bolus of vasopressors were the sole intraoperative risk factor [27]. That points out the dilemma of the physician: such techniques as hepatic clamping or vasopressive drugs have a very interesting effect on the intraoperative blood loss and the quality of surgery it allows (with probably less postoperative surgical complications and a better treatment of the cancer). However hepatic clamping such as the use of vasopressors could have deleterious effect on the liver but also on other organs such as the lung. That should focus our efforts on patient-centered outcomes, which are clearly more important than process outcomes [28].

Thirdly, the third independent risk-factor for PPCs identified in our study was an elevated preoperative concentration of GGT. Because GGT and APL, are very sensitive parameters of liver diseases [29], it was rational to study if preoperative GGT could be predictive of PPCs. In a large cohort of 278,419 patients, recorded over 7 years, Sung KC et al. [30] have shown that GGT > 35 UI/l was an independent factor of mortality, not regarding the presence of fatty liver. In this Korean cohort, there was no significant association between ALT and all causes of mortality. Preoperative GGT has not often been studied in liver surgery. Rau HG et al. [31] analyzed the postoperative risk of liver failure in a series of 570 patients: preoperative GGT was one of the three independent parameters predicting liver failure. Because of the small number of patients with postoperative liver failure in our series, the preoperative GGT as a risk factor for liver failure could not be analysed. However, GGT and not transaminases, as for liver failure in the Rau’s study, was a risk factor for PPCs. The direct role of the liver in the association between elevation of GGT and postoperative morbidity could however be discussed. Indeed, some authors reported the association between baseline levels of GGT and specific cardiovascular complications [32‐34] in the general population.

Also, the rate of PPC in our study was higher than other studies [4, 35, 36]. This can be explained by our large definition of pulmonary infection and the fact that we wanted specifically to study PPCs. Indeed, we defined pulmonary infection with large clinical criteria such as fever, expectoration of mucus and the presence of a suspect image on thoracic radiography. One criteria we used for PPCs is “oxygen therapy >3L O2/min on POD2”, that is not present clearly in Dindo-Clavien classification [37], could significantly increase the number of patients with PPC. In fact this criteria could be considered as Grade 2 complication of the Dindo-Clavien Classification, because, in our current practice most patient did not need oxygen therapy on POD 2. This was present in the criteria definition recommended by an ESA-ESICM statement [38].

It could be asked in what way the results of our study are dependent of the choice of the definition of PPCs. To explore this aspect, we have performed a sensitivity analysis where “oxygen therapy >3L O2/min on POD2” was not considered as a criterion for having at least one PPC. All the results can be found in Additional file 1. In short, according to this second definition, there would be 21 (22.3%) patients with PPCs instead of 32. Univariate comparisons by the presence or absence of PPCs would show results closed to the main analysis: four characteristics that were not found to be associated with PPCs would become associated (diabetes, portal hypertension, steatosis and etiology), and two characteristics that were associated to PPCs in the main analysis would become not associated (major hepatectomy and clamping by TVEL). All the others associations would remain the same. After performing multivariate logistic regression with the same procedure and criteria for model selection as the main analysis, the model that would be retained is a model with the same characteristics as the main analysis (GGT, ischemia duration and use of vasopressive drugs) alongside with two more characteristics (cardiopathy and non-cancer hepatopathy). To summarize, this sensitivity analysis confirms the three characteristics retained in the main analysis, but allow discussing the role of two other factors.

Finally, obesity, defined in our article by BMI > 30, was not demonstrated as a risk factor. This result could be compared with that of Cucchetti et al. [39]. COPD, like in the study of Nobili et al. [4] was not an independent risk factor and diabetes, in contrary of that study was also not an independent risk factor. The cause of this discrepancy could be the lack of power of our study, where the number of patients presenting COPD or diabetes was low. Another cause could be that the pathophysiology of PPCs after liver surgery was first liver ischemia reperfusion induced lung injury and not direct pulmonary infection. This hypothesis should be confirmed by a larger prospective study.

PPC remains a frequent postoperative complication. Our study allowed identifying three independent risk factors of PPC: prolonged time of ischemia, the use of vasopressor and elevated preoperative GGT. Preoperative GGT is not controllable. However, use of vasopressors for correction of all episodes of arterial hypotension and the time of hepatic ischemia are completely decided upon by the medical team. The necessity of hepatic clamping is decided after surgical evaluation and the actual evolution is to minimize it or to prefer the intermittent. Concerning the treatment of intraoperative hypotension, definition of hypotension requiring vasopressors and the benefit/risk of such treatment should be probably finely defined in a randomized trial.