Visual quality assessment of the liver graft by the transplanting surgeon predicts postreperfusion syndrome after liver transplantation: a retrospective cohort study

According to the Organ Procurement and Transplantation Network of the United States (US) Department of Health and Human Services, the numbers of liver transplantations (LT) in the US have constantly been rising since the beginning of LT [1]: In 2016, a total of 7841 LTs were performed in the US. By contrast, the numbers of LTs in Germany and in the Eurotransplant region have been decreasing since a peak in 2010 [2, 3]. The decreasing organ donations combined with the persisting high morbidity and mortality of patients on the waiting list has led to a discrepancy between organ supply and demand [2, 4] and therefore to the more frequent acceptance extended criteria donors (ECD) to the pool of eligible donors [5, 6].

The most critical intraoperative moment for the anesthesiologist during LT is the reperfusion of the liver graft. An immediate and severe complication following reperfusion is hemodynamic instability, the so called post-reperfusion syndrome (PRS). PRS is defined as a decrease of mean arterial pressure (MAP) of more than 30% during the first 5 min after reperfusion and continuing for at least 1 min [7‐12]. PRS occurs in around 10–60% of LT recipients [8, 9, 13]. Several risk factors for PRS have been described: Older donor age, higher donor risk index, longer cold ischemic time (CIT), severity of the recipient’s liver disease, operation time and technique, hemodynamics at time of reperfusion, and steatosis of the graft organ [8, 13‐20]. The mechanisms of PRS appear to be complex and not fully understood [10, 21, 22]. In addition, the transplantation of ECD livers has led to a decrease in mortality for LT recipients on the wait lists at the cost of increased perioperative complication [23]. Both these factors have made it difficult to predict PRS. In particular, the quality of the donor organ and its role as risk factor for the occurrence of PRS as well as its associations with patients’ outcome have been neglected in the past.

In this single center retrospective analysis of 122 LT recipients, we found that approximately half of the patients suffered from PRS (53.3%). Multivariable regression analysis identified two predictors of PRS: Only visually assessed acceptable (OR 12.2, P = 0.002) and poor (OR 13.4, P = 0.02) donor organ quality by the transplanting surgeon and intraoperative norepinephrine infusion rate before reperfusion of the liver transplant (OR 2.2 per 0.1 μg/kg/min, p = 0.01) were associated with the occurrence of PRS. LT recipients suffering from PRS were hemodynamically more instable after reperfusion compared to recipients not suffering from PRS. They had lower MAPs from reperfusion until the end of surgery (P < 0.001). Hence, they received more epinephrine and norepinephrine before reperfusion (P < 0.001 and P = 0.01, respectively) and higher rates of continuous infusion of norepinephrine (P < 0.001) and vasopressin (P = 0.02) after reperfusion of the transplant. Moreover, LT recipients with intraoperative PRS had a more complicated postoperative course compared to recipients without intraoperative PRS: They had more adverse cardiac events (P = 0.05) and suffered more often from postoperative delirium (P = 0.04).

All 122 LT recipients were operated at the same center with the same technique (venovenous/portalvenous bypass) and with uniform anesthesiologic management, guided by a SOP that included instructions for the management of PRS. No changes in operating technique or anesthesiologic management were made during the study period, leading to a homogenous single center study sample. Nevertheless, this study has several weaknesses. Due to the retrospective design of the analyses, data quality could be impaired. Histological data for our analyses was limited to steatosis assessment. Analyses of hemodynamics from hand-written anesthesia protocols are often suspected to present “sugar-coated” hemodynamic values. Although we cannot rule out that this was the case, our data present clinically and empirically plausible hemodynamics. At worst, hemodynamic stability is overestimated and the effects shown are even more pronounced. The fact that this was a single center analysis combined with the uncommon but consistent use of intraoperative venovenous/portalvenous bypass limits the external validity of our results.

Although rather on the top end of the range, PRS incidence of 53% in our sample concurs with several other studies [8, 10, 12, 17, 19]. The pathophysiology of PRS is complex and not entirely understood. The abrupt influx of cold, hyperkalemic and acidic blood into the circulation, air or thrombotic embolization and the release of vasoactive substances from the graft liver contribute to PRS [29‐31]. The rationale for using a venovenous/portalvenous bypass during the anhepatic phase at our center is to ensure maximum safety of the procedure [24]. The combination of a femoro-brachial and porto-axillary bypass reduces lower limb and mesenterial congestion and therefore reduces the abrupt influx of hyperkalemic and acidic blood into systemic circulation at the time of reperfusion. Nevertheless, the use of a bypass during LT surgery remains controversial [13, 32, 33]. The PRS effect seen in our study is therefore likely to be predominantly caused by vasoactive substances released from the graft. With that in mind, PRS incidence does appear rather high.

A possible explanation for this contradiction may be the fact that we used an extended definition for PRS. PRS is commonly defined as a decrease in MAP of more than 30% from the baseline value for more than one minute during the first five minutes after reperfusion [7, 8, 10‐12]. We extended the definition for PRS for mainly two reasons. First, the treatment of PRS seeks to preserve hemodynamic stability. At the time of reperfusion, anesthesiologists expect a certain degree of hemodynamic instability and preemptively treat a (soon to be) falling MAP with catecholamines [11], either as bolus or by increasing continuous infusion of the very same. Since this preemptive treatment conceals the occurrence of by-definition-PRS, we added (i) the administration of an intravenous bolus of norepinephrine > 2 μg kg (BW) ^-1 as well as (ii) the increase of continuous norepinephrine infusion of ≥0.1 μg kg BW^− 1 within the first 5 min after reperfusion as new criteria to the definition. Second, our clinical empiricism and the literature have presented cases of prolonged vasodilation after reperfusion [34‐36]. We thus included (iii) an increase of continuous norepinephrine infusion up to 30 min after reperfusion and (iv) the initiation of continuous vasopressin infusion after reperfusion as additional criteria for PRS. This altered definition impedes comparability with other studies and may overestimate PRS incidence compared to the commonly used definition. However, our data appear clinically plausible and concur with description of hemodynamic recovery after PRS by other authors [17].

In this study, multivariable regression modelling identified one donor related and one recipient related predictor of PRS: graft quality – visually assessed by the transplanting surgeon – and cumulative intraoperative norepinephrine dosage before reperfusion of the liver graft. The variety of risk factors and risk prediction models described in different studies suggests that PRS may occur in an unpredictable manner or may be highly subjective to center-specific effects [8, 13‐20, 22]. The most frequent reported risk factors for PRS are longer CIT [15, 16, 18, 20] and intraoperative hemodynamics prior to reperfusion [14, 15, 17, 20], followed by graft steatosis [8, 16], operation time and technique [15, 18], severity of recipient’s liver disease [8, 19], and donor age [13, 17]. One single center study retrospective analysis was unable to identify even a single risk factor in a retrospective analysis of a sample of 261 LT recipients [37]. Our findings that graft quality and hemodymic impairment prior to reperfusion are risk factors for PRS concur with several similar studies. However, it remains cryptic why the multitude of these observational studies generate such a diversity of findings when it comes to predicting postreperfusion syndrome. Prospective multi-center trials, the creation of an LT register or meta-analyses investigating the prediction of PRS could help to shed light on this matter in the future.

The increasing number of liver transplantations has led to an increased demand of donor organs [1] and to the growing use of organs from extended criteria donors (ECD) [23]. A common quality marker for organs is fatty change, or steatosis. Steatosis has indeed been linked to the occurrence of PRS [8, 16]. Studies have shown that transplantation of liver grafts with moderate to severe steatosis can very well be conducted without sacrificing recipient survival but in particular, that increased efforts are needed when dealing with these organs [26, 38]. However, macroscopic assessment of the organ may not be qualified to assess steatosis: Rey and colleagues have examined 36 livers of organ donors which were explanted but not allocated. The authors found that macroscopic appearance and coloring compared to histopathologic evaluation for steatosis unveiled discrepancies: Most of the graft livers with histopathologically confirmed mild degree of steatosis macroscopically appeared yellow, thus of false poor quality and could have been transplanted. In our analyses, microscopic steatosis evaluation was a worse predictor for PRS than the visual evaluation of the transplanting surgeon. This may be the case because surgeons likely assess the organ not only by its color, but also by general appearance, organ consistency, and size. It may even be the case that information of the recipient’s medical history and the knowledge that the organ was already rejected once by another center or had a long CIT additionally influenced the assessment. Therefore, the multitude of information may have enabled the surgeon to predict PRS better than any single variable (recipient status, CIT, steatosis, etc.).

We also found intraoperative norepinephrine infusion rate before reperfusion to be a predictor of PRS. This also concurs with other studies that have identified markers of hemodynamic instability prior to reperfusion as predictors of PRS. One possible explanation for norepinephrine infusion or hemodynamic instability is hypovolemia. This would also concur with other studies that have shown lower central venous pressure at time of dissection [14] or at time of reperfusion [17] as well as a higher requirement for transfusion units [15] to be predictors of PRS. A second possible explanation is vasodilatation or vasplegia which has been linked to end stage liver disease [39, 40]. Case reports have described vasoplegic syndrome to occur during liver transplantation [31, 41]. However, differentiation between the two can sometimes be complicated as both appear alike [34]. A third possible explanation may be that these patients already had a compromised cardiac function. E.g., Xu and colleagues could link left ventricular diastolic dysfunction to the occurrence of PRS in a cohort of 330 Chinese LT recipients [20]. Unfortunately, our data did not contain systematic and detailed cardiac function assessment of the recipients and the literature lacks systematic investigation of its impact on the occurrence of PRS. Moreover, it has been suggested by several authors that PRS could be only a sign of an underlying problem. It could be that anhepatic hypovolemia, vasoplegia and/or impaired cardiac function demask as PRS intraoperatively. On top of it, patients receiving a LT from ECD appear especially susceptible to severe PRS. In this light, these patients should especially be optimized regarding hemodynamics and acid-base balance when awaiting reperfusion.

Similar to the prediction of PRS, data on the effect of PRS on LT recipients’ outcome is inconclusive. In particular, the impact of PRS on hard clinical endpoints such as primary graft nonfunction with requirement for retransplantation and mortality is unresolved. E.g., some studies show an effect of PRS on mortality [15, 18‐20] and others did not [13, 17]. In fact, we were not able to show a significant association of PRS with primary non-function (4.6 vs. 1.8%) and mortality (15.4% vs. 12.3%) in this study population. This is most likely due to the small sample size, again underscoring the need for multicenter registries. Regarding post-transplantation morbidity, we could demonstrate in our sample that the occurrence of PRS was associated with a higher frequency of postoperative delirium (12.3% vs. 1.8%) and major cardiovascular events (16.9% vs. 5.3%). While other studies have demonstrated early allograft dysfunction [42, 43], ICU length of stay [43] and the need for postoperative renal dysfunction [19, 20] we could not substantiate these associations in our study sample.

The data of this study demonstrated that both the visual assessment of the liver graft by the transplanting surgeon as well as the intraoperative recipients need for catecholamines before reperfusion of the graft were predictors for PRS at our center. Besides the known risk factors for PRS, anesthesiologists should keep these newly described risk factors in mind when preparing for reperfusion. As the impact of PRS on short- and long-term hard clinical endpoint remains inconclusive, further investigations and possibly multicenter prospective registries could ultimately resolve the impact of PRS on LT recipients’ outcome.