Gamma-glutamyltransferase is a strong predictor of secondary sclerosing cholangitis after lung transplantation for COVID-19 ARDS

Background Lung transplantation (LTx) can be considered for selected patients suffering from COVID-19 acute respiratory distress syndrome (ARDS). Secondary sclerosing cholangitis in critically ill (SSC-CIP) patients has been described as a late complication in COVID-19 ARDS survivors, however, rates of SSC-CIP after LTx and factors predicting this detrimental sequela are unknown. Methods This retrospective analysis included all LTx performed for post-COVID ARDS at 8 European LTx centers between May 2020 and January 2022. Clinical risk factors for SSC-CIP were analyzed over time. Prediction of SSC-CIP was assessed by ROC-analysis. Results A total of 40 patients were included in the analysis. Fifteen patients (37.5%) developed SSC-CIP. GGT at the time of listing was significantly higher in patients who developed SSC-CIP (median 661 (IQR 324-871) vs 186 (109-346); p = 0.001). Moreover, higher peak values for GGT (585 vs 128.4; p < 0.001) and ALP (325 vs 160.2; p = 0.015) were found in the ‘SSC’ group during the waiting period. Both, GGT at the time of listing and peak GGT during the waiting time, could predict SSC-CIP with an AUC of 0.797 (95% CI: 0.647-0.947) and 0.851 (95% CI: 0.707-0.995). Survival of ‘SSC’ patients was severely impaired compared to ‘no SSC’ patients (1-year: 46.7% vs 90.2%, log-rank p = 0.004). Conclusions SSC-CIP is a severe late complication after LTx for COVID-19 ARDS leading to significant morbidity and mortality. GGT appears to be a sensitive parameter able to predict SSC-CIP even at the time of listing.

Gamma-glutamyltransferase is a strong predictor of secondary sclerosing cholangitis after lung transplantation for COVID-19 ARDS RESULTS: A total of 40 patients were included in the analysis. Fifteen patients (37.5%) developed SSC-CIP. GGT at the time of listing was significantly higher in patients who developed SSC-CIP (median 661 (IQR 324-871) vs 186 (109-346); p = 0.001). Moreover, higher peak values for GGT (585 vs 128.4; p < 0.001) and ALP (325 vs 160.2; p = 0.015) were found in the 'SSC' group during the waiting period. Both, GGT at the time of listing and peak GGT during the waiting time, could predict SSC-CIP with an AUC of 0.797 (95% CI: 0.647-0.947) and 0.851 (95% CI: 0.707-0.995). Survival of 'SSC' patients was severely impaired compared to 'no SSC' patients (1-year: 46.7% vs 90.2%, log-rank p = 0.004). CONCLUSIONS: SSC-CIP is a severe late complication after LTx for COVID-19 ARDS leading to significant morbidity and mortality. GGT appears to be a sensitive parameter able to predict SSC-CIP even at the time of listing. J Heart Lung Transplant 2022;41:1501−1510 Ó 2022 The Author(s). Published by Elsevier Inc. on behalf of International Society for Heart and Lung Transplantation. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/) Lung transplantation (LTx) is an established treatment option for end stage chronic lung diseases. In addition, LTx has recently been established as a last resort in patients suffering from acute respiratory distress syndrome (ARDS) who do not recover despite several weeks of extra-corporeal membrane oxygenation (ECMO). 1 The detrimental coronavirus disease 2019 (COVID- 19) pandemic has resulted in an increase of ventilator-and ECLS-dependent ARDS cases worldwide. As a result, ARDS has become an important indication for LTx. In recent reports, it represented 7-29% of the transplant volume in institutions offering transplantation for ARDS. [2][3][4] Secondary sclerosing cholangitis in critically ill patients (SSC-CIP) is a rare complication following ARDS and entails severe morbidity and mortality. It is marked by increased total bilirubin (TBi), gamma-glutamyltransferase (GGT) and alkaline phosphatase (ALP) reflecting cholestatic liver injury. 5 SSC-CIP has been recognized as an underdiagnosed and underreported disease entity associated with intensive care treatment in different settings such as sepsis, shock, trauma and burns. 6,7 It can affect the intraand extrahepatic biliary tree and usually evolves rapidly to biliary cirrhosis. The prognosis of SSC-CIP is poor. 8 In few cases, the disease can be stabilized by medical and endoscopic treatment, however the most severe cases require liver transplantation, when possible, in the overall clinical context. Without liver transplant, survival is limited to 55% at 1 year and 14% at 6 years. 9 The median survival for SSC-CIP is only 13 months, and thereby significantly lower than median survival rates of other forms of SSC. 10 Several risk factors of SSC-CIP have been proposed, including sepsis/SIRS, ischemia/hypoxia, prolonged prone positioning and abdominal obesity, ECMO and potentially hepatotoxic medication. 6,11,12 In critically ill COVID-19 patients, liver injury has been a widely reported extrapulmonary manifestation with deranged liver parameters in 10-58% of hospitalized patients. 13,14 Severe cholestatic liver dysfunction resembling SSC-CIP has also been observed in COVID-19 patients without the need for ECMO treatment. [15][16][17][18][19] Several studies have shown a direct damage of SARS-COV-2 on the small bile ducts. 20,21 LTx for post-COVID-19 ARDS has shown promising short-and mid-term survival. 2,22 Several centers have reported liver damage with features of SSC-CIP as a late complication in these patients, but comprehensive data are still lacking. In this multicenter-study, we therefore aimed to (i) analyze the incidence of SSC-CIP in patients after LTx for COVID-19 ARDS and to (ii) explore predictive factors which could aid in patient selection and therefore avoid this severe complication.

Patients and methods
We retrospectively analyzed double LTx performed for post-COVID-19 ARDS within the Eurotransplant (ET) region between May 2020 and January 2022. The ET area covers 138 million inhabitants across 8 countries and currently has an annual lung transplant activity of around 1200 LTx in 22 centers. Eight of 22 ET-LTx centers performed LTx for post-COVID-19 ARDS patients within the study period. All 8 agreed to participate in this study and contributed patient data for transplanted patients for this analysis. Ethics approval was granted by the institutional review board of the Medical University of Vienna (EK-Nr 1528/2021) and the participating institutions.
The cohort was divided into 2 groups. Group 'SSC' included patients with clinically suspected SSC or SSC proven by endoscopic retrograde cholangiopancreatography and/or magnetic resonance cholangiopancreatography. 23 All other patients formed group 'no SSC'.

Management of COVID-19 ARDS patients
In all centers, the primary ECMO configuration of choice was VV ECMO either in femoro-femoral or femoro-jugular configuration or a double-lumen cannula. The type of cannulation was dependent on the preference of the treating intensive care teams, patient characteristics (thrombosis, anatomical situation) or expectation to achieve awake bridging. VA or VVA configurations were employed in case of additional hemodynamic instability.
For anticoagulation during ECMO bridging, either subcutaneous low-molecular weight heparin twice daily with a target antiXa at 4 hours of 0.4 to 0.7 or intravenous unfractionated heparin with a target aPTT of 60 to 80 sec was used. In case of heparin-induced thrombocytopenia, argatroban was used.
All centers used lung-protective low tidal volume ventilation strategy as their standard for COVID-19 ARDS patients. This involved volume-limited or pressure controlled ventilation mode aiming for tidal volumes of ≤ 6 ml/kg ideal body weight, driving pressure limited to 15 cm H2O, and a target peak pressure of ≤ 30 cm H2O. Prone positioning was employed by all centers.
All centers aimed to bridge patients to transplantation in an awake. Otherwise, sedative medications included a combination of propofol and remifentanil (mainly employed in the first 7 days), midazolam, sufentanil, esketamine or dexmedetomidine (mostly used after the first 7 days) according to patient requirements.

Donor data
Each center provided basic donor data retrieved from the Eurotransplant donor registry. In addition, the Oto score 24 was calculated for each donor. This scoring system includes donor age, smoking history, chest radiograph assessment, bronchial secretions observed in bronchoscopy and paO2/FiO2 ratio. Lower numbers correspond to favourable donor characteristics.

Recipient data
Recipient data included basic demographic parameters, details on the course of COVID-19 disease and specific treatment, details on mechanical ventilation and ECLS bridging, data of the transplant procedure as well as perioperative data. Liver serum biochemistry parameters at the time of listing and at the time of transplantation were collected. In addition, the maximum pre-as well as posttransplant values were assessed. Liver parameters included TBi in milligrams per deciliter (mg/dL), alanine aminotransferase (ALAT) in units per liter (U/L), aspartate aminotransferase (ASAT) in U/L, GGT in U/L and ALP in U/L.

Outcome parameters
Early recipient outcome analysis included primary graft dysfunction (PGD) grades at T0, T24, T48 and T72 hours, length of post-transplant intensive care unit (ICU) stay and length of total hospital stay. PGD was graded according to the current guidelines of the International Society for Heart and Lung Transplantation. 25 Patients with postoperatively prolonged ECMO support were graded as PGD 3 or PGD 'ungradable' depending on the chest X-ray. Total length of mechanical ventilation was defined as the time to successful extubation without early reintubation (<3 days). In case of tracheostomy, length of mechanical ventilation was defined as the time when the patient tolerated mere oxygen insufflation without any mechanical breathing assistance for more than six continuous hours. Furthermore, postoperative complications and in-hospital mortality were recorded.

Statistical analysis
Statistical analysis was performed in IBM SPSS 26 (IBM Analytics, Armonk, NY). P-values below 0.05 were considered statistically significant. Missing data (only single data points with a random pattern) were appropriately coded and missing cases were excluded from each respective sub analysis. Continuous variables were reported as means § standard deviation or medians with interquartile ranges (IQR) of 25% to 75% and compared using t-tests or Mann-Whitney-U-test according to distribution. Chi-square test or Fisher's exact test used for categorical variables. Comparison of PGD rates as well as in-hospital mortality was performed with Pearson's chi-square test or Fisher's exact test where applicable. Long-term outcome was analyzed by Kaplan-Meier curves and log-rank tests. For parameters significantly different between the 'SSC' group and the 'no SSC' group at listing and for the peak value between listing transplantation, receiver-operating characteristics (ROC) curve analysis was performed and the area under the curve (AUC) calculated. Sensitivity and specificity were determined according to the ROC curve coordinates and optimal thresholds assessed using the Youden index. Figures were created using GraphPad Prism 8 (GraphPad Software, La Jolla, CA).

Donor parameters
Donor characteristics are shown in Table 2. Overall, there were no significant differences in important donor parameters between patients who developed SSC and those who did not. Donor age, cause of death, history of aspiration or trauma and duration of ventilation before explantation were similar in both groups. The SSC group included 2 donors after circulatory death (13.3%) while it was 1 (4%) in the 'no SSC' group (p = 0.545). Oto scores in donors of 'no SSC' recipients with a median of 4 (IQR 2.5-5) were comparable to the SSC group. (median 6 (IQR 3-8)) (p = 0.545). PaO2/FiO2 ratios were similar in both groups (p = 0.740). PaC02 was significantly lower in the SSC group (median 37.8; IQR: 34-42; p = 0.033). ; p < 0.001) were significantly increased. Hepatology was consulted for elevated liver serum biochemistry in one case pre-transplant who developed SSC after transplantation. Since GGT at time of listing and peak GGT during the waiting period were significantly associated with the later diagnosis of a SSC, we determined their individual predictive value with ROC curve analysis ( Figure 2). GGT at the time of listing predicted SSC with an AUC of 0.797 (95% CI: 0.647-0.947). The optimal threshold value was 320 U/L, with a sensitivity of 80% and a specificity of 72%, resulting in a Youden index of 0.52. Peak GGT during the waiting period also provided an excellent prediction with an AUC of 0.851 (95% CI: 0.707-0.995). An optimal cut-off value was 633 U/L, where the sensitivity was 87%, the specificity was 81%, with a Youden index of 0.68. Various sensitivity and specificity values are given in Suppl. Table 2.

Liver serum biochemistry
Outcomes Short-term outcomes were similar between groups (Table 1). At 72 hours after transplantation, most patients had PGD 0 and no PGD 3 was observed at this point. Two patients (8.0%) in the 'no SSC' group were ungradable due to prolonged ECMO while showing clear chest radiographs. Median post-operative length of mechanical ventilation was 15 days in both groups (p = 0.377). Median post-transplant ICU stay was also similar with 34.5 days (IQR 27-52) in surviving 'SSC' and 36 days  in 'no SSC' patients. The diagnosis of SSC was triggered by clinical suspicion due to jaundice and exceedingly elevated cholestasis parameters in 3 patients (20%) but confirmed by ECRP and/or MRCP in the majority of patients. The morbidity and mortality this diagnosis entailed was significant. Four patients (26.7%) recovered, while GGT and ALP still remained elevated. One patient (6.7%) is currently being evaluated for liver transplantation (POD 237), one did not reach transplantation and died on the waiting list (POD 157), one is currently on the liver waiting list while also requiring kidney dialysis due to cholemic nephropathy (POD 300) and one patient successfully received liver  transplantation 216 days after the initial LTx. In 5 of 8 patients who died after transplantation, the cause of death was SSC. Figure 3 shows the 1-year survival for patients in the 'SSC' group (47%) and the 'no SSC' patients (90%) (log-rank: p = 0.004).

Discussion
SSC-CIP is a known complication of COVID-19 ARDS patients surviving the ICU. [15][16][17] With increasing experience in salvage LTx for this patient group, a growing number of centers have also observed SSC-CIP. However, data is lacking, given the rarity of SSC-CIP in general and the limited number of COVID-19 ARDS patients transplanted worldwide to date. To the best of our knowledge, this multi-center study is the first to examine this entity in patients transplanted for ARDS. Our results suggest that i) the incidence of SSC-CIP in LTx for COVID-19 ARDS is substantial and entails significant non-graft-related morbidity and mortality and ii) GGT is a sensitive early biomarker that could be useful for patient selection.
While long-term outcomes remain to be assessed, the reported early-and mid-term outcomes reported for LTx in COVID-19 ARDS are encouraging. 22,26 The largest singlecenter experience to date has recently been published by Kurihara et al The authors showed a short-term survival of 100% with a median follow-up of 351 days in a cohort of 30 patients. 2 Roach et al reported the largest case series to date in a registry study of the UNOS database. Their cohort consisted of 140 COVID-19 ARDS patients and 74 COVID-19 associated pulmonary fibrosis patients and showed a 3-month survival of 95.6%. 3 Looking at early outcomes and graft performance, the results in our study cohort corroborate the concept of LTx for ARDS. Both study groups had initially successful courses with no perioperative mortality and survival was excellent in patients who did not develop SSC-CIP.
Patients transplanted due to COVID-19 are not a homogenous group and two phenotypes can be distinguished. Patients unweanable from ECLS with chronic fibrotic parenchymal changes seen after COVID-19 represent a less complex group. Excellent post-transplant outcome similar to idiopathic pulmonary fibrosis can be achieved. On the other hand, the setting of acute critical illness with necrotizing lungs, bacterial superinfection, pleural empyema and frequent episodes of bacteremia poses more surgical and peri-operative challenges. Our cohort predominantly represents this latter phenotype. Naturally, these complex patients are more prone for complications, among them hepatic problems. SSC-CIP has been previously described in non-COVID-19 ARDS 11,12,27 as well as COVID-19 associated critical illness. [15][16][17] Several risk factors have been proposed. Ischemic injury to the biliary system is thought to be an important underlying factor for SSC-CIP. 6 While ARDS itself constitutes a hypoxic situation to the organism, aggressive intensive care treatment can further contribute to these effects. Sustained high PEEP values are thought to impair splanchnic perfusion. 28 Of note, high PEEP levels are often applied as a part of lung-protective ventilation concepts with low tidal volumes and are very commonly used in COVID-19 ARDS patients. Similarly, prolonged prone positioning has been employed almost universally to improve oxygenation in these patients. However, prone positioning is thought to negatively impact blood supply to the liver and biliary system, especially in obese patients. 12 Ketamine has been widely used in COVID-19 patients as an additional sedative. It has been proposed as an important risk factor for SSC-CIP. 29 Especially high cumulative ketamine doses have been suspected to cause progressive cholangiopathy. 30 Interestingly, all 15 patients in the 'SSC' cohort received ketamine pre-transplant, while only 50% of 'no SSC' patients did. Unfortunately, data on cumulative doses could not be collected due to the retrospective nature of this study. The severely destroyed lung parenchyma in ARDS is highly susceptible to recurring bacterial or fungal superinfections, potentially leading to septicaemia and potentially septic shock. Moreover, hepatotoxic antibiotics and antimycotics often cannot be avoided, which may add additional risk factors for the development of cholestatic liver injury. 5 In our cohort, all patients received various potentially hepatotoxic antibiotics or antimycotics over the course of the pre-transplant ICU treatment with no apparent differences between groups. In patients with severe elevations in serum liver parameters, these medications were avoided post-transplant as far as possible. While pre-transplant bridging with VV ECMO has become a routine in LTx, 31 the pandemic has led to very long bridging periods. Current guidelines for LTx in COVID-19 emphasize evaluating the potential for native lung recovery and suggest at least 4 weeks on optimal treatment including ECLS until considering listing. 32 As VV ECMO has been described as a risk factor for SSC-CIP, these prolonged ECMO-bridging durations could also contribute to its high incidence in LTx patients after COVID-19. 33 In our study, the vast majority of patients were bridged using VV ECMO in a femoro-jugular configuration. We found no association of ECMO mode and cannulation site with SSC.
In addition to the above mentioned risk factors, extrapulmonary viral effects of COVID-19 itself could be an important contributor to the development of SSC-CIP. Liver function impairment has been widely reported in critically ill COVID-19 patients. 13 In particular, cholestatic liver injury resembling SSC-CIP has been a known complication in these patients. [15][16][17] The ACE2-receptor as one of the main entry sites for the virus in the cell is also expressed in the bile ducts which have been shown to be infected by COVID-19. 14 This has been postulated as an explanation for the higher incidence of cholestatic problems in COVID-19 patients. 17,34 Most patients in our overall cohort show a broad combination of the above mentioned risk factors accumulating over time (Table 3). As the majority of COVID-19 ARDS patients share these factors, the assessment of a patient's clinical risk for SSC is of limited value. In addition, routine liver imaging during pre-transplant evaluation performed within a short period before transplantation in these cases was non-prognostic for a later SSC in our study cohort. Sensitive laboratory values as proposed by our study could be more useful to determine the risk for SSC-CIP development.
Guidelines to approach evaluation of COVID-19 patients for LTx have been previously suggested. 32 LTx is only recommended in case of mono-organ failure of the lungs. However, the distinction between an extrapulmonary organ failure precluding LTx and a transient organ dysfunction is difficult. Our study illustrates these difficulties. Synthetic liver function can be preserved in SSC-CIP, misrepresenting the true condition of the liver. 11 Generally, the lack of clinical symptoms in the initial phase can delay or prevent the diagnosis of SSC-CIP and often, only persistent indicators of cholestasis parameters raises the suspicion over time. 35 Given the highly-urgent nature of LTx in the setting of ARDS, the window for diagnosis of SSC-CIP may be easily missed before transplantation. In addition, more transient factors such as sepsis-associated cholestasis or druginduced liver injury are common differential diagnoses. 5,36 GGT is a non-specific and therefore frequently underestimated and highly sensitive marker for oxidative stress and bile duct injury. 37,38 Notably, in contrast to other cholangiopathies, the clinical value of ALP is increasingly controversially discussed in primary sclerosing cholangitis sharing several key features with SSC. 39 Increased GGT has been previously described as one of the earliest markers predicting SSC-CIP. 40 Our study corroborates these findings and suggests that persistent and excessive levels of GGT should raise a red flag and further diagnostic work-up should be recommended given the poor outcomes of SSC-CIP after LTx. Delisting should be considered. Our data further suggests that TBi and ALP should be closely observed on the waiting list.
Our study has several limitations. As a retrospective study, it is prone to missing or miscoded data. While the multi-center approach is invaluable to gain a substantial patient cohort, heterogenous recipient and donor selection, surgical standards and strategies in perioperative care may introduce bias. Patient numbers contributed by the different centers were not balanced and ranged between 1 and 23. The limited overall cohort size prevented us from meaningfully correcting our statistical analysis for volume of the individual centers. As the worldwide experience for LTx in COVID-19 patients is still limited and SSC-CIP is a rare disease, a limited cohort size is unavoidable. This may limit the generalizability of our findings. This fact also prevented us from meaningfully applying multivariate statistical calculations. Registry studies could provide larger cohorts but lack data granularity necessary to address the aims of our study. Lastly, while our study covers a relatively long follow-up period compared to other reports on LTx for COVID-19, it can still only assess a limited time frame and true long-term outcomes are beyond our scope.
In conclusion, our study shows that SSC-CIP is a severe complication after LTx for COVID-19 ARDS and entails significant morbidity and mortality in this cohort. While the risk-benefit ratio still should be considered favorable for LTx, this strongly underlines the importance of patient selection. GGT appears to be the most sensitive early parameter routinely available and predicts SSC-CIP even at the time of listing.

Disclosure statement
None of the authors have any relevant conflict of interest to declare.