Whipple's Pancreatoduodenectomy in the Background of Chronic Liver Disease (CLD): An Institutional Experience

Introduction Whipple's pancreatoduodenectomy (PD) is a formidable operation, associated with a high risk of morbidity and mortality. In the setting of an underlying chronic liver disease, the incidence of complications and mortality increases manifold. Patients and Outcomes. Of the 112 Whipple's PD performed between 2018 to 2020 at a high-volume HPB and liver transplant centre, 4 patients underwent the surgery in the background of an underlying chronic liver disease (CLD). All except one were performed in Child's A cirrhotics. There was a single 30-day mortality in this series of 4 patients that occurred in the background of Child's B cirrhosis. On follow-up at 1 year, there was one more mortality in the series, owing to liver decompensation following chemotherapy. Conclusion Judicious preoperative selection criteria, adequate preoperative nutritional and physiological optimisation, and prudent weighing of risk vs. benefit of undergoing Whipple's PD in periampullary malignancies in the setting of CLD are the major determinants of the surgical outcome.


Introduction
Chronic liver disease (CLD) has traditionally been regarded as a contraindication to most major gastrointestinal (GI) and hepatopancreaticobiliary (HPB) procedures because there is, in some reports, a mortality of as high as 35% in these patients. Even in patients with normal liver function, pancreatoduodenectomy (PD) which is the treatment of choice for resectable periampullary malignancies has a morbidity of over 40% and a mortality of nearly 5-8%, in highly specialised tertiary-care centres [1,2]. Apart from the intraoperative difficulties encountered during the surgery such as bleeding from dilated venous collaterals resulting from portal hypertension and coagulopathy from liver dysfunction, there may be difficult planes of dissection due to previous surgery and anaesthetic challenges. PD is also associated with a high risk of postoperative morbidity in the form of anastomotic leaks and hepatic decompensation [2].
Between January 2018 to December 2020 at a highvolume tertiary-care GI, HPB, and liver transplant centre in New Delhi, India, we operated on 4 patients with periampullary malignancies who also had CLD and retrospectively attempted to elucidate the surgical management of these patients who also had chronic liver disease and reviewed the published literature in an attempt to provide some guidelines for an experienced HPB surgeon.

Patients and Outcomes
Out of a total of 112 Whipple's pancreatoduodenectomy performed at our centre during these three years, there were 82 (73%) males and 40 females (38%) who had a mean age of 57 years (range 31-69 years). Of them, 3 patients had chronic liver disease (CLD) at the time of presentation and 1 was found to have features of CLD intraoperatively. Overall, 6 (5.3%) patients died 5 with normal livers and one case who had Child's B CLD. eir average length of postoperative hospital stay was 12 days (8)(9)(10)(11)(12)(13)(14)(15)(16)(17)(18)(19)(20)(21)(22)(23)(24). e aetiology of CLD was Non-Alcoholic Steatohepatitis (NASH) related in all but one (Case 3). e worst MELD score noted in the series was 19 (Case 4). e preoperative fitness was assessed using the ECOG performance score/ ASA scoring by the anaesthesiologist as well. Whipple's PD was performed in all for malignancies (adenocarcinoma) diagnosed preoperatively by EUS-guided biopsy. ree had some other comorbidities such as diabetes mellitus. Preoperative stenting was carried out in Case 2 in view of his bilirubin levels (>15 mg/dl) and in the other patient (Case 3) due to poor nutrition with decompensated CLD. None of the patients received any form of neoadjuvant chemo-or radiotherapy (Table 1). Table 2 summarises the trend of laboratory variables in the postoperative period in comparison to the preoperative (baseline) value. Case 3 exhibited features of pancytopenia (related to CLD with hypersplenism) in the preoperative period, which was optimised to an adequate level before surgery. However, there was persistent decline in the haemoglobin and platelet counts noted in the postoperative period (surrogate markers of CLD decompensation and sepsis).
e serum albumin levels also showed a similar trend in this patient, with persistent decline after POD 5, despite intravenous support. e operative duration ranged between 370 mins (Case 4) to 480 mins (Case 1) ( Table 3). e four cases were performed by different senior surgeons of the same unit, following a uniform protocol of pancreatoduodenectomy.
e intraoperative blood loss was aimed to be kept low to minimise the chances of decompensation and ranged from 700 ml to 900 ml. A single-loop reconstruction of the pancreatic-jejunal anastomosis with the duct to mucosa technique was performed in all but one (Case 4), where the isolated loop (Machado's) technique was used. Pancreatic dissection was deemed challenging in patients with liver disease owing to the presence of peripancreatic collaterals in 2 of the 4 cases (Case 1 and 3). ere was no significant vascular event noted in any of the 4 cases. e liver was cirrhotic in two patients (Cases 1 and 3) and steatotic in the other two (Cases 2 and 4) ( Figure 1). ere was moderate ascites present in Case 3. Table 4 summarises the postoperative outcome in the 4 patients. All required blood product transfusion in the postoperative period, the maximum (7 units packed cells with 8 units fresh frozen plasma) being required in Case 3. Local complications in the form of a clinically significant pancreatic leak were noted in 2 of the 4 cases (Cases 2 and 3), one of whom (Case 2) required placement of a percutaneous drain under USG guidance. One patient (Case 1) developed delayed gastric emptying on POD 5, with no evidence of any leak or collection (as documented by a CT of the abdomen on POD 5), which was managed with prokinetic drugs (metoclopramide and erythromycin started on POD 5). Case 3 developed severe ascites with features of hepatic encephalopathy (elevated ammonia levels) from POD 6, along with clinically a major pancreatic leak, eventually culminating into severe sepsis, disseminated intravascular coagulopathy (DIC), and multiorgan failure and died on POD 9. e other 3 patients were discharged after a postoperative stay of 10 days. Of the 3 patients, 1 (Case 4) was readmitted after 3 months of undergoing surgery with features of encephalopathy and decompensation related to CLD (probably secondary to two cycles of gemcitabine-based chemotherapy) and eventually died in the same admission after a hospital stay of 45 days. Case 1 and 2 have been followed up till 18 months after surgery and have shown no signs of recurrence or metastasis.

Discussion
e presence of underlying CLD in a patient with pancreatic carcinoma poses a formidable challenge to the surgeon planning a pancreatoduodenectomy (PD). e biggest dilemma is the "toss-up" situation between considering the risks of major surgery in a patient with CLD (documented mortality >30%) vs. the mortality associated with the primary malignancy and its effects on his or her quality of life.
ere have been very few studies in the literature assessing the impact of cirrhosis on the outcome of PD [3,4]. One of the earliest reports came from a study by Artinyan et al. in 2012 who concluded that cirrhosis increased the perioperative morbidity and mortality in patients undergoing major GI oncological procedures [5]. e major challenges faced in patients with cirrhosis range in the intraoperative period from an increased likelihood of blood loss (due to underlying coagulopathy or massive collateral bleeding due to underlying portal hypertension) to encephalopathy and other haemodynamic alterations due to the anaesthetic drugs. In the postoperative period, decompensation in the form of intractable ascites, gastrointestinal bleeding, an increased risk of anastomotic dehiscence (poor healing), and resultant local and systemic sepsis account for the major causes of morbidity and mortality. Warnick et al. compared the outcome of pancreatic resection in patients with underlying cirrhosis against a matched control set of patients and reported a significantly high incidence of complications (47% vs. 22%; P � 0.035) and reoperations (34% vs. 12%; P � 0.03) [6]. In another recent study by El Nakeeb et al. conducted in 67 (Child's A and early B) out of 442 patients undergoing PD, the intraoperative blood loss and transfusion requirements were found to be significantly higher in the cirrhotic group [7]. e authors also reported a higher incidence of pancreatic fistula, wound complications, and haemorrhage in the cirrhotic group. eir duration of postoperative stay was also longer in the cirrhotic. Sethi et al. performed a retrospective analysis of the outcomes in 4 patients with operable pancreatic tumours and well-compensated CLD (CTP 5-6) over a 6-year period at a highvolume centre [8]. ey reported a favourable short-term outcome in all their patients, with no requirement of blood transfusion, minimal intraoperative blood losses, and no postoperative hepatic decompensation. One death was reported at 18 months due to disease recurrence. Similar results were reported by the French group on a large multicentre study in 2015 [9]. Another group from Spain compared the outcome of PD in cirrhotics (n � 15) vs. noncirrhotics (n � 30) and reported a postoperative morbidity of nearly 60% in the former and also a higher duration of postoperative stay (25 ± 19 days) [10]. Findings of the studies are summarised in Table 5. However, they reported no difference in the rate of complications between the cirrhotic and noncirrhotic groups, as well as no difference in   the rate of haemorrhage and fistulae. In the present case series over a period of 3 years at a high-volume HPB and transplant centre, we also reported findings similar to the studies mentioned above in terms of complications. However, a notable difference was observed in the incidence of immediate postoperative and long-term complications in our series. Factors such as difference in ethnicity and nutritional status between the West and the East could be an attributable factor [11]. e high incidence of decompensation and pancreatic leaks could be due to the already diminished healing and nutritional capacity of the patients, compounded by chronic alcohol intake (Case 3). Measures such as preoperative TIPSS have been suggested by few in patients with ascites to have a favourable outcome [12]. e

Conclusions
PD is a feasible option in patients with periampullary malignancy with underlying CLD. However, meticulous case selection based on Child's status, nutritional parameters, and blood indices and surgical expertise of the centre (highvolume HPB and transplant) could be the way forward to deal with such complicated case scenarios. e risk of mortality due to CLD must be weighed against the poor QoL and complications and mortality secondary to periampullary malignancy, and the same be explained to the patient.

Data Availability
Data are available on request.

Conflicts of Interest
e authors declare no conflicts of interest.