Discussion and conclusions
During PD, particularly for gastric cancer, the rLHA arising from the LGA should be preserved as far as oncologically acceptable. Resection of the rLHA during PD can cause postoperative liver necrosis.
The liver can survive arterial ligation because of sufficient supply of oxygen and nutrients from the portal vein and several collateral arterial channel pathways [
12]. A study on surgical anatomy based on dissection of 200 cadavers demonstrated 26 possible collateral arterial pathways to the liver [
13]. Similarly, several radiologic studies showed many collateral arterial pathways in case of temporary interruption of blood supply from the hepatic artery [
14‐
16]. Immediately after hepatic artery embolization (HAE), the following arteries that supply the liver have been radiologically identified: (1) inferior phrenic arteries through the right and left triangular and coronary ligaments, (2) branches of the superior and inferior epigastric arteries through the round and falciform ligaments, and (3) unnamed gastric collaterals in the lesser omentum and hepatoduodenal ligament [
12,
13,
17].
For major hepato-biliary-pancreatic (HBP) surgeries including PD, extended hepatectomy, and hepato-pancreaticoduodenectomy (HPD), these abundant extrahepatic collateral arterial pathways are commonly divided by operative procedures, such as liver mobilization or dissection of the hepatoduodenal ligament [
18]. Nevertheless, postoperative HAE for ruptured pseudoaneurysm had been considered safe and was widely performed even after a major HBP surgery [
16,
18‐
20], probably because the remaining extrahepatic collateral pathways and the interlobar communicating artery (ICA) protect the liver from ischemic injuries [
14,
16,
18‐
20].
The ICA runs through the Glissonean sheath around the bilateral hepatic duct confluence [
14,
21,
22], which is known as the hilar plate [
23], to perfuse liver areas without arterial supply [
12,
14,
18,
21,
22]. Another study on seven cases of biliary tract carcinoma with unilateral hepatic artery involvement concluded that hepatic artery reconstruction was not always required as long as the hilar plate was preserved [
21]. The ICA plays an important role in preventing ischemic injury from interruption of hepatic artery supply, particularly during a major HBP surgery [
14,
21], and has been regarded an important communication into the liver [
14,
15,
24,
25].
Conversely, an 18-case study [
19] observed that 83% (15/18) developed ischemic liver injury after HAE for arterial hemorrhage following PD and HPD. However, most patients in that study had mild and transient ischemic liver damage and only one patient (7.7%) developed liver necrosis. The authors concluded that HAE following a major HBP surgery did not carry a great risk for serious ischemic liver injury.
Considering these evidences, arterial supply to the whole liver can be theoretically maintained after rLHA division during PD. Even if liver ischemic damage occurs, it would be mild and result in transient liver enzyme elevation. However, our two cases had serious postoperative liver necrosis. In both cases, the liver was not mobilized and the hilar plate was not dissected. The color of the liver surface was normal, and there was no demarcation line that appeared after rLHA division. In addition, blood flow from the residual hepatic arteries and portal vein was maintained perioperatively.
We assumed that the unexpected postoperative liver necrosis in this study was mainly caused by four possible mechanisms. First, the extent of lymph node dissection during PD for advanced gastric cancer tended to be larger than for other HBP diseases, such as those around the celiac axis and lesser omentum or those around the hepatoduodenal ligament; these may have reduced the extrahepatic collateral pathways to the liver. Some reports noted that extensive lymph node dissection around the liver could lead to liver necrosis after HAE following major HBP surgery [
18,
26]. In that report, total remnant liver necrosis after HAE occurred in one patient (11%) who underwent whole liver mobilization during HPD [
18]. In the field of liver transplantation, extrahepatic collateral pathways are absent immediately after operation and arterial supply depends on the reconstructed hepatic artery only [
27,
28]. As a result, hepatic artery thrombosis can cause fatal liver necrosis [
27,
28].
The second mechanism could be congenital underdevelopment of extrahepatic collateral pathways or the ICA. A report stated that the degree of collateral pathway development varied among individuals and that the course after HAE depended on the individual vascular anatomic characteristics [
12]. A study on 16 adult cadaveric livers indicated that the ICA was not always existent and was not detected in 2/16 (13%) cases [
25]. Indeed, a case report described liver necrosis with abscess formation in the lateral segment immediately after coil embolization of the rLHA [
3]. The patient had not undergone any abdominal operation that potentially destroyed collateral arterial pathways around the liver. The author concluded that the liver necrosis was probably related to congenital underdevelopment of the ICA.
The third mechanism could be based on the fact that PD requires hepaticojejunostomy, in contrast to standard gastrectomy; this can cause upstream bacterial infection into the intrahepatic bile duct. To our knowledge, there had been no report on liver necrosis due to rLHA resection after standard gastrectomy for gastric cancer. A large-scale study on standard gastrectomy for 1340 consecutive patients with gastric cancer noted that all 116 cases that underwent rLHA resection did not have postoperative liver necrosis with abscess formation, although some of the patients had transient liver enzyme elevation [
1]. Other studies on standard gastrectomy with rLHA resection reported the same outcomes in terms of ischemic liver injury [
2,
4]. In our cases, biliary infection through hepaticojejunostomy may have led to liver necrosis with abscess formation in ischemic areas.
The fourth possible mechanism for the development of liver necrosis in our cases could be the reduction of total hepatic arterial flow due to rLHA resection with extended lymphadenectomy around the liver. Reportedly, sufficient hepatic arterial flow plays an important role in clearing bacterial translocation from the gut into the liver [
29,
30]; that is, bacterial translocation under reduced hepatic arterial flow may have led to worse liver necrosis and abscess formation shortly after PD.
Although further work is required to clarify the potential risk factors of postoperative liver necrosis, we assumed that these four mechanisms mainly caused the unexpected postoperative liver necrosis in our patients, even though only the A2 + A3 or A2 was resected.
In usual clinical practice, it is impossible to identify whether these extrahepatic collateral pathways and the ICA are congenitally poor [
19]. As a result, surgeons should always reconstruct the divided rLHA during PD [
12]. However, microvascular reconstruction is technically demanding, particularly when the arterial stump is small, as in our cases. Judging from our clinical outcomes, we believe that surgeons should preserve the rLHA during PD as far as oncologically acceptable. In case the rLHA is resected during PD, surgeons should carefully observe the postoperative course of the patient while keeping in mind the possibility of urgent hepatectomy. If resection of the rLHA is inevitable due to tumor invasion, coil embolization of the rLHA before PD may be helpful in preventing postoperative ischemic complications, such as liver necrosis. In fact, a few case reports on requiring resection of rRHA from SMA stated that preoperative coil embolization of the rRHA prevented ischemic liver complications after PD [
31‐
34].
In case 1, we created an arterioportal shunt as an alternative to microvascular reconstruction after A2 + A3 resection [
35]. Although the effect remained unknown [
35], the procedure may have contributed to the response to conservative management. In case 2, liver necrosis occurred even though only the A2 was resected. We presumed that the collateral pathways to the liver or the ICA were congenitally poor in this patient. In both cases, any demarcation line did not appear on the liver surface. We should have performed Doppler ultrasonography to confirm sufficient blood flow in the intrahepatic arteries.
To the best of our knowledge, this was the first report on liver necrosis shortly after PD due to the rLHA resection. This is potentially because the rLHA hardly interfered with operative procedure in PD for HBP disease [
36], in contrast to that for gastric cancer. We consider it a pitfall that surgeons tend to underestimate the rLHA arising from the LGA [
1,
2,
4], compared with the rRHA, particularly in gastric cancer surgery.
In conclusion, resection of the rLHA arising from the LGA during PD can cause postoperative liver necrosis, even if only the A2 + A3 or A2 was resected. We underscore that the rLHA should be preserved during PD as far as oncologically acceptable. In case rLHA resection is inevitable during PD due to tumor conditions, surgeons should carefully observe the patient in the postoperative course while keeping in mind the possibility of urgent hepatectomy.