Portal Vein Stenosis Following Neoadjuvant Therapy With MRgART and Surgery for Pancreatic Cancer: A Case Report

A 63-year-old man with a past medical history of alcohol use disorder, recurrent pancreatitis, and colon cancer after sigmoidectomy and no adjuvant therapies presented to the emergency department for pancreatitis with obstructive jaundice. The patient was treated with a biliary stent, and CT showed a 4.0 × 3.8-cm mass located in the pancreatic head with greater than 180° involvement of the superior mesenteric vein (SMV). Endoscopic biopsy confirmed pancreatic adenocarcinoma. The patient was staged as clinical T2N1M0, stage IIB, and the tumor was deemed borderline-resectable after multidisciplinary tumor board review. The patient subsequently underwent 8 cycles of neoadjuvant FOLFIRINOX (fluorouracil, irinotecan, and oxaliplatin), followed by MRI-guided adaptive radiation therapy (MRgART) to 50 Gy in 5 fractions, with the SMV receiving a Dmax of 56.4 Gy (Figure 1). His CA19-9 level decreased from 766 U/mL to 51.8 U/mL, and SMV involvement was less than 180°, making him eligible for pancreatoduodenectomy (PD) with planned en bloc SMV resection with venovenous anastomosis (International Study Group type 3). At 104 days postsurgery and 150 days postradiation, the patient developed symptomatic ascites and CT imaging demonstrated severe stenosis of the SMV at the confluence with the main portal vein (PV). He underwent paracentesis, and 860 mL of ascitic fluid was removed; cytology was negative. Portal venography demonstrated portal vein stenosis (PVS) at the SMV orifice and occlusion of the splenic vein at the confluence with the main PV. Angioplasty of the PVS was performed and resolved the patient’s symptomatic ascites. The patient also experienced variceal bleeding, requiring endoscopic clipping and eventually embolization, and remains in clinical surveillance for pancreatic cancer without evidence of recurrent disease.

MRI performed as part of the initial diagnostic workup characterized the tumor as a 3.8-cm mass at the pancreatic head, along with a left 1.3-cm periaortic node suspected to be inflammatory. There was vascular encasement of the portosplenic confluence by the tumor as well as greater than 180° contact with the SMV. After neoadjuvant therapy, CT imaging revealed that the pancreatic mass had decreased to 2.6 cm, and there was a decrease in vascular involvement and patent PV (Figure 2A). Postsurgical CT showed small-volume ascites, along with narrowing of the PV, SMV, and splenic vein with peripancreatic edema (Figure 2B). In surveillance, CT showed increasing ascites and ultrasound showed stenosis at the PV-SMV confluence (Figure 2C). The patient underwent paracentesis, and CT demonstrated persistent narrowing of the portal venous confluence. Transhepatic portal venography demonstrated stenosis of the main PV at the SMV orifice and chronic occlusion of the splenic vein at the confluence with the main PV (Figure 3). A timeline of the imaging findings is shown in Table 1.

The diagnosis was consistent with symptomatic PVS secondary to occlusion of the splenic vein at the confluence with the main PV, as demonstrated by the imaging that followed the patient’s surgery as well as the evaluation by an interventional radiologist who performed portal venography. The patient only became symptomatic with ascites and esophageal varices after imaging findings began to show PVS. He had no prior history of varices, ascites, or liver failure despite a history of alcohol abuse. Additionally, the imaging taken early in his treatment course showed no evidence of PVS.

Here, we describe a case report of symptomatic PVS in a patient with adenocarcinoma of the pancreas treated with chemotherapy, stereotactic body radiation therapy (SBRT) with MRgART, and PD with vascular reconstruction. Following PD, the incidence of iatrogenic PVS was 3.4% to 6.1%, which can be associated with significant morbidity as well as a 3% mortality rate secondary to gastric bleeding. ^1-4 The risk factors for the development of PVS were tumor location in the pancreas, delivery of chemoradiation, and concomitant PV resection. ⁴ The patient, for example, developed both symptomatic ascites and bleeding varices as a result of PVS. Thus, it is important to continue to describe this phenomenon and offer potential etiologies and methods of mitigation.

Pancreatic surgery and porto-mesenteric reconstruction are known risk factors in the development of PVS due to inflammation, narrowing at the anastomotic site, and pancreatic leak. ^4-6 Ten days after PD with vascular reconstruction, 84% of patients have some degree of concentric or eccentric vascular stenosis. ⁶ In patients with pancreatic cancer assessed 5 years after PD, it was found that vascular resection confers 3.28 times the risk of developing PVS — increasing it from 17% to 51%. ⁴

Neoadjuvant chemotherapy may separately partially contribute to PVS. Two common chemotherapy regimens include gemcitabine and oxaliplatin. Gemcitabine monotherapy has been associated with coagulation cascade activation and endothelial damage and has been linked to increased thrombotic events, as well as increased risk when used in multiagent regimens. ^7,8 Oxaliplatin has been linked to hepatic sinusoidal obstructive syndrome (SOS) in colorectal cancer due to chronic injury to endothelial cells; however, there is limited literature documenting oxaliplatin-induced hepatic SOS with respect to pancreatic cancer. ⁹ Discussion of this phenomenon may be limited in patients with pancreatic cancer due to more limited survival and, thus, follow-up.

Regarding radiation therapy (RT), radiation-induced vascular inflammation, thrombus, and stenosis are well-described phenomena, and the pathology is generally thought to be limited to small caliber vessels in the myocardium or mandible. ^10-13 Reports on PVS after neoadjuvant chemoradiation are limited ⁷ ; however, a recent prospective study reported the safety of MRgART for patients with locally advanced or borderline resectable pancreatic cancer. ¹⁴ After neoadjuvant chemotherapy, patients were treated to 50 Gy in 5 fractions and a total of 44 patients (32%) proceeded to surgical resection, with more than half of the patients requiring vascular reconstruction (n = 23). Two patients experienced grade 5 toxicities of fatal gastrointestinal bleeding that were deemed possibly related to MRgART. Additionally, the 3 postoperative deaths in the study occurred in patients who had vascular reconstruction more than 5 weeks from the completion of radiation. In the study, there were no published PV dose constraints; as a result, we adopted a D0.03 cm³ (maximum dose to 0.03 cm³) of 50 Gy for the PV for all patients being treated with 5-fraction RT. We also included vascular injury in the informed consent form when planning RT to an upper abdominal disease site.

Nonetheless, acute-onset PVS may present with rapidly progressive abdominal ascites and endovascular therapies such as angioplasty and stenting can be performed in the setting of iatrogenic or tumor-related PVS. The success rate of stent deployment is approximately 93%, with median patency rates of at least 14 months. ^3,15 The primary stent patency in stenting performed for tumor-related PVS was found to be shorter than that for iatrogenic PVS, at 6.5 months compared with 16 months, respectively. ¹⁵ After stenting, 71% to 93% of patients reported clinical improvements in symptoms, with major complication rates of 0% to 7%. ^16-18 However, rapid symptom recurrence after stent deployment for radiation-induced venous disease within the lower extremities has been described, which suggests radiation changes may make intraluminal stenting less successful compared with stenting of non-irradiated veins. ¹⁹

We acknowledge several limitations in reviewing and reporting this rare toxicity. The patient had heterogeneous treatments, which make identification of a single inciting factor difficult, and we suspect that the development of PVS is multifactorial. Importantly, the patient’s variceal bleeding occurred prior to ascites, which we suspect was predominantly driven by splenic vein occlusion at the PVS rather than the PVS alone. We also acknowledge that there is likely a much larger group of patients with subclinical PVS without the need for therapeutic paracentesis or endovascular intervention if there is no significant pressure gradient across the vessel. Therefore, the incidence of PVS is likely under-reported in the literature.

Although rarely reported in the literature, PVS is a clinically significant side effect in patients with pancreatic cancer, which may be compounded by multimodality therapy. Although the cause of PVS in this population is not clearly identified and likely multifactorial (eg, extrinsic tumor compression, thrombus, narrowing at the venous anastomotic site, radiation-related vascular changes, and possibly underappreciated vascular and hepatic sinusoidal toxicity from FOLFIRINOX), its incidence and the potential influence of preoperative radiation should not be trivialized. Until clear dosimetric factors mitigating the risk of this phenomenon are more clearly defined, we caution against excessive radiation dose near the PV in operable patients and strongly recommend the adoption of PV dose constraints for patients planned for 5-fraction SBRT.