Portomesenteric Venous Stenting for Palliation of Ascites and Variceal Bleeding Caused by Prehepatic Portal Hypertension

This study evaluated the efficacy of percutaneous transhepatic portal vein stenting in cancer patients with symptomatic portal hypertension caused by portomesenteric venous compression, demonstrating that this technique can play a substantial role in improving quality of life in patients with portomesenteric stenoses.

Abstract

Background.

The purpose of this study was to evaluate percutaneous transhepatic portal vein stenting (PVS) for palliation of refractory ascites and/or variceal bleeding caused by extrahepatic portomesenteric venous stenosis in patients with pancreaticobiliary cancer.

Materials and Methods.

A single‐institution, retrospective review of patients who underwent PVS between January 2007 and July 2015 was performed. A total of 38 patients were identified, of whom 28 met the inclusion criterion of PVS performed primarily for refractory ascites or variceal bleeding. In addition to technical success and overall survival, clinical success was measured by fraction of remaining life palliated. The palliative effect of PVS was also quantified by measuring changes in liver and ascites volumes after the procedure.

Results.

Technical success was 93% (26/28). Stent deployment involved more than one portomesenteric vessel in most patients (20/26). The cumulative probability of symptom recurrence at 6, 12, 18, and 24 months was 12%, 16%, 26%, and 40%, respectively. There was a significant difference (p < .001) in the probability of symptom recurrence, recurrence of abdominal ascites, and increase in liver volume between patients whose stents remained patent and those whose stents demonstrated partial or complete occlusion. The mean fraction of remaining life palliated was 87%. All but two patients were found to have improvement in clinical symptoms for the majority of their lives after the procedure. There were no major or minor complications.

Conclusion.

As a low‐risk procedure with a high clinical success rate, PVS can play a substantial role in improving quality of life in patients with portomesenteric stenoses.

Implications for Practice.

Portomesenteric venous stenosis is a challenging complication of pancreaticobiliary malignancy. Portomesenteric stenoses can lead to esophageal, gastric, and mesenteric variceal bleeding, as well as abdominal ascites. The purpose of this study was to evaluate the safety and efficacy of portal vein stenting (PVS) in patients with cancer who have symptomatic portal hypertension caused by portomesenteric venous compression. As a low‐risk procedure with a high clinical success rate, PVS can play a substantial role in improving quality of life in patients with portomesenteric stenoses.

Introduction

Portomesenteric venous stenosis is a challenging complication of pancreaticobiliary malignancy [1], [2]. Encasement of the portal vein (PV), superior mesenteric vein (SMV), and splenic vein (SpV) frequently occurs in pancreaticobiliary cancers and can lead to hemodynamically significant stenoses or occlusion. Benign portomesenteric venous obstruction can also occur in 13%–20% of patients who undergo intraoperative radiation therapy or venous reconstruction during pancreaticoduodenectomy [3], [4], [5], [6]. Moreover, the risk for benign stenoses increases with time, affecting 27%–48% of patients by 3 years after surgery [5], [6].

Both benign and malignant portomesenteric stenoses result in prehepatic portal hypertension and can lead to equally profound ramifications as the more common causes of portal hypertension. For example, elevated mesenteric venous pressures can lead to esophageal, gastric, and mesenteric variceal bleeding, as well as abdominal ascites. Elevated splenic venous pressures can result in splenomegaly‐associated thrombocytopenia. Intrahepatic portal hypoperfusion can lead to liver atrophy, liver dysfunction, hypoalbuminemia, and malnutrition.

Portal vein stenting (PVS) is a minimally invasive procedure that has been shown to be safe in patients with portomesenteric venous stenosis [1], [3], [7], [8], [9], [10], [11], [12], [13]. The principal role of this procedure is palliation of symptoms caused by portal hypertension. The goal is either to reduce the volume of ascites, minimizing the need for therapeutic paracenteses, or to prevent variceal bleeding, decreasing the symptoms of anemia and the need for blood transfusions. Although prior reports have demonstrated a high technical success rate for this procedure, its impact on patient quality of life is not well understood. Specifically, a quantitative assessment of the influence of PVS on the quality of remaining life in this patient population is lacking.

The purpose of this study was to evaluate the efficacy of PVS in patients with cancer who have symptomatic portal hypertension (ascites or variceal bleeding) caused by portomesenteric venous compression. Volumetric imaging data were used to evaluate the duration and magnitude of the palliative effect of PVS. Demographic and anatomic variables were also analyzed to identify predictors of treatment response.

Materials and Methods

Study Population

Institutional review board approval was obtained for this single‐institution, retrospective study. An institutional radiology database was queried for patients who underwent PVS procedures from January 2007 to July 2015. The electronic medical records for all patients identified from the database were reviewed. Patients with extrahepatic portomesenteric venous stenosis or occlusion, for whom PVS was performed for refractory ascites or variceal bleeding, were included in this study. Patients with either malignant external compression or benign postoperative stenoses were included. Patients for whom PVS was performed prophylactically or empirically to prevent ascites or variceal bleeding were excluded. Likewise, patients for whom PVS was performed with the intention of reducing splenic volume in the setting of thrombocytopenia were excluded. Patients with direct tumor invasion into the portomesenteric vasculature were also excluded.

Clinical outcomes were assessed by reviewing patients’ medical records. Initial clinical outcomes were assessed at 1 month or at the first available clinic visit, if a 1‐month visit was not available. PVS was considered a clinical success if symptoms (ascites or variceal bleeding) were improved or resolved; PVS was considered a clinical failure if symptoms showed no improvement or had worsened since the procedure. Long‐term outcomes were determined by evaluating for the recurrence of symptoms for the duration of the study period. For patients who underwent PVS for ascites, recurrence was defined as reaccumulation of ascites requiring paracentesis. For patients who underwent PVS for variceal bleeding, recurrence was defined as a repeat episode of variceal bleeding. Given that the intention of PVS in this study was symptom palliation and that variable life expectancy in this patient population could lead to survival bias, long‐term outcomes were quantified by calculating the fraction of remaining life palliated (FRLP), using the following formula: FLRP = symptom recurrence free survival ÷ overall survival. In this way, the palliative efficacy of PVS could be decoupled from uncontrollable variables such as patient disease status, life expectancy, and overall survival.

Complications were categorized as major or minor based on the Society of Interventional Radiology classification system [14].

Portal Vein Stenting Technique

All PVS procedures were performed by a single interventional radiologist with over 20 years of experience in vascular oncologic interventions. If ascites was present, a 6 French (Fr) centesis catheter was first placed into the peritoneum to drain the ascites. Ultrasound guidance was used to obtain percutaneous access to the right portal venous system, usually a second‐order branch of the right posterior portal vein. The access was then dilated to accommodate an 8 Fr sheath, and the stenosis or occlusion was then crossed using a hydrophilic wire and angled catheter. Measurements for the size and length of the most appropriate stent were then made based on digital subtraction angiography. Stents were oversized by 10%–20% of the diameter of the vessel based on digital subtraction angiography measurements. Postdeployment angioplasty was performed to the same size as the stent. Stents were deployed in a variety of configurations, depending on the patient's anatomy: main portal vein (MPV) alone, SMV alone, MPV + SMV, PMV + SpV, and MPV + SMV + SpV. Venous pressure gradients were not routinely measured prior to or after stent deployment. Prior to removing the access sheath, metallic coils were deployed along the liver parenchymal tract to prevent bleeding. Technical success was defined as successful traversal of the portomesenteric stenosis or occlusion with successful deployment of a stent or stents resulting in angiographic demonstration of restored flow.

Imaging Analysis

Analysis of liver volumes on computed tomography (CT) was conducted using a standard clinical three‐dimensional image analysis software package (iNtuition; TeraRecon, Foster City, CA). Image analysis was performed by an interventional radiologist with subspecialty training in abdominal imaging and 5 years of experience in volumetric imaging. Whole liver volumes were measured using an automated liver segmentation algorithm; the segmented volumes were then manually adjusted as needed.

Likewise, ascites volumes were measured using a region‐growing segmentation analysis tool. The entire volume of abdominal ascites was measured on pre‐ and postprocedural CT imaging. The amount of fluid drained during paracentesis procedures was also included as additional data points in this analysis.

Statistical Analysis

Univariate and multivariate analysis of differences between patients with patent and occluded stents was performed using Fisher's exact test and the Wilcoxon rank sum test, as appropriate, using a standard statistical software package (R, R Foundation, Vienna) [15]. Postprocedure survival proportions were calculated by Kaplan‐Meier analysis. Competing risk analysis was performed to evaluate the probability of symptom recurrence after PVS [16], [17]. Gray's test was used to evaluate differences in recurrence probability between patients with patent or thrombosed stents.

Results

A total of 38 patients underwent PVS during the study period. Of these patients, 10 were excluded because of the criteria described in the Materials and Methods section. The demographics of the remaining 28 patients are summarized in Table 1. The procedure was technically successful in 26/28 patients (93%). Technical failures were due to inability to cross the stenosis or occlusion. A summary of procedural details is presented in Table 2. Balloon‐mounted stents (Palmaz Genesis; Cordis, Miami, FL; VisiPro; Medtronic Vascular, Santa Rosa, CA) were used in all but 3 patients, who received self‐expanding stents (SMART; Cordis, Miami, FL; EverFlex; Medtronic Vascular, Santa Rosa, CA). Stent deployment involved more than one portomesenteric vessel in most patients (20/26; Fig. 1), including stents extending from the SMV into the MPV, as well as Y configurations with an SMV stent deployed through the interstices of a stent extending from the SpV into the MPV. In addition to stent placement, one patient also underwent embolization of cecal varices.

Table 1. Study patient demographics.

Table 2. Portal vein stenting procedural details (n = 26).

Abbreviations: PV, portal vein; SMV, superior mesenteric vein; SpV, splenic vein.

^aclopidogrel; Bristol‐Myers Squibb, New York City, NY.

^benoxaparin sodium; Sanofi, Gentilly, France.

^cwarfarin; Bristol‐Myers Squibb, New York City, NY.

Figure 1.

Intraprocedural examples of the various stent deployment configurations.

Abbreviations: MPV, main portal vein; PV, portal vein; SMV, superior mesenteric vein; SpV, splenic vein.

Antiplatelet therapy with Plavix (clopidogrel; Bristol‐Myers Squibb, New York) for at least 3 months after the procedure was provided initially during the study period (6/26); however, therapeutic Lovenox (enoxaparin sodium; Sanofi, Gentilly, France) for at least 3 months became the preferred method of stent thrombosis prevention for the majority of the study period (14/26). One patient was continued on a preprocedural warfarin regimen after stenting, and 5 patients received no postprocedural anticoagulation because of the risk for bleeding complications.

Overall procedural outcomes are summarized in Table 3. All patients exhibited initial symptom improvement, with the exception of two patients, both of whom had persistent ascites after the PVS procedure (clinical success rate 24/26, 92%). The median duration of clinical success was 371 days for all patients. For patients who underwent PVS for recurrent ascites, 7/16 patients (44%) were recurrence‐free during the study period; 9/16 patients (56%) were found to have recurrent symptomatic ascites, although the ascites in 2 of these patients was noted to be malignant at the time of recurrence (Fig. 2). For 2/16 patients, there was no improvement in ascites after PVS (12.5%). The median duration of clinical success for patients who underwent PVS for ascites was 242 days (range 0–876 days).

Table 3. Clinical outcomes for portal vein stenting.

Figure 2.

Illustrative patient with ascites. Preprocedure imaging (A) demonstrated a central pancreatic mass with moderate volume ascites. Intraprocedural imaging demonstrated complete occlusion of the main portal vein (B). Resolution of ascites (C) occurred after stent placement and recanalization of the portal vein (D). Symbols: *, moderate volume ascites; **, central pancreatic mass.

For patients who underwent PVS for variceal bleeding, recurrent gastrointestinal (GI) bleeding occurred in 3/10 patients (30%). In one of these three patients, recurrent bleeding occurred 618 days after the PVS stenting and coincided with occlusion of the PV stent. In another of these three patients, recurrent bleeding occurred 1,263 days after the PVS procedure despite the persistent patency of the stent. In the third patient, hemorrhoidal bleeding occurred 357 after the PVS procedure, although at the time of PV stenting, the source of bleeding had been esophageal varices.

The 6‐month, 12‐month, 18‐month, and 24‐month survival rates were 87.5%, 69%, 53%, and 32%, respectively (Fig. 3). The cumulative probabilities of symptom recurrence at 6, 12, 18, and 24 months were 12%, 16%, 26%, and 40% (Fig. 4). There was a significant difference (p < .001) in the probabilities of symptom recurrence between patients whose stents remained patent and those whose stents demonstrated partial or complete occlusion. The mean fraction of remaining life palliated was 87%. As illustrated in Figure 5, all but two patients were found to have improvement in clinical symptoms for the majority of their lives after the procedure.

Figure 3.

Kaplan‐Meier curve demonstrating overall survival rates after portal vein stenting for patients in this study.

Figure 4.

Probability of symptom recurrence after portal vein stenting. (A): Cumulative symptom recurrence. (B): The statistically significant difference in recurrence probability between patients whose stents remained patent (blue line) and those whose stents demonstrated partial or complete thrombosis (red line). Dashed lines represent 95% confidence intervals.

Figure 5.

Duration of clinical success (red) plotted adjacent to duration of survival (blue) for all patients. For all but two patients, the duration of symptom relief closely matched the patient's survival.

There were no major or minor procedure‐related complications.

Stent Patency

Partial or complete stent thrombosis was found in 6/26 patients. For two patients, the thrombosis occurred early and was seen on the first postprocedural imaging study within 2 days and 6 months of stent placement, respectively. For the other four patients, the stent was noted to be patent for at least 200 days before thrombosis became evident on imaging. On both univariate and multivariate analysis, patient age (p = .256), gender (p = .357), indication for stenting (p = .96), type of underlying malignancy (p = .994), cause of portomesenteric stenosis (p = .997), number of stents deployed (p = .80), and type of anticoagulation used (p = .660) were not found to be significant risk factors for stent thrombosis.

Liver and Ascites Volumes After PVS

Liver volumes were measured prior to stenting and on all cross‐sectional imaging studies after stenting. The longitudinal percent change in liver volume after stenting is shown in Figure 6. For patients whose stents remained patent, there was a nonsignificant trend toward sustained increase in liver volumes throughout the study period, compared with patients whose stents exhibited partial or complete thrombosis.

Figure 6.

Liver volumes and ascites volumes after portal vein stenting. The trend lines (solid lines) are locally weighted regression curves, and the shaded gray areas are 95% confidence intervals. There was a nonsignificant trend toward increased liver volumes (A) and decreased ascites volumes (B) in patients with patent stents (blue lines) versus those whose stents developed in‐stent thrombosis (red lines).

Ascites volumes were measured on pre‐ and postprocedural imaging; the amount of fluid drained during paracentesis was also included in the analysis. There was a nonsignificant trend toward a sustained decrease in ascites volumes for patients with patent stents compared with patients whose stents developed partial or complete thrombosis.

Discussion

Portomesenteric stenosis is an important cause of morbidity and mortality in patients with pancreaticobiliary malignancies. Although infrequently caused by benign post‐treatment changes in patients who are cancer free, such stenoses are more commonly the result of external compression by tumor tissue in patients with incurable malignancy. As a result, the primary role of interventions such as PVS in this patient population with limited life expectancy is symptom palliation.

In this study, the palliative role of PVS was emphasized with the FRLP metric and quantitative assessments of liver and ascites volumes. Prior studies have evaluated the efficacy of PVS on portal venous pressures, ascites, and GI bleeding. These studies conclude that PVS is effective in patients with malignant portal venous stenosis, but objective metrics to quantify clinical outcomes are not clearly defined [8], [10], [11], [12], [13]. For example, pre‐and poststenting portal pressure gradients were measured in a number of studies [7], [10], [12], [13], but no direct correlation was made between immediate improvement in pressure gradients and long‐term success or palliation in the studied patients. Hasegawa et al. used a palliative prognostic index for stratifying patients by risk based on a calculated score after evaluating their performance states, oral intake, edema, dyspnea, and delirium. However, no long‐term metric was described to quantitatively assess the patients’ overall improvement or relief in symptoms [7].

Evaluating symptom palliation remains a challenging exercise. Standard definitions for “palliation” do not exist, and there is limited consensus on best practices for clinical trial design [18]. Even prospective trials can be invalidated by their reliance on untested outcome metrics [19]. One useful method that has been used in patients with cancer is the palliation index or ratio, calculated as the duration of response divided by the duration of survival [20]. The FRLP outcome used in this study was based on this technique and provides a result that is normalized by survival time and thus unaffected by biases such as survival bias. The observation that all but two patients demonstrated symptom palliation for the remainder of their lives highlights the substantial clinical impact of PVS.

This study identified stent patency as an important contributor to outcomes with regard to symptom palliation, liver regeneration, and ascites volumes. Previous studies have suggested that patients with malignant stenosis of the portal vein demonstrate an increased risk of stent occlusion compared with patients with benign stenosis [9], although this study did not identify any statistically significant risk factor for stent thrombosis. Although placement of covered stents to prevent tumor invasion may be helpful, there is the additional risk of covering minor portomesenteric branches that may nonetheless influence outcomes. The primary unassisted patency rate in the current study was 77% (20/26), with a median follow‐up duration of 415 days. This result agrees with Novellas et al., who demonstrated a primary unassisted primary patency of 75.6% (n = 11) with a mean follow‐up of 134.4 ± 123.3 days [10]. Despite anticoagulation, Yamakado et al. described a primary patency of 60% with a mean follow‐up of 11.9 months [12]. Of note, because the follow‐up duration of the current study was longer than other studies, and because most stent occlusions occurred after 200 days, the 6‐month primary stent patency in the current study was higher than previously published reports.

Anticoagulation was not shown to have a significant impact on stent patency either in this study or in prior studies. Anticoagulation remains a challenging consideration in this patient population. Given the frequently multiple simultaneous contraindications to anticoagulation, including prior gastrointestinal bleeding, presence of varices, thrombocytopenia, and platelet dysfunction, anticoagulation is often used conservatively. On the other hand, given the patients’ prothrombotic state and the substantial ramifications of stent thrombosis, anticoagulation after PVS must be considered. Striking the appropriate balance and identifying patients who would benefit from anticoagulation are important questions for future studies.

There are several limitations to this study. Because of the study's retrospective design, the accuracy of the clinical data is limited by reporting bias. The retrospective nature also limits the availability of technical details regarding each stenting procedure, such as the motivations for placement of the stents and the measurement of pressure gradients before and after stenting. It is conceivable that different selections in stent configurations could have had an impact on outcomes, a hypothesis that could not be tested with the available data. Another limitation to this single‐arm study is the lack of a control group to compare outcomes if PVS had not been performed. It is not unreasonable to argue, however, that the patients who developed stent thrombosis represent an internal control, as their portomesenteric hemodynamics are not unlike patients with portal vein occlusion. Lastly, although the emphasis of this study was on objectively assessing PVS from a palliative standpoint, the retrospective nature precluded the rigorous evaluation of PVS by patient‐reported outcomes.

Conclusion

As a low risk procedure with a high clinical success rate, PVS can play a substantial role in improving quality of life in patients with portomesenteric stenoses. Interventions, including minimally invasive interventions, are often deferred in patients with advanced malignancy based on the argument of futility. However, from a palliative perspective, procedures that can make a patient's remaining days more comfortable should be considered. We show in this study that the probability of symptom recurrence after PVS in the setting of preserved stent patency is 16% at 12 months. We view the fact that the mortality rate was 31% over the same period as a motivation rather than a deterrent to performing PVS in this patient population. That is, PVS should be considered a single‐encounter intervention that can provide durable symptom relief for patients with limited life expectancy.

Author Contributions

Conception/design: Michael J. Wallace

Provision of study material or patients: Philip Parmet, Roshon Amin, Joshua Kuban, Steven Huang, Armeen Mahvash, David Fogelman, Milind Javle, Michael J. Wallace

Collection and/or assembly of data: Rahul A. Sheth, Sharjeel Sabir

Data analysis and interpretation: Philip Parmet, Roshon Amin, Joshua Kuban, Steven Huang, Armeen Mahvash, David Fogelman, Milind Javle, Michael J. Wallace

Manuscript writing: Rahul A. Sheth, Sharjeel Sabir, Philip Parmet, Roshon Amin, Joshua Kuban, Steven Huang, Armeen Mahvash, David Fogelman, Milind Javle, Michael J. Wallace

Final approval of manuscript: Rahul A. Sheth, Sharjeel Sabir, Philip Parmet, Roshon Amin, Joshua Kuban, Steven Huang, Armeen Mahvash, David Fogelman, Milind Javle, Michael J. Wallace

Disclosures

Sharjeel Sabir: Neuwave, Terumo, Boston Scientific, Merit Medical, Endocare (Other [travel expenses]). The other authors indicated no financial relationships.

(C/A) Consulting/advisory relationship; (RF) Research funding; (E) Employment; (ET) Expert testimony; (H) Honoraria received; (OI) Ownership interests; (IP) Intellectual property rights/inventor/patent holder; (SAB) Scientific advisory board