Transjugular intrahepatic portosystemic shunt for the treatment of refractory ascites

Abstract

Background:

Randomized trials have shown that transjugular intrahepatic portosystemic shunt (TIPS) improves control of ascites compared with serial large volume paracentesis (LVP) in patients with refractory ascites. However, the effect of TIPS on liver transplant-free (LTF) survival is controversial. Our objective was to compare TIPS versus serial LVP on LTF survival in the general population of patients with refractory ascites.

Methods:

This is a retrospective, population-based cohort study using linked administrative health data from Ontario, Canada. Adult patients identified with refractory ascites from January 1, 2008 to December 31, 2016 were included and followed until December 31, 2017. A propensity score was used to match patients treated with serial LVP to those who received TIPS in a 2:1 ratio. LTF survival was evaluated using Kaplan–Meier analysis and Cox proportional hazards regression with TIPS treated as a time-varying exposure.

Results:

Overall, 4,935 patients with refractory ascites were identified and 488 patients were matched (325 serial LVP, 163 TIPS). The mean age was 58 years, 70% were male, 50% had viral hepatitis, the median model for end-stage liver disease (MELD) score was 12, 13% received liver transplant and the 1-year LTF survival was 72%. After TIPS, 80 patients (49%) had no further requirement for LVP by 6 months and 61 patients (37%) never required a repeat paracentesis. In survival analysis, there was marginally worse LTF survival in patients receiving TIPS (TIPS HR 1.29, 95% CI 1.00–1.67; p = .052).

Conclusion:

In this population-based study of patients with refractory ascites, TIPS was associated with improved control of ascites but not improved LTF survival.

The burden of cirrhosis in North America and Europe has increased substantially over the past several decades secondary to viral hepatitis and alcoholic and non-alcoholic fatty liver disease (1). Ascites is the most common complication of cirrhosis developing in over 50% of patients within 10 years of diagnosis (2). The development of ascites is associated with significant morbidity including fatigue, dyspnea, and bacterial infections (3,4). First-line treatment of ascites consists of dietary sodium restriction and combination oral diuretics. Approximately 10% of patients with cirrhosis will develop ascites that is resistant to first-line treatment, termed refractory ascites (5). The development of refractory ascites is associated with an especially poor prognosis with 1-year mortality estimated from 50% to 80% (6–8).

The standard of care for refractory ascites is serial large volume paracentesis (LVP) and consideration for liver transplantation. While LVP provides rapid symptomatic relief and is relatively safe (9), it does not address the underlying cause of ascites and therefore does not prevent recurrence. Transjugular intrahepatic portosystemic shunt (TIPS) has emerged as an alternative treatment option in this population. TIPS is a minimally invasive procedure that involves inserting a stent between the portal vein and the hepatic vein under fluoroscopic guidance. It seeks to treat portal hypertension, the underlying cause of refractory ascites.

Randomized controlled trials (RCT) have compared TIPS to serial LVP for the treatment of refractory ascites (10–16) and multiple meta-analyses have subsequently been published on these trials (17–23). There is overall agreement that TIPS leads to improved control of ascites, but an increased risk of hepatic encephalopathy (HE). The impact of TIPS on survival, however, remains unclear with some trials indicating a survival benefit in favour of TIPS (14–16) while others found that survival is equivocal (11–13) or worse with TIPS (10). The most recent RCT was the only one to evaluate TIPS using self-expandable polytetrafluoroethylene (ePTFE)-covered stents and found that 1-year liver transplant-free (LTF) survival was superior in patients randomized to TIPS (16). Based on these results, the authors suggested that TIPS should now be considered first-line treatment for the management of refractory ascites.

Given the uncertainty surrounding the impact of TIPS on survival in patients with refractory ascites, population-level data can serve as an important adjunct to RCTs. The objective of this study was to compare TIPS and serial LVP on LTF survival in the general population of patients with refractory ascites. The secondary objective was to evaluate the impact of TIPS on the control of ascites.

Methods

Study design

We conducted a retrospective, population-based cohort study utilizing routinely collected administrative health care data from the province of Ontario, Canada. Ontario provides universal health care coverage for its population of approximately 13.5 million through the Ontario Health Insurance Program (OHIP). The population of Ontario is ethnically diverse with 25% belonging to a visible minority and 2% being of Indigenous descent. The primary databases used in this analysis were the Registered Persons Database (RPDB) which includes demographic and vital status information for individuals covered under OHIP, the Canadian Institute for Health Information (CIHI) Discharge Abstract Database (DAD) which captures diagnostic and procedural information from inpatient hospital admissions, the National Ambulatory Care Reporting System (NACRS) which captures diagnostic and procedural information from ambulatory care visits, the OHIP Physician Claims Database which includes all claims made by physicians for universally insured services, and the Ontario Laboratory Information System (OLIS) which includes over 90% of all bloodwork results performed by hospitals and clinical laboratories in Ontario. These databases were linked using unique encoded identifiers at the individual level and analyzed at ICES. This study was approved by the Health Sciences Research Ethics Board at Queen’s University (DMED 1651-13).

Study population and primary exposure

All patients ≥18 years old with both cirrhosis and refractory ascites between January 1, 2008 and December 31, 2016 were identified and followed until December 31, 2017. This time frame was chosen as it occurred after ePTFE-covered stents were approved by Health Canada and after laboratory data became available through OLIS. Patients were identified as having cirrhosis using a validated administrative coding algorithm that requires one inpatient or outpatient code for cirrhosis or non-bleeding esophageal varices (24). Refractory ascites was defined as ≥3 therapeutic paracenteses during a 3-month period with at least one paracentesis occurring ≥1 month after the first. Therapeutic paracenteses were identified using OHIP billing code Z591. Patients were excluded if they 1) were missing in the RPDB, 2) had undergone liver transplantation before meeting the refractory ascites definition, or 3) received TIPS before meeting the refractory ascites definition. The receipt of TIPS was identified using OHIP billing code J057. In those who received TIPS after meeting the refractory ascites definition, we included those with a prior history of variceal bleeding; however, we excluded patients if during their admission for TIPS, procedural codes consistent with variceal hemorrhage were identified (Canadian Classification of Health Intervention codes 1.NA.13.BA-FA, 1.NA.13.BA-X7, and 1.NA.13.BA-Bd) as TIPS was likely for the indication of a refractory variceal hemorrhage in these cases.

Covariates, propensity score matching, and outcome measures

Baseline patient characteristics described at the time of meeting the refractory ascites definition included age, sex, cirrhosis etiology, income quintile, model for end-stage liver disease (MELD) score, congestive heart failure (CHF), hepatocellular carcinoma (HCC), HE and Charlson–Deyo Comorbidity Index (CCI). Cirrhosis etiology was determined using a hierarchical coding algorithm defined previously (1). Income quintiles were determined using Statistics Canada census data (25). MELD score was calculated using the individual values of total bilirubin, serum creatinine, and internalized normalized ratio (INR) from OLIS data (26). The closest lab values within 1 year to the patient’s date of meeting the refractory ascites definition were used to calculate the MELD score. In patients with incomplete MELD score components, multiple imputations for overall MELD score were performed using multiple regression with fully conditional specifications to generate 20 imputed datasets. These data were assumed to be missing at random (27) as evaluation of the baseline characteristics between those with and without complete MELD were similar (data not shown). Variables for the imputation included total bilirubin, creatinine, INR, sex, age, etiology of cirrhosis, and CCI. CHF and HCC were defined using validated coding algorithms (24,28). HE was identified using ICD-10 code K72.90. The CCI was calculated during the year prior to meeting the refractory ascites definition (29). The number of LVP per month after meeting the refractory ascites definition was also described. In those who received TIPS, the number of LVP per month was further described before and after TIPS.

Patients with refractory ascites who were only managed with serial LVP were matched to those who ultimately received TIPS in a 2:1 ratio by propensity score (30). This ratio was chosen as using an N:1 ratio higher than 2:1 when matching on propensity score can result in a biased estimate of the treatment effect (31). Propensity score was calculated using a multiple logistic regression model that included variables known to factor into the clinical decision to recommend TIPS in patients with refractory ascites. This included age, sex, cirrhosis etiology, income quintile, MELD score, HCC, HE, CHF, and rate of LVP per month as a priori predictors of receiving TIPS as a treatment. A nearest-neighbour matching algorithm within a caliper width of 0.2 times the standard deviation of the propensity score logit was employed to match TIPS patients to their most comparable serial LVP counterparts (32). Balance diagnostics between patients who received TIPS and serial LVP were assessed before and after propensity score matching using standardized differences of ≤0.1 for each baseline covariate as a criterion for acceptable balance.

The primary outcome of this study was LTF survival as defined from the date of meeting the refractory ascites definition to either 1) the date of death from the RPDB, or 2) the date of liver transplantation based on OHIP billing codes S294 and S266. The secondary outcome was the difference in the monthly rate of LVP before and after TIPS placement in those who received TIPS.

Statistical analysis

Means and standard deviations were used to describe normally distributed variables and medians and interquartile ranges were used for variables with non-normal distributions. Due to an ICES Privacy Agreement, data which contains small cells (n <5) are not reportable due to the risk of re-identification. Therefore, when small cells are present, the data must be suppressed or aggregated in a fashion where the exact number cannot be calculated. Median, 30-day, 90-day, 1-year and 5-year LTF survival rates were described separately for the overall refractory ascites cohort and the propensity matched cohort using the Kaplan–Meier method. In the propensity matched cohort, Cox proportional hazards regression was used to evaluate the association between TIPS and LTF survival. To account for the matched design, the clustered robust variance estimator was used. TIPS was treated as a time-varying exposure. The misclassification of unexposed time as exposed time in TIPS patients can potentially introduce immortal time bias in favour of TIPS (33). Treating TIPS as a time-varying exposure mitigates this by ensuring that the time between meeting the refractory ascites definition (baseline) and the date of TIPS was treated as unexposed and exposed after the date of TIPS (Figure 1). To evaluate the effect of TIPS on control of ascites, the rates of monthly LVP before and after TIPS were calculated using the total number of LVP during the overall unexposed person-time and the total number of LVP during the overall exposed person-time. These rates were compared using Poisson regression. All statistical analyses were performed using SAS (version 9.4; SAS Institute, Cary, NC).

Figure 1:

Survival analysis for the propensity matched cohort with refractory ascites (n = 488)

Results

Baseline demographics

A total of 4,935 patients met the definition for refractory ascites (Figure 2, Table 1). Overall, patients were predominantly male (68%) with a mean age of 62.7 (SD 11.9) years. The most common etiologies of cirrhosis were viral hepatitis (50%) and alcohol-related disease (26%) and the median MELD score was 14. Eight percent of patients received a liver transplant. After meeting the refractory ascites definition, 87% required an average of ≤2 LVP per month while 13% required >2 LVP per month. The median LTF survival was 6.1 (interquartile range [IQR] 5.8–6.5) months with 1-year and 5-year LTF survival rates of 36% and 15% (see supplemental Table S1).

Figure 2:

Cohort development

Table 1:

Baseline characteristics at refractory ascites definition for the overall cohort

	Overall	TIPS	Serial LVP	Standardized difference	p-value
	N = 4,935	n = 163	n = 4,772
Age, mean (SD)	62.7 (11.9)	57.7 (9.3)	62.9 (11.9)	0.48	<.001
Sex: male, % (no.)	67.6 (3,338)	71.2 (116)	67.5 (3,222)	0.08	.259
Income quintile, % (no.)
1	23.8 (1,173)	17.8 (29)	24.0 (1,144)	0.15	.291
2	21.2 (1,047)	24.5 (40)	21.1 (1,007)	0.08
3	18.9 (935)	16.6 (27)	19.0 (908)	0.06
4	19.0 (937)	23.3 (38)	18.8 (899)	0.11
5	16.5 (814)	17.8 (29)	16.5 (785)	0.04
Missing	0.6 (29)	0 (0)	0.6 (29)	0.11
Cirrhosis etiology, % (no.)
Viral hepatitis	50.3 (2,484)	53.4 (87)	50.2 (2,397)	0.06	<.001
Alcohol-related	25.5 (1,256)	31.9 (52)	25.2 (1,204)	0.17
NAFLD/cryptogenic	21.4 (1,056)	8.6 (14)	21.8 (1,042)	0.15
Other*	2.8 (139)	6.1 (10)	2.7 (129)	0.38
MELD†, median (IQR)	14 (11-18)	13 (10–16)	14 (11–18)	0.27	<.001
HCC, % (n)	13.0 (643)	1–5 (<5)§	10–15 (477–716)§	0.45	<.001
HE, % (n)	12.0 (591)	6.7 (11)	12.2 (580)	0.19	.037
CHF, % (n)	20.2 (999)	5.5 (9)	20.7 (990)	0.46	<.001
LVP per month‡, % (no.)
≤1	66.8 (3,298)	46.6 (76)	67.5 (3,222)	0.48	<.001
2	20.4 (1,009)	36.8 (60)	19.9 (949)	0.40
3	7.9 (388)	12.3 (20)	7.7 (368)	0.17
≥4	4.9 (240)	4.3 (7)	4.9 (233)	0.03
CCI, % (n)
0	25.9 (1,279)	35.0 (57)	25.6 (1,222)	0.20	<.001
1	12.4 (612)	17.2 (28)	12.2 (584)	0.14
2–3	28.0 (1,381)	29.4 (48)	27.9 (1,333)	0.04
≥4	33.7 (1,663)	18.4 (30)	34.2 (1,633)	0.13
Liver transplant, % (no.)	8.0 (396)	14.1 (23)	7.8 (373)	0.20	<.001

^*Includes hereditary hemochromatosis, primary biliary cholangitis, primary sclerosing cholangitis, autoimmune hepatitis, Wilson disease, and alpha-1 antitrypsin deficiency

^†MELD score closest to refractory ascites definition

^‡Average per month, only paracentesis before TIPS considered in those who received TIPS

^§Exact numbers not displayed due to ICES Privacy Policy for cells <5

TIPS = Transjugular intrahepatic portosystemic shunt; LVP = Large volume paracentesis; NAFLD = Non-alcoholic fatty liver disease; MELD = Model for end-stage liver disease; IQR = Interquartile range; HCC = Hepatocellular carcinoma; HE = Hepatic encephalopathy; CHF = Congestive heart failure; CCI = Charlson–Deyo co-morbidity index

Within the study cohort, 163 patients (3%) received TIPS after the development of refractory ascites. The median time from the diagnosis of refractory ascites to TIPS was 5.3 (IQR 2.5–9.7) months. Compared with patients managed with serial LVP, those who received TIPS were younger; more likely to have viral hepatitis or alcohol-related disease; had lower MELD scores, less CHF, HCC and HE, lower comorbid illness and higher monthly rates of LVP; and were more likely to receive a liver transplant (Table 1).

Liver transplant-free survival by TIPS in the propensity matched cohort

After propensity score matching, a total of 488 individuals were included for the survival analysis (163 TIPS, 325 serial LVP; Figure 2, Table 2). With the exception of CCI, the TIPS and serial LVP patients were comparable with respect to all matching variables with the standardized difference for each being ≤0.1. The TIPS group had more patients with a CCI of 0 (standardized difference = 0.16) and fewer with a CCI ≥4 (standardized difference = 0.14).

Table 2:

Baseline characteristics at refractory ascites definition for the propensity matched cohort

	Overall	TIPS	Serial LVP	Standardized difference	p-value
	n = 488	n = 163	n = 325
Age, mean (SD)	57.9 (10.9)	57.7 (9.3)	58.0 (11.6)	0.04	.781
Sex: male, % (no.)	69.5 (339)	71.2 (116)	68.6 (223)	0.05	.564
Income quintile, % (n)
1	15.0 (73)	17.8 (29)	13.5 (44)	0.02
2	24.6 (120)	24.5 (40)	24.6 (80)	0.00
3	19.7 (96)	16.6 (27)	21.2 (69)	0.06	.475
4	21.1 (103)	23.3 (38)	20.0 (65)	0.1
5	19.7 (96)	17.8 (29)	20.6 (67)	0.05
Cirrhosis etiology, % (no.)
Viral hepatitis	52.5 (256)	53.4 (87)	52.0 (169)	0.02	.985
Alcohol-related	32.2 (157)	31.9 (52)	32.3 (105)	0.01
NAFLD/cryptogenic	9.2 (45)	8.6 (14)	9.5 (31)	0.02
Other*	6.1 (30)	6.1 (10)	6.2 (20)	0.08
MELD†, median (IQR)	12 (10–16)	13 (10–16)	12 (10–16)	0.02	.809
HCC, % (no.)	2.3 (11)	1–3 (<5)§	1–3 (3–9)§	0.04	.663
HE, % (no.)	7.0 (34)	6.7 (11)	7.1 (23)	0.01	.893
CHF, % (no.)	5.9 (29)	5.5 (9)	6.2 (20)	0.03	.780
LVP per month‡, % (no.)
≤1	44.5 (217)	46.6 (76)	43.4 (141)	0.02
2	38.6 (188)	36.8 (60)	39.4 (128)	0.04
3	12.7 (62)	12.3 (20)	12.9 (42)	0.00	.927
≥4	4.3 (21)	4.3 (7)	4.3 (14)	0.06
CCI, % (n)
0	32.4 (158)	35.0 (57)	31.1 (101)	0.16
1	16.2 (79)	17.2 (28)	15.7 (51)	0.05
2–3	28.9 (141)	29.4 (48)	28.6 (93)	0.06	.468
≥4	22.5 (110)	18.4 (30)	24.6 (80)	0.14
Liver transplant, % (no.)	7.8 (38)	14.1 (23)	18.8 (61)	0.12	.446

^*Includes hereditary hemochromatosis, primary biliary cholangitis, primary sclerosing cholangitis, autoimmune hepatitis, Wilson disease, and alpha-1 antitrypsin deficiency

^†MELD score closest to refractory ascites definition

^‡Average per month, only paracentesis before TIPS considered in those who received TIPS

^§Exact numbers not displayed due to ICES Privacy Policy for cells <5

TIPS = Transjugular intrahepatic portosystemic shunt; LVP = Large volume paracentesis; NAFLD = Non-alcoholic fatty liver disease; MELD = Model for end-stage liver disease; IQR = Interquartile range; HCC = Hepatocellular carcinoma; HE = Hepatic encephalopathy; CHF = Congestive heart failure; CCI = Charlson–Deyo co-morbidity index

LTF survival in the propensity score matched cohort is shown in supplemental Table S1. The median LTF survival was 11.4 (IQR 9.2–13.9) months with 1-year and 5-year survival rates of 72% and 33%. Cox proportional hazards regression was performed to evaluate the association between TIPS and LTF survival. The CCI was included as a confounder in the model as the standardized difference between the groups after propensity score matching remained >0.1. After adjustment, patients who received TIPS had marginally worse LTF survival compared with those who received serial LVP alone with the lower limit of the confidence interval reaching 1 (HR 1.29, 95% CI 1.00–1.67; p = .052; Table 3, Figure 3).

Table 3:

Cox proportional hazards regression for the outcome of LTF survival in the propensity matched cohort of patients with refractory ascites (n = 488)

	Unadjusted			Adjusted
	HR	95% CI	p	HR	95% CI	p
TIPS*	1.32	1.02–1.71	.037	1.29	1.00–1.67	.052
CCI ≥4†	1.36	1.03–1.78	.029	1.32	0.99–1.75	.056

^*versus large volume paracentesis

^†versus CCI <4

LTF = Liver transplant-free; HR = Hazard ratio; TIPS = Transjugular intrahepatic portosystemic shunt; CCI = Charlson–Deyo comorbidity index

Figure 3:

Adjusted cumulative hazard function curves for death/LT in propensity matched cohort of refractory ascites (n = 488)

Control of ascites in patients with TIPS

In the 163 patients who received TIPS, the average number of LVP per month was reduced after TIPS compared with before TIPS (median LVP per month 0.21, 95% CI 0.20–0.22 versus 1.38 per month, 95% CI 1.31–1.44; p < .001) (see Figure 4). After TIPS, 80 patients (49%) had no further requirement for LVP by 6 months and 61 patients (37%) never required a repeat paracentesis during follow-up.

Figure 4:

Monthly rates of large volume paracenteses in patients with refractory ascites who received TIPS (n = 163) showing median monthly LVP rate per person (A) and monthly rate of LVP using overall unexposed and exposed person-time (B) compared using Poisson regression

Discussion

To our knowledge, this is the first study to evaluate the association between TIPS and survival in patients with refractory ascites at the population level. After propensity score matching, there was no evidence that TIPS was associated with improved LTF survival. However, the use of TIPS at the population level is associated with improved control of ascites as evidenced by a substantial reduction in the requirement for LVP.

A diagnosis of refractory ascites is made when symptomatic ascites is not managed with a combination of sodium restriction and diuretic therapy requiring the use of serial LVP (5). Given that our study relies on administrative data, we were unable to ascertain the use and effectiveness of diuretics and therefore had to define our population based on the need for multiple LVPs alone. It is therefore important to ensure our study cohort is comparable to previous clinical studies of patients with refractory ascites. Our cohort is similar in age, sex, cause of cirrhosis, and MELD score to previously published RCTs and observational studies (16,34), indicating that our cohort is representative of patients with refractory ascites in clinical practice.

There have been seven published RCTs evaluating the use of TIPS in patients with refractory ascites. However, these studies are plagued by small sample size and other methodological limitations. The largest study randomized 109 patients (13) and the most recent meta-analyses of the first six studies included a total of 390 patients (22,23). The results from the individual trials have been conflicting with some showing that TIPS is associated with improved survival while others do not. Further, the results from several meta-analyses have been equally equivocal. Six of the RCTs were performed in the setting of bare TIPS stents. Given the strong evidence that ePTFE-covered stents are associated with superior outcomes, there has been a suspicion that TIPS may provide a survival advantage in the setting of covered stents. The seventh and most recent RCT used ePTFE-covered stents and randomized 62 patients to TIPS or serial LVP with the results suggesting improvement in 1-year LTF survival with TIPS (93% versus 52%; p = .003) (16). However, concerns have been raised regarding the patient recruitment and methodology from this trial (35). It is also notable that the study took seven years and four tertiary care centres to recruit these 62 patients highlighting the challenges of using an RCT design to answer this question. Therefore, there remains clinical equipoise.

Observational population-based studies can be considered adjuncts to RCTs or substitutes when RCTs are not feasible, and are especially useful for evaluating the effectiveness of interventions in real-world clinical practice (36,37). However, careful study design and statistical considerations are required to avoid bias that can lead to inaccurate conclusions. This is particularly important when evaluating the effect of an intervention where the treatment exposure is time-dependent (38). Failure to account for this can lead to “immortal time bias” where survival in the treatment group is over estimated due to the fact that those in the treatment group must survive long enough to receive the intervention. Although there have been no other population-based observational studies, three single-centre studies have evaluated the association between TIPS and survival in this patient population (34,39,40). The results of the first two studies suggested that TIPS is associated with improved survival, however neither study design accounted for immortal time bias (34,39). Failure to account for this bias is exemplified in our study where TIPS occurred at a median of 5 months after the development of refractory ascites. To demonstrate the size of the problem of immortal time bias for our readers, we analyzed the association between TIPS and survival ignoring the immortal time bias issue (see supplemental Table S2). In contrast to our main finding, in this demonstration analysis we observed a biased hazard ratio of 0.49, (95% CI 0.39–0.60; p < .001). This is a stark example of the impact of attributing follow-up time to an intervention that has not yet occurred. By definition, the exposed group (TIPS) survives longer simply because they had to in order to receive the intervention. We think the observed association with TIPS and improved LTF survival in these previous studies is likely confounded by immortal time bias. The third study was designed similarly to ours with propensity score matching and TIPS treated as a time-varying exposure (40). Although their study cohort was much smaller (n = 184), the investigators found no association between TIPS and LTF survival but had an almost identical point estimate to ours (HR 1.39, 95% CI 0.38–5.10).

Our results support the current American (5) and European (41) guidelines which suggest that the decision to proceed to TIPS in patients with refractory ascites should be with the goal of improving control of ascites as defined by a decrease in the requirement for LVP as opposed to improved survival. Despite the fact that our observed HR suggests that LTF survival may be inferior in those who received TIPS, it should not be interpreted to imply that TIPS be avoided in this population. Given the limitations of observational research and recent literature debating the ubiquitous use of the p value cutoff of .05 to represent statistical significance (42), we believe these data support the position that TIPS is not associated with improved LTF survival compared with serial LVP. Therefore, if the goal is to improve control of ascites, TIPS is an appropriate therapeutic option. However, given that TIPS is not without risk, most commonly the development or worsening of HE, an informed discussion about the potential risks and benefits between the patient, their support system and providers is of utmost importance. Finally, recent data has shown that TIPS performed in high volume centres are associated with both short-term (43) and long-term (44) survival benefits. Therefore, if the decision to proceed to TIPS is made, the procedure should be performed in a high volume centre to optimize outcomes.

Consistent with previous RCTs, our population- level data confirms that the use of TIPS results in superior control of ascites. In our cohort, 49% patients who received TIPS had no need for LVP by 6 months post-TIPS and 37% never required further LVP with an over 6-fold reduction in the number of monthly LVP. This is consistent with the most recent meta-analysis where 51% of patients had complete resolution of ascites post-TIPS (22) and an observational study where 45% had complete clearance of ascites by 6 months (45).

Importantly, this study also provides an updated estimate of survival in a contemporary cohort of patients with refractory ascites. Our results highlight that in patients with refractory ascites, the overall prognosis is poor and has not changed substantially over the past several decades despite advances in the management of patients with liver disease. Previous survival estimates are based on single-centre cohorts in the early 2000s with 1-year survival estimated to be approximately 50% (7,46). Our cohort evaluated almost 5,000 patients with refractory ascites with a 1-year LTF survival of 36%, median overall survival of 6 months, 8% surviving to liver transplant, and less than 5% receiving TIPS. This supports the notion that the development of refractory ascites in the absence of liver transplantation should be considered a palliative situation where it may be appropriate to begin advanced care planning and goals of care discussions with patients and their families (47).

The strength of our study stems from the size and inclusivity of our patient cohort. Since we utilized linked administrative data sets from a province that provides universal health care, we were able to include essentially all patients with refractory ascites in a large, demographically diverse population over a long time period, with reliable follow-up and accurate outcome assessment.

The results of this study must also be considered in the context of methodological limitations. First, the ICD-10 code used to identify patients with a history of HE has not been previously validated in our data. Second, given the administrative nature of our data, we are unable to determine which patients received ePTFE-covered stents as opposed to bare stents. To mitigate this limitation, we chose a study period after the Health Canada approval of ePTFE-covered stents. Third, we were unable to ascertain patient-level details regarding the management of refractory ascites which may impact survival, including the use of intravenous albumin (48), non-selective beta-blockers (49), a history of spontaneous bacterial peritonitis, and the use of spontaneous bacterial peritonitis prophylaxis (5). Further, given that the data do not include details on the use of effectiveness of salt restriction or diuretics, the cohort likely contains a population with both refractory and recurrent ascites. Finally, as with all observational data, we are unable to account for residual confounding of variables that we are not able to capture within our datasets. However, we calculated the E-value, which is the minimum strength of association that an unmeasured confounder would need to have with both the treatment and the outcome to fully explain away a specific treatment-outcome association (50). For our hazard ratio to be shifted to 0.9 favouring TIPS with LTF survival, this unmeasured confounder would need to have risk ratio of 1.89 with the lower bound CI of 1.36. Given that we have accounted for the majority of known confounders in the relationship between TIPS and survival we feel that an unidentified confounder of this magnitude would be unlikely.

In conclusion, in this large, population-based study of patients with refractory ascites, there was no association between the use of TIPS and improved LTF survival when compared with serial LVP. However, the use of TIPS does result in significant improvement in ascites control. These results support current clinical practice guidelines that state that the use of TIPS should be with a goal of better ascites control but not necessarily improved survival. Finally, the prognosis of patients with refractory ascites is poor and in the absence of liver transplantation, goals of care discussions with patients and families should be considered.

Supplementary Material

Table S1:

Liver transplant-free survival in patients with refractory ascites in the overall and propensity matched cohorts

LTF survival	Overall	Propensity matched
	n = 4,935	n = 488
Median months (IQR)	6.1 (5.8–6.5)	11.4 (9.2–13.9)
30- day, % (StdErr)	86.9 (0.48)	99.4 (0.61)
90-day, % (StdErr)	66.2 (0.67)	93.9 (1.88)
1-year, % (StdErr)	35.9 (0.68)	71.8 (3.53)
5-year, % (StdErr)	14.8 (0.58)	33.0 (3.95)

LTF = Liver transplant-free; IQR = Interquartile range; StdErr = Standard error of the mean; TIPS = Transjugular intrahepatic portosystemic shunt; LVP = Large volume paracentesis

Table S2:

Cox proportional hazards regression for the outcome of LTF survival in the propensity matched cohort of patients with refractory ascites using TIPS as a standard exposure (n = 488)

	Unadjusted			Adjusted
	HR	95% CI	p	HR	95% CI	p
TIPS†	0.49	0.40–0.60	<.001	0.49	0.39–0.60	<.001
CCI ≥4‡	1.36	1.04–1.77	.025	1.42	1.11–1.80	.004

^†versus large volume paracentesis

^‡versus CCI <4

HR = Hazard ratio; TIPS = Transjugular intrahepatic portosystemic shunt; CCI = Charlson–Deyo comorbidity index

Funding Statement

Funding: This study was supported by an American Association for the Study of Liver Disease (AASLD) Foundation Clinical, Translational, and Outcomes Research Award in Liver Disease (JAF) and a Southeastern Ontario Academic Medical Association New Clinician Scientist Award (JAF).

Ethics approval:

The study protocol was approved by an ethics committee and the ethics certificate information is available from the authors upon request.

Informed consent:

A waiver for informed consent was obtained from the ethics committee.

Registry and registration no. of the study/trial:

N/A

Funding:

This study was supported by an American Association for the Study of Liver Disease (AASLD) Foundation Clinical, Translational, and Outcomes Research Award in Liver Disease (JAF) and a Southeastern Ontario Academic Medical Association New Clinician Scientist Award (JAF).

Disclosures:

The authors have nothing to declare.

Peer Review:

This article has been peer reviewed.