Design and rationale of a multicenter defeat alcoholic steatohepatitis trial: (DASH) randomized clinical trial to treat alcohol-associated hepatitis

Abstract

Background/aims:

Despite high mortality of alcohol-associated hepatitis, there has been limited advancement in treatment strategies. Defeat Alcoholic Steatohepatitis (DASH) is a multicenter, randomized, double-blind controlled trial whose primary objective was to evaluate the safety and efficacy of a novel combination of 3 drugs targeting different perturbations in AH.

Methods:

Severe AH was diagnosed by liver biopsy or clinical and biochemical criteria and model for end stage liver disease (MELD) score ≥ 20 stratified by MELD scores (20–25 and ≥ 26) and randomized to a combination of an interleukin receptor 1 antagonist, Anakinra(100 mg daily for 14 days) to suppress acute inflammation, pentoxifylline (400 mg three times a day for 28 days) to prevent hepatorenal syndrome, and zinc sulfate (220 mg orally once daily for 6 months) or the standard of care therapy including methylprednisolone 32 mg orally once daily for 28 days. The primary efficacy outcome was the unadjusted log-rank test of the Kaplan-Meier survival estimates for the two treatment groups at 180 days.

Results:

Between July 2012 to March 2018, 500 subjects with severe AH were screened of which 104 subjects were enrolled with MELD score of 25.6 ± 3.2 (20.0–35.0) in the investigational arm and 25.8 ± 4.5 (20.0–40.0) in the standard of care arm. Causes of screen failures included not meeting eligibility criteria (n = 347), declining to participate (n = 39), and other reasons (n = 10).

Conclusions:

Data from the DASH consortium studies will determine if a combination of drugs targeting multiple mechanisms of injury in the severe AH will improve clinical outcomes.

1. Introduction

Alcohol-associated liver disease (ALD), a major cause of liver disease worldwide, accounts for 50% of liver related deaths in the United States, 1% of the gross national product, and is estimated to become more prevalent over the next decade [1,2]. Alcohol-associated hepatitis (AH), a subset of ALD, occurs in 25–35% of patients with chronic heavy alcohol use [3]. The mortality of untreated patients with severe AH, defined as having a Maddrey discriminant function (MDF) > 32 and model for end stage liver disease (MELD) score > 20, is as high as 50% at 90 days [3-5]. From 2002 to 2010, emergency utilization and hospitalization rates for AH increased by 25–30% and costs increased from $27,124 to $46,264 [6-9]. The average cost of hospitalization for AH is now similar to that for diabetes and higher than costs for acute myocardial infarction, acute cerebrovascular disease and acute pancreatitis [6,8,10]. Total healthcare costs for commercially insured AH patients averaged $145,000 over 5 years with a morality of 67% during that time [8]. In Denmark, the 84-day mortality increased from 14 to 24% with 5-year mortality of 56% between 1999 and 2008 [11]. Similar data were reported from the VA Cooperative study in the 1990's [12].

Corticosteroids are recommended for AH by multiple society practice guidelines [4,13,14] and improve short-term survival (28 day), but 6 month mortality remains high (30–50%) [4,15]. However, the STOPAH trial and a recent meta-analysis of 2111 patients, both published after the initiation of this trial, found that corticosteroid monotherapy did not improve morality beyond 28 days [16,17]. At present, no specific therapies improve 6 month mortality despite major advances in critical care and supportive care over the last 3 decades [15], leading to the use of 6 months mortality as the primary endpoint in the current trial. Well-defined diagnostic criteria and end-points for clinical trials in severe AH are urgently needed to allow comparison of data across studies.

The DASH study was designed by the site principal investigators in consultation with NIAAA program officials. The criteria for selecting the proposed treatments for a Phase IIB study included current FDA approval of the medications for other indications and known safety profile in other disease states [18]. Medications were selected based on preclinical and clinical studies that showed high levels of IL-1β, TNFα, and other pro-inflammatory cytokines, likely triggered by gut derived endotoxins and other bacterial products as a consequence of loss of integrity of the gut mucosa [19,20]. Since there are a number of pathogenic mechanisms that contribute to the severity and outcomes in AH, we hypothesized that combining therapeutic agents that target different pathways would be more effective than monotherapy that targets only one of the potential mechanisms of injury.

In the DASH clinical trial, the current accepted standard of care, corticosteroids (methylprednisolone 32 mg/d for 28 days) therapy, was compared with a drug regimen designed to decrease inflammation (interleukin-1 receptor antagonist for 28 days) [21], protect against acute kidney injury (pentoxifylline for 28 days) [22] and improve gut integrity (oral zinc for 6 months) [23].

2. Trial design

The clinical trial was designed as a superiority trial to determine if a combination of drugs would be better than the current standard of care, corticosteroids with supportive measures, in patients with severe AH defined by MELD score > 20. Patients who were unable or unwilling to participate in the intervention study were enrolled in an observational study to determine the natural history of severe AH because most studies on the natural history of AH had been performed in an era prior to modern intensive care and improvement in nutritional status of the population [24]. Finally, a biorepository of blood and peripheral blood mononuclear cells and DNA was created to develop non-invasive diagnostic biomarkers and predictors of response to therapy including genome wide association studies. The study was registered in Clinicaltrials.gov (NCT04072822).

2.1. Primary study endpoints/secondary endpoints

The primary study endpoint for this study was 6-month mortality. Secondary outcomes included 30 and 90 day mortality; changes in MELD score at 30, 90, and 180 days and changes in gut mucosal integrity (serum endotoxin levels) and serum cytokine profiles (IL-1, TNFα, etc.) at 2, 7, 30, 90 and 180 days.

2.2. Study design/type

The trial was a randomized, placebo-controlled, parallel group clinical trial (Figs. 1,2). Eligible patients were randomized into one of two groups in a 1:1 ratio to either the combination group or methylprednisolone. Patients with severe AH were stratified and randomized separately based on MELD scores (patients with MELD 20–25, patients with MELD 26–31, and patients with MELD ≥32) to ensure comparable severity and equal numbers in the 2 treatment groups. The trial was double-blinded by re-packaging all oral medications into capsules to ensure that each group received an equal number of capsules each day for the entire duration of the study. Patients were treated with pentoxifylline 400 mg three times daily for 28 days PLUS zinc supplements for a total of 6 months (Group A) or methylprednisolone 32 mg daily for 28 days PLUSien placebo capsules daily for the remaining 5 months (Group B). An injection of either anakinra (100 mg) (Group A) or a placebo in a matched syringe (Group B) was administered subcutaneously each day for 14 days (Supplementary Table 1). Since there are no FDA approved treatments for severe AH, an investigational new drug application was obtained from the FDA prior to the start of the study. Patients, investigators and physicians involved in the care of the patients were masked to the treatment. Table 1 indicates the timing of samples takentaken both for research purposes and for ensuring standard of care for all subjects.

Fig. 1.

Overall therapeutic strategy of the clinical trial for severe alcohol-associated hepatitis in the DASH consortium. Patients with acute alcohol-associated hepatitis were stratified into moderate and severe disease based on MELD score cut-off and patients with severe disease were then randomized into standard of care or combination treatment arm.

Fig. 2.

Schematic of subject enrollment strategy in severe AH. Screening, evaluation for eligibility, and randomization of subjects with alcohol-associated hepatitis in the DASH consortium.

Table 1.

Assessment plan during follow up.

Purpose	Test	Day 0	Day 2–6*	Day 7	Day 8–27*	Day 28	2 mos *†	3 mos †	4 mos *†	5 mos *†	6 mos †	12 mos †
Standard	Clinical assessment	x	x	x	x	x	x	x	x	x	x
	Alcohol intake	x	x	x	x	x	x	x	x	x	x
	Liver panel (bilirubin, alkaline phosphatase, AST, ALT) and GGT	x	x	x	x	x	x	x	x	x	x
	Serum electrolytes, creatinine, BUN	x	x	x	x	x	x	x	x	x	x
	Serum albumin, total protein	x	x	x	x	x	x	x	x	x	x
	Prothrombin time, INR	x	x	x	x	x	x	x	x	x	x
	Complete blood count	x	x	x	x	x	x	x	x	x	x
	Haptoglobin	x									x
	Lipid panel	x									x
	Amylase, lipase	x
	HIV	x
	HAV Ab, HBsAg, HBcore Ab (total)	x
	HCV Ab	x
	PPD	x
	Blood cultures x2	x
	Urine culture & urinalysis	x
	Urine toxicology	x
	Ascites for albumin, total protein, cell count and differential (if present)	x
	Ascites culture	x
	Stool studies (if diarrhea): C. difficile, culture	x
	Complete abdominal ultrasound or cross-sectional radiology	x									x
Research	Gut permeability studies	x ‡				x		x			x
	Serum zinc, urine zinc	x				x					x
	Stool (DNA microbiome)	x									x
	Urine (KIM-1)	x	x day2	x		x		x			x
	Biomarkers = Cytokines	x	x day2	x		x		x			x
	DNA = Pharmacogenomics	x
	Serum micro RNA	x		x		x		x			x
	Monocyte prep (U Mass, U Louisville)	x		x							x
	Mortality assessment											x

Study subjects.

All human subjects were patients with a clinical presentation consistent with alcohol-associated hepatitis (AH) [25]. The total subject population was planned to be 130 patients with AH, 65 in each treatment group based on MELD strata. The age range was planned to be 21–70 years. Inclusion and exclusion criteria for patient recruitment are shown in Table 2. Each site had a study coordinator, and investigators formulated a recruitment strategy that was optimized for each site including coordinating with the emergency room, admission team and primary care physicians. The schedule of evaluations is shown in Supplementary Tables 2, 3.

Table 2.

Patient Recruitment criteria at inclusion.

Inclusion criteria Ability to provide informed consent by subject or appropriate family member Age between 21 and 70 years Recent alcohol consumption > 50 g/d for > 6 months, continuing within two months before enrollment At least 2 of the following symptoms of acute alcoholic hepatitis: Anorexia, nausea, right upper quadrant pain, jaundice, leukocytosis, hepatomegaly AND Liver biopsy showing alcoholic hepatitis (steatohepatitis) OR ultrasound of liver showing increased echogenicity OR CT scan showing decreased attenuation of liver compared to spleen OR MRI showing fatty liver (decreased signaling intensity on T1 weighted images) If liver biopsy confirms diagnosis of alcoholic hepatitis then requirement for AST elevation > 50 is waived. The liver biopsy must be done within 60 days of study enrollment. The AST levels: AST > Or equal to 50 IU/mL but less than 500 IU/mL AST > ALT, ratio AST/ALT > 1.5; ALT < 200 IU/mL or biopsy proven alcoholic hepatitis. Model for End-Stage Liver Disease (MELD) ≥ 20 and Maddrey DF ≥ 32. Willingness to utilize two reliable forms of contraception (both males and females of childbearing potential) from screening through the first 6 weeks of the study.

Exclusion criteria Hypotension with BP < 80/50 after volume repletion Pregnancy; incarceration; inability to provide consent or lack of appropriate family member Signs of uncontrolled systemic infection: Fever > 38 °C and positive blood or ascites cultures and on appropriate antibiotic therapy for ≥ 3 days within 3 days of inclusion Acute gastrointestinal bleeding requiring > 2 units blood transfusion within the previous 4 days Undue risk from immunosuppression: Positive HBsAg; a positive skin PPD skin test, a positive quantiferon, or history of treatment for tuberculosis; history of any malignancy except skin cancer but including hepatocellular carcinoma within the last five years; HIV infection Recent previous treatment with corticosteroids or other immunosuppressive medications including specific anti-TNF therapy (not including pentoxifylline), calcineurin inhibitors within the previous 3 months. Treatment with corticosteroids for ≤ 3 days prior to baseline is acceptable. Evidence of acute pancreatitis: CT evidence or amylase or lipase > 5 X upper limit of normal (ULN). Serious cardiac, respiratory or neurologic disease or evidence of other liver diseases such as autoimmune hepatitis, primary biliary cirrhosis, primary sclerosing cholangitis, Wilson disease, hemochromatosis, secondary iron overload due to chronic hemolysis, alpha-1-antitrypsin deficiency Acute or chronic kidney injury with serum creatinine > 3.0 mg/dL.

2.2.1. Sampling plan

There was no sampling per se. All patients who fulfilled the eligibility criteria and were willing to participate in the study were enrolled in the study for screening and, if found to be fully eligible were randomized to treatment, depending on their severity.

2.2.2. Potential benefits

The potential benefits to every subject participating in the clinical trial are related to receiving optimal patient-centered care for their condition. This includes ensuring mobilization of available resources for rehabilitation programs and education of all involved patients and their families on the possible consequences of alcohol consumption. While ideally occurring in the care of all patients with alcohol-associated hepatitis, the study team provided patients with the personal healthcare navigation assistance that is not available to every individual.

There may be immediate additional benefits from the medical therapy. Although corticosteroids or pentoxifylline may improve short term outcomes, neither provides sustained benefit when measured at 6 months. If the results of this trial show an improvement above the current standard care, all patients with severe alcohol-associated hepatitis may benefit from extrapolation of the findings. If the outcomes are negative, other subjects will be protected from the risks of treatment that has no additional benefit. Additional societal benefits include those with respect to alcohol consumption, alcoholism and alcohol associated liver disease.

2.3. Study Interventions

2.3.1. Rationale for anakinra in AH

Inflammation in AH is mediated by activation of innate immune responses and release of inflammatory cytokines in the liver [19,26-28]. Preclinical studies from our group identified the NLRP3 inflammasome- IL-1ß axis as a critical pathway in inflammation and hepatocyte injury in mouse models of alcohol-induced liver disease [29-31]. Blocking IL1β was less effective than use of anakinra, an inhibitor of both IL1α and β receptors [31]. Anakinra is FDA approved for treatment of rheumatoid arthritis, Still's disease, familial cold auto-inflammatory and Muckle-Wells syndromes [32-34]. Anakinra has an excellent safety profile compared to other biological therapies [35]. A meta-analysis of randomized studies using anakinra reported a 1.4% incidence of serious infections with anakinra, compared with 0.5% with placebo that was not significant after adjustment [35]. Other FDA approved IL-1 inhibitors, canakinumab and rilonacept, target IL-1β but not IL-1α [36]. The short half-life of anakinra may be beneficial in quickly stopping therapy in the event of infection.

2.3.2. Rationale for pentoxifylline (PTX) in severe AH

Pentoxifylline is a phosphodiesterase (PDE) inhibitor that increases intracellular cAMP levels, decreases generation of TNFα, increases the anti-inflammatory cytokine, IL10, and improves microcirculation and gut barrier function [37-42]. Kupffer cells from alcohol-fed rats showed decreased intracellular cAMP levels and increased expression of TNFα compared to those from pair-fed rats [43]. Phosphodiesterase inhibitors increase cAMP and prevent LPS-induced liver injury in alcohol-fed rats, supporting the rationale for PTX in AH [44]. PTX was as effective as, or possibly better than, corticosteroids in reducing mortality of severe AH in one study [45] but not in another [46], and was reported to reduce the risk of hepatorenal syndrome in patients with severe AH [47].

2.3.3. Rationale for zinc supplements in severe AH

Zinc, an essential trace element, is required for normal cell growth, development and differentiation, and is a critical component of critical zinc transcription factors and tight junction proteins [48]. Zinc deficiency is common in ALD, due to altered metabolism, decreased dietary intake, increased urinary excretion, activation of certain zinc transporters, and induction of hepatic metallothionein [49], and zinc deficiency results in skin lesions, poor wound healing, decreased hepatocyte regeneration, altered mental status and impaired immune function [48,50]. There is both clinical and preclinical evidence that zinc deficiency is common in patients with heavy ethanol use and contributes to impaired gut mucosal integrity [23]. Zinc replacement has been shown to improve alcohol induced hepatic injury in rodent models [51,52]. In this clinical trial, zinc supplementation was included to potentially improve gut mucosal integrity through its effects on tight junction proteins [51,52]. This manuscript will focus on the design and implementation of studies in patients with severe AH.

2.3.3.1. Treatment arms.

In this clinical trial, corticosteroids (methylprednisolone 32 mg/d for 28 days) were compared with a drug regimen designed to decrease inflammation using anakinra (interleukin-1 receptor antagonist for 28 days), improve gut permeability (oral zinc for 6 months) and pentoxifylline (oral for 6 months) to protect against acute kidney injury (AKI, hepatorenal syndrome).

Following enrollment, the Investigational Pharmacist designated the study arm assigned by the randomization schedule. Participants were randomized to either Group A or Group B. Prior to dispensing, appropriate study personnel would call the Interactive Voice Response System (IVRS) to obtain their assignment. The trial was double-blinded by re-packaging all oral medications into capsules to ensure that each group received an equal number of capsules each day for the entire duration of the study. Patients were treated with pentoxifylline 400 mg three times daily for 28 days PLUS zinc supplements for a total of 6 months (Group A) or methylprednisolone 32 mg daily for 28 days PLUS placebo capsules daily for the remaining 5 months (Group B). An injection of either anakinra (100 mg) (Group A) or a placebo (Group B) was administered subcutaneously daily for 14 days. Patients, investigators and physicians involved in the care of the patients were masked as to the treatment.

2.4. Statistical Considerations

2.4.1. Sample Size and Randomization

The primary outcome was 6-month mortality. Since this outcome has both censored (patients lost to follow-up) and uncensored (patients who died) observations, Kaplan-Meier Product-Limit estimates of six-month mortality with a log-rank statistic were used to compare the two survival curves (intervention vs. control), combining treatment groups across strata (MELD 20–25, 26–31, ≥32). Using the Lakatos approach to estimating the sample size for a log-rank test [53], we found that a two-sided log rank test with an overall sample size of 130 patients (65 patients each of the control and treatment groups) achieved 80.0% power at a 0.05 significance level to detect a hazard ratio of 0.4447 when the proportion surviving at 6 months in the control group was 0.64 and 0.82 in the treatment group. The total study duration was planned to be 2.5 years, of which subject accrual (entry) was planned for the first 2 years. We assumed that the accrual pattern would be uniform throughout the accrual period and that the proportion ofdrop-out in each group will be 0.10 during the 6 months with no treatment cross-overs. As a sensitivity analysis of these estimates, we used the Lachin-Foulkes approach [54,55], a slightly less flexible approach, and with the same assumptions as above, we estimated 53 patients in each group, reasonably close to the Lakatos estimates. Thus, to be conservative in our estimates, we set our sample size as 130 patients total or 65 patients in each treatment group. PASS II (NCSS, Kaysville, Utah) was used for all sample size calculations. The sample size was calculated based on the analysis of patients with alcohol-associated liver disease in the (SRTR) database, accounting for drop-out, lost-to-follow-up and non- evaluable results and including a factor for possible non-representativeness of the original database. In addition, preliminary data for 114 patients meeting criteria were used in the calculations and stratification decisions.

2.4.2. Subject Population(s) for Analysis

For the primary efficacy analysis, we used intention to treat principles, so that all patients randomized, regardless of adherence to study medication, werewere used in the analysis in the treatment group to which they were assigned. As secondary analyses, we performed a per-protocol analysis using patients who achieved at least 80% compliance with study medication. Significance was assessed using a two-sided alpha level of 0.05.

2.4.3. Statistical Methods

Our primary analysis of this outcome was the unadjusted log-rank test of the Kaplan-Meier survival estimates for the two treatment groups at a two-sided alpha level of 0.05 with survival censored at 6 months. Secondary analysis used a Cox proportional hazards regression model (with survival censored at 6 months) to adjust the treatment difference in survival for patient characteristics (e.g., age, gender, time since diagnosis) and other covariates (e.g., ALT, AST, other lab measures). Predictors were added to the model using reference cell coding. The Cox model generated estimates of the hazard ratio for each factor compared to the reference level; e.g., for male patients vs. female patients, assuming gender = 1 for male and 0 for females, a hazard ratio of 1.25 indicates that males have a 25% higher probability of death overall by 6 months compared to females. We added time-dependent covariates to the model, such as ALT levels at 2 and 4 months, to estimate their predictive value for mortality.

To assess the fit of the model, we used Nagelkerke's R^2, constructed from the ratio of the likelihoods for the fitted model to the intercept only model. As further secondary analysis, we eliminated the survival censoring at 6 months and tested the overall survival throughout the duration of follow-up.

Secondary outcomes included changes in MELD score at 30, 90 and 180 days and changes in gut mucosal permeability measured by serum endotoxin levels. Serum endotoxin levels and serum cytokine profiles (IL-1, TNFα, etc.) were measured in both groups at baseline, 2, 7, 30, 90 and 180 days. Changes in these parameters were analyzed using mixed effects models with the outcome of interest (say, MELD score) as the outcome and predictors including treatment group, time metric, and other factors of interest.

For experiments with fold change as the outcome, our initial analyses were descriptive in nature, allowing for investigation of appropriate normalization approaches. Initial comparative analyses of fold-change between treatment groups (or between other groups of interest) werewe through a standard two-group t-test at the p = 0.05 level with appropriate adjustment for unequal variances. If the fold change data are longitudinal in nature, we used mixed (random effects) models to estimate the change in fold- change over time. In these models, fold-change at each time point was the outcome with treatment (or other group indicator) and time as predictors. Additional patient characteristics can be added to the model to create adjusted estimates of treatment (or other group) differences. Mixed models can adjust for the correlation due to measures within patients and due to repeated microarrays at each time point.

2.4.4. Safety monitoring of research subject

The same patient population was used for safety analysis. Common terminology criteria for adverse events (CTCAE) were used to document all adverse events (AE) and serious adverse events (SAE) (Supplementary Table 4). A data safety monitoring board (DSMB) was generated including investigators who were independent of the clinical study/no conflicts with the investigators and with expertise in clinical trial design and monitoring and alcohol related tissue injury. Members of the DSMB met formally every 6 months and in addition to evaluating the study progress also performed an interim analysis to determine if safety or futility reached statistical significance. There were no formal statistical stopping boundaries proposed even though the DSMB had the option to request such boundaries to be determined at any time during the study for both safety and efficacy (including futility). However, an apparent, consistent and persistent evidence of net harm that tends to overwhelm any benefit would allow for premature termination of the study. Continuing enrollment into each cohort was determined on an ongoing basis and at periodic intervals.

The DSMB received a report on all reported adverse events and serious adverse events as well as the proportion of subjects with any specific event of interest. The finding of any unexpected serious adverse event (SAE) considered to be related to study drug in 2 of 6 patients with the unexpected SAE would lead to review and suspension of recruitment and review of the complete data by the external safety monitors.

In addition to the planned evaluations, DSMB received quarterly safety summary reports of enrollment, baseline demographics, withdrawals from treatment and all AEs and SAEs with information on relation to study treatment. Based on these reviews, the DSMB, did not observe any significant adverse events or serious adverse events thought to be due to study medication. Futility of continuing the study or overwhelming benefit in either treatment arm was also not observed. A major challenge in patients with severe alcohol-associated hepatitis is the occurrence of severe complications (sepsis, gastrointestinal bleeding, renal failure, encephalopathy) as part of the natural course of the disease. Hence, adjudicating the effect of disease or drug was based on review of reports from each site. In this study with a novel combination of medications, the Lille criteria were not used to stop the study because these criteria were developed as stopping criteria due to futility of treatment with steroids [56]. There are no data if such criteria will predict futility with Anakinra and we did not unblind the study to determine if the Lille criteria were predicting futility. These will be reported with the results of the clinical trial separately.

In addition to the clinical trial design as above, we also determined the natural history of severe AH (MELD > 20). The primary rationale for this aim was that most observational studies in severe AH were conducted over 4 decades ago and the impact of improved public health, nutrition and advances in intensive care on outcomes is not known. Patients with severe AH who declined to be randomized but consented to prospective data collection and minimal risk non-invasive studies (blood drawing and urine collection). This part of the study was the optional Natural History (minimal risk with urine and blood collection) component. All screened subjects meeting inclusion criteria and either excluded or who declined consent to randomization in the clinical trial were invited to join this study of the natural history of severe AH. As mentioned, the natural history of patients with severe AH is not well documented in the current era. Prospective collection of data in parallel with the randomized study subjects has great potential value. Every patient who agreed and gave consent was seen at follow-up as clinically indicated. At each visit that coincides with collection of samples for research, the subjects were offered the opportunity to participate.

Finally, subjects were asked for a DNA blood draw during the first day of the study to generate a biorepository. A biorepository would serve as a national resource for all studies related to acute alcohol-associated hepatitis. A major goal of the consortium is to share clinical samples. Our research group will use the samples and share with others for future studies on biomarkers for AH severity/outcome, cost analysis and rationales for new therapies.

3. Results

During the study period, a total of 500 subjects with severe AH were screened and 104 subjects were randomized using the strategy. Patient characteristics and severity of disease assessed by prognostic functions (i.e., Lille score, Maddrey Discriminant Function, and MELD score) were reported after analysis. Overall mean duration of follow up was 100 days. During the study, despite a robust recruitment and retention strategy (Table 3), overall recruitment was less than anticipated for a variety of reasons (Table 4) and the study was extended from the planned 2.5 years to a total of 5.5 years of recruitment, despite which the planned recruitment goals were not met. However, the number of patients enrolled was sufficient to obtain a signal that one of the arms was more effective than the other.

Table 3.

Patient retention strategies.

Contact potential participants as soon as possible. Obtain as many contact points (e.g., home phone, cell phone, work phone, address, etc.) on individual enrollee During enrollment, obtain contact information on one or more family members/close friends of the participant in order to be able to maintain contact with the enrollee Plans and suggestions for participant retention should be described and may include strategies such as: Monthly phone calls from coordinators/physicians Birthday and holiday cards Postcards and verbal reminders E-mail reminders, if applicable Stress the importance of compliance during the informed consent interview and throughout the study. Establish rapport with the participants. Remember that retention begins with the subject's first visit and is an ongoing process. Retention is everyone's responsibility Treat participants and their caregivers with respect. Assure a welcoming atmosphere where participants are seen. Be considerate of the participant's time. Identify and resolve issues in a timely manner. If contact is lost with an enrollee, determine best manner to contact him/her Utilize long-term therapy that will likely enhance sustained interest in continued participation. Monitoring Progress Identify barriers to recruitment. Track the number of participants enrolled against expected per site. Monitor recruitment and intervene quickly to change recruitment techniques that are proving unsuccessful.

Table 4.

Unanticipated obstacles toward achieving recruitment goals and potential solutions.

Challenges Strategies adopted to overcome challenge

Delay in developing protocolsDevelop standardized definitions, operating procedures

Have in person meetings rather than phone/webinars

Exclusion criteria were too stringentAdapting the inclusion/exclusion criteria, modifying limits of ALT/AST

Low numbers of patients at each siteEducate primary and admitting teams with seminars and individual discussions

Communicate and regularly conact community hospitals and primary care practice physicians

Weekend admissionsPI or designee to review admissions and review with research coordinator

An adaptive strategy was used for the diagnostic criteria based on the experience of the initial 2 years of the study with modification of the limits of AST and ALT to define AH. In some previous studies, AH required liver biopsy for diagnosis, but in patients with severe AH, thrombocytopenia/coagulopathy are contraindications to percutaneous liver biopsy and it is not the standard of care in the vast majority of centers in the United States. To preserve applicability to the population of patients with severe AH, we used laboratory and clinical criteria for diagnosis. Importantly, underlying cirrhosis and advanced fibrosis were not an exclusion.

Finally, of the patients who were recruited, 74% completed the study or were accounted for at 180 days while 26% were lost during follow up despite maximum efforts to have to return for follow up.

3.1. Ancillary studies

A number of ancillary studies are ongoing based on the clinical data and biorepository samples. Other ancillary studies are strongly encouraged using the database and biosamples as long as they are first considered by the steering committee, consisting of the 4 site principal investigators.

4. Discussion

This trial is the first multicenter treatment study in the United States since the Veteran Administration trials that evaluated the role of corticosteroids and nutrition in AH [57]. The protocols were developed by the site principal investigators in consultation with the NIAAA program staff. The efforts and teams at each site were supported by grant support from the NIAAA awarded by a competitive peer review process. The DASH consortium goals were developed based on the resources available, recognizing the challenges in recruiting a well characterized population and development of robust, uniform diagnostic criteria. Since this was a treatment trial aimed toward improving outcome in severe AH, a disease state for which there are no approved treatments, an investigational new drug application was obtained from the FDA.

We evaluated previous randomized controlled trials in AH using the infliximab trial in AH as a model, but other trials including the VA cooperative study was also reviewed [58,59]. Since previous studies on AH had limited evaluation of outcomes beyond 30 days and it is known that 3 and 6 month mortality remain high [3,4,15], the primary outcome of the DASH study was 6 months mortality in response to treatment. The STOPAH study which had similar inclusion and exclusion criteria was published after recruitment had begun for the DASH study [17]. In comparison with the STOPAH study, a significant proportion of patients with AH in the DASH study had liver biopsies and in those patients in whom a biopsy was not done, the diagnostic criteria that were initially proposed were modified based on the data from the enrolled subjects.

The major innovations of the proposed trial included: [1] two parallel studies to encompass patients with severe AH in a novel treatment evaluation or an observational arm; [2] three drug randomization with placebos for each drug to ensure double blinding; [3] compliance statistics for each drug; and [4] a survival outcome as well as a continuous longitudinal outcome (MELD). These approaches will allow for evaluation of a combination intervention to treat a complex, multifactorial disease with a high mortality and lay the foundation for future trials in similar high mortality acute disorders.

In addition to the treatment trial, the DASH study highlighted the major limitations to enrollment in trials in patients with severe AH that will create a foundation for future therapeutic studies in this population of patients. These are similar to those reported by others [60]. Even though the enrollment goals were not met, we anticipate that the DASH study with its well characterized, long term treatment and outcomes in severe AH provides a basis for novel combination therapies not only in severe AH but also for other acute severe illnesses.