European Review for Medical and Pharmacological Sciences 2021; 25: 2971-2980 Comparison of diuretic strategies in diuretic-resistant acute heart failure: a systematic review and network meta-analysis D. ORSO1,2, G. TAVAZZI3,4, F. CORRADI5,6, F. MEARELLI7, N. FEDERICI1,2, D. PERIC8, N. D’ANDREA1,2, G. SANTORI9, F. MOJOLI3,4, F. FORFORI5, L. VETRUGNO1,2, T. BOVE1,2 1Department of Medicine, University of Udine, Udine, Italy 2Department of Anesthesia and Intensive Care Medicine, ASUFC “Santa Maria della Misericordia” University Hospital of Udine, Udine, Italy 3Anesthesiology and Intensive Care Unit, Fondazione IRCCS Policlinico San Matteo, Pavia, Italy 4Department of Medical, Surgical, Diagnostic and Pediatric Science, University of Pavia, Pavia, Italy 5Department of Surgical, Medical and Molecular Pathology and Critical Care Medicine, University of Pisa, Pisa, Italy 6Anaesthesia and Intensive Care Unit, Ente Ospedaliero Ospedali Galliera, Genoa, Italy 7Department of Internal Medicine, ASUGI “Cattinara” University Hospital of Trieste, Trieste, Italy 8Department of Emergency Medicine, ASUGI “Cattinara” University Hospital of Trieste, Trieste, Italy 9Department of Surgical Sciences and Integrated Diagnostics (DISC), University of Genoa, Genoa, Italy Abstract. – OBJECTIVE: Up to 50% of patients randomized trials; SMD: Standardized Mean Difference; hospitalized for acute heart failure (AHF) show 95% CI: 95% Confidence Interval. resistance to diuretics. This condition contrib- utes to a prolonged hospital length of stay and a higher risk of death. This review aimed to inves- tigate whether a diuretic therapeutic approach Introduction more effective than furosemide alone exists for patients with diuretic-resistant AHF. Acute heart failure accounts for >26 million MATERIALS AND METHODS: We identified hospitalizations per year worldwide1-3, and a 1-year all randomized controlled trials (RCTs) evaluat- mortality rate as high as 20-30%4, with an additive ing diuretic therapy in patients with diuretic-re- 5 sistant AHF. We searched Pubmed, BioMed Cen- risk with each subsequent hospitalization . tral, and Cochrane CENTRAL databases. Congestion is the most frequent clinical mani- RESULTS: Six RCTs were identified, involving festation (70% of patients), with diuretics recom- a total of 845 patients. The P-score ranges from mended as first-line therapy6,7. Amongst those, 0.6663 for furosemide to 0.2294 for the tolvap- loop diuretics are the most prescribed drugs, al- tan-furosemide. We found no significant differ- though, despite the improvement in symptoms, at ences in efficacy for any drug comparison. least 50% of patients fail to experience any weight CONCLUSIONS: None of the diuretics con- sidered in RCTs performed to date (tolvaptan, loss due to reduced extracellular fluid retention, metolazone, hydrochlorothiazide, indapamide) and up to 50% leave the hospital with residual appear to be more effective than furosemide congestion, leading to further readmissions and therapy alone for the treatment of patients with mortality8,9. diuretic-resistant AHF. Diuretic resistance has not an accepted Key Words: unique definition, but it is commonly described Acute heart failure, Furosemide, Diuretic resistance, as the failure to achieve effective decongestion Tolvaptan, Efficacy, Network meta-analysis. despite an adequate dose of diuretic adminis- tered10. Since diuretic resistance contributes Abbreviations to worsening heart failure and outcome, great AHF: Acute Heart Failure; RCT: Randomized Controlled effort is directed towards identifying the best Trial; RoB 2: Revised Cochrane risk-of-bias tool for therapeutic strategies11. Corresponding Author: Luigi Vetrugno, MD; e-mail: [email protected] 2971 D. Orso, G. Tavazzi, F. Corradi, F. Mearelli, N. Federici, D. Peric, N. D’Andrea, G. Santori, et al The aim of this systematic literature review was to apy, usually a loop diuretic, to resolve a condition investigate whether a diuretic therapeutic approach of acute congestive heart failure). We included exists that is more effective in treating patients with studies performed in any hospital department. diuretic-resistant AHF than furosemide alone. The following publication types were exclud- ed: observational studies, conference abstracts, reviews, non-human studies, protocols, policy Materials and Methods statements, and guidelines. The electronic search strategy applied stan- Data Sources and Searches dard filters for identifying related studies. The We identified all randomized controlled trials search was performed on the Medline (PubMed), (RCTs) evaluating patients’ medical therapy with BioMed Central, Scopus, Web of Science, and diuretic-resistant AHF (i.e., the clinician-estab- Cochrane CENTRAL databases, from inception lished need to add a second diuretic to initial ther- to 31st January 2020, with language restrictions. Figure 1. PRISMA 2009 Flow-diagram for the search and selection process of articles. From: Moher D, Liberati A, Tetzlaff J, Altman DG, The PRISMA Group (2009). Preferred Reporting Items for Systematic Reviews and Meta-Analyses: The PRIS- MA Statement. PLoS Med 6(7): e1000097. doi: 10.1371/journal.pmed1000097. For more information, visit www.prisma-state- ment.org. 2972 Diuretic-resistant heart failure: a meta-analysis The meta-analysis was conducted according to Statistical Analysis the PRISMA guidelines12. Our search strategy’s A default fixed-effect model or random-ef- specific details can be seen in the Online Data fects model was used if a high degree of incon- Supplement appended to this paper (Supplemen- sistency/heterogeneity was detected through tary Material S1). The protocol for this system- Cochran’s Q value. The I2 was used to measure atic review was registered on PROSPERO (regis- the percentage of variation across studies due tration number CRD42020168905). to heterogeneity rather than chance. When I2 results between 50 and 90%, the heterogeneity Data Extraction and Quality Assessment is assumed as substantial, while I2 is 75-100%, Two independent couples of reviewers (DO/ the heterogeneity is considerable. Results were NF and FM/DP) identified all relevant titles and illustrated using forest plots. A funnel plot and abstracts. Full-text copies of all potentially rel- Egger’s test for asymmetry were planned to evant studies were then obtained for detailed evaluate potential publication bias. The stan- evaluation by each pair of reviewers, and the dardized mean difference (SMD) and 95% con- data from each study were independently ex- fidence interval (95% CI) were calculated ac- tracted using a standardized abstraction mod- cording to the intention-to-treat principle. Tau2 ule. One pair of reviewers (FM/DP) evaluated defined the between-study variance. The differ- the data in the absence of author and journal ence in the estimates of the treatment effect be- names, institutional affiliations, and publica- tween the treatment groups for each hypothesis tion date. The data extracted from the docu- was tested using a two-sided z test. A p-val- ments were checked by an additional reviewer ue <0.05 was considered statistically signifi- (LV) for accuracy. Each study was assessed for cant. All statistical analyses were performed methodological quality using the revised Co- using the R environment (version. 3.6.1. The chrane risk-of-bias tool for randomized trials R Foundation for Statistical Computing; Vien- (RoB 2). We examined each article for informa- na, Austria) with the ‘netmeta’ and ‘metacont’ tion on the sampling method, the presence of a packages14,15. control group, the comparability of the control group (if included), and methods used to obtain the results. Results Qualitative Analysis A narrative synthesis approach was used to Characteristics of the Included Studies explore each study’s characteristics and the vari- Diuretic resistance was considered as the per- ations between the studies. We collected demo- sistence of congestion (orthopnea, edema, pulmo- graphic data, information about the study’s design nary rales, elevated jugular venous pulse, or con- and planning objectives, and information about gestion on chest radiograph) despite the diuretic the diuretic drugs and control groups compared. administration. Cox et al16 defined “diuretic-resis- As an indication of therapeutic efficacy, the urine tance” as the production of less than 2 L of urine volume produced in the first 24 hours was consid- in 12 hours, despite a furosemide dosage greater ered. We did not consider a cut-off value of urine than 240 mg/die. volume, but instead considered the comparison in Six studies fulfilled the specified criteria in- terms of absolute quantities of urine produced in volving 845 patients16-21 (Figure 1). Four studies the post-treatment period. We adopted this out- were double-blind RCTs, and 2 were open-label come because the subjective assessment of dys- RCTs. All but 2 studies were multicenter studies. pnea improvement is prone to methodological Five out of the 6 studies investigated tolvap- bias, whereas measurement of the volume of urine tan; 2 studied metolazone, 1 hydrochlorothiazide, produced is a pragmatic outcome. Unlike the uri- and 1 indapamide. In 3 studies, one or more arms nary sodium concentration, many of the studies evaluated an incremental dose of furosemide. involved reported the total amount of urine pro- Only 2 of the 6 randomized trials considered a duced8. Furthermore, all biomarkers (creatinine, placebo-arm. All but 2 of the studies considered B-type natriuretic peptide,