This study was designed as a multicenter, randomized, single-blind, efficacy trial of patients recently discharged from an episode of AHF (< 180 days), with New York Heart Association functional class II-IV and CA-125 > 35 U/mL. A computer-generated randomization scheme was used to allocate participants (in a 1:1 ratio) to receive therapy guided by CA-125 (aiming to keep CA-125 ≤ 35 U/mL) or standard treatment. Patients (but not investigators) were masked to the group assignment. The study is being conducted in 5 centers in Spain. Independently of staggered entry, the minimum duration of a patient's participation is 12 months (from first to last visit). The study is being conducted in accordance with Good Clinical Practice, Declaration of Helsinki 2002. All patients provided signed informed consent and the protocol was approved by the research ethics committee of participating centers and Agencia Española de Medicamentos y Productos Sanitarios.

Candidate patients were selected from a consecutive cohort of patients with a recent admission for AHF. Inclusion and exclusion criteria are summarized in Table 1.

All end points are verified by an independent clinical monitor and subsequently submitted to an adjudication committee blinded to treatment assignments.

Based on the treatment strategies, the patient safety assessment is specifically focused on monitoring renal function, liver function, myopathy, and electrolyte disturbances.

Sample size determination for this study assumes 2-sided testing at the 0.05 significance alpha level. The effect size for the primary end point was based on prior results from the AHF registry at the Hospital Clínico Universitario (Valencia, Spain). In this pilot study, 1702 patients consecutively admitted to the hospital with a diagnosis of AHF were followed up for clinical end points. During the index hospitalization, 64.6% had CA-125 > 35 U/mL. For this registry, we have assumed that both CA-125 groups have received standard medical therapy, and thus the differences found in outcomes will not include the effect of an optimized therapy. A 1-year post-discharge rate for the composite end point of all-cause mortality or readmission for AHF was estimated as 42.8% for the group with CA-125 > 35 U/mL, and 32.6% for the group with CA-125 ≤ 35 U/mL, yielding an absolute rate difference of 10.4%. Therefore, we expect that optimizing treatment in the group with CA-125 > 35 U/mL will translate into 35% relative risk reduction of this combined end point. Thus, using 80% power to detect at least such difference at 1 year and a 1:1 allocation ratio, 180 patients were estimated for each group (n=360 patients). The total sample was increased to 378 patients (n=189 each group) in order to account for losses to follow-up, which was assumed to be around 5%.

All statistical comparisons will be made under the intention-to-treat principle. Continuous variables will be expressed as a mean (1 standard deviation) or median [interquartile range] as appropriate. Comparison of the both treatment strategies will be carried out with unpaired Student t test or Mann-Whitney U-test according to variable distribution. Discrete variables will be compared with chi-square test or Fisher exact test as appropriate.

We have envisioned 2 methodological approaches for the analysis of the primary end point:

• •

  Analyzed as time to first event: In the first approach, which relies on the traditional way of analyzing results from clinical trials, the difference between the treatment groups will be graphically depicted by Kaplan-Meier method and tested by log rank test. The main censoring criteria will be the end of the trial (administrative censoring at 1 year) and time to first event. Univariate hazard ratios will be estimated with Cox proportional hazard regression; however, multivariable Cox regression analysis will be used only in the event that important prognostic factors or patient characteristics at baseline show significant imbalance between the two randomized groups. All analyses will account for a potential clustering effect within centers.

• •

  Accounting for recurrent HF hospitalizations: To account for the patient's entire outcome history, we perform negative binomial regression analysis, where the count of events (including death, if it occurred) will be compared between the two treatment groups.

For the secondary end points, composite of death and any rehospitalization, a similar approach will be used. For the mortality endpoint, both groups will be compared using the joint modeling approach, which simultaneously adjusts for the longitudinal time-varying effect of CA-125 while accounting for death as a terminal event. For the end point of rehospitalization, the survival curve will be constructed as a cumulative incidence function, in order to account for mortality as a competing event. For these end points, however, we will also explore the use of joint frailty modeling to compare both groups in regard to recurrent hospitalizations along the entire follow-up, while adjusting for death as a terminal event. For the end point of days alive and out of the hospital, the length of the follow-up that will be used in the survival analysis will not include the length of time the patient has spent in the hospital on any admission.

Subgroup analyses will be performed among patients defined by age group (> 65 years, > 70 years, > 75 years), sex, statins use, LVEF (> 50%, > 40%), etiology, inflammatory status (relative lymphocyte count and high-sensitivity C-reactive protein), renal failure, CA-125 at baseline (above median), and natriuretic peptides at baseline (above median).

All analyses were planned to be intention-to-treat. A 2-sided P-value of < .05 is considered to be statistically significant for all analyses. All analyses will be performed with Stata 13.1 or the R package.

The protocol was approved by the ethics committee/institutional review board for each participating center. Patients started enrolling in December 2011. As of July 17, 2013, the study was fully enrolled. As of December 2013, >50% of patients had finished the follow-up.

In any national health system, AHF hospitalizations represent a health care burden1–3,19, being associated with prohibitively high morbidity and mortality rates soon after discharge.1–3 Paradoxically, most studies evaluating novel therapeutic strategies have been focused on stable HF patients and not on the AHF setting.17,18 It is our thinking that new research avenues need to be guided toward this vulnerable period in order to reduce the burden of this disease. In this regard, alternative strategies that seek to optimize decongestive therapies and control the inflammatory milieu seem to be pertinent goals, as both mechanisms play a critical role during this particular period.5,20 For instance, the severity of FO is related to poor prognosis5,21–23 and, far from being an epiphenomenon, it may promote the progression of the disease through complex neurohumoral (favoring sodium retention), renal (reducing glomerular filtration), cardiac, and endothelial interactions.24,25 Moreover, FO assessment and its treatment is frequently challenging and largely subjective, mainly because the clinical tools available, which include symptoms, signs, X-ray, and even natriuretic peptides, have shown limited accuracy in identifying or quantifying the degree of FO. This is particularly true for certain subgroups of patients carrying confounding conditions such as obesity, advanced age, and lung, liver, or peripheral vein diseases.5,8,9 Likewise, low-grade inflammation is common in patients with AHF, a fact that opens avenues for new treatment modalities such as statins and omega-3 polyunsaturated fatty acids,26,27 even though these treatments have failed to demonstrate substantial clinical benefits in previous studies.28,29

Based on the above background information, we designed this study in an effort to determine whether CA-125 values allow us to optimize treatment after an AHF episode.

Biomarker guided therapy has recently emerged as an attractive strategy for personalizing the therapeutic options already available in HF management. Several studies have evaluated the clinical utility of natriuretic peptides-guided therapy in HF but, despite the slight improvement shown from pooled data,30,31 other relatively large studies have failed to document a significant clinical improvement in terms of mortality and morbidity.30–32 In fact, current US guidelines33 do not systematically recommend it for improving outcomes (class IIb); other clinical practice guidelines (including those from the European Society of Cardiology) are awaiting the results from new clinical trials. This lack of consensus is due, at least in part, to several uncertainties: a) determining which plasma level should be assumed as the optimal target level; this threshold is not well established because most studies have used different criteria; b) common confounders such as age, weight, renal function, and other comorbidities that have shown to influence the predictive ability of natriuretic peptides; c) very few large studies evaluating the clinical value of natriuretic-guided therapy in patients recently discharged for AHF, and d) logistic disadvantages such as price and high variability may also play a role.30–32 In our opinion, larger and better conducted trials addressing these unresolved issues should be undertaken in the future.

CA-125, also called MUC16, is a glycoprotein synthesized by epithelial serous cells, with extremely complex structure and high molecular weight. Although CA-125 is widely used to monitor ovarian cancer therapy, high plasma levels have also been reported in other malignant and nonmalignant diseases (HF, nephrotic syndrome, liver cirrhosis, tuberculosis, or pelvic inflammatory disease, among others).10 In the HF stetting, plasma CA-125 correlates with clinical (New York Heart Association functional class), hemodynamic (pulmonary artery wedge pressure, right atrial pressure), and echocardiographic (E wave deceleration time, right ventricular systolic dysfunction) parameters related to the severity of the disease.10,11 High levels are present in the majority of patients admitted for AHF; for instance, in a recent cohort of 1111 unselected patients admitted for AHF, 65% of patients displayed CA-125 > 35 U/mL.12 In this setting, CA-125 has been related to symptoms and signs of FO,10,12 inflammation status,34 and 6-month mortality and readmission,12 independently of natriuretic peptides and other traditional risk factors. In addition, recent studies10,12–16 have highlighted the potential of CA-125 to monitor and guide HF therapy in HF by exploiting certain characteristics, such as close correlation with disease severity and clinical outcomes, and as a marker that has the potential to differentially characterize the prognostic effect attributed to the use of loop diuretics doses and of statins. In fact, our group found that at the first outpatient visit after an admission for AHF (mean 28 days after discharge), those patients who normalized CA-125 levels exhibited the lowest adjusted risk of death, intermediate for those who had decreased but not normalized CA-125, and higher for those in whom CA-125 increased.13 These findings were also reproduced when predicting 6-month AHF readmission.14 More recently, Husser et al35 reported, in 228 patients with aortic stenosis who had undergone transcatheter aortic valve implantation, that the longitudinal evolution of CA-125 levels predicted adverse clinical outcomes after the procedure. Moreover, we reported in a cohort of patients admitted for AHF that discharge with high-dose loop diuretics was associated with higher risk of mortality, with the exception of a subgroup of patients characterized with elevated blood urea nitrogen and CA-125, in which the use of high-dose loop diuretics was associated with lower adjusted mortality rates.15 Furthermore, the use of statins has correlated with lower mortality rates in patients with AHF, but only when CA-125 was elevated.16

Other characteristics such as wide availability, low cost (around 2€ per determination), levels not substantially modified by age and renal dysfunction, and prolonged half-life (>1 week) make this biomarker a promising clinical tool in HF.10