https://orcid.org/0000-0002-6036-2113
Abstract
Acetazolamide inhibits proximal tubular sodium and bicarbonate re-absorption and improved decongestive response in acute heart failure in the ADVOR trial. It is unknown whether bicarbonate levels alter the decongestive response to acetazolamide.
This is a sub-analysis of the randomized, double-blind, placebo-controlled ADVOR trial that randomized 519 patients with acute heart failure and volume overload in a 1:1 ratio to intravenous acetazolamide (500 mg/day) or matching placebo on top of standardized intravenous loop diuretics (dose equivalent of twice oral maintenance dose). The primary endpoint was complete decongestion after 3 days of treatment (morning of day 4). Impact of baseline HCO3 levels on the treatment effect of acetazolamide was assessed. : Of the 519 enrolled patients, 516 (99.4%) had a baseline HCO3 measurement. Continuous HCO3 modelling illustrated a higher proportional treatment effect for acetazolamide if baseline HCO3 ≥ 27 mmol/l. A total of 234 (45%) had a baseline HCO3 ≥ 27 mmol/l. Randomization towards acetazolamide improved decongestive response over the entire range of baseline HCO3− levels (P = 0.004); however, patients with elevated baseline HCO3 exhibited a significant higher response to acetazolamide [primary endpoint: no vs. elevated HCO3; OR 1.37 (0.79–2.37) vs. OR 2.39 (1.35–4.22), P-interaction = 0.065), with higher proportional diuretic and natriuretic response (both P-interaction < 0.001), greater reduction in congestion score on consecutive days (treatment × time by HCO3-interaction <0.001) and length of stay (P-interaction = 0.019). The larger proportional treatment effect was mainly explained by the development of diminished decongestive response in the placebo arm (loop diuretics only), both with regard to reaching the primary endpoint of decongestion as well as reduction in congestion score. Development of elevated HCO3 further worsened decongestive response in the placebo arm (P-interaction = 0.041). A loop diuretic only strategy was associated with an increase in the HCO3 during the treatment phase which was prevented by acetazolamide (day 3: placebo 74.8% vs. acetazolamide 41.3%, P < 0.001).
Acetazolamide improves decongestive response over the entire range of HCO3− levels; however, the treatment response is magnified in patients with baseline or loop diuretic-induced elevated HCO3 (marker of proximal nephron NaHCO3 retention) by specifically counteracting this component of diuretic resistance.
Sub-analysis of the ADVOR trial investigating the role of baseline bicarbonate values on the diuretic response to acetazolamide and the change in bicarbonate during the treatment phase. The diuretic effectiveness of acetazolamide is more pronounced with higher baseline bicarbonate. An increase in bicarbonate in patients in the loop diuretic arm is associated with development of diuretic resistance which can be prevented by acetazolamide. OR, odds ratio; AHF, acute heart failure; LOS, length of stay; prox, proximal.
See the editorial comment for this article ‘Serum bicarbonate and congestion: a potential biomarker for identifying and guiding management in diuretic resistance?’, by R.T. Campbell and K.F. Docherty, https://doi.org/10.1093/eurheartj/ehad209.
Introduction
The Acetazolamide in Decompensated heart failure with Volume OveRload (ADVOR) trial investigated the effect of adding acetazolamide on top of standardized intravenous loop diuretics in patients with acute heart failure (AHF) and clear signs of volume overload.1 Acetazolamide improved decongestive effectiveness measured by a higher likelihood of achieving decongestion. Acetazolamide also improved diuretic response, measured by increased diuresis and natriuresis, more pronounced reduction in congestion score overall resulting in a shorter length of stay of the index AHF admission.2
Carbonic anhydrase II catalyses the formation of HCO3− from H2O and CO2 in the proximal tubular cells, which generates H+ that is exchanged for tubular luminal Na+ in the proximal nephron by the NHE3 transporter. On the basolateral surface, Na+ transport occurs through the electrogenic Na+/HCO3− co-transporter (NBC) in the first segment of the proximal tubules that re-absorbs most sodium. NBC transports three HCO3− molecules for every Na+ molecule, which is the main responsible mechanism for acid–base buffering of the kidneys by generating net HCO3− absorption.3 Both the NHE3 transporter and NBC are stimulated by angiotensin II, which also explains why the proximal nephron in heart failure absorbs proportionally more of the filtered sodium.4,5 Therefore, elevated HCO3 in AHF can be seen as an indicator of neuro-hormonal activation via enhanced proximal sodium and HCO3 uptake.6–8 It is unclear if elevated levels of HCO3 are important for the treatment effect of acetazolamide or if the presence of elevated HCO3 levels results in a modulation of the treatment effect of acetazolamide. This is important as it is well recognized that treatment with loop diuretics further results in neuro-hormonal activation and development of an elevated HCO3 (due to proximal Na/HCO3 retention).9–11 This sub-analysis of the ADVOR trial aims to evaluate the effect of elevated baseline and treatment-induced HCO3− levels on treatment efficacy of acetazolamide in AHF.
Methods
Trial design and population
ADVOR was an investigator-initiated, multi-centre, randomized, parallel-group, double-blind, placebo-controlled trial. Detailed methods and results of the trial have been published previously.1,2,12 Briefly, patients aged above 18 years with an AHF admission and clinical signs of volume overload were eligible for participation. Patients were required to have an N-terminal pro-B-type natriuretic peptide (NT-proBNP) or B-type natriuretic peptide level >1000 pg/ml or >250 pg/ml, respectively, with at least one clinical sign of volume overload (i.e. ascites, pleural effusion, or oedema). In addition, oral maintenance therapy with at least 40 mg furosemide or an equivalent dose for at least 1 month was required for randomization.1 Main exclusion criteria were acetazolamide maintenance therapy or treatment with another proximal tubular diuretic agent including a sodium–glucose co-transporter 2 inhibitor (SGLT2i), a systolic blood pressure <90 mmHg, or an estimated glomerular filtration rate <20 ml/min/1.73 m². A full list of inclusion and exclusion criteria is provided in Supplementary data online, Table S1. All the participants provided written informed consent.
Trial intervention and endpoint collection
Patients were randomly assigned to receive an intravenous bolus of acetazolamide (500 mg once daily) or matching placebo (1:1) upon randomization (baseline) and during the next 2 days (treatment phase, see Supplementary data online, Figure S1). The primary endpoint was successful decongestion, defined as the absence of signs of volume overload, within 3 days of randomization without the need for escalation of diuretic therapy using a previously published congestion score (see Supplementary data online, Figure S2 for congestion score). A urine collection was started, after voiding empty, at the time of randomization and collected for two consecutive days. The congestion score was completed upon inclusion and thereafter daily before the morning dose of diuretics during the entire treatment phase. Next to the primary endpoint of complete decongestion, other predefined congestion/volume endpoints included (i) complete decongestion on the morning of day 3 (without taking the need of escalation therapy into account), (ii) cumulative diuretic response (l/40 mg furosemide equivalent), (iii) cumulative natriuretic response (mmol/40 mg furosemide equivalent) during the first 2 days after randomization, (iv) length of stay, (v) the combined endpoint of all-cause mortality and re-hospitalization for heart failure during 3 months of follow-up, and (vi) all-cause mortality separately.
Baseline bicarbonate levels
A baseline laboratory analysis at the time of randomization was performed with local lab assessment of venous blood bicarbonate levels. Based on their baseline HCO3− levels, patients were categorized as having an elevated HCO3− level (≥27 mmol/l) or not (<27 mmol/l). Additionally, treatment efficacy and safety of acetazolamide was assessed in patients with low (<23 mmol/l), normal (23–27 mmol/l), and elevated (>27 mmol/l), or HCO3 on a continuous scale.
Changes in bicarbonate levels
Bicarbonate levels were determined daily after randomization on day 1, day 2 and finally at day 3 (when the primary endpoint was collected). Change in bicarbonate levels were calculated from baseline to day 3 or earlier in case of complete decongestion within 72 h, with positive values indicating an increase in HCO3− levels. The proportion of patients developing an elevated HCO3− level (≥27 mmol/l) on every day was determined per treatment allocation.
Statistical analysis
Baseline characteristics are summarized as means and standard deviations, medians and 25th–75th percentile, or numbers and percentages and were evaluated using the independent-samples t-test, Mann–Whitney U test, chi-square test, ANOVA, and Kruskal–Wallis test, as appropriate. The primary endpoint (binary/binomial), as defined in the statistical analysis plan, was evaluated using a generalized linear mixed model (models equipped for response variables of different nature including binary endpoints or continuous endpoints). The generalized linear mixed model included a fixed treatment effect and random intercept to calculate odds ratios and 95% confidence interval (CI) (logit link function for binary endpoints). For interaction analysis with baseline HCO3−, HCO3− was entered into the model as a fixed effect interaction term with treatment allocation. Baseline HCO3 levels were used both binary (elevated or not elevated), categorical (low, normal, elevated) and on a continuous-scale enter in the model. As several cut-points exist to describe elevated HCO3− (e.g. based on normal values of venous, arterial, or mixed venous blood), baseline HCO3− was modelled against the odds ratio for the primary endpoint to identify the ideal cut-point associated with increased decongestive response. When using HCO3− as a continuous variable, restricted cubic splines based on three HCO3− knots were used to visualize the relation with the primary endpoint in a semi-parametric model. Continuous endpoints (urine output and natriuresis) were assessed using a similar generalized linear mixed model, but for continuous endpoints. Changes in congestion score on consecutive days were assessed using a linear mixed effect model for repeated measurements with a fixed treatment effect, its interaction by treatment day (time) and the interaction with baseline HCO3− and a random intercept. The combined endpoint of all-cause mortality and heart failure re-hospitalization after 3 months and all-cause mortality separately was assessed in a time-to-event analysis using a Cox-proportional hazard model including the treatment arm and interaction of HCO3− × treatment to calculate hazard ratios and 95% CI. Non-proportionality of hazards amongst treatment arms was assessed for the combined endpoint or all-cause mortality separately by adding an interaction of treatment groups with the logarithm of time to the Cox-proportional hazard model. Length of index hospitalization was compared with a linear mixed model after logarithmic transformation to calculate a geometric mean, geometric mean ratio and 95% CI. No multiplicity adjustments were done for all secondary analysis, so all reported values are exploratory. Hypotheses testing is two-sided and a significance level of α = 0.05 was used; however, for treatment interaction values <0.1 were assumed to indicate treatment effect modification. All statistical analyses were done using SPSS version 25 or RStudio 2022.02.3 + 492.
Results
Patient population
Of the 519 patients included in the ADVOR trial, 516 (99.4%) had a baseline HCO3− measurement. When plotting the odds ratio for successful decongestion with acetazolamide vs. placebo on a continuous scale according to baseline HCO3− levels, a value ≥27 mmol/l was associated with higher odds, with the entire 95% CI > 1 (Figure 1). At baseline, 234 patients (45%) had a HCO3− ≥ 27 mmol/l. Baseline characteristics of patients with vs. without elevated HCO3− are presented in Table 1. Patients with elevated HCO3− were more often women, treated with higher doses of loop diuretics, with more pronounced peripheral oedema, a higher serum sodium but lower serum chloride levels, and were less frequently treated with renin-angiotensin-aldosterone system blockers. Baseline characteristics of patients with low (n = 91, 18%), normal (n = 188, 36%), or high HCO3− (n = 234, 45%) are presented in Supplementary data online, Table S2, showing generally similar differences in baseline features.
Odds ratio for successful decongestion with acetazolamide vs. placebo according to baseline serum HCO3− level. Results of semi-parametric regression model reporting the OR for successful decongestion in patients assigned to acetazolamide vs. placebo modelled as a restricted cubic spline for baseline HCO3 using three knots. Solid line indicates mean OR and area in between dotted lines indicate the 95% CI. The vertical dotted line is where the lower bound of the 95% CI crosses the line unity equating to a HCO3 of 27 mmol/l.
Baseline characteristics per baseline bicarbonate levels
| Parameter | HCO3 < 27 mmol/l (n = 279) | HCO3 ≥ 27 mmol/l (n = 234) | P-value |
|---|---|---|---|
| Age (years) | 79 (72–84) | 80 (72–85) | 0.254 |
| Male sex | 190 (68%) | 131 (56%) | 0.006 |
| White race | 274 (99%) | 234 (100%) | 0.244 |
| Heart rate (b.p.m.) | 74 (62–90) | 77 (64–88) | 0.996 |
| SBP (mmHg) | 125 (113–141) | 120 (110–140) | 0.107 |
| DBP (mmHg) | 70 (62–80) | 70 (63–79) | 0.854 |
| Weight (kg) | 81 (72–95) | 82 (69–98) | 0.839 |
| Congestion score | 4 (3–6) | 4 (3–6) | 0.328 |
| Edema (>1+) | 254 (91%) | 218 (93%) | 0.560 |
| Pleural effusion | 150 (54%) | 118 (50%) | 0.514 |
| Ascites | 23 (9%) | 22 (9%) | 0.488 |
| Edema score | 0.026 | ||
| No or trace oedema | 24 (9%) | 17 (7%) | |
| Up to ankle | 42 (15%) | 29 (12%) | |
| Up to knee | 134 (48%) | 92 (39%) | |
| Above the knee | 78 (28%) | 97 (41%) | |
| Maintenance dose—furosemide equivalents (mg) | 50 (40–100) | 80 (40–150) | 0.001 |
| LVEF (%) | 43 (30–55) | 45 (34–55) | 0.347 |
| NT-proBNP (pg/ml) | 6629 (3187–11 087) | 5476 (2873–10 590) | 0.169 |
| NYHA | 0.141 | ||
| II | 43 (16%) | 23 (10%) | |
| III | 155 (56%) | 135 (57%) | |
| IV | 80 (29%) | 77 (33%) | |
| Ischaemic aetiology | 128 (46%) | 102 (43%) | 0.549 |
| Sodium (mmol/l) | 140 (136–142) | 141 (138–143) | 0.006 |
| HCO3 (mmol/l) | 24 (22–25) | 29 (28–31) | <0.001 |
| Chloride (mmol/l) | 102 (99–105) | 100 (97–102) | <0.001 |
| eGFR (ml/min/1.73 m2) | 37 (28–51) | 42 (32–54) | 0.051 |
| eGFR < 60 ml/min/1.73 m² | 230 (83%) | 189 (80%) | 0.501 |
| Comorbidities | |||
| History of AF | 194 (70%) | 178 (76%) | 0.132 |
| Diabetes | 139 (50%) | 104 (44%) | 0.194 |
| Hypertension | 206 (74%) | 177 (75%) | 0.752 |
| ACEi/ARB/ARNI | 162 (58%) | 104 (44%) | 0.002 |
| Beta-blocker | 222 (80%) | 195 (83%) | 0.366 |
| MRA | 129 (46%) | 85 (36%) | 0.019 |
| ICD | 42 (15%) | 37 (16%) | 0.842 |
| CRT | 33 (12%) | 27 (12%) | 0.894 |
| Parameter | HCO3 < 27 mmol/l (n = 279) | HCO3 ≥ 27 mmol/l (n = 234) | P-value |
|---|---|---|---|
| Age (years) | 79 (72–84) | 80 (72–85) | 0.254 |
| Male sex | 190 (68%) | 131 (56%) | 0.006 |
| White race | 274 (99%) | 234 (100%) | 0.244 |
| Heart rate (b.p.m.) | 74 (62–90) | 77 (64–88) | 0.996 |
| SBP (mmHg) | 125 (113–141) | 120 (110–140) | 0.107 |
| DBP (mmHg) | 70 (62–80) | 70 (63–79) | 0.854 |
| Weight (kg) | 81 (72–95) | 82 (69–98) | 0.839 |
| Congestion score | 4 (3–6) | 4 (3–6) | 0.328 |
| Edema (>1+) | 254 (91%) | 218 (93%) | 0.560 |
| Pleural effusion | 150 (54%) | 118 (50%) | 0.514 |
| Ascites | 23 (9%) | 22 (9%) | 0.488 |
| Edema score | 0.026 | ||
| No or trace oedema | 24 (9%) | 17 (7%) | |
| Up to ankle | 42 (15%) | 29 (12%) | |
| Up to knee | 134 (48%) | 92 (39%) | |
| Above the knee | 78 (28%) | 97 (41%) | |
| Maintenance dose—furosemide equivalents (mg) | 50 (40–100) | 80 (40–150) | 0.001 |
| LVEF (%) | 43 (30–55) | 45 (34–55) | 0.347 |
| NT-proBNP (pg/ml) | 6629 (3187–11 087) | 5476 (2873–10 590) | 0.169 |
| NYHA | 0.141 | ||
| II | 43 (16%) | 23 (10%) | |
| III | 155 (56%) | 135 (57%) | |
| IV | 80 (29%) | 77 (33%) | |
| Ischaemic aetiology | 128 (46%) | 102 (43%) | 0.549 |
| Sodium (mmol/l) | 140 (136–142) | 141 (138–143) | 0.006 |
| HCO3 (mmol/l) | 24 (22–25) | 29 (28–31) | <0.001 |
| Chloride (mmol/l) | 102 (99–105) | 100 (97–102) | <0.001 |
| eGFR (ml/min/1.73 m2) | 37 (28–51) | 42 (32–54) | 0.051 |
| eGFR < 60 ml/min/1.73 m² | 230 (83%) | 189 (80%) | 0.501 |
| Comorbidities | |||
| History of AF | 194 (70%) | 178 (76%) | 0.132 |
| Diabetes | 139 (50%) | 104 (44%) | 0.194 |
| Hypertension | 206 (74%) | 177 (75%) | 0.752 |
| ACEi/ARB/ARNI | 162 (58%) | 104 (44%) | 0.002 |
| Beta-blocker | 222 (80%) | 195 (83%) | 0.366 |
| MRA | 129 (46%) | 85 (36%) | 0.019 |
| ICD | 42 (15%) | 37 (16%) | 0.842 |
| CRT | 33 (12%) | 27 (12%) | 0.894 |
CRT, cardiac resynchronisation therapy; eGFR, estimated glomerular filtration rate; ICD, implantable cardioverter-defibrillator; LVEF, left ventricular ejection fraction; NT-proBNP, N-terminal pro-B-type natriuretic peptide; NYHA, New York Heart Association; ACEi, angiotensin-converting enzyme inhibitor; ARB, angiotensin II receptor blocker; ARNI, angiotensin receptor-neprilysin inhibitor; MRA, mineralocorticoid receptor antagonist.
Congestion score: see Supplementary data online.
Baseline characteristics per baseline bicarbonate levels
| Parameter | HCO3 < 27 mmol/l (n = 279) | HCO3 ≥ 27 mmol/l (n = 234) | P-value |
|---|---|---|---|
| Age (years) | 79 (72–84) | 80 (72–85) | 0.254 |
| Male sex | 190 (68%) | 131 (56%) | 0.006 |
| White race | 274 (99%) | 234 (100%) | 0.244 |
| Heart rate (b.p.m.) | 74 (62–90) | 77 (64–88) | 0.996 |
| SBP (mmHg) | 125 (113–141) | 120 (110–140) | 0.107 |
| DBP (mmHg) | 70 (62–80) | 70 (63–79) | 0.854 |
| Weight (kg) | 81 (72–95) | 82 (69–98) | 0.839 |
| Congestion score | 4 (3–6) | 4 (3–6) | 0.328 |
| Edema (>1+) | 254 (91%) | 218 (93%) | 0.560 |
| Pleural effusion | 150 (54%) | 118 (50%) | 0.514 |
| Ascites | 23 (9%) | 22 (9%) | 0.488 |
| Edema score | 0.026 | ||
| No or trace oedema | 24 (9%) | 17 (7%) | |
| Up to ankle | 42 (15%) | 29 (12%) | |
| Up to knee | 134 (48%) | 92 (39%) | |
| Above the knee | 78 (28%) | 97 (41%) | |
| Maintenance dose—furosemide equivalents (mg) | 50 (40–100) | 80 (40–150) | 0.001 |
| LVEF (%) | 43 (30–55) | 45 (34–55) | 0.347 |
| NT-proBNP (pg/ml) | 6629 (3187–11 087) | 5476 (2873–10 590) | 0.169 |
| NYHA | 0.141 | ||
| II | 43 (16%) | 23 (10%) | |
| III | 155 (56%) | 135 (57%) | |
| IV | 80 (29%) | 77 (33%) | |
| Ischaemic aetiology | 128 (46%) | 102 (43%) | 0.549 |
| Sodium (mmol/l) | 140 (136–142) | 141 (138–143) | 0.006 |
| HCO3 (mmol/l) | 24 (22–25) | 29 (28–31) | <0.001 |
| Chloride (mmol/l) | 102 (99–105) | 100 (97–102) | <0.001 |
| eGFR (ml/min/1.73 m2) | 37 (28–51) | 42 (32–54) | 0.051 |
| eGFR < 60 ml/min/1.73 m² | 230 (83%) | 189 (80%) | 0.501 |
| Comorbidities | |||
| History of AF | 194 (70%) | 178 (76%) | 0.132 |
| Diabetes | 139 (50%) | 104 (44%) | 0.194 |
| Hypertension | 206 (74%) | 177 (75%) | 0.752 |
| ACEi/ARB/ARNI | 162 (58%) | 104 (44%) | 0.002 |
| Beta-blocker | 222 (80%) | 195 (83%) | 0.366 |
| MRA | 129 (46%) | 85 (36%) | 0.019 |
| ICD | 42 (15%) | 37 (16%) | 0.842 |
| CRT | 33 (12%) | 27 (12%) | 0.894 |
| Parameter | HCO3 < 27 mmol/l (n = 279) | HCO3 ≥ 27 mmol/l (n = 234) | P-value |
|---|---|---|---|
| Age (years) | 79 (72–84) | 80 (72–85) | 0.254 |
| Male sex | 190 (68%) | 131 (56%) | 0.006 |
| White race | 274 (99%) | 234 (100%) | 0.244 |
| Heart rate (b.p.m.) | 74 (62–90) | 77 (64–88) | 0.996 |
| SBP (mmHg) | 125 (113–141) | 120 (110–140) | 0.107 |
| DBP (mmHg) | 70 (62–80) | 70 (63–79) | 0.854 |
| Weight (kg) | 81 (72–95) | 82 (69–98) | 0.839 |
| Congestion score | 4 (3–6) | 4 (3–6) | 0.328 |
| Edema (>1+) | 254 (91%) | 218 (93%) | 0.560 |
| Pleural effusion | 150 (54%) | 118 (50%) | 0.514 |
| Ascites | 23 (9%) | 22 (9%) | 0.488 |
| Edema score | 0.026 | ||
| No or trace oedema | 24 (9%) | 17 (7%) | |
| Up to ankle | 42 (15%) | 29 (12%) | |
| Up to knee | 134 (48%) | 92 (39%) | |
| Above the knee | 78 (28%) | 97 (41%) | |
| Maintenance dose—furosemide equivalents (mg) | 50 (40–100) | 80 (40–150) | 0.001 |
| LVEF (%) | 43 (30–55) | 45 (34–55) | 0.347 |
| NT-proBNP (pg/ml) | 6629 (3187–11 087) | 5476 (2873–10 590) | 0.169 |
| NYHA | 0.141 | ||
| II | 43 (16%) | 23 (10%) | |
| III | 155 (56%) | 135 (57%) | |
| IV | 80 (29%) | 77 (33%) | |
| Ischaemic aetiology | 128 (46%) | 102 (43%) | 0.549 |
| Sodium (mmol/l) | 140 (136–142) | 141 (138–143) | 0.006 |
| HCO3 (mmol/l) | 24 (22–25) | 29 (28–31) | <0.001 |
| Chloride (mmol/l) | 102 (99–105) | 100 (97–102) | <0.001 |
| eGFR (ml/min/1.73 m2) | 37 (28–51) | 42 (32–54) | 0.051 |
| eGFR < 60 ml/min/1.73 m² | 230 (83%) | 189 (80%) | 0.501 |
| Comorbidities | |||
| History of AF | 194 (70%) | 178 (76%) | 0.132 |
| Diabetes | 139 (50%) | 104 (44%) | 0.194 |
| Hypertension | 206 (74%) | 177 (75%) | 0.752 |
| ACEi/ARB/ARNI | 162 (58%) | 104 (44%) | 0.002 |
| Beta-blocker | 222 (80%) | 195 (83%) | 0.366 |
| MRA | 129 (46%) | 85 (36%) | 0.019 |
| ICD | 42 (15%) | 37 (16%) | 0.842 |
| CRT | 33 (12%) | 27 (12%) | 0.894 |
CRT, cardiac resynchronisation therapy; eGFR, estimated glomerular filtration rate; ICD, implantable cardioverter-defibrillator; LVEF, left ventricular ejection fraction; NT-proBNP, N-terminal pro-B-type natriuretic peptide; NYHA, New York Heart Association; ACEi, angiotensin-converting enzyme inhibitor; ARB, angiotensin II receptor blocker; ARNI, angiotensin receptor-neprilysin inhibitor; MRA, mineralocorticoid receptor antagonist.
Congestion score: see Supplementary data online.
Baseline HCO3 and decongestive response
Figure 1 illustrates the relation between baseline HCO3 levels (on a continuous scale) and the odds ratio for the primary endpoint for allocation towards acetazolamide vs. placebo, showing that over the entire range of HCO3, allocation towards acetazolamide was associated with a higher odds for successful decongestion. The P-value for interaction between the treatment effect of acetazolamide and HCO3 on a continuous scale was P = 0.091, potentially suggesting treatment effect modification. Table 2 indicates the decongestive response to acetazolamide in patients with or without elevated HCO3 further demonstrating that the proportional treatment effect expressed as odds ratio was larger in patients with an elevated HCO3 for both the primary endpoint (P-interaction = 0.065) or the primary endpoint with exclusion for the need for escalation therapy (P-interaction = 0.054). This higher odds ratio was predominantly driven by a lower rate of successful decongestion in the placebo arm (loop diuretics + placebo) if elevated HCO3 was present, while the proportion of patients attaining decongestion in the acetazolamide arm were comparable in numbers.
Treatment effect according to baseline HCO3 levels
| Parameter | N (%) placebo | N (%) acetazolamide | OR | 95% CI | *P-interaction |
|---|---|---|---|---|---|
| Primary endpoint | |||||
| Overall | 77 (30.2%) | 108 (42.5%) | 1.71 | 1.18–2.46 | 0.004 |
| HCO3 < 27 mmol/l | 49 (36.6%) | 62 (43.1%) | 1.37 | 0.79–2.37 | 0.065* |
| HCO3 ≥ 27 mmol/l | 28 (23.1%) | 46 (41.8%) | 2.39 | 1.35–4.22 | |
| Primary endpoint, excluding need for escalation | |||||
| Overall | 86 (33.25) | 115 (44.9%) | 1.77 | 1.19–2.64 | 0.005 |
| HCO3 < 27 mmol/l | 53 (39.6%) | 65 (45.1%) | 1.26 | 0.78–2.03 | 0.054* |
| HCO3 ≥ 27 mmol/l | 31 (25.6%) | 50 (45.5%) | 2.62 | 1.44–4.78 | |
| Parameter | N (%) placebo | N (%) acetazolamide | OR | 95% CI | *P-interaction |
|---|---|---|---|---|---|
| Primary endpoint | |||||
| Overall | 77 (30.2%) | 108 (42.5%) | 1.71 | 1.18–2.46 | 0.004 |
| HCO3 < 27 mmol/l | 49 (36.6%) | 62 (43.1%) | 1.37 | 0.79–2.37 | 0.065* |
| HCO3 ≥ 27 mmol/l | 28 (23.1%) | 46 (41.8%) | 2.39 | 1.35–4.22 | |
| Primary endpoint, excluding need for escalation | |||||
| Overall | 86 (33.25) | 115 (44.9%) | 1.77 | 1.19–2.64 | 0.005 |
| HCO3 < 27 mmol/l | 53 (39.6%) | 65 (45.1%) | 1.26 | 0.78–2.03 | 0.054* |
| HCO3 ≥ 27 mmol/l | 31 (25.6%) | 50 (45.5%) | 2.62 | 1.44–4.78 | |
| Parameter | mean ± SD placebo | mean ± SD acetazolamide | Absolute difference | 95% CI | *P-interaction |
|---|---|---|---|---|---|
| Cumulative diuretic response (in litre per 40 mg furosemide equivalent) | |||||
| Overall | 1.3±0.8 | 1.5±0.9 | −148 | −307 to 10 | <0.001 |
| HCO3 < 27 mmol/l | 1.5 ± 0.9 | 1.6 ± 0.9 | −101 | −325 to 122 | <0.001* |
| HCO3 ≥ 27 mmol/l | 1.1 ± 0.8 | 1.3 ± 0.9 | −138 | −358 to 82 | |
| Cumulative natiuretic response (in mmol per 40 mg furosemide equivalent) | |||||
| Overall | 120 ± 109 | 151 ± 125 | −31 | −52 to −11 | 0.003 |
| HCO3 < 27 mmol/l | 138 ± 124 | 166 ± 126 | −26 | −57 to 2.4 | <0.001* |
| HCO3 ≥ 27 mmol/l | 101 ± 86 | 133 ± 123 | −32 | −57 to −3.0 | |
| Parameter | mean ± SD placebo | mean ± SD acetazolamide | Absolute difference | 95% CI | *P-interaction |
|---|---|---|---|---|---|
| Cumulative diuretic response (in litre per 40 mg furosemide equivalent) | |||||
| Overall | 1.3±0.8 | 1.5±0.9 | −148 | −307 to 10 | <0.001 |
| HCO3 < 27 mmol/l | 1.5 ± 0.9 | 1.6 ± 0.9 | −101 | −325 to 122 | <0.001* |
| HCO3 ≥ 27 mmol/l | 1.1 ± 0.8 | 1.3 ± 0.9 | −138 | −358 to 82 | |
| Cumulative natiuretic response (in mmol per 40 mg furosemide equivalent) | |||||
| Overall | 120 ± 109 | 151 ± 125 | −31 | −52 to −11 | 0.003 |
| HCO3 < 27 mmol/l | 138 ± 124 | 166 ± 126 | −26 | −57 to 2.4 | <0.001* |
| HCO3 ≥ 27 mmol/l | 101 ± 86 | 133 ± 123 | −32 | −57 to −3.0 | |
| Parameter | Geometric mean Placebo | Geometric mean acetazolamide | GM ratio | 95% CI of GM ratio | *P-interaction |
|---|---|---|---|---|---|
| Length of stay (days) | |||||
| Overall | 9.8 | 8.7 | 0.89 | 0.86–0.93 | 0.008 |
| HCO3 < 27 mmol/l | 9.0 | 8.5 | 0.95 | 0.91–1.02 | 0.019* |
| HCO3 ≥ 27 mmol/l | 10.9 | 8.9 | 0.82 | 0.78–0.82 | |
| Parameter | Geometric mean Placebo | Geometric mean acetazolamide | GM ratio | 95% CI of GM ratio | *P-interaction |
|---|---|---|---|---|---|
| Length of stay (days) | |||||
| Overall | 9.8 | 8.7 | 0.89 | 0.86–0.93 | 0.008 |
| HCO3 < 27 mmol/l | 9.0 | 8.5 | 0.95 | 0.91–1.02 | 0.019* |
| HCO3 ≥ 27 mmol/l | 10.9 | 8.9 | 0.82 | 0.78–0.82 | |
| Parameter | N (%) placebo | N (%) acetazolamide | HR | 95% CI | *P-interaction |
|---|---|---|---|---|---|
| Risk for all-cause mortality and heart failure admission (HR) | |||||
| Overall | 72 (27.8%) | 76 (29.7%) | 1.07 | 0.78–1.48 | P = 0.667 |
| HCO3 < 27 mmol/l | 33 (24.8%) | 44 (30.3%) | 1.29 | 0.82–2.03 | 0.663* |
| HCO3 ≥ 27 mmol/l | 39 (32.5%) | 32 (29.4%) | 0.86 | 0.54–1.37 | |
| Parameter | N (%) placebo | N (%) acetazolamide | HR | 95% CI | *P-interaction |
|---|---|---|---|---|---|
| Risk for all-cause mortality and heart failure admission (HR) | |||||
| Overall | 72 (27.8%) | 76 (29.7%) | 1.07 | 0.78–1.48 | P = 0.667 |
| HCO3 < 27 mmol/l | 33 (24.8%) | 44 (30.3%) | 1.29 | 0.82–2.03 | 0.663* |
| HCO3 ≥ 27 mmol/l | 39 (32.5%) | 32 (29.4%) | 0.86 | 0.54–1.37 | |
| Parameter | N (%) placebo | N (%) acetazolamide | HR | 95% CI | *P-interaction |
|---|---|---|---|---|---|
| Risk for all-cause mortality (HR) | |||||
| Overall | 27 (26.5%) | 33 (28.4%) | 1.27 | 0.80–2.04 | P = 0.639 |
| HCO3 < 27 mmol/l | 12 (9.0%) | 22 (15.3%) | 1.75 | 0.87–3.54 | 0.656* |
| HCO3 ≥ 27 mmol/l | 19 (15.7%) | 17 (15.5%) | 0.96 | 0.50–1.85 | |
| Parameter | N (%) placebo | N (%) acetazolamide | HR | 95% CI | *P-interaction |
|---|---|---|---|---|---|
| Risk for all-cause mortality (HR) | |||||
| Overall | 27 (26.5%) | 33 (28.4%) | 1.27 | 0.80–2.04 | P = 0.639 |
| HCO3 < 27 mmol/l | 12 (9.0%) | 22 (15.3%) | 1.75 | 0.87–3.54 | 0.656* |
| HCO3 ≥ 27 mmol/l | 19 (15.7%) | 17 (15.5%) | 0.96 | 0.50–1.85 | |
OR, odds ratio; HR, hazard ratio; CI, confidence interval; GM, geometric mean.
* = P-value for interaction.
Treatment effect according to baseline HCO3 levels
| Parameter | N (%) placebo | N (%) acetazolamide | OR | 95% CI | *P-interaction |
|---|---|---|---|---|---|
| Primary endpoint | |||||
| Overall | 77 (30.2%) | 108 (42.5%) | 1.71 | 1.18–2.46 | 0.004 |
| HCO3 < 27 mmol/l | 49 (36.6%) | 62 (43.1%) | 1.37 | 0.79–2.37 | 0.065* |
| HCO3 ≥ 27 mmol/l | 28 (23.1%) | 46 (41.8%) | 2.39 | 1.35–4.22 | |
| Primary endpoint, excluding need for escalation | |||||
| Overall | 86 (33.25) | 115 (44.9%) | 1.77 | 1.19–2.64 | 0.005 |
| HCO3 < 27 mmol/l | 53 (39.6%) | 65 (45.1%) | 1.26 | 0.78–2.03 | 0.054* |
| HCO3 ≥ 27 mmol/l | 31 (25.6%) | 50 (45.5%) | 2.62 | 1.44–4.78 | |
| Parameter | N (%) placebo | N (%) acetazolamide | OR | 95% CI | *P-interaction |
|---|---|---|---|---|---|
| Primary endpoint | |||||
| Overall | 77 (30.2%) | 108 (42.5%) | 1.71 | 1.18–2.46 | 0.004 |
| HCO3 < 27 mmol/l | 49 (36.6%) | 62 (43.1%) | 1.37 | 0.79–2.37 | 0.065* |
| HCO3 ≥ 27 mmol/l | 28 (23.1%) | 46 (41.8%) | 2.39 | 1.35–4.22 | |
| Primary endpoint, excluding need for escalation | |||||
| Overall | 86 (33.25) | 115 (44.9%) | 1.77 | 1.19–2.64 | 0.005 |
| HCO3 < 27 mmol/l | 53 (39.6%) | 65 (45.1%) | 1.26 | 0.78–2.03 | 0.054* |
| HCO3 ≥ 27 mmol/l | 31 (25.6%) | 50 (45.5%) | 2.62 | 1.44–4.78 | |
| Parameter | mean ± SD placebo | mean ± SD acetazolamide | Absolute difference | 95% CI | *P-interaction |
|---|---|---|---|---|---|
| Cumulative diuretic response (in litre per 40 mg furosemide equivalent) | |||||
| Overall | 1.3±0.8 | 1.5±0.9 | −148 | −307 to 10 | <0.001 |
| HCO3 < 27 mmol/l | 1.5 ± 0.9 | 1.6 ± 0.9 | −101 | −325 to 122 | <0.001* |
| HCO3 ≥ 27 mmol/l | 1.1 ± 0.8 | 1.3 ± 0.9 | −138 | −358 to 82 | |
| Cumulative natiuretic response (in mmol per 40 mg furosemide equivalent) | |||||
| Overall | 120 ± 109 | 151 ± 125 | −31 | −52 to −11 | 0.003 |
| HCO3 < 27 mmol/l | 138 ± 124 | 166 ± 126 | −26 | −57 to 2.4 | <0.001* |
| HCO3 ≥ 27 mmol/l | 101 ± 86 | 133 ± 123 | −32 | −57 to −3.0 | |
| Parameter | mean ± SD placebo | mean ± SD acetazolamide | Absolute difference | 95% CI | *P-interaction |
|---|---|---|---|---|---|
| Cumulative diuretic response (in litre per 40 mg furosemide equivalent) | |||||
| Overall | 1.3±0.8 | 1.5±0.9 | −148 | −307 to 10 | <0.001 |
| HCO3 < 27 mmol/l | 1.5 ± 0.9 | 1.6 ± 0.9 | −101 | −325 to 122 | <0.001* |
| HCO3 ≥ 27 mmol/l | 1.1 ± 0.8 | 1.3 ± 0.9 | −138 | −358 to 82 | |
| Cumulative natiuretic response (in mmol per 40 mg furosemide equivalent) | |||||
| Overall | 120 ± 109 | 151 ± 125 | −31 | −52 to −11 | 0.003 |
| HCO3 < 27 mmol/l | 138 ± 124 | 166 ± 126 | −26 | −57 to 2.4 | <0.001* |
| HCO3 ≥ 27 mmol/l | 101 ± 86 | 133 ± 123 | −32 | −57 to −3.0 | |
| Parameter | Geometric mean Placebo | Geometric mean acetazolamide | GM ratio | 95% CI of GM ratio | *P-interaction |
|---|---|---|---|---|---|
| Length of stay (days) | |||||
| Overall | 9.8 | 8.7 | 0.89 | 0.86–0.93 | 0.008 |
| HCO3 < 27 mmol/l | 9.0 | 8.5 | 0.95 | 0.91–1.02 | 0.019* |
| HCO3 ≥ 27 mmol/l | 10.9 | 8.9 | 0.82 | 0.78–0.82 | |
| Parameter | Geometric mean Placebo | Geometric mean acetazolamide | GM ratio | 95% CI of GM ratio | *P-interaction |
|---|---|---|---|---|---|
| Length of stay (days) | |||||
| Overall | 9.8 | 8.7 | 0.89 | 0.86–0.93 | 0.008 |
| HCO3 < 27 mmol/l | 9.0 | 8.5 | 0.95 | 0.91–1.02 | 0.019* |
| HCO3 ≥ 27 mmol/l | 10.9 | 8.9 | 0.82 | 0.78–0.82 | |
| Parameter | N (%) placebo | N (%) acetazolamide | HR | 95% CI | *P-interaction |
|---|---|---|---|---|---|
| Risk for all-cause mortality and heart failure admission (HR) | |||||
| Overall | 72 (27.8%) | 76 (29.7%) | 1.07 | 0.78–1.48 | P = 0.667 |
| HCO3 < 27 mmol/l | 33 (24.8%) | 44 (30.3%) | 1.29 | 0.82–2.03 | 0.663* |
| HCO3 ≥ 27 mmol/l | 39 (32.5%) | 32 (29.4%) | 0.86 | 0.54–1.37 | |
| Parameter | N (%) placebo | N (%) acetazolamide | HR | 95% CI | *P-interaction |
|---|---|---|---|---|---|
| Risk for all-cause mortality and heart failure admission (HR) | |||||
| Overall | 72 (27.8%) | 76 (29.7%) | 1.07 | 0.78–1.48 | P = 0.667 |
| HCO3 < 27 mmol/l | 33 (24.8%) | 44 (30.3%) | 1.29 | 0.82–2.03 | 0.663* |
| HCO3 ≥ 27 mmol/l | 39 (32.5%) | 32 (29.4%) | 0.86 | 0.54–1.37 | |
| Parameter | N (%) placebo | N (%) acetazolamide | HR | 95% CI | *P-interaction |
|---|---|---|---|---|---|
| Risk for all-cause mortality (HR) | |||||
| Overall | 27 (26.5%) | 33 (28.4%) | 1.27 | 0.80–2.04 | P = 0.639 |
| HCO3 < 27 mmol/l | 12 (9.0%) | 22 (15.3%) | 1.75 | 0.87–3.54 | 0.656* |
| HCO3 ≥ 27 mmol/l | 19 (15.7%) | 17 (15.5%) | 0.96 | 0.50–1.85 | |
| Parameter | N (%) placebo | N (%) acetazolamide | HR | 95% CI | *P-interaction |
|---|---|---|---|---|---|
| Risk for all-cause mortality (HR) | |||||
| Overall | 27 (26.5%) | 33 (28.4%) | 1.27 | 0.80–2.04 | P = 0.639 |
| HCO3 < 27 mmol/l | 12 (9.0%) | 22 (15.3%) | 1.75 | 0.87–3.54 | 0.656* |
| HCO3 ≥ 27 mmol/l | 19 (15.7%) | 17 (15.5%) | 0.96 | 0.50–1.85 | |
OR, odds ratio; HR, hazard ratio; CI, confidence interval; GM, geometric mean.
* = P-value for interaction.
Figure 2 shows the result of a linear mixed effect model of the daily congestion score during the trial treatment phase (baseline until morning of day 3) after stratification according to absence (panel A) or presence (panel B) of elevated baseline HCO3−. Allocation towards acetazolamide was associated with lower congestion score, with a treatment effect that increased on subsequent days (treatment effect × day interaction P < 0.001 for all). However, baseline HCO3− altered the treatment effect of acetazolamide (treatment effect × HCO3−; P-interaction = 0.033) and the change in treatment effect over time (treatment effect × time × HCO3−; P-interaction < 0.001), indicating a more pronounced treatment effect and increase in treatment effect over time in the presence of baseline elevated HCO3− levels. Supplementary data online, Table S3 confirms that this higher treatment effect of acetazolamide was driven by a higher residual congestion score in the placebo arm in the setting of a high baseline HCO3. Supplementary data online, Table S4 illustrates the decongestive response after categorizing patients into low, normal, or elevated HCO3−, again showing that acetazolamide improved decongestive response in all patients, with the larger proportional effect in patients with an elevated HCO3−, being explained by a diminished decongestive response in the placebo arm.
Change in congestion score according to baseline HCO3−. Results of linear mixed effect model for repeated measurements of congestion score over time. (A and B) Patients and with elevated HCO3, respectively, with larger treatment effect observed in patients with elevated HCO3.
Baseline HCO3 and other endpoints
Table 2 illustrates the relation between baseline HCO3− and diuretic and natriuretic response in patients with or without baseline elevated HCO3−. Because of the differences in baseline loop diuretic dose, cumulative diuresis and natriuresis was indexed for loop diuretic dose (with expression of diuretic response in l/40 mg furosemide equivalents and natriuretic response in mmol/40 mg furosemide equivalents). While allocation to acetazolamide improved diuretic and natriuretic response in all patients across the baseline HCO3−-spectrum, a treatment interaction was found, with patients with an elevated baseline HCO3 showing a proportionally larger treatment effect with acetazolamide. This was associated by a lower diuretic response and natriuretic response in patients in the placebo arm if elevated HCO3− was present. Similarly, while randomization towards acetazolamide shortened length of stay expressed by geometric means in all patients, the proportional treatment effect (geometric mean ratio) of acetazolamide was larger in patients with baseline elevated HCO3− levels (P-interaction = 0.019). In the overall cohort, or in patients with or without an elevated baseline HCO3−, no significant effect of acetazolamide (or treatment interaction) was found on the combined endpoint of all-cause mortality and heart failure hospitalization or all-cause mortality separately (Table 2).
Change in HCO3− according to treatment allocation
Figure 3A illustrates the change in HCO3− according to treatment allocation during the treatment phase, showing a progressive increase in HCO3− in patients in the placebo arm in comparison to the acetazolamide arm (P < 0.001). This was associated with a progressive increase in the proportion of patients with elevated HCO3− in the placebo arm (Figure 3B, P < 0.001 at day 2 and day 3).
Change in HCO3− according to treatment allocation. (A) HCO3 according to time and treatment allocation. Results from linear mixed effect model, reporting (least square means and 95% CI). (B) Bar chart of proportion of patients with elevated HCO3 (>27 mmol/l) according to time and treatment allocation.
Changes in HCO3− and decongestive response according to treatment allocation
Supplementary data online, Figure S3 shows the relation between the change in HCO3− levels and the odds ratio for having successful decongestion according to treatment allocation. An increase in HCO3− was associated with a lower odds ratio for having successful decongestion in the placebo arm, while this was not the case in the acetazolamide arm (P-interaction = 0.041).
Discussion
This sub-group analysis of the ADVOR trial which investigated the decongestion success with acetazolamide on top of standardized loop diuretics in patients with AHF and volume overload provides numerous important insights into the mode of action of acetazolamide to diminish diuretic resistance. (i) Acetazolamide improves decongestive effectiveness over the entire range of baseline HCO3−. (ii) Patients with an elevated HCO3− (as a marker of neuro-hormonal activation and proximal nephron activation) exhibited a more pronounced response to acetazolamide, which seems to be mediated by the prevention of loop diuretic resistance. (iii) Almost half of patients with AHF have a HCO3 ≥ 27 mmol/l and these patients have significant more volume overload despite chronic treatment with higher doses of loop diuretics, again indicating neuro-hormonal activation. (iv) HCO3 progressively increases on consecutive days in patients treated with loop diuretics only (placebo arm), which does not occur in patients treated by the combination of acetazolamide with loop diuretics (Structured Graphical Abstract). These data strongly corroborate physiological reasoning that increased HCO3 both at baseline and during mono-therapy with loop diuretics are the result of increased neuro-hormonal stimulation with more proximal sodium re-absorption which can be specifically counteracted by the addition of acetazolamide.
High HCO3− levels are common in patients presenting with AHF, as for instance half of the patients in a pooled analysis of the DOSE-HF, ROSE-AHF, and CARRESS-HF trial had a baseline HCO3− > 28 mmol/l (median HCO3− value).9,13 Similar to our analysis, patients with higher baseline HCO3− in the DOSE-HF, ROSE-AHF, and CARRESS-HF trial exhibited features of more congestion and worse disease severity. Additionally, in these trials, treatment with loop diuretics resulted in further increase in HCO3−. While different cut-points exist to describe the normal range of HCO3− in relation to acid/base status exist (and often 29 mmol/l is cited as the upper range of normal), the goal of our analysis was to identify a point in HCO3− where treatment response to acetazolamide is significantly higher. The presence of an elevated HCO3− in the context of AHF and loop diuretic therapy is often termed ‘contraction alkalosis’, which is probably a misnomer in AHF.14 The term contraction alkalosis stems from the observation that in healthy individuals a reduction in effective circulating volume (ECV) leads to neuro-hormonal activation. Neuro-hormonal activation results in an increase of the filtration fraction which drives proximal sodium (and HCO3− re-absorption).7 Additionally, angiotensin II stimulates proximal nephron Na/HCO3− retention by regulating activity and expression of the luminal NHE3 transporter and the basolateral NBC (Figure 4), resulting in retention of sodium (correcting the ECV contraction), but also further retention of HCO3− (three HCO3− molecules per one Na molecule for basolateral NBC).4,5 In the setting of heart failure, the ECV is not contracted but the elevation of HCO3 is a reflection of neuro-hormonal activation. Similarly to the pooled analysis of DOSE-HF, ROSE-AHF, and CARRESS-HF trial, patients in the ADVOR trial with elevated HCO3− had more pronounced clinical signs of volume overload, with 41% of patients exhibiting peripheral oedema above the knees (adjudicated in all patients by heart failure experts in a double-blind manner), indicating that there is no volume depletion (contraction of the ECV). Additionally, the increasing HCO3− during decongestion is probably not caused by a contraction of the ECV around a fixed amount of HCO3 but relates to worsening of (intra-renal) neuro-hormonal activation during decongestion, further promoting proximal sodium re-absorption. Indeed, the fact that the increase was largely seen in the loop diuretic therapy only group, which clearly demonstrated significantly less reduction in congestion score, further corroborates this.
Influence of bicarbonate levels on the decongestive response to acetazolamide. Visual representation of study findings with regard to renal physiology of loop diuretic only vs. combination of loop diuretic therapy with acetazolamide used in the ADVOR trial. Loop diuretics and AHF induce neuro-hormonal activation. Neuro-hormonal activation induces vasoconstriction of the efferent arteriole which reduces renal blood flow and results in an increased filtration fraction (ratio of glomerular filtration rate/renal blood flow). An increase in filtration fraction further stimulates proximal NaHCO3 retention (glomerular–tubular balance). Additionally, neuro-hormonal activation also increases the expression and activity of the luminal NHE3 and basolateral NaHCO3 transporters. The result of enhanced proximal nephron sodium retention is lower tubular flow and less Na and Cl presentation to the TAL (further triggering renin release and the macula densa level), elevated HCO3 plasma levels and loop diuretic resistance. This sequence of events is (at least partially) prevented when prescribing acetazolamide.
It is well known that loop diuretics further increase HCO3−. In addition to the kaliuretic effects of loop diuretics (which result in an exchange of K+ for H+ in the distal nephron), loop diuretics inhibit the chloride uptake in the macula densa by blocking the NKCC receptor, leading to worsening of (intra-renal) neuro-hormonal activation. This neuro-hormonal activation results in enhanced proximal nephron sodium and HCO3− retention. Therefore, elevated HCO3− levels are reflective of a state of neuro-hormonal activation resulting in more proximal sodium re-absorption (see Figure 4). In the Chronic Renal Insufficiency Cohort study, a baseline HCO3− > 26 mmol/l was associated with a 1.66 higher risk to develop de novo heart failure in 3586 patients with chronic renal insufficiency.15
We now demonstrate for the first time that treatment allocation towards acetazolamide is associated with an impressive increase of the decongestive response in patients with elevated HCO3 (significant treatment interaction). Moreover, this proportionally higher treatment effect is mainly realized because loop diuretic efficacy decreases when an elevated HCO3− is present. If pronounced proximal nephron Na+ and HCO3− retention is present, the tubular flow and the amount of sodium and chloride presenting in the thick ascending limb of Henle diminishes, which results in less substrate for the NKCC receptor and thus diminished efficacy of loop diuretics.8 This is important in clinical practice as this form of diuretic resistance will often lead to prescription of higher doses of loop diuretics resulting in a vicious cycle of more neuro-hormonal activation and enhanced proximal nephron Na+ and HCO3− retention and worsening loop diuretic resistance. Our data suggest a paradigm shift is needed with regard to the contemporary clinical practice of escalation of loop diuretics and/or adding thiazides in such situations enhanced proximal NaHCO3 retention. In contrast, adding a proximal working natriuretic agent as acetazolamide is needed. Indeed, loop diuretics induce their own resistance, explaining the modest treatment effect observed in the DOSE trial with only about 15% of patients attaining decongestion at 72 h.16 A particular strength of the current study is the randomized double-blind design of the ADVOR trial with an intervention blocking the proximal nephron Na+ and HCO3− re-absorption. Indeed, only a double-blind randomized design allows to conclude that observed differences between treatment groups relate to the intervention. The interaction found with baseline HCO3 (consistently across multiple endpoints) provides unequivocal evidence about the importance of proximal sodium re-absorption and the profound effect of acetazolamide in state of enhanced proximal sodium re-absorption (including the situation where it is self-induced by loop diuretic therapies) (Figure 5).
Kidney physiologic basis for study findings.
Moreover, in the ADVOR trial, the decongestive effectiveness of acetazolamide increased on consecutive days with this increased treatment by day effect being significantly influenced by the baseline HCO3. An increase in HCO3− during decongestion in the placebo arm resulted in a lower odds ratio for successful decongestion, which was not observed in the acetazolamide arm. Similar findings have been reported in another study which indicated that diuretic response to loop diuretics diminishes during NaHCO3 infusion which was not the case during NaCl or NH4Cl infusion. This is likely explained by the fact that the proximal nephron will re-absorb virtually all glomerular filtered HCO3− (reabsorbing 80% of all glomerular filtered HCO3−) to prevent the development of metabolic acidosis, but this also leads to less tubular flow and chloride presentation in the thick ascending limb of Henle.17
The consistent interaction with baseline HCO−3 across several indices of decongestive response including the primary endpoint, the primary endpoint without the need for escalation therapy, diuresis, natriuresis, and length of stay supports the concept that elevated HCO3− (a marker of neuro-hormonal activation and proximal nephron sodium retention) in AHF is associated with loop diuretic resistance, which can be specifically abolished by treatment with acetazolamide, explaining the super-response to acetazolamide in patients with elevated HCO3.Our analysis however should not be interpreted that only patients with elevated HCO3 benefit from acetazolamide. Acetazolamide improved decongestion over the entire range of HCO3, but one of its auxiliary benefits is the capability of preventing loop diuretic resistance in patients with elevated HCO3. This is important as elevation of HCO3 is expected to almost universally occur in the current practice where loop diuretics are almost ubiquitously prescribed to all patients presenting with AHF. As such, upfront combination with acetazolamide and loop diuretic therapy seems to be strongly recommended in patients with AHF and volume overload.
Limitations
Several limitations should be acknowledged. First, as ADVOR was a pragmatic trial, HCO3 was measured in local labs which might have resulted in greater assay variability when compared with a central lab. Second, while treatment effect interaction with baseline (pre-randomization) covariates give a good indication of potential treatment effect modification, interaction analysis with the change in HCO3 after randomization should be interpreted with care as changes in HCO3 are influenced by treatment allocation, increasing the likelihood of finding a significant P-value for interaction. Third, the reported analytic coefficient of variation of HCO3 (how much the lab value would differ after re-laboratory analysis) is between 2.6% and 5.4% representing only a minor change in the overall HCO3 level. Thus, upon re-analysis some patients categorized as HCO3 ≥ 27 mmol/l might actually be categorized as <27 mmol/l. However, as HCO3 was handled both categorically and continuous in interaction analysis, this negates the effect of altered classification due to biologic/analytic variation.
Conclusion
Acetazolamide when added to high dose loop diuretics improves decongestive effectiveness, diuretic response, and shortens length of stay over the entire range of baseline HCO3. However, acetazolamide invokes a super-response in patients with an elevated baseline or treatment-induced elevated HCO3 by specifically counteracting poor loop diuretic response in this patient sub-group.
Acknowledgments
None.
Supplementary data
Supplementary data is available at European Heart Journal online.
Pre-registered clinical trial number
NCT: NCT03505788
Ethical approval
Ethical approval was not required.
Data availability
Anonymized will be made available upon request to the corresponding authors, taking into account the local legal and ethical framework around anonymized data-sharing.
Funding
This study (KCE-17001) is an independent research funded by Belgian Health Care Knowledge Centre under the KCE Trials Program. The views expressed in this publication are those of the author(s) and not necessarily those of Belgian Health Care Knowledge Centre which did not influence the analysis or reporting of the trial. P.M. is supported by a grant from the Belgian American Educational Foundation (BAEF) and by the Frans Van de Werf Fund.
References
Author notes
Conflict of interest P.M.: received consultancy fees from CLS Vifor Pharma. F.H.V.: none. J.D.: none. P.N.: none. E.M.: FWO research grant—1SF6822N. S.N.A.: none. J.M.T.M.: received consultancy fees from Boehringer Ingelheim and a Personal grant from the Dutch Heart Foundation. L.H.: none. K.D.: speakers fee from Abbott, Boehringer ingelheim and Astrazeneca and fees for participation in DSMB of FIRE1, REPREIVE, SEQUANA Medical. A.M.: none. G.F.: grants from the European Commission, and speaker fees or DSMB fees from Bayer and Boehringer Ingelheim and fees from the Heart Failure Association. Serves as committee member for Medtronics, Bayer, Boehringer Ingelhein, Vifor, Amgen, Servier, Novartis. F.R.: not received personal payments by pharmaceutical companies or device manufacturers in the last 3 years. The Department of Cardiology (University Hospital of Zurich/University of Zurich) reports research, educational and/or travel grants from Abbott, Amgen, Astra Zeneca, Bayer, Berlin Heart, B. Braun, Biosense Webster, Biosensors Europe AG, Biotronik, BMS, Boehringer Ingelheim, Boston Scientific, Bracco, Cardinal Health Switzerland, Corteria, Daiichi, Diatools AG, Edwards Lifesciences, Fresenius, Guidant Europe NV (BS), Hamilton Health Sciences, Kaneka Corporation, Kantar, Labormedizinisches Zentrum, Medtronic, MSD, Mundipharma Medical Company, Novartis, Novo Nordisk, Orion, Pfizer, Quintiles Switzerland Sarl, Sahajanand IN, Sanofi, Sarstedt AG, Servier, SIS Medical, SSS International Clinical Research, Terumo Deutschland, Swiss National Foundation, Trama Solutions, V-Wave, Vascular Medical, Vifor, Wissens Plus, ZOLL. The research and educational grants do not impact on Prof. Ruschitzka’s personal remuneration. The Department of Cardiology received consultancy fees form AstraZeneca (IMC), Bayer, Boehringer Ingelheim, Citi Research, Klub Class, Novo Nordisk, Radcliffe Group, Stiftung Pfizer Forschungspreis, Vifor. The Department of Cardiology (University Hospital of Zurich/University of Zurich) reports speaker fees from Abbott, Amgen, AstraZeneca (A + Science AB), Bayer (At the Limits), Boehringer Ingelheim, Boston Scientific (CCE Services), Brigham and Women’s Hospital Boston, C.T.I. GmbH, Hôpitaux Universitaires des Genève (GECORE), Luzerner Kantonsspital, Sanofi-Aventis, Servier, Medscape (WebMD), Medtronic, Medworld, Novartis, Roche, Ruwag, Swiss Heart Failure Academy, The Hong Kong Heart Failure Society, Trama Solutions SL, Inselspital Bern, Charité—Universitätsmedizin Berlin (Medical Education Global Solutions), Romanian Society of Cardiology, ÖKG—Österreichische Gesellschaft für Kardiologie. W.H.W.T.: received grants from NIH and consultancy fees from Owkin, Relypsa, ProCardia, Sequana Medical, Cardiol Therapeutics, Genomics plc, Zehna Therapeutics, Renovacor, Whiteswell, Boston Scientific, Kiniksa Pharmaceuticals, CardiaTec Biosciences, Applier Therapeutics. M.D.: received consultancy fees from Astra Zeneca and Boehringer Ingelheim. W.M.: speakers fees from Medtronic, Abbott, Novartis, Vifor Pharma, AstraZenica, Boehringer Ingelheim, Pfizer, Novo Nordisk.
