Introduction
Heart failure (HF) with preserved ejection fraction (HFpEF) and CKD share similar symptoms and risk factors. Furthermore, CKD worsens HF outcomes and predicts HF mortality better than left ventricular ejection fraction (LVEF).1 However, not all eGFR decline in HF is harmful. For example, sodium-glucose cotransporter-2 (SGLT2) inhibitors acutely reduce eGFR but exert long-term cardiorenal benefits.2,3 In addition, eGFR decline during successful decongestion is linked to better survival.4,5 Therefore, investigators should carefully evaluate the long-term effects on kidney function of therapies that acutely affect eGFR in HFpEF.
The Phosphodiesterase-5 Inhibition to Improve Clinical Status and Exercise Capacity in Diastolic HFpEF (RELAX) trial aimed to investigate the effects of sildenafil on peak oxygen consumption in people with HFpEF.6 Although the trial did not show any improvement in cardiopulmonary end points, it unexpectedly revealed a decline in kidney function in the sildenafil group.6,7 This finding seemingly contrasts with preclinical examples of kidney protective effects of sildenafil.8–12 However, the aforementioned therapies, such as SGLT2 inhibitors, that lead to both acute eGFR declines and long-term kidney benefits emphasize the need to better understand the mechanisms of these complex cardiorenal interactions.
To evaluate the effects of sildenafil on kidney function, we calculated the total eGFR slope for each participant. eGFR slope is a more reliable predictor of adverse kidney outcomes than change in eGFR.13 We also examined the association between eGFR slope and baseline data, cardiovascular imaging, and clinical outcomes, including a composite of hospitalization and mortality. In addition, we performed a binary assessment of substantial eGFR decline (≥20%) to provide more intuitive inferences and support the eGFR slope analyses.
Methods
RELAX Study Cohort
The RELAX trial included 216 participants with HFpEF (LVEF ≥50%), elevated filling pressures (measured invasively or inferred by elevated N-terminal pro–B-type natriuretic peptide [NT-proBNP]), reduced exercise capacity (reduced peak oxygen consumption, PVO2), and eGFR ≥20 ml/min per 1.73 m2. After collecting baseline data, including cardiopulmonary exercise testing, physical examination, echocardiography, cardiac magnetic resonance imaging (cMRI), laboratory data, and HF disease severity questionnaire, participants were randomized 1:1 to placebo or sildenafil. Sildenafil was initially administered at 20 mg three times daily until week 12 and then increased to 60 mg three times daily for the remainder of this study, if tolerated. Sildenafil levels were measured 2 hours after scheduled doses at the 12- and 24-week study visits. All participants provided informed consent for the parent trial, and this study was approved by the Duke University Institutional Review Board.
Baseline Clinical Data
We included the following baseline data: demographics; medical history; medications; clinical assessment and vital signs; laboratory results, including sodium, potassium, BUN, bicarbonate, creatinine, total cholesterol, hemoglobin, NT-proBNP, endothelin-1, and galectin-3; Minnesota Living with HF (MLWHF) Questionnaire; cardiopulmonary data, including PVO2 and 6-minute walk distance (6MWD); echocardiography; and cMRI. Full details on how these data were obtained and validated are published with the RELAX trial.6
Kidney Function and eGFR Decline
After exclusion of four participants for missing baseline creatinine, we applied the revised 2021 Chronic Kidney Disease Epidemiology Collaboration definition to calculate eGFR from serum creatinine in 212 participants.14 This revised definition removed the race modifier. Creatinine was measured at baseline, 3, 12, and 24 weeks. We used linear mixed models with eGFR as a continuous variable, a random intercept for subject, and a random slope for time to calculate eGFR slopes for each participant. Owing to the frequency of creatinine assessment, we extended the acute period to week 12 to provide three creatinine measurements. We were then able to apply the same modeling approach to assess acute eGFR slope as we used for the total and chronic eGFR slopes. The models were then adjusted for baseline eGFR. Time was windowed as a proportion of 1 year to yield an eGFR slope in units of ml/min per 1.73 m2 per year. We used the same mixed model approach to calculate acute slopes (baseline to week 12) and chronic slopes (week 3 to week 24) for each participant. We also performed a supplementary analysis that defined the acute slope as weeks 0–3. Because this time window includes only two data points, we calculated this modified acute eGFR slope from the 3-week change from baseline followed by conversion of time to years. We then used linear regression adjusted for baseline eGFR to compare the modified acute slopes between treatment and placebo groups. We also assessed whether change in BP influenced the association between treatment group and acute eGFR slope. Using the systolic BP obtained before 6MWD assessment at baseline and week 12, we calculated the change in systolic BP. We determined the association between this change in BP and acute eGFR slope before and after adjustment for baseline systolic BP. Furthermore, we evaluated whether addition of change in BP significantly modified the association between treatment group and acute eGFR slope. To supplement eGFR slope data, we stratified participants by the degree of eGFR change during the 24-week study period: substantial eGFR decline (≥20%) versus stable (<20% decline). Data from trials of SGLT2 inhibitors suggest that an inflection point occurs at this 20% threshold above which benign hemodynamic shifts are less common.15 If week 24 eGFR was not available (N=14), we imputed the week 12 eGFR to calculate change in eGFR (N=10) for this binary end point.
Longitudinal Data and Clinical End Points
Plasma sildenafil levels were measured at weeks 12 and 24. We assessed changes in the following end points and biomarkers after 24 weeks: PVO2, 6MWD, MLWHF questionnaire score, right ventricular systolic pressure (RVSP), LVEF, NT-proBNP, galectin-3, endothelin-1, and high-sensitivity C-reactive protein. We also assessed a composite of time-to-death or all-cause hospitalization.
Statistical Analyses
We used descriptive statistics to compare participants according to quartile of eGFR slope. To analyze the association between select baseline characteristics and eGFR slope, we used linear regression with eGFR slope as the dependent variable, the respective baseline data as the independent variable, and baseline eGFR as a covariate. We applied the same approach to evaluate the association between eGFR slope and baseline cardiovascular imaging (echocardiography and cMRI). Owing to the non-normal distribution of plasma sildenafil levels, we evaluated the association between eGFR slope and the natural log of plasma sildenafil level. To evaluate the association between eGFR slope and 24-week end points, we used linear regression with eGFR slope as the dependent variable, the respective end point data as the independent variable, and baseline eGFR and peak VO2 as covariates. Owing to the small amount of missingness for end points (<3%), we did not impute missing data. We applied Cox proportional hazards modeling to assess the association between eGFR slope and a time-to-event composite of all-cause hospitalization or death, adjusting for baseline eGFR and peak VO2. We included interaction terms for treatment allocation to assess whether sildenafil use modified the relationship between eGFR slope and these clinical parameters.
For the binary assessment of substantial eGFR decline during the study period, we applied logistic regression with substantial eGFR decline as the independent variable, the respective clinical data as the dependent variable, and the same covariate adjustment as the eGFR slope models. We similarly analyzed whether sildenafil treatment modified each association and repeated regression models stratified by treatment allocation. We used Cox proportional hazards modeling to assess the association between substantial eGFR decline and the composite of hospitalization or death, adjusting for baseline eGFR and peak VO2.
All analyses were performed using R version 3.6.3 (R Foundation for Statistical Computing, Vienna, Austria). Each analysis considered a two-sided P value of <0.05 as statistically significant.
Results
Sildenafil Use and eGFR Trajectory
The mean baseline eGFR for the 212 participants in this study was 60.8 ml/min per 1.73 m2. When stratified by quartile of eGFR slope, baseline eGFR was highest among participants in quartile 1 (most negative slope; Table 1). Sildenafil use did not differ by quartile of eGFR slope. Substantial eGFR decline (≥20%) occurred in 17% of the cohort (Table 2). Neither baseline eGFR nor prevalence of sildenafil use significantly differed for participants who experienced substantial eGFR decline. The mean eGFR slope for entire cohort was −3.21 ml/min per 1.73 m2 per year (95% confidence interval [CI], −6.76 to 0.34; Table 3). Sildenafil treatment did not significantly alter eGFR slope compared with placebo. The mean eGFR slope for sildenafil users was −2.99 ml/min per 1.73 m2 per year compared with −3.46 ml/min per 1.73 m2 per year for placebo (difference +0.47 ml/min per 1.73 m2 per year, 95% CI, −6.63 to 7.57 ml/min per 1.73 m2 per year; P=0.90; Table 3).
Table 1 -
Baseline characteristics according to quartile of eGFR slope
| eGFR Slope
a
, ml/min per 1.73 m2 per year |
Quartile 1
(N=53) |
Quartile 2
(N=53) |
Quartile 3
(N=53) |
Quartile 4
(N=53) |
Overall
(N=212) |
| −15.6 (−39.5, −10.5) |
−6.29 (−9.94, −3.18) |
−0.970 (−3.17, 4.12) |
8.95 (4.27, 38.8) |
−3.17 (−39.5, 38.8) |
|
Treatment allocation
|
| Sildenafil |
27 (50.9%) |
32 (60.4%) |
24 (45.3%) |
27 (50.9%) |
110 (51.9%) |
|
Demographics
|
| Age, yr |
67.6 (10.7) |
69.9 (9.07) |
69.8 (11.1) |
66.3 (9.73) |
68.4 (10.2) |
| Women |
25 (47.2%) |
24 (45.3%) |
24 (45.3%) |
30 (56.6%) |
103 (48.6%) |
| Non-White race |
4 (7.5%) |
3 (5.7%) |
4 (7.5%) |
8 (15.1%) |
19 (9.0%) |
|
Medical history
|
| HF vintage, yr |
2.32 (2.88) |
2.55 (3.77) |
3.32 (4.56) |
2.91 (4.01) |
2.78 (3.85) |
| Hypertension |
46 (86.8%) |
47 (88.7%) |
45 (84.9%) |
42 (79.2%) |
180 (84.9%) |
| Stroke |
2 (3.8%) |
2 (3.8%) |
5 (9.4%) |
5 (9.4%) |
14 (6.6%) |
| PVD |
7 (13.2%) |
8 (15.1%) |
7 (13.2%) |
8 (15.1%) |
30 (14.2%) |
| Diabetes |
26 (49.1%) |
24 (45.3%) |
17 (32.1%) |
25 (47.2%) |
92 (43.4%) |
| Smoking |
34 (64.2%) |
34 (64.2%) |
34 (64.2%) |
31 (58.5%) |
133 (62.7%) |
|
Medications
|
| Loop diuretic |
45 (84.9%) |
39 (73.6%) |
40 (75.5%) |
39 (73.6%) |
163 (76.9%) |
| ACEi or ARB |
38 (71.7%) |
37 (69.8%) |
36 (67.9%) |
40 (75.5%) |
151 (71.2%) |
|
β blocker |
44 (83.0%) |
41 (77.4%) |
36 (67.9%) |
40 (75.5%) |
161 (75.9%) |
| MRA |
8 (15.1%) |
4 (7.5%) |
6 (11.3%) |
5 (9.4%) |
23 (10.8%) |
| Statin |
39 (73.6%) |
32 (60.4%) |
35 (66.0%) |
28 (52.8%) |
134 (63.2%) |
|
Vital signs and functional assessment
|
| Systolic BP, mm Hg |
129 (18.1) |
128 (18.8) |
127 (16.2) |
128 (17.0) |
128 (17.4) |
| BMI, kg/m2
|
33.5 (6.56) |
34.5 (6.90) |
34.4 (6.99) |
34.2 (7.73) |
34.2 (7.02) |
| NYHA class |
|
|
|
|
|
| 2
|
27 (50.9%) |
22 (41.5%) |
22 (41.5%) |
28 (52.8%) |
99 (46.7%) |
| 3
|
26 (49.1%) |
31 (58.5%) |
31 (58.5%) |
25 (47.2%) |
113 (53.3%) |
| Peak VO2, ml/min per /kilogram |
12.3 (2.55) |
12.3 (3.63) |
12.4 (3.28) |
12.2 (2.95) |
12.3 (3.10) |
|
Laboratory values
|
| eGFR, ml/min per 1.73 m2
|
68.9 (21.3) |
58.2 (23.0) |
56.4 (25.2) |
59.8 (19.7) |
60.8 (22.8) |
| CKD stage |
|
|
|
|
|
| None
|
32 (60.4%) |
25 (47.2%) |
19 (35.8%) |
25 (47.2%) |
101 (47.6%) |
| 3a
|
15 (28.3%) |
10 (18.9%) |
12 (22.6%) |
12 (22.6%) |
49 (23.1%) |
| 3b |
6 (11.3%) |
12 (22.6%) |
13 (24.5%) |
14 (26.4%) |
45 (21.2%) |
| 4
|
0 (0%) |
6 (11.3%) |
9 (17.0%) |
2 (3.8%) |
17 (8.0%) |
| Hemoglobin, g/dL |
12.9 (1.61) |
12.7 (1.44) |
12.9 (1.43) |
13.2 (1.44) |
12.9 (1.48) |
| NT-proBNP, ng/L |
1200 (1380) |
1320 (1330) |
1020 (1120) |
866 (879) |
1100 (1200) |
| Endothelin-1, pg/mL |
2.58 (0.967) |
2.60 (0.926) |
2.94 (1.88) |
2.63 (1.15) |
2.69 (1.29) |
| Galectin 3, ng/mL |
15.7 (5.70) |
15.5 (6.61) |
14.6 (5.70) |
14.5 (5.24) |
15.1 (5.82) |
| hsCRP, mg/L |
6.81 (8.70) |
5.36 (5.54) |
5.54 (4.41) |
6.29 (6.68) |
5.99 (6.48) |
|
Cardiovascular imaging
|
| TRV, m/s |
2.85 (0.389) |
2.78 (0.437) |
3.02 (0.438) |
2.94 (0.421) |
2.89 (0.425) |
| RVSP, mm Hg |
41.3 (12.0) |
41.7 (12.1) |
46.0 (12.2) |
44.6 (12.2) |
43.3 (12.1) |
| LVEF, % |
61.5 (7.31) |
59.9 (6.55) |
60.9 (4.48) |
60.2 (6.58) |
60.6 (6.31) |
| LV Mass, g |
134 (40.4) |
169 (70.0) |
141 (60.3) |
141 (47.9) |
147 (58.0) |
| Cardiac output, L/min |
4.85 (1.28) |
5.79 (2.32) |
4.77 (1.41) |
4.97 (1.49) |
5.11 (1.73) |
Values represent number (%) or mean (±standard deviation) unless otherwise noted. HF, heart failure; PVD, peripheral vascular disease; ACEi, angiotensin-converting enzyme inhibitor; ARB, angiotensin receptor blocker; MRA, mineralocorticoid receptor antagonist; BMI, body mass index; NT-proBNP, N-terminal pro–brain natriuretic peptide; hsCRP, high-sensitivity C-reactive protein; TRV, tricuspid regurgitant velocity; RVSP, right ventricular systolic pressure; LVEF, left ventricular ejection fraction; LV, left ventricular.
aMedian (minimum, maximum).
Table 2 -
Baseline characteristics according to incidence of substantial eGFR decline
| Baseline Clinical Data |
Substantial (≥20%) eGFR Decline |
Overall (N=212) |
| No (N=172) |
Yes (N=36) |
|
Treatment allocation
|
| Sildenafil |
86 (50.0%) |
20 (55.6%) |
110 (51.9%) |
|
Demographics
|
| Age, yr |
68.5 (10.4) |
67.9 (9.08) |
68.4 (10.2) |
| Women |
89 (51.7%) |
12 (33.3%) |
103 (48.6%) |
| Non-White race |
17 (9.9%) |
1 (2.8%) |
19 (9.0%) |
|
Medical history
|
| HF Vintage, yr |
2.79 (4.08) |
2.76 (2.59) |
2.78 (3.85) |
| Hypertension |
145 (84.3%) |
31 (86.1%) |
180 (84.9%) |
| Stroke |
13 (7.6%) |
1 (2.8%) |
14 (6.6%) |
| PVD |
23 (13.4%) |
6 (16.7%) |
30 (14.2%) |
| Diabetes |
69 (40.1%) |
21 (58.3%) |
92 (43.4%) |
| Smoking |
106 (61.6%) |
25 (69.4%) |
133 (62.7%) |
|
Medications
|
| Loop diuretic |
125 (72.7%) |
34 (94.4%) |
163 (76.9%) |
| ACEi or ARB |
122 (70.9%) |
25 (69.4%) |
151 (71.2%) |
|
β blocker |
127 (73.8%) |
31 (86.1%) |
161 (75.9%) |
| MRA |
17 (9.9%) |
5 (13.9%) |
23 (10.8%) |
| Statin |
104 (60.5%) |
27 (75.0%) |
134 (63.2%) |
|
Vital signs and functional assessment
|
| Systolic BP, mm Hg |
128 (17.1) |
128 (19.9) |
128 (17.4) |
| BMI, kg/m2
|
34.2 (7.12) |
33.4 (6.74) |
34.2 (7.02) |
| NYHA class |
|
|
|
| 2
|
79 (45.9%) |
18 (50.0%) |
99 (46.7%) |
| 3
|
93 (54.1%) |
18 (50.0%) |
113 (53.3%) |
| Peak VO2, ml/min per kilogram |
12.4 (3.17) |
12.0 (2.40) |
12.3 (3.10) |
|
Laboratory values
|
| eGFR, ml/min per 1.73 m2
|
60.7 (22.6) |
61.4 (23.6) |
60.8 (22.8) |
| CKD stage |
|
|
|
| None
|
82 (47.7%) |
17 (47.2%) |
101 (47.6%) |
| 3a
|
39 (22.7%) |
9 (25.0%) |
49 (23.1%) |
| 3b
|
37 (21.5%) |
8 (22.2%) |
45 (21.2%) |
| 4
|
14 (8.1%) |
2 (5.6%) |
17 (8.0%) |
| Hemoglobin, g/dL |
13.0 (1.45) |
12.7 (1.67) |
12.9 (1.48) |
| NT-proBNP, ng/L |
972 (1030) |
1740 (1720) |
1100 (1200) |
| Endothelin-1, pg/mL |
2.65 (1.33) |
2.95 (1.15) |
2.69 (1.29) |
| Galectin 3, ng/mL |
14.9 (5.85) |
16.5 (5.71) |
15.1 (5.82) |
| hsCRP, mg/L |
5.70 (5.77) |
7.41 (9.49) |
5.99 (6.48) |
|
Cardiovascular imaging
|
| TRV, m/s |
2.89 (0.424) |
2.87 (0.419) |
2.89 (0.425) |
| RVSP, mm Hg |
43.0 (11.9) |
43.7 (13.0) |
43.3 (12.1) |
| LVEF, % |
60.3 (5.83) |
62.1 (8.13) |
60.6 (6.31) |
| LV mass, g |
60.3 (5.83) |
62.1 (8.13) |
60.6 (6.31) |
| Cardiac output, L/min |
5.08 (1.81) |
5.19 (0.932) |
5.11 (1.73) |
Values represent number (%) or mean (±standard deviation) unless otherwise noted. HF, heart failure; PVD, peripheral vascular disease; ACEi, angiotensin-converting enzyme inhibitor; ARB, angiotensin receptor blocker; MRA, mineralocorticoid receptor antagonist; BMI, body mass index; NT-proBNP, N-terminal pro–brain natriuretic peptide; TRV, tricuspid regurgitant velocity; RVSP, right ventricular systolic pressure; LVEF, left ventricular ejection fraction; LV, left ventricular.
Table 3 -
Treatment effects on eGFR slopes
| Treatment |
Total Slope |
Acute Slope |
Chronic Slope |
| Sildenafil |
−2.99 (−9.62 to 4.58) |
−6.78 (−21.31 to 1.46) |
−1.40 (−7.91 to 9.25) |
| Placebo |
−3.46 (−8.51 to 1.60) |
−3.63 (−11.73 to 4.48) |
−3.46 (−9.56 to 2.64) |
| Difference |
0.47 (−6.63 to 7.57) |
−3.15 (−14.53 to 8.24) |
2.06 (−6.51 to 10.65) |
Slopes reflect ml/min per 1.73 m2 per year change (95% confidence interval) over weeks 0–24 (total), 0–12 (acute), or 3–24 (chronic).
Both treatment groups experienced a steeper eGFR decline during the acute phase (Figure 1). Participants randomized to sildenafil experienced a more negative acute decline in eGFR (−6.78 ml/min per 1.73 m2 per year compared with −3.63 ml/min per 1.73 m2 per year for placebo; Table 3); however, this difference of −3.15 ml/min per 1.73 m2 per year did not reach statistical significance (P=0.59). When we limited the acute slope to weeks 0–3, the difference in this modified acute slope was −8.9 ml/min per 1.73 m2 per year for sildenafil compared with placebo (mean slope −20.4 ml/min per 1.73 m2 per year for sildenafil; −11.5 ml/min per 1.73 m2 per year for placebo). However, this difference also did not reach statistical significance (P=0.57). In the chronic phase, eGFR decline tended to be less negative in sildenafil users (difference of +2.06 ml/min per 1.73 m2 per year when adjusted for baseline eGFR, 95% CI, −6.51 to 10.65 ml/min per 1.73 m2 per year; P=0.64), although again the difference was insignificant.
;)
Figure 1: Change in eGFR by treatment during the study period. Values reflect mean change in eGFR from baseline. Error bars represent standard error.
Week 12 sildenafil levels were significantly and directly associated with eGFR slope (β coefficient 2.47, P=0.03 after adjustment for baseline eGFR; Table 4). Higher week 12 sildenafil level similarly associated with lower odds of substantial eGFR decline (odds ratio [OR], 0.54, 95% CI, 0.31 to 0.86; P=0.02). Week 24 sildenafil levels were not associated with eGFR slope or odds of substantial eGFR decline. During the acute phase, participants randomized to sildenafil experienced a more negative change in BP than the placebo group that did not reach statistical significance (−2.7 mm Hg for sildenafil, −1.3 mm Hg for placebo; unadjusted P=0.60, P=0.55 after adjustment for baseline BP). This change in BP also did not significantly modify the association between treatment group and acute eGFR slope (P value for interaction=0.85).
Table 4 -
Plasma
sildenafil level and eGFR trajectory
| Variable |
β Coefficient
a
|
P Value |
OR (95% CI)
a
|
P Value |
Sildenafil level, week 12
b
Missingness=10.8% |
2.47 |
0.03 |
0.54 (0.31 to 0.86) |
0.02 |
Sildenafil level, week 24
b
Missingness=16.0% |
−0.48 |
0.64 |
0.99 (0.66 to 1.70) |
0.97 |
OR, odds ratio; CI, confidence interval.
aAdjusted for baseline eGFR.
bNatural log of sildenafil level.
Baseline Clinical Characteristics and eGFR Trajectory
The baseline level of NT-proBNP was inversely associated with eGFR slope (Table 5). Each unit increase in the natural log of NT-proBNP was also associated with increased odds of substantial eGFR decline (OR, 2.20, 95% CI, 1.49 to 3.44). Loop diuretic use at baseline was also associated with significantly lower eGFR slope (β coefficient −4.69, P=0.02) and higher odds of substantial eGFR decline (OR, 6.97, 95% CI, 1.97 to 44.45; P=0.01). Baseline diabetes was also associated with higher odds of substantial eGFR decline (OR, 2.19, 95% CI, 1.04 to 4.69; P=0.04). Sildenafil significantly modified the association between baseline peak VO2 and odds of substantial eGFR decline, suggesting participants randomized to sildenafil experienced a stronger association between peak VO2 and lower odds of substantial eGFR decline (P value for interaction=0.05; Figure 2).
Table 5 -
Association between baseline clinical data and eGFR trajectory
| Variable |
eGFR Slope |
Substantial eGFR Decline |
|
β Coefficient |
P Value
a
|
Treatment Interaction P Value |
OR (95% CI) |
P Value
a
|
Treatment Interaction P Value |
| Hemoglobin (g/dl) |
0.76 |
0.22 |
0.73 |
0.83 (0.63 to 1.09) |
0.19 |
0.28 |
| NT-proBNP (ng/L)
b
|
−2.74 |
<0.001 |
0.81 |
2.20 (1.49 to 3.44) |
<0.001 |
0.17 |
| Endothelin-1 (pg/ml) |
−0.38 |
0.58 |
0.33 |
1.17 (0.89 to 1.51) |
0.22 |
0.27 |
| Galectin 3 (ng/ml) |
−0.21 |
0.17 |
0.69 |
1.05 (0.98 to 1.12) |
0.18 |
0.46 |
| NYHA classification
c
|
−1.93 |
0.27 |
0.25 |
0.85 (0.41 to 1.79) |
0.68 |
0.24 |
| Diabetes |
−1.72 |
0.32 |
0.49 |
2.19 (1.04 to 4.69) |
0.04 |
0.29 |
| Loop diuretic use |
−4.69 |
0.02 |
0.15 |
6.97 (1.97 to 44.45) |
0.01 |
0.99 |
| ACEi or ARB use |
1.06 |
0.57 |
0.25 |
0.93 (0.43 to 2.10) |
0.86 |
0.14 |
| MRA use |
−0.58 |
0.83 |
0.23 |
1.48 (0.46 to 4.09) |
0.47 |
0.35 |
| Systolic BP (mm Hg) |
0.02 |
0.76 |
0.91 |
1.00 (0.98 to 1.02) |
0.98 |
0.86 |
| Peak VO2 (ml/min per kilogram) |
0.18 |
0.53 |
0.68 |
0.95 (0.83 to 1.07) |
0.40 |
0.05 |
| TRV (m/s) |
2.58 |
0.31 |
0.21 |
0.91 (0.32 to 2.49) |
0.86 |
0.78 |
| RVSP (mm Hg)
b
|
3.39 |
0.37 |
0.78 |
1.12 (0.25 to 5.13) |
0.88 |
0.26 |
| LVEF (%) |
−0.18 |
0.39 |
0.02 |
1.05 (0.99 to 1.11) |
0.12 |
0.01 |
| LV Mass (g) |
0.003 |
0.89 |
0.93 |
1.00 (0.99 to 1.01) |
0.78 |
0.81 |
| Cardiac output (L/min) |
−0.60 |
0.36 |
0.96 |
1.04 (0.72 to 1.41) |
0.80 |
0.65 |
OR, odds ratio; CI, confidence interval; NT-proBNP, N-terminal pro–brain natriuretic peptide; ACEi, angiotensin converting enzyme inhibitor; ARB, angiotensin receptor blocker; MRA, mineralocorticoid receptor antagonist; TRV, tricuspid regurgitant velocity; RVSP, right ventricular systolic pressure; LVEF, left ventricular ejection fraction; LV, left ventricular.
aAdjusted for baseline eGFR.
bNatural log.
cClass III with class II as referent.
Figure 2: Association between clinical end points and substantial eGFR decline by treatment subgroup.
Baseline cardiovascular imaging was not associated with eGFR slope or odds of substantial eGFR decline (Table 5). However, sildenafil use significantly modified the relationship between baseline LVEF and eGFR slope (P value for interaction=0.02; Figure 2). In participants randomized to placebo, higher LVEF was associated with more rapid eGFR decline (β coefficient −0.41, P=0.03), whereas participants randomized to sildenafil experienced no relationship between LVEF and eGFR slope. The same interaction was present for the association between LVEF and the odds of substantial eGFR decline during the study period (P value for interaction=0.01), whereby each 1% increase in LVEF was associated with 12% higher odds of substantial eGFR decline in the placebo group only (OR, 1.12, 95% CI, 1.04 to 1.29, P=0.005).
Clinical End Points
Select changes in biomarkers and cardiovascular imaging were associated eGFR slope (Table 6). A 24-week increase in NT-proBNP was numerically associated with a more negative eGFR slope (β coefficient −1.32, P=0.14) and higher odds of substantial eGFR decline (OR, 1.60, 95% CI, 0.96 to 2.96, P=0.10), although neither association reached statistical significance. Each pg/ml increase in endothelin-1 was associated with 44% higher odds of substantial eGFR decline (OR, 1.44, 95% CI, 1.10 to 2.04, P=0.02). The eGFR slope was not significantly associated with time-to-death or hospitalization (hazard ratio [HR], 0.99, 95% CI, 0.97 to 1.02, P=0.65); however, participants with substantial eGFR decline experienced over two-fold higher risk of this composite (HR, 2.34, 95% CI, 1.21 to 4.53, P=0.01).
Table 6 -
Association between eGFR trajectory and clinical end points
| End Point (24-wk Change) |
eGFR Slope |
Substantial eGFR Decline |
|
β Coefficient or HR (95% CI) |
P Value
a
|
OR or HR (95% CI) |
P Value
a
|
| Peak VO2 (ml/min per kilogram) |
−0.29 |
0.50 |
0.97 (0.80 to 1.17) |
0.74 |
| 6MWD (m)
b
|
−0.002 |
0.86 |
1.03 (0.99 to 1.08) |
0.14 |
| MLWHF total score |
0.02 |
0.62 |
1.10 (0.92 to 1.32) |
0.31 |
| LVEF (%) |
0.02 |
0.82 |
0.99 (0.95 to 1.04) |
0.73 |
| NT-proBNP (ng/L) |
−1.32 |
0.14 |
1.60 (0.96 to 2.96) |
0.10 |
| Galectin-3 (ng/ml)
c
|
−0.11 |
0.27 |
1.13 (0.70 to 1.65) |
0.53 |
| Endothelin-1 (pg/ml) |
−0.32 |
0.62 |
1.44 (1.10 to 2.04) |
0.02 |
| Hospitalization or death |
0.99 (0.97 to 1.02) |
0.65 |
2.34 (1.21 to 4.53) |
0.01 |
For odds of substantial decline, the odds ratio reflects per one unit change of the continuous end point unless otherwise noted. OR, odds ratio; CI, confidence interval; 6MWD, 6-minute walk distance; MLWHF, Minnesota living with heart failure; LVEF, left ventricular ejection fraction; NT-proBNP, N-terminal pro–brain natriuretic peptide.
aAdjusted for baseline eGFR and peak VO2.
bPer 10-m increase.
cPer 10 ng/ml increase.
Discussion
In this population with HFpEF, sildenafil use did not affect the eGFR slope or significantly increase the odds of substantial eGFR decline over 24 weeks of treatment. Participants who did suffer substantial eGFR decline experienced significantly higher odds of hospitalization or death. While the overall eGFR slope was similar between treatment groups, sildenafil exerted more of an acute effect on the eGFR slope than placebo. Among those treated with sildenafil, higher plasma sildenafil levels at week 12 (although not week 24) were associated with less eGFR decline. Higher NT-proBNP levels, loop diuretic use, and diabetes were most strongly associated with eGFR decline. Endothelin-1 increase coincided with a substantial eGFR decline during the study period. Together, these results suggest that sildenafil exerts an acute but unlikely deleterious effect on eGFR when used in patients with HFpEF.
We found no significant difference in eGFR slope after sildenafil therapy compared with placebo. While this finding contrasts with the decrease in eGFR for the sildenafil group reported in the parent study,6,7 we assessed change in eGFR differently. First, we used the revised 2021 formula to calculate eGFR; this model excluded the race modifier and validated new coefficients.14 Furthermore, we applied linear mixed-effects modeling to estimate eGFR slopes for each participant. This robust approach aligns with the method used by the National Kidney Foundation, Food and Drug Administration, and European Medicines Agency cosponsored workshop, which endorsed eGFR slope as a surrogate end point for CKD clinical trials.13 The mixed modeling approach also retains all creatinine data rather than excluding participants or imputing missing data. By including more participants, we decrease the risk of survivor bias since only participants well enough to complete the study had week 24 data. We supplemented eGFR slope data with an assessment of substantial eGFR decline, defined as a ≥20% decline in eGFR during the study period. The inferences from this dichotomous outcome aligned with the eGFR slope data.
Beyond a lack of clear harm to kidney function, select observations in our study may signal potential eGFR benefit with sildenafil use. After exclusion of the acute phase (but continued adjustment for baseline eGFR), sildenafil users tended to experience less eGFR decline than the placebo group. Furthermore, higher week 12 sildenafil levels were associated with better eGFR slope. These observations align with data from other clinical populations and animal models that suggest a beneficial effect of sildenafil on kidney function. For example, patients with pulmonary arterial hypertension treated with sildenafil experienced improved eGFR compared with placebo after 12 weeks.16 Pretreatment with sildenafil also mitigated kidney injury in rodent models of diabetes, ischemia-reperfusion injury, and cisplatin toxicity.17–19 Furthermore, sildenafil reduced proteinuria in a murine model through decreased expression of podocyte transient receptor potential channel C6 (TRPC6) a prominent channel implicated in the pathophysiology of various glomerulopathies.20 To the best of our knowledge, no models have examined the kidney effects of sildenafil in HF. However, combined angiotensin receptor blocker (ARB) neprilysin inhibitors, which block the degradation of brain natriuretic peptide (BNP), also slow eGFR decline and lower the risk of kidney end points compared with ARB monotherapy in people with HFpEF.21 Since phosphodiesterase 5 acts downstream of BNP, this shared pathway may suggest that kidney benefit from sildenafil therapy is plausible in HFpEF.
The acute eGFR effect of sildenafil may be related to glomerular hemodynamics and autoregulation. Perturbations of the nitric oxide-cyclic guanosine monophosphate pathway impair the autoregulation of kidney blood flow.22,23 Once altered, relatively minor changes in systemic or glomerular BP can lead to substantial reductions in GFR.24 Indeed, phosphodiesterase-5 inhibition by sildenafil decreases the vasoconstrictive actions of angiotensin II in the afferent arteriole and dilates both the afferent and efferent arterioles.25 Sildenafil also influences systemic BP: In an ancillary study of the RELAX trial that performed detailed hemodynamic assessment in a subset of trial participants,26 sildenafil significantly reduced BP when compared with placebo. While a similar hemodynamically mediated acute decline in eGFR is kidney protective in the context of SGTL2 inhibitors and renin-angiotensin-aldosterone blockade, longer and larger studies are required to determine whether any acute influence on eGFR by sildenafil leads to long-term kidney changes in HFpEF.
Independent of sildenafil use, baseline levels of NT-proBNP and use of loop diuretics were significantly associated with more rapid loss of eGFR and higher odds of substantial eGFR decline in our study. These factors likely reflect HF severity or comorbidity burden. Since NT-proBNP levels increase in CKD,27 NT-proBNP may further serve as a marker for CKD severity. Similarly, the association between baseline diabetes and substantial eGFR decline likely reflects poor kidney reserve or overestimated GFR due to hyperfiltration. Endothelin-1 increase during the study period was associated with substantial eGFR decline. While endothelin-1 plays a putative role in CKD and HF progression,28,29 the increase in endothelin-1 may serve as a biomarker of reduced eGFR. Together, these associations between baseline clinical data and eGFR trajectory reaffirm the link between HF severity and the risk of eGFR decline.
Sildenafil significantly modified the association between baseline LVEF and eGFR trajectory in our study. In participants randomized to placebo, each percent increase in LVEF was associated with higher odds of substantial eGFR decline (and more negative eGFR slope). By contrast, sildenafil users demonstrated no clear relationship between LVEF and eGFR trajectory. The LVEF-eGFR relationship demonstrated in the placebo group aligns with data from patients with acute HF (AHF). In a secondary analysis of the RELAX-AHF-2 study, which randomized participants with AHF to serelaxin versus placebo, the highest quartile of baseline LVEF was associated with a 20% increased risk of worsening kidney function during the 5-day study period.30 Studies have replicated this finding in other AHF cohorts.31,32 However, this relationship is less evident in the chronic setting and inverse for HFrEF populations.33,34 Although sildenafil did not affect the primary cardiopulmonary end point, a substudy of the RELAX trial reported significant changes in select cardiovascular parameters, including decreased left ventricular contractility.26 However, whether this change in contractility is sufficient to blunt the relationship between LVEF and eGFR trajectory remains unclear.
We recognize limitations to the study design. The sample size and 24-week duration of this study limit the statistical power to assess select outcomes. For example, studies of SGLT2 inhibitors often require 12–18 months to demonstrate favorable eGFR separation versus placebo. The RELAX trial was also not designed to test the hypotheses examined in this study such that the exploratory nature and number of hypotheses tested increase the risk for type I error. Furthermore, lack of proteinuria data prevented more comprehensive characterization of kidney function. Additional echocardiography data may have better characterized the study cohort. For example, this study did not measure tricuspid annular systolic plane excursion, a valuable assessment of right ventricular function. The three times daily dosing regimen for sildenafil increases the risk of nonadherence which is difficult to capture with the present study design. In addition, the dose escalation after 12 weeks may also explain the absence of a relationship between week 24 sildenafil levels and eGFR slope despite the positive association demonstrated at week 12.
Our study design also provides several strengths. First, we used a mixed modeling approach to calculate the eGFR slope. This approach preserves more participant data and aligns with the method recommended to calculate an eGFR slope as a surrogate for hard kidney outcomes in clinical trials. We also used the revised eGFR equation that eliminated the race modifier. Furthermore, we supplemented the eGFR slope results with a binary threshold for substantial eGFR decline; this end point proved clinically significant, as evidenced by the higher risk of death or hospitalization in participants exceeding this threshold. Finally, the RELAX trial included granular data collection for relevant biomarkers, clinical assessments, cMRI, and echocardiography. These data provided a detailed cardiac phenotype of the study population.
In conclusion, sildenafil therapy induced an acute effect on eGFR in this HFpEF cohort which did not translate into a significant change in the overall eGFR slope after 24 weeks. Studies with a longer duration of follow-up are required to determine whether continued sildenafil therapy would improve kidney function in people with HFpEF.
Disclosures
D. Edmonston reports the following: Ownership Interest: AGNC, Allbirds, Apple, Cardinal Health, CVS, Eli Lilly, Google, Granite PT, Home Depot, and Microsoft. S. Rajagopal reports the following: Consultancy: Aerami Therapeutics, Altavant, APIE Therapeutics, Gossamer Bio, Insmed, Janssen Pharmaceuticals, Liquidia Technologies, Inc, Polarean, TotalCME, United Therapeutics Corp., and Visterra; Ownership Interest: Apie Therapeutics—stock options; Research Funding: Actelion USA, American Heart Association, Altavant, National Institutes of Health, and United Therapeutics; Patents or Royalties: Polarean; and Speakers Bureau: TotalCME. M. Sparks reports the following: Research Funding: Renal Research Institute; Honoraria: Elsevier—Nephrology Secrets; and Advisory or Leadership Role: American Board of Internal Medicine; Nephrology Board, Board of Director, NephJC; Editorial Board—American Journal of Kidney Diseases, Kidney360, Kidney Medicine, ASN Kidney News; KCVD Membership and Communications Committee—AHA; KCVD Scientific and Clinical Education Lifelong Learning Committee (SCILL)—AHA; NKF North Carolina—Medical Advisory Board. M. Wolf reports the following: Consultancy: Akebia, Amgen, Bayer, Enyo, Jnana, Pfizer, Pharmacosmos, and Reata; Ownership Interest: Akebia, Unicycive, and Walden; Research Funding: CSL-Behring; Honoraria: Akebia, Amgen, Bayer, Enyo, Jnana, Pfizer, Pharmacosmos, and Reata; and Advisory or Leadership Role: Akebia, Unicycive, and Walden.
Funding
This study was supported by Duke University School of Medicine.
Acknowledgments
This manuscript was prepared using HFN_RELAX Research Materials obtained from the NHLBI Biologic Specimen and Data Repository Information Coordinating Center and does not necessarily reflect the opinions or views of the HFN_RELAX or the NHLBI.
Author Contributions
Conceptualization: Daniel Edmonston.
Data curation: Daniel Edmonston.
Formal analysis: Daniel Edmonston.
Supervision: Matthew Sparks, Sudarshan Rajagopal, Myles Wolf.
Writing – original draft: Daniel Edmonston.
Writing – review & editing: Matthew Sparks, Sudarshan Rajagopal, Myles Wolf.
References
1. Hillege HL, Girbes ARJ, de Kam PJ, et al. Renal function, neurohormonal activation, and survival in patients with chronic heart failure. Circulation. 2000;102(2):203–210. doi:
10.1161/01.cir.102.2.203
2. Guthrie R. Canagliflozin and cardiovascular and renal events in type 2 diabetes. Postgrad Med. 2018;130(2):149–153. doi:
10.1080/00325481.2018.1423852
3. Heerspink HJL, Stefansson BV, Correa-Rotter R, et al. Investigators: dapagliflozin in patients with chronic kidney disease. N Engl J Med. 2020;383(15):1436–1446. doi:
10.1056/NEJMoa2024816
4. Anker SD, Butler J, Filippatos G, et al. Empagliflozin in heart failure with a preserved ejection fraction. N Engl J Med. 2021;385(16):1451–1461. doi:
10.1056/NEJMoa2107038
5. Testani JM, Chen J, McCauley BD, Kimmel SE, Shannon RP. Potential effects of aggressive decongestion during the treatment of decompensated heart failure on renal function and survival. Circulation. 2010;122(3):265–272. doi:
10.1161/circulationaha.109.933275
6. Redfield MM, Chen HH, Borlaug BA, et al. Effect of phosphodiesterase-5 inhibition on exercise capacity and clinical status in
heart failure with preserved ejection fraction: a randomized clinical trial. JAMA. 2013;309(12):1268–1277. doi:
10.1001/jama.2013.2024
7. Patel RB, Mehta R, Redfield MM, et al. Renal dysfunction in
heart failure with preserved ejection fraction: insights from the RELAX trial. J Card Fail. 2020;26(3):233–242. doi:
10.1016/j.cardfail.2020.01.003
8. Jeong KH, Lee TW, Ihm CG, Lee SH, Moon JY, Lim SJ. Effects of
sildenafil on oxidative and inflammatory injuries of the kidney in streptozotocin-induced diabetic rats. Am J Nephrol. 2009;29(3):274–282. doi:
10.1159/000158635
9. Cui W, Maimaitiyiming H, Qi X, et al. Increasing cGMP-dependent protein kinase activity attenuates unilateral ureteral obstruction-induced renal fibrosis. Am J Physiol Renal Physiol. 2014;306(9):F996–F1007. doi:
10.1152/ajprenal.00657.2013
10. Rodragguez-Iturbe B, Ferrebuz A, Vanegas V, et al. Early treatment with cGMP phosphodiesterase inhibitor ameliorates progression of renal damage. Kidney Int. 2005;68(5):2131–2142. doi:
10.1111/j.1523-1755.2005.00669.x
11. Tapia E, Sanchez-Lozada LG, Soto V, et al.
Sildenafil treatment prevents glomerular hypertension and hyperfiltration in rats with renal ablation. Kidney Blood Press Res. 2012;35(4):273–280. doi:
10.1159/000334952
12. Bae EH, Kim IJ, Joo SY, et al. Renoprotective effects of
sildenafil in DOCA-salt hypertensive rats. Kidney Blood Press Res. 2012;36(1):248–257. doi:
10.1159/000343414
13. Inker LA, Heerspink HJL, Tighiouart H, et al. GFR slope as a surrogate end point for kidney disease progression in clinical trials: a meta-analysis of treatment effects of randomized controlled trials. J Am Soc Nephrol. 2019;30(9):1735–1745. doi:
10.1681/ASN.2019010007
14. Inker LA, Eneanya ND, Coresh J, et al.; Chronic Kidney Disease Epidemiology Collaboration. New creatinine- and cystatin C-based equations to estimate GFR without race. N Engl J Med. 2021;385(19):1737–1749. doi:
10.1056/NEJMoa2102953
15. Oshima M, Jardine MJ, Agarwal R, et al. Insights from CREDENCE trial indicate an acute drop in estimated glomerular filtration rate during treatment with canagliflozin with implications for clinical practice. Kidney Int. 2021;99(4):999–1009. doi:
10.1016/j.kint.2020.10.042
16. Webb DJ, Vachiery JL, Hwang LJ, Maurey JO.
Sildenafil improves renal function in patients with pulmonary arterial hypertension. Br J Clin Pharmacol. 2015;80(2):235–241. doi:
10.1111/bcp.12616
17. Pofi R, Fiore D, De Gaetano R, et al. Phosphodiesterase-5 inhibition preserves renal hemodynamics and function in mice with diabetic kidney disease by modulating miR-22 and BMP7. Sci Rep. 2017;7(1):44584. doi:
10.1038/srep44584
18. Choi DE, Jeong JY, Lim BJ, et al. Pretreatment of
sildenafil attenuates ischemia-reperfusion renal injury in rats. Am J Physiol Renal Physiol. 2009;297(2):F362–F370. doi:
10.1152/ajprenal.90609.2008
19. Lee KW, Jeong JY, Lim BJ, et al.
Sildenafil attenuates renal injury in an experimental model of rat cisplatin-induced nephrotoxicity. Toxicology. 2009;257(3):137–143. doi:
10.1016/j.tox.2008.12.017
20. Sonneveld R, Hoenderop JG, Isidori AM, et al.
Sildenafil prevents podocyte injury via PPAR-gamma-mediated TRPC6 inhibition. J Am Soc Nephrol. 2017;28(5):1491–1505. doi:
10.1681/ASN.2015080885
21. Peikert A, Vaduganathan M, Mc Causland F, et al. Effects of sacubitril/valsartan versus valsartan on renal function in patients with and without diabetes and
heart failure with preserved ejection fraction: insights from PARAGON-HF. Eur J Heart Fail. 2022;24(5):794–803. doi:
10.1002/ejhf.2450
22. Kvam FI, Ofstad J, Iversen BM. Role of nitric oxide in the autoregulation of renal blood flow and glomerular filtration rate in aging spontaneously hypertensive rats. Kidney Blood Press Res. 2000;23(6):376–384. doi:
10.1159/000025986
23. Turkstra E, Braam B, Koomans HA. Impaired renal blood flow autoregulation in two-kidney, one-clip hypertensive rats is caused by enhanced activity of nitric oxide. J Am Soc Nephrol. 2000;11(5):847–855. doi:
10.1681/ASN.v115847
24. Griffin KA. Hypertensive kidney injury and the progression of chronic kidney disease. Hypertension. 2017;70(4):687–694. doi:
10.1161/hypertensionaha.117.08314
25. Wennysia IC, Zhao L, Schomber T, et al. Role of soluble guanylyl cyclase in renal afferent and efferent arterioles. Am J Physiol Renal Physiol. 2021;320(2):F193–F202. doi:
10.1152/ajprenal.00272.2020
26. Borlaug BA, Lewis GD, McNulty SE, et al. Effects of
sildenafil on ventricular and vascular function in
heart failure with preserved ejection fraction. Circ Heart Fail. 2015;8(3):533–541. doi:
10.1161/circheartfailure.114.001915
27. Takase H, Dohi Y. Kidney function crucially affects B-type natriuretic peptide (BNP), N-terminal proBNP and their relationship. Eur J Clin Invest. 2014;44(3):303–308. doi:
10.1111/eci.12234
28. Kohan DE, Barton M. Endothelin and endothelin antagonists in chronic kidney disease. Kidney Int. 2014;86(5):896–904. doi:
10.1038/ki.2014.143
29. Yanagisawa M, Kurihara H, Kimura S, et al. A novel potent vasoconstrictor peptide produced by vascular endothelial cells. Nature. 1988;332(6163):411–415. doi:
10.1038/332411a0
30. Feng S, Janwanishstaporn S, Teerlink JR, et al. Association of left ventricular ejection fraction with worsening renal function in patients with acute heart failure: insights from the RELAX-AHF-2 study. Eur J Heart Fail. 2021;23(1):58–67. doi:
10.1002/ejhf.2012
31. Kang J, Park JJ, Cho YJ, et al. Predictors and prognostic value of worsening renal function during admission in HFpEF versus HFrEF: data from the KorAHF (Korean acute heart failure) registry. J Am Heart Assoc. 2018;7(6):e007910. doi:
10.1161/JAHA.117.007910
32. Sweitzer NK, Lopatin M, Yancy CW, Mills RM, Stevenson LW. Comparison of clinical features and outcomes of patients hospitalized with heart failure and normal ejection fraction (≥55%) versus those with mildly reduced (40% to 55%) and moderately to severely reduced (<40%) fractions. Am J Cardiol. 2008;101(8):1151–1156. doi:
10.1016/j.amjcard.2007.12.014
33. Khan NA, Ma I, Thompson CR, et al. Kidney function and mortality among patients with left ventricular systolic dysfunction. J Am Soc Nephrol. 2006;17(1):244–253. doi:
10.1681/ASN.2005030270
34. de Silva R, Nikitin NP, Witte KK, et al. Incidence of renal dysfunction over 6 months in patients with chronic heart failure due to left ventricular systolic dysfunction: contributing factors and relationship to prognosis. Eur Heart J. 2006;27(5):569–581. doi:
10.1093/eurheartj/ehi696
