Retrospective analysis of influences of sacubitril/valsartan in patients with diabetes with poorly controlled blood pressure in Japan

ABSTRACT

Aims

Achieving optimal blood pressure control remains challenging, particularly for patients with diabetes. This post‐hoc study compared the efficacy and safety of switching to sacubitril/valsartan vs adding thiazide diuretics in this population.

Methods

This retrospective study included patients with diabetes and inadequate blood pressure control in the office or home settings despite combination therapy with renin‐angiotensin system inhibitors and calcium channel blockers. Patients were categorized into those receiving an additional thiazide diuretic (THZ group, n = 136) and those switching to sacubitril/valsartan (SacVal group, n = 199). The treatment effects over 12 months were analyzed using propensity score analysis with inverse probability weighting. Treatment discontinuation rates were assessed using the Cox proportional hazards model.

Results

In the propensity score analysis model, target pressure achievement rates were similar between the two groups. However, compared to the THZ group, the SacVal group exhibited significantly lower uric acid levels (P < 0.001), improved glycated hemoglobin A_1c (P = 0.02), and a smaller estimated glomerular filtration rate decline (P = 0.02). Treatment discontinuation due to adverse events was significantly higher in the THZ group (13% vs 1%), with a hazard ratio of 11.53 (95% confidence interval: 2.66–49.93, P < 0.001).

Conclusions

The combination of sacubitril/valsartan with calcium channel blockers provided a similar reduction in blood pressure compared with thiazide with renin‐angiotensin system inhibitors and calcium channel blockers, and reported more favorable changes in uric acid levels, glycated hemoglobin A_1c, and estimated glomerular filtration rate, along with a significantly better treatment tolerability.

Sacubitoril/valsartan vs thiazide in patients with diabetes.

INTRODUCTION

Hypertension is a major risk factor for cardiovascular complications and the most common lifestyle‐related disease ¹ . Despite the availability of numerous effective antihypertensive medications, the treatment and control of hypertension remain suboptimal. In high‐income countries, the prevalence of hypertension is approximately 40%, while blood pressure (BP) control rates remain below 25% ² . In Japan, among the 43 million individuals with hypertension, 33% are unaware of their condition, only 56% receive treatment, and only 27% are estimated to have well‐controlled BP ¹ . This “hypertension paradox” represents an urgent issue that must be addressed to prevent cardiovascular events and improve longevity worldwide, including in Japan.

Hypertension‐related complications occur at nearly twice the rate in patients with diabetes than in those without diabetes, while the prevalence of diabetes is 2–3 times higher in patients with hypertension ³ . Type 2 diabetes and hypertension share common underlying factors, such as obesity and insulin resistance, and are key components of metabolic syndrome ³ . In patients with type 2 diabetes, lower BP levels are associated with reduced mortality and improved clinical outcomes ⁴ . However, in Japan, only 52% of patients with type 2 diabetes achieve a BP target of <130/80 mmHg, highlighting the challenge of effective management in this population ⁵ .

Although guidelines emphasize evidence‐based BP management, gaps persist in public awareness and practical strategies for achieving target BP levels. The Japanese Society of Hypertension Treatment Guidelines 2019 (JSH 2019) ¹ recommend thiazide diuretics, particularly in Japan, where the estimated average daily salt intake remains high at 10.1 g (11.0 g for men and 9.3 g for women) ⁶ . However, the use of thiazide diuretics has significantly declined over the past decade ⁷ . Sacubitril/valsartan, a combination of neprilysin inhibitor and a renin‐angiotensin system (RAS) inhibitor, has gained global recognition as the cornerstone therapy for heart failure (HF) ⁸ , ⁹ , ¹⁰ . Although its potent antihypertensive effects have been documented, its efficacy and safety in routine clinical practice, particularly among patients with hypertension and diabetes, remain insufficiently explored.

We previously reported that in patients with poorly controlled BP despite treatment with RAS inhibitors and calcium channel blockers (CCBs), switching to sacubitril/valsartan resulted in comparable or greater BP reductions, along with improved metabolic parameters, renal function, and treatment tolerability ¹¹ . Based on these findings, this post‐hoc study aimed to evaluate the efficacy and safety of switching to sacubitril/valsartan compared with the addition of thiazide diuretics in patients with hypertension, type 2 diabetes, and inadequate BP control.

MATERIALS AND METHODS

Study participants and data collection

This study is a post‐hoc analysis of our previous research, in which the study design and subject details were described ¹¹ . In brief, we conducted a retrospective survey of patients who visited clinics or hospitals affiliated with the Kanagawa Physicians Association between January 2014 and December 2023. A schematic representation of the study design is provided in Figure S1.

Patients were included if they met the following criteria: (i) Diagnosed with hypertension and exhibiting poor BP control, defined as exceeding the target BP specified in clinical guidelines, despite receiving combination therapy with CCBs and RAS inhibitors, including angiotensin II receptor blockers (ARBs) and angiotensin‐converting enzyme inhibitors (ACEis), for ≥2 months, (ii) undergoing a switch from RAS inhibitors to sacubitril/valsartan (SacVal group) or receiving an addition of thiazide‐type diuretic, including thiazide and thiazide‐like diuretics (THZ group), for ≥4 months, (iii) continuous use of CCBs throughout observational periods, and (iv) availability of clinical data at baseline, 4, and 12 months after treatment, including age*, sex*, height, body weight (BW), systolic blood pressure* (SBP), diastolic blood pressure* (DBP), estimated glomerular filtration rate (eGFR), serum uric acid (UA), serum potassium (K), glycated hemoglobin A_1c (HbA_1c), urine albumin‐to‐creatinine ratio (ACR) or qualitative proteinuria, estimated daily salt intake, patient history*, comorbidities*, and concomitant medications* (*indicates essential data, except for office BP measurements at 12 months).

BP targets for both office and home settings were determined based on patient characteristics according to the JSH 2014 ¹² or 2019 ¹ guidelines, depending on the treatment period.

The eGFR was calculated using the following formula: eGFR (mL/min/1.73 m²) = 194 × age^−0.287 × serum creatinine^−1.094 × (0.739 for women) ¹³ . The gradient of the regression curve using eGFR values at baseline, 4, and 12 months was used as the GFR slope. Qualitative proteinuria values were converted to albuminuria values using the method reported by Sumida et al. ¹⁴ During the observation period, daily salt intake was estimated using Tanaka's formula ¹⁵ .

Patients were excluded if they met any of the following criteria: (i) age < 20 years, (ii) requirement for chronic dialysis, (iii) severe liver dysfunction (e.g., liver cirrhosis), (iv) terminal‐stage malignancy, (v) pregnancy, (vi) treatment discontinuation within the first 4 months, (vii) poor adherence to medication or clinic visits, (viii) missing essential data, and (ix) opting out of the study.

A schematic overview of study participants and the analyzed dataset is shown in Figure S2.

A total of 805 patients from 25 clinics and hospitals were initially included. Sixty‐six patients were excluded, and among the remaining 739 patients, 335 patients with diabetes (199 in the SacVal group and 136 in the THZ group) were used to construct a Cox proportional hazards model for treatment discontinuation after 4 months.

Among the 739 patients that we initially collected, 611 patients could be observed during the course of treatment for up to 12 months (Figure S2). In this study, the MI method for missing values was performed using this dataset of 611 patients. Because our database includes a variety of clinical backgrounds, stratified analysis is of clinical interest. In the post‐hoc stratified analysis, if multiple imputation is performed by extracting data from only the target patients, multiple data sets will be created, and this raises a concern that consistent analytical results may not be obtained. Therefore, in our study, we performed post‐hoc stratified analyses based on the initial multiple imputation data set.

For the multiple imputation (MI) method, 611 patients who completed the 12‐month treatment were analyzed (351 in the SacVal group and 260 in the THZ group). In this post‐hoc study, further analyses were performed in 155 patients in the SacVal group and 111 patients in the THZ group.

This study was approved by the Review Board of the Kanagawa Medical Association, Kanagawa Prefecture, Japan (Approval ID: krec2305, approval on December 18, 2023).

Office and home BP measurements

Office BP was measured at each medical institution using validated oscillometric cuff devices. In accordance with the JSH 2019 ¹ , office BP measurements were performed in a quiet environment after allowing the patient to rest in a seated position with uncrossed legs for several minutes. The average of two consecutive measurements taken 1–2 min apart was recorded as the office BP. Additionally, home BP was measured using oscillometric devices with upper‐arm cuffs. Patients were instructed to measure their BP each morning, and the average of the home BP readings taken during the week immediately before the clinic visit was calculated.

Statistical analysis

Data following a normal distribution are presented as mean ± SD, while skewed data are reported as median (25th, 75th percentile). Statistical analysis was performed using IBM SPSS Statistics software (version 28.0; IBM, Inc., Armonk, NY, USA). Statistical significance was defined as a P‐value < 0.05.

Missing value analysis

To address missing data, the MI method was performed ¹⁶ on 611 patients who completed 12 months of treatment with sacubitril/valsartan or a thiazide diuretic. Because our database includes a variety of clinical backgrounds, stratified analysis is of clinical interest. However, performing MI by extracting data from only the target patients could generate multiple datasets, risking inconsistent analytical results. Therefore, in our study, we performed post‐hoc stratified analyses based on the initial MI data set.

A breakdown of missing data is provided in Figure S3. Due to a higher‐than‐expected missing data rate, missing values were replaced with a set of plausible substituted values by generating 100 imputed datasets using the chained equation method.

Propensity score analysis

Given the retrospective observational nature of this study, adjustments were necessary to account for baseline differences in patient characteristics, which could act as confounding factors. Therefore, propensity score (PS) analysis was performed. For each dataset created through the MI method, the PS for the SacVal group was calculated using logistic regression analysis with the following baseline covariates: age, sex, height, BW, BMI, office SBP/DBP, home SBP/DBP, eGFR, UA, K, HbA_1c, logarithmic value of ACR (LnACR), estimated salt intake, hypertension, comorbidities, types of ARBs, and concomitant medications.

To minimize bias, stabilized average treatment effect (ATE) weighting was applied, and trimming was performed to exclude patients with PS < 0.05 or >0.95 with the inverse probability weighting method (PS‐IPW). The formulas for stabilized ATE weighting were as follows: patients in the SacVal group were weighted by (proportion of SacVal group patients)/PS. Patients in the THZ group were weighted by (proportion of THZ group patients)/(1 − PS). The balance of baseline characteristics between the two groups after PS‐IPW was evaluated using standardized differences. A generalized linear model was then employed for comparative analysis after PS‐IPW adjustment.

Analysis of treatment discontinuation events using the Cox proportional hazards model

A Cox proportional hazards model was used to assess differences in treatment discontinuation events between the two groups.

RESULTS

PS‐IPW analysis

Table 1 presents the baseline characteristics before and after the PS‐IPW. In the PS‐IPW model, the maximum standardized difference among clinical baseline characteristics was 0.13, indicating a well‐balanced model.

Table 1.

Baseline characteristics before and after PS‐IPW method

	Unadjusted (before PS‐IPW)			PS‐IPW with trimming by 0.05 ≦ PS ≦ 0.95
	THZ group, n = 111	SacVal group, n = 155	P‐value	THZ group, n = 101 †	SacVal group, n = 141 †	Standardized difference
Age (year‐old)	66.4 ± 13.0	72.6 ± 11.6	<0.001	69.5 ± 11.6	69.5 ± 12.7	0.001
Sex (female)	36 (32%)	58 (37%)	0.40	39 (39%)	57 (40%)	0.04
History of hypertension ≥10 years	170 (65%)	213 (61%)	0.48	71 (70%)	92 (65%)	0.11
Dyslipidemia	72 (65%)	108 (70%)	0.41	69 (69%)	92 (65%)	0.07
Chronic heart failure	8 (7%)	35 (23%)	<0.001	13 (13%)	18 (12%)	0.003
Atrial fibrillation	6 (5%)	16 (10%)	0.15	6 (6%)	10 (7%)	0.05
Ischemic heart disease	10 (9%)	27 (17%)	0.05	12 (12%)	18 (13%)	0.03
Cerebrovascular diseases	15 (14%)	18 (12%)	0.64	10 (10%)	15 (10%)	0.02
Smoke (current/past)	18 (16%)/30 (27%)	17 (11%)/30 (19%)	0.13	15 (15%)/22 (21%)	26 (19%)/29 (20%)	0.10/0.03
Alcohol (habitual drinking)	22 (20%)	15 (10%)	0.01	14 (14%)	21 (15%)	0.03
BW (kg)	71.0 ± 17.2	67.6 ± 6.5	0.11	69.2 ± 17.7	68.3 ± 20.2	0.05
BMI	26.8 ± 5.4	26.0 ± 4.8	0.22	26.2 ± 5.3	26.0 ± 6.1	0.03
Office SBP (mmHg)	150.2 ± 14.6	145.6 ± 16.6	0.02	148.8 ± 14.4	148.6 ± 16.8	0.01
Office DBP (mmHg)	79.9 ± 13.1	76.8 ± 13.3	0.06	77.5 ± 13.7	77.5 ± 14.4	0.004
eGFR (mL/min/1.73m2)	60.5 ± 23.3	55.7 ± 21.7	0.09	58.3 ± 22.2	56.2 ± 29.2	0.08
UA (mg/dL)	5.7 ± 1.3	5.5 ± 1.4	0.45	5.6 ± 1.2	5.6 ± 1.4	0.04
K (mEq/L)	4.3 ± 0.4	4.4 ± 0.5	0.02	4.3 ± 0.5	4.4 ± 0.6	0.001
HbA1c (mmol/mol (%))	52.9 ± 11.5 (7.0 ± 1.1)	52.0 ± 10.6 (6.9 ± 1.0)	0.52	52.1 ± 11.4 (6.9 ± 1.0)	51.2 ± 15.1 (6.8 ± 1.4)	0.08
Estimated salt intake (g/day)	9.7 ± 4.1	9.4 ± 3.2	0.42	9.6 ± 3.3	9.6 ± 3.1	0.006
Concomitant medications
ACEi before treatment	5 (5%)	5 (3%)	0.59	4 (4%)	5 (4%)	0.03
Types of ARB before treatment
Azilsartan	22 (20%)	47 (30%)	0.05	22 (22%)	31 (22%)	0.005
Candesartan	8 (7%)	6 (4%)	0.23	4 (4%))	6 (4%)	0.02
Irbesartan	7 (6%)	19 (12%)	0.11	11 (11%)	16 (11%)	0.02
Losartan	5 (5%)	9 (6%)	0.64	6 (6%)	8 (6%)	0.01
Olmesartan	24 (22%)	30 (19%)	0.65	22 (22%)	27 (19%)	0.07
Telmisartan	28 (25%)	30 (19%)	0.25	20 (20%)	35 (25%)	0.12
Valsartan	12 (11%)	9 (6%)	0.14	11 (11%)	13 (9%)	0.06
β blocker	18 (16%)	48 (31%)	0.006	21 (21%)	29 (21%)	0.006
MRB	7 (6%)	25 (16%)	0.02	11 (10%)	13 (9%)	0.06
α blocker	12 (11%)	9 (6%)	0.14	9 (9%)	11 (8%)	0.04
Loop diuretics	8 (7%)	18 (12%)	0.23	11 (11%)	14 (10%)	0.03
DPP4 inhibitor	47 (42%)	66 (43%)	0.97	43 (42%)	51 (36%)	0.13
Metformin	33 (30%)	48 (31%)	0.83	27 (27%)	36 (25%)	0.03
Sulphonyl urea	16 (14%)	20 (13%)	0.72	13 (13%)	15 (11%)	0.07
SGLT2 inhibitor	24 (22%)	68 (44%)	<0.001	31 (30%)	42 (29%)	0.02
GLP‐1Ra	11 (10%)	20 (13%)	0.45	10 (10%)	16 (11%)	0.05
Insulin	23 (21%)	24 (16%)	0.27	17 (17%)	25 (18%)	0.02
αGI	12 (11%)	9 (6%)	0.14	7 (7%)	9 (6%)	0.02
Glinide	3 (3%)	4 (3%)	0.95	2 (2%)	2 (2%)	0.04
Pioglitazone	5 (5%)	5 (3%)	0.59	5 (5%)	5 (4%)	0.07
Statin	85 (59%)	93 (60%)	0.81	62 (61%)	91 (64%)	0.07
Medication for hyperuricemia	28 (25%)	38 (25%)	0.90	26 (26%)	41 (29%)	0.08

The values represent the means ± SD, median (25th percentile, 75th percentile), or n/total n (%). P values were calculated by an unpaired t test or chi‐square test. ACEi, angiotensin‐converting enzyme inhibitor; ARB, angiotensin II receptor blocker; BMI, body mass index; BW, body weight; DBP, diastolic blood pressure; DPP4, dipeptidyl peptidase‐4; eGFR, estimated glomerular filtration; GLP‐1Ra, glucagon‐like peptide 1 receptor agonist; HbA_1c, glycated hemoglobin A_1c; HTN, hypertension; K, serum potassium; MRB, mineral corticoid receptor blocker; SacVal, sacubitril valsartan; SBP, systolic blood pressure; SD, standard deviation; SGLT2, sodium–glucose cotransporter inhibitor; THZ, thiazide diuretic; UA, serum uremic acid; αGI, alpha glucosidase inhibitor.

^† Calculated number of participants after weighting.

Figure 1 and Table 2 summarize the results of the generalized linear model, comparing the achievement rates of target BP and BP status after treatment using PS‐IPW analysis. No statistically significant differences in target BP achievement rates were observed between the SacVal and THZ groups.

Figure 1.

Changes in clinical characteristics after treatment after PS‐IPW. Error bars show SD. The numbers in the figure demonstrate the mean differences [95% CI], and P‐values. P‐values was calculated using generalized linear model. BW, body weight; CI, confidence interval; DBP, diastolic blood pressure; eGFR, estimated glomerular filtration; GLMM, generalized linear mixed model; HbA_1c, glycated hemoglobin A_1c; K, serum potassium; LnACR, logarithmic value of urine albumin‐to‐creatinine ratio; SacVal, sacubitril valsartan; SBP, systolic blood pressure; THZ, thiazide diuretic; UA, uremic acid.

Table 2.

Achievement rates for a target BP and clinical characteristics after treatment in PS‐IPW model

	4 months			12 months
	THZ group, n = 101 ‡	SacVal group, n = 141 ‡	OR † or mean difference (95%CI), P‐value by GLM	THZ group, n = 101 ‡	SacVal group, n = 141 ‡	OR or mean difference (95%CI) † , P‐value by GLM
(a) BP management
Achievement rate for a target office BP according to JSH guidelines	29 (29%)	40 (28%)	0.97 [0.48, 1.97], P = 0.94	21 (20%)	44 (31%)	1.75 [0.85, 3.61], P = 0.13
(b) BP values
Office SBP (mmHg)	135.8 ± 16.2	136.6 ± 16.6	0.2 [−4.4, 4.8], P = 0.93	133.7 ± 13.1	133.0 ± 14.9	−0.7 [−4.6, 3.2], P = 0.71
Office DBP (mmHg)	72.6 ± 14.9	73.9 ± 14.4	1.3 [−3.5, 6.1], P = 0.60	73.1 ± 13.7	71.9 ± 13.3	−1.2 [−5.6, 3.1], P = 0.57
c) Clinical characteristics
BW (kg)	68.7 ± 17.3	67.0 ± 19.4	−1.7 [−7.8, 4.4], P = 0.59	68.9 ± 17.4	66.3 ± 19.8	−2.7 [−8.9, 3.5], P = 0.40
BMI	26.0 ± 5.2	25.5 ± 5.8	−0.5 [−2.3, 1.4], P = 0.61	26.1 ± 5.2	25.2 ± 6.1	−0.9 [−2.8, 1.1], P = 0.38
UA (mg/dl)	6.3 ± 1.4	5.4 ± 1.5	−0.9 [−1.3, −0.5], P < 0.001	6.4 ± 1.6	5.6 ± 1.7	−0.9 [−1.4, −0.4], P < 0.001
K (mEq/l)	4.3 ± 0.5	4.3 ± 0.5	0.03 [−0.1, 0.2], P = 0.76	4.3 ± 0.6	4.3 ± 0.6	0.01 [−0.2, 0.2], P = 0.92
eGFR (mL/min/1.73m2)	52.7 ± 21.3	54.6 ± 27.8	1.9 [−6.5, 10.2], P = 0.66	51.4 ± 20.2	53.4 ± 29.7	2.0 [−6.7, 10.7], P = 0.65
HbA1c (mmol/mol (%))	53.0 ± 10.9 (7.0 ± 1.0)	50.6 ± 13.4 (6.8 ± 1.2)	−2.3 [−6.3, 1.6] (−0.2 [−0.6, 0.2]), P = 0.25	53.7 ± 12.6 (7.1 ± 1.2)	48.9 ± 12.9 (6.6 ± 1.2)	−4.8 [−8.8, −0.8] (−0.4 [−0.8, −0.1]), P = 0.02
Annual ΔeGFR (mL/min/1.73m2/year)	No data	No data		−6.2 ± 9.4	−2.6 ± 8.9	3.6 [0.6, 6.5], P = 0.02

The values represent n/total n (%), or means ± SD. BMI, body mass index; BP, blood pressure; BW, body weight; CI, confidence interval; DBP, diastolic blood pressure; eGFR, estimated glomerular filtration; GLM, generalized linear model; HbA_1c, glycated hemoglobin A1c; IPW, inverse probability weighting; JSH, Japanese society of hypertension; K, serum potassium; OR, odds ratio; PS, propensity score; SacVal, sacubitril valsartan; SBP, systolic blood pressure; THZ, thiazide diuretic; UA, serum uremic acid; Δ, change in.

^† Odds ratio or mean difference for patients in the SacVal group compared with those in the THZ group.

^‡ Calculated number of participants after weighting.

Figure 1 and Table 2 also provide details on clinical characteristics after treatment in the PS‐IPW model. Although BP levels were comparable between the two groups throughout the 12‐month treatment period, the change in UA in the THZ group was significantly larger than that in the SacVal group (0.7 ± 1.1 and −0.1 ± 1.1 mg/dL at 4 months (P < 0.001), and 0.8 ± 1.3 and 0.0 ± 1.5 mg/dL at 12 months (P < 0.001), respectively). The generalized linear model demonstrated a mean difference of −0.9 mg/dL at 4 months (95% confidence interval (CI): −1.3 to −0.5, P < 0.001) and a mean difference of −0.9 mg/dL at 12 months (95% CI: −1.4 to −0.4, P < 0.001). Additionally, the change in HbA_1c at 12 months in the THZ group was significantly larger than that in the SacVal group (1.6 ± 8.8 mmol/mol [0.2 ± 0.8%], and −2.3 ± 11.0 mmol/mol [−0.2 ± 1.0%] (P = 0.002), respectively). At 12 months, there was a mean difference of −4.8 mmol/mol (95% CI: −8.8 to −0.8 equivalent to −0.4%; 95% CI: −0.8 to −0.1, P = 0.02) of HbA_1c.

While there were no significant differences in eGFR between the two groups, the SacVal group demonstrated a milder decline in annual eGFR over 12 months: −6.2 ± 9.4 mL/min/1.73 m²/year in the THZ group and −2.6 ± 8.9 mL/min/1.73 m²/year in the SacVal group, with a mean difference of 3.6 (95% CI: 0.6–6.5, P = 0.02).

Because the missing rates were high, the data of home BP and LnACR on PS‐IPW model were described in Table S1.

Treatment discontinuation events

For the analysis of treatment discontinuation after 4 months, a dataset of 335 patients (199 in the SacVal group and 156 in the THZ group) was analyzed using a Cox proportional hazards model. Baseline characteristics of the two groups are detailed in Table S2.

Among the 335 patients with diabetes, 30 patients (9%) discontinued treatment within 12 months for various reasons outlined in Table 3. Twenty‐two patients in the THZ group and 8 patients in the SacVal group discontinued treatment. Kaplan–Meier curves illustrating treatment discontinuation rates are shown in Figure 2. Notably, adverse event‐related discontinuations were significantly more frequent in the THZ group than in the SacVal group (P < 0.001, calculated using the log‐rank test). Based on the Cox proportional hazards model, the hazard ratio (HR) for discontinuation due to adverse events was 11.53 (95% CI: 2.66–49.93, P < 0.001). After adjustment for age, sex, SBP, DBP, eGFR, BMI, and comorbidities, HR was 7.96 (95% CI: 1.74–36.36, P < 0.001).

Table 3.

Breakdown of discontinuation of treatment

	THIZ group (n = 136)	SacVal group (n = 199)
Termination of the treatment by the adverse events	17 (13%)	2 (1%)
	Excess of BP reduction (n = 8)	Hyperkalemia (n = 1)
	Worsening of renal function (n = 4)	BW loss (n = 1)
	Death by worsening of heart failure (n = 1)
	Worsening of hyperuricemia and HbA1c (n = 1)
	Muscle cramp (n = 1)
	New onset of stroke (n = 1)
	New onset of aortic dissection (n = 1)
Interruption of consultation	3 (2%)	2 (1%)
Transfer to other medical facility	1 (1%)	5 (3%)
Termination of the treatment due to the improvement of BP control	1 (1%)	0 (0%)

BP, blood pressure; BW, body weight; HbA1c, glycated hemoglobin A1c; SacVal, sacubitril valsartan; THZ, thiazide diuretic.

Figure 2.

Kaplan–Meier curve for the discontinuation of treatment by the adverse events. CI, confidence interval; HR, hazard ratio; SacVal, sacubitril valsartan; THZ, thiazide diuretic.

DISCUSSION

Sacubitril/valsartan as an antihypertensive drug

In the PS‐IPW model, both groups exhibited a reduction in BP levels. After 12 months of treatment, 20% of patients in the THZ group and 31% in the SacVal group achieved the target office BP. The odds ratio for the SacVal group compared to the THZ group was 1.75 (95% CI: 0.85–3.61, P = 0.13). Although the difference was not statistically significant, the 95% CI suggests that a larger sample size may reveal a potential advantage of switching to sacubitril/valsartan over adding thiazide diuretics in patients with poor BP control despite treatment with RAS inhibitors and CCBs. A similar trend was observed in home BP measurements (Table S1). Sacubitril/valsartan, formerly known as LCZ696, combines the angiotensin receptor blocker valsartan and the neprilysin inhibitor prodrug sacubitril in a 1:1 ratio in a sodium supramolecular complex. Therefore, it is possible that the neprilysin inhibitory effect of sacubitril/valsartan caused these differences.

Resistant hypertension is defined as uncontrolled BP despite treatment with three or more antihypertensive drugs, including diuretics ¹⁷ . While the patients in this study were classified as having “uncontrolled BP” rather than “resistant hypertension,” they shared similar clinical characteristics with patients diagnosed with resistant hypertension, such as inadequate salt intake, fluid retention, poor adherence to medication, chronic kidney disease (CKD), or secondary hypertension (e.g., primary aldosteronism, renovascular hypertension, and sleep apnea) ¹ . In this study, both groups had an estimated salt intake of 9 g/day, suggesting salt sensitivity or excessive salt consumption in many patients. The BP reduction in both the SacVal and THZ groups suggests that salt intake played a crucial role in BP control.

Comparison of sacubitril/valsartan and thiazide diuretics

Thiazide diuretics are highly effective antihypertensive agents, known for their cardiovascular ¹⁸ , ¹⁹ and cerebrovascular ²⁰ , ²¹ protective effects. The Systolic Blood Pressure Intervention Trial (SPRINT) found that thiazide diuretics were used more frequently in patients with strict BP targets (54.9%) than in those with standard BP targets (33.3%) ²² . A small retrospective cohort study of patients with hypertension reported that sacubitril/valsartan demonstrated a comparable BP‐lowering effect compared with thiazide diuretics ²³ . A larger study by Kobayashi et al. reported that sacubitril/valsartan had a similar antihypertensive efficacy for both office and home BP using generalized linear mixed models (GLMM) and confounder‐adjusted PS‐IPW analysis ¹¹ . To the best of our knowledge, this is the first study to report the comparable antihypertensive effect of sacubitril/valsartan versus thiazide diuretics in patients with hypertension and diabetes.

Our previous periodic cross‐sectional studies have demonstrated that thiazide diuretic use has been very limited over the past decade ⁷ , despite strong evidence supporting their antihypertensive efficacy and cardiovascular benefits. Potential reasons include concerns regarding adverse effects, such as hyperuricemia, hypokalemia, hyponatremia, deterioration of glucose levels, dehydration, and excessive BP reduction. To mitigate these risks, low‐dose thiazide diuretics are recommended in clinical practice ¹ . However, in this study, treatment discontinuation due to adverse events was notably higher in the THZ group (13%) than in the SacVal group (2%). Furthermore, serum UA levels were significantly lower in the SacVal group than the THZ group throughout the treatment period (P < 0.001). Thus, the adverse effects associated with thiazide diuretics may indeed contribute to their limited clinical use ²³ . In the JSH2019 ¹ , the risk of cerebrovascular disease is stratified based on office BP, and hypertension with diabetes is classified as the highest risk stratum, even if there are no other risk factors and regardless of BP value. Insulin resistance has been cited as a mechanism by which patients with diabetes develop hypertension ³ . Insulin resistance causes compensatory hyperinsulinemia, which increases both circulating plasma volume and peripheral vascular resistance due to sympathetic nerve activity and enhanced RAS, which leads to raised BP. In addition, mineralocorticoid receptors are pathologically activated in patients with obesity, CKD, and diabetes, and this may be the cause of treatment‐resistant hypertension ²⁴ . Because both fluid retention and vasoconstriction cause hypertension in patients with diabetes, the use of multiple medications is often needed to achieve a target BP. However, sufficient BP reduction has not been achieved in the management of hypertension with diabetes ⁷ . Among new hypoglycemic drugs, the antihypertensive effects of SGLT2 inhibitors ²⁵ and GLP‐1 receptor agonists ²⁶ were reported; however, further strategies for lowering BP in patients with diabetes are needed.

Sacubitril/valsartan has demonstrated a comparable safety profile to olmesartan in Japanese patients with hypertension ²⁷ , suggesting that switching from ARBs to sacubitril/valsartan may offer a more potent antihypertensive effect without additional adverse events associated with thiazide diuretics. In contrast, Matsumoto et al. reported that 22.6% of Japanese patients with HF discontinued sacubitril/valsartan within 3 months, and hypotension was the main cause of adverse effects ²⁸ . Although a direct a comparison cannot be performed, the present study included only patients who could continue treatment for at least 4 months. As a result, early adverse effects associated with sacubitril/valsartan could not be evaluated, resulting in a low frequency of adverse effects in this study.

Furthermore, the timing of the initiation of treatment may introduce bias. The distribution of patients by treatment initiation year is shown in Table S3. Most patients in the SacVal group started treatment during the coronavirus disease‐19 (COVID‐19) pandemic, whereas 60% of those in the THZ group started treatment beforehand. During the state of emergency, an increase in SBP was observed, along with poor adherence to medication and clinic visits in some cases. Although treatment discontinuations in the THZ group increased after the onset of the pandemic, the overall discontinuation rate remained higher than that of the SacVal group.

Effects on metabolites

In this study, sacubitril/valsartan treatment led to a reduced HbA_1c, suggesting potential benefits for metabolic control in patients with non‐communicable diseases. RAS inhibition may prevent the adverse effects of hyperglycemia, including oxidative stress, tissue inflammation, cell proliferation, and apoptosis, while also improving β‐cell function ²⁹ , ³⁰ . Additionally, it enhances glucose utilization in skeletal muscles, reduces adipocyte size, improves adipose tissue function, and alleviates insulin resistance ³¹ . Therefore, RAS inhibitors may reduce the incidence of diabetes ³² . Neprilysin, an endopeptidase, degrades multiple peptides, including natriuretic peptides, angiotensin I and II, and glucagon‐like peptide‐1 (GLP‐1) ³³ . Inhibition of neprilysin increases GLP‐1 levels, reduces dipeptidyl peptidase‐4 (DPP‐4) activity, and improves β‐cell function ³⁴ , ³⁵ . Moreover, neprilysin degrades over 50 vasoactive peptides, including natriuretic peptides, bradykinin, angiotensin I and II, endothelin 1, glucagon, and insulin‐B chains ³⁶ , ³⁷ . This effect of neprilysin inhibition may be associated with the improvement of glucose metabolism by sacubitril. Clinical trials have demonstrated the metabolic benefits of sacubitril/valsartan. In the prospective comparison of ARNI with ACEI to Determine Impact on Global Mortality and morbidity in Heart Failure (PARADIGM‐HF) trial, sacubitril/valsartan led to a greater reduction in HbA_1c than enalapril ³⁸ , and in the prospective comparison of ARNI with ARB Global Outcomes in HF with Preserved Ejection Fraction trial, it was more effective than valsartan in reducing the need for insulin or hypoglycemic agents ³⁹ . In patients with hypertension, sacubitril/valsartan improved insulin sensitivity and lipid mobilization more effectively than amlodipine ⁴⁰ . Although the SacVal group showed significantly lower HbA_1c levels at 12 months compared with baseline, the difference was small and clinical significance might be limited. The large difference in HbA_1c between the two groups appears to result primarily from a significant increase in the THZ group following treatment. A more accurate evaluation of the influence of sacubitril/valsartan requires a study comparing with placebo in patients with higher baseline HbA_1c. Similar results were observed in the comparison of UA levels, and this should be considered when evaluating the comparison between the two groups in this study.

Effects on renal function

Smaller eGFR value at 12 months was observed in the SacVal group than in the THZ group; however, there was no significant difference. In contrast, the SacVal group exhibited a significantly smaller annual decline in eGFR compared to the THZ group (P = 0.02). In contrast, LnACR levels were lower in the THZ group than those in the SacVal group (4.17 ± 2.04 vs 4.67 ± 2.52, respectively); however, the mean difference of 0.51 (95% CI: −0.21 to 1.23) was not statistically significant (P = 0.17, Table S1). Prospective comparison of ARNI with ARB on the management of HF with preserved ejection fraction demonstrated a smaller decline in eGFR by sacubitril/valsartan than valsartan ⁴¹ ; however, the short treatment period of 36 weeks was similar to the present study. The annual change in eGFR over an observation period of more than 2 years should be considered a surrogate marker for renal outcome ⁴² . Furthermore, the initial reduction in renal blood flow induced by thiazide diuretics may significantly affect the annual change in eGFR. Therefore, given the 12‐month observation period, this study cannot draw definitive conclusions regarding the renal protective effects of sacubitril/valsartan. In contrast, the PARADIGM‐HF trial, with a 44‐month follow‐up period, demonstrated that sacubitril/valsartan preserves eGFR and mitigates eGFR decline more effectively than RAS inhibitors ⁴³ . These effects are thought to result from afferent arteriole dilation, efferent arteriole constriction, and mesangial cell relaxation induced by natriuretic peptides ⁴⁴ , ⁴⁵ . However, sacubitril/valsartan has also been associated with increased ACR levels compared to RAS inhibitors ⁴¹ , ⁴³ , ⁴⁶ , potentially due to glomerular hyperfiltration. To date, no randomized controlled trials have been conducted with nephroprotection as a primary endpoint. Additionally, most studies on sacubitril/valsartan have focused on patients with heart failure, making it premature to draw definitive conclusions about its nephroprotective effects based on current evidence.

Limitations of the study

This study has some limitations. First, as a retrospective observational study, research bias, particularly selection bias, cannot be entirely excluded. Although PS‐IPW analysis was used to adjust for known confounders, unknown biases may still exist. For missing values, the MI method consisted of 100 datasets; however, some covariates had more than 30% missing values. Even though we use these statistical methods, large missing rates can cause bias. In contrast, Madley‐Dowd et al. reported that the proportion of missing data should not be used to guide decisions on MI ⁴⁷ . GLMM, another statistical method for handling missing data, was applied to 335 patients with diabetes, including those who discontinued treatment before 12 months. From the results of GLMM (Figure S4), office BP and home BP significantly decreased during the treatment, without interaction between the two groups. Regarding changes in BW, eGFR, UA, HbA_1c, and LnACR, significant interactions were observed between the two groups (P = 0.002, <0.001, <0.001, and <0.001, respectively). At 12 months, a significant decrease in BW and HbA_1c was observed in the SacVal group, and a decrease in LnACR was seen in the THZ group. To balance baseline characteristics between the groups, we performed PS‐IPW in the dataset with MI. Both statistical methods for dealing with missing data have limitations, and prospective outcome‐based studies are needed for definitive conclusions.

Second, this study lacked detailed information on antihypertensive drug dosages or data related to diabetes, for example, history of diabetes, fasting glucose, 1,5‐Anhydroglucitol, or glycated albumin level, which may lead to insufficient analysis in this post‐hoc study that focused on patients with diabetes. Specifically, thiazide diuretic dosages were not collected for patients in the THZ group. The THZ group included three different diuretics: hydrochlorothiazide (24%), trichlorothiazide (56%), and indapamide (21%). Because these medications have distinct pharmacologic properties, treating them as a single group (THZ group) may limit the accuracy of comparisons with sacubitril/valsartan. Thus, we did not collect data on the dosage of RAS inhibitors before switching to sacubitril/valsartan. In cases where the dose of RAS inhibitor contained within sacubitril/valsartan was lower than the previous dosage, the true antihypertensive effect of sacubitril/valsartan may have been underestimated.

Furthermore, information on the administration of antihypertensive or hypoglycemic drugs at the time of 12 months was not collected. Although baseline treatments were balanced between the two groups using the PS‐IPW method, changes in these medications during the study period may have influenced the outcomes. This represents a key limitation of the present observational study.

In patients with diabetes and uncontrolled BP despite combination therapy with a RAS inhibitor and a CCB, transitioning to sacubitril/valsartan resulted in an equivalent reduction in BP compared to adding thiazide diuretics. Additionally, sacubitril/valsartan demonstrated a favorable effect on uric acid levels and eGFR, along with superior treatment tolerability.

DISCLOSURE

Hiromichi Wakui has received research support from AstraZeneca, Otsuka Pharmaceutical, and Kaneka Corporation, and honoraria from AstraZeneca, Otsuka Pharmaceutical, Kaneka Corporation, Kyowa Kirin, Sanwa Kagaku, Chugai Pharma, Kowa, Otsuka Pharmaceutical, Novartis, Mitsubishi‐Tanabe Pharma, Bayer, and Daiichi‐Sankyo Pharma. Masao Toyoda received lecture fees from Boehringer Ingelheim, Eli Lilly, Novo Nordisk, Sumitomo, and Mitsubishi Tanabe and received subsidies from Super Light Water, TAKAGI, Roche DC, and LifeScan. Kouichi Tamura has received honoraria/lecture fees from AstraZeneca, Novartis, Bayer, Otsuka Pharmaceutical, Boehringer Ingelheim, Fuji Pharma, Kyowa Kirin, Ono Pharmaceutical, Sanwa Kagaku, Mochida Pharmaceutical, Kowa, Eli Lilly, Novo Nordisk, commissioned clinical trials, contract research, and joint research funding: AstraZeneca, Bayer, Novartis, Chinook, Otsuka Medical Devices, Novo Nordisk, Terumo, Variatris, and Kowa, and scholarship donations: Otsuka Pharmaceutical, Bayer, Mochida Pharmaceutical, and Boehringer Ingelheim. All other authors declare no conflicts of interest.

Approval of the research protocol: This study was approved by the Review Board of the Kanagawa Medical Association, Kanagawa Prefecture, Japan (krec2305, approval on December 18, 2023).

Informed Consent: N/A.

Registry and the Registration No. of the study/trial: N/A.

Animal Studies: N/A.

FUNDING INFORMATION

This study received no specific grant from any funding agency in the public, commercial, or not‐for‐profit sectors.

Supporting information

Figure S1. Schematic of the study design.

Figure S2. Schematic of the study participants.

Figure S3. Breakdown of the missing data.

Figure S4. Changes in clinical characteristics after treatment on GLMM.

Table S1. Change in home BP and LnACR (PS‐IPW method).

Table S2. Clinical characteristics and the concomitant drugs at baseline.

Table S3. Year of the initiation of treatment and occurrence of adverse events.

DATA AVAILABILITY STATEMENT

Data are available from the Review Board of Kanagawa Medical Association, Kanagawa Prefecture, to investigators bound by confidential agreements. Contact details: Kazuo Kobayashi, M.D., Ph.D., Committee of Hypertension and Kidney Disease, Kanagawa Physicians Association, Yokohama, Kanagawa Prefecture, Japan, E‐mail: k-taishi@xc4.so-net.ne.jp.