Above‐Threshold Ambulatory Systolic Pressure and Left Ventricular Mass in Chronic Kidney Disease: Does AUC Add Beyond Mean Pressure?

ABSTRACT

Above‐threshold ambulatory systolic blood pressure (SBP) area under the curve (AUC) captures both the magnitude and duration of SBP above daytime and nighttime reference values. We examined whether this measure provided information about left ventricular mass index (LVMI) beyond daytime and nighttime mean SBP. In this cross‐sectional study, 212 Chinese adults with non‐dialysis chronic kidney disease (CKD) underwent ambulatory blood pressure monitoring and echocardiography. Daytime AUC above 135 mmHg and nighttime AUC above 120 mmHg were entered jointly. Together, daytime and nighttime AUC were significantly associated with LVMI (P < 0.001; adjusted R ² = 0.345). From the 25th to the 75th percentile, adjusted LVMI was 3.46 g/m²·⁷ higher for daytime AUC and 5.56 g/m²·⁷ higher for nighttime AUC. AUC attributable to daytime and nighttime mean SBP accounted for 89.7% of the cohort‐average above‐threshold burden. The remaining AUC not captured by those period means was not significantly associated with LVMI after adjustment for daytime and nighttime mean SBP (0.69 g/m²·⁷ per interquartile range; 95% confidence interval, ‐2.00 to 3.37). Above‐threshold AUC was associated with greater LVMI but added limited information beyond daytime and nighttime mean SBP in this cohort.

1. Introduction

Chronic kidney disease (CKD) is associated with substantial cardiovascular risk, and blood pressure (BP) management is central to CKD care [1, 2, 3, 4]. Office BP does not fully characterize the ambulatory pressure profile, as masked hypertension, elevated nighttime BP, and a blunted day‐night decline are common in CKD [5, 6, 7, 8]. Ambulatory blood pressure monitoring (ABPM) provides clinically established daytime and nighttime mean systolic blood pressure (SBP) summaries.

Clinical Perspective

1. What is known: Daytime and nighttime mean SBP are established clinical summaries from ambulatory blood pressure monitoring (ABPM). Above‐threshold area under the curve (AUC) combines how high SBP rises and how long it remains above period‐specific reference values.

2. What this study adds: In 212 Chinese adults with non‐dialysis CKD, AUC identified patients with higher left ventricular mass index (LVMI). AUC attributable to daytime and nighttime mean SBP accounted for about 90% of the cohort‐average above‐threshold burden. The remaining AUC not captured by those means was not significantly associated with LVMI beyond daytime and nighttime mean SBP and did not significantly improve internal model performance.

3. Clinical implication: Daytime and nighttime mean SBP should remain the principal clinical ABPM summaries. These cross‐sectional data do not support above‐threshold AUC as a routine additional metric, treatment target, or replacement for established ABPM measures.

Left ventricular hypertrophy (LVH) and altered LV geometry are also common in CKD and are associated with adverse outcomes [9, 10, 11]. Persistent pressure exposure may increase LV wall stress and favor hypertrophic adaptation [12]. In CKD, arterial stiffness, impaired sodium excretion with volume expansion, and activation of sympathetic and renin‐angiotensin‐aldosterone pathways may sustain ambulatory and nocturnal BP. These mechanisms jointly induce the onset of LVH and concentric LV structural changes [1, 3, 9, 10].

Several measures have been proposed to describe ambulatory BP burden. Percentage load counts the proportion of readings above a reference value and has been studied in adult and pediatric CKD and in population cohorts [13, 14, 15, 16]. Curve‐area methods summarize exposure over time, but their definitions differ. Total BP‐time area integrates the complete recorded curve, whereas above‐threshold AUC includes only burden accumulated above a reference value [17, 18, 19]. These measures, together with BP variability and rhythm [20, 21, 22], offer a more detailed description of the ABPM profile.

Above‐threshold AUC may provide a more detailed summary than daytime and nighttime mean SBP because it captures both the magnitude and duration of SBP above the corresponding reference values. Whether AUC provides additional information about target‐organ damage beyond daytime and nighttime mean SBP remains uncertain.

We therefore examined whether above‐threshold AUC provided information about LVMI beyond daytime and nighttime mean SBP. We also decomposed AUC to determine which component accounted for any difference.

2. Methods

2.1. Study Population

This retrospective, single‐center, cross‐sectional study included 212 Chinese adults with non‐dialysis CKD. Each patient had technically adequate ABPM, echocardiography, and complete LVMI and core covariate data.

2.2. ABPM and Echocardiography

ABPM was performed with SpaceLabs 90207 or 90217 devices (SpaceLabs Healthcare). Daytime was defined as 06:00‐21:59 and nighttime as 22:00‐05:59. Eligible recordings had at least 20 daytime readings, 7 nighttime readings, and 20 h of integrable support. We connected adjacent valid spot readings linearly at their actual timestamps. We split segments that crossed 06:00 or 22:00 at the boundary. We did not extrapolate beyond the observed recording.

LVMI was indexed to height²·⁷ and to body surface area (BSA) [23, 24, 25]. Height²·⁷‐indexed LVMI was the primary outcome, and BSA‐indexed LVMI was examined to assess consistency across indexation methods. LVH, relative wall thickness (RWT), and LV geometry were exploratory structural outcomes [26, 27]. LVH and LV geometry were defined using height²·⁷‐indexed LVMI. LVH was defined as LVMI ≥48 g/m²·⁷ in men or ≥44 g/m²·⁷ in women. RWT was calculated as twice the posterior wall thickness at end‐diastole divided by LV end‐diastolic diameter; RWT >0.42 was considered increased. LV geometry was classified as normal geometry, concentric remodeling, eccentric LVH, or concentric LVH from LVH and increased RWT status.

2.3. Components of above‐Threshold AUC

Above‐threshold AUC was the cumulative area above 135 mmHg during the day and 120 mmHg at night. We connected adjacent valid ABPM spot readings with straight lines. Only the part of each line segment above the corresponding threshold contributed to the area. Daytime and nighttime areas were time‐normalized and retained as separate period‐specific exposures. Their fixed 16‐ and 8‐hour weighted sum was used as the total 24‐hour‐equivalent AUC in descriptive and structural analyses.

We used three terms to describe the AUC components. $AU{C}_{\mathrm{total}}$ is the total above‐threshold AUC after combining daytime and nighttime values. $AU{C}_{\mathrm{mean}}$ is the AUC expected if every reading within each period equaled that period's mean SBP. $AU{C}_{\mathrm{remaining}}$ is the part of $AU{C}_{\mathrm{total}}$ not captured by $AU{C}_{\mathrm{mean}}$. Patients with similar daytime and nighttime mean SBP can therefore have different $AU{C}_{\mathrm{total}}$ values because their readings are distributed differently around the thresholds. Figure 1 illustrates these quantities with two examples. Detailed definitions are provided in Table S1. Figure S1 shows the relation between the AUC components and mean‐threshold proximity.

FIGURE 1.

Visual explanation of total above‐threshold AUC and its two components. Panels A and B show two deidentified ambulatory blood pressure monitoring (ABPM) profiles with similar daytime and nighttime mean SBP but different total above‐threshold AUC. Points are recorded ABPM measurements, and adjacent points are connected linearly. Dashed gray lines show the period‐specific thresholds, gold lines show period mean SBP, and red shading shows the complete above‐threshold burden. Panel C separates $AU{C}_{\mathrm{total}}$ into the part attributable to daytime and nighttime mean SBP ($AU{C}_{\mathrm{mean}}$) and the remaining part not captured by those means ($AU{C}_{\mathrm{remaining}}$).

2.4. Statistical Analysis

All models adjusted for age, sex, body mass index (BMI), diabetes, estimated glomerular filtration rate (eGFR), and log10‐transformed albumin‐creatinine ratio (ACR). Daytime and nighttime mean SBP were entered jointly in the mean‐SBP model. Daytime and nighttime above‐threshold AUC were entered jointly in the AUC model. The primary incremental model added $AU{C}_{\mathrm{remaining}}$ to both period means. This model directly tested whether AUC contained information beyond daytime and nighttime mean SBP.

A secondary decomposition model estimated separate coefficients for $AU{C}_{\mathrm{mean}}$ and $AU{C}_{\mathrm{remaining}}$ while retaining the night‐minus‐day SBP difference to compare the two components of AUC. Technical parameterization is provided in the Supplement.

Effects were primarily reported as differences between the 25th and 75th percentiles to facilitate comparison of their magnitudes within the cohort. Total AUC was also expressed per 100 mmHg·h for structural and sensitivity analyses.

We compared model performance using the same patient‐level folds across 100 repetitions of 10‐fold cross‐validation. Uncertainty in model‐error differences was estimated with 2,000 paired patient‐level bootstrap samples.

BSA was calculated with the Du Bois equation:

$$\mathrm{BSA}=0.007184\times {\mathrm{H}}^{0.725}\times {\mathrm{W}}^{0.425},\mathrm{LVM}{\mathrm{I}}_{\mathrm{BSA}}=\frac{\mathrm{LVM}}{\mathrm{BSA}}$$

In this equation, H denotes height in centimeters and W denotes weight in kilograms. The height²·⁷‐indexed models were also adjusted for the use of angiotensin‐converting enzyme inhibitors, angiotensin receptor blockers, calcium‐channel blockers, beta‐blockers, and diuretics. Sensitivity analyses examined earlier and later fixed day/night windows. We also tested whether the association between $AU{C}_{\mathrm{remaining}}$ and LVMI varied according to the distance of daytime mean SBP from 135 mmHg and nighttime mean SBP from 120 mmHg. The two interaction terms were assessed jointly.

Exploratory models examined RWT, LVH, and four‐category LV geometry. Normal geometry was the reference for the multinomial model. For the descriptive AUC‐quartile analysis, the four‐category geometry distribution was tested with a 4×4 Pearson chi‐square test and a 100,000‐replicate Monte Carlo sensitivity test. Linear trends across ordered AUC quartiles were tested separately for concentric LVH, overall LVH, and increased RWT, with Holm adjustment across these three tests.

Other secondary analyses examined conventional BP variability (Table S2) and alternative AUC reconstruction methods (Table S3). Reconstruction sensitivity is shown in Figure S2. Detailed methods are provided in the Supplement.

All tests were two‐sided, with emphasis on effect sizes, confidence intervals, and absolute differences. P < 0.05 was considered statistically significant.

3. Results

3.1. Cohort Characteristics

The analysis included 212 Chinese adults with non‐dialysis CKD. Mean age was 54.7 years. Men accounted for 67.5% of the cohort, and 31.1% had diabetes. Median eGFR was 36.4 mL/min/1.73 m². Overall time‐weighted SBP averaged 140.0 mmHg. Mean daytime and nighttime SBP were 142.1 and 135.8 mmHg, respectively. Mean LVMI was 46.4 g/m²·⁷, and 87 patients (41.0%) had LVH (Table 1).

TABLE 1.

Clinical and ABPM characteristics.

Characteristic	Value, mean (SD), median (IQR), or n (%)
Age, years	54.7 (14.3)
Men	143 (67.5%)
BMI, kg/m2	26.0 (4.4)
Diabetes	66 (31.1%)
eGFR, mL/min/1.73 m2	36.4 (17.1‐63.3)
ACR, mg/g	949.2 (299.4‐2581.8)
LVMI, g/m2·7	46.4 (14.4)
Relative wall thickness	0.415 (0.063)
LVH	87 (41.0%)
LV geometry: normal geometry	91 (42.9%)
LV geometry: concentric remodeling	34 (16.0%)
LV geometry: eccentric LVH	40 (18.9%)
LV geometry: concentric LVH	47 (22.2%)
Overall time‐weighted SBP, mmHg	140.0 (20.0)
Daytime mean SBP, mmHg	142.1 (19.9)
Nighttime mean SBP, mmHg	135.8 (22.7)
Daytime above‐threshold AUC, mmHg·h	127.9 (48.6–339.3)
Nighttime above‐threshold AUC, mmHg·h	123.3 (33.2–243.6)
Total 24‐hour‐equivalent above‐threshold AUC, mmHg·h	279.4 (87.9–588.5)
Time‐normalized AUC mean, mmHg	8.91 (0.49–23.59)
Time‐normalized AUC remaining, mmHg	1.25 (0.29–2.64)

Note: Values are mean (standard deviation [SD]), median (interquartile range [IQR]), or n (%). ABPM, ambulatory blood pressure monitoring; ACR, albumin‐creatinine ratio; eGFR, estimated glomerular filtration rate; LVMI, left ventricular mass index; LVH, left ventricular hypertrophy; SBP, systolic blood pressure. ACR was back‐transformed from log10 ACR for descriptive presentation.

3.2. Association of above‐Threshold AUC with LVMI

Daytime and nighttime above‐threshold AUC were entered together. From the 25th to the 75th percentile, daytime AUC was associated with a 3.46‐g/m²·⁷ higher adjusted LVMI (95% confidence interval [CI], 0.00 to 6.92), reaching borderline statistical significance (p = 0.050), whereas nighttime AUC was associated with a 5.56‐g/m²·⁷ higher adjusted LVMI (95% CI, 1.58 to 9.55) and was statistically significant (p = 0.006). Together, the two AUC terms were significantly associated with LVMI (P < 0.001), and the model's adjusted R² was 0.345. Their coefficients did not differ significantly (p = 0.321).

In the corresponding mean‐SBP model, daytime mean SBP was not significantly associated with adjusted LVMI (3.52 g/m²·⁷ across the interquartile range; 95% CI, ‐0.33 to 7.36; p = 0.073), whereas nighttime mean SBP was significantly associated with adjusted LVMI (5.47 g/m²·⁷; 95% CI, 1.31 to 9.64; p = 0.010). The model's adjusted R² was 0.334, and the two mean‐SBP coefficients did not differ significantly (p = 0.773) (Table 2, Figure 2, and Table S4).

TABLE 2.

Association of ambulatory SBP summaries with LVMI.

Model and component	Scale	Adjusted LVMI difference, g/m2·7	95% CI	Model adjusted R2
Daytime/nighttime mean SBP model: daytime mean SBP	P25 to P75	3.52	−0.33 to 7.36	0.334
Daytime/nighttime mean SBP model: nighttime mean SBP	P25 to P75	5.47	1.31 to 9.64	0.334
Daytime/nighttime AUC model: daytime AUC	P25 to P75	3.46	0.00 to 6.92	0.345
Daytime/nighttime AUC model: nighttime AUC	P25 to P75	5.56	1.58 to 9.55	0.345
Daytime/nighttime mean‐SBP model plus total AUC remaining: total AUC remaining	P25 to P75	0.69	−2.00 to 3.37	0.332

Note: All models adjusted for age, sex, BMI, diabetes, estimated glomerular filtration rate, and log10 albumin‐creatinine ratio. Daytime and nighttime terms were entered jointly within their respective models.

Abbreviations: SBP, systolic blood pressure; AUC, area under the curve; LVMI, left ventricular mass index; CI, confidence interval.

FIGURE 2.

Associations of daytime and nighttime ambulatory SBP summaries with LVMI. Points and lines show adjusted LVMI differences and 95% confidence intervals from the 25th to 75th percentile of each exposure. Daytime and nighttime terms were entered jointly within the mean‐SBP and AUC models. $AU{C}_{\mathrm{remaining}}$ was added to the model containing daytime and nighttime mean SBP. All analyses used the same 212 patients and core covariates. SBP, systolic blood pressure; AUC, area under the curve; CI, confidence interval.

Height²·⁷‐indexed and BSA‐indexed LVMI were strongly correlated (r = 0.9617). Above‐threshold AUC remained significantly associated with BSA‐indexed LVMI (3.821 g/m² per 100 mmHg·h; 95% CI, 2.657 to 4.985; P < 0.001). The association of $AU{C}_{\mathrm{remaining}}$ with BSA‐indexed LVMI beyond daytime and nighttime mean SBP was not statistically significant (1.433 g/m² per IQR; 95% CI, ‐4.404 to 7.270; p = 0.629). The change in BSA‐LVMI internal RMSE after adding $AU{C}_{\mathrm{remaining}}$ was not statistically significant (0.093 g/m²; 95% interval, ‐0.035 to 0.243) (Table S5).

3.3. Contribution of Period Mean SBP to above‐Threshold AUC

Cohort‐average $AU{C}_{\mathrm{total}}$ was 15.77 mmHg after time normalization. Of this, $AU{C}_{\mathrm{mean}}$ averaged 14.14 mmHg and $AU{C}_{\mathrm{remaining}}$ averaged 1.62 mmHg. Thus, $AU{C}_{\mathrm{mean}}$ represented about 90% of cohort‐average $AU{C}_{\mathrm{total}}$. This percentage describes the cohort average, not a fixed proportion for every patient. Median $AU{C}_{\mathrm{remaining}}$ was 1.25 mmHg (interquartile range [IQR], 0.29‐2.64). The relation among total, mean‐expected, and remaining AUC is shown in Figure S1.

3.4. Incremental Association of Remaining AUC beyond Daytime and Nighttime Mean SBP

The primary model included daytime and nighttime mean SBP directly. The corresponding LVMI difference for $AU{C}_{\mathrm{remaining}}$ was 0.686 g/m²·⁷ per interquartile range (95% CI, ‐2.001 to 3.373; p = 0.615). The association was not statistically significant (Figure 2). Adding $AU{C}_{\mathrm{remaining}}$ changed adjusted R² by ‐0.0025, indicating very little additional information (Table 2).

3.5. Secondary Comparison of AUC Components

In the secondary decomposition model, $AU{C}_{\mathrm{remaining}}$ showed a nominal association with LVMI after adjustment for $AU{C}_{\mathrm{mean}}$, the night‐minus‐day SBP difference, and the core covariates (3.014 g/m²·⁷ per interquartile range; 95% CI, 0.081 to 5.948; p = 0.044; partial R ² = 0.0199; Table S6). The coefficients for $AU{C}_{\mathrm{mean}}$ and $AU{C}_{\mathrm{remaining}}$ did not differ significantly (p = 0.167; Table S7).

3.6. Internal Model Performance

The joint daytime/nighttime AUC model had a repeated internal cross‐validation root mean square error (RMSE) of 12.031 g/m²·⁷, compared with 12.087 g/m²·⁷ for the joint period‐mean model. The difference was ‐0.056 g/m²·⁷ (95% paired‐bootstrap interval, ‐0.284 to 0.154), or ‐0.46% of the reference error.

Adding $AU{C}_{\mathrm{remaining}}$ to the period‐mean model changed RMSE by 0.048 g/m²·⁷ (95% interval, ‐0.017 to 0.121), or 0.40%. Comparing $AU{C}_{\mathrm{mean}}+AU{C}_{\mathrm{remaining}}$ with $AU{C}_{\mathrm{mean}}$ alone changed RMSE by ‐0.074 g/m²·⁷ (95% interval, ‐0.283 to 0.153), or ‐0.61%. None of these internal performance differences was statistically significant (Table 3, Figure 3, and Table S9).

TABLE 3.

Internal performance of daytime/nighttime mean‐SBP and AUC‐component models.

Model	Adjusted R2	Repeated internal cross‐validation RMSE	RMSE difference	Bootstrap 95% interval	Interpretation
Daytime/nighttime mean SBP model	0.334	12.087	Reference	—	Clinical comparator
Daytime/nighttime AUC model	0.345	12.031	−0.056	−0.284 to 0.154	−0.46%; difference not statistically significant
Daytime/nighttime mean‐SBP model plus total AUC remaining	0.332	12.135	0.048	−0.017 to 0.121	+0.40%; improvement not statistically significant
AUC mean model	0.337	12.055	Reference	—	Reference for secondary AUC‐component comparison
Unrestricted AUC mean plus AUC remaining	0.347	11.982	−0.074	−0.283 to 0.153	−0.61%; difference not statistically significant

Note: RMSE, root mean square error, is in g/m²·⁷. AUC, area under the curve; SBP, systolic blood pressure. Differences are first model minus the stated reference; negative values favor the first model. The secondary AUC‐component models also included the night‐minus‐day SBP difference. Repeated internal cross‐validation evaluates performance within this cohort and does not establish external validity or clinical prediction utility.

FIGURE 3.

Internal model performance across the primary and secondary model comparisons. Points and lines show paired differences in LVMI error from repeated internal cross‐validation, with 95% paired‐bootstrap intervals. Negative values favor the first model named. The comparisons evaluate the joint daytime/nighttime AUC model against the joint daytime/nighttime mean‐SBP model, the mean‐SBP model plus $AU{C}_{\mathrm{remaining}}$ against the mean‐SBP model alone, and $AU{C}_{\mathrm{mean}}+AU{C}_{\mathrm{remaining}}$ against $AU{C}_{\mathrm{mean}}$ alone. The last comparison evaluates the two components within AUC and is not a test beyond the complete period means. These comparisons assess internal performance and do not establish external validity. AUC, area under the curve; SBP, systolic blood pressure; LVMI, left ventricular mass index.

3.7. LV Geometry and Related Structural Findings

Normal geometry was present in 91 patients (42.9%), concentric remodeling in 34 (16.0%), eccentric LVH in 40 (18.9%), and concentric LVH in 47 (22.2%). Across AUC quartiles, the prevalence of normal geometry declined from 66.0% in Q1 to 30.2% in Q4. Concentric LVH was present in 7, 3, 15, and 22 patients from Q1 to Q4. The four‐category distribution differed significantly across quartiles (Pearson chi‐square = 45.86, 9 df; P < 0.001; Monte Carlo P < 0.001). The trend in concentric LVH was statistically significant (P < 0.001; Holm‐adjusted P < 0.001), although the Q2 decrease meant that the proportion was not strictly monotonic. Trends were also statistically significant for overall LVH (P < 0.001) and increased RWT (p = 0.00445) (Figure 4A and Table S10).

FIGURE 4.

Left ventricular geometry distribution and adjusted associations with AUC. Panel A shows normal geometry, concentric remodeling, eccentric LVH, and concentric LVH across AUC quartiles; each quartile included 53 patients. The overall four‐category distribution differed significantly across quartiles (Pearson P < 0.001). The concentric LVH trend was P < 0.001, although its proportion was not strictly monotonic because Q2 was lower than Q1. Panel B shows adjusted category‐specific odds ratios and 95% confidence intervals per 100 mmHg·h, with normal geometry as the reference. The four‐category geometry outcome is nominal; therefore, the adjusted multinomial model, rather than an ordered‐category trend, was used for inference. These analyses were exploratory. AUC, area under the curve; LVH, left ventricular hypertrophy; OR, odds ratio.

Above‐threshold AUC was significantly associated with the four‐category geometry outcome (likelihood‐ratio p = 0.0039). Each 100‐mmHg·h increase in AUC was associated with higher odds of concentric LVH versus normal geometry (adjusted odds ratio [OR], 1.225; 95% CI, 1.074 to 1.397; p = 0.002). Associations with concentric remodeling (OR, 0.955; 95% CI, 0.802 to 1.138; p = 0.610) and eccentric LVH (OR, 1.095; 95% CI, 0.946 to 1.268; p = 0.224) versus normal geometry were not statistically significant (Figure 4B and Table 4). Higher AUC was also significantly associated with greater RWT (difference, 0.0042 per 100 mmHg·h; 95% CI, 0.0013 to 0.0070; p = 0.004) and higher odds of LVH (OR, 1.181; 95% CI, 1.059 to 1.316; p = 0.003).

TABLE 4.

Association of above‐threshold AUC with LV geometry and related structural outcomes.

Outcome or contrast	Model	n	Events	Effect	Estimate per 100 mmHg·h	95% CI	p value
Relative wall thickness	Linear regression	212	—	Difference	0.0042	0.0013 to 0.0070	0.004
LVH	Logistic regression	212	87	OR	1.181	1.059 to 1.316	0.003
Concentric remodeling vs normal geometry	Multinomial logistic	212	34	OR	0.955	0.802 to 1.138	0.610
Eccentric LVH vs normal geometry	Multinomial logistic	212	40	OR	1.095	0.946 to 1.268	0.224
Concentric LVH vs normal geometry	Multinomial logistic	212	47	OR	1.225	1.074 to 1.397	0.002

Note: All models adjusted for age, sex, BMI, diabetes, estimated glomerular filtration rate, and log10 albumin‐creatinine ratio.

AUC, area under the curve; LVH, left ventricular hypertrophy; RWT, relative wall thickness; OR, odds ratio; CI, confidence interval. AUC was expressed per 100 mmHg·h. The four‐category multinomial model used normal geometry as the reference; its global likelihood‐ratio p value was 0.0039. Across AUC quartiles, the four‐category Pearson p value was <0.001; trend p values were <0.001 for concentric LVH and overall LVH and 0.004 for increased RWT. Category‐specific multinomial p values are exploratory and were not adjusted for multiple comparisons; quartile trend tests were Holm‐adjusted in Table S10.

3.8. Medication‐Adjusted and Additional Analyses

After adjustment for antihypertensive medication classes, above‐threshold AUC remained significantly associated with LVMI (1.664 g/m²·⁷ per 100 mmHg·h; 95% CI, 1.106 to 2.221; P < 0.001). The coefficient was 3.7% lower than in the core model. The primary $AU{C}_{\mathrm{remaining}}$ association was not statistically significant (0.672 g/m²·⁷ per IQR; 95% CI, ‐2.129 to 3.473; p = 0.636) (Table S11). Neither within‐period standard deviation nor average real variability was significantly associated with LVMI after adjustment for daytime and nighttime mean SBP, and the remaining AUC component was not statistically significant in the corresponding joint models (Table S2). We tested whether the association of $AU{C}_{\mathrm{remaining}}$ with LVMI changed as daytime mean SBP moved above or below 135 mmHg and nighttime mean SBP moved above or below 120 mmHg. The joint test of the two interaction terms was not statistically significant (p = 0.389; Table S8 and Figure S3). Thus, the data did not identify a mean‐SBP range in which $AU{C}_{\mathrm{remaining}}$ provided clearer additional information. Together, daytime and nighttime AUC were significantly associated with LVMI, but their coefficients did not differ significantly (Table S7). Under the early, primary, and late fixed‐clock windows, $AU{C}_{\mathrm{remaining}}$ was not statistically significant beyond the corresponding period means. Adding it did not improve repeated internal cross‐validation in any window (Table S8). Results were also consistent across alternative AUC reconstructions (Table S3 and Figure S2).

4. Discussion

Above‐threshold AUC identified patients with higher LVMI in this non‐dialysis CKD cohort. However, $AU{C}_{\mathrm{mean}}$ represented about 90% of cohort‐average $AU{C}_{\mathrm{total}}$. $AU{C}_{\mathrm{remaining}}$ was not significantly associated with LVMI beyond the complete period‐mean profile. The secondary decomposition model showed a nominal association, but it did not significantly improve internal model performance.

Above‐threshold AUC combines the magnitude and duration of pressure above clinical reference values. Blood pressure risk is continuous, so treatment targets and diagnostic reference values do not mark levels below which pressure is totally harmless [28]. Nevertheless, these values identify clinically actionable pressure ranges. In this study, AUC was calculated using period‐specific ABPM reference values, thereby giving greater weight to pressure above those values; the reference values are supported by outcome‐based ambulatory threshold studies [19, 29]. Therefore, above‐threshold AUC not only provides a richer description of the ambulatory BP profile but may also offer a way to quantify the clinically relevant “damage burden” of hypertension.

We used two models to test whether this more detailed ABPM summary provided information about target‐organ damage beyond daytime and nighttime mean SBP. The primary model retained the complete daytime and nighttime mean‐SBP profile and tested whether $AU{C}_{\mathrm{remaining}}$ added information beyond those means. The secondary model instead compared $AU{C}_{\mathrm{mean}}$ with $AU{C}_{\mathrm{remaining}}$ within the total AUC. Neither model showed that AUC added information beyond conventional summaries based on period means.

Despite incorporating additional detail, the joint daytime/nighttime AUC model reduced the repeated internal cross‐validation errorby only 0.056 g/m²·⁷ compared with the joint period‐mean model. This difference was less than 1% of the overall model error and was not statistically significant.

These findings help explain why previous studies have reported associations between BP load and adverse outcomes while reaching different conclusions about added clinical value. Wang et al. found that nighttime percentage load added a small amount of explained variation beyond 24‐hour BP in selected adults with nondiabetic CKD [13]. Li et al. reported longitudinal associations between nighttime percentage load and outcomes in nondialysis CKD [14]. By contrast, percentage load added little beyond ambulatory BP level in an international population cohort and beyond mean BP in pediatric CKD [15, 16]. Zhou et al. linked cumulative BP‐time area with LVH, but their measure integrated the full recorded curve rather than only the area above a reference threshold [18]. Definitions, thresholds, sampling, outcomes, and adjustment for mean BP differ across these studies. In this study, we directly compared above‐threshold AUC with simpler daytime and nighttime mean SBP summaries and used the decomposition to identify the source of any difference.

We also tested whether the association of $AU{C}_{\mathrm{remaining}}$ with LVMI changed according to how far daytime mean SBP was above or below 135 mmHg and nighttime mean SBP was above or below 120 mmHg. The joint test of the two interaction terms was not statistically significant (p = 0.389). Thus, this cohort did not identify a mean‐SBP range in which $AU{C}_{\mathrm{remaining}}$ provided clearer additional information.

We also explored LV structural patterns. Sustained afterload can favor increased wall thickness and LV mass, whereas eccentric LVH may also reflect volume burden, anemia, chamber dilation, and other CKD‐related influences [12]. Concentric remodeling reflects increased RWT without increased LV mass and represents a distinct structural pattern. As expected, above‐threshold AUC, as a marker of increased afterload, was associated with concentric LVH. Above‐threshold AUC was also significantly associated with both height²·⁷‐indexed and BSA‐indexed LVMI. Because the study was cross‐sectional, LVMI remains a structural parameter rather than a validated substitute for clinical outcomes [9, 10, 11, 23, 24, 25, 30].

Overall, this study did not establish clinically meaningful incremental information from AUC beyond daytime and nighttime mean SBP. For clinical ABPM interpretation, daytime and nighttime mean SBP should remain the principal summaries. The present data do not support the superiority of AUC as an additional assessment metric. Larger studies with repeated ABPM and interventional follow‐up may examine which patients benefit from AUC monitoring and whether changes in AUC track changes in cardiac structure after mean SBP has been optimized.

Some limitations should be noted. The study was retrospective, cross‐sectional, and conducted at one tertiary center. External validation is warranted. Referral for ABPM may have selected patients with more difficult hypertension. Individual sleep and wake times were unavailable. Future studies in external populations using repeated ABPM, individualized sleep‐wake periods, and longitudinal outcomes are warranted.

5. Conclusions

Above‐threshold AUC identified a higher‐pressure ambulatory phenotype associated with greater LVMI. Most cohort‐average $AU{C}_{\mathrm{total}}$ was represented by $AU{C}_{\mathrm{mean}}$, and $AU{C}_{\mathrm{remaining}}$ was not significantly associated with LVMI beyond daytime and nighttime mean SBP. Daytime and nighttime mean SBP should therefore remain the principal clinical ABPM summaries.

Authors Contributions

Zhen Zhang and Xinyue Dong contributed equally. Zhen Zhang, Wuhua Jiang, and Yi Fang conceived and designed the study. Zhen Zhang and Xinyue Dong curated the ABPM data, performed the statistical analyses, prepared the figures, and drafted the manuscript. Xiaoqiang Ding contributed to supervision of the study. Wuhua Jiang and Yi Fang supervised the study and critically revised the manuscript. Yi Fang acquired funding. All authors approved the final manuscript.

Funding

This work was supported by the National Key R&D Program of China (grant no. 2024YFC3607400) and the Science and Technology Commission of Shanghai Municipality (grant no. 25Y12800100).

Ethics approval

The study was approved by the Ethics Committee of Zhongshan Hospital, Fudan University (approval no. B2018‐269).

Consent

All patients provided written informed consent.

Conflicts of Interest

The authors declare no conflicts of interest.

Permission to reproduce material

No third‐party material was reproduced.

Clinical Trial Registration

Not applicable; this was a retrospective observational study.

Supporting information

Supporting Information: jch70378‐sup‐0001‐SuppMat.pdf

Data Availability Statement

Deidentified analytic data may be available from the corresponding authors on reasonable request, subject to institutional approval.