Abstract
Purpose
To construct and preliminarily validate a novel nomogram for primary aldosteronism (PA) risk estimation, with the goal of providing a potential reference tool for PA screening in clinical practice.
Methods
Hypertensive inpatients who underwent PA screening at the Department of Endocrinology and Metabolism, Cardiology, and Urology of Tianjin Medical University General Hospital from October 2021 to July 2024 were included. Basic information and laboratory test results of the patients were collected, and were randomly divided into training set and validation set at a 7:3 ratio. Differential analysis, Lasso regression, univariate logistic regression, and multivariate logistic regression were used to screen the variables associated with PA. The model results were presented as a nomogram. Predictive ability was assessed by plotting receiver operating characteristic (ROC) curves, and the diagnostic performance of the model was compared using the DeLong test. The Hosmer-Lemeshow test was used to evaluate the goodness of fit of the regression model. The clinical utility of the model was evaluated by clinical decision curve analysis (DCA). Kappa test was used to evaluate the agreement between the model and ARR.
Results
Six indicators including serum sodium-to-potassium ratio, anion gap, uric acid, urine sodium, 24-hour urine potassium and urine pH were used to construct a nomogram prediction model for PA. The training and validation sets achieved area under the curve (AUC) values of 0.860 (95% CI: 0.825–0.894) and 0.850 (95% CI: 0.800–0.899), respectively. The AUCs of the validation set for PFK score and ARR were 0.718 (95% CI: 0.656–0.781) and 0.965 (95% CI: 0.943–0.987), respectively. The model demonstrated good discrimination ability. The Hosmer-Lemeshow test indicated a good fit (training set: χ² = 7.897, P = 0.444;validation set: χ² = 6.208, P = 0.624). The DCA showed that the model had good clinical application value. The Kappa test showed good agreement between the model and ARR(Kappa = 0.364, P < 0.001).
Conclusion
The nomogram model based on serum sodium-to-potassium ratio, anion gap, uric acid, urine sodium, 24-hour urine potassium and urine pH can predict the risk of PA, facilitating PA screening in clinical practice.
Keywords: essential hypertension, nomogram, prediction model, primary aldosteronism, secondary hypertension
1. Introduction
Primary aldosteronism (PA) is one of the common causes of secondary hypertension. It was first described by American physician Jerome W. Conn in 1955, who reported a case of hypertension caused by an aldosterone-producing adrenal adenoma (1, 2). Among the hypertensive population, the prevalence of PA ranges from approximately 10% to 20%. Specifically, the prevalence rates of PA among individuals with high-normal blood pressure, Stage 1, Stage 2, and Stage 3 hypertension were 5.5%, 4.2%, 10.2%, and 16.4%, respectively. Among hypertensive patients in primary care, the estimated prevalence of PA ranges from 3.2% to 12.7% (3). Studies have shown that, compared to patients with essential hypertension (EH), those with PA exhibit significantly increased risks of atrial fibrillation, stroke, heart failure, and coronary artery disease (4). Unlike EH, however, PA has identifiable causes, and most patients can be cured through surgery or drug therapy. Therefore, early screening and diagnosis of PA are particularly important. Indeed, the Mayo Clinic recommends that all hypertensive patients undergo at least one PA screening (5). Despite this recommendation, PA screening rates remain low both in China and in other countries. A survey conducted in China revealed that only 0.1% of primary care hospitals conduct PA screening (6). International studies report that only 7%–8% of hypertensive patients have undergone PA screening. Even among those with hypokalemia, only 2.7% have been screened for PA (7, 8). The 2025 American Endocrine Society Clinical Practice Guidelines for PA recommend screening all hypertensive patients to prevent PA from being misdiagnosed as EH and to avoid missing targeted treatment opportunities (9).The aldosterone-to-renin ratio (ARR), defined as the ratio of plasma aldosterone concentration (PAC) to direct renin concentration (DRC), is currently the internationally recognized screening indicator for PA (10, 11). Our previous research established the optimal ARR cutoff point for PA screening in Tianjin, China (12). However, the use of ARR faces several practical limitations. First, patients must discontinue medications that affect ARR—such as aldosterone receptor antagonists, potassium-sparing diuretics, and angiotensin-converting enzyme inhibitors—prior to PA testing, which is significantly inconvenient for patients. Second, since aldosterone and renin testing predominantly relies on chemiluminescence methods that require specialized equipment, most primary care hospitals lack the necessary testing capabilities (6). Collectively, these factors limit the widespread adoption of ARR. Consequently, there is an urgent clinical need for a new, easily implementable PA screening tool.
To address this need, the present study developed a nomogram-based prediction model for PA. Specifically, we analyzed relevant serum and urine indicators in patients with EH and PA. Using Lasso regression and logistic regression methods, we established a PA prediction nomogram based on serum sodium-to-potassium ratio, anion gap, uric acid, urine sodium, 24-hour urine potassium, and urine pH. This model may serve as a foundation for developing more user-friendly tools for primary care settings in the future.
2. Materials and methods
2.1. Patient population
Hypertensive inpatients who underwent PA screening at the Department of Endocrinology and Metabolism, Cardiology, and Urology of Tianjin Medical University General Hospital from October 2021 to July 2024 were retrospectively enrolled. Patients were included if they had the following parameters measured: PAC, DRC, ARR, serum sodium, serum potassium, serum chloride, serum calcium, anion gap, total cholesterol, triglycerides, blood glucose, urea, creatinine, uric acid, urine sodium, urine potassium, and urine pH. The screening criteria for PA were based on the Expert Consensus on Diagnosis and Treatment of Primary Aldosteronism (2024 Edition) (13). Main criteria included: 1) Patients with persistent hypertension (BP > 150/100 mmHg, 1 mmHg = 0.133 kPa); patients whose blood pressure remained uncontrolled (≥ 140/90 mmHg) despite treatment with three conventional antihypertensive agents (including diuretics); patients requiring ≥ 4 antihypertensive agents to achieve control (< 140/90 mmHg); and newly diagnosed hypertensive patients; 2) Hypertensive patients with spontaneous or diuretic-induced hypokalemia; 3) Hypertensive patients with incidentally discovered adrenal tumors; 4) Hypertensive patients with a family history of early-onset hypertension or early-onset (age < 40 years) cerebrovascular accident; 5) Hypertensive first-degree relatives of patients with PA; 6) Hypertensive patients with obstructive sleep apnea. Exclusion criteria were: other types of secondary hypertension (secondary hyperaldosteronism, renal artery stenosis, Cushing’s syndrome, pheochromocytoma,etc.), stage IV heart failure, severe organic heart disease, severe arrhythmia, severe hepatic or renal impairment, pregnancy or lactation.
2.2. Data collection
Patient demographics were collected, including gender, age, body mass index, blood pressure, presence of diabetes, and hypokalemia.The initial laboratory results obtained upon admission included: PAC, DRC, ARR (LIAISON®XL fully automated chemiluminescent immunoassay analyzer, DiaSorin Italia S.p.A.); serum sodium, potassium, chloride, calcium, anion gap, total cholesterol, triglycerides, glucose, urea, creatinine, and uric acid (008AS fully automated biochemical analyzer, Hitachi High-Tech Corporation); urine sodium concentration, 24-hour urine sodium, urine potassium concentration, and 24-hour urine potassium (7180 fully automated biochemical analyzer, Hitachi High-Tech Corporation); urine pH (UF-1000i fully automated urine analyzer, Sysmex Corporation).
2.3. ARR screening
ARR screening was conducted according to the following standardized protocol:
1) Sampling conditions: Blood samples were collected in the morning (between 8:00 AM and 10:00 AM) after patients had been ambulatory (sitting, standing, or walking) for at least 2 hours, followed by 5–15 minutes of seated rest before collection. All samples were drawn with the patient in a seated position. 2) Potassium correction: Serum potassium levels were corrected to the normal range (≥ 3.5 mmol/L) before ARR measurement, as hypokalemia can suppress renin and falsely elevate ARR. 3) Assay methods and ARR cutoff: PAC and DRC were measured using chemiluminescent immunoassay. The ARR was calculated as PAC/DRC. An ARR ≥ 3.70 with PAC > 15 ng/dL was considered screening-positive. 4) Medication adjustment: To minimize the influence of antihypertensive medications on ARR results, patients were instructed to discontinue medications that significantly affect the renin-angiotensin-aldosterone system when clinically feasible. Specifically, mineralocorticoid receptor antagonists (e.g., spironolactone, eplerenone), potassium-sparing diuretics, potassium-wasting diuretics, and licorice extract were withdrawn for at least 4 weeks prior to testing. Angiotensin-converting enzyme inhibitors (ACEIs), angiotensin receptor blockers (ARBs), and calcium channel blockers (CCBs) were withdrawn for at least 2 weeks prior to testing. For patients in whom complete withdrawal was not feasible, antihypertensive therapy was switched to agents with minimal effects on ARR, such as slow-release verapamil, hydralazine, prazosin, or doxazosin (11).
2.4. PA and EH diagnostic criteria
The diagnostic criteria for PA were as follows. Patients with a positive ARR underwent confirmatory testing using either a captopril suppression test or a saline loading test. PA was confirmed if the captopril suppression test showed an aldosterone suppression rate of ≤ 30% at 2 hours or a PAC > 11 ng/dL, or if the saline loading test showed a PAC > 10 ng/dL. Patients with concomitant spontaneous hypokalemia, plasma renin levels below the detectable limit, and aldosterone levels > 20 ng/dL were directly diagnosed with PA without additional confirmatory tests (11). EH was diagnosed by excluding the following secondary causes of hypertension: renal parenchymal hypertension (e.g., IgA nephropathy, membranoproliferative glomerulonephritis, polycystic kidney disease, chronic pyelonephritis, obstructive nephropathy, lupus nephritis), renal artery stenosis, aortic stenosis, PA, pheochromocytoma/paraganglioma, Cushing’s syndrome, Liddle syndrome, Gordon syndrome, apparent mineralocorticoid excess syndrome, and other secondary causes.
2.5. Statistical analysis
Statistical analyses were performed using the R software package (R version 4.3.3). Missing values were imputed using the PMM method. Clinical baseline data were analyzed according to data type. Normally distributed continuous variables were expressed as mean ± standard deviation and were compared between groups using the independent samples t-test. Non-normally distributed continuous variables were expressed as median (interquartile range) and compared between groups using the Mann-Whitney U test. Categorical variables were expressed as frequencies and percentages, with intergroup comparisons performed using the χ² test. Variables associated with PA were identified using Lasso regression, univariate logistic regression, and multivariate logistic regression. Statistical significance was set at P < 0.05. The model results were presented as a nomogram. Receiver operating characteristic (ROC) curves were plotted to evaluate the predictive ability of the model. The Hosmer-Lemeshow test was used to assess the goodness of fit of the regression model, with P > 0.05 indicating adequate fit. DCA was used to evaluate the clinical utility of the model. The Kappa statistic was used to measure the agreement between the model and ARR screening results for PA, with P < 0.05 considered statistically significant.
3. Results
3.1. Basic clinical characteristics of study subjects
A total of 1601 patients were included in this study, comprising 1349 cases of EH and 252 cases of PA. These patients were randomly assigned to a training set (1120 cases) and a validation set (481 cases) at a ratio of 7:3. No statistically significant differences were observed between the training and validation sets in terms of serum sodium-to-potassium ratio, serum chloride, serum calcium, anion gap, total cholesterol, triglycerides, blood glucose, urea, creatinine, uric acid, or other indicators (P > 0.05), indicating that the sample populations were balanced and comparable (Table 1).
Table 1.
Baseline characteristics of training and validation sets.
| Variables | Training set (n = 1120) |
Validation set (n = 481) |
t/Z/χ² | P |
|---|---|---|---|---|
| Age (years) | 53.00 (40.00, 64.00) | 53.00 (40.00, 63.00) | -0.697 | 0.486 |
| Gender [n (%)] | ||||
| Male | 578 (51.61) | 248 (51.56) | <0.001 | 0.986 |
| Female | 542 (48.39) | 233 (48.44) | ||
| BMI (kg/m2) | 26.52 (24.06, 29.75) | 26.42 (24.15, 29.41) | -0.155 | 0.877 |
| SBP (mmHg) | 143.00 (130.00, 155.00) | 143.00 (130.00, 156.00) | -0.126 | 0.900 |
| DBP (mmHg) | 89.00 (80.00, 100.00) | 89.00 (80.00, 99.00) | -0.402 | 0.688 |
| PAC (ng/dL) | 13.20 (8.40, 19.62) | 12.50 (8.60, 20.20) | -0.238 | 0.812 |
| DRC (μ IU/mL) | 16.90 (5.80, 46.08) | 17.60 (5.10, 47.90) | -0.122 | 0.903 |
| ARR [(ng/dL)/(μ IU/mL)] | 0.68 (0.27, 2.28) | 0.64 (0.24, 2.36) | -0.233 | 0.815 |
| Serum Na (mmol/L) | 142.00 (141.00, 144.00) | 142.00 (141.00, 143.00) | -0.986 | 0.324 |
| Serum K (mmol/L) | 3.80 (3.50, 4.10) | 3.80 (3.50, 4.00) | -0.867 | 0.386 |
| Serum Na-to-K ratio | 37.37 (34.88, 40.86) | 37.63 (35.00, 40.57) | -0.660 | 0.509 |
| Serum Cl (mmol/L) | 105.00 (104.00, 107.00) | 105.00 (103.00, 107.00) | -1.118 | 0.263 |
| Serum Ca (mmol/L) | 2.30 (2.22, 2.36) | 2.30 (2.22, 2.37) | -0.122 | 0.903 |
| AG | 13.90 (12.30, 15.50) | 14.00 (12.50, 15.60) | -0.923 | 0.356 |
| TC (mmol/L) | 4.68 (4.03, 5.41) | 4.80 (4.09, 5.30) | -0.763 | 0.445 |
| TG (mmol/L) | 1.64 (1.14, 2.30) | 1.65 (1.23, 2.49) | -1.698 | 0.089 |
| GLU (mmol/L) | 5.00 (4.50, 5.80) | 5.10 (4.60, 5.80) | -1.210 | 0.226 |
| UREA (mmol/L) | 4.80 (4.10, 5.70) | 4.80 (4.00, 5.70) | -0.803 | 0.422 |
| CREA (μmol/L) | 64.00 (54.00, 75.00) | 63.00 (52.00, 76.00) | -0.692 | 0.489 |
| UA (μmol/L) | 344.00 (282.75, 421.00) | 347.00 (279.00, 428.00) | -0.561 | 0.575 |
| Urine Na (mmol/L) | 80.05 (53.58, 110.38) | 78.90 (51.90, 109.70) | -0.497 | 0.619 |
| 24h Urine Na (mmol/24h) | 151.55 (103.19, 215.26) | 156.90 (112.70, 227.70) | -1.743 | 0.081 |
| Urine K (mmol/L) | 20.09 (15.03, 28.74) | 19.76 (14.41, 27.82) | -0.866 | 0.386 |
| 24h Urine K (mmol/24h) | 39.66 (29.55, 51.63) | 41.27 (31.40, 54.50) | -1.864 | 0.062 |
| Urine pH | 6.00 (5.50, 6.50) | 6.00 (5.50, 6.50) | -1.021 | 0.307 |
| Diabetes [n (%)] | ||||
| Yes | 251 (22.41) | 113 (23.49) | 0.224 | 0.636 |
| No | 869 (77.59) | 368 (76.51) | ||
| Hypokalemia [n (%)] | ||||
| Yes | 175 (15.62) | 80 (16.63) | 0.255 | 0.614 |
| No | 945 (84.38) | 401 (83.37) | ||
| Diagnosis [n (%)] | ||||
| PA | 174 (15.54) | 78 (16.22) | 0.117 | 0.732 |
| EH | 946 (84.46) | 403 (83.78) | ||
3.2. Screening of predictors for the PA nomogram prediction model
Differential analysis was performed on the following variables in the training set: age, gender, BMI, blood pressure, serum sodium, serum potassium, serum sodium-to-potassium ratio, serum calcium, anion gap, total cholesterol, triglycerides, blood glucose, urea, creatinine, and uric acid. The results showed that systolic blood pressure, diastolic blood pressure, serum sodium, serum sodium-to-potassium ratio, urine potassium, 24-hour urine potassium and urine pH were significantly higher in the PA group than in the EH group (P < 0.05). Additionally, the proportion of patients with hypokalemia in the PA group was higher than in the EH group (P < 0.05). Conversely, the PA group exhibited lower levels of BMI, serum potassium, serum calcium, anion gap, total cholesterol, triglycerides, blood glucose, urea, uric acid, and urine sodium compared with the EH group (P < 0.05) (Table 2). These statistically significant indicators were included in the Lasso regression, and six indicators including serum sodium-to-potassium ratio, anion gap, uric acid, urine sodium, 24-hour urine potassium, and urine pH were screened as potential predictors of the PA prediction model (Figure 1).
Table 2.
Differential analysis of the training set.
| Variables | PA (n = 174) | EH (n = 946) | t/Z/χ² | P |
|---|---|---|---|---|
| Age (years) | 54.00 (44.25, 63.00) | 52.00 (39.00, 65.00) | -1.447 | 0.148 |
| Gender [n (%)] | ||||
| Male | 86 (49.43) | 492 (52.01) | 0.393 | 0.531 |
| Female | 88 (50.57) | 454 (47.99) | ||
| BMI (kg/m2) | 25.78 (23.66, 27.77) | 26.77 (24.11, 30.02) | -2.999 | 0.003 |
| SBP (mmHg) | 147.00 (137.00, 158.50) | 142.00 (130.00, 154.75) | -3.415 | <0.001 |
| DBP (mmHg) | 93.00 (84.00, 101.75) | 89.00 (80.00, 99.00) | -3.560 | <0.001 |
| Serum Na (mmol/L) | 143.00 (142.00, 145.00) | 142.00 (141.00, 143.00) | -5.929 | <0.001 |
| Serum K (mmol/L) | 3.30 (3.00, 3.60) | 3.90 (3.60, 4.10) | -12.568 | <0.001 |
| Serum Na-to-K ratio | 43.33 (38.98, 48.00) | 36.67 (34.52, 39.44) | -12.625 | <0.001 |
| Serum Cl (mmol/L) | 106.00 (104.00, 107.00) | 105.00 (104.00, 107.00) | -1.029 | 0.304 |
| Serum Ca (mmol/L) | 2.26 (2.20, 2.31) | 2.30 (2.23, 2.37) | -5.680 | <0.001 |
| AG | 13.00 (11.62, 14.57) | 14.10 (12.60, 15.60) | -4.859 | <0.001 |
| TC (mmol/L) | 4.33 (3.70, 4.94) | 4.73 (4.09, 5.49) | -5.230 | <0.001 |
| TG (mmol/L) | 1.39 (0.99, 2.09) | 1.68 (1.18, 2.32) | -3.566 | <0.001 |
| GLU (mmol/L) | 4.80 (4.50, 5.40) | 5.00 (4.50, 5.80) | -2.293 | 0.022 |
| UREA (mmol/L) | 4.40 (3.73, 5.40) | 4.80 (4.10, 5.80) | -3.747 | <0.001 |
| CREA (μmol/L) | 65.00 (55.00, 76.00) | 64.00 (54.00, 75.00) | -0.777 | 0.437 |
| UA (μmol/L) | 319.00 (259.25, 374.75) | 348.00 (287.25, 427.75) | -4.155 | <0.001 |
| Urine Na (mmol/L) | 71.45 (49.15, 102.30) | 82.45 (54.73, 113.68) | -2.915 | 0.004 |
| 24h Urine Na (mmol/24h) | 160.13 (106.07, 232.25) | 149.97 (102.67, 212.86) | -1.590 | 0.112 |
| Urine K (mmol/L) | 23.59 (17.28, 32.31) | 19.55 (14.67, 27.57) | -4.717 | <0.001 |
| 24h Urine K (mmol/24h) | 55.51 (39.52, 73.88) | 38.19 (28.69, 48.30) | -9.658 | <0.001 |
| Urine pH | 6.50 (6.00, 7.00) | 6.00 (5.50, 6.50) | -8.635 | <0.001 |
| Diabetes [n (%)] | ||||
| Yes | 39 (22.41) | 212 (22.41) | <0.001 | 0.999 |
| No | 135 (77.59) | 734 (77.59) | ||
| Hypokalemia [n (%)] | ||||
| Yes | 60 (34.48) | 115 (12.16) | 55.568 | <0.001 |
| No | 114 (65.52) | 831 (87.84) | ||
Figure 1.
Lasso regression for the training set. (A) LASSO regression cross-validation. (B) LASSO regression path diagram.
Univariate logistic regression and stepwise backward multivariate logistic regression were performed on these six indicators. Ultimately, the same six indicators—serum sodium-to-potassium ratio, anion gap, uric acid, urine sodium, 24-hour urine potassium, and urine pH—were selected for constructing the final PA prediction model (Tables 3, 4).
Table 3.
Univariate logistic regression for the training set.
| Variables | Univariate logistic regression | ||||
|---|---|---|---|---|---|
| β | S.E | Z | P | OR (95%CI) | |
| Serum Na-to-K ratio | 0.239 | 0.019 | 12.611 | <0.001 | 1.271 (1.224 ~ 1.319) |
| AG | -0.160 | 0.033 | -4.876 | <0.001 | 0.852 (0.799 ~ 0.909) |
| UA (μmol/L) | -0.004 | 0.001 | -4.309 | <0.001 | 0.996 (0.994 ~ 0.998) |
| Urine Na (mmol/L) | -0.007 | 0.002 | -3.144 | 0.002 | 0.993 (0.989 ~ 0.997) |
| 24h Urine K (mmol/24h) | 0.033 | 0.004 | 8.473 | <0.001 | 1.033 (1.025 ~ 1.041) |
| Urine pH | 0.998 | 0.116 | 8.599 | <0.001 | 2.712 (2.160 ~ 3.405) |
Table 4.
Multivariate logistic regression for the training set.
| Variables | Multivariate logistic regression | ||||
|---|---|---|---|---|---|
| β | S.E | Z | P | OR (95%CI) | |
| Serum Na-to-K ratio | 0.218 | 0.021 | 10.363 | <0.001 | 1.243 (1.193 ~ 1.295) |
| AG | -0.098 | 0.042 | -2.361 | 0.018 | 0.906 (0.835 ~ 0.983) |
| UA (μmol/L) | -0.003 | 0.001 | -2.454 | 0.014 | 0.997 (0.995 ~ 0.999) |
| Urine Na (mmol/L) | -0.009 | 0.003 | -3.485 | <0.001 | 0.991 (0.985 ~ 0.996) |
| 24h Urine K (mmol/24h) | 0.024 | 0.004 | 5.392 | <0.001 | 1.024 (1.015 ~ 1.033) |
| Urine pH | 0.406 | 0.142 | 2.866 | 0.004 | 1.501 (1.137 ~ 1.983) |
3.3. Construction and evaluation of the PA nomogram prediction model
The prediction model was visualized as a nomogram (Figure 2). The predictive performance of the model was assessed using ROC curves (Figures 3A, D). The AUC for the training set was 0.860 (95% CI: 0.825–0.894), with a sensitivity of 0.736 and a specificity of 0.876. The validation set achieved an AUC of 0.850 (95% CI: 0.800–0.899), sensitivity of 0.795, and specificity of 0.787. For the validation set, the AUC values of the PFK score and ARR were 0.718 (95% CI: 0.656–0.781) and 0.965 (95% CI: 0.943–0.987), respectively, with corresponding sensitivities of 0.846 and 0.936, and specificities of 0.469 and 0.923 (Figure 4). Notably, our predictive model showed significantly better diagnostic performance than the PFK score, as determined by the DeLong test (P < 0.001) (Table 5). The PFK score (14) was defined as follows: 1 point for female, 1 point for serum potassium < 3.5 mmol/L, and 1 point for urine pH > 7.0, with a maximum total score of 3 points.
Figure 2.
Nomogram of PA prediction model.
Figure 3.
Validation of PA prediction model. (A) ROC curve for training set of prediction model. (B) Calibration curve for training set of prediction model. (C) Decision curve analysis (DCA) curve for training set of prediction model. (D) ROC curve for validation set of prediction model. (E) Calibration curve for validation set of prediction model. (F) DCA curve for validation set of prediction model.
Figure 4.
ROC curves for validation set PFK score and ARR. (A) ROC curve for validation set PFK score. (B) ROC curve for validation set ARR.
Table 5.
DeLong test between nomogram, PFK score, and ARR.
| Comparison | Z | P |
|---|---|---|
| Nomogram vs ARR | -4.974 | <0.001 |
| PFK score vs ARR | 8.727 | <0.001 |
| Nomogram vs PFK score | 4.457 | <0.001 |
The Hosmer-Lemeshow test was used to assess the agreement between the model predictions and the actual results, yielding a χ² of 7.897 (P = 0.444) for the training set and 6.208 (P = 0.624) for the validation set, indicating good model fit. Calibration curves are shown in Figures 3B, E. Decision curve analysis (DCA) was employed to evaluate the clinical utility of the model in decision-making. DCA (Figures 3C, F) demonstrated that using this model to predict primary aldosteronism provided significant clinical benefit to patients. The predictive model showed good agreement with ARR in screening for PA (Kappa = 0.364, P < 0.001), which was better than the agreement between the PFK score and ARR (Kappa = 0.224, P < 0.001) (Table 6).
Table 6.
Consistency between nomogram, PFK score, and ARR.
| ARR | Nomogram | PFK score | ||
|---|---|---|---|---|
| + | - | + | - | |
| + | 71 | 33 | 90 | 14 |
| – | 92 | 285 | 190 | 187 |
| Kappa | 0.364 | 0.224 | ||
| P | <0.001 | <0.001 | ||
4. Discussion
PA was initially considered a rare cause of hypertension. However, with increasing research in recent years, its prevalence has risen annually. Despite this, PA screening remains suboptimal, largely due to the high technical demands of ARR measurement and poor patient compliance (6). To facilitate PA screening in hypertensive patients, we developed and internally validated a novel nomogram prediction model for PA based on six readily available indicators: serum sodium-to-potassium ratio, anion gap, uric acid, urine sodium, 24-hour urine potassium, and urine pH. This model may serve as a potential tool for risk stratification that may guide referral decisions in secondary or tertiary care settings, with future simplification needed for primary care adaptation. It is expected to improve patient compliance with PA screening and thereby promote the early diagnosis and treatment of PA.
As a mineralocorticoid, aldosterone plays a key role in maintaining extracellular volume homeostasis, blood pressure regulation, and serum potassium balance. In patients with PA, autonomous aldosterone secretion from the adrenal cortex leads to increased water and sodium retention, and enhanced potassium excretion, while suppressing the renin-angiotensin system. Clinically, this manifests primarily as hypertension and hypokalemia (15, 16). However, not all PA patients present with hypokalemia, despite excessive aldosterone secretion (17). Studies have reported a hypokalemia prevalence of approximately 37.5% in PA patients (3). In the present study, we observed hypokalemia in 34.48% of PA patients and 12.16% of EH patients (Table 2). Given the high prevalence of hypokalemia in PA, it is also of interest to examine the prevalence of PA among patients presenting with hypokalemia. Another study reported a PA prevalence of 28.1% among hypokalemic patients (18). To systematically investigate the relationship between serum potassium levels and PA, we comprehensively analyzed serum sodium, serum potassium, serum sodium-to-potassium ratio, and anion gap in both PA and EH patients. Additionally, alterations in serum electrolytes may influence urine electrolyte levels; therefore, we incorporated urine sodium and potassium into the predictive model. Our results indicated that elevated sodium-to-potassium ratio and 24-hour urine potassium, along with decreased anion gap and urine sodium, were independently associated with a higher predicted probability of PA. The molecular mechanisms underlying reduced serum potassium in PA patients remain incompletely understood. Studies indicate that somatic KCNJ5 mutations occur in some aldosterone-producing adenomas, leading to structural and functional alterations in the GIRK4 protein, which results in potassium loss and increased sodium permeability (19). Additionally, the ATP1A1 gene may also participate in aldosterone regulation. Approximately 5.2% of patients with aldosterone-producing adenomas carry ATP1A1 mutations, which affect the potassium-binding site and reduce the affinity of the ion pump for potassium ions (20).
Studies have shown a close association between PA and metabolic syndrome (21, 22). Elevated aldosterone and reduced serum potassium increase the risk of diabetes (23, 24). Therefore, in the present study, we analyzed blood glucose, serum lipid levels, and diabetes status in PA and EH patients. The results showed no significant differences in these parameters between the two groups, which is consistent with previous reports (25). Furthermore, studies indicate that PA patients have lower uric acid levels than EH patients, and both groups show a decreasing trend in serum uric acid with rising plasma aldosterone levels and declining renin activity. PA patients often have concomitant renal impairment (26–28). Therefore, we included renal function parameters in our statistical analysis of PA and EH patients. Results indicate that serum uric acid levels were independently associated with PA. Specifically, lower uric acid levels were associated with higher odds of PA. This finding may be related to elevated aldosterone levels in PA patients. High aldosterone induces sodium retention and potassium excretion, thereby increasing blood volume and renal blood flow perfusion, which in turn enhances uric acid excretion and ultimately lowering serum uric acid. Furthermore, studies have reported varying degrees of proximal renal tubular damage in PA patients, which reduces uric acid reabsorption and consequently leads to lower serum uric acid levels (29, 30).
In summary, this study analyzed clinical laboratory parameters associated with PA and developed a novel prediction model incorporating serum sodium-to-potassium ratio, anion gap, uric acid, urine sodium, 24-hour urine potassium, and urine pH, presented as a nomogram. Internal validation of the predictive model demonstrated good discriminatory ability based on the ROC curve. The calibration curves and DCA indicated sound predictive performance and clinical benefit, and the Kappa test showed satisfactory agreement with ARR. For clinical institutions lacking aldosterone and renin testing capabilities, all parameters included in this model are routine serum and urine tests available in clinical laboratories. Therefore, this model can effectively assists clinicians in screening for PA, enabling efficient diagnostic evaluation for patients with limited access to medical resources, and thereby promoting early diagnosis and treatment of PA.
Several limitations of this study should be acknowledged. First, the study population was derived from hypertensive inpatients at a single tertiary university hospital, and the cohort already included patients with indications for PA screening (e.g., resistant hypertension, hypokalemia, adrenal incidentaloma, family history of PA, and obstructive sleep apnea). This high-risk profile reflects a referral-biased population and may not be representative of the broader hypertensive population encountered in primary care settings. Consequently, the model’s performance in unselected hypertensive patients requires further validation. Further multicenter studies with external validation in community-based or primary care cohorts are needed before the model can be generalized to lower-risk populations. Second, some patients in the study had comorbidities such as pneumonia, cholecystitis, urinary tract infections, or kidney stones, and the medications they received for these conditions may have interfered with the study results. Third, although the nomogram demonstrated good predictive performance in our cohort, its current implementation relies on 24-hour urine collections and multiple electrolyte measurements, which may limit its direct applicability in resource-constrained primary care settings. Future studies should explore whether simplified versions of the model—using more readily available variables—could achieve comparable performance, thereby enhancing its utility in primary healthcare settings.
Acknowledgments
We thank all physicians, nurses, and researchers from the Department of Endocrinology and Metabolism, Cardiology, and Urology at Tianjin Medical University General Hospital for their assistance with data collection.
Funding Statement
The author(s) declared that financial support was not received for this work and/or its publication.
Footnotes
Edited by: Michael Stowasser, The University of Queensland, Australia
Reviewed by: Ilker Çordan, Konya City Hospital, Türkiye
Raul Ibarra-Salce, Autonomous University of Coahuila, Mexico
Data availability statement
The original contributions presented in the study are included in the article/supplementary material. Further inquiries can be directed to the corresponding author.
Ethics statement
The studies involving humans were approved by Ethics Committee of Tianjin Medical University General Hospital. The studies were conducted in accordance with the local legislation and institutional requirements. Written informed consent for participation was not required from the participants or the participants’ legal guardians/next of kin in accordance with the national legislation and institutional requirements. Written informed consent was obtained from the individual(s) for the publication of any potentially identifiable images or data included in this article.
Author contributions
XZ: Writing – original draft, Data curation, Software, Investigation, Methodology, Writing – review & editing, Conceptualization, Formal Analysis. YR: Data curation, Methodology, Conceptualization, Investigation, Writing – original draft. HW: Validation, Methodology, Writing – review & editing, Formal Analysis, Software. ZD: Validation, Project administration, Conceptualization, Visualization, Formal Analysis, Methodology, Supervision, Writing – review & editing.
Conflict of interest
The author(s) declared that this work was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
Generative AI statement
The author(s) declared that generative AI was not used in the creation of this manuscript.
Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.
Publisher’s note
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.
References
- 1. Conn JW, Louis LH. Primary aldosteronism, a new clinical entity. Ann Intern Med. (1956) 44:1–15. doi: 10.7326/0003-4819-44-1-1 [DOI] [PubMed] [Google Scholar]
- 2. Rimoldi SF, Scherrer U, Messerli FH. Secondary arterial hypertension: when, who, and how to screen? Eur Heart J. (2013) 35:1245–54. doi: 10.1093/eurheartj/eht534 [DOI] [PubMed] [Google Scholar]
- 3. Käyser SC, Dekkers T, Groenewoud HJ, van der Wilt GJ, Carel Bakx J, van der Wel MC, et al. Study heterogeneity and estimation of prevalence of primary aldosteronism: a systematic review and meta-regression analysis. J Clin Endocrinol Metab. (2016) 101:2826–35. doi: 10.1210/jc.2016-1472 [DOI] [PubMed] [Google Scholar]
- 4. Monticone S, D'Ascenzo F, Moretti C, Williams TA, Veglio F, Gaita F, et al. Cardiovascular events and target organ damage in primary aldosteronism compared with essential hypertension: a systematic review and meta-analysis. Lancet Diabetes Endocrinol. (2018) 6:41–50. doi: 10.1016/s2213-8587(17)30319-4 [DOI] [PubMed] [Google Scholar]
- 5. Young WF, Jr. Diagnosis and treatment of primary aldosteronism: practical clinical perspectives. J Intern Med. (2019) 285:126–48. doi: 10.1111/joim.12831 [DOI] [PubMed] [Google Scholar]
- 6. Yuan X, Lan X, Lihua Z, Qifu L, Shumin Y. Investigation on current status of diagnosis of primary aldosteronism in China. Int J Endocrinol Metab. (2022) 42:288–92. doi: 10.3760/cma.j.cn121383-20201221-1205830704229 [DOI] [Google Scholar]
- 7. Mulatero P, Monticone S, Burrello J, Veglio F, Williams TA, Funder J. Guidelines for primary aldosteronism: uptake by primary care physicians in Europe. J Hypertens. (2016) 34:2253–57. doi: 10.1097/hjh.0000000000001088 [DOI] [PubMed] [Google Scholar]
- 8. Ruhle BC, White MG, Alsafran S, Kaplan EL, Angelos P, Grogan RH. Keeping primary aldosteronism in mind: deficiencies in screening at-risk hypertensives. Surgery. (2019) 165:221–27. doi: 10.1016/j.surg.2018.05.085 [DOI] [PubMed] [Google Scholar]
- 9. Adler GK, Stowasser M, Correa RR, Khan N, Kline G, McGowan MJ, et al. Primary aldosteronism: an Endocrine Society clinical practice guideline. J Clin Endocrinol Metab. (2025) 110:2453–95. doi: 10.1210/clinem/dgaf284 [DOI] [PubMed] [Google Scholar]
- 10. Mulatero P, Monticone S, Deinum J, Amar L, Prejbisz A, Zennaro MC, et al. Genetics, prevalence, screening and confirmation of primary aldosteronism: a position statement and consensus of the Working Group on Endocrine Hypertension of the European Society of Hypertension. J Hypertens. (2020) 38:1919–28. doi: 10.1097/hjh.0000000000002510 [DOI] [PubMed] [Google Scholar]
- 11. Chinese Society of Endocrinology CMA . Expert consensus on the diagnosis and treatment of primary aldosteronism(2024). Chin J Endocrinol Metab. (2025) 41:12–24. doi: 10.3760/cma.j.cn311282-20241115-0054630704229 [DOI] [Google Scholar]
- 12. Yanv R, Xiaoyu Z, Zuoliang D. The optimal cut-off point of plasma aldosterone/renin ratio for screening primary aldosteronism among hypertensive population in Tianjin. Chin J Prev Control Chronic Dis. (2024) 32:827–31+36. doi: 10.16386/j.cjpccd.issn.1004-6194.2024.11.005 [DOI] [Google Scholar]
- 13. Hypertension. WGoCGftMo. League. CH. Cardiology. CSo. Committee. CMDAH. Care. HBoCIEaPAfMaH. Association. HBoCGM . 2018 Chinese guidelines for the management of hypertensionwriting group of 2018. (2019) 24:24–56. doi: 10.3969/j.issn.1007-5410.2019.01.002 [DOI] [Google Scholar]
- 14. Yamashita T, Shimizu S, Koyama M, Ohno K, Mita T, Tobisawa T, et al. Screening of primary aldosteronism by clinical features and daily laboratory tests: combination of urine ph, sex, and serum k. J Hypertens. (2018) 36:326–34. doi: 10.1097/hjh.0000000000001511 [DOI] [PubMed] [Google Scholar]
- 15. Szabo Yamashita T, Shariq OA, Foster TR, Lyden ML, Dy BM, Young WF, Jr., et al. Unilateral adrenalectomy for primary aldosteronism due to bilateral adrenal disease can result in resolution of hypokalemia and amelioration of hypertension. World J Surg. (2023) 47:314–18. doi: 10.1007/s00268-022-06780-x [DOI] [PubMed] [Google Scholar]
- 16. Gupta G, Teo AED, Swee DS, Loh LM, Chuah M, Loh WJ, et al. Prolonged hypokalemia and delayed diagnosis of primary aldosteronism: clinical course and risk factors. J Clin Endocrinol Metab. (2024) 109:e1574–e81. doi: 10.1210/clinem/dgad752 [DOI] [PubMed] [Google Scholar]
- 17. Funder JW, Carey RM, Mantero F, Murad MH, Reincke M, Shibata H, et al. The management of primary aldosteronism: case detection, diagnosis, and treatment: an Endocrine Society clinical practice guideline. J Clin Endocrinol Metab. (2016) 101:1889–916. doi: 10.1210/jc.2015-4061 [DOI] [PubMed] [Google Scholar]
- 18. Burrello J, Monticone S, Losano I, Cavaglià G, Buffolo F, Tetti M, et al. Prevalence of hypokalemia and primary aldosteronism in 5100 patients referred to a tertiary hypertension unit. Hypertension. (2020) 75:1025–33. doi: 10.1161/hypertensionaha.119.14063 [DOI] [PubMed] [Google Scholar]
- 19. Choi M, Scholl UI, Yue P, Björklund P, Zhao B, Nelson-Williams C, et al. K+ channel mutations in adrenal aldosterone-producing adenomas and hereditary hypertension. Sci (New York NY). (2011) 331:768–72. doi: 10.1126/science.1198785 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Beuschlein F, Boulkroun S, Osswald A, Wieland T, Nielsen HN, Lichtenauer UD, et al. Somatic mutations in Atp1a1 and Atp2b3 lead to aldosterone-producing adenomas and secondary hypertension. Nat Genet. (2013) 45:440–4, 4e1-2. doi: 10.1038/ng.2550 [DOI] [PubMed] [Google Scholar]
- 21. Fallo F, Pilon C, Urbanet R. Primary aldosteronism and metabolic syndrome. Horm Metab Res. (2012) 44:208–14. doi: 10.1055/s-0031-1295412 [DOI] [PubMed] [Google Scholar]
- 22. Sun K, Zhou C, Gong M, Zhang Y, Jiang Y, Song W. The prevalence of metabolic syndrome in primary aldosteronism and essential hypertension: a systematic review and meta-analysis. J Clin Hypertens (Greenwich). (2024) 26:879–89. doi: 10.1111/jch.14873 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Hu Y, Zhang J, Liu W, Su X. Determining the prevalence of primary aldosteronism in patients with new-onset type 2 diabetes and hypertension. J Clin Endocrinol Metab. (2020) 105. doi: 10.1210/clinem/dgz293 [DOI] [PubMed] [Google Scholar]
- 24. Saiki A, Otsuki M, Tamada D, Kitamura T, Shimomura I, Kurihara I, et al. Diabetes mellitus itself increases cardio-cerebrovascular risk and renal complications in primary aldosteronism. J Clin Endocrinol Metab. (2020) 105. doi: 10.1210/clinem/dgaa177 [DOI] [PubMed] [Google Scholar]
- 25. Liu Y, Wang M, Qiu X, Ma G, Ji M, Yang Y, et al. A novel clinical-imaging nomogram for predicting primary aldosteronism in patients with hypertension. Hypertens Res. (2023) 46:2603–12. doi: 10.1038/s41440-023-01374-z [DOI] [PubMed] [Google Scholar]
- 26. Miaoyuan Z, Xiuwei Z, Weikun C. Comparison of glucose and lipid metabolism and uric acid levels between primary aldosteronism and essential hypertension. Chin J Integr Med Cardio-Cerebrovascular Dis. (2020) 18:314–7. doi: 10.12102/j.issn.1672-1349.2020.02.031 [DOI] [Google Scholar]
- 27. Cohen DL, Wachtel H, Vaidya A, Hundemer GL, Tezuka Y, Davio A, et al. Primary aldosteronism in chronic kidney disease: blood pressure control and kidney and cardiovascular outcomes after surgical versus medical management. Hypertension. (2023) 80:2187–95. doi: 10.1161/hypertensionaha.123.21474 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. Nakano Y, Murakami M, Hara K, Fukuda T, Horino M, Takeuchi A, et al. Long-term effects of primary aldosteronism treatment on patients with primary aldosteronism and chronic kidney disease. Clin Endocrinol (Oxf). (2023) 98:323–31. doi: 10.1111/cen.14849 [DOI] [PubMed] [Google Scholar]
- 29. Namba M, Kikuchi K, Komura H, Suzuki S, Satoh N, Ohtomo T, et al. Study on uric acid metabolism in patients with primary aldosteronism. Nihon Naibunpi Gakkai zasshi. (1992) 68:51–61. doi: 10.1507/endocrine1927.68.1_51 [DOI] [PubMed] [Google Scholar]
- 30. Mandal AK, Mount DB. The molecular physiology of uric acid homeostasis. Annu Rev Physiol. (2015) 77:323–45. doi: 10.1146/annurev-physiol-021113-170343 [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
The original contributions presented in the study are included in the article/supplementary material. Further inquiries can be directed to the corresponding author.