Non-albumin proteinuria is associated with all-cause mortality in community-dwelling adults

Abstract

Non-albumin proteinuria (NAP), the presence of proteins other than albumin in urine, involves a wide range of urinary proteins that reflect mainly tubular injury. Although hospital-based studies have reported an association between NAP and mortality risk, little is known about this association in community-dwelling adults. This community-based prospective cohort study conducted in Japan included 6,601 participants (aged ≥ 40 years, 48.0% women). We assessed NAP as the difference between the urine albumin-to-creatinine ratio (ACR) and the urine protein-to-creatinine ratio. The association of NAP with all-cause mortality was assessed by multivariable Cox proportional hazard regression analysis, with adjustment for potential confounders. During the median follow-up of 11.2 years, a total of 1,182 deaths were documented. For a doubling of NAP, the adjusted hazard ratio (95% confidence interval) for mortality was 1.16 (1.11, 1.21). When further adjusted for ACR levels, this association remained significant (adjusted hazard ratio [95% confidence interval], 1.12 [1.06, 1.18]). We found that higher NAP was positively associated with all-cause mortality independent of albuminuria levels in community-dwelling adults. These findings suggest that NAP may serve as a complementary marker to albuminuria and is associated with elevated mortality risk in community settings.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-026-39833-5.

Introduction

The presence of proteins in the urine is a sign of kidney damage and is linked to adverse health outcomes such as all-cause and cardiovascular mortality as well as end-stage kidney disease, even in the general population¹. Clinical guidelines recommend screening for chronic kidney disease (CKD) in adults by assessing urine albumin or protein^2,3. Despite the recent expansion of treatment options for CKD and growing awareness of the need to reduce albuminuria, significant residual health risks remain, indicating the need for reliable predictive markers for those risks.

To date, no single residual risk marker has been identified, although previous studies have evaluated health risks, using specialized biomarkers such as a urine as well as blood biomarker panel and proteomics markers^4,5. In contrast, non-albumin proteinuria (NAP), calculated as the difference or ratio between albuminuria and proteinuria, might be a potential residual risk marker because elevated NAP reflects the progression of renal dysfunction in patients with kidney disease^6,7. NAP involves a wide range of urinary proteins that can reflect tubulointerstitial pathology⁸or the overflow of low-molecular-weight proteins due to overproduction from other non-renal systemic diseases. Although previous hospital-based studies have reported a positive association of NAP with patient mortality^9,10, no study has prospectively evaluated the impact of NAP on mortality risk in the general population.

Therefore, this study examined the association of NAP with mortality risk compared with albuminuria and proteinuria, using data from a Japanese community-based prospective cohort study, the Uonuma Cohort Study. CKD is one of the primary interests in this cohort¹¹, enabling us to simultaneously assess albuminuria and proteinuria in community-dwelling adults. The findings provide new insights into residual risk markers using NAP in conjunction with currently available clinical measures.

Methods

Study population

The present study analyzed data from urine samples collected as a subset of the Uonuma Cohort Study. The subset involved participants of governmental health check-ups aged 40 years or older between April 1, 2012, and March 31, 2015, in the cities of Minamiuonuma and Uonuma, Niigata, Japan (FYs 2012–2013 for Minamiuonuma City and 2014 FY for Uonuma City, n = 10,654) and consent was obtained from 7,426 participants (participation ratio, 69.7%); of these, 98 were excluded due to missing urine samples and 4 due to missing information on covariates. We previously compared the characteristics of overall participants of the Uonuma cohort study who submitted urine samples with those who did not¹². We excluded participants whose urine albumin-to-creatinine ratio (ACR) was higher than their urine protein-to-creatinine ratio (PCR) because of potential measurement error, leaving 6,601 participants for the present analysis (Supplementary Fig. 1). Compared with participants who had a higher PCR than ACR, those who had a higher ACR than PCR were more likely to be younger and women and to have a lower PCR, higher eGFR, lower blood pressure, and have a regular exercise habit; and were less likely to be a smoker and drinker and to have diabetes, take antihypertensive medication, and have a history of cardiovascular disease (Supplementary Table 1). All participants provided written informed consent before inclusion. The study protocol was approved by the Ethics Committee of Niigata University (approval numbers, 2012-1640 and 2017-0054).

Urine assessment

At the baseline health checkup, spot urine samples were collected for measurement¹¹. We assessed albuminuria as ACR, and proteinuria as PCR. In fresh urine, urine albumin and creatinine concentrations were measured using the latex agglutination method and the enzymatic method, respectively. Urine protein was measured in a stored frozen urine sample, using the Pyrogallol Red Method. NAP was assessed as the difference between albuminuria and proteinuria, calculated by subtracting ACR from PCR. In addition, we assessed the urine albumin-to-protein ratio (APR), calculated by dividing ACR by PCR. We assessed 3 × 3 levels of ACR and NAP as follows: ACR < 30, 30–299, and ≥ 300 mg/g; and NAP < 120, 120–199, and ≥ 200 mg/g. These cutoff values were taken from the Japanese CKD clinical guidelines³, and the NAP cutoff was determined by subtracting the ACR cutoff from the PCR cutoff.

Outcomes

The primary outcome was all-cause mortality until December 31, 2024. Secondarily, we assessed the cause of death, including cardiovascular disease (heart disease and stroke), cancer, and other causes until December 31, 2021. Information on the dates of deaths was obtained from the Basic Residential Registry managed by the governments of Minamiuonuma and Uonuma. Data on the causes of deaths occurring in Niigata Prefecture were obtained from the Ministry of Health, Labour and Welfare. Cause of death codes were assigned later by the Ministry of Health, Labour and Welfare, and further coded for research purposes, so the end dates for the available data on all-cause and cause of death were not the same in this study. If a study participant could not be reached for 3 years and their survival status was unknown, they were set as censored on the day contact was lost.

Other study variables

Demographic variables, clinical histories, and blood measurements were obtained from the baseline health checkups. Body weight and height were measured, and body mass index was calculated as body weight (kg) divided by height (m) squared. Systolic and diastolic blood pressures were measured with the participant in a sitting position. Information on whether or not the participant is a current smoker, drinks every day, has an exercise habit, and has a history of heart disease and/or stroke was obtained from a self-administered questionnaire, which was confirmed by public health nurses. We obtained the history of cancer from the responses to a self-administered questionnaire distributed to a subsample of the participants.

Serum creatinine was measured using the enzymatic method, and estimated glomerular filtration rate (eGFR) was calculated using the Japanese Society of Nephrology equation as follows: eGFR (mL/min/1.73 m²) = 194 × (serum creatinine [mg/dL]) (− 1.094) × age (years) (− 0.287) (× 0.739 [for women])¹³. Diabetes was defined according to the plasma glucose level (fasting glucose ≥ 126 mg/dL or casual glucose ≥ 200 mg/dL), HbA1c (≥ 6.5%)¹⁴ or use of antidiabetic medication. Hypertension was defined by systolic blood pressure (≥ 140 mmHg), diastolic blood pressure (≥ 90 mmHg)¹⁵, or use of any antihypertensive medications.

Statistical analysis

Because we obtained the lowest value of urinary protein concentration for proteinuria, but only ACR levels (and not the urinary albumin concentration itself) were available for albuminuria in this study, we imputed a value of one quarter of the lowest value for each of ACR and PCR in 80 and 36 participants for whom one or both of these values was below the limit of detection (i.e. ACR or PCR = 0), respectively. Baseline characteristics were assessed as the median (interquartile interval [IQI]), or numbers (%) according to quartiles of NAP and APR. Spearman’s correlation coefficients assessed the correlations between NAP, APR, ACR, and PCR.

Multivariable Cox proportional hazard regression analysis was used to test the associations of NAP, APR, ACR, and PCR with all-cause mortality. We set the urine sample collection date as the date that observation was started and assessed the proportional hazards assumption using cumulative sums of martingale residuals. In the model, NAP, ACR, and PCR were converted to the logarithm of base 2 and APR was assessed per 10% increment. In addition, NAP, APR, ACR, and PCR were divided by quartile and the lowest quartile was set as the reference. First, we ran Model 1 with adjustments for age, sex, smoking, drinking, and exercise habits, then we ran Model 2 with further adjustments for body mass index, diabetes, systolic blood pressure, use of any antihypertensive medication, history of cardiovascular disease, and eGFR. Given that albuminuria and proteinuria are risk factors of cardiovascular disease^16,17and are affected by antihypertensive medications¹⁸that can reduce mortality risk¹⁹, we selected these covariates as potential confounders. We ran Model 3 with further adjustments for ACR levels (< 30, 30–299, and ≥ 300 mg/g)³ for NAP and APR models and NAP quartile for ACR and PCR models in order to assess the independence of each. We assessed multicollinearity, using the variance inflation factor of independent variables in each model, and confirmed that each estimate was less than 5.

Second, we modeled the association of NAP and APR with all-cause mortality, using restricted cubic spline terms, setting knots at quartiles and a reference set at the 12.5 percentile (Models 2 and 3). Third, we estimated the hazard ratios and 95% confidence intervals (CIs) for all-cause mortality according to the 3 × 3 ACR and NAP levels, using the crude model and the multivariable adjusted model (Model 2).

We performed similar analyses to assess the associations of NAP, APR, ACR, and PCR with cause of death. As a sensitivity analysis, we ran the same models while excluding participants who had history of cardiovascular disease (for cardiovascular mortality), a history of any cancer (for cancer mortality), or who died due to kidney disease (for all-cause mortality).

Finally, a subgroup analysis was performed by sex, presence of diabetes, presence of hypertension, and eGFR level (below or above 60 mL/min/1.73 m²). P for interaction was assessed by testing the cross-term between each exposure variable and subgroup variable in Model 3. Statistical analyses were performed using SAS ver. 9.4 (SAS Institute Inc., Cary, NC) or Stata 16.1 (StataCorp, College Station, TX).

Results

Participants’ characteristics at baseline

Mean (standard deviation) age and eGFR were 68.4 (10.2) years old and 74.5 (16.1) mL/min/1.73 m², respectively. Median (IQI) NAP, APR, ACR, and PCR were 27.3 (16, 45.1) mg/g, 0.33 (0.20, 0.50), 11 (6, 25) mg/g, and 40.8 (25.6, 69.8) mg/g, respectively. Compared with the participants in the lowest quartile of NAP, those having higher NAP levels were more likely to be men, older, current smokers, and drinkers, to take any antihypertensive medications, and to have diabetes, a history of cardiovascular disease, and higher levels of ACR and PCR, lower eGFR, and higher blood pressure (Table 1). Compared with the participants in the lowest quartile of APR, those having higher APR levels were more likely to be women and older, to take any antihypertensive medications, and to have diabetes, a history of cardiovascular disease, and higher levels of ACR and PCR, lower eGFR, higher blood pressure, and a regular exercise habit, and less likely to be a current smoker and drinker (Supplementary Table 2).

Table 1.

Baseline characteristics according to quartile of urine non-albumin proteinuria.

	Quartile of urine non-albumin proteinuria, mg/g
	< 16.0	16.0–27.2	27.3–45.0	> 45.0
N	1,652	1,645	1,654	1,650
Urine albumin-to-creatinine ratio, mg/g	7.8 [5, 13]	8.4 [5, 15]	12 [7, 23]	34 [13, 88]
Urine protein-to-creatinine ratio, mg/g	18.2 [13.9, 23.8]	30.8 [26.3, 37.7]	47.7 [40.8, 59.1]	108.1 [76, 186]
eGFR, mL/min/1.73 m2	74.0 [65.6, 83.8]	75.0 [65.7, 84.8]	73.7 [64.4, 84.4]	71.8 [60.2, 83.5]
Male sex, n (%)	784 (47.5)	845 (51.4)	899 (54.4)	906 (54.9)
Age, years	67 [61, 73]	68 [62, 74]	70 [63, 76]	72 [66, 78]
Body mass index, kg/m2	22.6 [20.6, 24.5]	22.6 [20.6, 24.6]	22.5 [20.5, 24.6]	22.5 [20.6, 24.7]
Systolic blood pressure, mmHg	125 [115, 137.5]	127 [116, 140]	128 [119, 142]	133 [122, 148]
Diastolic blood pressure, mmHg	73 [66, 81]	75 [67, 82]	76 [68, 83]	76 [68, 83]
Diabetes, n (%)	121 (7.3)	112 (6.8)	165 (10.0)	266 (16.1)
Taking anti-hypertensive medication, n (%)	447 (27.1)	498 (30.3)	572 (34.6)	788 (47.8)
History of cardiovascular disease, n (%)	110 (6.7)	138 (8.4)	144 (8.7)	195 (11.8)
Current smoker, n (%)	245 (14.8)	268 (16.3)	287 (17.4)	277 (16.8)
Drinks alcohol every day, n (%)	487 (29.5)	520 (31.6)	586 (35.4)	610 (37.0)
Regular exercise habit, n (%)	606 (36.7)	590 (35.9)	618 (37.4)	583 (35.3)

Values are presented as the median [interquartile interval] or number (percentage). Non-albumin proteinuria was urine albumin-to-creatinine ratio subtracted from urine protein-to-creatinine ratio (PCR − ACR). History of cardiovascular disease included histories of stroke and heart disease. ACR, urine albumin-to-creatinine ratio, eGFR, estimated glomerular filtration rate, PCR, urine protein-to-creatinine ratio.

Spearman’s correlation coefficients (95% CI) of NAP with APR, ACR, and PCR were − 0.226 (− 0.249, − 0.203), 0.466 (0.447, 0.485), and 0.895 (0.890, 0.899), respectively. Correlation coefficients of APR with ACR and PCR were 0.698 (0.685, 0.710) and 0.135 (0.112, 0.159), respectively.

Association between NAP and all-cause mortality

During the median follow-up of 11.2 years, we observed 1,182 events (incidence rate, 17.1 per 1,000 person-years). In the multivariable Cox proportional hazard regression, NAP, APR, ACR, and PCR were positively associated with all-cause mortality (Model 1, Table 2). When the clinical factors were accounted for (Model 2), those associations were slightly attenuated but remained significant. Furthermore, adjusted for ACR levels, NAP remained significant, whereas APR did not.

Table 2.

Associations of NAP, APR, PCR, and ACR with all-cause mortality.

	Model 1	Model 2	Model 3
NAP (per double increase)	1.19 [1.14, 1.24]	1.16 [1.11, 1.21]	1.12 [1.06, 1.18]
Q1	Reference	Reference	Reference
Q2	0.92 [0.76, 1.11]	0.91 [0.75, 1.10]	0.91 [0.75, 1.10]
Q3	1.08 [0.90, 1.29]	1.04 [0.87, 1.25]	1.02 [0.85, 1.22]
Q4	1.62 [1.37, 1.91]	1.51 [1.27, 1.79]	1.37 [1.14, 1.65]
APR (per 10% increase)	1.04 [1.01, 1.07]	1.03 [1.002, 1.06]	0.98 [0.95, 1.02]
Q1	Reference	Reference	Reference
Q2	1.11 [0.93, 1.31]	1.09 [0.92, 1.29]	1.06 [0.90, 1.26]
Q3	1.32 [1.12, 1.56]	1.31 [1.11, 1.55]	1.17 [0.99, 1.39]
Q4	1.23 [1.04, 1.46]	1.19 [1.004, 1.41]	0.91 [0.74, 1.11]
ACR (per double increase)	1.13 [1.10, 1.17]	1.12 [1.08, 1.16]	1.09 [1.05, 1.13]
Q1	Reference	Reference	Reference
Q2	0.99 [0.82, 1.20]	1.01 [0.83, 1.22]	0.97 [0.80, 1.18]
Q3	1.13 [0.94, 1.36]	1.13 [0.93, 1.36]	1.05 [0.87, 1.28]
Q4	1.58 [1.33, 1.89]	1.53 [1.27, 1.84]	1.32 [1.08, 1.61]
PCR (per double increase)	1.20 [1.15, 1.25]	1.18 [1.13, 1.23]	1.17 [1.09, 1.24]
Q1	Reference	Reference	Reference
Q2	1.01 [0.83, 1.23]	1.01 [0.83, 1.22]	0.95 [0.77, 1.18]
Q3	1.18 [0.98, 1.42]	1.16 [0.96, 1.40]	1.04 [0.81, 1.35]
Q4	1.73 [1.46, 2.06]	1.63 [1.36, 1.95]	1.40 [1.02, 1.92]

Data are shown in adjusted hazard ratio (95% CI) of NAP, APR, PCR, and ACR for all-cause mortality. Model 1 was adjusted for age, sex, smoking, drinking, and exercise habits. Model 2 was further adjusted for body mass index, diabetes, systolic blood pressure, taking any antihypertensive medication, history of cardiovascular disease, and estimated glomerular filtration rate. Model 3 was further adjusted for ACR levels (< 30, 30–299, and ≥ 300 mg/g) in NAP and APR models or NAP quartile in ACR and PCR models. ACR, urine albumin-to-creatinine ratio, APR, urine albumin-to-protein ratio, NAP, urine non-albumin proteinuria, PCR, urine protein-to-creatinine ratio.

Restricted cubic spline curves between NAP and APR and all-cause mortality were positively linear in Model 2 (Supplementary Fig. 2). Further adjusted for ACR levels, the association between NAP and all-cause mortality was slightly attenuated but still linear, whereas that of APR did not demonstrate a linear relationship with all-cause mortality (Fig. 1).

Fig. 1.

Restricted cubic spline curves of non-albumin proteinuria and urine albumin-to-protein ratio for all-cause mortality Note. Restricted cubic spline curves were created with knots at the quartiles and a reference set at the 12.5 percentile. Cox proportional hazard model was adjusted for age, sex, smoking, drinking, exercise habits, body mass index, diabetes, systolic blood pressure, taking any antihypertensive medication, history of cardiovascular disease, estimated glomerular filtration rate, and urine albumin-to-creatinine ratio levels. Doted lines indicate 95% CI.

Next, we ran multivariable Cox proportional hazard regression models across the 3 × 3 ACR-NAP levels with the references set at ACR < 30 mg/g and NAP < 120 mg/g (Table 3). In the crude model, the highest hazard ratio (95% CI) for all-cause mortality was seen in the participants with ACR 30–299 mg/g and NAP ≥ 200 mg/g at 6.06 (3.80, 9.67), followed by those with ACR ≥ 300 mg/g and NAP ≥ 200 mg/g and those with ACR 30–299 mg/g and NAP 120–199 mg/g. After accounting for potential confounders, those associations were attenuated but remained significant.

Table 3.

Hazard ratio and 95% CI for all-cause mortality according to 3 × 3 ACR and NAP levels.

	Crude model			Multivariable-adjusted model
	NAP, mg/g			NAP, mg/g
ACR, mg/g	< 120	120–199	≥ 200	< 120	120–199	≥ 200
< 30	Reference	1.35 [0.60, 3.01]	1.06 [0.27, 4.25]	Reference	1.11 [0.50, 2.48]	1.23 [0.31, 4.95]
30–299	1.85 [1.62, 2.12]	3.31 [2.50, 4.38]	6.06 [3.80, 9.67]	1.40 [1.22, 1.61]	1.80 [1.35, 2.40]	2.37 [1.47, 3.82]
≥ 300	1.61 [0.52, 5.00]	1.48 [0.66, 3.31]	3.26 [2.30, 4.62]	1.02 [0.33, 3.17]	1.003 [0.45, 2.25]	1.91 [1.34, 2.72]

Multivariable adjusted model included age, sex, smoking, drinking, exercise habits, body mass index, diabetes, systolic blood pressure, taking any antihypertensive medication, history of cardiovascular disease, and estimated glomerular filtration rate. NAP was calculated as PCR minus ACR. ACR, urine albumin-to-creatinine ratio, NAP, non-albumin proteinuria, PCR, urine protein-to-creatinine ratio.

Association between NAP and cause of death

We observed 167 deaths due to cardiovascular disease (3.2 per 1,000-person years), 260 due to cancer (5.0 per 1,000-person years), and 253 due to other causes (4.8 per 1,000-person years). NAP, ACR, and PCR were associated with death due to cardiovascular disease, cancer, and other causes in Models 1 and 2 (Supplementary Table 3). When mutually adjusted for ACR levels or NAP quartile, these associations were slightly attenuated but remained significant in Model 3, except for those with cancer mortality.

Sensitivity analysis

Excluding participants with a history of cardiovascular disease at baseline (N = 6,014), the adjusted hazard ratio (95% CI) of NAP for cardiovascular mortality was 1.18 (1.03, 1.36) in Model 2, but this value was attenuated in Model 3 (an adjusted hazard ratio [95% CI], 1.13 [0.96, 1.34]) (Supplementary Table 4). Excluding participants who had history of cancer at baseline (N = 5,148), the associations of NAP, APR, ACR, and PCR with cancer mortality were not significant across all three models (Supplementary Table 5). Excluding participants who died due to kidney disease (N = 6,593), similar significant positive associations of NAP, ACR, and PCR with all-cause mortality were consistent across all three Models (Supplementary Table 6).

Subgroup analysis

In the subgroup analysis, the associations of NAP and APR with all-cause mortality were consistent across groups except for the diabetes group. Participants with diabetes demonstrated a stronger positive association between NAP and mortality risk than those without diabetes, but the interaction was not significant (Fig. 2) (P for interaction = 0.407). Compared with participants without diabetes, there was an inverse association between APR and all-cause mortality in those with diabetes (adjusted odds ratio [95% CI], 0.89 [0.81, 0.97], P for interaction = 0.101). Regarding ACR and PCR, the consistent positive associations with all-cause mortality were seen across subgroups. Among participants with diabetes, NAP demonstrated a significantly higher hazard ratio (95% CI) for all-cause mortality (1.30 [1.13, 1.50]), but that was not the case for ACR and PCR.

Fig. 2.

Associations of NAP, APR, ACR, and PCR with all-cause mortality by subgroup Note. Points and dotted lines show adjusted hazard ratios and 95% CI. In the model, NAP, ACR, and PCR were converted to the logarithm of base 2 and APR was assessed per 10% increment. Cox regression model was adjusted for age, sex, smoking, drinking, and exercise habits, body mass index, diabetes, systolic blood pressure, taking any hypertensive medication, history of cardiovascular disease, estimated glomerular filtration rate and ACR levels (for NAP and APR models) or NAP quartile (for ACR and PCR models). The variable used to define subgroups was not included in that subgroup analysis model (e.g., age was not included in the model when we analyzed age subgroup). ACR, urine albumin-to-creatinine ratio, APR, urine albumin-to-protein ratio, NAP, urine non-albumin proteinuria, PCR, urine protein-to-creatinine ratio.

Discussion

This prospective cohort study found significant positive associations of NAP with all-cause mortality independent of ACR levels in community-dwelling middle-aged and older adults. When participants were divided into 3 × 3 ACR and NAP levels, those with the highest NAP level (> 200 mg/g) concomitant with microalbuminuria (ACR 30–299 mg/g) were at the highest risk for all-cause mortality. NAP demonstrated an independent positive association with cardiovascular mortality but not with cancer mortality. We consistently observed positive associations of NAP with mortality risk across subgroups.

The present community-based cohort study confirmed the findings of prior hospital-based studies that NAP is associated with mortality risk. As a unique aspect of the present study, we prospectively assessed long-term mortality risk and rigorously adjusted for potential confounders, including lifestyle factors such as smoking, drinking, and exercise habits. Although the contents of NAP can differ between normal and pathological conditions, NAP might serve as a complementary marker for mortality risk even in community-dwelling adults.

To our knowledge, this study is the first to examine the association of NAP with cause of death. Our findings indicate that elevated NAP is a significant risk factor for cardiovascular mortality but not for cancer mortality. Some individual or panel tubule markers (potential contents of NAP) were found to be associated with the risk of cardiovascular mortality. For example, urine beta-2 microglobulin was found to be an independent risk factor for cardiovascular mortality in community-dwelling Japanese adults²⁰, while a panel of tubule markers was reported to improve discrimination of cardiovascular risk beyond eGFR and ACR in participants in the Systolic Blood Pressure Intervention Trial⁴. However, the present study assessed a small number of cause-specific deaths, particularly cardiovascular and cancer mortality, and the results had wide confidence intervals; therefore, the results that NAP is associated with cause of death should be interpreted with caution.

Although the interactions did not reach statistical significance, participants with diabetes and high NAP (i.e., less albuminuria relative to proteinuria) had a higher risk of mortality, while those having high APR (i.e., more albuminuria relative to proteinuria) had a lower risk of mortality, in contrast to participants without diabetes. Even more surprisingly, ACR was not associated with mortality risk in participants with diabetes in the present study. Recent clinical guidelines for diabetic kidney disease recommend reducing albuminuria as a therapeutic/management target²¹; thus, our findings may indicate a weakened association between albuminuria and mortality risk, reflecting the therapeutic effects on ACR in recent decades. Nevertheless, the results of this subgroup analysis did not show a significant interaction, and future studies are needed to confirm our observations.

The present study has some implications. Our findings suggest that assessing NAP might be a useful approach for stratifying people with increased mortality risk in routine clinical settings. ACR and PCR are readily measurable, and we can calculate NAP by subtracting PCR from ACR. Importantly, the individuals with the highest mortality risk would be missed if we measured only ACR. Indeed, a prior study involving a US database pointed out that testing for only albuminuria could miss up to 40% of people with severe proteinuria²². NAP should also enable noninvasive quantitative assessment regardless of ACR levels in large studies, given that it does not require expensive or specialized measurements in the laboratory setting. NAP might also facilitate the identification of high-risk populations for all-cause mortality, beyond those detected by albuminuria alone.

Although the present study is the first to demonstrate a prospective positive association between NAP and all-cause mortality in community-dwelling adults, the results should be interpreted with caution because NAP was strongly correlated with PCR in this population. However, in the Cox proportional hazard model with PCR as an independent variable, NAP levels did not violate multicollinearity. Moreover, our results indicated that NAP, but not ACR or PCR, could be used to discriminate mortality risk in people with diabetes. Further research is warranted to explore the biological plausibility of NAP in relation to mortality risk.

This study has some limitations that should be noted. First, we measured ACR and PCR only once in a spot urine sample, but the NAP content was not measured. Therefore, we cannot draw conclusions about the effects of chronicity or mechanism on the association between NAP and all-cause mortality. Second, the baseline information on histories of cardiovascular disease and cancer and that of lifestyles were based on self-administered questionnaires, introducing the possibility of reporting bias. Third, as in other observational studies, residual confounding cannot be ruled out. Although we adjusted for self-reported taking of anti-hypertensive medications, we were not able to account for information on inhibitors of the renin angiotensin system or sodium-glucose co-transporter 2. These therapeutic agents can substantially reduce albuminuria and improve cardiovascular outcomes^18,23; therefore, the absence of these treatment data raises concerns about residual confounding. Finally, because this study was conducted in a non-hospital setting in one area of Japan, the generalizability of the findings to hospital inpatient settings and other ethnicities should be carefully assessed. Nevertheless, the present study has some strengths, including being the first to analyze the associations of NAP with mortality and cause of death, compared with albuminuria and proteinuria, in the general population.

In conclusion, this prospective cohort study found that NAP was positively associated with all-cause mortality in community-dwelling middle-aged and older adults. This association was independent of albuminuria levels and might expand to that of cardiovascular mortality. These findings suggest that NAP may serve as a complementary marker to albuminuria, and further studies are warranted to explore the biological plausibility of applying NAP to the general population.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

KK, MH, YI, HK, and AS conceived and designed the project and review and editing. KK and YI performed data curation, and KK performed formal analysis and writing original draft. KN, NI, ST, and AS contributed to supervision. YI, JT, and SY contributed to project administration. All authors read and approved the manuscript.

Funding

This study was supported by funding from Niigata Prefecture and the Japan Society for the Promotion of Science KAKENHI (17K15850 and 23K07694 to KK). The funders had no role in the design, nor in the collection, analysis or interpretation of the data, nor in the writing of the report or in the decision to submit the paper for publication.

Data availability

All data analyzed in this study are contained in the manuscript or the supplementary material.

Declarations

Competing interests

The authors declare no competing interests.

Ethics statements

The study protocol was approved by the Ethics Committee of Niigata University (approval numbers, 2012-1640 and 2017-0054). The study was conducted in accordance with the Declaration of Helsinki. Participants provided informed consent to take part in the study before participating.