Residential altitude, aerosol optical depth, and diagnostic composition of biopsy-confirmed renal diseases: a single-center renal biopsy study from Yunnan, China

Abstract

Background

Residential altitude may influence human physiology and reflect broader environmental gradients, but its relationship with the diagnostic composition of biopsy-confirmed renal diseases remains unclear.

Methods

We retrospectively studied patients undergoing native kidney biopsy at The First People’s Hospital of Yunnan Province, China, between February 2023 and February 2026. Residential altitude was derived from geocoded permanent addresses and analyzed using altitude quartiles and continuous models per 500-m increase. Restricted cubic spline analyses were performed as exploratory analyses. Environmental exposures were assigned according to geocoded residential coordinates and matched to the calendar year of kidney biopsy. Logistic regression and Firth penalized logistic regression compared biopsy-confirmed ANCA-associated renal vasculitis with other biopsy-confirmed renal diagnoses, with an additional sensitivity analysis using an expanded clinical-or-pathological ANCA outcome.

Results

Among 1,602 patients, 1,593 had available altitude data. The overall distribution of renal pathological diagnoses did not differ across altitude quartiles. Exploratory restricted cubic spline analyses showed no significant altitude association for most major diagnoses, whereas ANCA-associated renal vasculitis demonstrated an inverse overall association without evidence of nonlinearity. In the selected multivariable model adjusted for age, sex, BMI, and ethnicity, each 500-m increase in altitude was associated with lower odds of biopsy-confirmed ANCA-associated renal vasculitis relative to other biopsy diagnoses (OR 0.51, 95% CI 0.34–0.77; P = 0.001). Similar results were observed in Firth penalized logistic regression. The association was attenuated after additional adjustment for distance to Kunming and AOD.

Conclusions

In this single-center biopsy-based cohort from Yunnan, residential altitude was not associated with broad differences in the overall spectrum of renal pathological diagnoses. Biopsy-confirmed ANCA-associated renal vasculitis accounted for a lower relative proportion of biopsy diagnoses at higher residential altitude, but this association was attenuated after accounting for geographic and environmental factors. These findings suggest that altitude-related patterns in renal biopsy diagnoses may reflect combined geographic, environmental, and referral-related gradients and should be interpreted within the context of a hospital-based biopsy population.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s12882-026-05386-y.

Introduction

Kidney biopsy remains essential for establishing renal pathological diagnoses and guiding disease-specific treatment in nephrology. The spectrum of biopsy-confirmed kidney diseases varies across regions and populations, reflecting differences in demographic structure, ethnicity, comorbidity burden, environmental exposures, biopsy indications, and healthcare practice. Large biopsy-based studies from China and other countries have shown substantial geographic variation in the relative frequencies of IgA nephropathy, membranous nephropathy, diabetic kidney disease, hypertensive nephropathy, tubulointerstitial disease, and other pathological diagnoses [1, 2]. These findings highlight the importance of evaluating renal pathological patterns within specific regional contexts.

Residential altitude is a distinct geographic exposure that may influence human physiology through chronic hypobaric hypoxia. High-altitude residence is associated with hematologic and cardiopulmonary adaptation, and the kidney plays an important role in oxygen sensing, erythropoietic regulation, and maintenance of fluid and electrolyte balance. The kidney may also be vulnerable to hypoxia-related stress, particularly within the tubulointerstitial compartment [3]. Previous studies of high-altitude populations have mainly focused on renal function, proteinuria, blood pressure, and high-altitude renal syndrome rather than the distribution of biopsy-confirmed renal pathological diagnoses [4].

Yunnan Province, located in southwestern China, provides a unique setting for studying the relationship between residential altitude and renal pathology. The province has complex terrain, substantial ethnic diversity, and marked altitude variation, with residential areas ranging from low-altitude valleys to plateau regions above 3,000 m. In addition to hypobaric hypoxia, residential altitude in Yunnan may also reflect broader geographic and environmental gradients, including climate, air pollution, aerosol burden, population distribution, and healthcare accessibility. Despite these features, limited evidence is available on whether residential altitude is associated with the spectrum of biopsy-confirmed renal diseases. This question is also relevant to antineutrophil cytoplasmic antibody (ANCA)-associated renal vasculitis, a rare but clinically important cause of rapidly progressive glomerulonephritis for which geographic and environmental heterogeneity has been reported [5].

Therefore, we conducted a retrospective single-center renal biopsy cohort study in Yunnan Province, China, to evaluate whether residential altitude was associated with the distribution of major biopsy-confirmed renal pathological diagnoses. Residential altitude was derived from geocoded permanent residential information and analyzed using altitude quartiles, continuous models, and exploratory restricted cubic spline analyses. We further performed focused analyses of biopsy-confirmed ANCA-associated renal vasculitis, including Firth penalized logistic regression, an expanded clinical-or-pathological outcome sensitivity analysis, and exploratory environmental attenuation analyses. This study was designed to characterize altitude-related patterns in biopsy-based renal diagnostic composition and to generate hypotheses for future multicenter studies with population-based denominators and more detailed environmental and healthcare-access data.

Methods

Study design and participants

This retrospective single-center study included patients who underwent native kidney biopsy at The First People’s Hospital of Yunnan Province between February 2023 and February 2026. Clinical and pathological data were retrieved from the renal biopsy registry and electronic medical records. The study population was restricted to patients with permanent residence in Yunnan Province. Patients without available residential altitude information were excluded from altitude-related analyses. The study protocol was approved by the Ethics Committee of The First People’s Hospital of Yunnan Province (approval number: KHLL2025-KY117). Written informed consent was obtained from all patients.

Clinical and pathological data collection

Demographic and clinical variables included age, sex, ethnicity, smoking status, alcohol consumption, hypertension, diabetes mellitus, height, and weight. Laboratory parameters closest to the time of kidney biopsy were collected, including white blood cell count, hemoglobin, platelet count, serum creatinine, uric acid, albumin, total cholesterol, triglycerides, and 24-hour urinary protein excretion. Body mass index (BMI) was calculated as weight divided by height squared (kg/m²). Estimated glomerular filtration rate (eGFR) was calculated using the 2021 Chronic Kidney Disease Epidemiology Collaboration creatinine Eq [6]. The main renal pathological diagnosis was assigned according to predefined adjudication rules based on biopsy labels, discharge diagnoses, and biopsy conclusion text. Diagnoses were categorized as primary IgA nephropathy, primary membranous nephropathy, tubulointerstitial disease, primary minimal change disease/focal segmental glomerulosclerosis (MCD/FSGS), lupus nephritis, hypertensive nephropathy, diabetic kidney disease, ANCA-associated renal vasculitis, and other diagnoses. The main outcome for regression analyses was biopsy-confirmed ANCA-associated renal vasculitis. A sensitivity analysis additionally used an expanded clinical-or-pathological ANCA-associated renal vasculitis outcome, defined by either the main pathological diagnosis or a compatible clinical diagnosis.

Residential altitude assessment

Permanent residential addresses were standardized into a unified residence field and geocoded to obtain longitude and latitude coordinates in the World Geodetic System 1984 (WGS84). Geocoding was performed using the Esri ArcGIS World Geocoding Service. Candidate locations were selected according to geocoding match scores and address types, with priority given to locality-, subregion-, and region-level matches when exact address-level matching was unavailable.

Residential altitude was retrieved according to the derived geographic coordinates and recorded in meters above mean sea level. According to the Open-Meteo documentation, elevation data are derived from the Copernicus Digital Elevation Model (Copernicus DEM; DOI: https://doi.org/10.5270/ESA-c5d3d65), which was cited as the underlying elevation data source together with attribution to Open-Meteo. All statistical analyses used altitude values derived from geocoded individual residential addresses rather than prefecture-level representative locations. Residential altitude was modeled both as a continuous variable per 500-m increase and as quartiles based on the distribution among patients with available altitude data.

For spatial visualization, Yunnan provincial and prefecture/city administrative boundaries were overlaid on a terrain background derived from a local Copernicus DEM GeoTIFF. Patient residential locations were plotted as point data using their geocoded longitude and latitude. Points were colored according to predefined categories of individual residential altitude (< 1000, 1000–1500, 1500–2000, 2000–2500, 2500–3000, and > 3000 m). The terrain layer was used only as a cartographic background. No inverse distance weighting, Kriging, or other spatial interpolation was applied to patient altitude values.

Environmental exposure assessment

Environmental exposures were assigned according to each patient’s geocoded residential coordinates and matched to the calendar year of kidney biopsy. For patients biopsied in 2023, 2024, and 2025, annual environmental exposures from the corresponding biopsy year were used. Only 3 patients in the altitude-analysis cohort underwent biopsy in 2026; because complete annual 2026 exposure data were not yet available, those patients were assigned 2025 annual exposure values as the closest complete-year proxy. Meteorological variables were retrieved from the NASA POWER Daily point API using each patient’s WGS84 latitude and longitude. The queried NASA POWER variables were T2M, RH2M, PRECTOTCORR, PS, WS2M, and ALLSKY_SFC_SW_DWN, corresponding to 2-m air temperature, 2-m relative humidity, precipitation, surface pressure, 2-m wind speed, and all-sky surface shortwave radiation. Daily values were extracted for the assigned calendar exposure year and aggregated to patient-level annual summaries. NASA POWER meteorological products are provided on their native gridded resolution, currently 0.5° × 0.625° for meteorological parameters. Air pollution-related variables were retrieved from the Open-Meteo Air Quality API using the same coordinates and the CAMS global atmospheric composition domain. The queried Open-Meteo/CAMS variables were pm2_5, pm10, carbon_monoxide, nitrogen_dioxide, sulphur_dioxide, ozone, and aerosol_optical_depth. The CAMS global air-quality domain used by Open-Meteo has an approximate spatial resolution of 0.4° (about 45 km) and a 3-hourly temporal resolution; aerosol_optical_depth represents column aerosol optical depth at 550 nm.

Distance to Kunming, the provincial capital and location of the study center, was calculated as the great-circle distance between each patient’s residential coordinates and Kunming. This variable was used as a referral-access/geographic proxy in exploratory environmental analyses because geographic proximity to the tertiary referral center may influence the probability of referral and kidney biopsy. For environmental exposure aggregation, daily NASA POWER observations and 3-hourly Open-Meteo/CAMS observations were summarized by calculating the arithmetic mean of all available non-missing observations within the assigned exposure year; cumulative annual precipitation was calculated as the sum of available daily precipitation values. Exposure extraction was performed at the geocoded coordinate level. Among the 1,593 patients included in altitude-related analyses, there were 217 unique coordinate pairs, 427 unique coordinate-year keys, and 359 unique annual environmental exposure profiles after biopsy-year matching, indicating that some patients sharing the same locality-level geocode or source grid were assigned identical exposure values. The exposure retrieval and aggregation workflow was implemented in the reproducible analysis code, which retains geocoded coordinates, exposure year, queried API variable names, exposure-source labels, and derived annual summaries. AOD was summarized to three decimal places for descriptive presentation and modeled per 0.01-unit increase in regression analyses.

Statistical analysis

Continuous variables are presented as median and interquartile range (IQR), whereas categorical variables are presented as counts and percentages. Baseline characteristics across altitude quartiles were compared using the Kruskal-Wallis test for continuous variables and the chi-square test for categorical variables. Logistic regression analyses compared biopsy-confirmed ANCA-associated renal vasculitis with all other biopsy-confirmed renal diagnoses within the hospital-based native kidney biopsy cohort. Residential altitude was modeled per 500-m increase. Accordingly, the odds ratios estimate the relative odds of receiving an ANCA-associated renal vasculitis diagnosis rather than another biopsy diagnosis among patients already selected into this single-center biopsy cohort. These odds ratios therefore describe diagnostic composition within the biopsy cohort, not population-based incidence or disease risk.

Three sequential multivariable models were constructed to limit overfitting and to separate clinical adjustment from exploratory environmental analyses. Model 1 was adjusted for age, sex, and BMI. Model 2 further adjusted for ethnicity, given its imbalance across altitude quartiles and potential role as a geographic and demographic confounder. Model 3 additionally adjusted for hemoglobin and eGFR as clinical severity-related covariates. Model 2 was considered the primary selected multivariable model, whereas Model 3 was interpreted as an additional adjustment model.

Because of the limited number of biopsy-confirmed ANCA-associated renal vasculitis cases, Firth penalized logistic regression was performed as a sensitivity analysis for the same model specifications. An additional sensitivity analysis repeated the altitude association analyses using the expanded clinical-or-pathological ANCA-associated renal vasculitis outcome.

Environmental exposure analyses were performed separately from the primary multivariable models. To evaluate whether geographic or environmental factors attenuated the altitude-ANCA diagnostic-composition association, attenuation analyses sequentially added distance to Kunming, AOD, PM2.5, CO, and combinations of these variables to the selected Model 2 reference model. Attenuation was quantified using the relative change in the log odds ratio for residential altitude compared with the Model 2 reference estimate. Variance inflation factors were calculated to assess multicollinearity among variables included in environmental exposure models. Spearman correlation coefficients were used for correlation matrices involving altitude, distance to Kunming, environmental variables, and diagnosis indicators. Pearson correlation was used only for the continuous bivariate association between residential altitude and annual mean AOD shown in the scatter plot.

Restricted cubic spline analyses were conducted as exploratory analyses to examine potential nonlinear associations between residential altitude and each major pathological diagnosis, excluding the other category. To reduce model complexity given the limited number of biopsy-confirmed ANCA-associated renal vasculitis events, exploratory restricted cubic spline analyses used a three-knot specification, with the median residential altitude used as the reference value. Diagnosis-specific altitude associations were also evaluated using false discovery rate correction to account for multiple testing. All statistical analyses were performed using Python with pandas, NumPy, SciPy, statsmodels, scikit-learn, and matplotlib. All tests were two-sided, and a P value < 0.05 was considered statistically significant.

Results

Study population

A total of 1,602 patients from the single-center renal biopsy cohort in Yunnan, China, were included. Main renal pathological diagnoses were assigned for all patients. Residential altitude data were available for 1,593 patients, whereas 9 patients were excluded from altitude-related analyses because of missing residential altitude information. The altitude-analysis cohort was used for altitude-quartile comparisons, exploratory restricted cubic spline analyses, biopsy-confirmed ANCA-associated renal vasculitis regression analyses, and environmental exposure analyses. Among these patients, 27 had biopsy-confirmed ANCA-associated renal vasculitis and 1,566 were classified as non-ANCA cases. In a sensitivity analysis using the expanded clinical-or-pathological definition, 34 patients were classified as ANCA-associated renal vasculitis cases (Fig. 1).

Fig. 1.

Study flow diagram. Flow diagram of the single-center renal biopsy cohort from Yunnan, China. A total of 1,602 patients underwent native kidney biopsy and had a main renal pathological diagnosis assigned. After exclusion of 9 patients without available residential altitude data, 1,593 patients were included in altitude-related analyses. These analyses included altitude-quartile comparisons of main pathological diagnoses, exploratory restricted cubic spline analyses by pathological diagnosis, and analyses comparing biopsy-confirmed ANCA-associated renal vasculitis with other biopsy-confirmed renal diagnoses. Among the included patients, 27 had biopsy-confirmed ANCA-associated renal vasculitis and 1,566 were classified as non-ANCA biopsy cases. A sensitivity analysis was additionally performed using an expanded clinical-or-pathological ANCA-associated renal vasculitis outcome (n = 34). Environmental exposure, attenuation, and variance inflation factor analyses were performed within this biopsy-based diagnostic-composition framework. ANCA, antineutrophil cytoplasmic antibody; VIF, variance inflation factor. This figure was created by figdraw.com

Geographic distribution and baseline characteristics

Geocoded residential locations demonstrated broad geographic coverage across Yunnan Province and spanned regions with marked variation in residential altitude (Fig. 2A). The median residential altitude was 1,884 m (IQR 1,544-1,937). Altitude quartiles were defined as Q1, 246-1,544 m; Q2, 1,546-1,884 m; Q3, 1,886-1,937 m; and Q4, 1,952-3,650 m.

Fig. 2.

Geographic distribution of patient residences, residential altitude, and renal pathological diagnoses. (A) Spatial distribution of patients included in the altitude-related analyses across Yunnan Province. Each point represents the geocoded residential location of an individual patient. Points are colored according to predefined categories of individual residential altitude derived from geocoded residential addresses. The terrain layer was used only as a cartographic background, and no IDW, Kriging, or other spatial interpolation was applied to patient altitude values. Labels indicate the number of patients from each prefecture-level region., (B) Distribution of main renal pathological diagnoses in the overall cohort and across residential altitude quartiles. The pie chart shows the overall diagnostic composition of the renal biopsy cohort, and the stacked bar chart shows the proportional distribution of major pathological diagnoses by altitude quartile. Map boundaries are shown for geographic reference and do not imply any position regarding jurisdictional claims

In the altitude-analysis cohort, the median age was 41 years (IQR 30–54), and 761 patients (47.8%) were male. Across altitude quartiles, age differed significantly (P = 0.004), as did the proportion of ethnic minority patients (P < 0.001). Hemoglobin increased across altitude quartiles, from 139 g/L in Q1 to 147 g/L in Q4 (P < 0.001). White blood cell count also differed across quartiles (P = 0.005). BMI, hypertension, diabetes, creatinine, eGFR, platelet count, and urinary protein excretion were not significantly different across altitude quartiles.

Environmental exposures differed substantially across altitude quartiles. Higher altitude quartiles were associated with lower mean temperature, lower relative humidity, lower annual precipitation, lower mean surface pressure, lower PM2.5, lower CO, and lower AOD (all P < 0.001). Distance to Kunming also differed markedly across altitude quartiles (P < 0.001) (Table 1).

Table 1.

Baseline clinical and environmental characteristics according to residential altitude quartiles

Characteristic	Overall	Q1 246–1544 m	Q2 1546–1884 m	Q3 1886–1937 m	Q4 1952–3650 m	P value
Demographics
Age, years	41.0 (30.0, 54.0)	39.0 (29.0, 52.0)	40.0 (29.0, 53.0)	43.0 (32.0, 56.0)	41.0 (29.0, 54.0)	0.004
Male sex, n (%)	761 (47.8)	181 (45.2)	194 (47.0)	206 (46.1)	180 (54.1)	0.075
Residential altitude, m	1884.0 (1544.0, 1937.0)	1266.0 (1072.8, 1454.0)	1758.0 (1634.0, 1875.0)	1910.0 (1902.0, 1937.0)	2289.0 (2115.0, 2407.0)	< 0.001
Ethnic minority, n (%)	338 (21.2)	126 (31.5)	52 (12.6)	53 (11.9)	107 (32.1)	< 0.001
Lifestyle
Current smoking, n (%)	148 (9.3)	33 (8.2)	33 (8.0)	48 (10.7)	34 (10.2)	0.425
Alcohol use, n (%)	300 (18.8)	64 (16.0)	74 (17.9)	92 (20.6)	70 (21.0)	0.234
Clinical history
BMI, kg/m²	23.6 (21.3, 26.5)	23.4 (21.2, 26.2)	23.4 (21.0, 26.5)	24.2 (21.6, 26.9)	23.3 (21.1, 26.3)	0.088
Hypertension, n (%)	683 (42.9)	162 (40.5)	176 (42.6)	204 (45.6)	141 (42.3)	0.500
Diabetes, n (%)	180 (11.3)	38 (9.5)	43 (10.4)	59 (13.2)	40 (12.0)	0.335
Blood tests
WBC, 10⁹/L	6.9 (5.7, 8.7)	7.0 (6.0, 9.0)	6.9 (5.5, 8.6)	6.8 (5.7, 8.2)	7.1 (5.8, 9.2)	0.005
Hemoglobin, g/L	143.0 (127.0, 158.0)	139.0 (124.0, 154.0)	142.5 (125.0, 158.0)	145.0 (130.5, 159.0)	147.0 (130.0, 161.0)	< 0.001
Platelet, 10⁹/L	260 (210, 310)	266 (219, 322)	259 (210, 309)	257 (208, 302)	260 (208, 315)	0.108
Kidney function
Creatinine, µmol/L	87 (66, 133)	91 (68, 142)	85 (64, 138)	84 (65, 124)	88 (68, 135)	0.171
eGFR, mL/min/1.73 m²	85.1 (51.2, 109.8)	81.4 (46.5, 109.3)	87.0 (50.5, 110.8)	84.6 (55.6, 109.0)	85.9 (50.3, 110.1)	0.436
UPE, mg/day	1517 (580, 3730)	1467 (577, 3569)	1335 (520, 3275)	1575 (565, 3908)	1680 (715, 4315)	0.053
Environmental exposure
Distance to Kunming, km	160.9 (45.8, 279.8)	260.5 (160.9, 360.2)	119.4 (79.0, 191.7)	20.7 (3.3, 20.7)	257.4 (196.0, 316.8)	< 0.001
Mean temperature, °C	16.0 (14.5, 17.1)	17.9 (16.4, 19.0)	15.8 (14.8, 17.0)	16.2 (15.3, 16.3)	14.3 (10.9, 15.0)	< 0.001
Relative humidity, %	73.4 (71.3, 75.7)	76.0 (73.7, 78.8)	73.2 (71.2, 76.7)	71.9 (71.2, 75.2)	71.5 (70.5, 74.0)	< 0.001
Annual precipitation, mm	981.6 (853.3, 1193.0)	1193.9 (904.6, 1502.7)	1099.6 (835.2, 1218.9)	941.4 (790.4, 1190.4)	969.8 (865.3, 1098.2)	< 0.001
Mean surface pressure, kPa	80.1 (79.2, 82.7)	84.5 (82.7, 86.3)	80.9 (79.2, 81.7)	79.9 (79.8, 80.4)	78.3 (72.7, 79.2)	< 0.001
PM2.5, µg/m³	14.6 (12.6, 16.9)	14.6 (13.2, 16.5)	14.6 (13.0, 16.2)	16.9 (13.9, 19.4)	11.2 (7.9, 14.6)	< 0.001
CO, µg/m³	287.9 (261.8, 327.3)	291.3 (259.5, 313.4)	281.3 (270.4, 308.4)	360.9 (283.4, 391.8)	241.5 (226.3, 286.2)	< 0.001
AOD	0.262 (0.241, 0.289)	0.302 (0.259, 0.334)	0.259 (0.241, 0.290)	0.268 (0.241, 0.280)	0.211 (0.167, 0.259)	< 0.001

Values are presented as median (interquartile range) or n (%). P values compare altitude quartiles using Kruskal-Wallis tests for continuous variables and chi-square tests for categorical variables. Environmental variables reflect the updated year-matched exposure dataset. AOD, aerosol optical depth; BMI, body mass index; CO, carbon monoxide; eGFR, estimated glomerular filtration rate; UPE, urinary protein excretion; WBC, white blood cell count

Distribution of pathological diagnoses

Primary IgA nephropathy was the most common pathological diagnosis, accounting for 581 patients (36.3%) in the total cohort. Other common diagnoses included primary membranous nephropathy in 207 patients (12.9%), tubulointerstitial disease in 178 (11.1%), primary MCD/FSGS in 121 (7.6%), lupus nephritis in 120 (7.5%), hypertensive nephropathy in 78 (4.9%), diabetic kidney disease in 75 (4.7%), and ANCA-associated renal vasculitis in 27 (1.7%) (Fig. 2B). In the altitude-analysis cohort, the overall distribution of pathological diagnoses did not differ significantly across altitude quartiles. However, ANCA-associated renal vasculitis showed a decreasing pattern with increasing altitude, accounting for 11 cases (2.8%) in Q1, 8 cases (1.9%) in Q2, 5 cases (1.1%) in Q3, and 3 cases (0.9%) in Q4 (Fig. 2B).

Exploratory restricted cubic spline analyses

Exploratory restricted cubic spline analyses showed no significant overall association between residential altitude and most major pathological diagnoses, including primary IgA nephropathy, primary membranous nephropathy, tubulointerstitial disease, primary MCD/FSGS, lupus nephritis, hypertensive nephropathy, and diabetic kidney disease. In contrast, residential altitude was associated with ANCA-associated renal vasculitis, with lower odds observed at higher altitude levels (P-overall = 0.034). There was no evidence of a nonlinear association (P-nonlinear = 0.476), supporting an approximately inverse linear pattern in the exploratory spline analysis (Fig. 3).

Fig. 3.

Restricted cubic spline analyses of residential altitude and major renal pathological diagnoses. Restricted cubic spline curves showing the exploratory associations between residential altitude and major renal pathological diagnoses: (A) primary IgA nephropathy; (B) primary membranous nephropathy; (C) tubulointerstitial disease; (D) primary MCD/FSGS; (E) lupus nephritis; (F) hypertensive nephropathy; (G) diabetic kidney disease; and (H) ANCA-associated renal vasculitis. Odds ratios were estimated using the median residential altitude as the reference. Solid lines represent estimated odds ratios, dashed lines represent 95% confidence intervals, and histograms show the distribution of residential altitude. The spline analyses used a reduced-complexity three-knot exploratory approach. ANCA-associated renal vasculitis showed an inverse overall association without evidence of nonlinearity.

Association between residential altitude and biopsy-confirmed ANCA-associated renal vasculitis

In univariable logistic regression within the biopsy cohort, each 500-m increase in residential altitude was associated with lower odds of biopsy-confirmed ANCA-associated renal vasculitis relative to other biopsy diagnoses (OR 0.57, 95% CI 0.39–0.84; P = 0.004). Older age, lower hemoglobin, higher creatinine, lower eGFR, higher relative humidity, and higher AOD were also associated with ANCA-associated renal vasculitis in univariable analyses (Supplementary Table 1).

In multivariable analysis, the inverse diagnostic-composition association between altitude and biopsy-confirmed ANCA-associated renal vasculitis remained significant after adjustment for age, sex, and BMI (Model 1: OR 0.53, 95% CI 0.36–0.78; P = 0.002). After further adjustment for ethnicity, the association remained significant and slightly strengthened (Model 2: OR 0.51, 95% CI 0.34–0.77; P = 0.001). Additional adjustment for hemoglobin and eGFR produced a similar estimate (Model 3: OR 0.51, 95% CI 0.32–0.80; P = 0.004). Firth penalized logistic regression yielded consistent results across models, including Model 2 (OR 0.51, 95% CI 0.35–0.75; P < 0.001) and Model 3 (OR 0.52, 95% CI 0.33–0.80; P = 0.003) (Table 2 and Fig. 4A).

Table 2.

Association between residential altitude and biopsy-confirmed ANCA-associated renal vasculitis

Analysis	Model	Adjustment	N	Events	OR (95% CI)	P value
Conventional logistic	Model 1	Age + sex + BMI	1593	27	0.53 (0.36, 0.78)	0.002
Firth penalized logistic	Model 1	Age + sex + BMI	1593	27	0.53 (0.36, 0.77)	< 0.001
Conventional logistic	Model 2	Age + sex + BMI + ethnicity	1593	27	0.51 (0.34, 0.77)	0.001
Firth penalized logistic	Model 2	Age + sex + BMI + ethnicity	1593	27	0.51 (0.35, 0.75)	< 0.001
Conventional logistic	Model 3	Model 2 + hemoglobin + eGFR	1593	27	0.51 (0.32, 0.80)	0.004
Firth penalized logistic	Model 3	Model 2 + hemoglobin + eGFR	1593	27	0.52 (0.33, 0.80)	0.003

Odds ratios are expressed per 500-m increase in residential altitude and compare biopsy-confirmed ANCA-associated renal vasculitis with other biopsy-confirmed renal diagnoses within the single-center renal biopsy cohort. These estimates describe biopsy-cohort diagnostic composition rather than population-based incidence or disease risk. Model 1 adjusted for age, sex, and BMI. Model 2 additionally adjusted for ethnicity. Model 3 further adjusted for hemoglobin and eGFR. Firth penalized logistic regression was used as a sensitivity analysis because of the limited number of events. ANCA, antineutrophil cytoplasmic antibody; BMI, body mass index; CI, confidence interval; eGFR, estimated glomerular filtration rate; OR, odds ratio

Fig. 4.

Multivariable and attenuation analyses for the association between residential altitude and biopsy-confirmed ANCA-associated renal vasculitis. (A) Forest plot showing the association between residential altitude and biopsy-confirmed ANCA-associated renal vasculitis relative to other biopsy diagnoses across sequential multivariable models. Model 1 adjusted for age, sex, and body mass index. Model 2 additionally adjusted for ethnicity. Model 3 further adjusted for hemoglobin and estimated glomerular filtration rate. Odds ratios are expressed per 500-m increase in residential altitude and describe biopsy-cohort diagnostic-composition odds., (B) Attenuation analysis showing changes in the altitude-associated odds ratio after sequential addition of geographic and environmental covariates to the Model 2 reference model. Log-OR change represents the relative change in the altitude-associated log odds ratio compared with the Model 2 reference. AOD, aerosol optical depth; BMI, body mass index; CO, carbon monoxide; eGFR, estimated glomerular filtration rate; OR, odds ratio; PM2.5, fine particulate matter

Attenuation analyses suggested that distance to Kunming and AOD contributed most to attenuation of the altitude-ANCA diagnostic-composition association. Compared with the Model 2 reference estimate (OR 0.51, 95% CI 0.34–0.77; P = 0.001), additional adjustment for distance to Kunming attenuated the association to OR 0.61 (95% CI 0.40–0.92; P = 0.020), corresponding to a 25.9% log-OR change. Additional adjustment for AOD attenuated the association to OR 0.63 (95% CI 0.39–1.02; P = 0.060), corresponding to a 31.6% log-OR change. In contrast, PM2.5 and CO alone produced minimal attenuation. Models including both distance to Kunming and AOD produced greater attenuation, and the full exploratory model including distance to Kunming, PM2.5, CO, and AOD showed a 56.7% log-OR change (Fig. 4B).

In the sensitivity analysis using the expanded clinical-or-pathological ANCA-associated renal vasculitis outcome, the inverse association with altitude remained directionally consistent. Each 500-m increase in residential altitude was associated with lower odds of the expanded outcome in univariable analysis (OR 0.61, 95% CI 0.43–0.86; P = 0.005), Model 2 (OR 0.54, 95% CI 0.38–0.79; P = 0.001), and Model 3 (OR 0.55, 95% CI 0.36–0.83; P = 0.005). Firth penalized logistic regression produced similar results (Supplementary Table 2).

Diagnosis-specific multiple-testing assessment

In exploratory diagnosis-specific analyses across major pathological diagnoses, ANCA-associated renal vasculitis was the only diagnosis that remained significantly associated with residential altitude after false discovery rate correction (raw P = 0.004; FDR-adjusted P = 0.033). No other major pathological diagnosis showed a statistically significant altitude association after false discovery rate correction (Supplementary Table 3).

Environmental collinearity and correlation analyses

Variance inflation factor analysis showed moderate collinearity for PM2.5 (VIF = 5.78), whereas CO (VIF = 3.35), AOD (VIF = 3.10), distance to Kunming (VIF = 1.73), altitude (VIF = 1.57), ethnicity (VIF = 1.10), age (VIF = 1.03), sex (VIF = 1.01), and BMI (VIF = 1.01) were below commonly used collinearity thresholds (Fig. 5A).

Fig. 5.

Collinearity and environmental correlation analyses. (A) Variance inflation factor analysis for variables included in the environmental exposure and attenuation models. The dashed vertical line indicates a VIF threshold of 5., (B) Spearman correlation matrix of residential altitude, distance to Kunming, climate variables, air pollution indicators, and AOD. AOD, aerosol optical depth; BMI, body mass index; CO, carbon monoxide; PM2.5, fine particulate matter; VIF, variance inflation factor

Spearman correlation analysis showed that residential altitude was inversely correlated with several environmental exposures, particularly mean surface pressure, temperature, and AOD. AOD was positively correlated with PM2.5 and CO, indicating shared spatial environmental patterns among pollution-related variables (Fig. 5B).

AOD analyses

Residential altitude was inversely correlated with annual mean AOD in Pearson correlation analysis (r=-0.57, P < 0.001) (Fig. 6A). Spatial mapping demonstrated opposing geographic patterns between residential altitude and AOD across Yunnan Province. Biopsy-confirmed ANCA-associated renal vasculitis cases were visualized on the AOD terrain map according to their individual residential altitude categories (Fig. 6B).

Fig. 6.

AOD analyses in relation to residential altitude and ANCA-associated renal vasculitis. (A) Scatter plot showing the association between residential altitude and annual mean AOD, with fitted regression line and 95% confidence band. Pearson correlation was used for this continuous bivariate analysis. (B) Spatial distribution of annual mean AOD across Yunnan Province with biopsy-confirmed ANCA-associated renal vasculitis cases overlaid. ANCA cases are colored according to predefined categories of individual residential altitude. Map boundaries are shown for geographic reference and do not imply any position regarding jurisdictional claims. AOD, aerosol optical depth

Discussion

In this single-center renal biopsy cohort from Yunnan, residential altitude was not significantly associated with most major renal pathological diagnoses. In contrast, biopsy-confirmed ANCA-associated renal vasculitis demonstrated an exploratory inverse diagnostic-composition association with higher residential altitude. The association remained directionally consistent in multivariable logistic regression, Firth penalized logistic regression, and an expanded clinical-or-pathological outcome sensitivity analysis. However, the altitude-associated estimate was attenuated after additional adjustment for distance to Kunming and AOD, suggesting that the observed pattern may be embedded within broader geographic, environmental, and referral-related gradients.

A key interpretive issue is the choice of comparator. The non-ANCA group in this study consisted of patients with IgA nephropathy, membranous nephropathy, MCD/FSGS, lupus nephritis, diabetic kidney disease, tubulointerstitial disease, hypertensive nephropathy, and other biopsy diagnoses, rather than population-based or disease-free controls. Therefore, the observed odds ratio describes the relative odds of ANCA-associated renal vasculitis among patients undergoing kidney biopsy at our center. It could be influenced by altitude-related differences in the frequency of other biopsy diagnoses, the threshold for kidney biopsy, or referral pathways, even if the underlying occurrence of ANCA-associated vasculitis in the source population were unchanged.

The largely null findings for most renal pathological diagnoses should first be interpreted in the context of the heterogeneous nature of biopsy-confirmed kidney disease. In our cohort, IgA nephropathy was the predominant diagnosis, whereas membranous nephropathy, MCD/FSGS, lupus nephritis, diabetic kidney disease, hypertensive nephropathy, and tubulointerstitial disease contributed variable proportions, broadly consistent with previous Chinese biopsy registries [1, 7]. Similar regional variability has also been reported in international biopsy registries, where differences in renal pathological spectra were influenced by biopsy indications, population structure, and healthcare practice patterns [2, 8]. Against this background of substantial pathological heterogeneity, the specific association observed for ANCA-associated renal vasculitis becomes particularly noteworthy.

The progressive increase in hemoglobin across altitude quartiles is biologically plausible and consistent with previous studies demonstrating altitude-related erythropoietic adaptation, thereby providing an internal validity check for the residential altitude exposure metric [9, 10]. The exploratory spline analysis did not show statistically significant evidence of a nonlinear association between residential altitude and tubulointerstitial disease. Nevertheless, a potential biological link warrants further investigation, given the sensitivity of tubulointerstitial injury to impaired oxygen delivery and hypoxia-related remodeling [3, 11]. Our previous IgA nephropathy validation study also suggested that altitude-related physiology may influence the relationship between clinical biomarkers and tubulointerstitial lesions [12].

The inverse diagnostic-composition association observed for ANCA-associated renal vasculitis deserves attention because ANCA-associated vasculitis has a distinct immunopathological basis compared with many other renal pathological diagnoses. ANCA-associated vasculitis is a pauci-immune necrotizing small-vessel vasculitis characterized by ANCA-mediated neutrophil activation, complement pathway amplification, endothelial injury, and crescentic glomerulonephritis. Both innate and adaptive immune mechanisms contribute to disease pathogenesis [13, 14]. Epidemiological and geoepidemiological studies suggest that ANCA-associated vasculitis has substantial geographic and environmental heterogeneity across populations. Environmental exposures including occupational silica exposure, air pollution, infectious triggers, seasonal variation, pollen and other airborne particles, latitude-related patterns, and ultraviolet radiation exposure have all been proposed as potential contributors to disease susceptibility [15, 16]. A systematic mapping review also identified environmental exposures, including pollution-related factors, as important but incompletely characterized domains in ANCA-associated vasculitis [5]. In addition, regional clustering of MPO-ANCA-associated vasculitis has been reported following major environmental disturbances, supporting the possibility that environmental factors may influence disease patterns in susceptible populations [17].

In the environmental analyses, AOD and distance to Kunming produced the largest attenuation of the altitude-ANCA association. AOD is a satellite-derived measure of columnar aerosol loading and has been widely used in environmental epidemiology as a spatially resolved indicator related to particulate air pollution, particularly PM2.5, although this relationship varies by region, humidity, vertical aerosol distribution, and aerosol composition [18–20]. Therefore, AOD should be interpreted as an integrated marker of regional aerosol burden rather than a single pollutant or direct causal exposure. The attenuation observed after adding AOD and distance to Kunming suggests that residential altitude may partly capture broader spatial environmental and referral-access patterns. These attenuation findings are best viewed as exploratory evidence of overlapping geographic exposure patterns rather than proof of mediation by AOD.

Several potential explanations may be considered, including altitude-related differences in hypoxia adaptation, innate immune responses, infection exposure, occupational exposure, air quality, airborne particulate exposure, ethnicity, lifestyle, and healthcare access. In the present study, however, the inverse association between altitude and ANCA-associated renal vasculitis was attenuated after adjustment for AOD and distance to Kunming, suggesting that altitude may partly represent a broader spatial exposure gradient rather than an isolated biological factor. Distance to Kunming was included specifically as an exploratory referral-access proxy, because patients living farther from the provincial capital may differ in referral probability, diagnostic delay, access to tertiary nephrology care, and likelihood of undergoing kidney biopsy. Kunming, the provincial capital of Yunnan and the location of the study center, serves as a major tertiary referral hub for patients from geographically diverse regions across the province. The observed geographic distribution of biopsy-confirmed renal diseases may therefore be influenced by referral structure, regional nephrology resources, transportation accessibility, healthcare access, and variability in kidney biopsy practice across clinical settings. Because ANCA-associated vasculitis is often associated with severe systemic manifestations and rapidly progressive kidney injury, referral patterns may differ from those of more indolent diseases such as IgA nephropathy. Previous studies have demonstrated substantial regional variation in kidney biopsy practice and have shown that geographic remoteness may reduce access to specialist nephrology care and related diagnostic evaluation [21, 22]. Thus, referral and ascertainment bias cannot be excluded.

The statistical approach supports the internal consistency of the finding while acknowledging sparse events and exploratory inference. Because only 27 biopsy-confirmed ANCA-associated renal vasculitis cases were available, conventional logistic regression may be vulnerable to small-sample bias and instability, and overly complex models or extensive subgroup analyses are inappropriate. We therefore used sequentially adjusted models, treated the ethnicity-adjusted model as the selected primary multivariable model, performed Firth penalized logistic regression as a sensitivity analysis, and used reduced-complexity restricted cubic spline analyses only for exploratory visualization. The consistency between standard multivariable models and penalized logistic regression provides supportive but not definitive evidence that the altitude association was not solely a modeling artifact [23].

This study has several limitations. First, the retrospective single-center design limits generalizability and prevents causal inference. Second, this biopsy-based cohort reflects patients who underwent kidney biopsy and cannot be interpreted as a population-incidence study or a measure of disease risk in the general population. Similarly, biopsy-confirmed diabetic kidney disease and hypertensive nephropathy should not be interpreted as reflecting the clinical burden of diabetes- or hypertension-related chronic kidney disease in the source population, because many such patients may not undergo kidney biopsy. Third, the non-ANCA comparator group consisted of patients with other biopsy-confirmed renal diseases rather than population-based, disease-free, or etiologically neutral controls. Therefore, the reported odds ratios may reflect changes in the numerator of ANCA-associated renal vasculitis, changes in the denominator of other biopsy diagnoses, or altitude-related differences in referral and biopsy practice. Fourth, the number of biopsy-confirmed ANCA-associated renal vasculitis cases was small, with only 27 events, which limits precision and precludes reliable subgroup analyses. Fifth, residential altitude was estimated from permanent residence and may not reflect lifetime exposure, migration, occupational location, indoor exposure, or duration of residence. Sixth, environmental exposures were assigned from geocoded residential coordinates using annual mean values matched to biopsy year; this approach does not capture short-term exposure windows, seasonal variation, indoor exposures, or personal activity patterns. For patients biopsied in early 2026, 2025 annual mean exposure values were used as the closest complete-year proxy because complete annual 2026 data were unavailable. Seventh, participants sharing the same locality-level geocode may have been assigned identical or similar gridded environmental exposure values, and within-location clustering could not be fully modeled because of the small number of ANCA-associated renal vasculitis events. Eighth, residual confounding by ethnicity, socioeconomic status, healthcare access, occupational exposures, infection history, medication exposure, ANCA serotype, environmental mixtures, and referral patterns could not be excluded. Although the expanded clinical-or-pathological ANCA sensitivity analysis supported the direction of the main finding, clinical diagnoses may be subject to heterogeneity in documentation and diagnostic certainty.

Conclusion

This biopsy-based Yunnan cohort identified no broad altitude-related shift in major renal pathological diagnoses, but suggested an exploratory inverse association between residential altitude and the relative odds of biopsy-confirmed ANCA-associated renal vasculitis compared with other biopsy diagnoses. The association was attenuated after adjustment for AOD and distance to Kunming, indicating that altitude may partly reflect broader spatial environmental and referral-related gradients rather than an isolated biological exposure. These findings require validation in larger multicenter or population-based studies with detailed environmental, clinical, immunological, and healthcare-access data.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

AOD

Aerosol optical depth

ANCA

Antineutrophil cytoplasmic antibody

BMI

Body mass index

CAMS

Copernicus atmosphere monitoring service

CKD-EPI

Chronic kidney disease epidemiology collaboration

CO

Carbon monoxide

DEM

Digital elevation model

eGFR

estimated glomerular filtration rate

FDR

False discovery rate

IQR

Interquartile range

MCD/FSGS

Minimal change disease/focal segmental glomerulosclerosis

PM2.5

fine particulate matter

PM10

Particulate matter with an aerodynamic diameter ≤ 10 μm

RCS

Restricted cubic spline

VIF

Variance inflation factor

WGS84

World geodetic system 1984

Author contributions

Research idea and study design: Xinyu Wang, Jian Xu and Qinyuan Deng; data acquisition: Meiyu Chen, Qiao Wang, Yaling Li, Chunli Liu, Yaling Yu and Feifei Liu; data analysis, interpretation, and statistical analysis: Xinyu Wang, Jian Xu and Qinyuan Deng; manuscript drafting: Xinyu Wang and Qinyuan Deng; critical revision of the manuscript for important intellectual content: Masashi Mukoyama, Yutaka Kakizoe, Ying Shen and Qinyuan Deng; supervision and overall study oversight: Ying Shen and Qinyuan Deng. All authors reviewed and approved the final manuscript.

Funding

This work was supported by the First People’s Hospital of Yunnan Province, National Priority Clinical Specialty Cultivation Project for Nephrology (2026ZDZK-SZ01, 2026ZDZK-SZ02) and the Special and Joint Program of the Yunnan Provincial Science and Technology Department and Kunming Medical University (Grant No. 202601AY070001-083).

Data availability

Analysis scripts, data dictionaries, and templates are provided in Supplementary Material 2. The deidentified dataset and local geospatial files are available from the corresponding author upon reasonable request and appropriate institutional and ethical approval.

Declarations

Ethics approval and consent to participate

The study protocol was approved by the Ethics Committee of The First People’s Hospital of Yunnan Province (approval number: KHLL2025-KY117) and was conducted in accordance with the principles of the Declaration of Helsinki. Written informed consent was obtained from all patients.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.