Nomogram to Predict Rapid Kidney Function Decline in Population at Risk of Cardiovascular Disease

Qiuxia Zhang, Junyan Lu, Li Lei, Guodong Li, H. Liang, Jingyi Zhang, Yun Li, Xiangqi Lu, Xinlu Zhang, Yaode Chen, Jiazhi Pan, Yejia Chen, Xinxin Lin, Xiaobo Li, Shiyu Zhou, Shengli An, Jiancheng XiuResearch 2021 · DOI

Chronic Kidney Disease (CKD) represents a significant global health burden, particularly due to its close association with cardiovascular disease (CVD). Patients with hypertension or diabetes are at a heightened risk of "Rapid Kidney Function Decline" (RKFD), a condition linked to increased mortality and heart failure. While many existing models focus on predicting the onset of CKD in healthy individuals, they often lose utility once a patient already has mild kidney impairment. This research presents a practical clinical tool—a nomogram—designed to predict the 5-year risk of RKFD specifically in populations already at risk for cardiovascular disease.

Methodology and Data

The study utilized an ambispective cohort design, analyzing data from 8,455 subjects who underwent community annual health examinations in Guangzhou, China, between 2015 and 2020. Participants included individuals aged 65 or older, or those aged 35 and older with existing diabetes or hypertension.

The primary endpoint, RKFD, was defined as a reduction in the estimated glomerular filtration rate (eGFR) of 40% or more during the follow-up period. To develop the predictive model, the researchers randomly assigned participants to a development cohort (n = 5,645) and a validation cohort (n = 2,810). Cox regression analysis and a backward stepwise approach were employed to identify the most significant clinical predictors from a wide array of laboratory and demographic variables.

A Practical Tool for Risk Assessment

The final predictive model was distilled into a graphical nomogram. This tool allows clinicians to assign points based on five readily accessible clinical markers: age, baseline eGFR, hemoglobin levels, systolic blood pressure (SBP), and diabetes status. The cumulative score corresponds to a specific 5-year probability of experiencing a rapid decline in renal function.

Nomogram for predicting the 5-year risk of rapid kidney function decline (RKFD).

Key Results and Validation

During a median follow-up period of 3.72 years, the incidence of RKFD was approximately 12% across the entire cohort. The nomogram demonstrated robust predictive power, achieving an Area Under the Curve (AUC) of 0.763 in the development cohort and 0.740 in the validation cohort.

Crucially, the study identified a total nomogram score of 150 as an optimal threshold for risk stratification. Patients scoring above 150 were classified as "high-risk," with a 47.98% chance of rapid decline over five years, compared to just 11.90% for those in the "low-risk" group (score ≤ 150).

Clinical Insights

Several key insights emerged from the modeling process:

  • The "Big Five" Predictors: The selection of age, eGFR, hemoglobin, SBP, and diabetes ensures the tool is highly applicable in primary care settings where advanced biomarkers may not be available.
  • The Significance of Hemoglobin: Lower hemoglobin levels (anemia) were identified as a significant independent risk factor for RKFD, likely due to chronic hypoxia and oxidative stress within the kidneys.
  • Personalized Metrics: By using a percentage-based decline (≥ 40%) rather than an absolute numerical drop, the model provides a more personalized assessment that accounts for a patient's unique baseline health.

Limitations and Future Directions

While the nomogram is a powerful tool for early intervention, the authors acknowledge certain limitations. The dataset lacked baseline albuminuria (protein in the urine) data, which is a traditional marker of kidney damage and might have further enhanced the model's accuracy. Additionally, as the study was conducted on a South Chinese population, the nomogram requires external validation in other ethnic and geographic groups to ensure its global generalizability. Future research may also explore the integration of novel genomic data or biomarkers to further refine these predictions.

Summary updated

Abstract

Abstract Background To develop a simple model to predict risk of rapid kidney function decline (RKFD) in population at risk of cardiovascular disease. Methods 8455 subjects aged ≥ 65 years or complicated diabetes or hypertension undergoing community annual health examinations between January 2015 and December 2020 were included. All participants were randomly assigned to a development cohort and a validation cohort in a 2:1 ratio. Rapid kidney function decline was defined as the reduction of estimated glomerular filtration rate (eGFR)≥40% during follow-up period. Cox regression analysis and stepwise approach were used to identify the risk factors. A nomogram based on these predictors was then developed, and discrimination, calibration and decision curve analysis were assessed. Results During the median follow-up period of 3.72 years, the incidence of rapid kidney function decline was 11.96% (n = 1011), 11.98% (n = 676) and 11.92% (n = 335) in the entire cohort, development cohort and validation cohort, respectively. Age, eGFR, hemoglobin, systolic blood pressure, and diabetes were identified as predictors for RKFD. The nomogram demonstrated a good discriminative power with the 5-year AUCs of 0.763 and 0.740 in the development and the validation cohort, respectively. Calibration plots also demonstrated a good fitness between the observed and predicted risk in both cohorts. Conclusions Risk stratification for rapid kidney function decline is achievable using a risk prediction nomogram based on clinical factors that are readily accessible in primary care. The utility of this nomogram in identifying individuals at high risk of RKFD in the community needs further investigation.
BibTeX
@inproceedings{zhang2021nomogram,
  author = {Qiuxia Zhang and Junyan Lu and Li Lei and Guodong Li and H. Liang and Jingyi Zhang and Yun Li and Xiangqi Lu and Xinlu Zhang and Yaode Chen and Jiazhi Pan and Yejia Chen and Xinxin Lin and Xiaobo Li and Shiyu Zhou and Shengli An and Jiancheng Xiu},
  title = {{Nomogram to Predict Rapid Kidney Function Decline in Population at Risk of Cardiovascular Disease}},
  booktitle = {Research},
  year = {2021},
  doi = {10.21203/rs.3.rs-576079/v1},
}
Figure