#614 Identifying distinct patient phenotypes in patients with chronic kidney disease

Abstract

Abstract Background and Aims Patients with chronic kidney disease (CKD) are heterogeneous in terms of age, causes of CKD, and comorbidities. Although individual risk factors greatly influence patients’ risk of clinical outcomes, this heterogeneity is rarely used to guide clinical decision-making. In this study we aimed to derive phenotypes in patients with CKD based on characteristics commonly available in routine clinical practice and evaluated whether these phenotypes were associated with clinical outcomes. Method Patients from the Utrecht Patient Oriented Database (UPOD) who were 18 years or older with CKD (defined as an eGFR <60 ml/min/1.73 m2 or an UACR > 3 mg/mmol present for >3 months) and visited one of the cardiovascular outpatient clinics of the University Medical Center Utrecht between 2012 and 2019 were included. Patient phenotypes were derived with latent class analyses based on age, sex, CKD stage, cause of kidney disease, and (cardiovascular) comorbidities. The incidence rate/100 person years (IR/100PY) for all-cause mortality and kidney failure (defined as initiation of kidney replacement therapy or a sustained eGFR <15) and 95% confidence interval (95% CI) was calculated for each phenotype. Results Overall 9005 patients were included (mean age 65 ± 14 years, 42% women, mean estimated glomerular filtration rate (eGFR) 49 ± 17 ml/min/1.73 m2). The median follow-up was 3.8 years [interquartile range 1.8–6.4]. Latent class analysis identified 7 phenotypes with the following characteristics: phenotype 1 were young patients with primary kidney disease (10%); phenotype 2 were patients with diabetes and a preserved eGFR (9%); phenotype 3 were patients with low comorbidity and inflammation (28%); phenotype 4 were older patients with a history of cardiovascular disease (27%); phenotype 5 were older patients with a history of cardiovascular disease and severe CKD (16%); phenotype 6 were younger patients with a kidney transplant and low comorbidity (6%); and phenotype 7 were older patients with a kidney transplant and cardiovascular disease (6%). The risk of all-cause mortality and kidney failure differed significantly between the identified phenotypes. The phenotypes of younger kidney transplant recipients, younger patients with primary kidney disease, and diabetes with preserved eGFR had the lowest risk for all-cause mortality (for all, IR/100 PY <5) (Fig. 1). The phenotypes with an intermediate incidence rate for all-cause mortality were the low comorbidity and inflammation phenotype (IR/100PY 7.4, 95% CI 6.9–8.0), the phenotype of older patients with cardiovascular disease (IR/100PY 7.2, 95% CI 6.7–7.8), and the phenotype of older patients with a kidney transplant and cardiovascular disease (IR/100PY 6.5, 95% CI 5.5–7.6). The phenotypes with highest incidence rate for all-cause mortality was the phenotype of older patients with cardiovascular disease and severe CKD (IR/100PY 12.5, 95% CI11.5–13.4). The phenotypes with the lowest risk for kidney failure were the diabetes phenotype, the low comorbidity and inflammation phenotype, and the phenotype of older patients with cardiovascular disease (all IR/100 PY <1). The phenotype of older patients with cardiovascular disease and severe CKD also had the highest risk to develop kidney failure (IR/100PY 5.8, 95% CI 5.1–6.6), followed by the phenotype of patients with primary kidney disease (IR/100PY 4.4, 95% CI 3.7 −5.5). Conclusion We identified 7 different phenotypes of patients with CKD including 2 phenotypes of patients with a kidney transplant with different clinical characteristics and prognosis. The differences in risk of all-cause mortality and kidney failure warrants a different monitoring and follow-up across the different phenotypes within the setting of CKD. Future studies should invest In evaluating the implementation of our identified phenotypes in routine clinical practice or clinical trial design with respect to differences in treatments.