1.1 Clinical Characteristics
This study retrospectively included inpatients with hypertension aged 18–60 years admitted to a local hospital from January 2017 to December 2020. All patients underwent an enhanced abdominal CT scan and laboratory tests during their hospitalization. Patients with a history of urinary tract infection (n=29), urinary calculus (n=42), malignant tumor or autoimmune disease because of an unknown kidney injury induced by long-term drugs (n=159), diabetes (n=52), renal congenital variation such as lobulated kidney, ectopic kidney, abnormal renal rotation, polycystic kidney (n=4), renal masses or cysts >1 cm in diameter (n=28), asymmetrical kidneys (n=9), or renal artery stenosis (n=7) were excluded (Figure 1). Clinical data collected included age, sex, hypertension grade on admission, hypertension course and poorly controlled hypertension. Poorly controlled hypertension was defined as systolic blood pressure [SBP]≥140 mmHg and/or diastolic blood pressure [DBP]≥90 mmHg while taking hypertension medication. Laboratory tests on admission included cystatin C, serum creatinine, serum urea nitrogen, total cholesterol (TC), triglycerides (TG), and low-density lipoprotein (LDL) level assessment. Hypertension and classification of blood pressure levels were defined according to the 2018 edition of the Chinese hypertension guidelines [12]. Hypertension course was divided into three categories: 0–9 years, 10–19 years, and greater than or equal to 20 years.
Cystatin C is typically used as an indicator of early renal injury in clinical situations [13, 14]. Therefore, the patients were divided into an ERI group or a CP group according to the following criteria for ERI: cystatin C > 1.02 mg/L, creatinine ≤ 127 μmol/L, and urea nitrogen ≤ 8.3 mmol/L.
1.2 Imaging Acquisition
Several CT equipments (GE Optima 64, Toshiba Aquilion One 320, Siemens Sensation 16, and Somatom Definition Flash) were used at a single medical center. The CT scan parameters were as follows: tube voltage 120 kV, tube current 10 mA - 300 mA, slice thickness 5 mm, slice increment 5 mm, field of view 35 cm ~ 40 cm, matrix 512×512. The CT scan of the corticomedullary phase was conducted at 30–35 s after starting iopromide contrast (Ultravist 350 or 370, Bayer Schering Pharma, Berlin, Germany) injection at an injection flow rate of 2.5–5.0 mL/s and a dose of 1.0–1.5 mL/kg.
1.3 Kidney Segmentation and Quantitation of Renal Surface Nodularity
CT image data at the corticomedullary phase (DICOM files) were analyzed by a radiologist (Kaixiang Wang). Twenty of 143 cases were sampled randomly and analyzed by another radiologist (Tao Wang) to assess interrater agreement of qRSN. The automated algorithm in ITK SNAP (version 3.8, www.itksnap.org) was employed to segment the kidney, as shown in Figure 2. A 3D surface mesh of the segmented kidney was generated as indicated in a previous study [15], and point coordinates were adjusted by using a windowed sinc function interpolation kernel [16]. The Euclidean distance between the generated 3D and the smoothed 3D surface mesh was computed. The median Euclidean distance was used to quantify qRSN and was normalized to the minimum value (Figure 3).
1.4 Statistical Analysis
All analyses were performed using SPSS or Medcalc. Categorical variables are expressed as frequencies (%). Continuous variables are expressed as the mean ± standard deviation. The Bland-Altman test was performed to assess interrater agreement of qRSN. The linear relationship between age and qRSN was tested by Pearson's correlation. Variables between the ERI and CP groups were compared using c2, Wilcoxon rank sum, or two independent sample t tests, as appropriate. To avoid eliminating potentially meaningful variables, P<0.1 was considered statistically significant. Multiple logistic regression analysis was applied to explore the relationship of qRSN with ERI, and P<0.05 was considered statistically significant. A variance inflation factor value <10 was considered to indicate no obvious collinearity.