Study design, setting, and population
We retrospectively reviewed patients with LN at Okayama University Hospital. Data from 1990-2006 were collected from paper-based records and data from 2007-2015 were collected from electronic-based records. Data collection was completed in 2016-2017. All of the enrolled patients fulfilled the 1997 American College of Rheumatology revised criteria for the classification of SLE . Patients were eligible for participation in this study if they had a histologically proven diagnosis of LN (class III, IV and V) meeting the International Society of Nephrology/Renal Pathology Society (ISN/RPS) classification . Eligible patients were followed up from their first renal biopsy for 10 years until December 2015.
The following information was collected at the time of the renal biopsy, before treatment: age, sex, SLE disease activity index 2000, daily maximum dose of prednisone, use of immunosuppressants, SCr and eGFR levels, urinary protein excretion (g/gCr), hematuria (dipstick test >2+, and >5 erythrocytes per high power field), and active urine sediments. The eGFR was evaluated by the equation defined by the Japanese Society of Nephrology .
For all participants, the histological result of the first renal biopsy sample was classified according to the ISN/RPS classification by experienced nephrologists and/or pathologists. Active glomerular lesions were defined by endocapillary hypercellularity, leukocyte infiltration, subendothelial hyaline deposit, interstitial inflammation, karyorrhexis, fibrinoid necrosis, and cellular crescent, and active interstitial lesions were defined by monocellular infiltration. Chronic glomerular lesions were also defined by glomerular sclerosis, fibrosis adhesion, and fibrous crescent, whereas chronic interstitial lesions were defined by interstitial fibrosis. Arteriosclerosis was also defined by chronic lesions. Each renal biopsy sample was processed using light and immunofluorescence microscopy with standard methods of fixation and staining. For semiquantitative analysis, the histological score was calculated as described in our previous study , where histological score = (0.5 × number of glomeruli with segmental lesions + 1 × number of glomeruli with global lesions)/total number of glomeruli. Interstitial lesions such as interstitial fibrosis, arteriosclerosis, and monocellular infiltration to interstitial, tubular, and vascular lesions were semi-quantitatively graded on a scale of 0, 1, 2, or 3 (absent, mild, moderate, or severe, respectively). Interstitial lesions were categorized as being at a high or low grade according to a cutoff score of 2.
The primary outcome measures were the urinary protein and SCr levels. The secondary outcome measure was the cumulative renal survival rates from the date of the renal biopsy for 10 years. The renal endpoint was defined as >40% decline in eGFR.
Statistical analyses in this study were performed using the Statistical Package of JMP® 14 (SAS Institute Inc., Cary, NC, USA) and STATA v15 (StataCorp, College Station, TX, USA). All statistical tests were 2-sided; p <0 .05 was considered statistically significant in this study. Complete case analysis was performed, excluding patients with missing clinical data at the time of the first biopsy. The descriptive statistics were expressed as mean and standard deviation (SD) for continuous variables and as n (%) for categorical variables. The cumulative renal survival rates were calculated using the Kaplan-Meier analysis. We censored patients that did not reach the renal endpoint when they completed the 10-year follow-up or at the date of the last recorded visit until December 31, 2015, and we calculated the number at risk for reaching the endpoint from the date of the renal biopsy. Survival curves were compared among the patients divided by qualities of urinary protein level and eGFR using log-rank tests.
Subsequently, multiple linear regression (ordinary least squares regression) analysis was performed to explore whether the histological findings contribute to urinary protein and SCr levels. The primary dependent variables were urinary protein and SCr levels at the time of renal biopsy, which were recorded as continuous variables, and the candidate variable was the scored renal histological findings. Urinary protein levels were log-transformed to fulfill the assumption of a normal distribution of the residuals. To address the issue with multicollinearity, which was assessed using variance inflation factor (VIF) [18, 19], we analyzed our data as two separate models excluding highly correlated covariates. As sensitivity analysis, we also performed same multiple linear regression analysis including age and sex.