Analysis of associated factors and construction of prediction model for primary Sj?gren′s syndrome complicated with renal damage
10.3760/cma.j.cn141217-2022110-00464
- VernacularTitle:原发性干燥综合征合并肾损害关联因素分析及预测模型构建
- Author:
Qingxiu FENG
1
;
Ying ZHANG
;
Bingzhu HUA
;
Hong WANG
;
Xuebing FENG
Author Information
1. 南京大学医学院附属鼓楼医院风湿免疫科，南京 210008
- Keywords:
Sj?gren′s syndrome;
Renal damage;
Predictive models
- From:
Chinese Journal of Rheumatology
2023;27(10):666-672
- CountryChina
- Language:Chinese
-
Abstract:
Objective:To investigate the key factors of renal damage in patients with primary Sj?gren′s syndrome (pSS), to construct a risk prediction model.Methods:Clinical data of 419 pSS inpatients in the Nanjing Drum Tower Hospital from 2017 to 2020 were retrospectively analyzed. The patients were randomly divided into the model group (315 cases) and the validation group (104 cases) in a 3∶1 ratio by R software randomized package. T test was used in accordance with the normal measurement data, Chi square test was used for counting data, and Kruskal-Wallis H and Mann-Whitney U test were used for non-normal distributed data. A nomogram model was created based on independent factors of renal damage by using LASSO Cox regression analysis, and the receiver operating characteristic (ROC) curve, calibration curve and decision curve analysis (DCA) were applied to evaluate the sensitivity, accuracy and clinical practicability of the model. Results:Renal impairment occurred in 127 (30.3%)pSS patients, including 89 (70.1%) cases with renal insufficiency, 86 cases with hematuria (67.7%), 107 cases with proteinuria (84.2%) and 49 cases with renal tubular acidosis. Compared with those without renal damage, patients with renal damage had a longer course of disease [3.00 (1.00, 8.00) years vs 1.00 (0.30, 5.00) years; Z=2.33, P=0.005], presence of fatigue [35.1% (33/94) vs 23.5% (52/221), Z=4.49, P=0.038], decreased blood cells count [white blood cell 4.50(3.30, 6.75)×10 9/L vs 5.40(3.70, 8.05)×10 9/L, Z=2.02, P=0.043; hemoglobin 99.00(79.00, 111.00)g/L vs 120.00 (109.00, 128.00)g/L, Z=6.59, P<0.001; platelet 152.00 (89.25, 204.25)×10 9/L vs 188.00 (117.99, 241.00)×10 9/L, Z=2.61, P=0.009], stronger inflammatory reactions [ESR 48.50 (29.75, 86.25)mm/1 h vs 26.00 (11.00, 52.00)mm/1 h, Z=5.83, P<0.001; CRP 5.80 (3.40, 18.45)mg/L vs 4.40 (2.60, 11.40) mg/L, Z=2.33, P=0.020] and higher positive rate for anti-SSA/B antibodies [79.79%(75/94) vs 61.99%(137/221), χ2=9.49, P=0.002; 37.23%(35/94) vs 18.10%(40/221), χ2=13.31, P<0.001], but relatively less pulmonary involvement [20.2%(19/94) vs 33.9%(75/221), χ2=5.93, P=0.016], the differences were statistically significant ( P<0.05). Lasso binary logistics regression analysis showed that EULAR Sj?gren′s syndrome disease activity index score >9[ OR(95% CI)=9.019(3.294, 24.689), P<0.001], dry eye[ OR(95% CI)=2.853(1.502, 5.422), P=0.011], anemia[ OR(95% CI)=3.819(1.913, 7.626), P<0.001], low complement C3 level[ OR(95% CI)=2.453(1.233, 4.879), P=0.011]and hypoalbuminemia [ OR(95% CI) =6.898 (3.007, 15.821), P<0.001] as well as C-reactive protein [ OR (95% CI)=2.168 (1.136, 4.139), P=0.019] were significant related factors of renal impairment. The AUC (95% CI) of the prediction group and the validation group were 0.851 (0.806, 0.896) and 0.832 (0.742, 0.922), respectively, and the model plot and DCA showed that the model had good specificity and clinical efficacy. Conclusion:We propose a new nomogram model with good differentiation and calibration to assist in clinical screening of renal damage in pSS.
