Abstract：
OBJECTIVE: To systematically evaluate the diagnostic performance of procalcitonin (PCT) and C-reactive protein (CRP) for distinguishing bacterial infections from lupus flares in systemic lupus erythematosus (SLE) via meta-analysis.
METHODS: Electronic databases were comprehensively searched. The pooled standard mean difference (SMD) and 95% confidence interval (CI) were calculated to estimate the differences of serum PCT and CRP levels between bacterial infections and flares in SLE. Sensitivity, specificity and summary receiver operating characteristics (SROC) curve were used to assess the diagnostic values of PCT and CRP. The use of fixed or random effects model depended on heterogeneity.
RESULTS: Fifteen studies were included in the analysis. Serum PCT and CRP levels were significantly higher in SLE patients with bacterial infections compared to SLE patients with flares (PCT: SMD = 1.035, 95 %CI = 0.708 to 1.362; CRP: SMD = 1.000, 95 %CI = 0.758 to 1.242). The overall sensitivity, specificity, area under the SROC curve, positive likelihood ratios (PLR) and negative likelihood ratios (NLR) of PCT were 0.62, 0.88, 0.862, 6.63 and 0.36, respectively, while the same indicators for CRP were 0.72, 0.70, 0.784, 2.45 and 0.38, respectively.
CONCLUSIONS: Serum PCT and CRP levels were significantly increased in SLE with bacterial infections. PCT had a better diagnostic performance than CRP. PCT had a high value of PLR and could serve as a rule-in marker, while CRP testing may result in a high false-positive rate due to low PLR; both markers had a suboptimal value of NLR and are not appropriate for ruling out bacterial infections.
METHODS: Electronic databases were comprehensively searched. The pooled standard mean difference (SMD) and 95% confidence interval (CI) were calculated to estimate the differences of serum PCT and CRP levels between bacterial infections and flares in SLE. Sensitivity, specificity and summary receiver operating characteristics (SROC) curve were used to assess the diagnostic values of PCT and CRP. The use of fixed or random effects model depended on heterogeneity.
RESULTS: Fifteen studies were included in the analysis. Serum PCT and CRP levels were significantly higher in SLE patients with bacterial infections compared to SLE patients with flares (PCT: SMD = 1.035, 95 %CI = 0.708 to 1.362; CRP: SMD = 1.000, 95 %CI = 0.758 to 1.242). The overall sensitivity, specificity, area under the SROC curve, positive likelihood ratios (PLR) and negative likelihood ratios (NLR) of PCT were 0.62, 0.88, 0.862, 6.63 and 0.36, respectively, while the same indicators for CRP were 0.72, 0.70, 0.784, 2.45 and 0.38, respectively.
CONCLUSIONS: Serum PCT and CRP levels were significantly increased in SLE with bacterial infections. PCT had a better diagnostic performance than CRP. PCT had a high value of PLR and could serve as a rule-in marker, while CRP testing may result in a high false-positive rate due to low PLR; both markers had a suboptimal value of NLR and are not appropriate for ruling out bacterial infections.
摘要：
目的：通过Meta分析系统评价降钙素原（PCT）和C反应蛋白（CRP）对系统性红斑狼疮（SLE）细菌感染与狼疮发作的鉴别诊断价值。
方法：全面检索电子数据库。计算合并标准平均差（SMD）和95％置信区间（CI），以估计SLE细菌感染和耀斑之间血清PCT和CRP水平的差异。灵敏度,使用特异性和汇总受试者工作特征(SROC)曲线评估PCT和CRP的诊断价值.固定或随机效应模型的使用取决于异质性。
结果：15项研究纳入分析。有细菌感染的SLE患者血清PCT和CRP水平明显高于有耀斑的SLE患者(PCT:SMD=1.035,95CI=0.708~1.362;CRP:SMD=1.000,95CI=0.758~1.242)。整体灵敏度,特异性，SROC曲线下的面积，PCT的正似然比(PLR)和负似然比(NLR)分别为0.62、0.88、0.862、6.63和0.36,而CRP的相同指标分别为0.72、0.70、0.784、2.45和0.38。
结论：SLE合并细菌感染患者血清PCT和CRP水平显著升高。PCT比CRP具有更好的诊断性能。PCT具有很高的PLR值，可以作为规则标记，而CRP检测可能由于PLR较低而导致假阳性率较高;这两种标志物的NLR值均次优,不适合排除细菌感染.
方法：全面检索电子数据库。计算合并标准平均差（SMD）和95％置信区间（CI），以估计SLE细菌感染和耀斑之间血清PCT和CRP水平的差异。灵敏度,使用特异性和汇总受试者工作特征(SROC)曲线评估PCT和CRP的诊断价值.固定或随机效应模型的使用取决于异质性。
结果：15项研究纳入分析。有细菌感染的SLE患者血清PCT和CRP水平明显高于有耀斑的SLE患者(PCT:SMD=1.035,95CI=0.708~1.362;CRP:SMD=1.000,95CI=0.758~1.242)。整体灵敏度,特异性，SROC曲线下的面积，PCT的正似然比(PLR)和负似然比(NLR)分别为0.62、0.88、0.862、6.63和0.36,而CRP的相同指标分别为0.72、0.70、0.784、2.45和0.38。
结论：SLE合并细菌感染患者血清PCT和CRP水平显著升高。PCT比CRP具有更好的诊断性能。PCT具有很高的PLR值，可以作为规则标记，而CRP检测可能由于PLR较低而导致假阳性率较高;这两种标志物的NLR值均次优,不适合排除细菌感染.
