Predicting active SLE using reliablemarkers is of particular importance when it comes to implementing useful preventive strategies in clinical practice. As we known, SLEDAI is the most commonly used to calculate disease activity. However, the numeration of SLEDAI need to synthesize many clinical symptoms and abnormal laboratory tests. Thus, If there are some indexs that can be obtained from laboratory data using blood samples, clinicians would be able to objectively, simply and continuously evaluate the activity of SLE patients. Studies have confirmed the presence of hematological abnormalities, such as the content of anti-dsDNA, L, Fbg, CRP, ESR, HDL-C, LDH, ALB, et al, in SLE patients [6, 8, 12, 22–24]. The present study systematically explored the levels of routine laboratory indicators between SLE with active group and SLE without active group and showed that many laboratory indicators including anti-dsDNA, ANUA, Anti-SSA, C3, ESR, CRP, TBIL, TP, ALB, A/G, CREA, HDL-C, D-D, Fbg, HGB, HCT, L were differential expression in SLE with active group than that in SLE without active group. These factors are important indicators of inflammation and immune response. Thus, we analyzed the potential of these novel serological indicators calculated by conventional indexs in predicting the activity in SLE patients.
Among the 12 novel serological indicators including AAR, THR, AFR, PNI, NLR, dNLR, PLR, LMR, SII, CAR and ECR, the levels of AFR, PNI were significantly lower in SLE with active group, only CAR was significantly higher in SLE with active group than that in SLE without active group. Only one study has investigated the level of AFR and CAR in SLE . Our findings agreed with those reported by He et al., where AFR was decreased and CAR was increased in SLE patients. However, the relationship between AFR, CAR, and clinical disease activity have not been investigated previously in SLE in previous study. Spearman method showed that AFR, CAR and PNI were all associated with disease activity measured by SLEDAI, autoantibodies, ESR, CRP, et al. As far as we are aware, there were three study has investigated the association of PNI with disease activity in SLE [14–16]. Our findings agreed with the results from those reports and suggest that PNI may be a useful index for the evaluation of disease activity. In addition, ROC curve showed that lower AFR and PNI could provide better predictive vaule in SLE activity (AUC > 0.7), but the AUC of CAR was inferior to 0.7. More importantly, logistic regression indicated lower AFR and PNI were risk factors for SLE activity. Thus, the present study considered together with previous work suggested that AFR and PNI could be a useful index for the evaluation of disease activity in SLE patients.
To the best of the authors knowledge, the present study is the first to constructe a prediction model of AFR and PNI and evaluate the value of prediction model in SLE activity. Our study demonstrated that the predictive model based on combination of AFR and PNI performed best value in distinguishing SLE with activity group from SLE without activity group, with AUC of 0.765, which superior to single AFR and single PNI. These results suggested that AFR and PNI have synergic effect on predicting activity occurrence in SLE. The predictive value based on combination of AFR and PNI = -0.102* AFR-0.008* PNI + 4.890. Moreover, our results indicated that the predictive value were correlated with commonly used indicators for SLE activity[6, 8, 21, 25, 26, 27, 28] including TP, ALB, A/G, D-D, Fbg, ESR, CRP, HCT, CAR, AFR, PNI, SLEDAI, TBIL, LDH, HDL-C, RBC, HGB, L, THR, C3, DBIL, CREA, UA, L%, N%, NLR, dNLR, PLR, LMR, ECR, anti-SSA, anti-dsDNA, ANUA, which suggested that the predictive model based on combination of AFR and PNI could be a useful novel index for the evaluation of disease activity in SLE patients.
As we known, anti-dsDNA is the traditional and most commonly used assessing activity marker for SLE [29, 30]. Furthermore, our study revealed that the AFR-PNI model + anti-dsDNA combination model could effectively discriminated the SLE with activity and SLE without activity, with a sensitivity of 82.81% (53/64), a specificity of 88.46% (23/26), which were superior to AFR-PNI model and anti-dsDNA. These results indicated that the combination of AFR-PNI model and traditional autoantibody could further improve the predictive value.
The research from Ahn. et al showed that PNI was associated with lupus nephritis . In this study, our data indicated PNI was associated with RI in agreement with the previous research, which suggesting PNI correlated with clinical symptoms of SLE. In addition, we found PNI was associated with fever, pleurisy and pericarditis. Moreover, the results demonstrated AFR was associated with pleurisy, pericarditis and AFR-PNI model was associated with fever, pleurisy, pericarditis.
There were potential limitations to this study. First, the assessment of clinical and laboratory datawas performed by reviewing medical records of patients. Second, the number of patients with inactive SLE patients included was relatively small. Third, serial changes in AFR, PNI, prediction model were not assessable and the adjustment for treatment was not possible owing to the retrospective study design. Fourth, we only included Chinese patients, these findings cannot be generalised to other ethnicities. Additional studies are required to validate our findings and reveal the association between prediction model and the active in SLE patient.