PF713 THERE ARE NO PREDICTIVE FACTORS ASSOCIATED WITH GOOD THERAPY RESPONSE TO SPLENECTOMY FOR PRIMARY IMMUNE THROMBOCYTOPENIA – CONCLUSION BASED ON SYSTEMATIC REVIEW AND META‐ANALYSIS

Background: Splenectomy may lead to a good response in 60–80% of adults with corticosteroid refractory immune thrombocytopenia (ITP). However, recent advent of new drugs has encouraged and generalized tendency to delay splenectomy. Appropriate selection of patients for splenectomy become essential for optimization of the treatment outcome. Nevertheless, available literature does not seem to contain any precise indications concerning possible factors predicting the response to splenectomy. Aims: To perform systematic review of literature and meta‐analysis in an attempt to identify potentially predictive clinical or laboratory parameters for good therapy response (GTR) to splenectomy in adult ITP patients. Methods: MEDLINE, Web of Science and EMBASE were searched from 1966 to 2018 for studies where splenectomised ITP patients were followed to determine predictive parameters for GTR. The search was limited to English‐language articles, with ≥15 consecutive splenectomised adult primary ITP patients. We performed two analyses: the primary analysis included all patients. In secondary analysis, papers were divided in two groups according to response criteria. The first group using the International Working Group criteria defined complete remission (CR) as platelet count (PC) >100x10 9 /L, partial remission (PR) as PC >30x10 9 /L and no remission (NR) as PC <30x10 9 /L. The second group defined CR as PC >150x10 9 /L, partial remission as PC >50x10 9 /L and NR as PC <50x10 9 /L. Both groups were subdivided according to the tested type of response: predictors of initial response, predictors of sustained remission and predictors of relapse. Random effects method was used to summarize outcomes. The effect estimates were expressed as odds ratio (OR) or standardized mean difference (SMD) with 95% confidence interval. Strength of evidence was assessed by New‐Castle Otawa protocol. Results: The literature search identified 410 articles; 75 of them, with 6151 patients, met our criteria. The first group includes 16 articles, with 1665 patients (subgroups: initial response 10 articles, 867 patients; sustained remission 6 articles, 868 patients, relapse 3 articles, 407 patients). The second group includes 29 articles, with 1518 patients (subgroups: initial response 15 articles, 879 patients; sustained remission 15 articles, 799 patients, relapse 3 articles, 105 patients). Meta‐analysis did not reveal any significant predictor for successful splenectomy, in both primary and secondary analysis (p>0.05). Age, sex, the time between diagnosis to surgery, response to steroids, response to IVGs, pre‐ and post‐operative PC, antiplatelet antibody positivity, spleen weight, accessory spleen, number of therapy lines, and platelet kinetic were not significantly correlated with response to splenectomy. Low quality of evidence was found for all outcomes. Summary/Conclusion: Our study did not identify any clinical or laboratory parameter clearly predictive of GTR after splenectomy. Further studies based on appropriate use of standardized criteria for patient recruitment and outcomes under assessment are needed to better define splenectomy effect persistence in long term.