Cox Regression Analysis Of Clinical Correlated Prognotic Factors For Hemphagocytic Lyphohistiocytosis

Objective:Hemophagocytic lyphohistiocytosis(HLH) is aimmunologic derangement leading to multiple organ disfunctioncharacterized by by fever, hepatosplenomegaly, hypocytosis,hypertriglyceridaemia or hypofibrinogenemia. It’s also calledhemophagocytic syndrome(HPS)with the important finding ofhematophage in bone marrow, lymph nodes and other tissues. HLH occurspredominantly in children and it comprises primary and acquired formswith different diease severity.Althought chemotherapy and/orimmuotherapy is often required for immediate survical,but a largepropotion of deaths occur early into treatment.Primary HLH typicallyrequires a haematopoietic stem cell transplant for permanent cure，whileacquired HLH can achieve relapse-free remission withouttransplantation.Death mostly occur before the start of treatment.Currently little is known about the consolidated predictors ofHLH.We collected data of HLH children and analyzed the prognosticfactors with COX univariate and multivariate analysis and to find out the factors influencing chance of surviving,thus representing clinical evidencefor early judgment of prognosis.Methods:The analysis was performed in115patients with HLHadmitted in our hospital between January2008and December2012abouttheir clinical feature and early laboratory markers.The diagnosis of HLHwas made according to the diagnostic criteria of HLH-2004criteria.Theseclinical parameters were selected for analysis:age,sex,days of fever beforeadimission, gastrointestinal hemorrhage, splenomegaly, neutrophilcount，platelets count, haemoglobin，lactate dehydrogenase, serum albumin,alanine aminotransferase level， activated partial thromboplastin time,fibriogen, ferritin,NK-cell activity, triglyceride, hormone therapy,CsAtherapy and VP16therapy.We used Kaplan-Meier method to draw survivalcurve and COX regression model for both univariate analysis andmultivariate analysis to screen independent prognosis factors and evaluatedthe relative risk.Results:1It was found that24of the115patients had bad prognosis.The rateof the incidence was20.87％ranging in age from6month to15years.There were9boys and15girls in the24patients,and the ratio of boyto girl is1:1.67.The bad prognosis happened from7days to2months afterdiagnosis.2The following factors were assciatied with an increasing risk of death by univariate analysis: gastrointestinal hemorrhage(p=0.000,RR=0.234)， APTT (p=0.00， RR=5.243)，FIB(p=0.001,RR=0.252)，temperature recovery after hormone(p=0.006，RR=0.129) related to prognosis(p<0.05);age,sex, duration of fever beforeadmitted, splenomegaly(≥4cm), neutrophil count， platelets count,hemoglobin, lactic dehydrogenase, fibrinogen，ALT(≥3times of normal)，ferritin，NK-cell activity，triglyceride，hormone therapy，CsA therapy andVP16therapy not relative with prognostis， and had no significantdifference(p>0.05).3These risk factors were entered into a multivariate model andgastrointestinal hemorrhage and APTT were independent prognosisfactors for survival.The independent risk factors of survival were：gastrointestinal hemorrhage(p=0.005，RR=3.276) and APTT(p=0.000，RR=5.803，95％CI=2.448-13.757) were identifies as being significantlyassociated with the death of HLH(B>0，RR>1).Conclusions1It was found that24of the115patients developed death,and therate of the incidence of death was20.87％.There were9boys and15girlsin24patients involvement.The ratio of boy to girl is1:1.67,and girls weremore than boys.The rate of the incidentce of boy’s death was16.98％.Compared to boys, the rate of girls was24.19％.The median ageof onset of death was4.6years old. 2Gastrointestinal hemorrhage was independent prognostic factor ofsurvival.The probability of poor prognosis is3.276times more than thosewithout gastrointestinal hemorrhage.This provide evidence of earlyjudgement and curative observation.3The probability of death proceeding in paitents with APTT>60s is5.803times more than those with APTT<60s. Coagulation disorders wasone of the independent prognostic factor s of death.This provide clinicalevidence for the prognosis and perfect effect to correct blood clottingfunction as soon as possible to decrease death.4Temperature recovery after hormone had relationship withprognosis by univariate analysis.But it was not statistically significant bymultivariate analysis and need more research. In patients receivinghormone therapy，temperature recovery in one week after treatment wassignificantly related to prognosis(p=0.034， RR=4.781，95％CI=1.124-20.343).5FIB<0.5g/L may have intimate relationship with prognosis byunivariate analysis，but no statistically significant by multivariate analysis.It was inconsistent with clinical work and need more research.6Whether the patients on admission had hyperlactacidemia，serumalbumin level，alanine aminotransferase levels，highly elevated ferritinlevels，hypertriglyceridemia，NK-cell activity and CsA therapy，VP16therapy were not related with prognosis.