| Objectives:The aim of the study was to assess the risk low level viremia(LLV)on clinical prognosis,to determine the factors associated with virological failure(VF)and to construct a VF binary logistic equation to predict the risk of VF.Methods:1:Experimental subject:A retrospective study was based on an open study cohort of people infected with HIV-1 treated with antiretroviral therapy in Shenyang Large General Hospital from 2002 to 2018.All subjects were treated with first-line ART recommended by the Chinese guidelines for AIDS diagnosis and treatment and followed up every 3 to 6months to evaluate the treatment effect to evaluate VL,CD4T cells count,blood routine and liver function and other routine clinical parameters.2:LLV and VF definition criteria:referring to the WHO guidelines,we defined LLV as 50–1000 copies/ml and defined VF as two or more consecutive detections of VL more than 1000 copies/ml that could not spontaneously control viral replication without altering the treatment protocol.3:Statistical methods:use statistical analysis software(SPSS22.0)to organize and analyze the database.The Kaplan-Meier(K-M)curve,Cox proportional risk model and linear mixed effect model were used to evaluate the effect of LLV on VF.Multivariate Logistics regression was used to analyze the related factors of VF.Using whether the subjects had VF as the outcome variable,the EM method supplemented the gender,economic status,baseline virus load level,first-line antiviral therapy duration,treatment plan,LLV level,total duration of LLV as the covariate,performed binary Logistic regression analysis and established regression equation.The ROC curve was established to analyze the accuracy of the regression equation by using the predictive probability of VF calculated by binary Logistic equation.Results:1.Characteristics of patients with LLV during long-term first-line ART1.1 Incidence rate of LLVDuring the 16-year observation period in the first-line long-term antiviral treatment cohort,the first affiliated hospital of China Medical University collected 4389 cases of HIV-1 infection and followed up 40649 times.Of these,2155 patients with ART treatment were longer than one year and whose treatment-related information was complete,with cumulative follow-up of 8348 person-years and median follow-up of 3.4person-years(IQR:2.3–5.0).38.7%(835/2155)patients occurred LLV.1.2 Sociodemographic information of LLV patientsIn this study,we compared the sociodemographic information of LLV patients.Then,we found that patients with LLV were older(42 vs 39)and the proportion of unmarried was less than those with long-term suppression of patients(53%vs 59%).However,there was no statistically significant difference among the factors such as gender,education level,nationality,economic status,transmission route.1.3 Clinical follow-up data for patients with LLVIn this study,by analyzing clinical follow-up data of patients with LLV,we found that patients with LLV had a longer follow-up period(4.5 vs 3.5 years)and fewer treatment options with TDF+3TC+EFV(58.8%vs 74.7%)than patients with long-term suppression.1.4 Baseline laboratory data for LLV patientsIn this study,baseline laboratory data from patients with LLV were compared to those with long-term suppression,the proportion of CRF01＿AE subtype in patients with LLV was higher(79.03%vs 64.69%),the baseline viral load was higher(1.27*10~5 vs.2.08*10~5 copies/ml),and the baseline CD4T cells count was lower(285 vs 291cells/mm~3).2.Risk of VF in patients with LLV with different characteristics2.1 Risk of VF in patients with LLV at different levelsThe rates of VF increased with elevated levels of LLV:50–200 copies/ml:1.6%,200–400 copies/ml:4.7%,400-1000 copies/ml:13.4%.K-M curve analysis revealed that the higher the level of LLV,the higher the risk of VF(p<0.01).2.2 Risk of VF in patients with LLV with different total durationsVirological failure rates increased with increased total duration of LLV:blip:3.3%,3-6 months:7.4%,6-12 months:6.3%,>12 months:11.6%.K-M curve analysis showed that the longer the total duration,the higher the risk of VF in patients with LLV(p<0.01).2.3 High-risk LLV which were defined in combination with LLV levels and total duration indicatorsWe combined levels and duration of LLV analysis and found that high level of LLV were at high risk of VF even if they had a short duration,while low level of LLV were at low risk of VF,even if they had a longer duration.The LLV that increased the risk of VF was defined as high-risk LLV.A total of 152 high-risk LLV patients in this study,of those14.5%patients have VF during long-term treatment,which was higher ratio than that in Low-risk LLV patients(2.0%).3.Factors associated with VF3.1 Characteristics of patients with VFThe patients with VF had specific baseline characteristics,including:slightly older(43 vs.40 years),higher proportion of stable income,shorter duration of first-line antiviral therapy(2.61 vs.3.91 years),fewer proportions of preferred treatment options,higher logarithmic baseline viral load level(1.50*10~5 vs.4.64*10~5 copies/ml),lower baseline CD4T cells count(226 vs.290 cells/mm~3)and higher proportion of B’subtype.3.2 Regression analysis of VFMultivariate Logistics regression analysis further supported that the patients with higher zenith baseline VL(>10~6 vs.<10~4 copies/ml:aOR=5.71,95%CI:1.82–17.84;105-106 copies/ml:aOR=4.24,95%CI:1.55-11.57,p<0.01),lower nadir baseline CD4T cells counts(<200 cells/mm~3 vs.>350 cells/mm~3:aOR=2.29,95%CI:1.07–4.89,p<0.05),Age(≥50 vs.<50 years:aOR=1.82,95%CI:1.07-3.09,p<0.05),ART for less than 36months(aOR:4.28,95%CI:1.87–9.76,p<0.01),receiving D4T+3TC+NVP regimen(aOR:9.95,95%CI:1.73–57.23,p<0.01)or DDI+1 NRTI+1 NNRTI regimen(aOR:22.16,95%CI:4.19–117.07,p<0.01)or subtype B’infection(aOR:10.77,95%CI:1.32–87.76,p<0.05)were factors of VF.4.Constructing a binary Logistic regression equation for VF4.1 Supplementary missing dataIn this study,there were different degrees of missing data in different variables(1.35%-13.64%),in which there were more than 5%missing data in education,economic status,transmission mode and virus subtype.Marital status and economic conditions differ in supplementary data from the EM methods and regression methods.4.2 Construct binary logistic regression equation and calculate sensitivity and specificityAfter stepwise backward screening of variables,variables such as gender,economic status,baseline viral load level,first-line antiviral therapy duration,LLV level,total LLV duration,and first-line antiviral treatment regimens were eventually included in the equation.Omnibus test results of this equation less than 0.01.Cox&Snell and Nagelkerke test results more than 0.05.Hosmer and lemeshow tested 0.92.Overall forecast rate 97.6%.Negative predictive rate 99.7%.Positive predictive rate of 27.4%.The area under the ROC curve is 0.919(95%CI:0.882-0.957).Standard error is 0.019.The Jorden index is 0.704.The corresponding Logit(p)value is 0.026.The sensitivity and specificity were 85.5%and 84.9%,respectively.Conclusions:1.In this study,a large sample and long-term retrospective study showed that 38.7%of patients with first-line antiviral therapy had LLV,of those,18.2%were high risk LLV.2.By analyzing the K-M curves of different levels and different total durations of LLV,we found that the higher the level of LLV,the higher the risk of VF for the longer the total duration.but Cox regression model and linear mixed effect model analysis showed that LLV>400 copies/ml increased the risk of VF even if the duration was very short;whereas LLV<200 copies/ml even if the duration was long,the risk of VF was low,focusing on the High-risk LLV population.3.Factors associated with VF include Age,duration of first-line ART,use of D4T+3TC+NVP or DDI+1 NRTI+1 NNRTI treatment regimen,higher baseline VL levels,lower baseline CD4T cells count and B’subtype.4.Using the binary logistic regression equation to predict the probability of VF,and to strengthen the treatment of patients with high probability. |
PDF Full Text Request