Prognosis of HIV Patients Receiving Antiretroviral Therapy According to CD4 Counts: A Long-term Follow-up study in Yunnan, China

We aim to evaluate the overall survival and associated risk factors for HIV-infected Chinese patients on antiretroviral therapy (ART). 2517 patients receiving ART between 2006 and 2016 were prospectively enrolled in Yunnan province. Kaplan-Meier analyses and Cox proportional hazard regression analyses were performed. 216/2517 patients died during a median 17.5 (interquartile range [IQR] 6.8–33.2) months of follow-up. 82/216 occurred within 6 months of starting ART. Adjusted hazard ratios were10.69 (95%CI 2.38–48.02, p = 0.002) for old age, 1.94 (95%CI 1.40–2.69, p < 0.0001) for advanced WHO stage, and 0.42 (95%CI 0.27–0.63, p < 0.0001) for heterosexual transmission compared to injecting drug users. Surprisingly, adjusted hazard ratios comparing low CD4 counts group (<50 cells/µl) with high CD4 counts group (≥500 cells/µl) within six months after starting ART was 20.17 (95%CI 4.62–87.95, p < 0.0001) and it declined to 3.57 (95%CI 1.10–11.58, p = 0.034) afterwards. Age, WHO stage, transmission route are significantly independent risk factors for ART treated HIV patients. Importantly, baseline CD4 counts is strongly inversely associated with survival in the first six months; whereas it becomes a weak prognostic factor after six months of starting ART.

It is important to identify prognostic factors for survival among HIV-infected patients receiving ART. Based on previous studies, the main risk factors for death include baseline low CD4 cell count, old age and advanced WHO stage. Among them, CD4 cell count was suggested as the most important prognostic factor based on the estimated hazard ratio values [9][10][11][13][14][15][16][17][18][19][20][21] . According to those results, the prognostic value of CD4 cell count seems to be well-established. However, a recent study, examining European and North American patients, suggested that patients with low baseline CD4 cell count only carry the burden of increased risk of death up to 5 years after ART 16 . This indicated the impersistence of the CD4 cell count on the increased mortality, although no other study has reported similar observation. In order to better understand the treatment outcome of HIV patients, we have carried out a large prospective cohort study with long-term follow up in China, enrolling patients from Zhaotong, a prefecture-level city located in the northeast corner of Yunnan province. In this study, we aim to evaluate the overall survival and associated risk factors for the HIV-infected patients on ART in this cohort, with particular focus on the prognostic value of CD4 cell count.

Results
Baseline Characteristics. A total of 2517 patients (adult) before starting ART were enrolled between July 2006 and April 2016 in this study. The baseline characteristics of the study population at start of ART are summarized in Table 1 Table S1).

Prognostic values of baseline factors.
The associations of baseline factors at start of ART with mortality, estimated from crude and adjusted Cox models with time-dependent coefficients for CD4 cell count are shown in Table 2. Age and CD4 cell count were significant prognostic factors for survival (p < 0.0001 by log-rank test) based on crude survival analysis as shown in Fig. 2

Discussion
This is a large prospective study pioneering the assessment of patient survival and associated risk factors in Chinese population with HIV infection receiving ART. We described the treatment outcome and identified baseline age, WHO clinical stage as independent predictors for patients survival for all time. The low CD4 cell count at baseline had a strong inverse association with survival at first six months of starting ART. The survival benefit from ART have been demonstrated intensively among the HIV infected patients 2, 3 . A high tolerance of ART regimen and fewer regimen switches were found among our patients (Supplementary Figure S2). Of the patients who died during the follow-up, more than two thirds occurred within the first six months since ART initiation, similar to previous studies 11,14,22 . The patient characteristics between those dead at first six months and the entire study population were clearly described in this study. The former group had a higher percentage of patients with old age, advanced WHO clinical stage (stage III/IV) and low CD4 cell count compared with the entire study population. Thus, poor baseline patient characteristics seem to contribute largely to the worse survival, supporting the importance of early diagnosis and treatment 11,14,15,[23][24][25] . Overall, we feel that the first six months of ART is critical for improving survival outcome in HIV-infected patients.
The varying coefficients for CD4 cell count was demonstrated among our patients. Many previous studies only reported CD4 cell count as a strong prognostic factor for all time of their study period, but the potential varying effects of CD4 cell count on survival were hardly discussed in their studies 9,11,21,26,27 . A potential reason could be the lack of testing the proportional hazard assumption for Cox model, or such assumption was met in those studies. Another potential explanation could be the relatively short follow-up time in many previous studies. Taking this into consideration, a recent retrospective study with long-term follow-up (more than ten years) from the Antiretroviral Therapy Cohort Collaboration (ART-CC) have suggested that the baseline CD4 cell count is less prognostic after five years since starting ART 16 . In other words, the patients with low baseline CD4 cell count, who survived the first five years since ART, may expect similar mortality to that of patients with high baseline CD4 cell count. In our prospective cohort study, a strong inverse association between baseline CD4 cell count and risk for mortality was also observed, but only for the first half year after starting ART. We observed a 10-fold increased mortality for low CD4 cell count, which was much higher than the 2.8-fold increase in the ART-CC study 16 . Of note, there are several differences between these two studies, in particular European Americans vs Chinese  28,29 , our data might reflect more the treatment effect on HIV infected patients nowadays. These factors may contribute to the discrepancies observed between our study and the previous ones. Besides, the CD4 cell count at six months after starting ART was also examined in our study for assessing its potential association with survival after six months of starting ART. In contrast to values at baseline, the six-month CD4 cell count was strongly associated with the worse survival for the period after six months of ART initiation. Consistently, the prognostic value of six-month CD4 cell count has been demonstrated previously 30 . To be noticed, due to the missing values of six-month CD4 cell count among the patients who had survived for six months (displayed in Supplementary Figure S3), these data were not shown in this study.
A potential limitation of this study is the lacking of baseline viral load. Although the primary aim of ART is to inhibit the viral replication and reduce viral load, the resulted increase of CD4 cell count is the main goal as it serves as the most important indicator of immune function in HIV-infected patients. Despite this, WHO clinical stage and particularly CD4 cell counts were analyzed. Therefore, the potential bias from not analyzing viral load have been largely circumvented by including the analysis of WHO clinical stage and particularly CD4 cell counts. Besides, the present study only analyzed the data which were collected at start of ART initiation. Although baseline CD4 cell count has been proven to be an important predictor for the long-term outcome of ART and patient survival 19,31,32 , the dynamics of CD4 cell count across follow-up period could also be very important, deserving further investigation.
In conclusion, this is a large prospective study with long-term follow-up investigating the treatment outcome and prognostic factors for the HIV-infected patients receiving ART. Baseline characteristics including gender, age and WHO clinical stage are significantly associated with all-cause mortality. Importantly, we reported the time-dependent coefficients for CD4 cell count over different time intervals among Chinese population. The strong inverse association between CD4 cell count and risk for mortality has been demonstrated for the first half year after starting ART. However, CD4 cell count at six month has a strong inverse association with survival after six months of starting ART.

Study population.
A prospective cohort study was conducted by enrolling HIV patients at start of ART between 2006 July to 2016 April in Zhaotong, Yunnan province, China. Patients were enrolled from different areas of Zhaotong (a prefecture-level city), including Zhaoyang District, Ludian County, Qiaojia County, Yanjin County, Daguan County, Yongshan County, Suijiang County, Zhenxiong County, Yiliang County, Weixin County and Shuifu County. All patients were treated based on the criteria of the "National AIDS Free Antiviral Treatment Manual", HIV drug resistance, individual health, and other factors. We collected data on demographics (age, sex, marital status, transmission category), histological parameter (WHO clinical stage), and laboratory maker (CD4 cell count) at baseline when patients initiated ART.
The institutional ethical committee of the First Affiliated Hospital of Kunming Medical University has approved this study. Informed consent was obtained from all participants in this study. No personally identifiable information was seen and used in our data analysis. All the methods of this study were performed in accordance with the guideline and regulation of the institutional ethical committee of the First Affiliated Hospital of Kunming Medical University.
Statistical Analysis. All patients were followed up from the date since starting ART until the date of death, loss to follow-up, or the end of follow-up. Patients who were lost to follow-up or the event did not occur within the study duration were considered as censored cases. For the patients with treatment changes or interruptions, we analyzed the data as their intent to continue treatment, same as the other patients. In descriptive statistics, continuous variables were reported as mean with standardized deviation (SD) and median with interquartile range (IQR). Categorical variables were reported as number with percentage. A crude survival analysis (Kaplan-Meier curve) was utilized to analyze the patients' survival on ART during 10-year follow up. An adjusted survival analysis, Cox proportional hazard regression analysis, was used to evaluate the factors related to survival outcome. Variables with a p-value below 0.20 on univariate analysis were included in multivariate analysis [33][34][35] . All statistical analyses were performed in R (version 3.3.1) 36 . Particularly, "survival" package 37 was used for survival analyses. The loss to follow up and not loss to follow up against survival time were plotted to depict the their patterns (Supplementary Figure S4). The assumption of proportional hazards was tested both statistically (Supplementary Table S2) and graphically (Supplementary Figure S5) using function cox.zph. Besides, Concordance was calculated to assess the Cox model accuracy. One of the strengths of Cox model is its ability to encompass the time-varying coefficients. A step function was used to analyze the time-dependent effect of baseline CD4 cell count over different time intervals after breaking the data set into time dependent parts using survSplit function 38 . P < 0.05 (two-tailed sides) was considered as significant. The R code for performing Cox proportional regression analyses in this study is shown in Supplementary Figure S6. Data availability. The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.
Role of the Funder/Sponsor. The funding organizations had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.