Monitoring HIV infection in Minas Gerais state: 15-year assessment of adults living with HIV initiating Antiretroviral Therapy

Abstract INTRODUCTION The first Brazilian HIV treatment recommendation was put forward in 1996, resulting in 12 subsequent guidelines. Several changes were made regarding “when” and “how” to begin treatment. The latest guideline recommends immediate initiation of antiretroviral therapy (ART). This study aimed to describe the evolution of HIV treatment among people living with HIV (PLHIV) who initiated ART between 2004 and 2018 based on the national guideline recommendations concerning T-CD4+ and VL measurements. METHODS A cross-sectional analysis of data of PLHIV aged >18 years, in Minas Gerais who received ART between 2004 and 2018 was conducted. Clinical, therapeutic, and demographic information were obtained from national healthcare databases. The study was divided into four periods: 2004-2007, 2008-2012, 2013-2016, and 2017-2018. Descriptive analyses were performed. RESULTS A total of 60,618 PLHIV initiated ART (67% male and 48% aged 25-39 years), 36% of whom had CD4 counts at ART initiation and 51% documented VL after ART initiation. The median CD4 count ranged from 288 to 373 cells/µL. The median time to ART initiation decreased from 604 to 28 days and was lower among males (p <0.01). The median time from ART initiation to the first VL result decreased from 101 to 62 days over the study period, while the median VL after ART initiation ranged from 2.3 to 1.7 log10 copies/ml. CONCLUSIONS Although our results demonstrated that most recommendations were followed, there seemed to be little impact on CD4 counts and VL testing. This may result in an inadequate evaluation of ART effectiveness.


INTRODUCTION
The primary goal of antiretroviral therapy (ART) is to suppress the viral load (VL) to undetectable levels in order to restore the immunologic response, to reduce opportunistic infections, and to improve survival 1 . In the general population, ART is associated not only with a substantial decrease in the probability of HIV transmission due to the reduction of VL but also with an improvement in the health of people living with HIV (PLHIV) due to an increase in T-CD4+ lymphocyte counts (T-CD4+) 2 .
ART was introduced in Brazil at the beginning of the 1990s 3 . Since then, national guidelines based on international recommendations have been formulated by the government to guide treatment. The first guideline was introduced in 1996, resulting in the current 12 guidelines [4][5][6][7][8][9][10][11][12][13][14][15] . When new guidelines were introduced, recommendations were updated, especially concerning "when" and "how" to begin treatment. The key goal was to assess the benefits and risks beginning from the early phase of treatment.
Patients were not to be exposed to the long-term adverse effects of ART or to the possibility of development of drug resistance. Drug resistance could lead to virologic failure and the achievement of poor clinical outcomes 16 . ART was not considered an emergency. Therefore, the postponement of ART initiation was justified to keep patients healthy [5][6][7] .
The first treatment of an AIDS case based on the Centers of Diseases Control and Prevention (CDC) criteria took place in Brazil in 1987. However, it was based on specific and definitive diagnosis methods that required complex and sophisticated exams, which are not feasible in many countries 17 . In 1992, the T-CD4+ count was introduced as a marker of immunosuppression, resulting in a more accurate classification of AIDS cases 17 . In this way, the timing of ART initiation was not only based on clinical criteria but also on the T-CD4+ count [5][6][7] . Determining the best cut-off value for T-CD4+ for beginning treatment also changed over time: from T-CD4+ counts < 200 cells/µl to any level, as is the currently adopted guideline.
From 2004 to 2007, Brazilian guidelines recommended initiating ART for all patients who matched the clinical disease criteria and/or T-CD4+ count ≤ 200 cells/µl. The combination of zidovudine/lamivudine (ZDV/3TC) plus efavirenz (EFV) was chosen as the first-line regimen 10,11 . In 2008, the guideline's recommendations were to increase the T-CD4+ count cut-off to 350 cells/µl 12 . After 2013, evidences indicated that early treatment decreased HIV transmission 18 . Thus, either T-CD4+ counts ≤ 500 cells/µl or VL >100,000 copies/ml were the laboratory criteria to initiate treatment. In addition, ART initiation based on tenofovir/lamivudine (TDF/3TC) plus EFV became the first-line regimen 13,14 . Finally, after 2017, the national guidelines have recommended the immediate initiation of ART to all PLHIV regardless of the T-CD4+ count or VL. Moreover, the EFVbased first-line regimens have been replaced with dolutegravir (DTG)-based first-line regimens ( Table 1) 15 .
All these changes in the recommendations regarding ART initiation have contributed to a potential positive impact on Brazilian HIV indicators. Between 2012 and 2018, the proportion of PLHIV maintained in public HIV healthcare services and on ART increased from 55% to 71% and from 44% to 66%, respectively 19 . A reduction in AIDS-related mortality rate (per 100.000 inhabitants) in Brazil has been observed since 2004: mean mortality rate, 6.0 (2004-2007), 5.7 (2008-2012), 5.5 (2013-2016), and 4.6 (2017-2018: preliminary data). However, in spite of the updated recommendations on a higher T-CD4+ threshold over the years, the mean incidences of AIDS cases (per 100.000 inhabitants) has not been reduced as expected: 20.4 (2004-2007 Despite the immediate treatment after HIV diagnosis, measurements of T-CD4+ counts and VL continue to play an important role in HIV monitoring. Current guidelines worldwide have recommended measurement of the T-CD4+ count prior to initiation of ART to provide information on the overall immune function with a focus on early treatment. VL has been recommended two to eight weeks after ART initiation to monitor the effectiveness of ART with a focus on viral suppression 24,25 . We hypothesized that the measurements of T-CD4+ counts before ART initiation and VL after ART initiation have improved over a 15year period in the state of Minas Gerais, Brazil. Minas Gerais, located in the southeastern region of Brazil, is a large state (approximately 21 million inhabitants) 26 , where 3,418 HIV-diagnoses were recorded in 2018. This accounted for approximately 8% of Brazilian HIV-notifications 23 . The objective of this study was to describe the evolution of HIV treatment among adult patients (18 years old and above) living in Minas Gerais, that initiated ART from 2004 to 2018 based on national guidelines regarding T-CD4+ and VL measurements.

Study design and population
We performed cross-sectional analyses of PLHIV data in the state of Minas Gerais, Brazil. All HIV-infected individuals 3/8 (aged 18 years or older) who initiated ART between January 2004 and December 2018 were eligible for the analysis. The study was approved by an appropriate Ethics Research Committee.

Data source and variables
Data on clinical/therapeutic (VL measurement, T-CD4+ count, ART use) and demographics (age, sex) variables were obtained from two national healthcare databases of the public Brazilian Unified Health System (SUS): Medication Logistics Control System (SICLOM) and Laboratory Tests Control System (SISCEL). Interaction with these databases is required for ART dispensing and obtaining laboratory results (VL and T-CD4+ count measurements) nationwide. The health databases were linked through a deterministic and probabilistic deduplication of records based on the patient's full name, sex, date of birth, and city of residence. This created a common identifier in both the SISCEL and SICLOM systems. Data management, programming, and analysis were performed using MySQL, Pareia, and R software systems 27,28 . The matching identification number for each patient allowed the linkage of the databases and formed a single database with all data processed through the Statistical Analysis System (SAS, version 9.4).
For analysis, we adopted four study periods according to the main changes in the recommendations of Brazilian guidelines concerning the T-CD4+ count cut-off for ART initiation: Period I, between 2004 and 2007; Period II, between 2008 and 2012; Period III, between 2013 and 2016; and Period IV, between 2017 and 2018 ( Table 1). Age was calculated based on the date of ART initiation, and the median age was estimated and categorized into groups: 18-24, 25-39, 40-49, 50-59, and >60 years. The date of ART initiation was based on the first record of ART dispensing, verifying that no other prescription was recorded at least in the previous three years. T-CD4+ counts comprised the last result preceded the date of ART initiation by 90 days 24 . This was categorized in accordance with the T-CD4+ count cut-offs of each period: Period I, <200 cells/µl; Period II, >200-349 cells/µl; Period III, 350-499 cells/µl and Period IV, >500 cells/µl. VL comprised the first result after ART initiation and was considered as a log 10 value and was characterized according to the detection limits at the designated time period: undetectable VL, < 2.6 log 10 copies/ml (from 2004 to 2009) or <1.7 log 10 copies/ml (from 2010 to 2018); detectable VL, from > 2.6 log 10 copies/ml (from 2004 to 2009) or > 1.7 log 10 copies/ml (from 2010 to 2018) to < 5.0 log 10 copies/ml (from 2004 to 2018); and very high VL, >5.0 log 10 copies/ml (from 2004 to 2018) 29 .
Additionally, the date of the first T-CD4+ count prior to ART initiation was considered to estimate the median time to ART initiation 30 . The time to ART initiation was categorized according to the T-CD4+ count cut-offs of each period. Similarly, the median time to the first VL measurement after ART initiation was calculated and stratified into the VL categories.

Data analysis
Descriptive analyses were performed for each period. Measures of central tendency (median), measures of dispersion (interquartile [IQR]), and absolute and relative frequencies were calculated for selected variables. The Kruskal-Wallis test was used to compare the continuous variables, and multiple comparisons were based on the Bonferroni method. Data management and data analyses were performed using the Statistical Analysis System (SAS, version 9.4). Table 2 shows the descriptive analysis of 60,618 PLHIV initiating ART between 2004 and 2018 in the state of Minas Gerais, Brazil. The number/proportion of individuals increased in the first three periods: from 11,701 to 17,844 (52%) and from 17,844 to 19,716 (10%); however, it decreased from period III to IV, from 19,716 to 11,357 (42%). Overall, we noticed a greater proportion of men (67%) and individuals aged 25-39 years (48%). Over the study period, the median (IQR) age decreased from 40 (33-47) to 34 (28)(29)(30)(31)(32)(33)(34)(35)(36)(37)(38)(39)(40)(41)(42)(43) years (p-value < 0.01). ART first-line regimens including ZDV /3TC plus EFV, or lopinavir (LPV), or nevirapine (NVP), were gradually changed to regimens based on TDF/3TC plus EFV or DTG. Among the study population, we noticed a large proportion of missing data due to the lack of routine exams. Overall, 64% (n= 38,535) of participants did not have T-CD4+ counts at ART initiation, ranging from 59% in period I to 61% in period IV. The proportion of individuals presenting with T-CD4+ counts of less than 200 cells/μl were 13%, 16%, 7%, and 12% in each period, respectively. The medians (IQR) of T-CD4+ baseline counts in periods I and II with 288(163-469) and 255(120-353) cells/μl were lower than those of periods III and IV with 373(205-520) and 349(151-553) cells/μl, respectively (p <0.01). Similarly, 49% (n= 29,554) of participants did not document VL measurements after ART initiation, and this proportion increased from 23% to 67%. The proportion of undetectable VL in periods III and IV ranged from 15% to 16%, respectively, while the median (IQR) VL decreased from 1.9 (1.4-2.8) to 1.7 (1.4-2.6) in the same periods (p <0.01) Table 2. Table 3 presents the estimated time to ART initiation and the time to the first VL measurement after ART initiation stratified by cut-off exams and sex. Among 22,083 participants (36%) with T-CD4+ counts at ART initiation, the median time (IQR) to ART initiation decreased dramatically from 604 days (84-804) in the first period to 28 days  in the last period and was lower among males between periods I, II, and III (p < 0.01). Furthermore, the median time to ART initiation was lower for individuals presenting with lower T-CD4+ counts, except in the last period, in which there was no difference between the T-CD4+ category and time to ART initiation. Approximately 51% of participants (n=31,064) documented VL measurements after ART initiation. The median time from ART initiation to the first VL result was considerably reduced from 101 days in the first period to 62 days in the last period (p <0.01), with a notable difference between the undectable and very high VL in the last period (96 days vs. 8 days). Overall, there was no difference between females and males in the time to the first VL measurement after ART initiation.   24  334  3  1024  6  2313  12  1598  14  5269  9  25-39  5092  43  8136  45  9926  50  5748  51  28902  48  40-49  3977  34  5206  29  4183  21  2235  20  15601  25  50-59  1699  15  2465  14  2321  12  1192  10  7677  13  > 60  599  5  1013  6  973  5  584  5  3169  5  Total  11,701

DISCUSSION
Our study described selected results based on the updated recommendations of HIV monitoring from 2004 to 2018 in the state of Minas Gerais, Brazil. Certain findings were in accordance with the national guidelines adopted for each period.
According to the Brazilian guideline recommendations, ART first-line regimens based on ZDV were gradually switched to TDF as well as EFV to DTG. Meireles and collaborators presented similar frequencies of initial ART regimens in Brazilian HIV-infected individuals between 2014 and 2015 (n=76,950): TDF+3TC+EFV (75.6%), ZDV+3TC+EFV (5.3%), and ZDV+3TC+LPV (7.0%) 31 . Due to the decrease in the VL values, especially in the last period, this ART modification may have had a great impact on VL reduction.
The national guidelines changed their recommendations to initiate ART earlier based on higher T-CD4+ thresholds. After 2017, these recommendations were changed to immediate ART initiation for all HIV-diagnosed individuals, regardless of the T-CD4+ count, with a preference for those with T-CD4+ counts below 350 cells/µl 15 . In fact, our results showed that the median time to ART initiation decreased and was prioritized in accordance with the T-CD4+ counts. A decreased time to ART initiation was associated with lower T-CD4+ counts. Our estimated findings in the state of Minas Gerais are similar to the national results. According to the Report of Clinical HIV Monitoring (2018), the median time between the first T-CD4+ count and ART initiation in 2009 and 2018 decreased markedly from 657 to 32 days in Brazil 30 .
National and international guidelines have recommended performing VL tests eight weeks after ART initiation to assess the virologic response and to identify virologic failure promptly 15,24,25 . We also noticed that the median time between ART initiation and the first VL measure is close to these recommendations, since it was reduced to just 62 days in the last period. We also observed a large proportion of HIV individuals on ART who had not completed routine tests. Even in the era of early ART and national/international guidelines recommend laboratory testing 15,24,25 , our results showed that 49% of individuals had no VL measures after ART initiation. This gap may compromise the "90-90-90" target to end the AIDS epidemic. The third "90" signifies that all people receiving ART should be virologically suppressed 34 . Furthermore, individuals without VL routine testing may impair the estimation of VL suppression after ART initiation. This measurement takes into account only the individuals with VL testing instead of the total number of individuals on ART 19,35,36 . As a VL test remains the key predictor of improvement with ART, low VL testing rates provide limited information about regarding the success of ART programs 37 . Additionally, VL testing after ART initiation improves adherence of ART 38 .

6/8
Our study comprised a large period of time; thus, uncontrolled limitations inherent to the dynamic changes in the HIV epidemic must be addressed. In the early 2000s, the Health Ministry launched the SICLOM and SISCEL systems nationwide 39 . Thus, due to the implementation phase, the findings from the first years of our study require a more careful interpretation. The change in detection limits for undetectable VL in 2009 (from 2.6 log 10 to 1.7 log 10 copies/ml) impaired the comparison of the results from the first periods. Additionally, the number of VL tests conducted in the last months of 2018 is most likely underestimated due to the ending of the study period. As a result of these limitations, these VL results should be interpreted with caution. Second, the proportion of no values of both T-CD4+ count and VL testing may have affected the estimation of the median. Missing data can be partially explained due to a proportion of the population (27%) that received healthcare at private facilities 40 . T-CD4+ counts and VL data were not compulsorily recorded in the SUS databases. From December 2015, however, VL data from both public and private healthcare facilities have been required for ART dispensing 41 .
Despite the recognized importance of T-CD4+ counts in evaluating the disease status at ART initiation, ART could have been initiated without performing a T-CD4+ count by the symptomatic patients who presented late for HIV/AIDS care (especially in the first years of the study period) or those who were clinically stable but did not perform T-CD4+ counts in view of the recommendation of early initiation of ART. Finally, the increasing demand for laboratory monitoring generated from earlier ART initiation can be a challenge for low-and middle-income countries in providing adequate coverage for both T-CD4+ counts and VL tests. In an African study to predict T-CD4+ count recovery, the number of participants was reduced by 83% (from 3,981,104 to 1,070,900) mainly due to the lack of T-CD4+ lymphocyte exams 42 . Similar results can be observed in Sub-Saharan Africa, where 11 million PLHIV are receiving ART. An estimated six million ART patients do not have access to VL testing 43 .
There is scarce information regarding the availability of T-CD4+ counts and VL tests in Brazil. We believe that national studies on the frequency of routine exams might improve the availability of exams and indicate a more realistic scenario regarding immune status prior to ART initiation and VL suppression after ART initiation. Since 2014, data regarding new diagnoses of HIV infections have been incorporated into the national databases as a compulsory notification 44 . This action provided additional data on HIV surveillance including the date of HIV diagnosis, time of ART initiation, and immune status at the time of diagnosis. These steps would also likely improve adherence to routine exams. Despite all limitations, the large sample size and the wide time range improved the accuracy of our findings, providing a picture of the monitoring of people initiating ART in the state of Minas Gerais, Brazil.
Our study demonstrated that a majority of the recommendations regarding the national guidelines were followed over the study period. However, despite the national policy of universal access to ART (confirmed by a remarkable increase in ART availability noticed in our study), there remains a substantial gap between routine testing and ART initiation, which may lead to an inadequate evaluation of both immunological and virologic responses to ART. In addition, in the general population, it may contribute neither to a reduction in HIV transmission nor to the health of PLHIV.

ACKNOWLEDGMENTS
We gratefully acknowledge the Chronic Conditions and Sexual Transmissible infections Department, Ministry of Health-Brazil, for making this study possible, providing data from SISCEL and SICLOM systems.

AUTHORS' CONTRIBUTION
CCPM: Conception and design of the study, Acquisition of data, Performed statistical analysis, Analysis and interpretation of data, Drafting the paper; GJS: Reviewed analysis and interpretation of data, Final approval of the version to be submitted statistical analysis; LPB: Reviewed analysis and interpretation of data, Final approval of the version to be submitted statistical analysis MDCG: Reviewed analysis and interpretation of data, Final approval of the version to be submitted statistical analysis. EAC: performed statistical analysis, Interpreted the data, Final approval of the version to be submitted statistical analysis CAMP: Conception and design of the study, Reviewed analysis and interpretation of data, Final approval of the version to be submitted statistical analysis.