Compromised CD4:CD8 ratio recovery in people living with HIV aged over 50 years: an observational study

Objectives Persistent CD4:CD8 ratio inversion (< 1) is associated with mortality in older people. We investigated the interaction of the effects of baseline CD8 count and age at HIV diagnosis on CD4:CD8 ratio recovery with antiretroviral therapy (ART). Methods An observational study (1 January 2007 to 31 December 2016) was carried out using routinely collected data from the HIV outpatient services at Imperial College Healthcare NHS Trust, London, UK. CD4 and CD8 counts, prior to and during ART, treatment during primary HIV infection (PHI) and HIV‐1 viral load were included in univariate and multivariate analyses using Cox proportional hazard regression. Results Data were included for 876 patients starting ART, where HIV suppression was achieved. Of these patients, 741 of 876 (84.6%) were male and 507 of 876 (57.9%) were Caucasian. The median time on ART was 38 [interquartile range (IQR) 17–66] months. CD8 count change on ART was bidirectional; low CD8 counts (≤ 600 cells/μL) increased and high CD8 counts (> 900 cells/μL) decreased. The median pre‐ART CD4:CD8 ratio was 0.41 (IQR 0.24–0.63), and recovery (≥ 1) occurred in 274 of 876 patients (31.3%). Pre‐ and post‐ART CD4:CD8 ratios were lower in those aged > 50 years compared with young adults aged 18–30 years (P < 0.001 and P = 0.002, respectively). After adjustment, younger age at HIV diagnosis (P < 0.001) and treatment during PHI (P < 0.001) were favourable for CD4:CD8 ratio normalization. Conclusions Older age (> 50 years) at HIV diagnosis was associated with persistent CD4:CD8 ratio inversion, whereas treatment of PHI was protective. These findings confirm the need for testing and early treatment of people aged > 50 years, and could be used in a risk management algorithm for enhanced surveillance.


Introduction
Globally, the HIV epidemic is ageing; not only are people living with HIV (PLWH) growing older, but the incidence of HIV infection remains significant in people aged 50 years and over in the United States, and has not followed the decline observed in younger age groups in the UK [1,2]. The aged immune system loses functionality, rendering older people vulnerable to infectious disease and potentially cancer (reviewed in Foster et al. and Pera et al. [3,4]). The ageing process affects T-cell subsets, particularly CD8 T-cells [5]. Loss of clonal diversity, increased CD8 count, and age-related silencing of the interleukin (IL)-7 receptor gene have been reported [5]. These changes may be reflected in CD4:CD8 ratio inversion (< 1) in older people.
The impact of acute and chronic HIV infection on the immune system in older people is not fully understood. Despite advances in life expectancy, PLWH continue to be at higher risk of all-cause mortality, and mortality from infectious and liver disease than the general population [6]. HIV infection is associated with an increased prevalence of age-related diseases, despite antiretroviral therapy (ART), and the prevalence of non-AIDS-related morbidity, such as cardiovascular disease, is higher than in the general population [7,8]. The mechanisms that drive this difference for PLWH on suppressive ART could be a consequence of persistent immune activation and inflammation [9]. Whilst most PLWH are now successfully managed on ART by nonmedical staff, development of a clinical algorithm to select those at risk of complications for additional surveillance is now required to optimize care and resources.
Monitoring the CD4:CD8 ratio may serve such a purpose [10]. The peripheral CD4 count has, for decades, been a reliable biomarker for monitoring immunosuppression and recovery with ART [11,12]. Widespread access to ART has greatly improved the prognosis in HIV infection, and, with continuous viral suppression, reduced the clinical need for CD4 count monitoring. Recovery of the CD4:CD8 ratio to normal levels (≥ 1) with ART is often incomplete, however, which could reflect chronic immune activation occurring in persistent viral infection, including cytomegalovirus (CMV) and HIV infections [13][14][15][16]. Treatment of HIV during early infection protects the CD4:CD8 ratio, improving recovery rates, but late diagnosis, when significant immune damage and viral exposure have occurred, is still commonplace [6, 16,17].
Decline of the CD4:CD8 ratio occurs in the ageing general population. This has been associated with increased mortality rates in large Swedish studies and may be linked to CMV persistence [18,19]. Age-related mechanisms underlying CD4:CD8 ratio decline include loss of thymic output, CD8 T-cell oligoclonal outgrowth and accumulation of senescent T cells, including CD8 CD28 null T cells, which can be virally driven [20,21]. These changes contribute to an adaptive immune system that is less responsive and less flexible in older people. In PLWH, persistent CD4:CD8 ratio inversion can involve incomplete CD4 count recovery and/or persistently high CD8 count, and early initiation of ART supports normalization of the CD8 count [15,22,23]. The HIV reservoir could be a driver for persistent CD4:CD8 ratio abnormality [24]. Given the central role CD8 T cells play in anticancer and antiviral immunity, the recovery of this T-cell compartment is important for older people living with HIV. Identifying the associations with persistent CD4:CD8 ratio abnormality in PLWH despite ART may indicate selection criteria for those at risk of complications.
Previous studies have considered the association of CD4:CD8 ratio recovery in HIV-1 infection with morbidity and mortality outcomes, but with conflicting results regarding the utility of this as a predictive biomarker in PLWH [25,26]. In the general population, the association of CD4:CD8 ratio inversion with excess mortality affecting older people indicates that persistent inversion might be of particular importance in older people diagnosed with or living with HIV infection [18,27]. We therefore investigated the impact of age at HIV diagnosis and baseline CD8 count on CD4:CD8 ratio recovery. Here we show that treatment during primary HIV infection (PHI) was associated with recovery of the CD4:CD8 ratio to ≥ 1. Conversely, even after adjustment, older age at HIV diagnosis had a negative impact on CD4:CD8 ratio recovery, despite CD4 count gains, particularly in those aged > 50 years.

Study design and setting
A retrospective study of PLWH starting ART during the 10-year period from 1 January 2007 to 31st December 2016 was carried out, using routinely collected clinical data from the HIV out-patient services at Imperial College Healthcare NHS Trust, London, UK. The flow diagram for data collection is shown in Figure 1.

Inclusion criteria
Data were collected from the electronic records system and anonymized prior to inclusion. All records of a first positive HIV test between the dates of 1 January 2007 and 31 December 2016 were included (n = 1903). Tests for HIV infection using a third-or fourth-generation assay were conducted in the Clinical Pathology Laboratory as part of routine diagnostic procedures. Cross-linked anonymized database matching of data from the HIV Reservoir Targeting with Early Antiretroviral Therapy (HEATHER) study was undertaken to identify the records of those who had been diagnosed with primary HIV infection (PHI) and started on ART within 4 weeks [these patients were considered to have 'treated PHI' (tPHI)] [28]. Entry criteria for the HEATHER study defined PHI as one or more of: positive HIV antibody test within 6 months of a negative test; negative HIV antibody test with positive polymerase chain reaction (PCR), protein p24 antigen or viral load; positive recent incident assay test algorithm consistent with recent infection; equivocal HIV antibody test with rising optical density on retest within 2 weeks; or symptomatic HIV seroconversion illness with positive antigen test and fewer than four bands in the western blot assay. Information on diagnosis and treatment of PHI is routinely available to members of the clinical care team.

Exclusion criteria
The following were excluded from the final data set: duplicate medical records, those with no last recorded clinic appointment, those with no record of starting ART, those for patients aged < 18 years, those for patients who acquired HIV infection through vertical transmission, those with fewer than two CD4:CD8 measurements, those with no record of ART initiation, and those where HIV viral load was < 50 HIV-1 RNA copies/mL when ART was initiated. Those already on ART previously started elsewhere, or where there was exceptional natural viral control (elite controllers) were therefore excluded from the study. Records indicating failure to suppress HIV infection whilst on ART were excluded by identifying those with an HIV viral load ≥ 50 copies/ml at the end of the study period. All inclusion and exclusion criteria were applied to the raw data to generate the final data set on which all analyses were conducted.
Demographic data included age in years at the time of HIV testing, sex and ethnicity. CD4 and CD8 counts (cells/ lL) and HIV viral load (copies/mL) measurements were performed by the accredited Clinical Pathology Laboratory using standard flow cytometry and molecular techniques. Measurements at baseline and all subsequent measurements during the study period were included. Categorical variables were selected after graphical visualization of the spread of the data. Baseline CD8 counts were categorized as low (0-350 cells/lL; CD8 Lo ), low-normal (351-600 cells/ lL; CD8 LoN ), high-normal (601-900 cells/lL; CD8 HiN ), high (901-1500 cells/lL; CD8 Hi ) and very high (> 1500 cells/lL; CD8 VHi ) using reference ranges derived from the Clinical Pathology Laboratory performing the assays. Age at baseline was categorized into four groups, according to clinical relevance and spread of the data: 18-30 years (n = 284); 31-40 years (n = 306); 41-50 years (n = 177); and > 50 years (n = 109). The CD4:CD8 ratio and time on treatment during the study were derived mathematically. A normal CD4:CD8 ratio was defined as ≥ 1 [19].

Statistical analysis
The anonymized data set was made available in MICROSOFT EXCEL format. The data were cleaned and analysed using EX-CEL 2013 and IBM SPSS statistics v24.0 software (IBM, New York, NY, USA). The primary outcome was the most recent CD4:CD8 ratio measurement in those on ART with suppressed HIV infection (viral load < 50 copies/mL). Quantitative values were assessed for normality using the Shapiro-Wilk test. In univariate analyses, paired data were investigated using the Wilcoxon signed-rank test for nonparametric data. Analysis of inter-group variance was performed with the Kruskal-Wallis test. Post hoc analyses were conducted using Dunn's test for multiple comparisons. Univariate analyses of continuous data were conducted using Spearman's rank correlation. HIV viral load was log 10 transformed for univariate and multivariate analyses.
Multivariate analyses were conducted using the Cox proportional hazard regression model to estimate hazard ratios (HRs) with 95% confidence intervals (CIs) with duration of treatment as the timescale and persistent CD4:CD8 ratio inversion (< 1) as the outcome variable. Each participant accrued person-time of follow-up from the date of first starting HIV treatment until the last recorded CD4:CD8 ratio (the end of the study was 31 December 2016). Entry time was defined as the date of starting ART, and exit time was defined as the last recorded CD4:CD8 ratio prior to 1 January 2017. The multivariable model included age in years at date of diagnosis as a continuous variable, duration of ART, sex, baseline log 10 HIV viral load (copies/mL), whether the patient was treated during PHI, ethnicity and baseline CD4 and CD8 counts (cells/lL).

Cohort characteristics
Data from 876 records were analysed (Table 1)
The direction of CD8 count change with ART was contingent upon baseline CD8 count. In the CD8 Lo and CD8 LoN groups, CD8 counts increased (P < 0.001 and P < 0.001, respectively; Table 2). Those in the CD8 HiN group had no significant change in CD8 count (P = 0.165; Table 2). In the CD8 Hi and CD8 VHi groups (CD8 count > 900 cells/lL), ART reduced the CD8 count (P < 0.001 and P < 0.001; Table 2). Significant CD4:CD8 ratio gain was observed in all CD8 categories (P < 0.001 for all CD8 categories) ( Table 2). In pairwise analyses, those in the CD8 VHi category had a lower baseline and final CD4:CD8 ratio than those in any other CD8 category (Table S1). CD4:CD8 ratio recovery closest to optimal levels was observed in those in the CD8 LoN group, from a    Table 3). After adjustment, a higher baseline CD8 count was associated with persistent CD4:CD8 ratio inversion (P = 0.025) and there was a trend for lower CD8 counts (CD8 Lo , CD8 LoN and CD8 HiN ) to be associated with a lower risk of persistent CD4:CD8 ratio inversion (Table 3).

Nonlinear relationship of baseline CD8 count and viral load
Almost half the cohort, 423 of 876 patients (48.3%), had high (CD8 Hi ) or very high (CD8 VHi ) baseline CD8 counts. Previous work has indicated a potential association of AIDS mortality with high CD8 count [26]. Given that unchecked HIV replication might drive a high CD8 count, we investigated the association of baseline CD8 count with viral load. A nonlinear relationship of baseline CD8 count with viral load was observed; those in the CD8 VHi and CD8 Lo categories had similarly high viral loads (P = 1.00; Table 4). Compared with CD8 VHi , those in the CD8 LoN , CD8 HiN and CD8 Hi categories had lower baseline HIV viral load (P = 0.001, P < 0.001 and P < 0.001, respectively; Table 4). HIV viral load correlated negatively with CD8 count in the lowest three categories (r s = À0.159; P = 0.001) and positively with CD8 count in the highest two categories (r s = 0.257; P < 0.001). After adjustment, a higher baseline HIV viral load was associated with persistent CD4:CD8 ratio inversion (adjusted HR 1.245; 95% CI 1.112-1.393; P < 0.001; Table 3).

Impact of timing of ART on CD4:CD8 ratio recovery
In univariate analyses, there was an inverse association between CD4 count at baseline and duration of ART (r s = À0.399; P < 0.001), reflecting change in clinical practice over the past decade. There were 75 records where there had been a diagnosis of PHI and ART was started (tPHI). Of these patients, 70 of 75 (93.3%) were male, the median age was 31 (IQR 26-42) years and the median time from first positive HIV test was 1 (IQR 0-1) month. In those with tPHI, the median CD4 count was 530 (IQR 400-710) cells/lL, the median CD8 count was 1080 (IQR 820-1600) cells/lL, the median CD4:CD8 ratio was 0.53 (IQR 0.28-0.72) and the median HIV viral load was 4.88 (4.39-5.70) log 10 copies/mL. Of the 423 patients who had CD8 Hi or CD8 VHi counts, 49 of 423 (11.6%) had tPHI. In multivariate analysis, where recent HIV seroconversion was identified and treated, this had a protective effect on the CD4: CD8 ratio (adjusted HR 0.450; 95% CI 0.302-0.671; P < 0.001; Table 3). Increasing duration of ART was associated with a moderate risk reduction in persistent CD4:CD8 ratio inversion (adjusted HR 0.955; 95% CI 0.951-0.958; P < 0.001; Table 3).

Discussion
We present data from records of PLWH starting ART over one decade who achieved viral suppression, showing that older age at diagnosis was associated with suboptimal CD4:CD8 ratio recovery, particularly in those aged > 50 years. ART was beneficial for CD4:CD8 ratio normalization, but recovery to ≥ 1 occurred in only a third of patients (31.3%). Although CD4 count recovery was similar in younger and older age groups, CD8 count change did not occur in those diagnosed with HIV infection in the fifth decade or above. Those aged > 50 years had lower baseline and post-ART CD4:CD8 ratios, and older age at diagnosis was associated with persistent CD4:CD8 ratio inversion even after adjustment and including those who started ART during PHI. These findings have implications for the long-term health of PLWH into old age and for older people diagnosed with HIV infection.
With age, the immune system undergoes several maladaptive changes associated with physiological decline and the cumulative effect of chronic viral infection, especially CMV infection. Changes that affect adaptive immunity include loss of thymic output, reduced frequency of na€ ıve cells and oligoclonal outgrowth of T cells and accumulation of CD8 CD28 null T cells [20,21,29]. The immune risk profile, which encompasses several of these features, including CD4:CD8 ratio inversion, has been linked to increased mortality in older people [19]. CD8 T cells are particularly vulnerable to epigenetic changes with age [5]. Coupled with unfavourable changes in adaptive immunity over time in chronic HIV infection, as a consequence of immune activation, this may compound immunocompromise in older PLWH [30,31].
We observed a wide range of baseline CD8 counts, with predominance in the high or very high range, reflecting the impact of HIV replication on cytotoxic T lymphocytes. The nonlinear relationship with viral load may explain why adjusting for viral load reduced the impact of baseline CD8 count on CD4:CD8 ratio recovery. Reduction in CD8 count with ART occurred in those with a baseline CD8 count > 900 cells/lL, which included those with a baseline CD4 count > 700 cells/lL. CD8 count reduction is therefore a feature of immune recovery even where CD4 count is not suppressed. CD8 counts are not usually stratified against clinical outcomes, but as part of the CD4:CD8 ratio might be useful in this regard [10]. The most favourable CD4:CD8 ratio recovery rate occurred in those with a baseline CD8 count of 351-600 cells/lL and normalization of the CD4:CD8 ratio was rare outside the range of 350-900 cells/lL.
Lower baseline CD4 counts were observed in older PLWH, which may reflect both immune decline and delay in diagnosis leading to greater immunosuppression prior to starting ART [2,32]. We demonstrated a powerful protective effect of early diagnosis and treatment of HIV infection in promoting immune recovery (CD4:CD8 ratio ≥ 1), in line with previous reports [16,33]. In the tPHI group, the median CD8 count was high [1080 (IQR 820-1600) cells/lL], but those with tPHI accounted for less than one-fifth of all those with a baseline CD8 count in the > 900 cells/lL range. Treatment of PHI was therefore not solely accountable for the reduction in CD8 count with ART described. Initiation of ART, not just early treatment of HIV infection, could account for this effect. However, our data indicate that, whilst ART may be beneficial for CD4 count recovery, CD8 count and CD4:CD8 ratio normalization may be challenging in older PLWH.
A CD4:CD8 ratio of ≤ 1 was selected as the threshold for immune recovery in this study, reflecting previous work in HIV-infected cohorts [16,26]. A CD4:CD8 ratio threshold of 1.2 has been used in some studies, and mean CD4:CD8 ratio may lie above this threshold, depending on the population in question [14,34,35]. Evidence linking the CD4:CD8 ratio with undesirable immunological changes and poor clinical outcomes, including oxidative stress, cognitive disability and mortality, is associated with its inversion (CD4:CD8 ratio < 1) [18,[36][37][38]. There are insufficient data to determine whether a difference of CD4:CD8 ratio recovery of 0.2 (between 1 and 1.2) is immunologically or clinically meaningful in PLWH.

Limitations
We were unable to test for the effect of coinfecting pathogens, comorbidities or duration of viral suppression on CD4:CD8 ratio recovery, because of limitations in our data set. The presence of coinfecting pathogens, such as CMV, adversely affects the CD4:CD8 ratio [39][40][41]. Teasing out the difference between HIV-and CMV-mediated effects requires a large cohort, as the majority of PLWH are CMV-antibody positive [39]. Data from such studies, whilst of mechanistic significance, have limited clinical application in the absence of a safe and scalable vaccine or treatment that could meaningfully alter outcomes related to CMV persistence. The effects of other viruses, including hepatitis C virus, are more immediately modifiable with treatment [42], but data were insufficient to investigate this in the present study. Finally, although we showed relationships of age at diagnosis and baseline CD8 count with immune recovery, we did not have data on subsequent clinical outcomes.

Concluding remarks
The incidence and prevalence of HIV infection are significant in older people, in whom immunity is vulnerable to both age-related decline and unfavourable adaptation to chronic infection. In this study, treatment during PHI was protective, but CD4:CD8 ratio recovery was nonetheless suboptimal in those aged >50 years. These data indicate that targeted testing and treatment of older people should be a priority for PLWH, alongside enhanced surveillance where the CD4:CD8 ratio fails to normalize.