The associations between age and the development of laboratory abnormalities and treatment discontinuation for reasons other than virological failure in the first year of highly active antiretroviral therapy
Abstract
Objective
The aim of this study was to describe the relationship between age and the time to treatment discontinuation in the absence of virological failure as well as the development of specific laboratory abnormalities, in patients starting highly active antiretroviral therapy (HAART) for the first time.
Methods
Analyses included 8708 antiretroviral-naïve patients from the UK Collaborative HIV Cohort (CHIC) study who started HAART from 1998 onwards. We considered time to the first discontinuation of any drug in the initial HAART regimen for reasons other than virological failure; the association between this and age at the start of HAART was determined using proportional hazards regression after adjustment for potential confounders. The incidence of specific laboratory abnormalities in the first year after starting HAART was compared in those of different ages using multiple logistic regression.
Results
A total of 2650 patients discontinued at least one drug in their HAART regimen in the first year for reasons other than virological failure; after controlling for confounders, those aged < 30 years at the time of starting HAART were more likely to discontinue than those aged 30–39 years [relative hazard (RH) 1.12; 95% confidence interval (CI) 1.01, 1.24] as were those aged 50 years (RH 1.14; 95% CI 1.00, 1.31). There were strong associations between greater age and raised total cholesterol, decreased haemoglobin and raised triglycerides over the first year, although the latter disappeared after adjustment for pre-HAART levels, suggesting that this finding reflected higher pre-HAART triglyceride levels in older individuals.
Conclusions
Continued attempts to improve the tolerability of HAART regimens may help to sustain the good outcomes in all age groups over the longer term.
Introduction
As a consequence of improvements in survival following the introduction of highly active antiretroviral therapy (HAART), an increasing proportion of individuals living with HIV are surviving to greater ages [1–3]. While virological and immunological responses to HAART have been described in those of different ages [3–5], the relationship between an individual's age and his/her risk of development of drug toxicities remains unclear. Two large studies have recently reported that individuals aged 50 years have a higher risk of adverse events on HAART than younger individuals [5,6]. However, opinion remains divided as to whether there is an impact of age on the time to discontinuation of HAART, particularly where the discontinuation is a consequence of a drug toxicity [7–15].
We have assessed the relationship between age and the time to treatment discontinuation in the absence of virological failure as well as the development of specific laboratory abnormalities in the first year after starting HAART in a large, diverse cohort of individuals in the UK.
Methods
The UK Collaborative HIV Cohort (UK CHIC) Study is a collaboration of some of the largest centres for the care of HIV-infected individuals in the UK. The criteria for inclusion of an individual in the study were that a person was HIV positive, was aged over 16 years, and had attended one of the collaborating centres for care at any time after 1 January 1996 [16]. The data set used for the present analysis contains information on 25 274 patients seen for care at 10 centres (Chelsea & Westminster NHS Trust; King's College Hospital; Mortimer Market Centre; St Mary's Hospital; Royal Free Hospital; Barts and the London; Homerton University Hospital NHS Foundation Trust; North Middlesex University Hospital; Brighton and Sussex University Hospital; Western General Hospital, Edinburgh) (see Appendix A). Each centre provided electronic data in a standardized format on demographic characteristics, AIDS diagnoses and mortality, laboratory data (CD4/CD8 cell counts, viral loads and markers of drug toxicity) and antiretroviral treatment. The analyses presented here are based on data collected to late 2005. The project was approved by a Multicentre Research Ethics Committee and by local ethics committees.
All antiretroviral (ARV)-naïve patients starting HAART for the first time after 1 January 1998 were included. HAART was defined as any treatment combination including one or more of a protease inhibitor (PI), a nonnucleoside reverse transcriptase inhibitor (NNRTI), abacavir, tenofovir or enfuvirtide, regardless of the number of drugs received. Follow-up was defined from the start of HAART until 1 year later or 3 months after the patient's last clinic attendance if this was earlier (3 months being the usual interval between scheduled clinic appointments).
Analyses of treatment discontinuation for reasons other than virological failure
For each individual, the date of the first discontinuation of any drug in the initial regimen for reasons other than virological failure was identified. To be classified as a discontinuation the patient had to remain off the drug for at least 14 days. Changes of drug formulations (e.g. switching from zidovudine (ZDV) and lamivudine to combivir) or reductions in doses were not counted as discontinuations (information on dose is not collected in UK CHIC). As information on the reasons for discontinuation is not available from all participating centres, drug discontinuations were classified as being for reasons other than virological failure if (1) the discontinuation occurred in the first 3 months after starting HAART; (2) the discontinuation occurred in months 4–6 after HAART while the post-HAART viral load was <1000 HIV-1 RNA copies/mL; (3) the discontinuation occurred in months 7–12 after HAART while the viral load was <50 copies/mL; or (4) the discontinuation occurred from month 3 onwards but prior to the first post-HAART viral load measurement. All other drug discontinuations were classified as attributable to virological failure. For (2) and (3), only the latest viral load prior to discontinuation was considered.
Kaplan–Meier analyses described the time to discontinuation for reasons other than virological failure. Follow-up on patients who discontinued a drug for virological failure was right-censored on the date of discontinuation – thus, this analysis is based on the cause-specific hazard in the first year after starting HAART. Age at start of HAART was categorized into four groups (<30, 30–39, 40-49 and 50 years). Cox regression models assessed the impact of age on the cause-specific hazard, both before and after adjusting for possible pre-HAART confounders (sex, primary mode of HIV transmission, ethnicity, calendar year, type of initial regimen, CD4 cell count and HIV RNA). Visual inspection of plots of the log [–log(survival)] function against log (time) suggested that it was reasonable to assume that the hazards in the four age groups were proportional over the study period. Further analyses considered whether any association with age could be explained by different immunological responses in those of different ages, through the incorporation of the latest CD4 cell count as a time-updated covariate.
Analyses of the incidence of laboratory-defined abnormalities
All patients were studied for the presence of any laboratory abnormality in the first year after HAART, regardless of whether drugs were discontinued over this time. Abnormal markers considered were: triglycerides 2.3 mmol/L; total cholesterol 6.2 mmol/L; haemoglobin <11.5 g/dL for women and <13.5 g/dL for men; bilirubin >17 μmol/L; alanine aminotransferase (ALT) >40 IU/L. These thresholds are generally less extreme than those used to define adverse events in randomized trials but were chosen to reflect levels at which concern would be raised about the patient's status. Patients were included in each analysis if they had at least one measurement of the marker in the first year after starting HAART. It was not a requirement for most of these analyses that patients had pre-HAART levels, but patients whose pre-HAART values were known to be abnormal were excluded from the relevant analysis. The only exception to this was for analyses that included adjustment for the baseline level of each marker in which those with missing baseline data were excluded. The frequency of an abnormal measurement was compared in the four age groups using χ2 tests, and multivariable logistic regression methods assessed independent relationships with age after controlling for (a) clinical and demographic characteristics at the start of HAART, and (b) clinical and demographic characteristics as well as pre-HAART marker levels (in the subset where these were available). As the relationship between greater age and the development of toxicities was approximately linear, age was included as a continuous covariate in these analyses. Further sensitivity analyses also considered the relationship between age and the highest/lowest (as appropriate) measurement within the first year after starting HAART, and considered the incidence of abnormal laboratory markers only in patients whose initial HAART regimens included drugs known to be associated with these abnormalities (i.e. nevirapine for analyses of ALT; lopinavir, ritonavir, stavudine, ZDV or nelfinavir for analyses of total cholesterol and triglycerides; ZDV for analyses of haemoglobin; indinavir for analyses of bilirubin – atazanavir was not included in this latter analysis because of the small number of patients receiving this drug).
All statistical analyses were performed using sas version 9.1 (SAS Institute, Cary, NC, USA).
Results
A total of 8708 patients in UK CHIC who were naïve to antiretroviral treatment started HAART from 1 January 1998 onwards. The age distribution of patients at the time of starting HAART was: <30 years, 1631 (18.7%); 30–39 years, 4260 (48.9%); 40–49 years, 2045 (23.5%); and 50 years, 772 (8.9%). Median [interquartile range (IQR)] follow-up was 2.7 (1.1–4.8) years; 6726 (77.2%) were followed for at least 1 year after starting HAART. Patient characteristics stratified by age group are shown in Table 1. Those who were older were more likely to be white males, infected via sex between men, to have started HAART in more recent years, to have started non-PI HAART regimens and to have lower CD4 cell counts and higher HIV RNA levels at the time of starting HAART. Follow-up after starting HAART was slightly shorter in older individuals.
| Overall | Age group | P-value | ||||
|---|---|---|---|---|---|---|
| <30 years | 30–39 years | 40–49 years | 50 years | |||
| Number of patients | 8708 | 1631 | 4260 | 2045 | 772 | |
| Male sex | 6330 (72.7) | 911 (55.9) | 3094 (72.6) | 1680 (82.2) | 645 (83.6) | 0.0001 |
| Primary mode of HIV infection | ||||||
| Sex between men | 4538 (52.1) | 690 (42.3) | 2247 (52.8) | 1156 (56.5) | 445 (57.6) | |
| Injecting drug use | 319 (3.7) | 35 (2.2) | 178 (4.2) | 93 (4.6) | 13 (1.7) | |
| Sex between men and women | 3435 (39.5) | 808 (49.5) | 1664 (39.1) | 692 (33.8) | 271 (35.1) | |
| Other/not known | 416 (4.8) | 98 (6.0) | 171 (4.0) | 104 (5.1) | 43 (5.7) | 0.0001 |
| Ethnicity | ||||||
| White | 4615 (53.0) | 670 (41.1) | 2212 (51.9) | 1235 (60.4) | 498 (64.5) | |
| Black African | 2580 (29.6) | 626 (38.4) | 1327 (31.2) | 497 (24.3) | 130 (16.8) | |
| Other | 828 (9.5) | 207 (12.7) | 381 (8.9) | 165 (8.1) | 75 (9.7) | |
| Not known | 685 (7.9) | 128 (7.9) | 340 (8.0) | 148 (7.2) | 69 (8.9) | 0.0001 |
| Year of starting HAART | ||||||
| 1998–1999 | 2036 (23.4) | 399 (24.5) | 1050 (24.7) | 406 (19.9) | 181 (23.5) | |
| 2000–2001 | 2071 (23.8) | 405 (24.8) | 1036 (24.3) | 460 (22.5) | 170 (22.0) | |
| 2002–2003 | 2418 (27.8) | 436 (26.7) | 1193 (28.0) | 568 (27.8) | 221 (28.6) | |
| 2004–2005 | 2183 (25.1) | 391 (24.0) | 981 (23.0) | 611 (29.9) | 200 (25.9) | 0.0001 |
| Type of initial regimen | ||||||
| PI-based | 2155 (24.8) | 439 (26.9) | 1048 (24.6) | 496 (24.3) | 172 (22.3) | |
| NNRTI-based | 5861 (67.3) | 1078 (66.1) | 2891 (67.9) | 1362 (66.6) | 530 (68.7) | |
| Other combination | 692 (8.0) | 114 (7.0) | 321 (7.5) | 187 (9.1) | 70 (9.1) | 0.03 |
| CD4 count (cells/μL) | ||||||
| Median (IQR) | 181 (90, 277) | 210 (110, 311) | 180 (90, 272) | 172 (80, 260) | 160 (70, 251) | 0.0001 |
| HIV RNA (log10 copies/mL) | ||||||
| Median (IQR) | 4.9 (4.3, 5.4) | 4.8 (4.1, 5.3) | 4.9 (4.3, 5.4) | 5.0 (4.4, 5.5) | 5.1 (4.5, 5.6) | 0.0001 |
| Follow-up after starting HAART (years) | ||||||
| Median (IQR) | 2.7 (1.1, 4.8) | 2.7 (1.1, 4.9) | 2.8 (1.2, 5.0) | 2.4 (1.0, 4.6) | 2.5 (1.0, 4.7) | 0.0001 |
| More than 1 year of follow-up | 6726 (77.2) | 1264 (77.5) | 3370 (79.1) | 1519 (74.3) | 573 (74.2) | 0.0001 |
- Entries are n (%) unless otherwise stated.
- IQR, interquartile range; NNRTI, nonnucleoside reverse transcriptase inhibitor; PI, protease inhibitor.
Treatment discontinuation for reasons other than virological failure
Overall, 5071 patients (58.2%) discontinued at least one drug in their initial regimen; 3146 (62.0%) of these discontinuations were in the first year after starting HAART. A total of 2650 patients (84.2%) who discontinued at least one drug in their regimen in the first year of HAART were classified as discontinuing the drug for reasons other than virological failure. This total included 1420 patients who discontinued the drug in the first 3 months, 588 patients who discontinued the drug in months 4–6 with a pre-discontinuation viral load <1000 copies/mL, 501 patients who discontinued the drug in months 7–12 with a pre-discontinuation viral load <50 copies/mL and 141 patients who discontinued the drug from month 3 onwards prior to their first post-HAART viral load measurement (there was a tendency for patients in the youngest group to be more likely to stop treatment in the first 3 months). Of the 2650 patients who did discontinue treatment for reasons other than virological failure in the first year of HAART, most (1880; 70.9%) remained on treatment (either a new or a reduced regimen). Of the 770 who discontinued all treatment, 498 (64.7%) subsequently restarted at a later time.
Kaplan–Meier estimates of the cumulative percentage discontinuing for reasons other than virological failure were 15.1, 25.1, 29.8 and 33.5% at 3, 6, 9 and 12 months after starting HAART, respectively. By 12 months, 36.2, 32.2, 33.5 and 35.1% of those aged <30, 30–39, 40–49 and 50 years, respectively (P=0.008), had discontinued for these reasons. In univariable analyses, those aged <30 years at the time of starting HAART were more likely to discontinue HAART for reasons other than virological failure when compared with individuals aged 30–39 years [relative hazard (RH) 1.19; 95% confidence interval (CI) 1.07, 1.31; P=0.0008]. Those aged 50 years were also more likely to discontinue for reasons other than virological failure, although this was nonsignificant (RH 1.10; 95% CI 0.96, 1.26; P=0.19). Those aged 40–49 years had discontinuation rates that were similar to those seen in individuals aged 30–39 years (RH 1.04; 95% CI 0.94, 1.14; P=0.46). After controlling for potential confounders (Table 2) the youngest (<30 years) and oldest (50 years) groups remained at higher risk of discontinuation for reasons other than virological failure. Discontinuation for reasons other than virological failure was also more common in women and in those whose pre-HAART CD4 count was <50 or 350 cells/μL. Discontinuation was less common in those with another/unknown mode of infection, in those starting an NNRTI-based regimen and in those starting HAART in 1998–1999 (possibly because of the smaller number of treatment options available) or from 2004 onwards. Additional analyses that incorporated the latest CD4 count revealed that the risk of treatment discontinuation increased by 2% (RH 1.02; 95% CI 1.01, 1.03; P=0.006) for each 50 cells/μL increment in the latest CD4 count. Adjustment for this did not, however, greatly modify the reported associations with age, although the association with the oldest age group was strengthened (RH 1.18; 95% CI 1.03, 1.36; P=0.02).
| Factor | Relative hazard | 95% CI | P-value |
|---|---|---|---|
| Age group (years) | |||
| <30 | 1.12 | 1.01, 1.24 | 0.03 |
| 30–39 | 1 | – | – |
| 40–49 | 1.06 | 0.97, 1.17 | 0.21 |
| 50 | 1.14 | 1.00, 1.31 | 0.06 |
| Male sex | 0.72 | 0.64, 0.81 | 0.0001 |
| Primary mode of HIV infection | |||
| Sex between men | 1 | – | – |
| Sex between men and women | 1.06 | 0.87–1.30 | 0.56 |
| Injecting drug use | 0.91 | 0.81–1.02 | 0.11 |
| Other/not known | 0.77 | 0.63–0.95 | 0.02 |
| Year of starting HAART | |||
| 1998–1999 | 0.82 | 0.73, 0.91 | 0.0003 |
| 2000–2001 | 0.97 | 0.88, 1.07 | 0.54 |
| 2002–2003 | 1 | – | – |
| 2004–2005 | 0.87 | 0.78, 0.97 | 0.01 |
| Type of initial regimen | |||
| PI-based | 1.22 | 1.12, 1.34 | 0.0001 |
| NNRTI-based | 1 | – | – |
| Other combination | 1.35 | 1.18, 1.54 | 0.0001 |
| CD4 count (cells/μL) | |||
| <50 | 1.16 | 1.02, 1.32 | 0.02 |
| 50–199 | 0.91 | 0.82, 1.01 | 0.06 |
| 200–349 | 1 | – | – |
| 350 | 1.44 | 1.27, 1.63 | 0.0001 |
| Not known | 1.12 | 0.99, 1.27 | 0.07 |
- CI, confidence interval; NNRTI, nonnucleoside reverse transcriptase inhibitor; PI, protease inhibitor.
The incidence of laboratory-defined abnormalities
Pre-HAART triglyceride, total cholesterol, bilirubin and ALT levels were higher in those who were older, whereas haemoglobin levels were lower (Table 3). The incidence of abnormal laboratory measurements increased with age for triglycerides, total cholesterol and haemoglobin, and varied by age (although the linear trend was not as strong) for ALT (Table 4). There did not appear to be a strong relationship between age and bilirubin levels. Adjustment for patient characteristics at baseline had only a small impact on the relationships between age and these abnormalities (Table 5), with the exception of elevated ALTs which did not remain associated with greater age after adjustment. After adjusting for pre-HAART triglycerides, the relationship between greater age and elevated triglycerides was removed, suggesting that older individuals were at increased risk of an elevated triglyceride level on HAART because their pre-HAART levels were already higher. In contrast, elevated cholesterol and lowered haemoglobin remained more common in older individuals after adjusting for pre-HAART levels of these markers. In particular, each 5-year increase in age was associated with a 12% increase in the risk of elevated cholesterol (adjusted odds ratio 1.12; 95% CI 1.05, 1.20; P=0.0009) and a 6% increase in the risk of lowered haemoglobin (adjusted odds ratio 1.06; 95% CI 1.00, 1.12; P=0.05).
| Overall | Age group (years) | P-value | ||||
|---|---|---|---|---|---|---|
| <30 | 30–39 | 40–49 | 50 | |||
| Triglycerides (mmol/L) | ||||||
| n | 2817 | 456 | 1395 | 718 | 248 | |
| Median (IQR) | 1.4 (1.0, 2.1) | 1.2 (0.9, 1.8) | 1.4 (1.1, 2.3) | 1.6 (1.1, 2.3) | 1.7 (1.2, 2.2) | 0.0001 |
| n (%) with abnormal value | 581 (20.6) | 64 (14.0) | 276 (19.8) | 182 (25.4) | 59 (23.8) | 0.0001 |
| Total cholesterol (mmol/L) | ||||||
| n | 3017 | 493 | 1486 | 760 | 278 | |
| Median (IQR) | 4.1 (3.5, 4.8) | 3.9 (3.3, 4.5) | 4.0 (3.5, 4.7) | 4.3 (3.6, 5.0) | 4.4 (3.8, 5.2) | 0.0001 |
| n (%) with abnormal value | 126 (4.2) | 13 (2.6) | 51 (3.4) | 46 (6.1) | 16 (5.8) | 0.004 |
| Bilirubin (μmol/L) | ||||||
| n | 5039 | 879 | 2495 | 1222 | 443 | |
| Median (IQR) | 8 (6, 11) | 8 (6, 10) | 8 (6, 11) | 8 (6, 11) | 8 (6, 11) | 0.01 |
| n (%) with abnormal value | 284 (5.6) | 41 (4.7) | 132 (5.3) | 80 (6.6) | 31 (7.0) | 0.13 |
| Haemoglobin (g/dL) | ||||||
| n | 4372 | 707 | 2162 | 1096 | 407 | |
| Median (IQR) | 13.2 (11.7, 14.5) | 13.2 (11.3, 14.5) | 13.3 (11.8, 14.5) | 13.2 (11.9, 14.4) | 12.8 (11.4, 14.1) | 0.0006 |
| n (%) with abnormal value | 2048 (48.8) | 305 (43.1) | 955 (44.2) | 547 (49.9) | 241 (59.2) | 0.0001 |
| ALT (IU/L) | ||||||
| n | 4036 | 664 | 2009 | 1005 | 358 | |
| Median (IQR) | 29 (20, 45) | 25 (17, 38) | 30 (20, 47) | 30 (21, 45) | 30 (21, 46) | 0.0001 |
| n (%) with abnormal value | 1193 (29.6) | 151 (22.7) | 633 (31.5) | 296 (29.5) | 113 (31.6) | 0.0003 |
- ALT, alanine aminotransferase; IQR, interquartile range.
| Abnormal: | Total | Age group (years)† | P-value | |||
|---|---|---|---|---|---|---|
| <30 | 30–39 | 40–49 | 50 | |||
| Triglycerides | 1202/3232 (37.2) | 166/601 (27.6) | 622/1618 (38.4) | 298/742 (40.2) | 116/271 (42.8) | 0.0001 |
| Total cholesterol | 811/3856 (21.0) | 96/687 (14.0) | 368/1912 (19.3) | 233/913 (25.5) | 114/344 (33.1) | 0.0001 |
| Bilirubin | 774/4780 (16.2) | 125/846 (14.8) | 376/2386 (15.8) | 202/1135 (17.8) | 71/413 (17.2) | 0.26 |
| Haemoglobin | 1238/2420 (51.2) | 202/427 (47.3) | 616/1252 (49.2) | 319/566 (56.4) | 101/175 (57.7) | 0.0006 |
| ALT | 1145/3000 (38.2) | 172/528 (32.6) | 572/1461 (39.2) | 308/748 (41.2) | 93/263 (35.4) | 0.01 |
- * Patients were excluded from these analyses if their pretreatment values were known to be abnormal.
- Entries shown are the number of patients with an abnormal value divided by the number of patients with any post-HAART measurement of the marker.
- † † Values in parentheses indicate percentages.
- ALT, alanine aminotransferase.
| All patients | Patients receiving drugs known to be related tospecific toxicities† | |||||
|---|---|---|---|---|---|---|
| OR | 95% CI | P-value | OR | 95% CI | P-value | |
| Abnormal triglycerides | ||||||
| Unadjusted | 1.11 | 1.06, 1.15 | 0.0001 | 1.11 | 1.06, 1.16 | 0.0001 |
| Adjusted for baseline characteristics* | 1.07 | 1.02, 1.12 | 0.004 | 1.07 | 1.02, 1.12 | 0.005 |
| Also adjusted for pre-HAART triglycerides | 0.98 | 0.92, 1.05 | 0.53 | 0.98 | 0.91, 1.05 | 0.56 |
| Abnormal cholesterol | ||||||
| Unadjusted | 1.21 | 1.16, 1.26 | 0.0001 | 1.21 | 1.16, 1.27 | 0.0001 |
| Adjusted for baseline characteristics | 1.23 | 1.17, 1.28 | 0.0001 | 1.23 | 1.17, 1.29 | 0.0001 |
| Also adjusted for pre-HAART cholesterol | 1.12 | 1.05, 1.20 | 0.0009 | 1.14 | 1.06, 1.22 | 0.0004 |
| Abnormal bilirubin | ||||||
| Unadjusted | 1.03 | 0.99, 1.08 | 0.19 | 1.10 | 0.89, 1.35 | 0.38 |
| Adjusted for baseline characteristics | 1.02 | 0.97, 1.77 | 0.51 | 1.06 | 0.85, 1.32 | 0.61 |
| Also adjusted for pre-HAART bilirubin | 1.01 | 0.95, 1.06 | 0.82 | 1.01 | 0.77, 1.31 | 0.97 |
| Abnormal haemoglobin | ||||||
| Unadjusted | 1.10 | 1.05, 1.16 | 0.0002 | 1.09 | 1.02, 1.16 | 0.02 |
| Adjusted for baseline characteristics | 1.11 | 1.06, 1.17 | 0.0001 | 1.11 | 1.03, 1.19 | 0.005 |
| Also adjusted for pre-HAART haemoglobin | 1.07 | 1.01, 1.14 | 0.03 | 1.06 | 0.98, 1.16 | 0.15 |
| Abnormal ALT | ||||||
| Unadjusted | 1.05 | 1.00, 1.09 | 0.04 | 1.09 | 0.98, 1.20 | 0.11 |
| Adjusted for baseline characteristics | 1.02 | 0.98, 1.07 | 0.38 | 1.03 | 0.92, 1.15 | 0.58 |
| Also adjusted for pre-HAART ALT | 1.00 | 0.95, 1.06 | 0.97 | 1.01 | 0.89, 1.15 | 0.85 |
- * Baseline characteristics included in adjustment: sex, ethnicity, risk group, calendar year, type of initial HAART regimen, CD4 cell count and HIV RNA level at initiation of HAART.
- † The drugs known to be related to specific abnormalities are: triglycerides/cholesterol - zidovudine, stavudine, lopinavir, ritonavir and nelfinavir; bilirubin - indinavir; haemoglobin - zidovudine; ALT - nevirapine.
- ALT, alanine aminotransferase; CI, confidence interval; HAART, highly active antiretroviral therapy; OR, odds ratio.
Similar conclusions were reached when analyses considered the maximum (or minimum for haemoglobin) measurement recorded in the year after starting HAART (data not shown). The only exception to this was that minimum haemoglobin levels did not differ significantly by age after controlling for pre-HAART haemoglobin levels. Sensitivity analyses including only patients who were receiving drugs previously reported to be associated with each specific abnormality gave similar results.
Discussion
While HAART has undoubtedly had a positive impact on patient health, virtually all ARV drugs are now known to have the potential to cause some degree of toxicity [17,18]. While many of these toxicities are manageable, a substantial proportion of patients discontinue their initial HAART regimen because of toxicities [8]. In one Spanish study [19], 7% of hospital admissions from 1998 to 2004 were attributable to toxicities, and the development of toxicities is associated with a higher rate of discontinuation of subsequent HAART regimens [20,21] as well as more rapid clinical progression [7].
In this large study, we have demonstrated that those aged <30 years, as well as those aged 50 years, were at higher risk of HAART discontinuation for reasons other than virological failure. Our decision to classify discontinuations into those occurring because of virological failure and those occurring for other reasons was a pragmatic one and reflects the fact that in observational studies such as this it may be difficult to differentiate between treatment discontinuations that are a result of real or perceived toxicities and those that are described as being a result of patient or clinician ‘choice’. We restricted follow-up to the first year after starting HAART, as most treatment switches in this period are a result of toxicities rather than virological failure [7,8] and to minimize any concerns regarding bias caused by competing events (i.e. discontinuations resulting from virological failure). Analyses that modified the definition of discontinuation for reasons other than virological failure (e.g. with the use of a viral load threshold of 400 copies/mL in months 4–6 rather than 1000 copies/mL) produced similar conclusions. Our finding of a higher rate of discontinuations in younger individuals is consistent with another study [22] in which treatment interruptions were more frequent in younger individuals, largely because of poorer adherence in this group. While this factor might also contribute to our findings, poorer adherence would be expected to lead to a rebound in viral load, and thus in this situation it is more likely that discontinuation would be attributed to virological failure. Our finding of only a weak association between treatment discontinuation and greater age is consistent with a lack of effect noted in other studies [9,10,14] but may reflect the fact that any increased risk of toxicity-discontinuation in older individuals may be countered by an improved adherence in these individuals.
We found a strong association between greater age and levels of several laboratory markers, including elevated total cholesterol, elevated triglycerides and lowered haemoglobin, consistent with findings from other studies [23–26]. While many studies have been limited by a small number of older individuals, two large studies have recently reported on this issue. Orlando et al. [6] reported findings from a case–control study of 159 individuals starting HAART aged 50 years and 118 controls starting HAART aged 25–35 years. Over the first year of HAART, cases were more likely than controls to experience abnormal glucose, total cholesterol, high-density lipoprotein cholesterol, triglyceride, and creatinine levels, but were less likely to experience abnormal ALT levels. These comparisons were not, however, adjusted for the demographic and clinical status of patients at the start of HAART; as cases were more likely to be male, to be heterosexual and to have more advanced HIV disease and more abnormal pretreatment levels of each marker than controls, these differences could be explained by differences in patient characteristics (including a possible higher rate of hepatitis C virus coinfection) between cases and controls. Silverberg et al. [4] reported that abnormal glucose, lipid, neutrophil, haemoglobin and creatinine levels were all more common in patients older than 50 years compared with younger individuals. The results from this latter study were remarkably similar to our own, despite a very different ethnic and HIV risk profile of the patients included, and the requirement for patients to have received medical care through a single health insurance system.
There are a number of reasons why a relationship might be expected between greater age and the development of toxicities. Firstly, as the HIV-infected population ages, infected individuals are likely to develop other clinical conditions typical of an aging population [27] – the medications used to treat any such comorbidities may interact with ARV drugs to leave these individuals at a higher risk of toxicity development [18,28]. Secondly, as older individuals are likely to be more adherent to therapy [29], the incidence of some toxicities may also be higher in this group because cumulative exposure to the drug is greater. Thirdly, it is possible that changes in some of the markers considered (e.g. total cholesterol levels) may reflect a return to health of patients starting HAART rather than a real toxicity of treatment [30]. Finally, we may also expect to see changes in some laboratory markers as individuals age. One of the limitations of most studies that have considered this issue is that it is difficult to determine the extent to which many of these apparent ‘toxicities’ would have occurred anyway regardless of HAART. Our analyses that adjust for pre-HAART marker levels go some way towards addressing this issue, as older individuals would be expected to have more abnormal levels of each marker prior to starting treatment. While a preferable analysis would incorporate a control group of similarly aged patients who did not receive HAART, the number of older patients who remain off therapy in most studies, including our own, is generally too small for such an analysis.
Our analyses of laboratory-defined abnormalities are restricted to individuals with a measurement in the first year after starting HAART. While the routine monitoring of these laboratory markers is now recommended [31], the speed at which monitoring has become routine has varied [32]. In the early years of HAART, information on laboratory markers prior to starting HAART was often unavailable or may have been performed only on a subgroup thought to be at high risk. Thus, restricting the data set in this way may overestimate the incidence of these toxicities if a high-risk group has been selected for monitoring.
Some further points about our cohort should be noted. While patients included in the UK CHIC Study are broadly representative of those infected with HIV in the UK [33], there is a slight overrepresentation of white males and those who acquired their HIV through sex between men. As care for HIV in the UK is not dependent on a patient's ability to pay for treatment or their health insurance status, our cohort includes a broad and diverse group of patients. Many of the clinics that participate in the UK CHIC Study are large university hospital-based clinics; prescribing patterns and responses to adverse events in these clinics may differ from those elsewhere in the country.
In conclusion, we report a high rate of treatment discontinuations for reasons other than virological failure among patients starting HAART for the first time. Discontinuation rates were higher in those aged <30 and 50 years, even after controlling for potential confounders. Those who were older were more likely to experience elevated total cholesterol and triglycerides and lowered haemoglobin than younger individuals. Continued attempts to improve the tolerability of HAART regimens may help to sustain the good outcomes in these age groups over the longer term.
Acknowledgements
Funding for this work has been obtained from the Medical Research Council, UK (grants G0000199 and G0600337). The views expressed in this manuscript are those of the researchers and not necessarily those of the MRC.
References
Appendix A
UK CHIC Steering Committee: Andrew N. Phillips, Richard Gilson, Philippa Easterbrook, Martin Fisher, Brian Gazzard, Margaret Johnson, John Walsh, Clifford Leen, Chloe Orkin, Jane Anderson, Deenan Pillay, Valerie Delpech, Achim Schwenk, David Dunn, Mark Gompels, Teresa Hill, Kholoud Porter, Abdel Babiker and Caroline Sabin.
Central Co-ordination: Royal Free and University College, London (Loveleen Bansi, Teresa Hill, Andrew Phillips and Caroline Sabin); Medical Research Council Clinical Trials Unit (MRC CTU), London (Abdel Babiker, David Dunn, Kholoud Porter and Stephen Sheehan).
Participating Centres: King's College Hospital, London (Philippa Easterbrook, Anele Waters, Dorian Crates and Siti Mohamed-Saad); Brighton and Sussex University Hospitals NHS Trust (Martin Fisher, Nicky Perry, Anthony Pullin, Duncan Churchill and Wendy Harris); Chelsea and Westminster NHS Trust, London (Brian Gazzard, Steve Bulbeck, Sundhiya Mandalia and Jemima Clarke); Mortimer Market Centre, Royal Free and University College Medical School (RFUCMS), London (Richard Gilson, Julie Dodds, Andy Rider and Ian Williams); Health Protection Agency – Centre for Infections, London (Valerie Delpech); Royal Free NHS Trust and RFUCMS, London (Margaret Johnson, Mike Youle, Fiona Lampe, Colette Smith, Helen Gumley, Clinton Chaloner and Dewi Ismajani Puradiredja); St Mary's Hospital, London (John Walsh, Jonathan Weber, Shane Cashin, Christian Kemble, Nicola Mackie and Alan Winston); Barts and The London NHS Trust, London (Chloe Orkin, Rachel Thomas and Kevin Jones); Homerton Hospital, London (Jane Anderson, Selina Gann and Kevin Jones); Western General Hospital, Edinburgh (Clifford Leen and Alan Wilson); North Middlesex (Achim Schwenk and Jonathan Ainsworth).
