Lipodystrophy-associated morphological, cholesterol and triglyceride abnormalities in a population-based HIV/AIDS treatment database

Objective: To provide population-based estimates of the prevalence of lipodystrophy syndrome and constituent symptoms and to identify correlates of prevalent symptomology.

Cited Here...: Participants in a province-wide HIV/AIDS treatment programme reported morphological and metabolic abnormalities. Probable lipodystrophy was defined as self-report of at least one morphological abnormality or both high cholesterol and triglyceride levels. Explanatory variables investigated included: age; sex; ethnicity; transmission risk group; CD4 cell count; plasma viral load; AIDS diagnosis; duration of infection; alternative therapy use; past, current and duration of use of antiretroviral therapy (ART) by class and specific drug; total duration of ART; and current adherence. Stepwise logistic regression identified possible determinates of lipodystrophy.

Cited Here...: Of 1035 participants, 50% appeared to have probable lipodystrophy, with 36% reporting peripheral wasting, 33% abdominal weight gain, 6% buffalo hump, and 10 and 12% increased triglyceride or cholesterol levels, respectively. In multivariate analysis, lipodystrophy was associated with older age (per year) (AOR 1.03; 95% CI 1.01, 1.04), the use of ingested alternative therapies (AOR 1.46; 95% CI 1.06, 2.01), having ever used protease inhibitors (PI) (AOR 2.63; 95% CI 1.89, 3.66), and duration of stavudine treatment (per year) (AOR 1.35; 95% CI 1.15, 1.58). In analysis limited to participants exposed to PI, after similar adjustment, the duration of lamivudine rather than stavudine treatment was associated with lipodystrophy (AOR 1.32; 95% CI 1.13, 1.53).

Cited Here...: Increased risk of abnormalities is associated with the use of PI, and the duration of stavudine and lamivudine treatment after adjustment for personal characteristics, clinical disease stage, duration of infection and detailed treatment history.

Introduction

In an era of highly active antiretroviral therapy (HAART), individuals with HIV/AIDS are utilizing novel therapies in ever increasingly complex regimens. Consequently, they may experience significant repression of HIV replication translating to increased survival times ^[1,2]. Coincident with these advances, reports have identified the emergence of morphological abnormalities involving the redistribution of body fat (lipodystrophy) characterized by peripheral fat wasting, increased visceral abdominal fat, breast hypertrophy in women, and enlargement of the dorsocervical fat pad (buffalo hump) ^[3-10]. Metabolic abnormalities have also been reported, including hypercholesterolemia and hypertryglyceridemia ^[10,11]. With abnormalities appearing in constellations, it appears that these symptoms represent a syndrome of HIV-associated lipodystrophy ^[5,10-12].

Reports estimating the prevalence of symptoms of lypodystrophy syndrome vary widely ^[13-17], with abnormalities occurring in 18-70% of patients exposed to protease inhibitor (PI)-containing regimens. Although initial identification was coincident with the widespread use of PI-inclusive therapy, the etiology of these abnormalities remains obscure. Several reports indicate a significant increase in patient-identified morphological changes associated with PI use ^[18-20], or identify specific PI associated with various morphological ^[5,21] and metabolic ^[22-27] abnormalities. Others have concluded that these abnormalities are not associated with the type of therapy or may occur in the absence of PI ^[3,26-30]. Several recent reports ^[31-33] indicated that these symptoms may be a function of nucleoside analogue reverse transcriptase inhibitor (NRTI) therapy rather than, or in addition to, PI use.

In-depth studies of select patient groups can contribute to our understanding of the natural history of symptoms and help to frame etiological investigation. Small clinical samples may allow for an assessment of the validity and reliability of outcome measures and extensive data collection through clinical evaluation based on standardized protocols.

Such studies have, however, inherent limitations. Clinical follow-up is costly and time consuming for both clinicians and subjects making large-scale or long-term commitment economically and logistically unfeasible. Smaller studies are also often hampered by a lack of statistical power. Meta-analysis is often inapplicable as a result of variations in outcomes assessed, measurement tools, information regarding confounders and sample populations. More importantly, non-random sample selection of clinical study participants may make study results susceptible to biases that can be difficult to identify or ameliorate.

Large, population-based observational databases can make a unique contribution to our understanding of emerging patterns of adverse events. At their best, such databases include large samples or entire populations of interest. Because of the large number of subjects followed, issues of sample size are negated and sub-samples are amenable to hypothesis-generating analyses. Broad-based observational studies also include a wide spectrum of individuals without the constraints of age, sex, disease stage and therapeutic regimen, often imposed or occurring by chance or selection processes in clinical studies; thereby reducing the risk of associated biases. Finally, findings of observational studies are useful in fulfilling additional criteria for causal inference, establishing consistency in comparison with studies of differing design and coherence with results of studies using similar design but based on other populations.

In this study, we report on estimates of prevalence and correlates of self-reported lipodystrophy and constituent symptoms within a large, population-based cohort of individuals treated with antiretroviral therapies (ART).

Methods

The HIV/AIDS drug treatment program (DTP) of British Columbia, Canada, provides antiretroviral medications free of charge to all eligible province residents. To receive ART at no cost individuals must be confirmed HIV positive, have a CD4 cell count of 500 cells/mm³ or less or a plasma viral load (pVL) of more than 5000 copies/ml, as is commensurate with current recommendations ^[34,35].

Individuals are automatically entered into the DTP when they are first prescribed any antiretroviral agent. At DTP entry and subsequent physician visits the participant's complete history (if any) of antiretroviral use, CD4 cell count, pVL, and disease stage are recorded. Typically, patients are followed-up at 3 month intervals, at which time prescriptions are renewed or altered based on treatment success and other clinical factors. For all patients, a complete prospective record of ART prescription is thus maintained, including the exact medications prescribed, the date of each prescription and actual prescription fill dates.

At DTP entry and at each annual anniversary participants are requested to provide additional information through a voluntary self-administered mailed questionnaire. Data collection includes information about sociodemographic characteristics, general health, the use of alternative therapies and detailed information regarding established therapy-related side-effects, as well as the occurrence of health problems or symptoms that may be attributed to HIV medications. Since 1998, these have included instances of symptoms considered to be key possible identifiers of lipodystrophy syndrome ^[36], specifically: fat wasting in the face, arms or legs; abdominal weight gain or breast enlargement; buffalo hump; increased triglyceride levels and increased total cholesterol.

Information regarding abnormalities is reliant on self-report; however, patients were directed to report metabolic disturbances only if they had been confirmed by their physician. In British Columbia, provincial guidelines for the treatment of HIV are distributed to all physicians who have prescribed antiretroviral agents, and recommend fasting laboratory tests for total cholesterol and triglyceride levels as standard care; however, the actual fasting state at the time of tests cannot be verified ^[35]. Local laboratory standards define the upper limit of normal for triglyceride levels and total cholesterol as 2.3 and 5.2 mg/dl, respectively.

As no standard definition of lipodystrophy exists, for the purposes of this analysis, patients were deemed to have probable lipodystrophy if they had at least one of the following: peripheral wasting; weight gain confined to the abdomen, or breast enlargement in women; buffalo hump; or both increased triglyceride levels and increased total cholesterol. Instances of high blood pressure, type 2 diabetes mellitus, peripheral neuropathy, sexual dysfunction (men only), menstrual changes, loss of hair, dry skin, high blood pressure or ingrown toenail were recorded, but because of their non-specific association with lipodystrophy were not included in our case definition.

The identification of correlates of prevalent symptoms and lipodystrophy syndrome (dichotomized as present or absent) was achieved through initial univariate comparisons. The explanatory variables and possible confounders investigated included: participant's age (continuous); sex; ethnicity (Caucasian, Aboriginal Canadian or other); transmission risk group (homosexual sex, heterosexual sex, injection drug user, or medical procedure); most recent CD4 cell count in cells/mm³; pVL in copies/ml; whether the patient had ever had an AIDS diagnosis; and their use of ingested alternative therapies (those not generally prescribed within Western medicine, including Chinese and other herbs, coenzyme Q10, N-acetylcysteine and dinitrochlorobenzene). Finally. past, current and the duration of use of antiretroviral agents was investigated. This included having ever used, current use of, total real-time duration of exposure in months and number of concurrent therapies of three classes of agents: PI, NRTI and non-nucleoside reverse transcriptase inhibitors (NNRTI). Past use, current use and duration of exposure to each individual antiretroviral agent was also recorded. Adherence to the current regimen was considered as a continuous measure, calculated as the percentage of prescriptions written that were actually filled in 10% increments. Chi squared or Fisher's exact tests were used to identify univariate associations between dichotomous or categorical variables and Wilcoxin rank sum tests for continuous explanatory variables.

Subsequent forward stepwise logistic regression assessed the independent effects of explanatory variables on the presence of lipodystrophy. All variables significant at the 0.05 level in univariate analysis were offered for inclusion in multivariate models. In cases in which variables were found to be collinear, only one variable (that considered most clinically relevant to the outcome or the one most strongly associated with the outcome in univariate analysis) was included in each model-building run. All P values reported are two sided.

Results

From October 1998 to September 1999, 1091 annual questionnaires were returned, 1035 (95%) with complete data regarding side-effects. The majority (92%) were from men, 78% were Caucasian, 10% Canadian Aboriginal and 12% other ethnic groups. The median age of participants was 41 years [interquartile range (IQR) 36-47]. A small majority (59%) had greater than a high school education and 44% were employed. The majority (80%) reported sexual contact as the most likely mode of infection, whereas 12% reported injection drug use. Ingested alternative therapies were used by 21% of participants. The median CD4 cell count was 370 cells/mm³ (IQR 210-525), some 26% had been diagnosed as having AIDS, and 50% of patients had pVL below 400 copies/ml.

In terms of treatment modalities, 642 participants (62%) were currently using PI-inclusive regimens, and 74% of participants had ever used PI. Of the current PI users, 63% were using regimens containing one PI and 37% were using two PI in combination. Indinavir was most commonly used (57%), whereas nelfinavir, ritonavir and saquinavir were each used by approximately 20% of current users. The remaining 393 participants were restricted to NRTI or NNRTI-based regimens of whom 273 were PI naive.

Among those who had ever utilized PI the median duration of PI therapy was 18.5 months (IQR 11.3-24.9), with median duration of NRTI and NNRTI-inclusive therapies before PI use of 9.1 and 0 months, respectively. Among PI-naive participants the median duration of NRTI or NNRTI therapy was 25.4 and 0 months, respectively.

Overall, 518 or 50% of participants were identified as having probable lipodystrophy syndrome. Peripheral fat wasting or abdominal fat accumulation occurred in 36 and 33% of participants, respectively. Buffalo hump was noted by 6%. Increased triglyceride and cholesterol levels were reported by 10 and 12%, respectively.

Other symptoms were reported by 673 participants, and included high blood pressure (6%), type 2 diabetes mellitus (4%), peripheral neuropathy (29%), sexual dysfunction (23% of men), menstrual changes (28% of women), loss of hair (16%), dry skin (38%) or ingrown toenail (10%).

Individuals reporting peripheral wasting (Table 1) were older, more likely to be Caucasian, to have completed high school, to be unemployed, and to be using complementary therapies. In terms of antiretroviral regimens, lipoatrophy was associated with having ever used PI, the current use of PI, and the duration of PI and NRTI therapies (all P < 0.001).

Table 1

Education level and alternative therapy use were also associated with centripetal or breast fat accumulation (Table 2). Those who had used PI (ever or currently), and who had a greater duration of treatment with either PI or NRTI-inclusive regimens, as well as those who had never used NNRTI were more likely to experience this symptom (all P < 0.05).

Table 2

Older age and unemployment were associated with dorsocervical fat pad enlargement (Table 3) as was the use and duration of PI therapy and the duration of NRTI therapy (P ≤ 0.02).

Table 3

Similar results were noted for metabolic abnormalities (data not shown) with older age, less than a high school education, Caucasian ethnicity, PI use (ever and current) and duration of PI therapy associated with both increased triglyceride and cholesterol levels (all P < 0.05).

Factors associated with probable lipodystrophy syndrome (Table 4) included older age, being unemployed, the use of alternative therapies, past or current use of PI-containing regimens, duration of PI therapy, and duration of NRTI therapy. In a sub-analysis of each of the four specific PI available, the current use of each was significantly associated with lipodystrophy univariately (all P < 0.02, data not shown).

Table 4

The CD4 cell count, pVL and diagnosis of AIDS were not associated with any specific symptom or with the lipodystrophy syndrome overall.

Table 5 summarizes the findings of the multivariate analysis. The initial analysis (a) indicates that each additional year of age was associated with a 3% increase in risk [adjusted odds ratio (AOR) 1.03; 95% confidence interval (CI) 1.01, 1.04] of lipodystrophy. Current employment (AOR 0.56; 95% CI 0.42, 0.73) and the use of ingested alternative therapies (AOR 1.46; 95% CI 1.06, 2.01) were independently associated with a decreased and increased risk of lipodystrophy, respectively. In terms of the independent effects of treatment, those having ever used PI therapies were more than twice as likely to suffer lipodystrophy (AOR 2.63; 95% CI 1.89, 3.66) as those who had never been exposed. The duration of treatment with stavudine was associated with a 35% increase in risk of lipodystrophy for every 12 months of stavudine exposure (AOR 1.35; 95% CI 1.15, 1.58). No significant interactions were observed.

Table 5

A second model (b) was developed to investigate further the contribution of various PI therapies and their combinations to the risk of lipodystrophy among participants who had ever been exposed to PI. Additional variables offered for inclusion were: the number of PI taken concomitantly; the cumulative duration of all PI therapies with months on each drug summed; and the contribution of each of the four PI both individually and in therapy regimens including more than one PI. Older age (AOR 1.03; 95% CI 1.01, 1.05), current employment (AOR 0.67; 95% CI 0.49, 0.91) and the use of ingested alternative therapies (AOR 1.51; 95% CI 1.05, 2.19) were again significantly associated with lipodystrophy. In this sub-group, however, each year of exposure to lamivudine treatment was associated with a 32% increase in risk (AOR 1.32; 95% CI 1.13, 1.53). Again, no significant interactions were observed.

The consistent relationships between the increased risk of lipodystrophy and both unemployment and the use of complementary therapies were further explored, as these factors may be associated with disease stage despite the lack of association with AIDS diagnosis, CD4 cell count and pVL in univariate analyses. This analysis revealed no associations between either complementary therapy use or employment status and these markers of disease stage (all P > 0.05, data not shown).

Discussion

To our knowledge, this study is the first to estimate the prevalence and describe the correlates of HIV-related lipodystrophy syndrome based on patient self-report through systematic enquiry within a large population-based sample. Patient report of adverse reactions are traditionally monitored through open-ended questioning, systematic symptom checklists or spontaneous report. Systematic enquiry, while optimizing event capture may be associated with an increased risk of false positives whereas other techniques may result in underestimates ^[37-40]. This technique, in conjunction with what may be a liberal definition of lipodystrophy in comparison with that recently proposed ^[18], may overestimate the prevalence of morphological symptoms in our population. As regards metabolic abnormalities, these may be underreported here as above normal triglyceride and cholesterol levels may not be communicated to patients or the recall of such communications may be poor. Reliance on self-reported data for morphological abnormalities in the absence of clinical evaluation and validation can also be problematical. Clearly, however, clinical validation can only occur when adverse events are associated with measurable signs whose boundaries are well defined. The limits of morphological changes that indicate lipodystrophy have yet to be clarified. Moreover, the use of individual clinical examination is impractical for the broad-scale population-based monitoring of known abnormalities or in routine post-marketing surveillance for emerging side-effects in general. In terms of morphological abnormalities associated with lipodystrophy, in the most comprehensive study to date, Carr and colleagues ^[18] reported a 98% concordance between patient self-report using systematic enquiry and the findings of clinical examination.

We have estimated that lipodystrophy affects approximately half the individuals accessing ART in British Columbia. The most commonly reported symptoms were changes in fat distribution, in the form of either peripheral wasting or centripetal obesity, whereas buffalo hump and laboratory abnormalities were less commonly reported. Others have reported similar ^[41,42] or higher ^[4,43] rates of prevalent morphological changes, whereas most clinical studies ^[4,24,43] indicated a greater rate of triglyceride and cholesterol abnormalities.

Among participants naive to PI, 27% were identified as having probable lipodystrophy, illustrating that key symptoms of lipodystrophy may occur in persons never exposed to PI. This finding is consistent with that of studies that have identified symptoms in PI-naive populations or specifically related to NRTI or NNRTI therapies ^{[3,4,12,29,31,44,45]}.

Univariate analyses revealed statistically significant associations between patient and treatment characteristics and symptoms providing an important basis for comparison with previous reports. Most notably, the use of specific PI has been associated with various morphological and metabolic abnormalities ^{[5,10,21,22,27]}. In univariate analyses our findings were similar, with a significantly increased risk of lipodystrophy associated with each PI examined, a finding not confirmed in multivariate analysis. This exemplifies the importance of adjustment for confounding and collinear variables.

In multivariate analysis, older age, unemployment and the use of alternative therapies were associated with an increased risk of prevalent lipodystrophy. Although many studies have not reported on factors other than treatment modalities, older age has been identified as a risk factor by others ^[17,32]. The stability of this association after comprehensive adjustment for clinical and therapeutic factors indicates either a complex function of duration of infection, disease stage at therapy initiation and therapeutic strategies or, conversely, a true age effect. The latter is consistent with the theory that symptoms of lipodystrophy may be associated with mitochondrial abnormalities that may accrue with age ^[46].

The increased risk of lipodystrophy among those using ingested alternative treatments is intriguing. The use of such therapies has been related to disease stage among individuals with HIV/AIDS ^[47]. These analyses, however, included non-invasive or non-ingested therapies. In fact, further analysis revealed no significant association between adjunct utilization and AIDS diagnosis, CD4 cell count or pVL. A second possibility - an association between compliance with antiviral agents and the use of complementary therapies - was also ruled out in our cohort. It may be that the relationship is a function of temporality, with those suffering from side-effects implementing alternative therapies to treat such adverse responses. Additional research is currently under way to explore factors motivating the use of ingested complementary therapies. Finally, these findings may be the result of reporting bias, with those using adjuncts being more aware of morphological alterations.

The role of employment status has not previously been reported. In all likelihood, the observed relationship is a function of temporality, with those experiencing morphological abnormalities being more likely to cease work as a result. It does not appear that employment acts as a marker of disease stage or severity in this cohort, as no associations between employment status and CD4 cell count, viral load or AIDS diagnosis were found. Given the Canadian system of universal and free access to primary and specialized healthcare, in conjunction with the centralized overseeing by experts of all antiretroviral prescriptions written within the DTP, it is unlikely that this population is receiving sub-optimal or differential care. Nor does employment act as a marker of inadequate nutrition, resulting in weight loss that might be interpreted as lipodystrophy-associated wasting. No difference was found in lipodystrophy status on the basis of participants' responses to a survey question asking directly whether they had experienced weight loss as a result of inadequate nutrition (chi squared P = 0.413).

An increased risk of lipodystrophy was associated with PI use and the duration of stavudine therapy after adjustment for the duration of any therapy, the use and duration of class-specific therapies, adherence and clinical characteristics. Among those not naive to PI, the risk of lipodystrophy was associated with the duration of lamivudine therapy after adjustment for the duration of PI use, the duration of use of each specific ART and the number of concurrent therapies. The duration of stavudine and lamivudine therapy does not appear simply to reflect total therapy or NRTI duration, as a repeated analysis forcing the entry of these latter two variables resulted in similar odds ratios and 95% confidence intervals for all previously significant variables.

The results implicating NRTI use are in agreement with others who have identified an increased risk of symptoms associated with these agents ^[31,44]. Recently presented data ^[17] indicated that the duration of stavudine and lamivudine as well as indinavir and saquinavir are associated univariately with an increased risk of lipodystrophy. Carr and colleagues ^[32] showed the use of stavudine to be associated with peripheral lipoatrophy, and lamivudine duration to be associated with abdominal obesity and buffalo hump, whereas PI duration was associated with all three symptom groups. The preliminary results of a French study ^[33] also indicated a relationship between lipoatrophy and stavudine use.

Caution is warranted in interpreting these findings, as the cross-sectional nature of the study precludes drawing conclusions regarding causality. Moreover, although this study is population based, it is reliant on voluntary responses from participants who may not be representative of all persons treated in British Columbia or elsewhere.

Conclusion

These issues not withstanding, the inordinately high prevalence of both lipodystrophy syndrome and its component symptoms identified through self-report within a large observational, population-based treatment programme by individuals representing a broad spectrum of those living with HIV/AIDS and utilizing a wide variety of antiretroviral regimens is of grave concern. Most pressing are the implications regarding PI and nucleoside analogues, drugs widely prescribed both as first-line agents and throughout the course of the disease. At a time when these therapies can finally offer some hope to individuals with HIV/AIDS, the emergence of such side-effects further complicates an already difficult task. Morphological abnormalities can be disfiguring and often frightening, as they may be experienced as one of the first overt physical manifestations related to HIV disease. The subjective perception of these events by those receiving therapy may be of importance beyond that of strict clinical relevance. Significant declines in morbidity and mortality noted among those living with HIV/AIDS directly attributable to therapeutic advances may be offset if individuals decide to alter, abandon or eschew HAART because of the occurrence or fear of such events. The prevention and treatment of side-effects as well as an understanding of their impact on treatment decisions is imperative if we are to continue successfully towards a goal of long-term care through pharmacological management.