icon star paper   HIV Articles  
Back grey_arrow_rt.gif
Buffalo Hump May be Associated with High Insulin
  'Buffalo Hump Seen in HIV-Associated Lipodystrophy is Associated With Hyperinsulinemia But Not Dyslipidemia"
JAIDS Journal of Acquired Immune Deficiency Syndromes: Volume 38(2) 1 February 2005
Mallon, Patrick W. G MB, BCh, BAO, BSc*†; Wand, Handan PhD*; Law, Matthew PhD*; Miller, John‡; Cooper, David A MD, DSc*†; Carr, Andrew MD†; on behalf of the HIV Lipodystrophy Case Definition Study and Australian Lipodystrophy Prevalence Survey Investigators
From the *National Centre in HIV Epidemiology and Clinical Research, Faculty of Medicine, University of New South Wales, Sydney, Australia; †HIV, Immunology and Infectious Diseases, Clinical Services Unit, St. Vincent's Hospital, Sydney, Australia; and ‡Roche Products, Sydney, Australia.
Accumulation of dorsocervical fat, or a buffalo hump (BH), is commonly reported in adults with HIV-associated lipodystrophy (HIVLD). The pathogenesis underlying this aspect of a syndrome characterized by loss of subcutaneous fat from other body sites is poorly understood.
We aimed to identify risk factors for a BH in HIV-infected adults in cross-sectional analyses of 2 HIV-infected ambulatory populations. The first group (Australian Lipodystrophy Prevalence Survey [APS]) consisted of 1348 Australian HIV-infected adults (95% male) irrespective of changes in body composition. The second group (Lipodystrophy Case Definition [LDCD] study) comprised 417 subjects (83% male) with at least 1 reported moderate or severe feature of HIVLD. A BH was reported in 24 (2%) APS subjects and 79 (19%) LDCD study subjects. A BH was not an isolated finding. Patients with a BH had a high prevalence of other features of HIVLD, similar to lipodystrophic patients without a BH, such as facial lipoatrophy reported in 100% and 61% BH-positive subjects from the APS and LDCD study, respectively.
In both groups, those with a BH had higher fasting insulin (P <= 0.007), a higher body mass index (P <= 0.003), a higher waist/hip ratio (P <= 0.001), higher limb fat (P <= 0.003), and higher systolic blood pressure (P < 0.05). On multivariate analysis, higher serum insulin, systolic blood pressure, age, and duration of exposure to ritonavir were independently associated with a BH in the APS group. In the LDCD group, higher insulin, diastolic blood pressure, and duration of exposure to zidovudine were independently associated with a BH. There was no association between a BH and hyperlipidemia.
These data show that a BH is associated with other physical features of the lipodystrophy phenotype and suggest that hyperinsulinemia, a feature common to HIVLD, obesity, and hypercortisolism, is an important component of this phenotype, thus warranting closer monitoring of BH-positive patients for glucose intolerance and diabetes.
This is the first systematic assessment of factors associated with a BH in a large number of HIV-infected adults. The data suggest that a BH is seen in the context of a wider syndrome consisting of fat loss and fat accumulation at other body sites, which was more prevalent in the LDCD study participants, all of whom had lipodystrophy. The presence of a BH was strongly associated with hyperinsulinemia but not with the hyperlipidemia commonly seen in HIVLD, suggesting that hyperinsulinemia may be involved in the pathogenesis of a BH, whereas dyslipidemia may not be involved.
In both cohorts examined, a BH was associated with duration of treatment with antiretroviral drugs from the PI and NRTI classes, with multivariate analyses showing antiretroviral therapies to be independent predictors of a BH in both cohorts. Similar associations between NRTI and PI therapy and a BH have been previously described. Although it could be argued that this may simply reflect a subgroup of overweight or obese HIV-infected patients, the higher prevalence of BHs in those with HIVLD, the absence of BHs in those without lipodystrophy (from the LDCD control group), and the fact that BH-positive individuals also report a high prevalence of peripheral fat wasting (85% within the LDCD cohort) and maintenance of lean mass suggest that this process is, at least in part, a component of a treatment-induced lipodystrophy syndrome and not simply age- or diet-related central fat gain.
It is notable that both multivariate analyses indicate a significant independent association between insulin, blood pressure, and a BH. Associations between central fat accumulation and insulin resistance in HIV-negative populations have been well described. In addition, in a smaller cohort of HIV-uninfected men (n = 27) with a BMI similar to the LDCD cohort (BMI = 27), serum leptin concentrations correlated with systolic and diastolic blood pressure, independent of insulin concentrations. Therefore, the hyperinsulinemia and hypertension seen in those with a BH may occur independently but have the common etiologic link to central fat accumulation. Hyperinsulinemia and hypertension are independent risk factors for cardiovascular disease and, together with the increased visceral adipose tissue (VAT) seen in those with a BH, give rise to a phenotype not unlike that of the metabolic syndrome. Unlike the metabolic syndrome, however, BHs are not associated with dyslipidemia but are highly associated with duration of exposure to antiretroviral medications.
Common causes of a BH in patients without HIV infection are Cushing syndrome and Cushing disease. Although hypercortisolemia has not been demonstrated in the setting of HIV infection and antiretroviral therapy or HIVLD, altered tissue responsiveness to cortisol, thought to have a genetic basis, may alter lipid metabolism in favor of cortisol, resulting in increased central adiposity in susceptible individuals, independent of circulating cortisol concentrations. The relation between tissue cortisol sensitivity and BHs in HIV-infected cohorts has yet to be studied.
Data from this study, together with other published data, suggest that a strong common link to BHs in HIVLD, obesity, and hypercortisolism is hyperinsulinemia. The cross-sectional nature of many studies such as this one makes it difficult to determine if hyperinsulinemia is simply secondary to central fat accumulation or if insulin may have a direct role in the development of BHs. A prospective study would be required to examine such a relation adequately.
As with all cross-sectional analyses, this study was not designed to determine cause-and-effect relations, and associations demonstrated must be viewed in the context of the unmeasured bias inherent in these types of studies. Each group studied has its own limitations, such as the disparity between lipodystrophy assessments between the 2 groups, the fact that the LDCD group was not recruited as a cohort, and the incomplete imaging data in the APS cohort. In addition, there were relatively few women and no children, and no HIV-negative control group was studied. Even though previous studies have failed to reveal hypercortisolemia in subjects with HIVLD because serum cortisol concentrations were not recorded in these studies, we are unable to comment on the dynamics of cortisol metabolism in the patients with a BH. Although the lack of objective measures of BHs may be viewed as a limitation, there is no consensus objective definition for a BH, and the methods of reporting used in these studies are consistent with those of previously published studies.
Nevertheless, this study reveals consistent and highly significant associations in both groups that shed some light as to the possible causes of BH. The higher serum insulin observed in those with a BH and the higher incidence of diabetes in the LDCD BH-positive group led us to recommend closer monitoring for insulin resistance and diabetes in subjects with a BH. In addition, these results suggest that some potential therapies, such as growth hormone, which induces hyperinsulinemia, may not be an ideal therapy for patients with a BH.
HIV-infected patients receiving antiretroviral therapy can experience lipodystrophy (HIVLD) characterized by progressive selective loss of limb fat, persistent increases in central fat, maintained lean mass, insulin resistance, dyslipidemia, and hyperlactatemia. Accumulation of fat over the dorsocervical spine, or a buffalo hump (BH), is reported in 2% to 13% of HIV-infected patients, with a higher prevalence in cohorts with HIVLD Ethnicity, antiretroviral exposure, and methods of assessment all likely contribute to the variation in prevalence between studies.
Unlike the lipoatrophy of HIVLD, for which a substantial body of evidence points to causative roles of protease inhibitor (PI) and nucleoside reverse transcriptase inhibitor (NRTI) therapy, less is known about the pathogenesis underlying the gain of central fat and, in particular, the development of a BH. Hypercortisolism (seen in Cushing syndrome), a common cause of central obesity and a BH in patients without HIV infection, has not been demonstrated in the setting of HIVLD and antiretroviral therapy. Although BHs are seen in HIV-uninfected subjects, they tend to be smaller and occur less often than those seen in HIV-infected adults, suggesting that factors other than diet and age, such as HIV disease or its therapy, play a role. Objective studies have demonstrated that antiretroviral therapy leads to sustained increases in central fat and lean mass10 coupled with increases in prevalence of insulin resistance and diabetes, raising concerns that this population may experience greater long-term diabetic complications such as increased cardiovascular disease.
Few studies have investigated risk factors for a BH in HIV-infected patients, and none has done so taking into account all potentially relevant demographic, clinical, HIV disease, metabolic, and body composition parameters using appropriate multivariate analyses. We hypothesized that a BH forms part of an overall lipodystrophy syndrome, consisting of gain and loss of regional adipose tissue, induced by antiretroviral therapy. Because the prevalence of BHs differs considerably depending on the type of HIV-infected population studied, we analyzed data from 2 relevant but different populations to identify factors associated with BHs in HIV-infected adults in an attempt to gain further insights into its pathogenesis.
A BH was reported in 24 (2%) subjects in the APS group and in 79 (19%) subjects in the LDCD group, which was within the range of previously reported prevalence data.
Clinical and HIV-Related Variables
In the APS group, those with a BH were older; more likely to be female; had a higher body mass index (BMI) and systolic blood pressure; and had a longer exposure to didanosine, stavudine, lamivudine, saquinavir, and full-dose ritonavir. In the LDCD group, those with a BH weighed significantly more, had a higher BMI and blood pressure (systolic and diastolic), had more exposure to nucleoside analogues (particularly zidovudine), and had more exposure to PIs (particularly indinavir).
Metabolic and Body Composition Variables
Metabolic and body composition parameters are detailed in Table 3. In the APS group, those with a BH also had higher insulin, C-peptide, HDL cholesterol, and alanine aminotransferase (ALT) levels. In the LDCD group, those with a BH had higher serum leptin and insulin levels, more estimated insulin resistance, and a higher prevalence of diabetes mellitus. In both the APS and LDCD groups, there were no significant differences in serum total or LDL cholesterol and triglyceride concentrations between those with and without a BH. In the LDCD group, those with a BH had significantly higher regional fat mass (absolute weight and percentage of total tissue) and lean mass. The APS group showed similar results.
Reported Changes in Body Composition
Associations between a BH and other reported changes in body composition in both groups are outlined in Table 4. In the APS group, 100% and 97% of those with a BH reported facial and buttock lipoatrophy, respectively, as compared with 87% and 84%, respectively, of those without a BH. All (100%) BH subjects reported abdominal fat accumulation compared with 86% of subjects without a BH (P = 0.5).
Although subjects with a BH in the LDCD group had a higher prevalence of abdominal fat accumulation than those without a BH, 85% of subjects with a BH also experienced regional lipoatrophy, although the prevalence of lipoatrophy was significantly less than in those without a BH (95%). The most common body regions affected by fat loss in those with lipodystrophy and a BH were the legs, buttocks, and face. In those with lipodystrophy and a BH, 86% also experienced some fat accumulation at a body site other than the dorsocervical region, with the abdomen being the most common site of fat accumulation. The prevalence of abdominal fat accumulation was significantly higher in those with lipodystrophy and a BH than in those with lipodystrophy alone (75% and 45%, respectively; P < 0.0001). Most BH cases (97%) exhibited a mixed picture, having evidence of fat loss and fat accumulation at different body regions.
Multivariate Analyses
Increasing insulin, age, duration of exposure to ritonavir, and systolic blood pressure were independently associated with a BH in the APS group. In the LDCD group, increasing insulin, diastolic blood pressure, and duration of exposure to zidovudine were independently associated with a BH. The effects of serum insulin and diastolic blood pressure were negated when serum leptin was included in the LDCD analysis. The strong association between a BH and serum leptin in the LDCD group was similarly negated when the analysis was controlled for BMI or total body fat.

The Australian Lipodystrophy Prevalence Survey (APS) was a cross-sectional cohort study of 1348 HIV-infected Australian adults (3% female). Because the primary aim of this study was to report the prevalence of and risk factors for lipodystrophy in a cohort of HIV-infected Australian adults, subjects were recruited regardless of exposure to therapy or changes in body composition. Subjects were recruited consecutively from 15 sites within Australia between November 1998 and July 1999. The study collected data on demography, HIV illnesses and disease stage, antiretroviral therapy, anthropometry, and fasting lipid and glycemic parameters. All recruited subjects were assessed for the presence of lipodystrophy by physician examination, and a subset of 20% of the cohort underwent body composition studies. The overall prevalence of lipodystrophy within this study was 53%.
The HIV Lipodystrophy Case Definition (LDCD) study was a case-control study whose primary aim was to develop an objective case definition of HIVLD. HIV-infected adults were consecutively recruited at 32 sites globally between November 2000 and July 2001, with approximately equal numbers of lipodystrophic patients and nonlipodystrophic controls at each site. Lipodystrophic patients (n = 417, 15% female) had at least 1 moderate or severe feature of lipodystrophy on physical examination and patient report, based on a standardized severity scale developed for the HIV Outpatient Study (HOPS) lipodystrophy study (except for isolated abdominal obesity, which may be related to age). No lipodystrophic patient had a BH alone. Body composition was assessed in all patients and controls. Controls (n = 371) had no feature of HIVLD identified, no cases of an isolated BH, and no reported regional fat change since the diagnosis of HIV. We examined the subjects from the LDCD study to determine risk factors for a BH in this group, who, unlike the APS subjects, were all diagnosed with lipodystrophy.
All subjective and objective assessments for both studies were performed at a single time point. In addition to demographic data and a detailed antiretroviral history, subjects from both studies had the following blood tests: CD4-positive lymphocyte count and plasma HIV RNA, fasting total cholesterol, estimated low-density lipoprotein (LDL) cholesterol, high-density lipoprotein (HDL) cholesterol, triglyceride, glucose, and insulin levels performed as previously described.1,17 Adiponectin was assessed by radioimmunoassay (LINCO Research, St. Charles, MO). Leptin levels were analyzed initially by enzyme-linked immunosorbent assay (ELISA), with a lower limit of detection of 2 ng/mL (LINCO Research); samples with values <2 ng/mL were reanalyzed using LINCO radioimmunoassay. Total and regional body composition was quantified by dual-energy absorptiometry (DEXA) and single-slice abdominal (L4) computed tomography (CT).
For both studies, a BH was defined clinically as the presence of diffuse or focal fat accumulation over the dorsocervical spine. In the APS, cases of HIVLD were identified by report on the basis of physician examination only, whereas in the LDCD study, cases were identified on the basis of physical examination and patient report. Because the APS predated the LDCD study, clinical assessment in the APS involved self-reported fat loss from the face and buttock only, whereas a more detailed regional assessment was included in the LDCD study. Because no objective definition exists for diagnosing a BH, objective measures of dorsocervical fat were not performed.
Statistical Analysis
Cross-sectional analyses using simple regression were used to examine relations between BHs and measured parameters in the 2 patient groups. A probability value of 0.05 was considered significant. Multiple forward stepwise regression analysis was also performed to identify parameters independently associated with a BH. Parameters with a probability value <=0.2 on univariate analysis were considered for inclusion in multivariate models. Body composition parameters and leptin were omitted from the multiple regression models because they were not considered to be independent of a BH. No corrections were made for multiple comparisons.
  icon paper stack View Older Articles   Back to Top   www.natap.org