10.2.4 Endocrine abnormalities in HIV infection

Approximately 33 million people worldwide are living with HIV infection, and more than 2 million individuals are newly infected each year (1). Sub-Saharan Africa bears the majority of the disease burden, with 67% of all HIV cases and 75% of all HIV/AIDS related deaths occurring in this region (2). Although access to antiretroviral therapy has improved significantly over the past decade, antiretrovirals are available to only about 30% of those who need them (2). Availability of antiretroviral therapy greatly impacts the endocrine manifestations of HIV infection: individuals treated with antiretrovirals may develop peripheral fat loss, abdominal obesity, insulin resistance, and hyperlipidemia, whereas untreated individuals may develop undernutrition, wasting, and end-organ effects of opportunistic infections such as primary adrenal insufficiency secondary to adrenal destruction (Box 10.2.4.1). In all individuals with HIV infection, regardless of treatment, gonadal function, thyroid function, and bone mineral density may also be decreased, and salt and water balance may be affected (Box 10.2.4.2). The purpose of this chapter is to review the endocrine manifestations of HIV infection, including pathogenesis and treatment.

Depending on the severity of infection and the use of antiretroviral therapy, HIV infection may have significant effects on body weight, fat distribution, and, in children, on growth. These changes often accompany and may contribute to the HIV-associated metabolic and endocrine abnormalities described below. Consequently, an understanding of the potential anthropometric consequences of HIV infection is necessary to evaluate and treat the endocrine and metabolic effects of the disease.

Originally termed ‘slim disease’ due to the cachexia that accompanies untreated infection, HIV commonly causes weight loss, even in the era of antiretroviral therapy. Wasting is an AIDS-defining condition, described by the US Centers for Disease Control (CDC) as involuntary weight loss of greater than 10% of usual body weight, accompanied by diarrhoea or weakness and fever lasting 30 days in the absence of other illness. In practice, wasting is more broadly defined and includes: unintentional weight loss of more than 10% of baseline weight, even in the absence of other symptoms, a resulting body weight less than 90% of ideal body weight (BMI below 20 kg/m²), and rapid unintentional weight loss of over 5% in a 6 month period, sustained over 1 year (3). Using these definitions, HIV-associated wasting remains relatively common even in individuals treated with antiretroviral therapy, with one study estimating incidence at 33.5% (3). Although initial descriptions of AIDS wasting indicated that lean body mass decreased disproportionately to fat mass, more recent studies suggest that the relative proportions of fat and lean mass lost depend on baseline body composition, with lean mass preferentially lost in those patients with low baseline body fat (3). Unintentional weight loss predicts decreased survival in individuals with HIV infection, and AIDS wasting is associated with GH resistance and hypogonadotropic hypogonadism as described below.

With the advent of highly active antiretroviral therapy (HAART), individuals with HIV infection began to demonstrate changes in body fat distribution, including increased abdominal fat accumulation, dorsocervical fat accumulation (buffalo hump), and lipoatrophy of the face and limbs. The changes, termed ‘HIV lipod-ystrophy’ by some and HIV-associated adipose redistribution syndrome (HARS) by others, may not represent a single syndrome, and are characterized by differing degrees of fat redistribution in individual patients. For example, patients may experience peripheral fat atrophy, centripetal fat accumulation with visceral fat gain and central subcutaneous fat loss, or a combination of both (Fig. 10.2.4.1). Although estimates of prevalence vary, clinically apparent changes in fat distribution are consistently reported in more than half of adults and approximately one quarter of children receiving HAART. Moreover, a recent report showed decreased extremity fat and increased visceral fat in a cohort of HIV-infected men without clinical lipodystrophy, suggesting that subtle changes in fat distribution may be present even in individuals without apparent lipodystrophic changes (4).

The aetiology of altered fat accumulation in HIV is multifactorial. Protease inhibitors impair adipocyte differentiation through effects on sterol regulator element binding protein 1 (SREBP1) and inhibit the GLUT4 transporter, reducing glucose uptake in muscle and fat (6). In addition, nucleoside reverse transcriptase inhibitors (NRTIs), particularly the thymidine analogues stavudine and zidovudine, impair mitochondrial function, decreasing mitochondrial DNA and inhibiting mitochondrial gene transcription (6). This mitochondrial toxicity is thought to cause adipocyte apoptosis and, in conjunction with impaired adipocyte differentiation caused by protease inhibitors, may be one of the main contributors to peripheral lipoatrophy. Finally, increased inflammatory cytokines, including TNF-α and IL-6, may contribute to abnormal fat distribution in HIV-infected individuals. The buffalo hump and centripetal fat accumulation seen among many HAART-treated patients are similar to Cushing’s syndrome. Although elevated serum and urine cortisol concentrations have been documented in a small minority of patients, appropriate suppression to dexamethasone in such cases argues against true Cushing’s syndrome. In addition, serum cortisol levels in HIV-infected patients with changes in fat distribution demonstrate normal diurnal variation (7).

As described below, HIV-associated fat redistribution that involves visceral fat accumulation is associated with reduced growth hormone levels, dyslipidaemia, insulin resistance, and impaired glucose metabolism. In addition, patients with HIV and fat redistribution demonstrate elevated CRP and decreased adiponectin compared to HIV-infected patients without such changes (8). Intramyocellular lipids and rates of hepatic steatosis are also increased. All of these factors may contribute to an increased cardiometabolic risk.

Children with perinatally acquired HIV infection often have decreased height-for-age and weight-for-age, and growth failure is strongly associated with increased mortality risk. Decreased growth in children with HIV has been associated with increased viral load, and growth rates commonly increase with effective antiretroviral therapy (9). In addition to decreased height and weight, fat-free mass is often decreased in children with HIV infection and increases with antiretroviral therapy. Finally, abdominal fat accumulation and/or peripheral fat atrophy are also described in children receiving antiretroviral therapy. As in adults, the severity of these changes depends on the particular antiretroviral agents used as well as the appropriateness of paediatric dosing.

Growth hormone secretion is commonly altered in individuals with HIV infection, often in relation to changes in body composition. Patients with HIV-associated weight loss often demonstrate elevated growth hormone levels and reduced IGF-1, consistent with growth hormone resistance (10). Endogenous overnight growth hormone secretion in this population is inversely associated with albumin and fat mass, and IGF-1 levels are directly associated with calorific intake. Patients with HIV infection and abdominal fat accumulation may have decreased growth hormone production. Men with HIV and abdominal fat accumulation demonstrate reductions in both mean overnight endogenous growth hormone secretion and peak growth hormone levels, following standard growth hormone releasing hormone (GHRH)/arginine stimulation testing, as compared to both healthy volunteers and men with HIV infection and normal body composition (10). As a result, a large percentage of men with HIV infection and abdominal fat accumulation demonstrate relative growth hormone deficiency, with almost 40% showing peak growth hormone response below 7.5 μg/l to standard GHRH/arginine stimulation testing and 18% showing peak growth hormone of less than 3.3 μg/l (10). Visceral fat area is negatively associated with both mean overnight growth hormone and peak stimulated growth hormone in this population. Although fewer studies describe growth hormone secretion in women with HIV infection, increased abdominal adiposity is a negative predictor of peak stimulated growth hormone in women as well as in men (10). The aetiology of relative growth hormone deficiency in individuals with HIV infection and abdominal fat accumulation is multifactorial, and includes increased somatostatin tone, increased free fatty acids, and decreased ghrelin (10).

In patients with AIDS, lean body mass may decline significantly and disproportionately to weight. Moreover, lean body mass correlates with survival. High-dose growth hormone has been used as an anabolic therapy to increase lean body mass in patients with AIDS wasting. Studies of recombinant human growth hormone (rhGH), at supraphysiological doses of 0.1 mg/kg, or a fixed dose of 6 mg, demonstrate increases in lean body mass along with corresponding improvements in muscle function and/or exercise capacity. The relatively high doses of growth hormone in these studies are thought to be necessary to overcome the growth hormone resistance found in individuals with AIDS wasting. These supraphysiologic doses also result in numerous side effects, however, including hyperglycaemia, arthralgia, and fluid retention, limiting long-term use of growth hormone in this population.

Although not currently approved by regulatory agencies in Europe or the USA for this indication, growth hormone has been shown to reduce visceral fat in HIV-infected patients with abdominal fat accumulation. Studies using an rhGH dose of 4 mg/day have demonstrated approximately 20% reduction in visceral adipose tissue, decreased LDL, and increased HDL (10). In one of these studies, however, insulin sensitivity as measured by oral glucose tolerance test decreased significantly, with increases in both fasting and 2-hour glucose concentrations as well as fasting insulin (11). In a more recent study using a physiological rhGH dosing algorithm to maintain IGF-1 levels in the upper quartile of the normal range, visceral adipose tissue decreased significantly, but to a lesser magnitude (−8.5%) than that seen with supraphysiological dosing. This was accompanied by a small but significant decrease of 0.07 mmol/l in triglyceride and no changes in HDL or LDL (12). Importantly, fasting insulin and glucose levels did not change significantly, although there was a 1.2 mmol/l increase in 2-hour glucose concentrations (12). A preliminary study of rhGH in adolescents with abdominal fat accumulation also demonstrated significant reduction in visceral fat, without significant adverse effects on glucose or lipid (13).

An alternative strategy to increase growth hormone levels in individuals with HIV-associated fat redistribution and dyslipidemia is through the use of growth hormone releasing hormone (GHRH). In theory, GHRH may have two advantages over the use of rhGH: it maintains growth hormone pulsatility, thus more closely mimicking physiological secretion, and it preserves the negative feedback of IGF-1 on pituitary growth hormone release. A large study of synthetic GHRH (tesamorelin) 2 mg daily for six months demonstrated a 15% reduction in visceral fat, decreased triglyceride and total cholesterol, and increased HDL, without changes in fasting or 2-hour glucose or insulin measured by oral glucose tolerance tests. A six-month extension demonstrated that the beneficial effects on visceral fat and lipids were maintained over 12 total months of treatment, without effect on insulin sensitivity, but that patients who discontinued treatment had rapid reaccumulation of visceral adipose tissue. As of this writing, tesamorelin is not approved by US or European regulatory agencies.

Although growth is often impaired, prepubertal HIV-infected children typically demonstrate normal growth hormone levels (14, 15). IGF-1 levels in HIV-infected children are variably reported as decreased (15) or normal (14), a discrepancy that may result from failure to control for nutritional status. IGFBP-3 appears to be decreased (15). Low IGF-1 and IGFBP-3 in the context of normal growth hormone levels, as well as sub-normal IGF-1 and IGFBP-3 responses to IGF-1 generation testing (via administration of growth hormone) suggest a degree of growth hormone resistance in this population (15). IGF-1 and IGFBP-3 both appear to increase with effective antiretroviral therapy. As in adults, HIV-infected adolescents with visceral adiposity demonstrate decreased growth hormone response to GHRH-arginine testing (16).

Hypogonadism is common in men with HIV infection, particularly in the advanced stages of disease and among patients with HIV-associated weight loss. In an early study, 6% of asymptomatic HIV-infected men demonstrated hypogonadism compared to 50% of men with CDC-defined AIDS (17). In the era of HAART, literature is conflicting regarding both the prevalence of hypogonadism in asymptomatic HIV-infected individuals and the effect of HAART on gonadal function. One recent report described a 70% prevalence of hypogonadism with no change in testosterone levels after initiating HAART (18), whereas another, larger study demonstrated only a 6% prevalence of hypogonadism in asymptomatic individuals, with significant increases in testosterone levels after starting HAART (19). Much of this variability may be due to assay differences, as bioavailable testosterone measurement is necessary to evaluate gonadal function in this population, because sex hormone binding globulin (SHBG) is elevated by 40–50% in individuals with HIV infection.

The mechanisms underlying hypogonadism in men with HIV infection are shown in Box 10.2.4.3. Hypogonadism is most commonly secondary, and may be related to an effect of severe acute illness or undernutrition on gonadotropin production. Studies of HIV-infected men with hypogonadism demonstrate low or inappropriately normal luteinizing hormone secretion in a majority of patients. Less commonly, men with HIV infection may develop primary hypogonadism secondary to anatomic destruction of testicular tissue by opportunistic infection. An autopsy series in men with AIDS demonstrated that 25% of men with opportunistic infections had direct testicular involvement (20). Hypothalamic and/or pituitary destruction from opportunistic infection (e.g. CMV) severe enough to cause panhypopituitarism and gonadal failure have also been reported in a small number of patients. Medications may also suppress the pituitary gonadal axis. Megestrol acetate suppresses gonadotropin secretion because of its glucocorticoid-like properties, and ketoconazole may cause primary hypogonadism by inhibiting enzymes involved in testicular steroidogenesis.

Hypogonadism in men with HIV infection is associated with decreased lean body mass and diminished exercise capacity as well as increased indices of depression. Physiological testosterone replacement in this population increases lean body mass and improves patient report of overall quality of life, appearance, and wellbeing. In eugonadal patients with HIV-associated weight loss, supraphysiological testosterone administration also increases lean body mass and may improve some functional strength measures. Dehydroepiandrosterone (DHEA) supplementation also increases testosterone levels and may improve mood in men with HIV infection (21). Oxandrolone administration also increases lean body mass among eugonadal men with AIDS wasting, but may have an adverse effect on liver function. In men with HIV-associated abdominal fat accumulation, testosterone levels may also be decreased, but testosterone therapy in this population has no effect on visceral adiposity in spite of reductions in total and subcutaneous abdominal fat mass (22).

Laboratory monitoring during testosterone administration should include prostate specific androgen (PSA) measurement and monitoring of HDL, which may decrease during therapy. Among patients who achieve stable weight and/or experience improved virological control, discontinuation of testosterone and reassessment of gonadal function by morning measurement of bioavailable testosterone levels may be appropriate, as androgen concentrations may improve with nutritional and immunological recovery.

Amenorrhoea is common among women with advanced HIV disease, with the prevalence approaching 40% in women with severe HIV-associated wasting. One study demonstrated that in HIV-infected women without an AIDS-defining illness, oligomenorrhoea may be up to 10 times more prevalent, and amenorrhoea up to 7 times more prevalent, than in HIV-negative controls (23), although other cohorts have demonstrated no increase in menstrual abnormalities due to HIV. Compared to the HIV-negative population, amenorrhoea in women with HIV infection is less likely to indicate ovarian failure and may be more commonly due to the effects of severe illness on the hypothalamic-pituitary gonadal axis. HIV infection and decreased CD4⁺ T cell counts are also associated with early menopause (24).

In addition, androgen deficiency is highly prevalent among HIV-infected women, particularly among those with wasting. In a cohort of HIV-infected women studied prior to the HAART era, over 50% of women with wasting and more than one third of normal weight patients demonstrated serum free testosterone concentrations below the lower limit of normal for healthy age-matched women. In a more recent cohort, 27% of women with HIV-associated weight loss and 19% of normal weight women with HIV infection demonstrated androgen deficiency. Importantly, androgen deficiency in women with HIV infection is associated with decreased bone mineral density. The aetiology of the androgen deficiency is unknown, but may relate to intra-adrenal shunting away from androgen production toward cortisol synthesis; in women with AIDS wasting, dehydroepiandrosterone sulfate (DHEAS) levels are low compared to controls and correlate with decreased androgen levels.

Multiple studies have investigated physiological testosterone replacement in women with HIV infection. Two studies of testosterone replacement at 150 μg/day in women with HIV-associated weight loss and relative androgen deficiency demonstrated that testosterone was well tolerated and, in one cohort, improved muscle function. A more recent study of 300 μg/day testosterone via transdermal delivery over 6 months showed no effects on lean body mass, exercise capacity, or quality of life (25). In contrast, a longer study demonstrated that 300 μg/day testosterone replacement over 18 months increased lean body mass, increased bone mineral density at the hip, and improved sexual function and depression indices. In both of these studies, transdermal testosterone was safe and well tolerated. Transdermal testosterone is not approved for the treatment of androgen deficiency in HIV-infected women in Europe, and is not yet approved for any indication in the USA. A study of nandrolone in women with HIV-associated weight loss has also shown increased lean body mass without significant adverse effects. DHEA supplementation has been investigated as a means of increasing testosterone and dihydrotestosterone as well as improving mood, but the safety profile and effects of DHEA on body composition are not known (21).

The adrenal axis may be affected in advanced HIV disease, with subclinical impairment of adrenal function more common than frank adrenal insufficiency. Adrenal impairment is most often associated with anatomic destruction of the adrenal glands or anterior pituitary due to opportunistic infection, or to use of medications that suppress adrenal function. In contrast, increased cortisol concentrations may also be seen in HIV-infected patients, typically due to stress activation of the hypothalamic-pituitary-adrenal axis or intra-adrenal shunting toward cortisol synthesis.

Although adrenal insufficiency is rare in HIV-infected individuals with virological control from HAART, adrenal impairment is relatively common in patients with progressive HIV infection or AIDS. Impaired adrenal reserve may precede clinical adrenal insufficiency, as asymptomatic HIV-infected men demonstrate progressively increased plasma corticotropin (ACTH) concentrations over time in spite of normal cortisol responses to synthetic ACTH (26). Individuals with advanced HIV disease are at higher risk for frank adrenal insufficiency. In a relatively large study of 93 patients with AIDS or AIDS-related complex (ARC), 4% of patients exhibited clinical adrenal insufficiency, whereas 54% had subnormal cortisol responses to synthetic ACTH indicating marginal adrenal reserve (27). Adrenal insufficiency in HIV-infected individuals is most often related to tissue destruction of the adrenal glands by opportunistic infections (Box 10.2.4.4). Cytomegalovirus (CMV), mycobacterium avium intracellulare (MAI), and cryptococcus may damage adrenal tissue. CMV, for example, is commonly found in the adrenal glands of AIDS patients at autopsy. Glandular destruction does not typically exceed more than 50% of adrenal tissue, however, and is therefore unlikely to result in frank adrenal

insufficiency in most cases. Kaposi’s sarcoma, lymphoma, and haemorrhage may also cause adrenal damage in advanced HIV. Secondary adrenal insufficiency related to pituitary infiltration from disseminated toxoplasma gondii, cryptococcus, and CMV infection has also been reported in patients with AIDS. Idiopathic adenohypophyseal necrosis was shown in 11% of patients with AIDS or ARC in one autopsy series (28).

Assessment of adrenal function should be performed in AIDS patients with significant fatigue, inanition, hypotension, or hyponatraemia. Patients with known disseminated CMV or MAI are at increased risk. Testing should proceed with either a morning cortisol concentration or with cosyntropin administration (0.25 mg of ACTH 1–24). The cosyntropin test is misleading, however, if secondary adrenal insufficiency is of relatively acute onset, in which case testing with metyrapone or insulin tolerance testing is indicated. Of note, the insulin tolerance test should not be performed in certain circumstances, e.g. if patients are older, or have known heart disease or seizures. Long-term therapy includes mineralocorticoid and glucocorticoid replacement in primary adrenal insufficiency, and glucocorticoid administration alone in secondary adrenal insufficiency.

Medications such as megestrol acetate (Megace), ketoconazole, rifampicin, and opiates are known to affect adrenal function. Ketoconazole inhibits multiple cytochrome-P450 dependent steroidogenic enzymes, including the side chain cleavage and 11-hydroxylase enzymes, and decreases cortisol synthesis in a dose dependent manner. Fluconazole and itraconazole, now more commonly used than ketoconazole, are much less likely to cause adrenal insufficiency. Rifampicin, an antituberculous agent, increases the metabolism of cortisol and may precipitate adrenal insufficiency in the setting of known hypoadrenalism or decreased adrenal reserve. In addition, megestrol acetate, a synthetic progestational agent approved for use as an appetite stimulant in patients with AIDS wasting, decreases adrenal function through a steroid-like effect on the hypothalamic–pituitary–adrenal (HPA) axis. Chronic use of megestrol acetate may result in Cushing’s disease-like symptoms, and rapid withdrawal of megestrol acetate may cause adrenal insufficiency. Furthermore, megestrol acetate use can aggravate underlying glucose intolerance and diabetes mellitus in HIV-infected patients.

Individuals with HIV infection may demonstrate increased cortisol levels. Although this is most often due to a stress response to illness, studies have also demonstrated intra-adrenal shunting toward cortisol synthesis, potentially due to 17,20 lyase dysfunction. Cytokine modulation of the HPA axis is another potential mechanism for hypercortisolism in HIV-infected patients. Anomalous cases of clinical adrenal insufficiency with normal or elevated cortisol concentrations have also been reported in HIV-infected individuals in association with glucocorticoid resistance, due to abnormal glucocorticoid receptors (29). Finally, Cushing’s syndrome has been described in individuals with HIV infection concurrently taking fluticasone and ritonavir, secondary to ritonavir’s inhibition of CYP3A4, which prevents metabolism of fluticasone. Discontinuation of fluticasone in this setting can lead to severe adrenal insufficiency.

Abnormal thyroid function may accompany HIV infection, although subtle laboratory abnormalities are more common than overt hyper- or hypothyroidism. Thyroid binding globulin (TBG) levels increase in both adults and children with HIV infection, correlating positively with severity of illness. Several studies indicate that subclinical hypothyroidism, indicated by elevated TSH with normal T₃ and T₄ levels, is also relatively common in HIV-infected individuals, with prevalence ranging from 3–12% (30). Antithyroid peroxidase antibodies are commonly negative in these patients (30), arguing against an autoimmune aetiology. Instead, studies suggest that low CD4⁺ T cell counts and the use of HAART may be associated with development of subclinical hypothyroidism in this population (31). Although various data have implicated specific antiretroviral agents, particularly stavudine, in the development of subclinical hypothyroidism (31), other studies have refuted these findings. In addition to subclinical hypothyroidism, the prevalence of isolated low free thyroxine levels (without increased TSH) may also be increased in HIV infection (30). Paediatric data demonstrate a high prevalence (approximately 20%) of isolated low free T₄ in HIV-infected children, with free T₄ levels directly associated with CD4⁺ count (32). Development of hyperthyroidism due to Graves’ disease has also been reported in HIV-infected individuals several months following initiation of HAART, potentially as a late manifestation of immune reconstitution (30).

In patients with advanced HIV disease, overt hypothyroidism may result from infiltration of the pituitary or thyroid by opportunistic infections. Pneumocystis jiroveci may cause pneumocystis thyroiditis characterized by painful gland enlargement; depending on the extent of necrosis, thyroid function may remain normal, or patients may develop hypothyroidism, sometimes preceded by a brief period of hyperthyroidism (30). In addition, the severity of HIV disease, as well as undernutrition and weight loss, may lead to thyroid function test abnormalities similar to those in non-thyroidal illness (i.e. ‘euthyroid sick syndrome’). In advanced HIV infection, however, both T₃ and reverse T₃ (rT₃) tend to be reduced in direct association to serum albumin and surrogate measures of muscle mass, in contrast to the rT₃ elevations typically found in non-thyroidal illness.

Hypocalcaemia occurs in about 6.5% of patients with HIV infection and is more prevalent in those with advanced disease (33). Mechanisms of hypocalcaemia in HIV infection include malabsorption, vitamin D deficiency, abnormal protein binding, medication effects, hypomagnesaemia, and altered parathyroid function. The intestinal tract of patients with advanced HIV is often the target of opportunistic infections, which may result in malabsorption of calcium and vitamin D. Hypocalcaemia associated with severe illness most often results from hypoalbuminaemia and abnormal protein binding. Medications are also associated with hypocalcaemia. In this regard, foscarnet forms complexes with calcium and decreases serum concentrations, whereas pentamidine administration can result in severe renal magnesium wasting and impaired parathyroid hormone release and action.

Although vitamin D insufficiency is relatively common in the general population, there may be a greater prevalence in HIV infection. In a recent cohort of ambulatory HIV-infected patients, 37% had 25-hydroxyvitamin D levels of 50 nmol/l or below (34). In addition, 1,25-dihydroxyvitamin D levels also may be lower in patients with more severe HIV-disease and may correlate with increased mortality (35). Decreased 1,25-dihydroxyvitamin D may result from impaired 1α-hydroxylation secondary to numerous factors, including increased TNF-α or ketoconazole use. In addition, in vitro studies demonstrate that protease inhibitors (PIs) strongly inhibit both 25-hydroxylation and 1α-hydroxylation, while also mildly inhibiting 24-hydroxylation, with a net effect of decreasing 1,25-dihydroxyvitamin D production (36). Finally, as demonstrated by a recent case report of osteomalacia in an HIV-infected patient given rifabutin to treat MAI, severe vitamin D deficiency and hypocalcaemia may be caused by medications that induce the CYP450 enzymes that catabolize vitamin D (37).

As might be expected from the increased prevalence of hypocalcaemia and vitamin D deficiency, bone mineral density (BMD) is decreased in both children and adults with HIV infection (38, 39). In children, lower BMD appears to be related to increasing severity of disease as well as decreased height-for-age, weight-for-age, and IGF-1 levels (38, 40). There has not been a consistent association between antiretrovirals and BMD in pediatric studies, but cohorts may have been too small to detect an effect. In adults, decreased bone density has been associated with decreased gonadal steroids, decreased vitamin D levels, lower IGF-1, lower BMI and/or weight loss, visceral fat accumulation, the duration of HIV infection, and increasing severity of disease. In a large longitudinal analysis, decreased BMD over time was predicted by low body weight, albumin, corticosteroid use, and menopause, whereas strength training was protective (41). The effect of antiretrovirals on bone density remains controversial: although many studies show no association between HAART and BMD, a recent meta-analysis demonstrated that HAART conferred a 2.5-fold increased risk of osteopenia (42). Moreover, PI use was associated with lower BMD than use of other antiviral agents (42).

Reduced BMD in HIV-infected individuals translates into increased fracture risk. In an epidemiological study in a large healthcare system, women with HIV infection were more likely to sustain vertebral and wrist fractures than non HIV-infected patients, and men with HIV infection were more likely to sustain any fracture compared to non HIV-infected patients. The overall fracture prevalence in the combined cohort was 2.87 fractures per 100 persons in the HIV-infected group vs. 1.77 fractures per 100 persons in the uninfected controls. A study of Canadian women demonstrated a similar increase in risk, with HIV-infected women 1.7 times more likely to sustain low-impact fractures than uninfected women (43).

Numerous strategies may be useful for treatment of low bone density in HIV-infected individuals. Self-reported exercise was associated with increased BMD in a longitudinal observational study of HIV-infected individuals (41). Physiological testosterone replacement in HIV-infected women with reduced androgen levels significantly increased bone density over 18 months. Although not specifically studied, long-term low dose testosterone is likely to increase BMD in hypogonadal HIV-infected men. Bisphosphonates also appear to be a safe and effective means to increase BMD in HIV infection. Studies using alendronate 70 mg weekly in combination with calcium and vitamin D for one year demonstrated increased BMD at the lumbar spine compared to treatment with calcium and vitamin D alone (44). One of these studies also showed that alendronate increased total hip BMD (44). Interestingly, twice daily treatment with vitamin D 200 IU and calcium carbonate 500 mg alone also significantly increased total hip BMD, and tended to increase lumbar spine BMD, suggesting the benefit of optimizing calcium and vitamin D supplementation in patients with HIV (44). Zoledronate 4 mg annually also decreases bone turnover markers and increases bone density in HIV infection (39). Moreover, in a cohort receiving zoledronate 4 mg annually for two years, suppression of bone turnover markers and increased BMD persisted for two additional years of non-treatment follow-up (45). There have not yet been studies assessing the effects of exercise, androgen, or bisphosphonate on fracture risk in the HIV-infected population.

HIV-infected individuals are at increased risk for osteonecrosis, typically occurring at the femoral head (46). Though the mechanisms are unknown, case-control studies demonstrate a significantly increased risk with exposure to antiretroviral medications (46). Other identified risk factors include alcohol consumption, use of corticosteroids, a prior AIDS-defining illness, and low CD4⁺ T cell count. Although the absolute incidence of osteonecrosis remains relatively low, estimated at approximately 0.2 to 0.6 cases per 100 person-years, HIV-infected individuals have approximately 100-fold increased risk over the general population (47).

Sodium and water balance are often disturbed in advanced HIV disease. Hyponatraemia is seen in 30–50% of hospitalized patients with AIDS (48). The syndrome of inappropriate antidiuretic hormone hypersecretion (SIADH), typically secondary to a concomitant infectious process, is the most common cause of hyponatraemia in this population, with other causes including adrenal insufficiency. Certain medications such as vidarabine, miconazole, and pentamidine are associated with hyponatraemia of unknown aetiology. Hypernatraemia and nephrogenic diabetes insipidus have been reported with foscarnet therapy, and also in association with CMV infection of the hypothalamus.

Diarrhoeal opportunistic infections may lead to hypokalaemia in patients with advanced HIV-disease. In addition, tenofovir has been associated with development of renal Fanconi’s syndrome, with accompanying hypokalaemia, hypophosphataemia, and acidosis. Foscarnet therapy for CMV has also been associated with hypokalaemia, hypomagnaesemia, and hypophosphataemia. Hyperkalaemia may be present in HIV infection due to primary adrenal insufficiency as discussed above. In particular, insufficient aldosterone secretion and abnormal potassium handling have been reported in some HIV-infected individuals (49). In addition, trimethoprim is associated with hyperkalaemia due to its similar action to potassium sparing diuretics on renal tubular function.

HIV infection is associated with abnormalities in lipid metabolism, with decreased HDL and LDL occurring early in the disease (50). Triglyceride levels also increase in HIV infection in conjunction with elevated interferon-α (IFNα) levels and increased de novo hepatic lipogenesis (50). These lipid abnormalities are seen in HIV-infected children as well as in adults. In addition, individuals with HIV infection demonstrate increased lipolysis and impaired peripheral fatty acid trapping (50). The net result is increased serum concentration of free fatty acids (FFAs), both fasting and postprandial, which may contribute to insulin resistance. Increased FFA concentrations are associated with visceral adiposity and inversely associated with subcutaneous fat, such that patients with HIV infection and fat redistribution are more likely to demonstrate altered lipid metabolism.

In HIV-infected patients receiving HAART, hypertriglyceridaemia may be much more pronounced, particularly in patients with altered fat distribution (Fig. 10.2.4.2) (51). Moreover, in contrast to patients with untreated HIV, in whom LDL is often low, patients receiving HAART may demonstrate an increase in LDL to pre HIV infection levels, or even an elevated LDL (6, 50). The lipid profile in the HIV-infected patient often depends on the specific components of the antiretroviral regimen, with some protease inhibitors demonstrating particularly adverse effects on triglyceride and LDL. In one large cohort comparing antiretroviral naïve patients to those on different HAART regimens, use of PI and NRTI was associated with 27% prevalence of hypercholesterolaemia (total cholesterol ≥6.2 mmol/l) and 40% prevalence of hypertriglyceridaemia (triglyceride ≥2.3 mmol/l), compared to 8% and 15% prevalence, respectively, in antiviral naïve patients (52). A significant effect of PIs on lipids is also seen in paediatric cohorts. Ritonavir, which is used at a low dose in many HAART regimens for ‘boosting’ the serum concentrations of other PIs through its inhibition of CYP3A4, has significant effects on lipids independently of HIV or body composition changes. In one study, relatively low dose (100 mg twice daily) ritonavir administration in healthy non-HIV infected volunteers for two weeks resulted in a 26% increase in fasting triglyceride, 16% increase in LDL, and 5% decrease in HDL (53). In contrast, newer protease inhibitors such as atazanavir and darunavir appear to have fewer adverse effects on the lipid profile.

As in the general population, lifestyle changes and exercise are considered to be an important initial treatment strategy for dyslipidaemia in HIV infection, and formal exercise programs effectively reduce triglyceride and total cholesterol in HIV-infected individuals (54). If pharmacotherapy is needed, medical treatment of dyslipidaemia in HIV-infected patients follows the same principles as treatment of lipid abnormalities in the general population, but interactions between lipid-lowering agents and antiretrovirals may alter therapeutic options (54, 55). For instance, many PIs are metabolized by CYP3A4 and may cause dangerous elevations in HMG-CoA reductase concentrations. Lovastatin and simvastatin are contraindicated with many PIs, whereas pravastatin, fluvastatin, and rosuvastatin are not metabolized by CYP3A4 and may be given safely. Studies have confirmed that statins decrease LDL by 20–25% in HIV-infected patients with hyperlipidaemia (56), and a study comparing treatment in HIV-infected vs uninfected individuals showed similar LDL-lowering benefit in both groups (56). In addition, studies suggest that statins improve endothelial function in HIV-infected patients (55). There are no studies to date investigating mortality benefit of statins in the HIV-infected population.

Many HIV-infected patients demonstrate severe hypertriglyceridaemia. The Infectious Disease Society of America and Adult AIDS Clinical Trials Group guidelines for the evaluation and management of dyslipidaemia in HIV recommend treatment with gemfibrozil or fenofibrate if triglycerides exceed 5.65 mmol/l (500 mg/dl) (54). A study of fenofibrate in HIV-infected patients with triglyceride levels of 2 mmol/l or higher demonstrated that three months of treatment decreases triglyceride by 40% and non-HDL cholesterol by 17%, while increasing HDL by 15% and apolipoprotein A1 (ApoA1) by 11% (57). The safety and efficacy of gemfibrozil have also been demonstrated in the HIV-infected population, but gemfibrozil appears to be less effective in HIV-infected patients than in uninfected individuals (56), with efficacy varying according to HAART regimen. Medication interactions may contribute to reduced efficacy, as a recent study demonstrated that concurrent use of the PI lopinavir, plus ritonavir, significantly reduces serum gemfibrozil concentrations (58). Omega-3 fatty acids may also be effective in HIV-associated hypertriglyceridaemia, with one study showing a 25% reduction in triglyceride after 8 weeks of treatment with fish oil (59). Niacin also improves lipid profile in HIV-infected individuals but may impair glucose homeostasis.

In HIV-infected individuals naïve to antiretroviral treatment, glucose abnormalities are relatively rare; they are caused most often by pancreatic damage and reduced insulin secretion from pentamidine administration or hyperglycaemia from megestrol acetate. For individuals treated with HAART, however, impaired glucose tolerance is relatively common, particularly in patients with HIV-associated fat redistribution. In one cohort, 35% of patients with HIV lipodystrophy had impaired glucose tolerance, and 7% had frank type 2 diabetes that had been previously undiagnosed (Fig. 10.2.4.2) (51). A more recent study demonstrated a 14% prevalence of diabetes in a large group of HIV-infected individuals receiving HAART (with or without lipodystrophy) compared to a 5% prevalence in uninfected controls (60). In addition, HIV-infected individuals had a fourfold increased risk of developing diabetes during follow-up (60). Increased fasting insulin and glucose levels are also seen in HIV-infected children compared to uninfected controls (61). These glucose abnormalities are largely secondary to effects of protease inhibitors, many of which block GLUT4 (6), induce suppressor of cytokine signaling-1 (SOCS-1) in insulin-sensitive tissues (62), and impair adipocyte differentiation (6). In addition, certain NRTIs, particularly thymidine containing analogues, may contribute to insulin resistance by diminishing mitochondrial DNA (6). Stavudine has a particularly adverse effect on glucose homeostasis; 1 month of stavudine administration has been shown to result in insulin resistance in otherwise healthy HIV-negative volunteers.

With the high prevalence of impaired glucose tolerance and diabetes in HIV-infected individuals, evaluation of glucose homeostasis is important in this population, particularly in patients with changes in fat distribution. Notably, a recent study suggests that haemoglobin A1c values may underestimate glycaemia in HIV-infected individuals, particularly those receiving NRTIs (63). For treatment of insulin resistance, both metformin and thiazoledinediones (TZDs) have proven beneficial in the HIV-infected population (55). In a cohort of patients with both fat redistribution and hyperinsulinaemia and/or impaired glucose tolerance, metformin 500 mg twice daily for 3 months improved insulin sensitivity and decreased body weight. These benefits were sustained in a 6-month open-label extension, in which waist circumference and BMI also significantly decreased. The addition of regular exercise to metformin therapy may further improve insulin sensitivity. In patients with peripheral lipoatrophy, TZDs are an alternative potential strategy to both improve insulin resistance and increase limb fat, but studies have not shown consistent increases in extremity fat. Although one study of rosiglitazone for 3 months demonstrated a significant increase in subcutaneous leg fat and improved insulin sensitivity, a larger study showed no change in limb fat over 12 months of treatment (64). The efficacy of TZDs in HIV-infected individuals is likely to depend on the concomitant antiretroviral regimen. For instance, one recent study showed that TZD’s effects on peroxisome proliferator-activated receptor gamma (PPAR-γ) are dependent on mitochondrial function, which is impaired by many NRTIs (65). Use of TZDs may also be limited by the potential adverse cardiovascular effects of this drug class, which have not yet been studied specifically in HIV. Choice of insulin-sensitizing agent in patients with HIV should take into consideration the antiretroviral regimen, phenotype of fat distribution, and any existing comorbidities such as hyperlipidaemia. A head-to-head comparison of rosiglitazone and metformin demonstrated that, while both agents improve insulin sensitivity, metformin has a beneficial effect on lipid profile and flow mediated dilation, whereas rosiglitazone increases adiponectin (66).

Numerous studies have demonstrated increased rates of coronary heart disease, and approximately twice the rate of myocardial infarction, in HIV-infected individuals compared to the general population. To a large degree, this excess cardiovascular risk appears attributable to the effects of HAART therapy (Fig. 10.2.4.3) (52, 67), but HIV infection itself, and the attendant immune response, may also adversely affect cardiovascular health. Furthermore, body composition changes and abnormalities of glucose and lipid metabolism associated with HIV infection increase risk for heart disease. Rates of smoking also tend to be higher in HIV-infected cohorts compared to the normal population, which may also contribute to the increased cardiovascular risk described in this population (55).

The immune response to HIV infection may play a key aetiological role in the cardiovascular disease associated with HIV. Levels of C-reactive protein (CRP) are elevated in individuals with HIV infection, and higher CRP is associated with increased myocardial infarction rate. Serum concentrations of IL-6 are also increased in HIV infection, and appear to be further elevated in patients with fat redistribution (68). A recent study comparing HIV-infected men to obese HIV-negative men demonstrated that HIV-infected men had a systemic inflammatory profile similar to obese men, with comparable serum levels of CRP, adiponectin, TNF-α, and IL-6, despite of significantly lower BMI and body fat (8).

Endothelial function, arterial stiffness, and carotid intima-media thickness (cIMT) are also altered in HIV infection. Pulse wave velocity, which indicates arterial stiffness, is increased in HIV-infected individuals and further elevated in PI-treated patients, in whom arterial stiffness is comparable to hypertensive non-HIV infected individuals (69). In addition, HIV-infected individuals have impaired flow-mediated dilation (6) and increased levels of soluble intercellular adhesion molecule (sICAM) and soluble vascular cell adhesion molecule (sVCAM) (70). The cIMT is also increased in both adults (6) and children (71) with HIV infection compared to uninfected controls. This difference in cIMT persists even in ‘HIV long-term controllers’ who maintain undetectable HIV viral loads without antiviral therapy, suggesting an effect on cIMT of HIV infection itself, independently of treatment (72).

Interestingly, although antiviral therapy has been linked to increased risk of myocardial infarction, initiation of antiviral therapy decreases sICAM, sVCAM, and D-dimer (70), and improves endothelial function as measured by brachial artery flow-mediated dilation (73). These beneficial effects of virological control on endothelial function and inflammation may be one mechanism behind the recent finding that CD4⁺ count-guided interruption of antiretroviral therapy increases all-cause mortality and tends to increase cardiovascular events (74). In fact, subsequent analysis demonstrated that treatment interruption increased IL-6 and D-dimer, levels of which were strongly associated with all-cause mortality (75).

Given the adverse cardiometabolic effects of HIV infection described above, effective treatment of risk factors for cardiovascular disease is crucial for HIV-infected patients. As in the general population, prevention strategies include smoking cessation and screening and treatment for dyslipidaemia, hypertension, and disordered glucose metabolism (55). In addition, while virological control remains the top priority in HIV care, careful modification of the antiretroviral regimen for select HIV-infected patients with increased cardiometabolic risk may decrease cardiovascular risk factors (55).

HIV infection is associated with a number of endocrine abnormalities, largely dependent on the stage of disease and the medications used for therapy. Individuals with advanced HIV disease or AIDS may demonstrate high growth hormone and low IGF-1 levels, in a pattern consistent with growth hormone resistance, hypogonadotropic hypogonadism and/or menstrual irregularity, and relative or frank adrenal insufficiency with hyponatraemia and hyperkalaemia. In these patients, both opportunistic infections and the medications used to treat them often play a role by damaging endocrine organs and altering hormone synthesis. Weight loss is a predictor of mortality in this group, and clinicians should be aware of strategies to increase lean body mass, including physiological androgen replacement, and potentially, growth hormone supplementation. In contrast, HIV-infected individuals treated with HAART may present quite differently, with visceral adiposity, peripheral lipoatrophy, relative growth hormone deficiency, hyperlipidaemia, and altered glucose homeostasis. In these patients, management of hyperlipidaemia, impaired glucose metabolism, and cardiovascular risk is of primary concern. Regardless of disease stage, HIV-infected patients generally demonstrate decreased bone density, increased prevalence of subclinical thyroid disease, and increased cardiovascular risk as evidenced by increased myocardial infarction rate and increased cIMT. Children with HIV infection also demonstrate many of these abnormalities and may also present with growth failure, particularly in severe disease.

This work was partially supported by NIH Grants RO1-DK49302, RO1-DK54167, K24 DK064545-08, MO1-RR01066, and 1 UL1 RR025758-01, Harvard Clinical and Translational Science Center, from the National Center for Research Resources. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Center For Research Resources or the National Institutes of Health. The authors would also like to thank all of the participants in HIV-research and the nursing staff of the Clinical Research Centers of the Massachusetts General Hospital and the Massachusetts Institute of Technology for their dedicated patient care.