Fat City: Understanding HIV Lipodystrophy By Charles Flexner, M.D.

Changes in body habitus following highly active antiretroviral therapy (HAART) were first reported publicly only a few months ago at the 1998 Chicago Retrovirus Conference. This week in Geneva, the HIV lipodystrophy syndrome (LDS), which consists of unusual distribution of body fat (buffalo hump, increasing abdominal girth, breast enlargement, thinning of face and arms) was one of the hottest topics of conversation. In several poster presentations and an oral session early in the conference, investigators shared what they know and tried to come to terms with this new side effect of successful treatment of HIV infection.

You can't understand a syndrome until you can define it, and, so far, no one has been able to develop an acceptable case definition for HIV lipodystrophy. This is especially confusing since some investigators lump together the serum lipid and glucose abnormalities often seen with HAART with these changes in body habitus; others separate the physical findings from the laboratory abnormalities. While it is true that most patients with abnormal fat redistribution also have changes in their lipid profile, a causal relationship between the two is far from established.

An additional problem is that most of the descriptions of this syndrome consist of uncontrolled anecdotes. Case series often revolve around asking patients taking protease inhibitors if they have noted any changes in body fat. Given the heightened sensitivity about this syndrome, there is some concern that trivial changes in appearance, or any weight gain while on HAART are being termed "lipodystrophy." It is no surprise, then, that estimates of the prevalence of this condition ranged from 5 to 60% of those taking protease inhibitors.

Epidemiology: Several case series in Geneva tried to estimate the prevalence of LDS. In a survey of HIV-infected women in Providence, RI, 21 of 116 (18%) expressed some concern about noted changes in body habitus while taking HAART [Abstract 12373]. Seven of 29 indinavir (IDV) recipients (24%) in a report from Hong Kong reported facial thinning without overall weight loss while taking this protease inhibitor [Abstract 12391]. Whether lipodystrophy is more common with some drugs than with others and whether switching regimens can lead to reversal of this syndrome are unsettled issues. David Cooper reported in an oral session on Tuesday that LDS was more prevalent in his clinic in patients on ritonavir-containing regimens [Oral Presentation 176]. However, this has not been a uniform observation, and Cooper's conclusions suffer from the same potential biases as other case series of LDS presented in Geneva. One study from Australia [Abstract 12287] reported that switching LDS patients from indinavir or ritonavir/saquinavir to nelfinavir led to improved triglycerides but no change in cholesterol. In addition, 11 of 21 subjects reported some improvement in body habitus. However, this was a small, unblinded, and potentially biased report.

Endocrinology: Changes in serum lipids and blood glucose, which may be mechanistically related to LDS, were also a hot topic in Geneva, with studies describing both epidemiology and potential mechanism of these toxicities. Hyperglycemia and insulin-dependent diabetes have been reported anecdotally in patients taking HIV protease inhibitors, but the causal role of the drug in these events has not been clear. Several studies examined the association between PI use and elevated blood glucose in large cohorts. A German study compared 336 patients taking PIs to 447 HIV-infected patients not taking PIs [Abstract 12377]. Glucose intolerance, defined as a random glucose between 140 and 180 mg/dL, was seen in 17% of those taking PIs, 8% of those taking RTIs, and 4% of those on no therapy (p <.001). However, frank diabetes, defined as a random glucose of >180 mg/dL, was just as common with RTI therapy or no therapy as with PI therapy (prevalence 3% vs. 3% vs. 4%, respectively). In support of the low prevalence of diabetes in PI-users was a retrospective chart review of 232 patients with more than one month of PI treatment in which no cases of hyperglycemia were found [Abstract 12381]. Although increased insulin secretion and glucose intolerance have been reported with all of the PIs, the balance of studies now suggests that the incidence of true diabetes caused by these drugs is likely to be very low, i.e., less than 1-2% of patients.

Hyperlipidemia, in contrast to diabetes, appears to be more common than previously reported. A chart review of 158 PI-users at Walter Reed Army Medical Center, for example, found a mean cholesterol elevation of 28.9 mg/dL, as compared to 3.5 in RTI users and 0.1 in untreated HIV infected patients [Abstract 12269]. Although most of these patients were taking indinavir, there appeared to be a similar mean increase in cholesterol regardless of which protease inhibitor was used. Only 5.7% of these patients had a cholesterol increase of >100 mg/dL. Overall, some increase in serum cholesterol and/or triglycerides has been seen in 35-60% of PI users [Abstract 12381] although most of this is of a mild nature.

Two controversies related to PI-associated hyperlipidemias are whether the drug is a direct or indirect factor and whether the changes seen are clinically consequential. Keith Henry and colleagues from the University of Minnesota reported three cases of acute coronary artery disease in young PI- users, prompting comments about the need for "HEART therapy." One of these patients was 26 years old but had only taken a PI for a few weeks, raising questions about the relationship between drug-induced hyperlipidemia and myocardial infarction. Henry reported that 64/135 patients (47%) he was treating with PIs had elevations in their serum lipid profiles high enough to meet intervention guidelines of the National Cholesterol Education Program [Abstract 12319]. These elevations were seen in 35/53 patients (66%) on ritonavir + saquinavir, 20/62 (32%) on indinavir, and 7/18 (39%) on nelfinavir.

There have been few reports thus far detailing whether or not the hypertriglyceridemia of HAART is associated with complications like pancreatitis. Henry reported four cases of acute pancreatitis in patients with PI-associated hypertriglyceridemia, but two of these had a history of ethanol abuse, and one was also taking d4T.

What Causes Lipodystrophy? A fascinating question is whether the lipodystrophy syndrome and the associated hyperlipidemias of HAART are a direct pharmacologic effect of the drugs involved or represent a reversal syndrome in patients who have experienced overt and covert wasting, often for years prior to the institution of suppressive antiretroviral therapy. A number of studies have now looked at specific physiologic and endocrine abnormalities that might provide a mechanism for LDS. So far, most reports show no consistent changes in any parameter that might indicate a single unifying mechanism for this syndrome. Although it appears that a number of patients on HAART have peripheral insulin resistance, as manifested by glucose intolerance [Abstract 12377], it is unclear how this could explain the characteristic fat redistribution. Serum cortisol is not consistently altered in LDS and does not appear to explain this syndrome [Abstract 12299].

A recent hypothesis paper in the Lancet (June, 1998) suggested that HIV protease inhibitors might interfere with a family of proteins that regulate lipid metabolism. Although a nice explanation was given for how blocking these proteins, called CRABP-1 and LRP, might lead to LDS, this was sheer speculation, not yet backed up by data.

Don Kotler provided a more balanced view on LDS in an oral presentation on Wednesday [Abstract 32173]. Kotler pointed out that some cases of LDS have occurred in patients who have never taken protease inhibitors but were treated with nucleoside RTIs or NNRTIs. He also pointed out that LDS is not unique to HIV; the kind of fat redistribution associated with HAART is seen in HIV- negative patients with something called "Syndrome X." This line of reasoning seems to point to LDS as a secondary consequence of suppressing chronic HIV infection, rather than as a primary side effect of drug therapy.

What to Tell Patients: Unfortunately, the data presented in Geneva provided no clear direction for counseling patients about risks and management strategies for LDS. Right now, it seems best to separate the blood abnormalities from the fat redistribution and to deal with each on its own terms. Dangerously high triglycerides or cholesterol should probably be treated as they would be in HIV-negative subjects. The same principle applies to blood glucose. In the absence of better studies, changes in body habitus should be viewed as largely cosmetic and the patient reassured that this is an unfortunate side effect of knocking out HIV. A change in antiretroviral regimen specifically to reverse LDS should probably be avoided unless the patient finds the changes in body fat intolerable and more conservative interventions fail to work.