Hyperglycemia associated with protease inhibitors in HIV-1-infected patients

Authors


Severe adverse effects have been described with protease inhibitors (PIs), potent antiretroviral drugs used against advanced infection with human immunodeficiency virus (HIV). Hyperglycemia and diabetes mellitus (DM) during treatment with these drugs have been described previously, and the Food and Drug Administration released a public health advisory notice in 1997 [1,2]. The occurrence of these effects, however, seems relatively infrequent, and it is not clear whether the treatment should be maintained. We observed three patients who developed DM and hyperglycemia with early onset when treated with saquinavir, indinavir and ritonavir plus saquinavir.

From April 1996 through April 1998, 632 patients with advanced HIV infection received treatment with PI in our facility. Patients were assessed monthly during the first quarter and every 3 months subsequently. Clinical evaluation, CD4+ cell counts, viral load determination and hematologic and biochemical analysis were performed at each visit. Adherence to the treatment, efficacy and adverse effects were registered. When persistent basal hyperglycemia (glucose > 140 mg/dL) was detected, peptide C, immunoreactive insulin, glycated hemoglobin (HbAlc), islet cell antibodies (ICA), glutamic acid decarboxylase antibodies (GADA) and levels of triglycerides were assessed.

Three patients (0.47%) developed DM; all were males, aged 34, 36 and 37 years, without a family history of diabetes, and their glucose levels before treatment with PI were normal. Two of them received prophylaxis with co-trimoxazole, but no other drug was administered and diseases related to hyperglycemia were not detected. Because of a deteriorating immunologic situation, patients were treated with stavudine plus lamivudine, and a PI was added: saquinavir (patient 1), indinavir (patient 2), and indinavir and later ritonavir plus saquinavir (patient 3).

Antiretroviral treatment produced a favorable response in patients 1 and 2. Patient 3 had received treatment with indinavir for 3 months without improvement, and with elevated glucose basal levels. In this patient, the treatment was then switched to ritonavir (400 mg twice-daily) plus saquinavir (400 mg twice-daily), and 1 month later he was hospitalized with acute pancreatitis and ketoacidosis. The time of treatment before the onset of hyperglycemia ranged between 1 and 3 months. Data are shown in Table 1.

Table 1.  Details of treatment and outcome of three patients who developed hyperglycemia following dosage with protease inhibitors
PatientDrugTime a (days)Glycemia (mg/dL)C-peptide (ng/mL)HbAlc (%)Insulin (mc/mL)GADA (U/L) bTriglycerides (mg/dL)
  • a

    Time after the onset of treatment with protease inhibitor.

  • b

    b Glutamic acid decarboxylase antibodies.

  • c

    c This patient developed acute pancreatitis.

1
Saquinavir
30
310
7.10
8.6
30.4
< 0.5
216
2
Indinavir
86
402
4.40
13.9
22.4
< 0.1
245
3 c
Indinavir
Saquinavir
+ ritonavir
96
30
482
7.30
Not done
8.4
0.5
1073

Treatment with PI was discontinued in two patients (1 and 3). Hyperglycemia was controlled with diet and glibenclamide (1 month) in patient 1, and he remained well after cessation of the hypoglycemic drug. Patients 2 and 3 needed insulin. Treatment with PI was started again after withdrawal in patient 1 (after 6 months without PI) and patient 3 (after 2 months). Patient 1 was switched from saquinavir to indinavir, and normal basal glucose levels were maintained after 6 months of follow-up. In patient 3, after the acute episode of pancreatitis, indinavir was reintroduced and hyperglycemia persists, despite the administration of insulin (see Table 1).

Hyperglycemia and DM have been reported as adverse effects of treatment with PI, with a frequency ranging from 0.67% to 9.4% [3–10]. These effects occur in the first few months of treatment [3–10] as in our report, and seem not to increase with time [3,7]. Cases with all currently used PIs (saquinavir, ritonavir, indinavir and nelfinavir) have been described [3–13], and no clear demographic, clinical or associated drug risk factors for diabetes are consistently present [4]. A family history of diabetes has been reported in 17–66% of patients [3, 5, 7], and several authors have described greater age and impaired glucose tolerance prior to initiating therapy as risk factors [14–16], but this was not our experience. New-onset DM has also been described with several agents commonly used to manage HIV disease, such as pentamidine, megestrol, ddI and prednisone [17–19], but not with co-trimoxazole, which was the only drug that our patient had received.

The pathogenesis of this effect is unknown. PIs block aspartate protease, which is required to produce new virions. Human processing of proinsulin to cleave C-peptide requires serine endopeptidases (PC1/PC3 and PC2); these enzymes are not affected by protease inhibitors [20, 21]. In patients who are functionally deficient in these endopeptidases, however, the mammalian homolog of yeast aspartic protease (YAP3p) may play a more critical role in processing proinsulin [21–23] and could be inhibited by the PI. The lack of ketosis in most of the cases reported suggested a pathogenesis similar to that of adult-onset, non-insulin-dependent diabetes and, in this situation, high levels of C-peptide and insulin and impaired oral glucose tolerance have been reported [6,7], revealing significant peripheral insulin resistance, as we observed in our patients ( Table 1).

Recently, partial lipodystrophies have also been reported after treatment with PIs and are associated with DM or insulin resistance, with higher triglyceride and cholesterol levels [6, 8, 24], as we observed in our patients. Peripheral fat wasting (fat loss from the face, limbs, and upper trunk), accumulation of intra-abdominal fat, and fat in the breasts of women and over the cervical vertebrae (‘buffalo hump’) have been described [25–27]. Not all patients develop all features, but the data suggest overall fat loss and a progressive process related to the use of PIs. An interesting hypothesis has been proposed by Carr et al [28]: HIV-1 protease inhibitors have considerable homology with the low-density lipoprotein-receptor-related protein (LRP) and with a C-terminal region of the cytoplasmic retinoic-acid binding protein type 1 (CRABP-1), which presents retinoic acid to cytochrome P450 3A. This cytochrome catalyzes the conversion of retinoic acid to cis-9-retinoic-acid, which binds to the retinoid X receptor (RXR). In adipocyte nuclei, RXR functions as a heterodimer with peroxisome-proliferator-activated receptor type gamma (PPAR-γ). Ligand binding to RXR or PPAR-γ inhibits adipocyte apoptosis and upregulates adipocyte differentiation and proliferation with preferential expression of peripheral versus central fat. The impaired CRABP1-mediated cis-9-retinoic acid stimulation of RXR-PPAR-γ induced by the PI then results in reduced differentiation and increased apoptosis of peripheral adipocytes, with impaired fat storage and lipid release. The syndrome's severity is in part proportional to a PI's capacity to inhibit cytochrome P450 3A, which is also inhibited by the PI, especially ritonavir. In addition, the inhibition of the LRP that is co-expressed on capillary endothelium with lipoprotein lipase (LPL) and cleaves fatty acids from circulating triglycerides leads to secondary hyperlipidemia, central obesity and breast fat deposition in the presence of estrogen. Insulin resistance could be produced as a consequence of hyperlipidemia, by interference with post-receptor insulin signaling, competition between glucose and lipid oxidation pathways in skeletal muscles, or inhibition of glycogen synthase [29,30]. An alternative hypothesis proposed by Martinez and Gatell [31] speculates that PI may also inhibit the insulin-degrading enzymes (glutathione–insulin transhydrogenase or insulysin). The hyperinsulinemia would stimulate hepatic lipogenesis, and liver-derived triglycerides would be transported as very low-density lipoproteins for deposition in adipose tissue. Insulin-induced stimulation of lipoprotein lipase and inhibition of hormone-sensitive lipase would result in a net increase of fat. As long as hyperinsulinemia is maintained, it would finally lead to insulin resistance, and the net effect at this stage would be the inhibition of lipogenesis and the stimulation of lipolysis, resulting in an inappropriate increase of plasma free fatty acids. The increased availability of free fatty acids would further increase hepatic triglyceride synthesis and enhance hepatic glucose output, which would induce compensatory insulin hypersecretion to maintain glucose uptake homoeostasis. In some predisposed individuals, overstimulation of β-cells might lead to its failure and give rise to the emergence of type 2 DM.

Therapeutic management of hyperglycemia seems difficult, and no clear guides are provided. Good control of glucose levels has been obtained with diet alone and, more frequently, adding insulin and hypoglycemic agents, without discontinuing treatment [3,4,6, 9,32], but in other reports control was difficult if treatment was maintained [5,7]. New PPAR-γ agonists such as troglitazone have been recently used [33], but it should be noted that the PI effect on CRABP-1 and LRP is not permanent, and hyperglycemia is therefore reversible if PI treatment is stopped [5,6], as occurred in patient 1 in our report. Surprisingly, in this patient, the reintroduction of a different PI did not produce hyperglycemia, and the patient maintained normal glucose levels with diet. Although metabolic changes seem to be a class-specific characteristic of PIs, their CRABP-1 inhibitory effect could be modulated for the different degrees of cytochrome P450 inhibition. The use of another PI does not appear to be an effective strategy [34], but some authors have reported benefit in switching to nelfinavir from another PI [35,36].

Patients with advanced HIV disease clearly benefit from protease inhibitors in terms of disease progression, survival and reversal of opportunistic infections, but survival advantage in early HIV disease is still unproven. An alternative treatment is now offered with the use of non-nucleoside reverse transcriptase inhibitors (NNRTIs). Reversal of metabolic abnormalities in patients using PIs have been reported after switching to nevirapine [37] or efavirenz [38]

In conclusion, treatment with PIs is associated with important lipidemic alterations and impaired glucose homeostasis. Although DM is relatively infrequent, clinicians need to be aware of this potential complication and consider the risks and benefits of withdrawal of PI therapy if hyperglycemia occurs.

Ancillary