Basics
HAART
AIDS
Special Topics
Drugs


Other Languages
German

HIV Medicine 2005
Portuguese
Russian
Spanisch

Copyright Removal
Mailing List
Privacy
About

Ulrich A. Walker and Grace A. McComsey

Introduction

Two years after the introduction of protease inhibitors into the armamentarium of antiviral therapy, reports of HIV-infected individuals experiencing clinically relevant changes in body metabolism began to surface. These "metabolic" symptoms were initially summarized under the term "lipodystrophy" (Carr 1998). Today, ten years after the introduction of highly active antiretroviral therapy (HAART), this lipodystrophy syndrome is increasingly understood as the result of overlapping, but distinct effects of the different drug components within the HAART antiretroviral cocktail. The main pathogenetic mechanism through which nucleoside analogs are thought to contribute to the metabolic changes and organ toxicities is mitochondrial toxicity (Brinkman 1999).

NRTIs are prodrugs (Kakuda 2000) because they require activation in the cell through phosphorylation before they are able to inhibit their target, e.g. HIV reverse transcriptase. In addition to impairing the HIV replication machinery, the NRTI-triphosphates also inhibit a human polymerase called "gamma-polymerase", which is responsible for the replication of mitochondrial DNA (mtDNA). Thus, the inhibition of gamma-polymerase by NRTIs leads to a decline (depletion) in mtDNA, a small circular molecule normally present in multiple copies in each mitochondrion and in hundreds of copies in most human cells (Lewis 2003). The only biological task of mtDNA is to encode for enzyme subunits of the respiratory chain, which is located in the inner mitochondrial membrane. Therefore, by causing mtDNA-depletion, NRTIs also lead to a defect in respiratory chain function.

An intact respiratory chain is the prerequisite for numerous metabolic pathways. The main task of the respiratory chain is to oxidatively synthesize ATP, our chemical currency of energy. In addition, the respiratory chain consumes NADH and FADH as end products of fatty acid oxidation. This fact explains the micro- or macrovesicular accumulation of intracellular triglycerides, which often accompanies mitochondrial toxicity. Last but not least, a normal respiratory function is also essential for the synthesis of DNA, because the de novo synthesis of pyrimidine nucleosides depends on an enzyme located in the inner mitochondrial membrane. This enzyme is called dihydroorotate dehydrogenase (DHODH) (Löffler 1997). The clinical implications of this fact are detailed below.

The onset of mitochondrial toxicity follows certain principles (Walker 2002b):

1. Mitochondrial toxicity is concentration dependent. High NRTI-concentrations cause a more pronounced mtDNA-depletion compared to low concentrations. The clinical dosing of some nucleoside analogs is close to the limit of tolerance with respect to mitochondrial toxicity.

2. The onset of mitochondrial toxicity requires prolonged time. Changes in mitochondrial metabolism are observed only if the amount of mtDNA-depletion exceeds a certain threshold, an effect observed solely with prolonged NRTI-exposure. As a consequence of this effect, the onset of mitochondrial toxicity is typically not observed in the first few months of HAART. Furthermore, long-term NRTI exposure may also lead to mitochondrial effects despite relatively low NRTI concentrations.

3. There are significant differences in the relative potencies of nucleoside and nucleotide analogs in their ability to interact with gamma-polymerase. The hierarchy of gamma-polymerase inhibition for the active NRTI metabolites has been determined as follows: zalcitabine (HIVID^TM) > didanosine (Videx^TM) > stavudine (Zerit^TM) > lamivudine (Epivir^TM) ≥ abacavir (Ziagen^™) ≥ tenofovir (Viread^™) ≥ emtricitabine (Emtriva^™).

4. Zidovudine may be peculiar because its active triphosphate is only a weak inhibitor of gamma-polymerase. However, another mechanism can explain how zidovudine could cause mtDNA-depletion independent from gamma-polymerase inhibition. Zidovudine is an inhibitor of mitochondrial thymidine kinase type 2 (TK2), and, as such, interferes with the synthesis of natural pyrimidine nucleotides, thus potentially impairing the formation of mtDNA (McKee 2004). Indeed, inborn defects of TK2 are known to cause mtDNA-depletion in muscle tissue of humans (Saada 2001). It has also been demonstrated recently that zidovudine can be non-enzymatically converted into stavudine within the body, at least within some cells (Becher 2003, Bonora 2004).

5. Mitochondrial toxicity is tissue specific. Tissue specificity is explained by the fact that the uptake of the NRTI-prodrugs into cells and their mitochondria, as well as activation by phosphorylation may be different among individual cell types.

6. There may be additive or synergistic mitochondrial toxicities if two or more NRTIs are used in combination.

7. Some data suggest that mitochondrial transcription may also be impaired without mtDNA-alterations (Mallon 2005, Galluzzi 2005). However, the mechanism and clinical significance of this observation are not yet understood.

MtDNA-depletion may manifest clinically in one or several main target tissues (Fig. 1).

In the liver mitochondrial toxicity is associated with increased lipid deposits, resulting in micro or macrovesicular steatosis. Steatosis may be accompanied by elevated liver transaminases. Such steatohepatitis may progress to liver failure and lactic acidosis, a potentially fatal, but fortunately rare complication. Although steatohepatitis and lactic acidosis were already described in the early 90s in patients receiving didanosine monotherapy (Lambert 1990), mitochondrial liver toxicity is now observed under treatment with all NRTIs that have a relatively strong potential to inhibit gamma-polymerase, especially with the so called "D-drugs" didanosine (Videx^TM), stavudine (Zerit^TM) and zalcitabine (HIVID^TM). However, liver complications were also described with zidovudine (Retrovir^TM). It has been demonstrated in the hepatic tissue of HIV patients, that each of the D-drugs leads to a time dependent mtDNA-depletion. On electron microscopy, morphologically abnormal mitochondria were observed.

A typical complication of mitochondrial toxicity is an elevation in serum lactate. Such hyperlactatemia was more frequently described with prolonged stavudine treatment (Saint-Marc 1999, Carr 2000), especially when combined with didanosine. The toxicity of didanosine is also increased through the interactions with ribavirin and hydroxyurea. The significance of asymptomatic hyperlactatemia is unclear. When elevated lactate levels are associated with symptoms, these are often non-specific such as nausea, right upper quadrant abdominal tenderness or myalgias. In the majority of cases, levels of bicarbonate and the anion gap (Na⁺ - [HCO3^- + Cl^-]) are normal, although liver transaminases are mildly increased in the majority of cases (Lonergan 2000a). Therefore, the diagnosis relies on the logistically more cumbersome direct determination of serum lactate. In order to avoid artifacts, venous blood must be drawn without the use of a tourniquet from resting patients. The blood needs to be collected in fluoride tubes and transported to the laboratory on ice for immediate analysis. Non-mitochondrial causes must also be considered in the differential diagnosis of lactic acidosis (Table 1) and underlying organ toxicities should be looked for.

Prolonged treatment with D-drugs may also frequently lead to a predominantly symmetrical, sensory and distal polyneuropathy of the lower extremities (Simpson 1995, Moyle 1998). An elevated serum lactate level may help to distinguish this axonal neuropathy from its HIV-associated phenocopy, although in most cases the lactate level is normal. The differential diagnosis may also take into account the fact that the mitochondrial polyneuropathy mostly occurs weeks or months after initiation of D-drugs. In contrast, the HIV-associated polyneuropathy generally does not worsen and may indeed improve with prolonged antiretroviral treatment.

In its more narrow sense, the term "lipodystrophy" denotes a change in the distribution of body fat. Some subjects affected with lipodystrophy may experience abnormal fat accumulation in certain body areas (most commonly abdomen or dorsocervical region), whereas others may develop fat wasting (Bichat's fat pad in the cheeks, temporal fat, or subcutaneous fat of the extremities). Both fat accumulation and fat loss may at times occur simultaneously in the same individuals. Fat wasting (also called lipoatrophy) is partially reversible and generally observed not earlier than one year after the initiation of antiretroviral therapy. In the affected subcutaneous tissue, ultrastructural abnormalities of mitochondria and reduced mtDNA levels have been identified, in particular in subjects treated with stavudine (Walker 2002a). In vitro and in vivo analyses of fat cells have also demonstrated diminished intracellular lipids, reduced expression of adipogenic transcription factors (PPAR-gamma and SREBP-1), and increased apoptotic indices. NRTI treatment may also impair some endocrine functions of adipocytes. For example, they may impair the secretion of adiponectin and through this mechanism may promote insulin resistance. Stavudine has been identified as a particular risk factor, but other NRTIs such as zidovudine may also contribute. When stavudine is replaced by another NRTI, mtDNA-levels and apoptotic indices improve (McComsey 2005a) along with an objectively measurable, albeit small increase of subcutaneous adipose tissue (McComsey 2004a). In contrast, switching away from protease inhibitors did not ameliorate lipoatrophy or adipocyte apoptosis. Taken together, the available data are consistent with a predominant effect of mitochondrial toxicity in the pathogenesis of lipoatrophy.

Some studies have suggested an effect of NRTIs on the mtDNA levels in blood (Coté 2003, Miro 2003). The functional consequence of such mitochondrial toxicity on lymphocytes is still unknown. In this context, it is important to note that a delayed loss of CD4+ and CD8+ T-lymphocytes was observed, when didanosine plasma levels were increased by comedication with tenofovir or by low body weight (Negredo 2004). Recent in vitro investigations with exposure of mitotically stimulated T-lymphocytes to slightly supratherapeutic concentrations of didanosine also detected a substantial mtDNA-depletion with a subsequent late onset decline of lymphocyte proliferation and increased apoptosis (Setzer 2005a, Setzer 2005b). Thus, mitochondrial toxicity is the most likely explanation for the late onset decline of lymphocytes observed with didanosine. The data suggest that the mitochondrial toxicity of NRTIs on lymphocytes has immunosuppressive properties.

Asymptomatic elevations in serum lipase are not uncommon under HAART, but of no value in predicting the onset of pancreatitis (Maxson 1992). The overall frequency of pancreatitis has been calculated as 0.8 cases/ 100 years of NRTI-containing HAART. Clinical pancreatitis is associated with the use of didanosine in particular. Didanosine reexposure may trigger a relapse and should be avoided. A mitochondrial mechanism has been cited to explain the onset of pancreatitis, but this assumption remains unproven.

New studies have also raised the question, whether or not zidovudine can be used safely to reduce the risk of HIV vertical transmission. Pregnant monkeys were treated with zidovudine plus lamivudine for a period of 10 weeks prior to delivery and zidovudine was found to be incorporated into mtDNA. The NRTI combination was also associated with mtDNA-depletion in skeletal muscle, heart and brain (Gerschenson 2004). Perinatally acquired lesions were shown to persist for months after cessation of NRTI exposure in some models (Walker 2004a).

Mitochondrial symptoms were found at increased frequency in infants perinatally exposed to NRTIs (Blanche 1999). Hyperlactatemia is not infrequently observed and may persist for several months after delivery (Noguera 2003). Very low mtDNA levels were measured in the placenta, as well as in the peripheral cord blood of neonates (Shiramizu 2003, Divi 2005). Other clinical trials in contrast did not detect an increased perinatal risk in association with perinatal zidovudine prophylaxis although key parameters of mitochondrial dysfunction were not assessed. Long-term follow-up data are urgently needed. The present information however does not justify deviating from the currently recommended strategy to use zidovudine to prevent vertical HIV transmission.

The existence of mitochondrial damage to the kidney is controversial. Supratherapeutic doses of the nucleotide analog reverse transcriptase inhibitor tenofovir (Viread^TM) induced a Fanconi syndrome with tubular phosphate loss and consecutive osteomalacia in animals (Tenofovir review team 2001). Tenofovir is taken up into the renal tubules by means of a special anion transporter and it cannot be ruled out that an excessively high intracellular drug concentration may lead to a clinically relevant gamma-polymerase inhibition and mtDNA depletion, despite the fact that tenofovir has only a low potency to impair the replication of mtDNA. Decreased mtDNA levels have recently been found in renal biopsies from patients exposed to tenofovir plus didanosine, a NRTI combination that for several reasons is no longer recommended (Côté 2005). Renal mtDNA levels from individuals treated with tenofovir did not differ from those who remained untreated. However, the glomerular and tubular function was not assessed in this study. Furthermore, information about the indication for renal biopsy was not provided, and a didanosine-only control group was not included. It should be noted that neither the trials leading to the approval of tenofovir, nor the subsequent field data were able to prove the mitochondrial toxicity of tenofovir in the renal tubules. However, most trials only measured creatinine clearance and serum phosphate (Izzedine 2005), even though a compromise in renal function is not expected in Fanconi's syndrome and normal serum phosphate levels may be preserved by increased phosphate mobilization from bone, thus masking increased renal loss. More sensitive methods have recently revealed a diminished renal phosphate resorption and an elevated alkaline phosphatase in patients treated with tenofovir (Kinai 2005). Cases of phosphate diabetes were also reported under treatment with other nucleoside analogues.

There is currently no method to reliably predict the mitochondrial risk of an individual patient. Routine screening of asymptomatic NRTI-treated subjects with lactate levels is not warranted, since elevated lactate levels in asymptomatic subjects are not predictive of clinical mitochondrial toxicity (McComsey 2004b). In contrast, there should be a low threshold to promptly check lactate levels in subjects who experience symptoms consistent with mitochondrial toxicity. The determination of mtDNA-levels in PBMCs is subject to systematic errors and high variability; the method is not internationally standardized. Quantifying mtDNA within affected tissues is likely to be more sensitive; however this form of monitoring is invasive and not prospectively evaluated with regard to clinical endpoints.

Once symptoms are established, histological examination of a biopsy may contribute to the correct diagnosis. The following findings in tissue biopsies point towards a mitochondrial etiology: ultrastructural abnormalities of mitochondria, diminished histochemical activities of cytochrome c-oxidase, the detection of intracellular and more specifically microvesicular steatosis, and the so-called ragged-red fibers.

An impairment of mitochondrial metabolism may also result from ibuprofen, valproic acid and acetyl salicylic acid as these substances impair the mitochondrial utilization of fatty acids. Numerous cases have been described, in which a life-threatening lactic acidosis was triggered by valproic acid, both in HIV-infected patients and in patients with inherited mutations of mtDNA. Acetyl salicylic acid may damage mitochondria and such damage to liver organelles may result in Reye's syndrome.

Amiodarone and tamoxifen also inhibit the mitochondrial synthesis of ATP. Acetaminophen and other drugs impair the antioxidative defense (glutathione) of mitochondria, allowing for their free radical mediated damage. Aminoglycoside antibiotics and chloramphenicol not only inhibit the protein synthesis of bacteria, but under certain circumstances may also impair the peptide transcription of mitochondria as our bacteria-like endosymbionts. Adefovir and cidofovir are also inhibitors of gamma-polymerase. Alcohol as a mitochondrial toxin is advised against.

The most important clinical intervention is probably the discontinuation of the NRTI(s) responsible for mitochondrial toxicity. Several studies have demonstrated that switching stavudine (Zerit^TM) to a less toxic alternative led to an objective and progressive improvement in lipoatrophy (McComsey 2004, Madruga 2005, Martin 2004, Moyle 2005, Milinkovic 2005, Tebas 2005). In contrast, a switch from protease inhibitors to NNRTIs was not associated with an improvement of lipoatrophy. These findings stress the importance of mitochondrial toxicity in the pathogenesis of these fat abnormalities.

The supplementation of uridine is a new, but promising strategy. As outlined above, any respiratory chain impairment also results in the inhibition of DHODH, an essential enzyme for the synthesis of uridine and its derived pyrimidines (Fig 2). This decrease in intracellular pyrimidine pools leads to a relative excess of the exogenous pyrimidine nucleoside analogs, with which they compete at gamma-polymerase. A vicious circle is closed and contributes to mtDNA-depletion. By supplementing uridine either prophylactically or therapeutically, this vicious circle may be interrupted, resulting in increased mtDNA-levels. Indeed, uridine abolished in hepatocytes all the effects of mtDNA-depletion and normalized lactate production, cell proliferation, the rate of cell death and intracellular steatosis. (Walker 2003). In contrast, vitamin cocktails were not beneficial in this model. New data indicate that uridine is able to also prevent the loss of mtDNA, lipids and mitochondrial functions in adipocytes exposed to stavudine (Walker 2004a). Adipocyte apoptosis was also prevented.

Figure 2: Suggested mechanism of Mitocnol (NucleomaxX^TM) in the prevention and treatment of mitochondrial toxicity.

The oral substitution of uridine as a pyrimidine precursor is well tolerated by humans, even at high doses (van Groeningen 1986, Kelsen 1997). A food supplement called Mitocnol was shown to have a more than 8-fold uridine bioavailability over conventional uridine (Venhoff 2005). After positive experiences in individual cases of mitochondrial toxicity (Walker 2004d), data from clinical trials are now surfacing. Mitocnol was studied in a randomized placebo-controlled double-blind trial in lipoatrophic subjects under continued therapy with stavudine or zidovudine where it has shown improvement in objectively measured subcutaneous fat (Sutinen 2005). The effect of Mitocnol on subcutaneous fat gain was more rapid and quantitatively more pronounced in comparison with switch strategies (e.g. the replacement of stavudine and zidovudine by antivirals with a reduced potential of mitochondrial toxicity (Fig 3).

A second trial has also suggested Mitocnol to be efficacious with regard to patient and physician assessed lipoatrophy scores, although fat and PBMC mtDNA levels were unchanged (McComsey 2005b). A third trial examined the effect of Mitocnol on the function of hepatic mitochondria by means of a ¹³C methionine breath test (Banasch 2006). Mitocnol improved mitochondrial liver function despite unchanged therapy with thymidine analogs. The effect of a 3-day course of Mitocnol was noticeable after 2 weeks, persisted over several weeks, and was reproducible on Mitocnol reexposure.

Figure 3: Subcutaneous fat gain with Mitocnol under stavudine and zidovudine treatment (in comparison with NRTI-sparing strategies).

In symptomatic hyperlactatemia and in lactic acidosis, all NRTIs should be immediately discontinued (Brinkman 2000). The supplementation of vitamin cocktails has been recommended, but there are no data that demonstrate the efficacy of this intervention with respect to mtDNA-depletion (Walker 1995, Venhoff 2002). After discontinuation of NRTIs, normalization of lactate may require several weeks. More mitochondrial friendly NRTIs may then be reintroduced, but patients should be monitored closely (Lonergan 2003). The proposed supportive treatment of hyperlactatemia and lactic acidosis is summarized in Table 2.