Mitochondrial toxicities of nucleoside reverse transcriptase inhibitors

Kees Brinkman, MD, PhD Department of Internal Medicine, Onze Lieve Vrouwe Gasthuis (olvg), Amsterdam

Summary by Tim Horn, Edited by Marshall Glesby, MD, PhD, and Thomas Kakuda, PharmD

It all began with an Ethiopian refugee under Dr. Brinkman’s care, beginning in 1996. She tested positive for HIV in 1996, along with her partner and two-year-old son. At the time of her diagnosis, she was asymptomatic and had a CD4+ count of 240 cells/mm3.

Her initial responses to therapy were frustrating, at best; she experienced anemia, polyneuropathy, and various gastrointestinal complications while receiving combination therapy consisting of AZT (Retrovir), ddC (Hivid), ritonavir (Norvir), and indinavir (Crixivan). A cocktail consisting of d4T (Zerit), 3TC (Epivir), and saquinavir (Invirase) eventually proved to be well tolerated, at least initially.

In August 1997 – six months after initiating the last HAART regimen – Dr. Brinkman’s patient was responding well to therapy; her viral load was undetectable (<400 copies/mL), her CD4 count had increased by more than 100 cells/mm3, and her triglyceride level was normal. A month later, she presented with symptoms of malaise, nausea, and vomiting. Upon being admitted to the hospital, endoscopy, ultrasound of her upper abdomen, and CT scans all appeared to be normal.

However, blood tests revealed that she had slightly elevated transaminases and hypertriglyceridemia (8.1 nmol/L), and on the 12th day of hospitalisation, it turned out that she was experiencing severe lactic acidosis and ketoacidosis. She had a lactate level of 19.9, a lactate/pyruvate ratio of 50, and a B-hydroxybutyrate/acetoacetate ratio of 3.5-all three levels significantly higher than normal ranges.

‘When we placed her in the intensive-care unit,’ explained Dr. Brinkman, ‘we perfomed bicarbonate dialysis, which did not work. Her lactate levels remained high. She soon developed liver failure, followed by arrhythmias, and died shortly thereafter.’

‘There is only one explanation for these biochemical parameters,’ Dr. Brinkman stated, ‘and that is mitochondrial dysfunction.’

Mitochondria: function and dysfunction

All cells in the body, with the exception of erythrocytes, contain hundreds of mitochondria. Found within the ribbon-like structure of these energy powerhouses are the enzyme complexes and mitochondrial DNA (mtDNA) needed to help carry out oxidative phosphorylation-the aerobic process of forming high-energy bonds, primarily ATP, that can be broken down and used by cells to generate energy.

The mitochondrial enzyme complexes, which make up the oxidative phosphorylation system (OXPHOS), are composed of subunits, several of which are encoded by mtDNA. Mitochondrial DNA-between two and ten copies are inside each mitochondria-consist of a double-stranded circular DNA molecule that is prone to errors during replication. It contains no protective histones and has a mutation rate 17 times higher than that of the nuclear genome.

A number of enzymes-polymerases (a, b, g, d, and e) – are required to catalyze the formation of new nuclear DNA. Of these, only polymerase-g is responsible for the replication of mtDNA. As discussed below, inhibition of polymerase-g can have a profound effect on mtDNA synthesis.

During cellular division, there is an even distribution of cellular DNA but not mtDNA in each daughter cell. Both mutated and wild-type forms of mtDNA are randomly segregated into the cellular progeny. The severity of a defect caused by mtDNA mutation depends on the nature of the genetic error and on the proportion of mutant mtDNA within the mitochondria and the cell. For example, a cell that contains approximately 80% dysfunctional mitochondria is likely to result in a down regulation of the energy-producing OXPHOS. And, depending on the proportion of cells with dysfunctional mitochondria in each organ system-along with the energy demands of each organ system-symptomatic complications can occur.

those at nucleotide positions 3243 and 8344), duplications, or deletions in mtDNA-has been associated with a number of clinical manifestations (Johns, 1996, 1995; Wallace, 1992; Munnich, 1992; Chinnery, 1999). These include complications associated with structural mitochondrial gene mutations (i.e., Leber’s herditary optic neuropathy [LHON] and neurogenic muscle weakness, ataxia, and retinitis pigmentosa [NARP]); tRNA and rRNA mutations (i.e., mitochondrial encephalopathy, lactic acidosis, and stroke-like episodes [MELAS] and myoclonic epilepsy with ragged red fibres [MERRF]); and large-scale mtDNA deletions (i.e., Kearns-Sayre syndrome [KSS] and chronic progressive external opthalmoplegia [CPEO]).

Non-hereditary factors that have been shown to impair mitochondrial DNA synthesis include alcohol and recreational drug usage, obesity, and aging (the accumulation of mtDNA mutations above a threshold level).

NRTIs and mitochondrial toxicity

In HIV-positive patients, reports of these specific mitochondrial diseases have been few and far between. However, there are a number of manifestations related to these diseases that overlap with hallmark side effects of nucleoside analogue therapy for HIV, particularly: neurological disorders, such as peripheral neuropathy and dementia; muscular complications including hypotonia, myopathy, and cardiomyopathy; hepatocellular manifestations such as steatosis and lactic acidosis; pancreatitis; pancytopenias; and various renal problems (Brinkman, 1998).

Like cellular DNA polymerase-g, HIV contains its own polymerase: reverse transcriptase. After nucleoside reverse transcriptase inhibitors (NRTIs) are triphosphorylated intracellularly to nucleotides, they are unknowingly incorporated in the growing DNA chain by reverse transcriptase. Because they lack the hydroxyl group needed for further chain elongation, the NRTIs inhibit further DNA elongation. These nucleotides can be mistaken for natural substrates by polymerase-g and, in turn, may do irreparable harm to mtDNA during replication. Other cellular DNA polymerases involved in nuclear DNA replication are not inhibited by NRTIs.

Clinical evidence of NRTI-associated mitochondrial toxicities date back several years, including a handful of reports linking AZT-associated myopathy and mitochondrial damage (Arnaudo, 1990; Dalakas, 1990; Chariot, 1995; Peters, 1995; Casademont, 1996). In addition, at least two reports have demonstrated a clear association between neuropathy and mitochondrial toxicity in rats receiving ddC (Hivid) (Anderson, 1994; Feldman, 1994). And more recently, in a case report published in the Journal of Hepatology, a patient who experienced severe liver steatosis and lactic acidosis while receiving AZT was found to have seriously depleted mtDNA in his skeletal muscle and in liver tissue samples (Chariot, 1999).

Unfortunately, few clinical studies have examined the possible associations between other NRTIs, mitochondrial toxicity, and specific adverse events. While the widening interest in mitochondrial toxicity, particularly as it relates to lipodystrophy (see below), may lead to the development and implementation of additional studies, a primary concern is the absence of a reliable, non-invasive test. In order to measure oxidative phosphorylation activity-such as in the brain, the heart, the pancreas, and the liver-tissue biopsies are necessary, embodying a certain amount of risk for patients involved.

Turning to in vitro studies, a few have examined the effect of individual NRTIs on mitochondrial DNA synthesis. One study published by Burroughs Wellcome in1994 demonstrated that the potency of these drugs against mitochondrial DNA varies considerably, based on an examination of drug concentrations (aucs) needed to inhibit mtDNA synthesis by 50% in cell cultures (Martin, 1994). The aucs reported to have this effect on mtDNA were 0.002 mmol/L for ddC, 10 mmol/L for d4T, >100 mmol/L for AZT and abacavir, and >200 mmol/L for 3TC.

These results were similar to a study conducted a few years earlier, which found the inhibitory concentrations (IC50) to be 0.022 mmol/L for ddC, 3 mmol/L for d4T, 19 mmol/L for AZT, and 290 mmol/L for ddI (Chen, 1991). Accordingly, exposure of cells to ddC and d4T resulted in dose-dependent reductions in mtDNA synthesis. [Editor’s note: The anti-HIV IC50s of the five approved NRTIs are 0.01 mmol/L for AZT, 3.9 mmol/L for ddI, 0.5 mmol/L for ddC, 0.7 mmol/L for 3TC, and 5.3 mmol/L for abacavir.]

With respect to in vitro neurite regeneration-a model designed to allow elucidation of the cellular and molecular events involved in NRTI-related peripheral neuropathy-study results published in 1997 suggested that neither AZT nor 3TC had any inhibitory effect (Cui, 1997). Yet ddC, ddI, and d4T were found to significantly impair neurite regeneration (IC50s of 1 mmol/L, 5 mmol/L, and 15 mmol/L, respectively), in a dose-dependent manner.

As intriguing as these data are, no clear associations between in vitro findings and in vivo signs and symptoms of NRTI-induced side effects have yet been determined. At the same time, it is not at all clear why different NRTIs lead to toxicity in a variety of tissues. It might be that each tissue has different NRTI kinetics or activation enzymes, hence differing levels of active drug. And while it has been established that NRTIs vary in their ability to inhibit polymerase-g and mtDNA synthesis, it is also possible that the role of the mitochondria may be more or less important in certain tissues.

Lactic acidosis

One of the most severe and potentially life-threatening side effect of NRTI therapy is lactic acidosis (Brinkman, 1999). To date, it has been shown to occur in patients receiving either AZT, ddI, or d4T and carries an incidence rate of 1.3 per 1000 person-years (Fortgang, 1995). Although this figure was calculated during the era of NRTI monotherapy, the estimate fits well with Dr. Brinkman’s own experience in Amsterdam, where four fatal cases of lactic acidosis-of approximately 3,000 patients receiving antiretroviral therapy-were documented in 1999 (Brinkman, 2000).

The symptoms of lactic acidosis, as seen in Dr. Brinkman’s Ethiopian patient, include episodes of malaise, nausea, and vomiting, along with abdominal pain and hyperventilation (compensatory for the acidosis). These symptoms are usually accompanied by various biochemical abnormalities, chiefly lactic acidemia and increased ratios of lactate/pyruvate and b-hydroxybutyrate/acetoacetate.

‘Not many clinicians have had the opportunity to diagnose HIV-positive patients with lactic acidosis,’ Dr. Brinkman suggested. ‘But this might be because they’re not looking for it. Ordering the appropriate biochemical assays may yield surprising results. When our patient from Ethiopia died, we agreed that we had never seen a patient like her. But when my Dutch colleagues and I went back in our files, several cases were found, compatible with the clinical diagnosis of lactic acidosis.’

Several reports describing lactic acidosis in HIV-positive patients being treated with NRTIs have been published over the past year (Brivet, 2000; Megarbane, 2000; Mokrzycki, 2000; Blanche, 1999; Allaouchiche, 1999; Roy, 1999; Megarbane, 1999; Acosta, 1999).

NRTI-related lactic acidosis begins with disturbances in polymerase-g function and mtDNA synthesis. As explained above, these changes are accompanied by impairment of the oxidative phosphorylation system, along with a shift in the redox state (increased NADH/NAD+ ratio). This leads to a shift in the Krebs cycle, particularly the pyruvate/lactate equilibrium, whereby excess lactic acid is generated and accumulates within the cell, along with a build-up of triglycerides and fatty acids.

Lipodystrophy and mitochondrial toxicity

In 1997, when the term lipodystrophy was coming into vogue, an interesting report linking protease inhibitor use with a condition called multiple symmetric lipomatosis (MSL) went virtually unrecognised (Hengel, 1997). This condition, sometimes referred to as Madelung’s disease or Launois-Bensaude adenolipomatosis, yields symptoms that overlap with those seen in patients with HIV-related lipodystrophy. MSL patients generally have a low body-mass index and show symmetrical accumulation of non-encapsulated masses of fatty tissue, especially in the subcutaneous regions of the neck and shoulders and inside the mediastinum.

In addition, there is pronounced atrophy of subcutaneous fat in the extremities, similar to fat loss seen in HIV-infected patients. These body-shape changes may also be accompanied by hypertriglyceridemia and insulin resistance.

While the etiology of MSL has not been fully elucidated, a number of studies have suggested that underlying mitochondrial disease is to blame. Most patients with msl also present with peripheral neuropathy and a buildup of mitochondria-rich brown adipose tissue (found in the viscera), along with key single or multiple mutations or deletions in their mtDNA.

Interestingly, MSL usually is not associated with increases in truncal size-a hallmark symptom of HIV-related lipodystrophy-despite an apparent increase in brown adipose tissue. Yet Dr. Brinkman argues that most mitochondrial toxicities of NRTIs do not produce symptoms that are identical to those seen in patients with hereditary mitochondrial disease. In turn, mitochondrial-related body-habitus changes and metabolic alterations in HIV-infected patients might simply represent different parts of a clinical spectrum.

Some of the possible mechanisms by which NRTI-induced mitochondrial damage can contribute to body-habitus changes and metabolic abnormalities were discussed by Dr. Thomas Kakuda and his colleagues at the 1st International Workshop on Adverse Drug Reactions and Lipodystrophy in San Diego last June (Kakuda, 1999). While not usually seen in patients with msl, the increase in visceral fat associated with lipodystrophy might result from mitochondria damage of brown fat tissue, ultimately slowing lipolysis and a buildup of fat mass.

As for peripheral fat loss, Dr. Kakuda’s team hypothesised that mutations or deletions of mtDNA can lead to the release of apoptosis-inducing factor and caspase C-two proteins within the mitochondrial membrane that are responsible for cell breakdown-that may lead to premature death of adipocytes.

Several studies reported over the last year have helped to advance Dr. Brinkman’s hypothesis and to confirm earlier data indicating that protease inhibitors were not solely to blame for the metabolic/body shape abnormalities being seen in patients receiving antiretroviral therapy. An indication that NRTIs might be partly to blame, with a damning spotlight on d4T, first came from a study reported by Dr. Thierry Saint-Marc and his colleagues in Lyon (Saint-Marc, 1999). According to Dr. Saint-Marc, nine people receiving a dual combination of ddI and d4T and eight patients taking d4T and 3TC-all of whom were naive to a protease inhibitor-were found to have partial or generalised lipodystrophy, confirmed by CT scans and by ratio scoring of visceral adipose tissue (VAT) to total adipose tissue (TAT).

At the San Diego adverse events and lipodystrophy conference, Dr. Andrew Carr-a champion of the initial CRABP-1/LRP hypothesis that implicated protease inhibitors as the most likely culpritÑpresented data from a case-control study involving 14 protease inhibitor-naive, NRTI-treated patients experiencing symptoms of lipodystrophy (Carr, 2000). According to Dr. Carr, DEXA scanning and morphologic testing found that body-habitus changes in patients receiving only NRTIs were similar to those seen in protease inhibitor-treated patients with lipodystrophy. Factors believed to be associated with lipodystrophy in NRTI-treated patients included older age, d4T use, and time on therapy. Interestingly, there was an absence of hyperlipidemia or abnormalities in fasting glucose and insulin in patients receiving only NRTIs. Several NRTI-treated patients with body-habitus changes also presented with low serum bicarbonate concentrations and elevated lactic acid concentrations, leading Dr. Carr’s team to conclude that NRTIs produce lipodystrophy as a result of mitochondrial toxicity.

Results of another study, reported by Dr. Mina John and her colleagues in Western Australia, looked at the incidence rate of lipoatrophy (peripheral fat loss)-confirmed by means of physical examination and dexa in a subgroup of patients-in a cohort of 277 patients (John, 1999). Approximately 51% of patients receiving a protease inhibitor-based combination were found to have symptoms of lipoatrophy, compared to 13% of patients receiving only an NRTI-based regimen. According to Dr. John’s report, lipodystrophy was significantly associated with d4T use and total time on NRTI therapy.

Finally, data from a large Centers for Disease Control-sponsored cohort study conducted by Drs. Doug Ward and Ken Lichtenstein and their colleagues demonstrated that the cause of lipodystrophy is probably multifactorial (Ward, 1999). Of 1077 patients taking NRTIs, non-nucleoside analogues, and/or protease inhibitors, lipodystrophy was highly correlated to time on therapy as well as use of d4T, 3TC, and indinavir.

With respect to all the finger-pointing at d4T as a primary culprit, Dr. Brinkman remains cautious with this implication. ‘Many people ask me, ‘Is d4T to blame?’ D4T may be playing some role, but it’s certainly not the only answer. All NRTIs need to be looked at carefully. It’s simply too early to blame it all on d4T. This issue needs to be addressed in well-controlled prospective studies.’

‘We know that haart always includes at least two NRTIs,’ Dr. Brinkman commented. ‘We also know that lipodystrophy is occurring in patients receiving therapies that include only NRTIs. Based on these findings, along with what is known about mitochondrial toxicity, we hypothesise that NRTIs have a key role in the pathogenesis of lipodystrophy.’

Putting it to the test

In September 1999, Dr. Brinkman – in collaboration with Drs. Jan Smeitnik, Johannes Romijn, and Peter Reiss, all from The Netherlands – published an elaborate hypothesis in the Lancet (Brinkman 1999a). Along with a detailed summary of background information, Dr. Brinkman and his colleagues support their postulate with a detailed overview of how proof of this hypothesis might be achieved.

First, a working case definition of lipodystrophy needs to be established, enabling objective rating and quantification of the features. But as explained in an update regarding the epidemiology and pathogenesis of lipodystrophy in the March 2000 issue of The PRN Notebook, attempts to reach a consensus upon the criteria have not yet been successful (see ‘The Epidemiology and Pathogenesis of Lipodystrophy in HIV Disease: An Update’ in Volume 5, Number 1).

Second, prospective studies linking lipodystrophy to either protease inhibitors, NRTIs, or other potential variables need to be conducted. ‘We have some studies examining NRTI-only HAART combinations and we need to look at the incidence of lipodystrophy-related symptoms very carefully.’

Third, the occurrence of mitochondrial dysfunction during NRTI therapy must be further investigated. As explained by Dr. Brinkman, studies will most likely need to be done with tissue samples, thus requiring invasive procedures. As for non-invasive methods, Dr. Brinkman contends that serum lactate concentrations may be useful in making a diagnosis of an underlying mitochondrial toxicity, but say little about what is occurring in the tissues involved. ‘We really need tissue samples to figure out what’s going on.’

Finally, with respect to lipodystrophy, it will be necessary not only to demonstrate mitochondrial dysfunction in adipocytes, but also to conduct sound animal and human studies. ‘We know lipodystrophy can take several months to develop,’ Dr. Brinkman said in his concluding remarks. ‘Studies to test this hypothesis will take a significant amount of time to produce convincing results. The answers we are looking for will not be immediate.’


Non-invasive, reliable and predictive tests for the early development of mitochondrial dysfunction are urgently needed.

The following report details some preliminary studies suggesting that blood lactate levels and exercise tolerance testing may be of utility.

Additionally, prospective studies will be required to ensure that multinucleoside analogue combinations (ie. triple nucleosides) do not carry additional risks of mitochondrial toxicities


  • Acosta BS, Grimsley EW. Zidovudine-associated type B lactic acidosis and hepatic steatosis in an HIV-infected patient. South Med J 92(4):421-3, 1999.
  • Allaouchiche B, Duflo F, Cotte L, et al. Acute pancreatitis with severe lactic acidosis in an HIV-infected patient on didanosine therapy. J Antimicrob Chemother 44(1):137-8, 1999.
  • Anderson T, Davidovich A, Feldman D, et al. Mitochondrial schwannopathy and peripheral myelinopathy in rabbit model of dideoxycitidine neurotoxicity. Lab Invest 70:724-39, 1994.
  • Arnaudo E, Dalakas M, Shanske S, et al. Depletion of muscle mitochondrial DNA in AIDS patients with zidovudine-induced myopathy. Lancet 337:508-10, 1991.
  • Blanche S, Tardieu M, Rustin P, et al. Persistent mitochondrial dysfunction and perinatal exposure to antiretroviral nucleoside analogues. Lancet 354(9184):1084-9, 1999.
  • Brinkman K, ter Hofstede HJ. Mitochondrial toxicity of nucleoside analogue reverse transcriptase inhibitors: lactic acidosis, risk factors and therapeutic options. AIDS 1:141-148, 1999.
  • Brinkman K, Smeitnik JA, Romijn JA, et al. Mitochondrial toxicity induced by nucleoside-analogue reverse-transcriptase inhibitors is a key factor in the pathogenesis of antiretroviral-therapy-related lipodystrophy. Lancet 354:1112-5, 1999a.
  • Brinkman K, ter Hofstede HJ, Burger DM, et al. Adverse effects of reverse transcriptase inhibitors: mitochondrial toxicity as common pathway. AIDS 12: 1735-1744, 1998.
  • Brivet FG, Nion I, Megarbane B, et al. Fatal lactic acidosis and liver steatosis associated with didanosine and stavudine treatment: a respiratory chain dysfunction? J Hepatol. 32(2):364-5, 2000.
  • Carr A, Miller J, Law M, Cooper DA. A syndrome of lipoatrophy, lactic acidaemia and liver dysfunction associated with HIV nucleoside analogue therapy: contribution to protease inhibitor-related lipodystrophy syndrome. AIDS 14(3):F25-32, 2000.
  • Casademont J, Barrientos A, Grau J, et al. The effect of zidovudine on skeletal muscle mtDNA in HIV-1 infected patients with mild or no muscle dysfunction. Brain 119:1357-64, 1996.
  • Chariot P, Drogou I, de Lacroix-Szmania I, et al. Zidovudine-induced mitochondrial disorder with massive liver steatosis, myopathy, lactic acidosis, and mitochondrial DNA depletion. J Hepatol 30:156-60, 1999.
  • Chariot P, Gherardi R. Myopathy and HIV infection. Curr Opin Rheumatol 7:497-502, 1995.
  • Chinnery P, Howell N, Andrews R, et al. Clinical mitochondrial genetics. J Med Genet 36:425-39, 1999.
  • Cui L, Locatelli L, Xei, MY, et al. Effect of nucleoside analogs on neurite regeneration and mitochondrial DNA synthesis in PC-12 cells. J Pharm Exp Ther 12:1228-34, 1997.
  • Dalakas M, Illa I, Pezehkpour G, et al. Mitochondrial myopathy caused by long-term zidovudine toxicity. N Engl J Med 322:1098-105, 1990.
  • Feldman D, Anderson T. Schwann cell mitochondrial alterations in peripheral nerves of rabbits treated with 2′,3′-dideoxycitidine. Acta Neuropathol Berl 87:71-80, 1994.Fortgang I, Belitsos P, Chaisson R, et al. Hepatomegaly and steatosis in HIV-infected patients receiving nucleoside analogue antiretroviral therapy. Am J Gastroenterol 90:1433-6, 1995.
  • Hengel RL, Watts NB, Lennox JL, et al. Benign symmetric lipomatosis associated with protease inhibitors. Lancet 350:1596, 1997.
  • Johns D. The other human genome: Mitochondrial DNA and disease. Nat Med 2:1065-8, 1996.
  • Johns D. Seminars in medicine of the Beth Israel Hospital, Boston. Mitochondrial DNA and disease. N Engl J Med 333:63-44, 1995.
  • Megarbane B, Brivet F, Guerin JM, et al. Lactic acidosis and multi-organ failure secondary to anti-retroviral therapy in HIV-infected patients. Presse Med 28(40):2257-64, 2000.
  • Megarbane B, Fromont C, Nion I, et al. A very, very short of breath HIV-infected patient. Rev Med Intern 20 Suppl 2:256s-259s, 1999.
  • Mokrzycki MH, Harris C, May H, et al. Lactic acidosis associated with stavudine administration: a report of five cases. Clin Infect Dis 30(1):198-200, 2000.
  • Munnich A, Rustin P, Rotig A, et al. Clinical aspects of mitochondrial disorders. J Inherit Metab Dis 15:448-55, 1992.
  • Peters B, Winer J, Landon D, et al. Mitochondrial myopathy associated with chronic zidovudine therapy in AIDS. Q J Med 86:5-15, 1993.
  • Roy PM, Gouello JP, Pennison-Besnier, et al. Severe lactic acidosis induced by nucleoside analogues in an HIV-infected man. Ann Emerg Med 34(2):282-4, 1999.
  • Saint-Marc T, Partisiani M, Poizot-Martin J, et al. A syndrome of peripheral fat wasting (lipodystrophy) in patients receiving long-term nucleoside analogue therapy. AIDS 13:1659-67, 1999.
  • Wallace D. Mitochondrial genetics: A paradigm for aging and degenerative diseases? Science 5057:628-32, 1992.

Source: PRN Notebook, (c) 1998 – 2000. Physicians’ Research Network, Inc. All rights reserved.

Links to other websites are current at date of posting but not maintained.