| |
Perinatal genotoxicity and carcinogenicity of
anti-retroviral nucleoside analog drugs
|
| |
| |
Download the PDF here
AZT and 3TC are DNA replication chain terminators known to induce various types of genotoxicity......evidence of genotoxicity in mouse and monkey models and in the infants themselves would suggest that exposed children should be followed well past adolescence for early detection of potential cancer hazard.
Studies conducted in two strains of mouse and presented here have demonstrated that: (1) AZT is a moderately strong transplacental carcinogen in offspring at 1 and 2 years after birth; (2) at birth, AZT can be found incorporated into the DNA of multiple fetal organs; (3) the mutagenic consequences of AZT-DNA damage are dose related and significant; and (4) some organs have shortened telomeres. In many different types of experimental models, causative associations among DNA damage, mutagenesis, and tumor formation have been demonstrated, and the dose-response relationships shown here for all three parameters suggest that AZT exposure is no exception. The mouse studies make a strong case for the potential carcinogenicity of transplacental NRTI exposure in transplacentally exposed infants, but the mouse pregnancy is very different from the human pregnancy.
Toxicology and Applied Pharmacology 199 (2004) 151- 161
Miriam C. Poirier,a,* Ofelia A. Olivero,a Dale M. Walker,b and Vernon E. Walkerb
a Center for Cancer Research, National Cancer Institute, NIH, Bethesda, MD 20892-4255, USA
b Lovelace Respiratory Research Institute, Albuquerque, NM 87108, USA
Abstract
The current worldwide spread of the human immunodeficiency virus-1 (HIV-1) to the heterosexual population has resulted in approximately 800 000 children born yearly to HIV-1-infected mothers. In the absence of anti-retroviral intervention, about 25% of the approximately 7000 children born yearly to HIV-1-infected women in the United States are HIV-1 infected. Administration of zidovudine (AZT) prophylaxis during pregnancy reduces the rate of infant HIV-1 infection to approximately 7%, and further reductions are achieved with the addition of lamivudine (3TC) in the clinical formulation Combivir. Whereas clinically this is a remarkable achievement, AZT and 3TC are DNA replication chain terminators known to induce various types of genotoxicity. Studies in rodents have demonstrated AZT-DNA incorporation, HPRT mutagenesis, telomere shortening, and tumorigenicity in organs of fetal mice exposed transplacentally to AZT. In monkeys, both AZT and 3TC become incorporated into the DNA from multiple fetal organs taken at birth after administration of human-equivalent protocols to pregnant dams during gestation, and telomere shortening has been found in monkey fetuses exposed to both drugs. In human infants, AZT-DNA and 3TC-DNA incorporation as well as HPRT and GPA mutagenesis have been documented in cord blood from infants exposed in utero to Combivir. In infants of mice, monkeys, and humans, levels of AZT-DNA incorporation were remarkably similar, and in newborn mice and humans, mutation frequencies were also very similar. Given the risk-benefit ratio, these highly successful drugs will continue to be used for prevention of vertical viral transmission, however evidence of genotoxicity in mouse and monkey models and in the infants themselves would suggest that exposed children should be followed well past adolescence for early detection of potential cancer hazard.
Keywords
· HIV/AIDS;
· Zidovudine (AZT);
· Lamivudine (3TC);
· Combivir;
· NRTIs;
· Fetus;
· Mice;
· Monkey;
· DNA incorporation;
· Mutagenesis;
· Tumor;
· Telomere shortening;
· Newborn;
· Infant
Abbreviations
· AZT, zidovudine or 3'-azido-3'-deoxythymidine;
· 3TC, lamivudine or [-]-ß-l-2',3'-dideoxy-3'-thiacytidine;
· Combivir, clinical formulation that includes both AZT and 3TC;
· HPRT, hypoxanthine guanine phosphoribosyl-transferase;
· GPA, glycophorin A;
· Mf, mutant frequency;
· NRTI, nucleoside reverse transcriptase inhibitor;
· Vf, variant frequency;
· TK, thymidine kinase;
· TPA, 12-O-tetradecanoylphorbol-13-acetate;
· 8-oxo-dG, 8-hydroxy-deoxyguanine;
· RIA, radioimmunoassay
Introduction
Drug and chemical exposures that are carcinogenic for the adult population are likely to be even more potent carcinogens when exposure occurs during development. Exposure of pregnant females of numerous mammalian species, including nonhuman primates, to some chemical carcinogens results in neoplasms in their offspring (Napalkov et al., 1989). The human relevance of transplacental carcinogenesis was established with the discovery that diethylstilbestrol (DES) caused vaginal adenocarcinomas in the children of women treated during pregnancy (Herbst et al., 1971). Subsequent experimental studies in mice demonstrated a similar effect (Newbold, 1995) and provided an opportunity to apply studies designed to elucidate the underlying mechanisms. Epidemiological evidence has implicated transplacental exposures to radiation, some medications, some occupational chemicals, and metals [Bunin et al., 1993], [Savitz and Chen, 1990] and [Zahm and Devesa, 1995] as possible contributors to human cancer risk. In certain instances, including the maternal-fetal transmission of the human immunodeficiency virus (HIV-1), there are compelling reasons to administer medications that have known genotoxicity in experimental models and may therefore have the potential for long-term carcinogenic effects. This review will consider the perinatal genotoxicity of two anti-retroviral nucleoside analog drugs that play a prominent role in the highly successful drug combinations currently used for the treatment of HIV-1, and inhibition of maternal-fetal HIV-1 transmission.
Clinical use of anti-retroviral nucleoside analog drugs in human pregnancy
The current spread of HIV-1 to the heterosexual population in the United States has resulted in approximately 7000 children born yearly to HIV-1-infected mothers. With no protective treatment, about 25% of these children will also be HIV-1 infected [Connor et al., 1994] and [Sperling et al., 1996]. Perinatal transmission is the primary mode of HIV-1 acquisition in children, and pediatric HIV-1 infection is a major cause of morbidity and mortality in children worldwide. Prevention of perinatal transmission became a reality in 1994 when the results of Pediatric AIDS Clinical Trials Group (PACTG) Protocol 076 demonstrated that a three-part regimen of zidovudine (3'-azido-3'-deoxythymidine, AZT) chemoprophylaxis administered to the mother during the last 6 months of pregnancy and during labor, and to the newborn for the first 6 weeks of life, could substantially reduce the risk of vertical viral transmission (Connor et al., 1994). In August 1994, a US Public Health Service (USPHS) Task Force recommended the use of AZT prophylaxis for reduction of perinatal HIV-1 transmission (Centers for Disease Control and Prevention, 1995). When given as the sole anti-retroviral prophylaxis during pregnancy, AZT reduces the rate of infant HIV-1 infection from approximately 25% to 7% [Cooper et al., 1996], [Fiscus et al., 1996] and [Sperling et al., 1996].
Combination anti-retroviral therapy, consisting of two nucleoside analogue reverse transcriptase inhibitors (NRTIs) and a non-nucleoside reverse transcriptase inhibitor or a protease inhibitor, is currently employed as standard-of-care for treatment of HIV-1-infected adults, including pregnant women. The NRTI combination of AZT plus lamivudine ([-]-ß-l-2',3'-dideoxy-3'-thiacytidine, 3TC), marketed as a single drug, Combivir, may achieve transmission rates as low as 1% (2003 PHS Task Force Recommendations, online at http://www.aidsinfo.nih.gov/guidelines/perinatal) when used in combination with Cesarean delivery. Whereas the use of prophylactic anti-retroviral drug combinations in pregnancy constitutes one of the great success stories in the war on HIV or AIDS, in utero exposure to drug "cocktails" during pregnancy is likely to become increasingly frequent, and little is known about the long-term consequences of these exposures in HIV-1-uninfected infants and children.
Mechanisms of drug efficacy and genotoxicity
The "dideoxy-type" anti-retroviral NRTIs are similar in structure to the DNA bases (Fig. 1); however, as they lack a ribose 3'-hydroxyl moiety, the incorporated species are unable to support replicative DNA chain extension [Barile et al., 1998], [Morse et al., 1993], [Peter and Gambertoglio, 1998] and [Veal and Back, 1995]. In most viral and mammalian systems, 5'-phosphorylation is readily accomplished and drug-DNA incorporation, with DNA chain termination, occurs in both nuclear and mitochondrial DNA. The virustatic activity of the drug is thought to result from a direct inhibition of the viral reverse transcriptase as well as from truncation of proviral DNA replication [Barile et al., 1998], [Palmer and Cox, 1997] and [Peter and Gambertoglio, 1998]. As these drugs also become incorporated into the host nuclear and mitochondrial DNA, there is the opportunity for short- and long-term toxic consequences. In this review, we focus only on the genotoxicity of NRTIs, although a large literature documents the mitochondrial toxicity of this class of drug [Dagan et al., 2002] and [Lewis and Dalakas, 1995].
The genotoxicity of NRTIs, resulting from DNA chain termination, has been well established in experimental model systems in vitro (Zidovudine, IARC Monograph Vol. 76, 2000). AZT is incorporated into the DNA of cultured bacteria and mammalian cells, and has displayed genotoxic activity primarily in assays that reflect clastogenic effects. In bacterial systems, the data indicate that AZT does not cause DNA lesions that are repaired by the excision (uvr A) or error-free post-replication repair (recA) processes, but rather that AZT, acting as a DNA chain terminator, generates an SOS-inducing response leading to the inhibition of DNA replication [Beskid et al., 1981], [Elwell et al., 1987] and [Mamber et al., 1990]. In mammalian cells, the observed genotoxic effects of AZT have included gene mutations, sister chromatid exchanges, micronucleus formation, chromosomal aberrations, and cell transformation (Zidovudine, IARC Monograph Vol. 76, 2000). The results of in vitro studies using gene mutation and clastogenicity assays in AZT-exposed human cells have been consistently positive, both at high exposures and at low exposures approximating plasma levels in AZT-treated patients. In TK6 cells (human lymphoblastoid cells heterozygous for the thymidine kinase or TK gene) exposed to an AZT dose about 15-fold higher than human peak plasma concentrations, increases in HPRT mutant frequency, total gene deletions, and point mutations were observed (Sussman et al., 1999). In TK6 cells exposed to physiological AZT concentrations (Meng et al., 2000a), mutant frequencies at the TK locus were increased at all AZT exposure levels, while Southern blot analyses showed a loss of heterozygosity through large deletions for 84% of AZT-induced TK mutants. Mutagenic responses were observed in the HPRT and TK genes of TK6 cells exposed in culture to AZT, or equimolar combinations of AZT + ddI, at exposure concentrations ranging from 3 to 30 times the peak plasma levels found in some patients (Meng et al., 2000b). Furthermore, the multiplicative synergistic enhancement of the mutagenic responses to the combined drug exposures correlated well with the potentiation of AZT-DNA incorporation by AZT-ddI co-exposures as compared with single drug exposures. All of these indicators of genotoxicity are consistent with the known mechanism of AZT as a DNA chain terminator.
Perinatal AZT exposure in rodent models
Tumor induction
Transplacental AZT exposure in mice
In adult mice, AZT is a weak carcinogen. Ayers et al. (1996) reported 10% and the National Toxicology Program reported 22% (NTP, 1996) incidences of vaginal squamous papillomas and carcinomas after lifetime oral administration of the drug to CD-1 and B6C3F1 mice, respectively. Because AZT is a known carcinogen in adult mice, and because this drug is used in human pregnancy, studies were initiated to examine AZT transplacental tumorigenicity in mice. Investigators at two different institutions have exposed pregnant mice using different mouse strains, and both have concluded that AZT is a moderately strong transplacental carcinogen in this species. An additional transplacental mouse study that employed much lower AZT doses [Ayers et al., 1996] and [Ayers et al., 1997] showed no tumorigenic effect, suggesting a dose-response and/or a no effect level.
In experiments with CD-1 mice [Diwan et al., 1999], [Olivero et al., 1997] and [Zhang et al., 1998], pregnant mice were exposed by gavage to 12.5 and 25.0 mg AZT daily on days 12-18 of gestation. At the high dose level, the total AZT given during gestation was approximately 3.0 g kg-1 body weight, similar to the total dose recommended for women during the last 6 months of pregnancy (approximately 1.5 g AZT kg-1 body weight). In the offspring at 1 year of age (Olivero et al., 1997), dose-dependent and statistically significant increases in incidences and multiplicities of lung, liver, skin, and female reproductive organ tumors were observed. In lungs of males and females and livers of males, tumor incidences (38-50%) in mice exposed to the high AZT dose were 4- to 5-fold higher than tumor incidences (5-12%) observed in unexposed controls. In females, a 17% tumor incidence was observed in reproductive organs at the higher dose compared to no tumors in the controls (Olivero et al., 1997). In the offspring at 2 years of age (Diwan et al., 1999), the background tumor incidences were elevated in unexposed controls; however, statistically significant increases in lung tumors, histiocytic sarcomas, mammary tumors, and ovarian tumors in females, as well as seminal vesicle tumors in males, were observed in the AZT-exposed groups. In a separate experiment in which transplacentally exposed mice were subjected to skin tumor promotion, the incidence of skin papillomas in 1-year-old mice exposed transplacentally to AZT and given topical 12-O-tetradecanoylphorbol-13-acetate (TPA) promotion was double that of mice given the TPA alone (Zhang et al., 1998).
The carcinogenic potential of transplacental AZT exposure was confirmed in a second mouse strain (Walker et al., 2004). Pregnant B6C3F1 mice were given AZT daily at three dose levels during the last 7 days of gestation. The high and middle AZT doses (480 and 240 mg kg-1 bw day-1) were essentially the same as those used in CD-1 mice [Diwan et al., 1999] and [Olivero et al., 1997]. The low-dose (80 mg kg-1 bw day-1) was 2-fold higher than the highest dose reported by Ayers et al. (1997) to give no tumors in offspring of AZT-exposed CD-1 mice. Groups of offspring were subjected to complete necropsy at 1 and 2 years of age. There was a significant increase in the incidence of lung neoplasms (adenomas and carcinomas combined) and a significant dose-related shift from benign to malignant hepatocellular neoplasms (with multiplicity and metastases to lung) in AZT-exposed male mice at 2 years after birth. The carcinogenic effects in lung and liver of transplacentally exposed male B6C3F1 mice were consistent with those found in transplacentally exposed CD-1 mice [Diwan et al., 1999] and [Olivero et al., 1997]; however, in B6C3F1 male mice, there were no significant increases in tumor incidence at 1 year of age, and in female mice, tumors were not increased at 1 or 2 years of age. The data suggest that AZT may be a less potent transplacental carcinogen in the B6C3F1 mouse compared to the CD-1 mouse.
Perinatal NRTI exposures in newborn mice
To model the human post-natal AZT exposure, AZT was administered to unexposed newborn CD-1 mouse pups on post-natal days 1 through 8 at doses of 25, 50, 100, and 200 mg AZT kg-1 bw day-1(Diwan et al., 1999). At 2 years of age, statistically significant increases in lung and liver tumors were observed in female mice. In these organs, highly significant dose-response trends were observed. The unexposed animals had lung and liver tumor incidences of 17% and 4%, respectively, while mice exposed to the high AZT dose (1.6 g AZT kg-1 bw) had lung and liver tumor incidences of 70% and 40%, respectively. In another study (Von Tungeln et al., 2000), newborn B6C3F1 mice were given either AZT or 3TC at doses of 125 and 250 mg kg-1 bw on post-natal days 8 and 15, for total doses of 0.25 and 0.50 g kg-1 bw. These animals showed no increases in lifetime tumor incidences. Taken together, these studies demonstrate the carcinogenic influence of NRTI exposures in newborn mice, but underscore the importance of mouse strain, treatment timing, and total dose on NRTI-induced tumorigenesis.
Transplacental AZT exposure in rats
A weak, but significant, carcinogenic potential of transplacental AZT exposure was demonstrated in a second rodent species, the rat (Walker et al., 2004). Pregnant female F344 rats were exposed to the same AZT doses used in the B6C3F1 mouse study (480, 240, and 80 mg AZT kg-1 bw day-1) for the last 7 days of gestation. In the rat offspring, a significant increase in the incidence of gliomas was observed in AZT-exposed female rats (2-4%) compared to the unexposed controls (0%). The average historical incidence of gliomas in control F344 rats was reported as 0.5% in females from 2-year bioassays (up to 26 months) versus 1.5% in females allowed to live out their life span (up to 34 months) [Solleveld et al., 1984] and [Walker and Swenberg, 1989], suggesting that the increased occurrence of gliomas in AZT-exposed female rats may be drug induced.
Biomarkers of AZT Genotoxicity in Mice
Alterations in DNA structure
In studies designed to examine the genotoxic effects of in utero AZT exposure, pregnant mice were given 25 mg AZT day-1 on days 12-18 of gestation. This was the high dose used in the tumor studies (see above), and pups were examined at birth for AZT-DNA incorporation and 8-hydroxy-deoxyguanine (8-oxo-dG) levels. Levels of AZT incorporated into organ DNA were measured by AZT-radioimmunoassay (AZT-RIA) (Olivero et al., 1997), while levels of 8-oxo-dG were measured by electrochemical detection (Bialkowska et al., 2000), and telomere length was examined by Southern blot (Olivero et al., 1997). Each litter was considered one experimental unit, as all organs (e.g., kidneys) from the pups in one litter were combined to obtain sufficient DNA for assay. The organs examined were brain, lung, liver, kidney, and skin. AZT-DNA incorporation was measurable in nuclear and mitochondrial DNA of some organs from each of three litters (Olivero et al., 1997). Nine of 15 samples were positive, with values ranging from 7.7 to 100.9 molecules of AZT/106 nucleotides (Table 1). Levels of 8-oxo-dG were significantly elevated in nuclear DNA from mouse livers and kidneys to AZT-exposed pups (n = 5 litters), with values ranging from 1.4- to 1.8-fold higher than the unexposed controls (Bialkowska et al., 2000).
To reproduce and extend the above study (Olivero et al., 1997), a dose response for AZT-DNA incorporation was evaluated (Walker and Poirier, 2002) in offspring of pregnant B6C3F1 mice exposed to AZT during the last 7 days of gestation using the same doses that induced transplacental tumors in CD1 and B6C3F1 mice [Diwan et al., 1999], [Olivero et al., 1997] and [Walker et al., 2004]. In pups taken immediately after birth, AZT-DNA incorporation was detected in all tissues evaluated (brain, kidney, liver, lung, and ovary) at levels of 10-120 molecules AZT per 106 nucleotides (Table 1). These values were strikingly similar to those observed in CD-1 mice.
HPRT mutagenesis
Inactivation of the HPRT gene, which codes for a nonessential purine salvage pathway enzyme, allows for subsequent cloning and selection of HPRT mutant T-cells (Albertini et al., 1990). In this study, pregnant CD-1 mice were exposed as in the tumor studies (see above), and T cells were isolated from thymus and spleen. The maximum HPRT mutation induction was observed in cultured T-cells taken from mice at about 2 weeks of age (Walker and Poirier, 2002). A dose-related increase in HPRT mutant frequency was observed in AZT-exposed mice with statistical significance for both spleen (P = 0.05) and thymus (P = 0.01) (Table 1). The dose-responses seen with increases in AZT-induced T-cell mutant frequency were consistent with the dose-response patterns for tumor incidence observed in the same mouse strain (Olivero et al., 1997).
Telomere length studies
Studies in cell culture demonstrating that AZT becomes incorporated preferentially into telomeric DNA [Gomez et al., 1995], [Olivero and Poirier, 1993] and [Olivero et al., 1994] suggested that telomeres might also be targeted in in vivo models. Therefore, telomere length was examined in organ DNA samples from AZT-exposed and unexposed newborn mice (Olivero et al., 1997). Compared to unexposed animals, telomeres were shortened (from about 22 to 2 kb) in DNA from livers, brains, and lungs of most AZT-exposed mice, confirming that AZT damages telomeres in some organs from offspring of pregnant mice given the drug during gestation (Table 1).
Transplacental NRTI exposures in primate models
Fetal organ DNA incorporation of AZT or 3TC after in utero exposure
Compared to humans, differences in rodent metabolism and differences in rodent gestational dynamics suggest that the mouse is not an ideal model in which to study nucleoside analog exposures in human fetuses. However, the nonhuman primate is an exceptionally useful model for several reasons. The metabolism and single-birth dynamics of AZT in primate gestation have been studied extensively [Garland et al., 1998], [Hankins et al., 1990], [Patterson et al., 1997], [Poirier et al., 1999] and [Tuntland et al., 1998]. The drug becomes mono-, di-, and tri-phosphorylated, and AZT pharmacokinetics, excretion of AZT and AZT-glucuronide, and accumulation of AZT in the amniotic fluid measured in pregnant monkeys [Garland et al., 1998], [Hankins et al., 1990], [Patterson et al., 1997], [Poirier et al., 1999] and [Tuntland et al., 1998] are very similar to those observed in pregnant women [O'Sullivan et al., 1993], [Pons et al., 1991] and [Sperling et al., 1992]. Therefore, monkey models have been used to examine the genotoxicity of transplacental AZT or 3TC exposures.
AZT-DNA incorporation and pharmacokinetics after short duration exposures
Whereas the recommended AZT treatment for HIV-1-infected pregnant women composed the last 6 months of pregnancy [Centers for Disease Control and Prevention, 1995] and [Mofenson, 1998], recent studies have shown that short dosing times may also be effective in reducing maternal-fetal viral transmission (PHS Task Force Recommendations for the Use of Anti-retroviral Drugs in Pregnant HIV-1-Infected Women, online at http://www.aidsinfo.nih.gov/guidelines/perinatal). To examine the genotoxicity of short-term dosing in a monkey model, four pregnant rhesus (Macaca mulatta) monkeys were given a 4 h AZT infusion at a dose equivalent (mg kg-1 bw) to the daily dose received by pregnant women (Poirier et al., 1999). Nuclear DNA from eight fetal organs showed a substantial number of positive samples with values ranging from 28.7 to 1944.4 molecules of AZT/106 nucleotides. In addition, AZT pharmacokinetics, evaluated on a single monkey, were almost identical to those reported for eight pregnant women given a similar daily dose of drug (O'Sullivan et al., 1993). This study demonstrates that, in the primate, AZT crosses the placenta and becomes rapidly incorporated into DNA of multiple fetal organs, exhibiting pharmacokinetics very similar to the human.
NRTI-DNA incorporation after long-duration exposures modeled on human clinical protocols
Genotoxicity was determined in fetal patas monkeys taken at term after in utero exposure to no drug (n = 2), AZT alone (n = 3), or AZT plus 3TC (n = 3) (Olivero et al., 2002). To model typical clinical protocols, pregnant Erythrocebus patas monkeys were given 40.0 mg AZT daily, 5 days week-1, for the last 10 weeks (approximately 50%) of gestation with or without the addition of 25.0 mg 3TC daily, 5 days week-1, for the last 4 weeks of gestation. Examination of multiple (n = 9-10) fetal organs at term revealed that AZT-DNA incorporation was similar in fetuses exposed to AZT alone or AZT plus 3TC, with values ranging from 7 to 296 molecules of AZT/106 nucleotides in positive samples (Table 1). Values for 3TC-DNA in fetal organs (Olivero et al., 2002) were greater than or equal to values for AZT-DNA and ranged from 20 to 1081 molecules of 3TC/106 nucleotides in positive samples (Table 1), indicating that the total DNA damage sustained by fetuses exposed to both drugs was at least double that observed in fetuses exposed to AZT alone.
Telomere length in NRTI-exposed fetuses
Telomere length was determined by Southern blot with a telomeric probe in DNA samples from 9 to 10 fetal patas organs obtained at birth (Olivero et al., 2002). In animals exposed to the combination of AZT and 3TC, there was significant telomere shortening (from about 23 kb to 2 kb) in heart, skeletal muscle, gonads, spleen, lung, and liver DNA (Table 1). No telomere shortening was evident in the unexposed fetuses. In fetuses exposed to AZT alone, occasional telomere shortening was found; however, the frequency of this phenomenon was far less than in mouse fetuses exposed to AZT alone (Olivero et al., 1997).
Studies in infants born to HIV-1-infected pregnant women
AZT-DNA incorporation in mother and infant leukocyte DNA samples
To assess the genotoxic potential of transplacental AZT exposure in HIV-1-uninfected infants born to HIV-1-infected women, peripheral blood leukocytes from 12 pregnant women given AZT therapy, and cord blood leukocytes from 22 infants of HIV-1-infected mothers were examined for AZT-DNA incorporation (Olivero et al., 1999). DNA was prepared from all of these samples and assayed by AZT-RIA. The majority (about 68%) of samples from AZT-exposed individuals, either pregnant women (8 of 12) or infants at delivery (15 of 22), had detectable AZT-DNA levels (Table 1). The range of positive values was 35-215 molecules of AZT/106 nucleotides for the mothers and 22-452 molecules of AZT/106 nucleotides for the infants (Olivero et al., 1999). Analysis of 11 mother-infant pairs showed variable AZT-DNA incorporation with no correlation by pair or by duration of drug treatment during pregnancy (Olivero et al., 1999). The 20-fold range for AZT-DNA values in both adults and infants suggested large interindividual differences in AZT phosphorylation. The data demonstrate that human adults and infants incorporate AZT into leukocyte DNA to the same extent as the other primates studied [Olivero et al., 1997], [Olivero et al., 2002] and [Poirier et al., 1999], and to a greater extent than mouse pups exposed in utero to a tumorigenic dose of AZT (Olivero et al., 1997).
An observational cohort study of AZT metabolism, AZT-DNA incorporation, and 3TC-DNA incorporation
The primary metabolic pathways involved in human AZT disposition include glucuronidation and phosphorylation, and concentrations of AZT and AZT-glucuronide have been determined in plasma of mother-child pairs to assess whether the genotoxic and mutagenic effects of AZT are associated with AZT bioavailability in plasma. Blood samples collected from mother-child pairs in an observational cohort study (Walker et al., 2001) included HIV-1-infected pregnant women receiving AZT alone or Combivir, and uninfected and unexposed pregnant women. Assay of plasma for levels of AZT and AZT-glucuronide did not reveal a correlation between the fraction of AZT available in plasma and levels of AZT incorporated into DNA. However, intracellular formation of the AZT-triphosphate, which becomes incorporated into DNA, is likely to be a major determinant in the availability of active drug. Therefore, cellular markers that included AZT, AZT-monophosphate, AZT-diphosphate, AZT-triphosphate, and total phosphorylated AZT were also compared in the infants from the observational cohort study (Meng et al., 2001). Biological specimens from 18 children were selected to include 9 infants with no evidence of AZT in plasma and 9 infants with positive AZT values measured in plasma. When metabolite values were compared with AZT-DNA levels, there was a positive correlation between intracellular levels of AZT-triphosphate and AZT-DNA incorporation cord blood lymphocytes (r = 0.612), as well as a confirmatory positive correlation between intracellular levels of AZT-diphosphate and AZT-triphosphate (P = 0.016) (Meng et al., 2001). This study demonstrates that intracellular metabolism to AZT-triphosphate correlates directly with the level of AZT incorporated into host cell DNA.
As described above (Meng et al., 2001), AZT-DNA incorporation was detectable in DNA samples from 63% of the mothers exposed to AZT alone, giving values of 57-100 AZT molecules/106 nucleotides. In DNA samples from women given both AZT and 3TC during pregnancy, 71% had detectable AZT-DNA levels with values ranging from 10 to 242 AZT molecules/106 nucleotides. In newborns, fewer infants had detectable AZT-DNA incorporation levels in the group exposed to AZT alone (60%) compared to the group exposed to Combivir (83%). In addition, the average AZT-DNA incorporation levels in the group exposed to AZT alone was 6-33 AZT molecules/106 nucleotides, a range several folds lower than the 2-152 AZT molecules/106 nucleotides observed in the Combivir-exposed group (P = 0.028) (Table 1).
Cord blood mononuclear cells from a small set of Combivir-exposed infants (n = 13) were evaluated for DNA incorporation of both AZT (Meng et al., 2001) and 3TC (Olivero et al., unpublished). All samples were positive for AZT-DNA incorporation, with levels ranging from 2 to 152 molecules of AZT/106 nucleotides (Table 1). 3TC-DNA incorporation was measurable in only four infants, with values ranging from 8 to 13 molecules of 3TC/106 nucleotides (Table 1). A comparison of 3TC-DNA incorporation with AZT-DNA incorporation in the same infants showed a positive correlation between the levels of these two markers of drug exposure, despite the lower levels of 3TC-DNA incorporation. In contrast, in maternal leukocytes, AZT-DNA incorporation was measurable in seven of nine mothers while 3TC-DNA incorporation was only detected in one mother. Similar to the transplacental monkey model, these preliminary data provide evidence that transplacental exposure to both AZT and 3TC, although more effective in the prevention of HIV-1 vertical transmission, may be more genotoxic than transplacental exposure to AZT alone.
Mutagenesis in Combivir-exposed infants
In infant blood samples from the observational cohort study (see above), mutations in the HPRT and GPA genes were evaluated after Combivir prophylaxis. The HPRT(Albertini et al., 1990) and GPA(Grant and Bigbee, 1993) assays characterize somatic mutations occurring during the development of T-cells and red blood cells, respectively.
Inactivation of the HPRT gene by a genotoxic agent allows quantification of mutant T-cells as well as elucidation of a mutation spectrum. In infants from the observational cohort study (Walker et al., 2001), HPRT mutant frequencies (Mfs), measured in cord blood T-cells, were increased significantly only in children exposed in utero to Combivir and not in children exposed to AZT alone (O'Neill et al., 2001). The mean value for HPRT Mf in cord blood T-cells of Combivir-exposed infants (2.5 x 10-6) is about 2-fold higher than that observed in unexposed infants (1.2 x 10-6, P < 0.005) (O'Neill et al., 2001) and approximates the AZT-induced HPRT Mfs observed in spleen T-cells from in utero-exposed CD-1-mice (Table 1). These data provide compelling evidence that the net mutagenic effects of in utero Combivir exposure in infants, low daily doses for months of gestation, are similar to those in AZT-exposed mice given high daily doses for the last approximately 33% of an approximately 21-day gestation (Table 1). In addition, the average HPRT Mf in infants exposed to AZT-3TC was at a level that one expects to see in children entering adolescence (Finette et al., 1994).
The GPA assay is based upon allele-specific loss of the gene coding for the major erythroid cell-specific sialoglycoprotein (Grant and Bigbee, 1993). In heterozygous GPAM/N individuals (approximately 50% of the population), the "M" and "N" alleles are co-dominantly expressed. The GPA assay measures loss of one allele, as well as the frequency of duplication variants [Grant and Bigbee, 1993] and [Langlois et al., 1986]. In cord red blood cells from infants born to mothers taking Combivir (Table 1), the mean frequency of GPA N/N variants (Vfs) at birth was 5.0 ± 3.7 x 10-6 compared to a mean frequency of 2.2 ± 1.5 x 10-6 in unexposed infants [Bigbee et al., 2001] and [Walker and Poirier, 2002]. In addition, the average-induced treatment effect of Combivir exposure was nearly twice that observed in children exposed to AZT as the only NRTI. In peripheral red blood cells of mothers taking Combivir, the GPA N/N Vfs were also significantly elevated, with mean values of 7.5 ± 4.8 x 10-6 compared to 6.3 ± 8.7 x 10-6 observed in uninfected and unexposed mothers (P = 0.05).
Summary and conclusions
The widespread and striking success of the NRTIs (AZT and 3TC) in the chemoprophylaxis of vertical HIV-1 transmission has led to the birth of thousands of virus-free children who might otherwise have been born infected. However, the fact that this treatment is imperative and that these children have been exposed in utero to genotoxic agents provides a rationale for the implementation of transplacental studies in various experimental models.
Studies conducted in two strains of mouse and presented here have demonstrated that: (1) AZT is a moderately strong transplacental carcinogen in offspring at 1 and 2 years after birth; (2) at birth, AZT can be found incorporated into the DNA of multiple fetal organs; (3) the mutagenic consequences of AZT-DNA damage are dose related and significant; and (4) some organs have shortened telomeres. In many different types of experimental models, causative associations among DNA damage, mutagenesis, and tumor formation have been demonstrated, and the dose-response relationships shown here for all three parameters suggest that AZT exposure is no exception. The mouse studies make a strong case for the potential carcinogenicity of transplacental NRTI exposure in transplacentally exposed infants, but the mouse pregnancy is very different from the human pregnancy.
The transplacental genotoxicity studies in primate models show that both short-term and long-term in utero exposures to AZT and 3TC induce NRTI-DNA damage in most fetal organs examined at birth. In addition, the combination of two nucleoside analogs (AZT plus 3TC) enhances the burden of genotoxicity compared to transplacental AZT exposure alone. These studies provide clear indication that similar events could occur in human pregnancy, and imply that evidence occurring in exposed children should not be ignored.
Finally, whereas it is not feasible to monitor most human infant organs for genotoxicity, it has been possible to use cord blood to assess both NRTI-DNA damage and mutagenesis. In a total of more than 50 infant cord blood DNA samples generated from multiple different institutions, the observed level of AZT-DNA damage was similar to that observed in transplacentally exposed mice and monkeys. In a small pilot study of 13 infants exposed to Combivir, the incorporation of 3TC into DNA was 50- to 100-fold lower than AZT-DNA incorporation in the same samples. However, AZT-DNA damage appears to be higher in infants exposed to Combivir compared to infants exposed to AZT alone, suggesting that Combivir induces a higher level of genotoxicity in infants than AZT alone.
Consistent with the observed NRTI-DNA damage is the significant induction of HPRT Mfs and GPA Vfs in cord blood from Combivir-exposed infants compared to unexposed infants. The mutation level at the HPRT locus in newborn infants exposed in utero to Combivir is in the same range as that seen for young adults in the general population and similar to that found in mice exposed to tumorigenic doses of AZT. The increase in GPA N/N variants in the same Combivir-exposed infants was almost 3-fold higher than in the unexposed population. These increases in mutation frequency demonstrate the unequivocal genotoxic consequences of anti-retroviral NRTI exposures and are consistent with the observed NRTI-DNA incorporation.
The most striking interspecies parallels observed here are in the levels of NRTI-DNA damage and HPRT mutagenesis. The AZT-DNA levels found in infants are equal to those found in mice at risk for tumor formation, and the drug-induced HPRT Mfs were similar in magnitude in both species. The monkey studies show that levels of AZT-DNA damage similar to those in human cord blood can be found in many fetal organs, suggesting that human infant organs may also be vulnerable to NRTI-induced genotoxic damage. The similarities in mutagenic responses following very different in utero-exposure regimens for human infants (low AZT dose and long treatment duration) and rodents (high AZT dose and short treatment duration) can be explained partly by the fact that host cell mutations resulting from AZT-DNA incorporation can accumulate over time during intrauterine life. Given the risk-benefit ratio, these highly successful drugs will continue to be used for the prevention of vertical viral transmission; however, evidence of genotoxicity in mouse and monkey models and in the infants themselves would suggest that exposed children should be followed well past adolescence for early detection of potential cancer hazard.
| |
| |
| |
|
|
|
