Nitrogen mustards

Soon after the initial report of clinical activity of mechlorethamine in 1946,^{4, 5} many nitrogen mustards were evaluated for their antitumor potential through structure–activity relationships. Of the handful that advanced to clinical trials, a few were approved by the FDA and are still in use today. In addition to mechlorethamine, these include cyclophosphamide, ifosfamide, melphalan, chlorambucil, and bendamustine, which are shown in Figure 2. Also shown in this figure is 4-hydroperoxycyclophosphamide, which is the prodrug form of active 4-hydroxycyclophosphamide.

The sequence of events and specificity of the reaction between nitrogen mustards and DNA have been well described by Colvin et al.¹⁰ The bis-chloroethyl group of these agents is critically important for the predominant alkylation of N7-guanine in DNA, although alkylation can occur but infrequently at O6-guanine and at N3- and N7-adenine. The alkylation is sequence specific and requires guanine in the 5′-GXC-3′ sequence (where X = any nucleotide). For interaction, the chloroethyl moiety cyclizes to form the highly reactive aziridinium ion intermediate that reacts readily with electron-rich sites, as shown with N7-guanine in Figure 3. It is possible that the aziridinium ion may first rearrange to the reactive carbonium ion intermediate for the alkylation reaction, but this remains uncertain. As there are two chloroethyl groups present, this results in sequential bifunctional alkylation to crosslink DNA. As indicated above, interstrand crosslinks are preferred by nitrogen mustards, with 1,3-crosslinks being the favored interaction. However, DNA–DNA intrastrand, DNA–protein, and DNA–glutathione crosslinks are also formed, but lack cytotoxic potential. Interstrand DNA crosslinks are the critical lesions, and according to Roberts et al.,¹¹ bifunctional interstrand DNA crosslinks are about 100-fold more effective at killing cells than monofunctional DNA adducts.

All nitrogen mustards are neutral and must be activated to interact with DNA target sites. Mechlorethamine, for instance, is activated spontaneously at physiological pH, but the rapid rate of activation is a major cause of much of the unwanted side effects. Chemical modification of the molecule, however, has played a crucial role not only to stabilize nitrogen mustard congeners, but also to influence their physicochemical properties that regulate transport, distribution, and reactivity of the molecule in order to enhance clinical utility.¹² Cyclophosphamide, which has an oxazaphosphorine ring, was selected from over a 1000 oxazaphosphorine candidates, and its chemical inertness and dependence on metabolic activation were considered as desirable characteristics.^{13, 14} The primary site for this activation is the liver, where the P450-mediated mixed function oxidase reaction initiates a complex series of chemical transformation events^15–18 that is depicted in part in Figure 4. The 4-hydroxycyclophosphamide is the most significant metabolite, which is relatively nonpolar and can readily distribute throughout the body, including tumor cells. This metabolite can be formed directly from the prodrug 4-hydroperoxycyclophosphamide (Figure 1), which has demonstrated utility in purging tumor cells in bone marrow aspirates of patients undergoing autologous bone marrow transplant.¹⁹ At physiologic pH, a small amount of the 4-hydroxycyclophosphamide exists in equilibrium with aldophosphamide, which is relatively unstable and spontaneously decomposes to form the highly reactive phosphoramide mustard and then nornitrogen mustard.

An important by-product of the metabolism of cyclophosphamide, resulting from the conversion of aldophosphamide to phosphoramide mustard (Figure 4), is acrolein,¹⁸ which induces hemorrhagic cystitis as a serious side effect,²⁰ but whether it has antitumor activity is of interest. Although a study with the model compound didechloro-4-hydroperoxycyclophosphamide indicated that the spontaneously released acrolein lacked cytotoxicity,²¹ other studies have demonstrated that acrolein indeed has antitumor and carcinogenic activities but is inactivated by O6-alkylguanine-alkyltransferase.^22–24 Aldehyde dehydrogenase can also play a significant role in moderating activity by converting aldophosphamide to the inactive carboxyphosphamide metabolite, which accounts for about 80% of the cyclophosphamide dose excreted in the urine. Not surprisingly, this enzyme can confer resistance to cyclophosphamide in tumor cells when ectopically overexpressed.²⁵ Thus, high levels of aldehyde dehydrogenase in the liver, early hematopoietic cells, and intestinal stem and mucosal absorptive cells are consistent with reduced hepatic, gastrointestinal (GI), and hematopoietic toxicities associated with cyclophosphamide as compared to other alkylating agents. Urinary excretion of metabolites can be compromised at high doses of cyclophosphamide by an antidiuretic effect owing to sodium loss and water reabsorption by renal tubules, with resultant edema.²⁶ The electrolyte imbalance may also contribute to the reported dose-limiting and fatal cardiotoxicity of high-dose chemotherapy.²⁷ This limitation has been recognized and doses are now monitored clinically to prevent this lesion from becoming serious.

Ifosfamide (Figure 2) is a structural isomer of cyclophosphamide that is important in the treatment of testicular cancer and sarcomas. Like cyclophosphamide, it also requires activation by hepatic mixed function oxidase,¹³ but the low rate of metabolic activation by 4-hydroxylation and conversion to the corresponding active isophosphoramide mustard decreases its potency approximately fourfold.^{28, 29} On the other hand, increase in the dose to compensate for the lower potency leads to a greater cumulative production of acrolein, which increases the incidence of urothelial toxicity that requires concurrent administration of the thiol-based mesna to inactivate the toxic by-product. Renal and bladder toxicity, as well as neurotoxicity, of ifosfamide have also been ascribed to the formation of chloroacetaldehyde,³⁰ the product of oxidation of the chloroethyl side chain (Figure 4) that is generated in greater quantities with ifosfamide than with cyclophosphamide (∼50% vs ∼10% of the dose). The primary metabolite of ifosfamide, aldoifosfamide, is also a substrate for aldehyde dehydrogenase, and this is consistent with low toxicity of the drug to the bone marrow and the GI tract due, as mentioned previously with cyclophosphamide, to the presence of high levels of the enzyme in these organs.

Melphalan, chlorambucil, and bendamustine do not have the oxazaphosphorine ring of cyclophosphamide and ifosfamide in their structure, and so do not require metabolic activation, although metabolism of these agents does occur in the liver, but only for its disposition. These agents instead have an unsaturated electron-withdrawing aromatic ring (Figure 2) that moderates activation of the bis-chloroethyl group and formation of the aziridinium ion in order to minimize side effects. Owing to its phenylalanine-based structure, melphalan can be transported actively into cells and across the blood–brain barrier.^31–33 These nitrogen mustards have found important applications against several cancers: melphalan against multiple myeloma, breast cancer, and ovarian carcinoma³⁴; chlorambucil against ovarian cancer, chronic lymphocytic leukemia (CLL), and Hodgkin’s and non-Hodgkin’s lymphoma^35–37; and bendamustine against CLL and lymphomas.³⁸ Owing to its improved tolerance, bendamustine is reported to be more cost-effective as a therapeutic drug than the structurally similar chlorambucil.³⁹

Aziridines

Also known as ethyleneimines, the aziridines alkylate through the three-membered heterocyclic aziridine ring. Aziridine family of drugs includes thiotepa (N,N′,N″-triethylenethiophosphoramide), mitomycin C, and triethylenemelamine, which are shown in Figure 5. Also shown is hexamethylmelamine, which is similar to triethylenemelamine, but lacks the classical aziridine ring. Of these, thiotepa and mitomycin C are of clinical interest for a number of diseases, including refractory osteosarcoma and bladder cancer, respectively.^{40, 41} High-dose thiotepa has also been used in drug combination regimens with⁴² or without⁴³ autologous stem cell support for breast cancer. Hexamethylmelamine, on the other hand, has found application in salvage therapy for recurrent ovarian cancer.⁴⁴

The aziridine ring in the chemical structure of thiotepa is not charged, which makes it less reactive with electron-rich centers in RNA (ribonucleic acid), DNA, protein, and thiols (such as glutathione) than the structurally similar aziridinium ring formed with nitrogen mustards. In DNA, thiotepa alkylates sites in all four nucleotides, particularly at O2-cytosine, N1-thymine, N1-, N2-, and N7-adenine, and N1-, N7-, and O6-guanine.⁴⁵ However, N7-guanine remains the preferential target, as with other alkylating agents, and results in guanine–guanine (GG) and adenine–guanine (AG) interstrand crosslinks, primarily 1,2-crosslinks.⁴⁶ These crosslinks are formed sequentially; the first reaction at N7-guanine leads to the monofunctional adduct, and the second reaction with a nearby guanine results in the crosslink, as shown by pathway A in Figure 6. Crosslinking may also proceed via metabolism of thiotepa to tepa (sulfur replaced by oxygen), followed by opening of the aziridine ring and conversion to the chloroethyl-containing crosslinking species, as found in nitrogen mustards.⁴⁷ Alternatively, the aziridine ring is cleaved to mediate alkylation of N7-guanine and forms a monofunctional DNA adduct (Figure 6, pathway B), which induces DNA single-strand breaks and then cell death.^{48, 49}

Unlike thiotepa, interaction of mitomycin C with DNA requires metabolic reduction to leucoaziridinomitosene for the formation of the monofunctional adduct.⁵⁰ Spontaneous intramolecular rearrangement and elimination of the carbamate group at C10 of mitomycin C results in a second alkylation reaction with the opposite strand of DNA to induce cytotoxic 1,2-GG interstrand crosslink lesions (Figure 7). Interestingly, alkylation at the N2-guanine site of DNA is preferred by mitomycin C, particularly if guanine is present in the 5′-purine-CG-pyridine-3′ sequence in both strands of the duplex.⁸

On the basis of the presence of the aziridine ring in the triethylenemelamine structure, alkylation by this agent likely occurs in the same manner as that described for thiotepa. However, with hexamethylmelamine, metabolic activation in the liver is required.^{51, 52} The metabolism generates a carbinolamine intermediate, N-methylol-pentamethylmelamine, which rearranges to the reactive iminium ion as the species responsible for alkylation of guanine in DNA (Figure 8). As hexamethylmelamine metabolism can also result in the dicarbinolamine product N,N′-dimethylol-tetramethylmelamine,⁵³ bifunctional interstrand DNA crosslink is the likely outcome, particularly as the tricarbinolamine-based prodrug of hexamethylmelamine, trimelamol, readily forms such crosslinks.⁵⁴ It is noteworthy that trimelamol has been investigated in clinical trials,⁵⁵ but the development did not continue, owing likely to limited clinical activity⁵⁶ and/or drug formulation/stability issues.⁵⁷

Hexitol epoxides

The di-epoxide 1,2:5,6-dianhydrogalactitol (DAG) and its prodrug 1,2-dibromodulcitol (DBD), like the aziridines, also alkylate via a strained tricyclic ring (Figure 9). Chemical interaction between DAG and DNA results in alkylation at N7-guanine,⁵⁸ a site commonly preferred by several other alkylating agents. In cell-free, cellular, and in vivo rodent systems, N7-monoguanyl and N7-diguanyl derivatives are detected as two major products that represent monofunctional and interstrand crosslinked DNA adducts, respectively.^58–61

Both DBD and DAG have shown antitumor properties that were similar to alkylating agents in some experimental antitumor models, but different in others. Lack of cross-resistance with DAG, for instance, was observed in a nitrosourea (BCNU)-resistant L1210 leukemia model.⁶² Of the two hexitols, DAG has greater chemical stability and better preclinical antitumor activity profile.^{63, 64} Nevertheless, both DBD and DAG entered clinical trials in 1970s and demonstrated some activity in GI and lung cancer.^{65, 66} Interestingly, these hexitols cross the blood–brain barrier in patients,^67–69 and follow-up clinical trials found a 44% response rate with DAG in brain cancer.⁶⁶ DAG was recently granted orphan designation in the United States and Europe for brain cancer, and clinical trials are currently underway in the United States (ClinicalTrials.gov Identifier: NCT01478178). In China, however, it has been marketed for the treatment of chronic myelogenous leukemia (CML) and lung cancer for over 20 years.

Alkyl sulfonates

Busulfan and the analog hepsulfam are the best known alkyl sulfonates, which are symmetrical and have either two sulfonates (busulfan) or sulfamates (hepsulfam) separated by a linear alkyl chain. Busulfan was approved in 1999 as a standard of care for CML, but its clinical application subsequently diminished by first the introduction of the less-toxic hydroxyurea and then by the targeted therapeutic Gleevec. However, busulfan is still used in the allogeneic stem cell or bone marrow transplantation setting.^{70, 71} Although hepsulfam was more potent than busulfan in human tumor model systems,^{72, 73} it has not been effective in clinical trials.

The sulfonate or the sulfamate is the critical group responsible for the alkylation reaction at N7-guanine to form GG interstrand crosslink,^{74, 75} as shown with busulfan in Figure 10. However, this crosslink formation by busulfan has been disputed⁷⁶ and the alternative GA intrastrand crosslink has been proposed.⁷⁷ Significantly, where any crosslinking has been reported, this has been correlated to cytotoxicity.

Nitrosoureas

Much of the structure–activity understanding, and the basis for clinical development of 2-chloroethylnitrosoureas (CENUs) as derivatives of urea, came from the work of Montgomery and his colleagues.^78–80 With the ability to cross the blood–brain barrier, four analogs entered the clinic,^{81, 82} and these are shown in Figure 11. BCNU (Carmustine) and CCNU (Lomustine) are effective against several types of cancers, including glioma, glioblastoma multiforme, GI tumors, breast, and multiple myeloma, with streptozotocin (Zanosar) effective against cancers of the pancreatic islet cell, owing primarily to its uptake via the glucose transport protein GLUT2 that is present at relatively high levels in these cells.⁸³ Absence of any advantage over other nitrosoureas has halted the use of methyl-CCNU (Semustine). The limitation in clinical utility of nitrosoureas is ascribed to their severe side effects, including hematopoietic and renal toxicities, with streptozotocin in particular demonstrating substantial toxicity to β cells of the normal pancreatic islets owing to the GLUT2 transporter.⁸⁴ The development of novel nitrosoureas, therefore, continues, and a third generation analog, fotemustine (Muphoran) (Figure 11), has shown activity in brain cancer and metastatic melanoma, with or without brain involvement, and has received approval in Australia, Brazil, China, and Europe, but not as yet in the United States.^{85, 86}

It is widely acknowledged that CENUs as bifunctional DNA interstrand crosslinkers are unstable, as exemplified by a short half-life of about 50 min for BCNU⁸²; the several products of decomposition are shown in Figure 12. An unusual feature of some CENUs is that they can generate isocyanate, which carbomylates the ϵ-amino group of lysine in proteins, including histone and other nuclear proteins, but there is disagreement whether the carbomylated-protein contributes to cytotoxicity.^{79, 87, 88} A second unusual feature is that CENUs can alkylate at O6-guanine in DNA, likely through formation of the 2-chloroethyl diazene hydroxide and the analogous 2-hydroxyethyl diazene hydroxide species (Figure 12).^{81, 82} Rearrangement of these species to carbonium ions leads to alkylation at the O6-site of guanine to form monoadducts.⁸⁹ The O6-alkylation reaction is considered important as upregulation of O6-alkylguanine-DNA alkyltransferase, which repairs the O6-adduct, increases resistance and, conversely, inhibition of this enzyme with O6-benzylguanine sensitizes tumor cells to nitrosoureas.^89–91 However, adducts from N7-alkylation of guanine (Figure 12) are normally the predominant species, particularly where this purine is in the middle of a GGG sequence.⁹² These DNA monoadducts form rapidly (often within an hour), but then slowly convert to the cytotoxic N7-diguanyl interstrand crosslink that may take up to 8 h.⁸² It is noteworthy that N3-cytosine–N1-guanyl crosslinks are also formed through cyclization of the O6-chloroethyl-guanine monoadduct to the O6,N1-ethanoguanine pentacyclic intermediate, which reacts with N3-site of cytosine and becomes crosslinked to the N1-site of guanine following opening and rearrangement of the ethanoguanine ring.^{82, 93, 94} There is also a proposal from steric considerations that only the CG crosslink of CENUs is interstrand, with the similarly cytotoxic GG adduct being intrastrand.⁸⁹

Hydrazines

More than 400 hydrazine compounds have been tested preclinically, but only procarbazine has received clinical approval for the treatment of cancers, specifically Hodgkin’s lymphoma, lung cancer, and melanoma. Its activation is poorly understood, but metabolic oxidation to azoxy-procarbazine is an important step. This metabolite rearranges to a reactive methyl diazonium intermediate that forms O6- and predominantly N7-methyl-guanine adducts, which then spontaneously depurinate to induce single-strand breaks.^95–97 DNA–DNA or DNA–protein crosslinks have not been reported. Another hydrazine that had been studied since the early 1970s was hydrazine sulfate, but the clinical activity was equivocal, and the compound never received approval in the United States.

Triazines

The most recognizable triazines are hexamethylmelamine (Altretamine), dacarazine (DTIC-Dome), and temozolomide (Temodar). Of these, hexamethylmelamine is likely a bifunctional alkylator and, therefore, has been grouped above with its closely resembling trimethylmelamine under “aziridines.” Dacarbazine has been used against a variety of cancers, including Hodgkin’s lymphoma, sarcoma, malignant melanoma, and pancreatic islet cell carcinoma. Temozolomide is an imidazotetrazine derivative of dacarbazine that is particularly active against astrocytoma, as well as melanoma. Triazines methylate O6- and N7-guanine,⁹⁸ but, like the hydrazines, require metabolic activation for cytotoxic alkylation and DNA single-strand breaks. For dacarbazine, hepatic activation to [methyl-triazenyl]-imidazole-carboxamide (MTIC) generates the reactive DNA-methylating methyldiazonium ion.⁹⁹ Temozolomide, in contrast, functions as a prodrug of MTIC and spontaneously generates the reactive species.¹⁰⁰

Isoquinoline alkaloids

The two naturally occurring tetrahydroisoquinoline alkaloids of interest are trabectedin (ecteinascidin-743, Yondelis) and Zalypsis. Trabectedin is the first marine-derived antitumor agent to be approved in Europe, Russia, and South Korea for cancer treatment, specifically advanced soft tissue sarcoma and platinum-resistant ovarian cancer.¹⁰¹ Its approval in the United States, however, is pending. The drug initially binds noncovalently to TCG, CGG, AGC, or GGC sequence of one strand in the DNA minor groove, where dehydration of the carbinolamine group generates the iminium intermediate that then alkylates the unusual N2-site of guanine in the opposite strand.¹⁰² Trabectedin-induced monofunctional adducts then recruit the transcriptionally coupled nuclear excision repair (TC-NER) protein Rad13/ERCC5 to the DNA damage site and induce strand breaks and cell death. An additional unusual feature of trabectedin, not found in several bifunctional alkylating agents, is that absence of the p53 tumor suppressor activity increases antitumor activity about threefold.¹⁰³ Zalypsis is essentially a synthetic derivative of trabectedin that has an identical mode of action as the parent molecule.^{104, 105} It is presently undergoing clinical trials.

Cyclophosphamide

A systemic dose of 50–75 mg/kg cyclophosphamide produces plasma levels of up to 400 µmol/L of the intact drug, which then decays with a half-life of 3–10 h.^139–141 Peak plasma concentration of 50–100 µmol/L phosphoramide mustard metabolite was also observed at about 3 h. Interestingly, the AUC of this and the 4-hydroxy-cyclophosphamide metabolite were similar after intravenous and oral administration, and this indicated both routes to be therapeutically effective. However, there is considerable variation in the PK of the drug among patients, but this is likely due to a prior dose of cyclophosphamide autoinducing its own metabolism or to administration of metabolism-inducing drugs, such as phenobarbital.^{109, 142, 143} At conventional doses, however, variations in PK do not significantly affect its toxicity or therapeutic effects.¹⁴⁴ One explanation for this is that the final AUC for the active metabolites is unaffected, which is supported by the demonstration that cyclophosphamide given iv over 90 min yields 4-hydroxycyclophosphamide AUC of 105–110 µmol/L that is similar to an AUC when the dose is infused over 4 days.¹⁴⁵ Although high-dose therapy is an attractive approach in bone marrow transplantation regimens, the benefits can be limited by saturable PK.¹⁴⁶ Urinary excretion of cyclosphosphamide, predominantly as inactive metabolites, is the main route of excretion and accounts for about 60–70% of the dose.^{147, 148} Renal function does not appear to correlate with the toxicity of cyclophosphamide, and this supports drug clearance of active metabolites being determined by spontaneous decomposition, and not renal excretion.

Ifosfamide

The clinical PK data on ifosfamide is limited, but appears to be similar to that of cyclophosphamide. However, levels of the active 4-hydroxy metabolite from ifosfamide are lower than from cyclophosphamide.^149–151 As with cyclophosphamide, ifosfamide is orally bioavailable and can autoinduce its own metabolism.^{141, 152} Although such similarities may make the choice between these two agents difficult, the potential for larger interpatient variation in metabolism of the side chain and the impact of concomitantly administered drugs altering the metabolic pathway may limit the utility of ifosfamide more than that of cyclophosphamide.^153–155

Melphalan

An iv dose of 0.6 mg/kg results in peak melphalan plasma levels of 4–13 µmol/L, which then decays biexponentially, with a rapid α-phase half-life (T½α) of 8 min and a relatively slower β-phase half-life (T½β) of 1.8 h.^{156, 157} At this dose, the mean AUC of 8 µmol h/L of melphalan was achieved and 24-h urinary excretion of melphalan was about 13% of the dose. Similar plasma PK has been observed following high-dose iv melphalan, and as spontaneous degradation of melphalan plays a major role in drug elimination, renal insufficiency was reported not to impact drug clearance.¹⁵⁸ Another study, however, has demonstrated that renal insufficiency reduces drug clearance and increases myelosuppression, which disappears with dose reduction.¹⁵⁹ Melphalan has good oral bioavailability, and conventional doses of 0.15–0.25 mg/kg melphalan given orally resulted in peak plasma levels of up to 0.2–0.6 µmol/L by 1–2 h, which decayed with a half-life of 0.6–3 h.¹⁶⁰ However, absorption after oral dosing can vary, with food reducing drug absorption and high doses subject to saturable absorption kinetics.^161–163

Chlorambucil

An oral dose of 0.6 mg/kg chlorambucil produced peak plasma levels of 2–6 µmol/L of the intact drug an hour later and 2–4 µmol/L of the metabolite phenylacetic acid mustard by 2–4 h.¹⁵⁷ The β-phase half-life of the parent compound was 92 min, with that of the metabolite being 145 min. In another study, PK analysis of an oral daily dose of 0.8–0.9 mg/kg over 4 days indicated that the AUC of about 10 µmol h/L of chlorambucil on day 1 decreased 17% by day 4, with further progressive decreases noted with each 4-week treatment cycle.¹⁶⁴ This suggests that chlorambucil, as with cyclophosphamide and ifosfamide, may also be subject to autoinduction of its metabolism and excretion, although an alternative possibility is that repeat dosing reduces absorption from the gut. A two- to fourfold inter- and intraindividual variation in the PK of oral chlorambucil has also been reported.¹⁶⁵ Interestingly, plasma AUC of the drug or its metabolite phenylacetic acid mustard was unaffected by food intake.¹⁶⁶

Bendamustine

A bolus iv dose of 5 mg/kg bendamustine was eliminated rapidly from the plasma, with T½α of about 10 min and T½β of 36 min, and a resultant AUC of about 28 µmol h/L.¹⁶⁷ The oral bioavailability is good and in the range 0.25–0.94 (mean, 0.57). The drug is metabolized by hepatic mixed function oxidases to N-demethyl- and γ-hydroxy-bendamustine as major metabolites.¹⁶⁸ Glutathione conjugates of the metabolites are also known, and an additional metabolite, dihydroxy-bendamustine, has been observed in patients plasma after an iv dose.¹⁶⁹

Related posts:

Aberrant signaling pathways in cancer Disparities in cancer care Monoclonal serotherapy Neoplasms in acquired immunodeficiency syndrome Chemical carcinogenesis Skeletal complications

Stay updated, free articles. Join our Telegram channel

Join