Nucleoside and nucleobase analogs in cancer treatment: not only sapacitabine, but also gemcitabine

Abstract

Nucleoside analogs are widely used for treatment of various malignancies. Benchmark drugs are cytarabine for acute myeloid leukemia and gemcitabine for pancreatic and lung cancer. Sapacitabine is a novel cytidine analog currently in development. This editorial focuses on the potential of new nucleoside analogs and on novel possibilities of gemcitabine. Gemcitabine is a nucleoside analog with many faces, which shows a remarkable activity in a variety of cancers, most likely because it has a unique metabolism, a so-called self-potentiation. Gemcitabine is taken up by nucleoside transporters, is activated by deoxycytidine kinase and incorporated into both RNA and DNA. Inhibition of ribonucleotide reductase and dCMP deaminase enhances its activation, while cytidine deaminase converts gemcitabine to its presumably inactive metabolite 2′,2′-difluorodeoxyuridine, which in its nucleotide form may inhibit thymidylate synthase. Gemcitabine is widely used in combination, predominantly with a platinum analog, with other combinations less frequently used or being explored. Standard administration of gemcitabine is with a 30-min weekly infusion at 1000 mg/m², but alternatives are being explored such as prodrugs (e.g., CO-1.01, which can bypass transport deficiency), the fixed-dose rate infusion (10 mg/m²/min), and local routes of administration by a 24-h hepatic artery infusion, by instillation in the bladder or by intraperitoneal administration to treat advanced ovarian cancer. Other alternatives for combinations of gemcitabine in ovarian cancer consist of increasing the inhibition of ribonucleotide reductase with triapine or hydroxyurea. Gemcitabine's action on signaling also provides a rational concept for combination with signal transduction pathways.

Keywords::

• cancer
• nucleoside
• nucleoside base analogs
• sapacitabine

The article ‘Sapacitabine for cancer' serves as a reminder that much remains to be learned about this class of agents. Their tabular description of ‘Diversity of mechanisms of action and clinical activities of nucleoside and nucleoside base analogs' represents a wake-up call. Not only do we have incomplete information on the mechanism of action for many of these drugs, but most remarkable is the diversity of mechanisms and disease targets. Moreover, this listing must be regarded as incomplete, considering mechanisms of action and disease types. For example, several oral prodrugs such as capecitabine, UFT and S-1 have a different activity profile [1], while 5-fluoro-2′-deoxyuridine because of both pharmacologic and pharmacodynamic (PD) reasons seems to have selective activity via hepatic artery infusion (HAI) [2]. Furthermore, depending on dose-scheduling and other unknown factors, the extent of the RNA incorporation varies with unclear implication on drug effects. While studies overwhelmingly supports that the incorporation of 5-fluorouracil (5-FU) into RNA is associated with toxicity [3], the impact of gemcitabine incorporation into RNA is unknown [4]. To add to the complexity of factors influencing specific drug actions, prodrugs of cytarabine and gemcitabine, CP-4055 and CP-4126 (now CO-1.01), respectively, may bypass transport and seem sequestered in intracellular organelles [5]. Finally, a nucleoside–alkylating moiety complex, bendamustine, after years of clinical development, has recently emerged as a drug with a very favorable therapeutic index against many lymphoid malignancies. What this ‘rediscovery' tells us, is that drug developers have neglected this area for some time. This Editorial expands on some newer aspects of gemcitabine's development, to encourage further research on antimetabolites for solid tumors and not confined to sapacitabine.

Gemcitabine's mechanism of action in the Expert Opinion paper is listed as causing inhibition of ribonucleotide reductase (RNR) and DNA replication. This list is incomplete: although many antimetabolites have multiple mechanisms of action, gemcitabine has a unique mode of self-potentiation. Its uptake via the equilibrative nucleoside transporters (ENT1-4), the role of activation by deoxycytidine kinase (dCK) and inactivation via dCMP deaminase (dCMPD) play a key role in this self-potentiation (Figure 1) [6-8], which is possibly responsible for the enhanced selectivity of gemcitabine by preferential intracellular retention and accumulation of its di- and triphosphates in solid tumor cell lines and tumor samples [9]. The self-potentiation occurs as follows: after phosphorylation dFdCTP competes with the natural substrate, dCTP, to be incorporated into DNA while dFdCDP inhibits RNR, the enzyme responsible for producing the four key deoxynucleotides required for DNA synthesis and repair, leading to blockade of these cellular functions [7]. Moreover, the diminished dCTP level does not exert its feedback inhibition on dCK, with the consequent increase of gemcitabine phosphorylation. Also, since dFdCTP negatively regulates dCMPD, dFdCMP is not only not deaminated but is further phosphorylated contributing to enhance dFdCTP and dFdCDP concentrations.

Figure 1. Self-potentiation of gemcitabine. hNTs (hENT and hCNT) transports dFdC into the cell. dFdC is intracellularly phosphorylated by dCK to its monophosphate dFdCMP [1] and subsequently into its active dFdCDP and dFdCTP metabolites by ribonucleotide mono- and diphosphate kinases, respectively. DNA polymerase (8) catalyzes the incorporation of dFdCTP into DNA, competing with dCTP. dFdCDP inhibits RNR (5), which inhibits the conversion of dinucleotides to deoxynucleotides, including CDP to dCDP and inhibits synthesis of dCTP. The feedback inhibition of dCK by dCTP is reduced, leading to an increase in the phosphorylation of dFdCTP. This will elevate the intracellular dFdCTP/dCTP ratio and enhance incorporation of dFdCTP into DNA.CDA deaminates dFdC to dFdU (2), which can occur in and outside the cell. dFdCMP can be dephosphorylated to dFdC by 5′-NT (4) and be deaminated to dFdUMP by dCMPD (3). dFdCTP can inhibit dCMPD, which stimulates its own formation. dFdCMP can also inhibit CTP synthetase (6) leading to a reduction in CTP, which has a dual effect, less competition of CTP with dFdCTP for incorporation into RNA (9), while the decrease in CTP will also decrease CDP, leading to an additional decrease in the reduction to dCDP. Finally, the deaminated product dFddUMP can inhibit thymidylate synthase (7), which will decrease dTTP pools but increase dUTP pools, leading to additional DNA damage. dFdUMP can also be formed from intra- and extracellularly produced dFdU, which can be taken up by hNTs, and subsequently phosphorylated to dFdUMP. This phosphorylation may be catalyzed by dCK, since dCK-deficient cells are also resistant to dFdU. However, dFdU is also a substrate for the mitochondrial thymidine kinase 2.

Moreover, dFdCTP is also an inhibitor of CTP synthetase, thus decreasing CTP synthesis as well, which will further deplete dCTP and as a consequence more dFdCTP will be incorporated into DNA. Furthermore, in contrast to other deoxynucleoside analogs, the presence of two fluorine atoms in the sugar apparently gives gemcitabine some ribonucleoside properties enabling its incorporation into RNA, which is further enhanced by CTP depletion [10,11]. Finally, intracellular formation of dFdUMP has been demonstrated by several groups [6,12]; dFdUMP can inhibit thymidylate synthase [4]. The implication of this finding is not clear, but this phenomenon leads to an increase in dUTP synthesis which may be misincorporated into DNA, adding to the DNA damage already caused by dFdCTP incorporation [12]. Plunkett's group demonstrated that dFdCTP incorporation led to a so-called ‘masked chain termination', which means that after incorporation into DNA, one more deoxynucleotide may be incorporated [13]. However, in cellular systems, incorporation into DNA seems to continue over a prolonged period [10]. The dFdCTP incorporation into DNA and the deoxynucleotide depletion seem important features in the interaction with other drugs, such as cisplatin, since this decreases DNA repair enhancing the cisplatin effect [14]. Furthermore, these features also lead to decreased repair of radiation-induced damage, making gemcitabine one of the most potent radiosensitizers studied clinically [15].

Mechanistic considerations have not been sufficiently utilized in the clinical development of gemcitabine. In the 1990s, Plunkett and clinical co-workers at MD Anderson worked out relevant PD parameters of gemcitabine activation in leukemia patients leading to testing the ‘fixed-dose-rate' (i.e., 10 mg/m²/min) administration of its most ‘tumor-selective' weekly schedule. With a dose of 1500 mg/m² this led to an infusion of 150 min, during which dFdCTP continued to accumulate in lymphocytes (as a surrogate tissue) with a twofold higher C_max compared with the 30-min schedule [16]. While initial clinical studies supported this hypothesis in patients with pancreatic cancer [16] and uterine sarcoma [17], a subsequent larger Phase III study in pancreatic cancer failed to meet its end point [18]. Nevertheless, an extensive PD study of these two schedules supports that a ‘carryover-effect' of one dose to the subsequent one is stronger when the fixed dose rate is utilized [19]. With the standard 30-min weekly infusion no carry-over of plasma gemcitabine or lymphocyte dFdCTP was observed, although dFdU was still present in the low µM range after 1 week [20]. Based on the cell-cycle properties of gemcitabine and experience with other antimetabolites it can be argued whether this infusion was long enough. In an animal study, the most efficacious schedule was a weekly 24-h infusion [21]. This was further investigated in order to determine whether gemcitabine would be a suitable agent for therapy of localized hepatic disease. Gemcitabine given as HAI was well tolerated [22]. At the recommended Phase II dose of 135 mg/m², the hepatic extraction rate was 0.75 which is comparable with that of 5-fluoro-2′-deoxyuridine; plasma levels at i.v. administration were 263 nM compared with 80 nM at HAI. This study underlines the potential of gemcitabine for localized treatment, not only as HAI, but also for bladder instillation or intraperitoneal infusions. Since gemcitabine is active in advanced bladder cancer (with cisplatin), a bladder instillation for localized treatment is a logical step and the first clinical studies clearly demonstrated the feasibility [23]. During bladder instillation no systemic gemcitabine or its metabolite dFdU were measured. Another form of localized treatment is intraperitoneal administration, which has been studied in Phase I trials for gemcitabine as a single agent in patients who had chemotherapeutically unresponsive malignancies [24] or in combination with cisplatin [25] in patients with microscopic or macroscopic disease after initial surgical cytoreduction and adjuvant platinum-based chemotherapy. The mean peritoneal advantages (AUC-ip/AUC-plasma) were 847 and 759, respectively, confirming an increase in local dose intensity with acceptable side effects. Hence, these studies demonstrate that using appropriate pharmacokinetics (PK) and PD gemcitabine administration can further be optimized, possibly with disease-specific adjustments, based on pharmacogenetic (PG) features. An example of relevant new PG information includes the prolonged progression-free survival (PFS) of pancreatic cancer associated with a low miR-21 expression, which regulates the PI3K pathway [26]. However, miR-21 does not seem to play the same role in lung cancer [27]. Another example includes decreased cytidine deaminase activity (CDA), which increases toxicity [28]. An integrated PK/PD/PG approach may improve current therapeutic expectations, especially when the drug is used to treat more sensitive tumors such as ovarian cancer.

Recurrent ovarian cancer is a major target of gemcitabine treatment, whether given by itself in platinum-resistant disease or in combination with carboplatin in the platinum-sensitive population. More recently, a Phase III placebo-controlled study sponsored by Genentech, (OCEANS, Dic, South San Francisco, CA, USA) [29] documented a hazard ratio (HR) of 0.48 (95% confidence interval 0.39 – 61) for gemcitabine + carboplatin + bevacizumab relative to the cytotoxic doublet + placebo, and this HR compares favorably with other control or experimental regimens (Table 1). It is unfortunate that there has been little follow-up using PG to improve the PD of gemcitabine in these combinations, or suggestions that platinums given together with gemcitabine alter its PD [36] or alter platinum resistance [37]. Future studies should aim to bypass potential resistance mechanisms to gemcitabine, which consists predominantly of impaired uptake, decreased activation, increased inactivation by CDA or increased RNR [4,28]. The novel gemcitabine prodrug CO-1.01, with an elaidic acid attached to the 5′-carboxy position, has shown similar activity in cells with inhibited transporter function and also showed oral activity against xenografts [1]. CP-4126 seems sequestered in cell lines leading to a longer intracellular retention and enhanced formation and retention of dFdCTP [5], hence bypassing multiple resistance parameters. In advanced pancreatic cancer the drug is currently compared with gemcitabine, using companion diagnostics for ENT. A similar prodrug for ara-C, CP-4055 [38] has recently shown activity in acute myeloid leukemia (AML) independent of the ENT expression [39].

Table 1. Outcome in patients with platinum-sensitive ovarian cancer recurrence.

Eligibility (months)	Control regimen	Patient number	Comparator	Comments outcome (months)
Platinum sensitive (> 6) [30]	Carboplatin + epirubicin	190	Carboplatin	Powered for response differences; OS 17 vs 15
Platinum sensitive (≥ 12) [31]	Cisplatin + doxorubicin + cyclophosphamide	97	Paclitaxel	PFS 15.7 vs 9; OS 34.7 vs 25.8
Platinum sensitive (> 6) [32]	Carboplatin + gemcitabine	356	Carboplatin	PFS 8.6 vs 5.8; OS 18 vs 17
Platinum sensitive [29]	Carboplatin + gemcitabine + placebo	484	Carboplatin + gemcitabine + bevacizumab	HR PFS: 0.48 (95% CI 39.0 – 61.9%, p < 0.0001)
Platinum sensitive (> 6) [33]	Cisplatin or carboplatin + paclitaxel	802	Single or non-taxane + platinums	PFS 11 vs 9; OS 24 vs 19
Platinum sensitive (> 6) [34]	Carboplatin + PLD	976	Carboplatin + paclitaxel	Median PFS 11.3 vs 9.4; median OS not assessed
Platinum partially sensitive (> 6 & < 12) [35]	PLD	672	Trabectedin + PLD	Subset analysis

Adapted from PDQ.

CI: Confidence interval; HR: Hazard ratio; OS: Overall survival; PFS: Progression-free survival; PLD: Pegylated liposomal doxorubicin.

Since RNR overexpression is a potential resistance mechanism, RNR inhibition is an attractive way to enhance gemcitabine activity, which should be exploited further in the clinic. A sequence-dependent interaction between the RNR inhibitor triapine [40] or hydroxyurea [41] and gemcitabine has been demonstrated and linked to a transient lowering of dCTP pools favoring dFdCTP formation and its incorporation into DNA [40]. Clinical studies of such PD-based doublet (either triapine or hydroxyurea) did not advance beyond Phase I [42]. We have added pretreatment with hydroxyurea in some gemcitabine-responding patients to extend waning responses (reflected by rises in their tumor markers) in patients with ovarian cancer, clinically obvious evidence of modulation ensued (such as sudden development of total alopecia and in some instances marker improvements; ([43] and I Diaz, unpublished). Unfortunately, a randomized Phase II study of hydroxyurea → gemcitabine versus gemcitabine alone in platinum-resistant recurrences of ovarian cancer is difficult to realize in the current investigator-initiated clinical trials environment that is focused exclusively on ‘novel targets'. Hopefully, knowledge of signaling pathways will also further enhance gemcitabine's activity when this agent is combined with some tyrosine and/or serine–threonine kinases inhibitors [44].

Taken within this broad context, the renewed interest generated by sapacitabine may catalyze additional questions in these neglected areas of therapeutic research. Nucleoside analogs, in particular gemcitabine, are likely to be more active when the administration route is changed, while its action on signaling provides a rational concept for combination with anti-signaling targeted drugs.

Declaration of interest

The authors state no conflict of interest and have received no payment in preparation of this manuscript.

Notes