The Alkylating Chemotherapeutic Temozolomide Induces Metabolic Stress in IDH1-Mutant Cancers and Potentiates NAD⁺ Depletion–Mediated Cytotoxicity

Somatic mutations in the isocitrate dehydrogenase genes IDH1/2 define a class of adult diffuse gliomas with a distinct etiology and natural history (1–6). Molecular correlative analyses of international randomized trial cohorts have suggested that patients with IDH-mutant glioma, including both those with and without chromosome 1p/19q codeletion, gain a survival benefit from treatment with DNA-alkylating chemotherapy (7, 8). As a result of this emerging evidence, chemotherapy is now frequently integrated into the treatment regimen of these patients, even though they typically present with lower grade histology when compared with IDH wild-type gliomas. The oral alkylating agent temozolomide is commonly utilized by clinicians for this treatment, due to its tolerability in the adjuvant setting.

Unfortunately, the vast majority of these cancers still recur after adjuvant or salvage temozolomide treatment. The activities of the O-6 methylguanine DNA methyltransferase (MGMT) repair enzyme (9) and the mismatch repair (MMR) pathway (10) are critical mechanistic determinants of temozolomide-induced cancer cell cytotoxicity (11) and subsequent evasion and resistance to therapy (12–14). Salvage therapeutic strategies for post-temozolomide glioma recurrences are complicated by acquired mutations inactivating the MMR pathway, with the resulting alkylator-induced hypermutation driving a treatment-resistant malignant phenotype (15, 16). Improved chemotherapeutic strategies are needed to secure durable clinical responses in patients with IDH-mutant gliomas.

In addition to gliomas, mutations in IDH1 are found in a diverse spectrum of histopathologic tumor types, including leukemia, chondrosarcoma, cholangiocarcinoma, and a minor fraction of melanomas and breast cancers. Across each of these cancer types, IDH1 mutation is typically found in different background genetic contexts. As a common feature, however, mutant IDH1 drives widespread metabolic alterations in cancer cells (17). These include the production of 2-hydroxyglutarate (2HG; ref. 18), modulation of HIF1α (19), pyruvate dehydrogenase (20), and lactate dehydrogenase (21), as well as altered citric acid cycle flux (22), and depleted steady-state pools of several canonical metabolites, including glutathione (23) and NAD⁺ (24). This altered baseline metabolism results in the exposure of distinct enzymatic targets, including glutaminase (25) and the NAD⁺ biosynthetic enzyme nicotinamide phosphoribosyltransferase (NAMPT; ref. 24), to selective inhibition with small molecules, resulting in genotype-specific metabolic vulnerabilities in IDH-mutant cancer cells.

We hypothesized that study of the metabolic consequences of temozolomide exposure in IDH1-mutant cancers could uncover novel opportunities for therapeutic targeting. Despite the important role of O6-methylguanine adducts in mediating adjuvant temozolomide sensitivity, the majority (>80%) of temozolomide-induced DNA lesions are actually N3-methyladenine and N7-methylguanine adducts. These lesions are rapidly processed by the base excision repair (BER) machinery (26), as opposed to the O6-methylguanine–dependent MGMT and MMR systems. Importantly, the dynamic capacity of BER does not become saturated with these lesions (27), which is why they are not rate-limiting determinants of cytotoxicity in adjuvant temozolomide-treated cancers. Their abundance nevertheless does induce a significant stress response, through PARP, which polymerizes NAD⁺ into poly(ADP-ribose) (PAR) as the molecular repair signal activating recruitment of downstream BER proteins. Recognizing this activated PARP pathway, alongside the sirtuin (SIRT) pathway, is a primary mediator of NAD⁺ consumption in cells (28), we assessed whether chemotherapeutic targeting of these nonredundant NAD⁺ pathways could be exploited in IDH1-mutant cancer cells.

In experiments we describe here, we observed a burst of NAD⁺ consumption associated with PARP activation during the initial time period immediately following temozolomide treatment. In IDH1-mutant cancer cells, this consumption resulted in a transient but critical reduction of the already abnormally lowered basal steady-state levels of NAD⁺, introducing a window of hypervulnerability to NAD⁺ biosynthesis inhibitors. This finding provided a rationale for the therapeutic combination of temozolomide and NAMPT inhibitors, which resulted in improved efficacy when compared with their administration as single agents in an in vivo IDH1-mutant cancer model.

Under IRB-approved protocols, the patient-derived glioma lines used in this study (MGG18, MGG23, MGG85, MGG91, MGG119, MGG152, and MGG171) were obtained from 2008 to 2014 and were cultured in serum-free neural stem cell medium as described previously (29–31). BT142 (IDH1^R132H-mutant anaplastic oligoastrocytoma) line was obtained from ATCC in 2014 and was not further authenticated. UACC257 line (IDH1/2 wild-type melanoma), HT1080 (IDH1R132C), and U87 (IDH1 wild type) lines were authenticated in 2017 by comparison of STR profiles to the ATCC public dataset. They were cryopreserved at passage number 3 or less prior to use for in vitro experiments. Normal human astrocytes (NHA) were obtained from ScienCell in 2014 and cultured in Astrocyte Medium (ScienCell) and were not further authenticated. All standard cell line media were supplemented with 10% FBS and penicillin–streptomycin–amphotericin B.

IDH1 genomic DNA PCR products (Platinum Taq polymerase) spanning coding exons were Sanger sequenced (Beckman Coulter Genomics). To assess MGMT promoter methylation status, methylation-specific PCR on genomic and bisulfite-modified DNA (Qiagen DNeasy Blood & Tissue Kit and EpiTect Bisulfite Kit) was performed in a two-step approach as described previously (32).

To generate IDH1^R132H overexpressing lines, UACC257 cells were transduced with IDH1^R132H lentivirus (ViraPower HiPerform T-Rex Gateway Expression System, Invitrogen), with the details of generation in Supplementary Methods. MGG18-IDH1-R132H cells have been described previously (24). After incubation with tetracycline (1 μg/mL) for 2 months, cells were used for experiments.

To knockdown MSH6, MGG152 or HT1080 cells were infected with lentivirus containing human MSH6 shRNA (#1, TRCN0000286578, Sigma Aldrich, #2 V3LHS_318784, GE Dharmacon), with the details of generation described in the Supplementary Methods. To knock down NAPRT1, UACC257 cells were infected with GIPZ Lentiviral Human NAPRT shRNA (V3LHS_359032, GE Dharmacon). GIPZ nonsilencing lentiviral shRNA control (RHS4348, GE Dharmacon) was used as a matched control.

To assess cell viability, under treatment conditions, CellTiter-Glo (Promega) assays were performed at the indicated time points, and the IC₅₀ values were determined. PARP activity was quantified by HT Colorimetric PARP/Apoptosis Assay Kit (Trevigen) according to the manufacturer's recommendations. To evaluate caspase-3/7 activities, 7,000 to 8,000 cells were treated with DMSO or NAMPT inhibitors (12.5 nmol/L) and/or temozolomide (200 μmol/L), or Z-VAD-FMK (50 μmol/L) for 6 or 24 hours and were tested by Caspase-Glo 3/7 Assay (Promega) according to the manufacturer's recommendations. Details are described in Supplementary Methods.

To evaluate qualitative values of NAD⁺, NADH, NADP⁺, and NADPH, NAD/NADH-Glo Assay and NADP/NADPH-Glo Assay (Promega) were used according to the manufacturer's recommendations. To measure NAD⁺ quantitatively, NAD⁺/NADH Quantification Colorimetric Kit (BioVision Incorporated) was used according to the manufacturer's recommendations. Protein concentrations were measured and used for normalization. NADt (NADH and NAD) and NADH signals were measured at OD 450 nm (BIOTEK). NAD⁺ concentrations were calculated by subtracting NADH from NADt. Data were expressed as pmol/1 × 10⁶ cells or pmol/mg protein.

All mouse experiments were approved by the Institutional Animal Care and Use Committee at Massachusetts General Hospital (Boston, MA). HT1080 cells (2.5 × 10⁶) were subcutaneously implanted into the right flank of 7- to 10-week-old female SCID mice. When maximum tumor diameter reached 5 mm, cohorts were randomized to vehicle (oral gavage, n = 6), FK866 (15 mg/kg, i.p, n = 6), temozolomide (50 mg/kg, oral gavage, n = 6), or FK866 (15 mg/kg, i.p) plus temozolomide (50 mg/kg, oral gavage, n = 6). Distilled water with 20% Captisol (Cydex) and 5% dextrose (Sigma-Aldrich) was used as vehicle control. Treatment was given four times a week for 2 weeks. Tumor diameters were measured three times a week using a digital caliper. Calculated volume (mm³) = length (mm) × width (mm)² × 0.5. Tumor volume was normalized at day 0 and expressed as percent change.

Tumor tissue sections were incubated with anti-Ki-67 (1:125, Wako), LC3A/B (1:100, Cell Signaling Technology), cleaved-PARP (1:100, Cell Signaling Technology), or cleaved-caspase-3 (1:100, Cell Signaling Technology) primary antibody.

All of the experiments were performed in three replicates per condition. For parametric analyses, two-tailed Student t tests or one-way ANOVA were used, and for analysis of frequencies of nominal data, two-tailed Fisher exact test was used, without Bonferroni correction. Data were expressed as mean ± SE. P values less than 0.05 were considered statistically significant.

The PARP-activated BER pathway consumes NAD⁺ during repair of the DNA lesions induced by temozolomide exposure. Because of decreased NAD⁺ biosynthesis, IDH1-mutant cancer cells contain a lower baseline NAD⁺ level compared with IDH wild-type cancer cells. We therefore investigated the effect of temozolomide on NAD⁺ levels in mutant IDH1 cells. We initially tested the cytotoxic and PARP-inducing effect of temozolomide in a panel of endogenous IDH1-mutant and wild-type glioma tumorspheres (Fig. 1A; Supplementary Fig. S1A; Supplementary Table S1).

Overall, sensitivity to temozolomide monotherapy varied across the panel, irrespective of IDH1 mutation or MGMT promoter methylation status (Supplementary Table S1). In MGG152 cells (IDH1 mutant, MGMT promoter methylated), temozolomide cytotoxicity was observed after day 4 (Fig. 1B). We observed PARP activation after temozolomide treatment, resulting in a marked induction of PAR in IDH1-mutant cancers in a time- and dose-dependent manner (Fig. 1C and D; Supplementary Fig. S1B). Notably, PARylation was induced rapidly, within 1 hour, and significantly decreased thereafter, consistent with its role in BER signaling response. Accordingly, temozolomide decreased NAD⁺ levels in a dose-dependent manner in multiple IDH1-mutant cell lines (Fig.1E; Supplementary Fig. S1C).

To investigate whether this effect was PARP specific, we repeated these experiments with the addition of the PARP inhibitor olaparib. Olaparib significantly suppressed PARP activation and PAR expression observed with temozolomide treatment (Fig. 1F). Olaparib also reversed the NAD⁺ reduction induced by temozolomide (Fig. 1G; Supplementary Fig. S1D). Thus, temozolomide promotes acute NAD⁺ consumption via PARP activation in IDH-mutant cancer cells.

As noted above, IDH1-mutant cancer cells contain a lower baseline NAD⁺ level compared with IDH wild-type cancer cells, rendering IDH1-mutant cells selectively sensitive to NAD⁺ biosynthesis blockade by NAMPT inhibition (24). We hypothesized that temozolomide-induced NAD⁺ reduction could therefore potentiate the metabolic cytotoxicity of NAMPT inhibitors in IDH-mutant cells. To investigate this further, we first tested the temozolomide sensitivity of several endogenous IDH1-mutant cancer cell lines. Exposure to temozolomide alone for 24 to 36 hours had no or marginal impact on viability of IDH1-mutant cells, consistent with the effective BER of DNA damage during this time interval.

However, the same exposure to temozolomide significantly augmented the acute cytotoxic effect of the NAMPT inhibitors FK866 and GMX1778 (Fig. 2A and B; Supplementary Fig. S2A). Testing a range of NAMPT inhibitor doses revealed that low dose NAMPT inhibitor (2.5 nmol/L) combined with temozolomide had more potent cytotoxicity at this early time point (24 hours) than high dose (250 nmol/L) NAMPT inhibitor monotherapy in IDH1-mutant cancer cells (Fig. 2C). Importantly, this metabolic combination effect seen in IDH1-mutant cancer cells was distinct from, and not dependent upon the rate-limiting genotoxic damage of single-agent temozolomide itself, nor its well-established modifier, MGMT promoter methylation status (Supplementary Table S1). In contrast, we did not detect any effect on cell viability with temozolomide, NAMPT inhibitors, and their combination in several IDH1 wild-type glioma tumorsphere cells at 24 hours (Fig. 2D; Supplementary Fig. S2B). At 72 hours, temozolomide combination with a higher dose range of NAMPT inhibitors showed some cytotoxic effects in endogenous IDH1 wild-type glioma tumorsphere lines (Supplementary Fig. S2C and S2D). In addition, combination treatment was not detectably toxic to NHA (Fig. 2E). Thus, combined treatment with NAMPT inhibitors and temozolomide had potent and additive effects on acute cytotoxicity, in a manner that had greater selectivity for IDH1-mutant cells.

To further explore the mechanism of this genotype-selective combination effect, we tested whether the combination of NAMPT inhibitor and temozolomide, each of which reduces intracellular NAD⁺ levels as single agents, would have an additive effect on NAD⁺ levels in IDH1 mutant cancer cells. We found that combined NAMPT inhibition and temozolomide exposure significantly decreased NAD⁺ and NADH levels compared with single-agent treatment, and the effect was dose dependent (Fig. 3A; Supplementary Fig. S3A). We also found that NADP⁺ and NADPH levels were only moderately decreased by the combination treatment (Supplementary Fig. S3A). Measurement of NAD⁺ levels at different time points revealed that, in contrast to the sustained NAD⁺ reduction induced by NAMPT inhibitors, the temozolomide-induced decrease in NAD⁺ was transient and recovered to baseline at 6 to 12 hours posttreatment (Fig. 3B; Supplementary Fig. S3A and S3B). Interestingly, preincubation with temozolomide for 36 hours prior to addition of NAMPT inhibitor attenuated combination benefit (Supplementary Fig. S3C), underscoring the role of the transient temozolomide-induced NAD⁺ decrease in the additive effect on NAD⁺ levels and efficacy. Nevertheless, the combination of NAMPT inhibitors and temozolomide progressively decreased NAD⁺ levels over time when given concurrently (Fig. 3B).

To investigate whether the additive effect of concurrent NAMPT inhibitor and temozolomide treatment on NAD⁺ levels in mutant IDH1 cells were mediated through PARP, we tested the combination with and without PARP inhibition with small-molecule inhibitors. Indeed, olaparib and veliparib were each sufficient to near-completely rescue the decrease in NAD⁺ levels as well as cell viability when given concurrently with NAMPT inhibitor and temozolomide in multiple IDH1-mutant cancer cells (Fig. 3C and D; Supplementary Fig. S3D and S3E). In addition, exogenous supplementation with NAD⁺ or nicotinamide mononucleotide, the product of the NAMPT reaction and the immediate precursor of NAD⁺, restored NAD⁺ levels and near-completely abrogated the cytotoxic effect of temozolomide and NAMPT inhibitor combination on IDH1-mutant cell viability (Fig. 3E and F; Supplementary Fig. S3F and S3G). These experiments indicate the effects of combined temozolomide and NAMPT inhibitor were both NAD⁺- and NAMPT specific. Together, these data indicate that combined treatment with NAMPT inhibitor and temozolomide results in additive NAD⁺ depletion in IDH1-mutant cancer cells.

To experimentally confirm whether combined temozolomide and NAMPT inhibitor treatment is selective for IDH1 mutation itself, we engineered an isogenic system using the UACC257 melanoma line to stably express mutant IDH1 (UACC257-IDH1^R132H) or control GFP (UACC257-GFP; Fig. 4A). There was no significant difference in proliferation between these lines (Supplementary Fig. S4A). Consistent with previous observations in a different model system (24), expression of mutant IDH1 lowered NAPRT1 expression (Fig. 4A) and steady-state NAD⁺ levels (Fig. 4B).

Next, we tested the combination on cell viability in the engineered lines. Consistent with our previous findings, expression of mutant IDH1 induced sensitivity to NAMPT inhibitors. Although a minor additive cytotoxic effect was seen in control GFP-expressing cells, combination treatment potently decreased cell viability in the mutant IDH1–expressing cell line (Fig. 4C). To confirm this IDH1 genotypic selectivity in a second line, we used a tetracycline-inducible IDH1-mutant glioblastoma tumorsphere line, MGG18-IDH1-R132H (Fig. 4D; ref. 24). As a control, tetracycline did not affect MGG18 response to NAMPT inhibitor or temozolomide (Supplementary Fig. S4B). Again, when mutant IDH1 was induced, no significant difference in cell proliferation was observed (Supplementary Fig. S4C). Consistent with our aforementioned findings with the UACC257 engineered lines, MGG18-IDH1-R132H became more acutely sensitive to NAMPT inhibitor treatment and demonstrated marked additive sensitivity to combination treatment with temozolomide, while combined therapy was minimally cytotoxic without mutant IDH1 induction (Fig. 4E).

We hypothesized that the mechanism of enhanced cytotoxicity in IDH1-mutant cells, compared with wild-type cells, was mediated through the reset of basal steady-state NAD⁺ levels by inhibition of NAPRT1, a rate-limiting enzyme of NAD⁺ biosynthesis. To validate this threshold effect, we established a stable NAPRT1 knockdown line of UACC257 (Fig. 4F). Consistent with our results observed in the endogenous and enforced IDH1-mutant lines, NAPRT1 knockdown decreased steady-state levels of NAD⁺ (Fig. 4G). Importantly, single-agent NAMPT inhibitor and combination treatment decreased cell viability of the NAPRT1 knockdown line to a greater extent than that of controls (Fig. 4H). Altogether, these findings indicate that mutant IDH1 decreases basal NAD⁺ through NAPRT1 inhibition, mediating increased vulnerability to combination treatment with NAMPT inhibitor and temozolomide in a genotype selective manner.

In parallel with ATP-based cell viability assay, we confirmed cell death from combined NAMPT inhibitor and temozolomide treatment by counting viable cells after 48-hour treatment (Fig. 5A; Supplementary Fig. S5A and S5B). We further found that combination treatment significantly inhibited clone formation of HT1080 cells (Fig. 5B), even with lower doses of temozolomide (Supplementary Fig. S5C). We observed induction of LC3-II expression, a marker of autophagy, at 48 hours after NAMPT inhibitor treatment (Supplementary Fig. S5D). Combined treatment with temozolomide more strongly induced LC3-II expression than single-agent exposure with either NAMPT inhibitor or temozolomide, and LC3-II expression was detectable as early as 6 hours posttreatment (Fig. 5C), indicative of an earlier engagement of a metabolic stress response. Furthermore, exogenous 3-methyladenine (3-MA), an autophagy inhibitor, partially rescued the cytotoxicity of the combination treatment (Fig. 5D; Supplementary Fig. S5E). In addition to autophagy, we found combination treatment induced PARP cleavage (Supplementary Fig. S5F) and increased caspase-3/7 activity (Supplementary Fig. S5G), suggesting a contribution of apoptosis to cell death. However, the pan-caspase inhibitor Z-VAD-FMK, which strongly suppressed caspase-3/7 activity (Supplementary Fig. S5H), only mildly rescued cell viability after combined temozolomide and NAMPT inhibitor treatment (Supplementary Fig. S5I). These data provide evidence that the additive cytotoxicity of temozolomide when combined with an NAMPT inhibitor in IDH-mutant cells is primarily mediated through augmentation of a metabolic stress response.

The clonal emergence of resistance to alkylating chemotherapy in a population of cancer cells is known to be mediated by inactivation of genes in the MMR system (33), and an MMR deficiency is typically acquired in post-temozolomide recurrent gliomas (13–16). To assess the independence of our observed combination effect from known mediators of temozolomide genotoxic cytotoxicity, we examined whether temozolomide potentiation of NAD⁺ depletion in IDH1-mutant cancer cells varied in relation to MMR status. To this end, we established stable MSH6 knockdown lines of HT1080, which is MGMT unmethylated (Supplementary Table S1; Fig. 6A and B). As expected, knockdown of MSH6 (MSH6 #1 and MSH6 #2) induced resistance to temozolomide, confirming the critical role of MSH6 in MMR-mediated temozolomide sensitivity (Fig. 6C). However, MSH6 knockdown did not alter NAMPT inhibitor sensitivity in HT1080 cells (Fig. 6D). Importantly, and strikingly indicative of this second independent mechanism, temozolomide equally potentiated the effect of NAMPT inhibitor treatment in HT1080 cells with and without MSH6 knockdown (Fig. 6D). Similarly, in MGMT methylated MGG152 cells (Fig. 6A), we established stable MSH6 knockdown cells and confirmed MSH6 knockdown increased resistance to temozolomide (Fig. 6E and F). Again, irrespective of MSH6 status, NAMPT inhibitors displayed equal efficacy (Fig. 6G), and temozolomide treatment equivalently augmented NAMPT inhibitor cytotoxicity in MGG152 cells (Fig. 6G). These findings indicate the independent effectiveness of temozolomide metabolic augmentation targeting NAD⁺ in IDH1-mutant cells, regardless of the status of resistance determinants to temozolomide genotoxic monotherapy effect.

The NAMPT inhibitor FK866 potently inhibits the growth of HT1080 (IDH1^R132C) xenograft tumors implanted in the flank of immunocompromised mice (24). We tested whether temozolomide alteration of NAD⁺ metabolism could potentiate the in vivo growth inhibiting effects of NAMPT inhibitor in this IDH-mutant xenograft model. We first measured intratumoral NAD⁺ levels with temozolomide treatment, two doses of NAMPT inhibitor (15 or 30 mg/kg per day), and combined therapy with temozolomide and FK866 at 15 mg/kg per day. Monotherapy with temozolomide and both doses of NAMPT inhibitor significantly reduced intratumoral NAD⁺ levels compared with vehicle-treated tumor tissue, and there was a dose-dependent decrease in NAD⁺ levels with NAMPT inhibitor treatment (Fig. 7A). Notably, reduced-dose FK866 (15 mg/kg per day) plus temozolomide treatment reduced intratumoral NAD⁺ to the levels equivalent to that of high dose FK866 (30 mg/kg per day).

On the basis of these observations, we tested whether the combination of reduced-dose FK866 with temozolomide has antitumor effect in HT1080 xenografts. Compared with vehicle treatment, FK866 (15 mg/kg) alone, but not temozolomide alone, significantly suppressed tumor growth. Indeed, combination treatment was significantly superior in inhibiting tumor growth compared with monotherapy with either agent or vehicle (Fig. 7B). In contrast, we did not observe the combination effect in IDH wild-type U87 xenografts (Fig. 7C). Histopathologic analysis demonstrated the presence of necrotic regions in tumor tissues receiving combination treatment, but not in those receiving vehicle treatment. IHC analysis revealed that combination treatment significantly decreased Ki-67 expression (Fig. 7D and E) and induced the autophagy marker LC3-A/B (Fig. 7D and E), similar to patterns previously observed in single-agent NAMPT inhibitor treatment (24). In contrast, we did not find significant differences in cleaved PARP and cleaved caspase-3 expression between combination and control groups (Supplementary Fig. S6). Importantly, we did not observe significant adverse toxicity during treatment, and no differences in body weight were detected between all the treatment groups (Fig. 7F).

Steady-state levels of cellular NAD⁺ are established by the balance of metabolite influx from biosynthetic enzymes, such as NAPRT1 and NAMPT, and efflux via NAD⁺ consumption by enzyme complexes including the PARP and SIRT proteins (28). In IDH1-mutant cancer cells, the baseline pool of NAD⁺ is diminished compared with that in IDH1 wild-type cells, due to reduced expression of the NAD⁺ biosynthetic enzyme NAPRT1 (24). In our findings herein, we observed a burst of NAD⁺ consumption associated with PARP activation during the initial time period after temozolomide treatment. This NAD⁺ consumption critically depleted NAD⁺ levels and introduced a window of hypervulnerability to NAMPT inhibitors. This finding provided a rationale for the therapeutic combination of temozolomide and NAMPT inhibitors, which resulted in improved efficacy, when compared with their use as single agents, in both in vitro and in vivo IDH1-mutant cancer models.

Importantly, our evidence indicates the effect is genotype selective, as we observe greater potency against IDH1-mutant cells compared with wild-type (including noncancerous) cells. Whether our findings would extend across the diverse spectrum of IDH1-mutant cancers would require further validation. Nonetheless, the efficacy of combining these two agents may allow for optimized dosing to maintain therapeutic index without compromising anticancer activity; for instance, by enabling reduced doses of agents that induce significant toxicity at doses needed for efficacy in monotherapy (34). This is particularly relevant to NAMPT inhibitors, which have been observed to have dose-limiting thrombocytopenia and gastrointestinal toxicities (35). Also, given the time dependence of temozolomide-induced PARP activation observed in vitro, future studies would be required to delineate the optimal timing window for combined administration of these agents. Analogously, we can envision strategies that combine systemic administration of one agent with local administration of the partner drug (or prodrug with CNS penetration before activation), to maximize intratumoral pharmacodynamic effect.

Furthermore, our findings may offer a new mechanistic perspective on the accumulating evidence that IDH1-mutant gliomas can display clinical responsiveness to alkylating chemotherapy. Our data raise the possibility that metabolic vulnerabilities during the acute DNA damage response could, in part, mediate this effectiveness. The discovery of an alternative mechanism of chemotherapy-induced cytotoxicity in IDH-mutant cancer cells would also provide an explanation for the paradox of why MGMT promoter methylation is not a predictive biomarker of chemotherapy response in IDH-mutant gliomas (36), which is in contrast to the evidence base derived from IDH wild-type glioblastomas (9, 37).

Finally, and notably from a clinical standpoint, our findings suggest that exploiting this mechanism of temozolomide-induced metabolic stress may allow for the recovery of an effective therapeutic index even in recurrent cancers with MMR-mediated alkylator resistance. Indeed, Sobol and colleagues have previously highlighted this ability to shift the spectrum of rate-limiting DNA damage response by modulating PARP/NAD⁺ consumption to restore cytotoxicity in MMR-deficient IDH wild-type glioma cells (38). On the basis of our findings here in IDH1-mutant tumors, we propose that upfront combination treatment regimens could potentially minimize or even avoid the development of this escape pathway of MMR deficiency in recurrences, improving the durability of treatment response. Further studies of other chemotherapeutic agents in clinical use, such as carmustine and lomustine, are needed to fully understand how they may exploit metabolic vulnerability for IDH1-mutant cancers. Notably, these alkylating agents generate a spectrum of DNA adducts that differ compared with temozolomide, especially with regard to N3-methyladenine and N7-methylguanine lesions (27), and therefore, their effects in combination with NAD⁺-reducing agents may differ as well.

In conclusion, we demonstrate the rational conception of a combined therapy with the alkylating agent temozolomide and targeted metabolic inhibition for IDH1-mutant tumors. These findings offer an evidence base for treatment strategies with the potential to minimize toxicity, allow for salvage therapy, and ultimately provide a more effective treatment for patients with these cancers.

T. Batchelor is a consultant/advisory board member for Amgen, Merck, NXDC, Proximagen, and Roche. A.J. Iafrate has ownership interest (including patents) in Amgen and is a consultant/advisory board member for Roche. D.P. Cahill has received speakers bureau honoraria from Merck. No potential conflicts of interest were disclosed by the other authors.

Conception and design: K. Tateishi, J.J. Miller, T.T. Batchelor, H. Wakimoto, A.S. Chi, D.P. Cahill

Development of methodology: K. Tateishi, F. Higuchi

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): K. Tateishi, M.V.A. Koerner, N. Lelic, A.J. Iafrate

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): K. Tateishi, J.J. Miller, N. Lelic, G.M. Shankar, A.J. Iafrate, H. Wakimoto, D.P. Cahill

Writing, review, and/or revision of the manuscript: K. Tateishi, J.J. Miller, G.M. Shankar, S. Tanaka, D.E. Fisher, T.T. Batchelor, A.J. Iafrate, H. Wakimoto, A.S. Chi, D.P. Cahill

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): K. Tateishi, M.V.A. Koerner, S. Tanaka, T.T. Batchelor, D.P. Cahill

Study supervision: H. Wakimoto, A.S. Chi, D.P. Cahill

Other (extending study to melanomas with IDH1 mutation): D.E. Fisher

The investigators thank Drs. Hang Lee and Alona Muzikansky for biostatistical input and review. The investigators would additionally like to thank their colleagues within the DFHCC for helpful discussions, and timely and careful review of the manuscript.

This work was supported by NIH P50CA165962-01A1 to T.T. Batchelor (principal investigator), D.P. Cahill, H. Wakimoto, and A.S. Chi, K24 CA125440-06 to T.T. Batchelor, a Burroughs Wellcome Fund Career Award to D.P. Cahill, Grant-Aid for Scientific Research (16K10765 to K. Tateishi), Yokohama Academic Foundation (K. Tateishi), The Yasuda Medical Foundation (K. Tateishi), and a Society of Nuclear Medicine and Molecular Imaging Wagner-Torizuka Fellowship (K. Tateishi).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.