Purpose: We sought to identify predictive biomarkers for a novel nicotinamide phosphoribosyltransferase (NAMPT) inhibitor.

Experimental Design: We use a NAMPT inhibitor, GNE-617, to evaluate nicotinic acid rescue status in a panel of more than 400 cancer cell lines. Using correlative analysis and RNA interference (RNAi), we identify a specific biomarker for nicotinic acid rescue status. We next determine the mechanism of regulation of expression of the biomarker. Finally, we develop immunohistochemical (IHC) and DNA methylation assays and evaluate cancer tissue for prevalence of the biomarker across indications.

Results: Nicotinate phosphoribosyltransferase (NAPRT1) is necessary for nicotinic acid rescue and its expression is the major determinant of rescue status. We demonstrate that NAPRT1 promoter methylation accounts for NAPRT1 deficiency in cancer cells, and NAPRT1 methylation is predictive of rescue status in cancer cell lines. Bisulfite next-generation sequencing mapping of the NAPRT1 promoter identified tumor-specific sites of NAPRT1 DNA methylation and enabled the development of a quantitative methylation-specific PCR (QMSP) assay suitable for use on archival formalin-fixed paraffin-embedded tumor tissue.

Conclusions: Tumor-specific promoter hypermethylation of NAPRT1 inactivates one of two NAD salvage pathways, resulting in synthetic lethality with the coadministration of a NAMPT inhibitor. NAPRT1 expression is lost due to promoter hypermethylation in most cancer types evaluated at frequencies ranging from 5% to 65%. NAPRT1-specific immunohistochemical or DNA methylation assays can be used on archival formalin paraffin-embedded cancer tissue to identify patients likely to benefit from coadministration of a Nampt inhibitor and nicotinic acid. Clin Cancer Res; 19(24); 6912–23. ©2013 AACR.

Translational Relevance

Although two Nampt inhibitors have entered clinical trials, no attempt was made to select potentially responsive patients in these trials. Tumor-specific loss of nicotinate phosphoribosyltransferase (NAPRT1) is synthetically lethal with coadministration of a nicotinamide phosphoribosyltransferase (NAMPT) inhibitor and nicotinic acid. Therefore, identification of NAPRT1-deficient tumors can enable a diagnostically driven clinical strategy that entails selection of patients likely to benefit from coadministration of a NAMPT inhibitor and nicotinic acid. We demonstrate that the loss of NAPRT1 expression is due to DNA methylation, and that NAPRT1 expression is lost in a broader range of cancer indications than previously appreciated. However, the percentage of tumors that lack NAPRT1 suggests that prospective identification of patients will be required in most indications. To support this approach, we validate NAPRT1 immunohistochemical and DNA methylation assays to enable implementation of this novel diagnostic strategy in the clinic.

There has been a recent reawakening of interest in targeting the altered metabolic state of cancer cells to combat cancer. Although the observation that cancer cells utilize atypical metabolic pathways was originally made quite some time ago (1), it was unclear how to exploit this property for cancer therapeutics development. However, there are now a number of inhibitors of metabolic enzymes and a wealth of cancer-genome data that is opening new avenues to target cancer metabolism (2). One approach is to inhibit NAD biosynthesis (3). Cancer cells seem to require higher levels of NAD and NADH because they have high metabolic demands and rely heavily on glycolysis, a process that is far less efficient than oxidative phosphorylation for generating ATP. In addition, cancer cells may require more NAD due to increased activity of NAD-consuming enzymes such as the sirtuins and PARPS (4).

Nicotinamide phosphoribosyltransferase (NAMPT), also known as visfatin or pre-B-cell colony-enhancing factor, catalyzes the rate-limiting step in the primary salvage pathway used to generate NAD. It transfers a phosphoribosyl residue from 5-phosphoribosyl-1-pyrophosphate to nicotinamide to produce nicotinamide mononucleotide, which is subsequently converted into NAD+ by nicotinomide mononucleotide adenylyl transferase (NMNAT; ref. 5–7). Three NAMPT inhibitors of two distinct structural classes have entered clinical trials for cancer, APO866 (formerly FK866; ref. 8), GMX1778 (formerly CHS828), and a prodrug of GMX1778, GMX1777 (9–11). In these trials, thrombocytopenia and gastrointestinal toxicity were the most common adverse events, with thrombocytopenia being the dose-limiting toxicity (12–14).

A novel strategy for increasing the therapeutic index of NAMPT inhibitors in a defined subset of cancer has been proposed. An alternate pathway for NAD synthesis relies on nicotinate phosphoribosyltransferase (NAPRT1) to convert nicotinic acid (NA, niacin, vitamin B3) into nicotinic acid mononucleotide, which is converted to NAD by NMNAT (Fig. 1A). Addition of nicotinic acid to the media or food allows this pathway to generate sufficient levels of NAD to mitigate the cytotoxic effects of a NAMPT inhibitor in cultured cells and animal rodents (15–17) as well as in human megakaryocytes (18). Interestingly, some cancers do not express NAPRT1 (19), rendering this pathway nonfunctional. Therefore, supplementation with nicotinic acid has the potential to mitigate the toxicity of NAMPT inhibitors to normal tissues that express NAPRT1, but not in NAPRT1-deficient cancer cells, thereby increasing the potential therapeutic index of NAMPT inhibitors. Although this strategy has been described and tested in animal models, there was no attempt to select patients based on NAPRT1 status or coadminister nicotinic acid in clinical trials of NAMPT inhibitors. Thus, the strategy of selecting patients with NAPRT1-deficient tumors and coadministering nicotinic acid remains untested in the clinic.

Figure 1.

Response of cancer cell lines to GNE-617 in the presence or absence of nicotinic acid. A, the NAMPT- and NAPRT1-dependent NAD salvage pathways. B, response of a rescuable (Calu-6) and non-rescuable (NCI-H460) cell line to GNE-617 in the absence (solid line) or presence (dotted line) of 10 μmol/L nicotinic acid evaluated by total nucleic acid (CyQuant). C, average IC50 values for 53 non–small lung cancer cell lines, rescuable lines (gray), non-rescuable lines (black), error bars represent SD. D, fraction of cell lines that are not rescued by nicotinic acid across a panel of cancer cell lines derived from a variety of tissues, the number of each cell line type is indicated in parentheses.

Figure 1.

Response of cancer cell lines to GNE-617 in the presence or absence of nicotinic acid. A, the NAMPT- and NAPRT1-dependent NAD salvage pathways. B, response of a rescuable (Calu-6) and non-rescuable (NCI-H460) cell line to GNE-617 in the absence (solid line) or presence (dotted line) of 10 μmol/L nicotinic acid evaluated by total nucleic acid (CyQuant). C, average IC50 values for 53 non–small lung cancer cell lines, rescuable lines (gray), non-rescuable lines (black), error bars represent SD. D, fraction of cell lines that are not rescued by nicotinic acid across a panel of cancer cell lines derived from a variety of tissues, the number of each cell line type is indicated in parentheses.

Close modal

To enable clinical implementation of this novel diagnostic strategy, we sought to identify the mechanism by which functional NAPRT1 is lost, determine the frequency of NAPRT1 deficiency across a large number of indications, and develop assays to identify NAPRT1-deficient cancers in patients

Cellular and biochemical assays

Cell lines were obtained from the American Type Culture Collection or Deutsche Sammlung von Mikroorganismen und Zellkulturen (DSMZ) and stored in a central cell bank. Lines were authenticated by short tandem repeat and genotyped upon re-expansions. Cells were grown in RPMI-1640 medium supplemented with 10% FBS and 2 mmol/L glutamine (Invitrogen) and passaged not more than 20 times after thawing. To determine the IC50 values and nicotinic acid rescue status, cells were treated with nine point dose titrations of GNE-617 with or without 10 μmol/L nicotinic acid. At 96 hours post-drug addition, the GNE-617–treated cells were evaluated using CyQUANT Direct Cell Proliferation Assay (Invitrogen, Ltd.) followed by CellTiter-Glo Luminescent Cell Viability Assay (Promega Corporation) quantified with a Wallac EnVision 2104 Multilabel Reader. IC50 values were calculated using XLfit 5.1 (ID Business Solutions, Ltd). To examine the protein level, cells were lysed in ice-cold radioimmunoprecipitation assay buffer (Cell Signaling Technology), run on SDS-PAGE (4%–12% Bis-Tris; Invitrogen), and evaluated by Western blotting using antibodies directed against NAPRT1 and β-actin (Sigma-Aldrich). For RNA interference (RNAi), A549 cells were plated at 1,500 cells per well in 96-well plates, allowed to adhere for 24 hours, and transfected with 25 nmol/L siRNA oligonucleotide using Dharmafect 4 (Thermo Scientific). NAPRT1-directed siRNA oligonucleotides were purchased from Ambion (s41083 and s41084) and from Thermo Scientific (J-016912-09, and J-016912-10). Nontargeting control siRNA was purchased from Thermo Scientific (D-001810-10-20). Transfected cells were treated with the indicated concentrations of GNE-617 for 72 hours and viability was evaluated with CellTiter-Glo (Promega). Lysates for detection of NAPRT1 protein were collected 72 hours after transfection of 1 million A549 cells in 10 cm dishes. For NAPRT1 re-expression, RERF-LC-MS cells were transfected with pCMV6-AC.NAPRT1 and empty vector pCMV6-AC (OriGene Technologies, Inc) using Amaxa Nucleofector technology and selected with Geneticin (Life technologies).

LC/MS assay for NAD concentration

The concentration of NAD in cells was determined by a nonvalidated liquid chromatography/tandem mass spectrometry (LC/MS-MS) assay using 13C5-NAD as an internal standard. Detailed methods are in the Supplementary Methods.

Immunohistochemistry

Immunohistochemistry for NAPRT1 was performed on a Ventana Discovery XT autostainer (Ventana Medical Systems). Formalin-fixed paraffin-embedded (FFPE) whole-tissue and tissue microarray sections were deparaffinized and pretreated with CC1 solution (Ventana Medical Systems) for 60 minutes followed by incubation with either 0.5 μg/mL NAPRT1 rabbit polyclonal antibody (Novus Biologicals) or naïve rabbit immunoglobulin G (Cell Signaling Technologies) for 60 minutes at 37°C. Detection was performed with OmniMap anti-rabbit HRP and DAB (Ventana Medical System) followed by counter staining with Hematoxylin II (Ventana Medical System).

Deparaffinization and DNA extraction of FFPE section for QMSP

FFPE sections were deparaffinized by soaking 3 times in Envirene (Hardy Diagnostics Cat No. CE-016) for 5 minutes, then 2 times in 100% ethanol for 5 minutes, then dried for 10 minutes at room temperature. Sections were scraped off the slides and Proteinase K digested in Tissue Lysis Buffer (from Roche High Pure FFPE RNA Micro Kit, Cat. No. 04823125001) overnight at 56°C. DNA was then purified using the QIAamp DNA FFPE Tissue Kit (Qiagen, Cat. No. 56404) starting at step 12.

RNA-seq and copy number analysis

RNA-seq reads were aligned to the human genome version GRCh37 using GSNAP (20). Gene expression was obtained by counting the number of reads aligning concordantly within a pair and uniquely to each gene locus as defined by consensus coding DNA sequence. The gene counts were then normalized using the DESeq Bioconductor software package (21). Illumina HumanOmni2.5_4v1 arrays were used to assay 906 cancer cell lines for genotype, DNA copy, and LOH at approximately 2.2 million single-nucleotide polymorphism (SNP) positions following methods published previously (PMID:23033341 and PMID:22895193). Copy number for NAPRT1 was calculated as the average absolute copy number for all SNPs within or directly adjacent to the bounds of the genomic region covered by any RefSeq transcript for that gene. Samples with a copy number less than 2 were counted as having a deletion.

DNA methylation analysis

DNA methylation was measured by Illumina Infinium 450 K BeadChip and preprocessed using Bioconductor lumi package (22) with default settings, as previously described (23). For next-generation sequencing, DNA samples containing equal amounts of 8 bisulfite PCR products (∼500 ng DNA total) were treated with T4 DNA polymerase, Klenow large fragment, and T4 polynucleotide kinase to generate 5′-phosphorylated blunt ends. After concatemerization with T4 DNA ligase, the sample was sonicated to an average fragment length of 150 to 300 bp using a Misonix cuphorn sonicator 3000. Libraries were generated from these sonicated DNA samples using the standard Illumina protocol. The 24 samples were indexed with 7-bp barcodes (independent Illumina index read). Sequencing on Hi-Seq generated a total of 115 million 50-nt reads. Reads were aligned to the reference sequence using the bismark software (24). As reference sequence, either the PCR-amplified target region (in NAPRT1 gene promoter) or the entire chr8 (hg19) was used. Alignments were captured in SAM/BAM files, and percentage methylation at each CpG site was determined by running the appropriate bismark scripts. Read coverage at each CpG site was determined by generating a signal map at 1-bp resolution (using Active Motif software), and individual 50-nt reads with specific methylation status were counted using a combination of samtools (25) and standard UNIX commands. Gene schematic generated with FancyGene.

For quantitative methylation-specific PCR (QMSP) assays, sodium bisulfite converted DNA was amplified using previously described conditions (23).

Evaluation of rescue identifies cancer types that may benefit from coadministration of NAMPT inhibitor and nicotinic acid

GNE-617 is a novel small-molecule inhibitor of NAMPT (Supplementary Fig. S1A) that inhibits the biochemical activity of NAMPT with an IC50 of 5 nmol/L and exhibits efficacy in xenograft models of cancer (26). As expected, this molecule causes rapid depletion of cellular NAD followed by a decrease in ATP (Supplementary Fig. S1B). We evaluated the activity of GNE-617 on a panel 53 non–small cell lung cancer (NSCLC) cell lines in the presence or absence of 10 μmol/L nicotinic acid. The majority of cell lines exhibit a steep dose response to GNE-617 when evaluated by decrease in ATP or total nucleic acid, and the cytotoxicity is completely rescued by simultaneous addition of nicotinic acid. However, some lines are not rescued by nicotinic acid (Fig. 1B), and we refer to these as “non-rescuable” cell lines. The majority of the cell lines tested have IC50 values below 100 nmol/L, with approximately half with IC50 values less than 10 nmol/L. We find a significant correlation between baseline expression of NAMPT and IC50, consistent with previous reports (16), and between NAPRT1 expression and IC50 (Suuplementary Table S1). Eighteen cell lines were not rescued with nicotinic acid, and these non-rescuable cell lines tended to have lower IC50 values (P = 0.008, Fisher exact test, IC50 < 10 nmol/L vs. ≥10 nmol/L; Fig. 1C).

To explore the prevalence of the nicotinic acid rescue phenotype across indications, we evaluated a panel of cancer cell lines for response to 10 μmol/L GNE-617 in the presence or absence of 10 μmol/L nicotinic acid and found 16% (72/445) were not rescued with nicotinic acid (Supplementary Table S2). Cell lines derived from brain tumors or sarcomas exhibit the non-rescuable phenotype at a higher frequency than other types of cancer cell lines (45% and 67%, respectively), consistent with previous reports (16). In this study, we expand into a larger variety of cancer types and find at least one non-rescuable cell line from most indications tested. However, with the exception of sarcomas and brain tumors, the prevalence is less than 35% (Fig. 1D).

Rescue of NAMPT inhibitor toxicity by nicotinic acid is due to NAPRT1 deficiency

Previous studies have indicated that NAPRT1 is required for nicotinic acid rescue (16, 17). We wanted to test this observation in a larger panel of cell lines and identify additional determinants of nicotinic acid rescue across cancer cell lines from a diverse range of tissue types. We evaluated the cell line panel shown in Fig. 1D for correlation between gene expression and nicotinic acid rescue. The best correlate of the rescue phenotype is NAPRT1 (adjusted P = 6 × 10−30). The next best correlative gene is lymphotoxin β receptor (LTBR, TNFRSF3) with an adjusted P value of 10−9, leading us to conclude that NAPRT1 level is the dominant determinant of nicotinic acid rescue status. We examined NAPRT1 protein levels in 38 NSCLC cell lines and found a good correlation between NAPRT1 protein and rescue (Fig. 2A and B). To demonstrate that NAPRT1 is necessary for nicotinic acid rescue, we reduced the level of NAPRT1 by siRNA in A549 cells. NAPRT1 reduction did not change the IC50 for GNE-617, but the cells were no longer rescued with nicotinic acid. To demonstrate that NAPRT1 is sufficient for rescue, we expressed NAPRT1 in the NAPRT1-deficient RERF-LC/MS cell line and observed nicotinic acid rescue in this line compared with the same line transfected with the empty vector control, which remains non-rescuable (Fig. 2C).

Figure 2.

NAPRT1 level determines nicotinic rescue status in cancer cell lines. A, Western blot analyses of NAPRT1 in NSCLC lines; *, non-rescuable lines. B, NAPRT1 level evaluated by Western blot analysis versus rescue status, n = 38, P = 0.04 (two-tailed t-test), *, P < 0.05. C, left, GNE-617 dose response of A549 cells to GNE-617 with nontargeting siRNA (NTC, filled circles) or 4 independent NAPRT1 directed siRNA oligos from Thermo Scientific (D1, D2, filled symbols) or Ambion (A1, A2, open symbols), right, GNE-617 dose response of RERF- LC/MS cells stably transfected with NAPRT1 (squares) or empty vector (EV, triangles) in the presence (open symbols, dashed lines) or absence (filled symbols) of 10 μmol/L nicotinic acid. D, fraction of cell lines for each cancer type with NAPRT mRNA ≤7.2 RMA normalized from HGU-133P (Affymetrix), n = 663.

Figure 2.

NAPRT1 level determines nicotinic rescue status in cancer cell lines. A, Western blot analyses of NAPRT1 in NSCLC lines; *, non-rescuable lines. B, NAPRT1 level evaluated by Western blot analysis versus rescue status, n = 38, P = 0.04 (two-tailed t-test), *, P < 0.05. C, left, GNE-617 dose response of A549 cells to GNE-617 with nontargeting siRNA (NTC, filled circles) or 4 independent NAPRT1 directed siRNA oligos from Thermo Scientific (D1, D2, filled symbols) or Ambion (A1, A2, open symbols), right, GNE-617 dose response of RERF- LC/MS cells stably transfected with NAPRT1 (squares) or empty vector (EV, triangles) in the presence (open symbols, dashed lines) or absence (filled symbols) of 10 μmol/L nicotinic acid. D, fraction of cell lines for each cancer type with NAPRT mRNA ≤7.2 RMA normalized from HGU-133P (Affymetrix), n = 663.

Close modal

In the 32 cell lines for which we had NAPRT1 protein quantification and gene expression data, there was a strong correlation between mRNA and protein (Spearman R = 0.88, P < 10−4; Supplementary Fig. S1C). In light of these data, we evaluated a panel of 551 cancer cell lines for NAPRT1 deficiency by gene expression to identify additional cancer types with and models that are likely to be non-rescuable (Fig. 2D and Supplementary Table S3). We set the threshold for NAPRT1 deficiency based on gene expression levels that allow nicotinic acid rescue in the NSCLC cell line panel. The threshold selected, 7.2, is close to the lower limit of detection on the Affymetrix array (Supplementary Fig. S1C), suggesting little or no gene expression. The most well represented indications, lung and breast cancer exhibited NAPRT1 deficiency at rates of 17% and 10%, respectively, similar to percentages of non-rescuable lines shown in Fig. 1D. There were no sarcoma cell lines in this panel, and the prevalence of NAPRT1 deficiency in gliomas, 9%, was lower than expected. Liver cancer cell lines had the highest prevalence of NAPRT1 deficiency (45%), slightly higher than the observation of 31% in the previous panel. This panel contains more lymphoma lines and these exhibited the second highest rate of NAPRT1 deficiency (38%), consistent with observations made on lymphoma tissue (19). In summary, lymphomas, sarcomas, and hepatocellular carcinomas tend to have higher rates of NAPRT1 deficiency.

Immunohistochemistry can be used to identify NAPRT1-deficient tumors

We next developed an immunohistochemical (IHC) assay for NAPRT1 using cell lines with known levels of NAPRT1 to validate the method. NAPRT1 staining on NSCLC tissue is cytoplasmic and has a large dynamic range. Benign cells stained with an intensity of immunohistochemistry 1+, whereas malignant cells stained across the full range of IHC scores (0–3+; Fig. 3A). To determine the level of NAPRT1 necessary for nicotinic acid rescue, we generated a cell pellet microarray from NCSLC cell lines with known rescue status and scored them for NAPRT1 on a scale of 0 to 3. With only one exception, cell lines with IHC scores more than zero were rescued by nicotinic acid (Fig. 3B). We next stained tumor samples from a variety of tumor types and determined the fraction with NAPRT1 IHC scores of zero (Fig. 3C). Small cell lung cancer (SCLC) had a high prevalence of NAPRT1 deficiency (60%), but it should be noted that the number of samples was low (n = 10). In general, the prevalence of IHC 0 by tumor type was consistent with the prevalence of NAPRT1 deficiency and loss of nicotinic acid rescue in the cell lines. Thus, an IHC assay could be used to select patients for coadministration of NAMPT inhibitor and nicotinic acid. On the basis of these preclinical data, we propose that an IHC score of zero will be predictive of a tumor likely to respond to the combination of a NAMPT inhibitor and nicotinic acid.

Figure 3.

NAPRT1 immunohistochemistry correlates with nicotinic acid rescue status. A, NAPRT1 IHC scores on NSCLC tissue are labeled in the top left of each image, note that the positively staining region in the sample scored as IHC 0 is noncancerous tissue. B, histogram of cancer cell lines binned by IHC score, colored by nicotinic acid rescue status [R = rescuable (black), NR = non-rescuable (striped), n = 35]. C, fraction of cancer tissue samples with an IHC score of 0 across indications, the asterisks indicate samples that were evaluated on a tissue microarray.

Figure 3.

NAPRT1 immunohistochemistry correlates with nicotinic acid rescue status. A, NAPRT1 IHC scores on NSCLC tissue are labeled in the top left of each image, note that the positively staining region in the sample scored as IHC 0 is noncancerous tissue. B, histogram of cancer cell lines binned by IHC score, colored by nicotinic acid rescue status [R = rescuable (black), NR = non-rescuable (striped), n = 35]. C, fraction of cancer tissue samples with an IHC score of 0 across indications, the asterisks indicate samples that were evaluated on a tissue microarray.

Close modal

NAPRT1 is methylated in a subset of lung, pancreatic, and ovarian tumors

The finding that loss of NAPRT1 expression has a strong negative association with nicotinic acid rescue suggested that NAPRT1 expression was completely absent in certain contexts. We examined copy number by SNP arrays in a panel of 906 cell lines and found that NAPRT1 underwent LOH in 18.6% of cell lines. In addition, we found four cell lines with single copy deletion at the NAPRT1 locus (BJAB, HCC1428, COLO-824, and NCI-H1882). We next examined NAPRT1 copy number in tumor tissue from The Cancer Genome Atlas (TCGA), and found the frequency of LOH to be 12.8% in ovarian cancer tissue (n = 405), and between 4.3% and 6.5% across 2066 breast, colon, glioblastoma, ovarian, and squamous lung cancer samples. There was single copy deletion of NAPRT1 in 1.5% of ovarian cancer samples, and 0.64% in the other cancer tissue samples that were evaluated for copy number. The observation that the NAPRT1 locus is subject to relatively frequent LOH and rare deletions suggests that NAPRT1 may have tumor suppressor activity in certain contexts. However, homozygous deletions were never observed, and thus cannot explain the absence of NAPRT1 mRNA and protein in some cancer cells. We therefore considered epigenetic mechanisms as an explanation for the loss of NAPRT1 expression in cancer.

Aberrant DNA methylation occurs frequently in cancer (27–31) and can be the major mechanism of loss of gene function at certain loci as in the example of RASSF1A (32). To determine whether NAPRT1 is silenced by promoter hypermethylation, we explored a previously published methylation dataset for lung cancer cell lines (23), as well as several new series for breast and pancreatic cancer. As shown in Fig. 4, hypermethylation of the NAPRT1 CpG island (indicated by high β values, colored red) is evident in a subset of NSCLC (14%), SCLC (18%), breast cancer (8%), and pancreatic cancer (18%) cell lines. In addition, partial methylation of the CpG island is evident in a subset of cell lines, indicated by the pink and light blue regions. Importantly, no methylation on the CpG island was detected in normal immortalized bronchial and small airway epithelial cells suggesting that the methylation we observed is tumor cell line specific. To demonstrate that hypermethylation of the NAPRT1 promoter causes transcriptional gene silencing, we treated 34 NSCLC cell lines with 5-aza-2′deoxycytidine (5-azadC). NAPRT1 expression was detectable both before and after 5-azadC treatment in cell lines that could be rescued by coadminstration of nicotinic acid. In contrast, low or no expression of NAPRT1 could be detected in cell lines that could not be rescued by coadminstration of nicotinic acid before 5-azadC treatment, whereas after 5-azadC treatment, NAPRT1 expression was comparable with rescuable cell lines (Fig. 4D).

Figure 4.

NAPRT1 is hypermethylated in a subset of cancer cell lines. A, heatmap of NAPRT1 DNA methylation in cell lines derived from lung cancer or nonmalignant lung epithelia (*) measured by Infinium arrays. β values represent the fraction of molecules methylated at the CpG site. B, NAPRT1 methylation in breast cancer and (C) pancreatic cancer cell lines. D, there is a significant increase in NAPRT1 gene expression after exposure to 5-aza-2′-deoxycitidine in the in the non-rescuable cell lines (P = 0.02, t test, n = 9), but not the rescuable lines (n = 25), *, P < 0.05. DMSO, dimethyl sulfoxide; NR, non-rescuable; R, rescuable.

Figure 4.

NAPRT1 is hypermethylated in a subset of cancer cell lines. A, heatmap of NAPRT1 DNA methylation in cell lines derived from lung cancer or nonmalignant lung epithelia (*) measured by Infinium arrays. β values represent the fraction of molecules methylated at the CpG site. B, NAPRT1 methylation in breast cancer and (C) pancreatic cancer cell lines. D, there is a significant increase in NAPRT1 gene expression after exposure to 5-aza-2′-deoxycitidine in the in the non-rescuable cell lines (P = 0.02, t test, n = 9), but not the rescuable lines (n = 25), *, P < 0.05. DMSO, dimethyl sulfoxide; NR, non-rescuable; R, rescuable.

Close modal

We next evaluated the correlation of NAPRT1 methylation and gene expression as measured by RNA-seq (Fig. 5A). As expected, the strongest relationship between expression and methylation was at the extreme ends of the range of βvalues. Overall, there is strong correlation between expression and methylation in all three cell line panels, with a correlation coefficient of −0.69 (n = 179; Fig. 5A). We then evaluated the correlation between methylation and expression in publicly available data from TCGA for breast cancer and lung adenocarcinoma. We find a similar strong negative association between expression and methylation in these data with correlation coefficients of −0.513 (n = 513) and −0.534 (n = 232), in breast and lung cancer, respectively (Fig. 5B and C). Next we evaluated the sensitivity and specificity of NAPRT1 methylation to predict nicotinic acid rescue. Using β values from the Infinium array, we calculated a positive predictive value of 1.0 (17/17), with a sensitivity of 0.94 (17/18) and specificity of 0.97 (34/35; Fig. 5D). Taken together, these data show that NAPRT1 expression is silenced by promoter hypermethylation in NSCLC and that promoter hypermethylation of NAPRT1 is inversely associated with nicotinic acid rescue in cancer cell lines.

Figure 5.

NAPRT1 hypermethylation is correlated with expression level and rescue status. A, NAPRT1 mRNA and DNA methylation in cancer cell lines derived from lung (n = 94), breast (n = 48), and pancreatic (n = 37) cancer. mRNA is represented as normalized count of unique RNA-seq short-reads mapped to the coding region of NAPRT1, methylation was calculated by averaging β values from Infinium arrays for all probes within the CpG island (correlation coefficient = −0.69). B, the same analysis applied to TCGA breast cancer tissue (correlation coefficient = −0.513, n = 513) and (C) lung adenocarcinoma tissue (correlation coefficient = −0.534, n = 232). D, cell lines were classified having hypermethylated NAPRT1 CpG islands (striped bar, average β ≥0.8) or not (solid bar) and binned by nicotinic acid rescue status. All hypermethylated lines were not rescuable. NAPRT1 hypermethylation has a positive predictive value of 1.0 (17/17), sensitivity of 0.94 (17/18), and specificity of 0.97 (34/35) on these cell lines.

Figure 5.

NAPRT1 hypermethylation is correlated with expression level and rescue status. A, NAPRT1 mRNA and DNA methylation in cancer cell lines derived from lung (n = 94), breast (n = 48), and pancreatic (n = 37) cancer. mRNA is represented as normalized count of unique RNA-seq short-reads mapped to the coding region of NAPRT1, methylation was calculated by averaging β values from Infinium arrays for all probes within the CpG island (correlation coefficient = −0.69). B, the same analysis applied to TCGA breast cancer tissue (correlation coefficient = −0.513, n = 513) and (C) lung adenocarcinoma tissue (correlation coefficient = −0.534, n = 232). D, cell lines were classified having hypermethylated NAPRT1 CpG islands (striped bar, average β ≥0.8) or not (solid bar) and binned by nicotinic acid rescue status. All hypermethylated lines were not rescuable. NAPRT1 hypermethylation has a positive predictive value of 1.0 (17/17), sensitivity of 0.94 (17/18), and specificity of 0.97 (34/35) on these cell lines.

Close modal

Bisulfite next-generation sequencing identifies tumor-specific methylation sites in NAPRT1

On the basis of the strength of the association between NAPRT1 methylation, expression, and rescue, we considered development of a quantitative methylation assay for use in clinical samples. We used a next-generation bisulfite sequencing (NGBS) strategy to identify the most appropriate target region for QMSP assay development within the NAPRT1 CpG island. We sequenced the NAPRT1 CpG island in DNA derived from nine peripheral blood mononuclear cells (PBMC) preparations from healthy volunteers, 20 tumor cell lines with known NAPRT1 expression and nicotinic acid rescue status, as well as DNA derived from a variety of benign tissues (Fig. 6A). Consistent with the Infinium data, the NAPRT1 promoter was hypermethylated in seven cell lines, whereas no methylation was detected in eight cell lines. Five cell lines exhibited partial methylation. Methylation was not detected in benign breast, colon, or lung tissues. Immune cell infiltrates are frequently a source of contaminating DNA in preparations derived from biopsies. Our data suggest that at least a subset of individuals have detectable methylation over most of the NAPRT1 CpG island in their PBMCs. However, there was a core region in the CpG island (hg19, chr8:144660496-144660520) where there was no methylation in any of the benign samples (Fig. 6A, bottom). These data suggest that a well-designed QMSP assay should be able to distinguish between tumors with complete promoter methylation and background signal in the tissue sample. We tested the QMSP assay in cell lines with known methylation and nicotinic acid rescue status. Consistent with bisulfite sequencing, QMSP distinguishes cell lines with methylation within the target region of the assay. Thus QMSP can be used to identify cell lines that lack NAPRT1 and are thus unable to be rescued by nicotinic acid (Fig. 6B).

Figure 6.

High-resolution mapping of NAPRT1 methylation enables design of a targeted methylation-specific PCR assay for detection of NAPRT1-low tumors. A, targeted amplicon bisulfite deep sequencing quantifies DNA methylation status of NAPRT1 promoter-associated CpG island in nicotinic acid rescuable and non-rescuable cell lines. Columns represent percent methylation at individual CpG sites (average coverage 80,568 reads per site) across the NAPRT1 CpG island. Note that bisulfite sequencing of peripheral blood lymphocyte DNA from healthy donors and of normal breast, colon, and lung tissues indicate lowest methylation levels near the NAPRT1 transcription start site, an optimal region for QMSP primer design. B, comparison of NGBS and QMSP in the NSCLC cancer cells from part A. Left y-axis shows the percentage methylation based on NGBS (grey bars), |$2^{-{\rm \Delta}C_{\rm t}}$|values (red dots, line) for the corresponding QMSP are shown on the right y-axis. C, representative QMSP results from DNA extracted from unstained sections of the tumors shown in the IHC slides above the graph. The positively stained area on the IHC 0 slide is benign tissue.

Figure 6.

High-resolution mapping of NAPRT1 methylation enables design of a targeted methylation-specific PCR assay for detection of NAPRT1-low tumors. A, targeted amplicon bisulfite deep sequencing quantifies DNA methylation status of NAPRT1 promoter-associated CpG island in nicotinic acid rescuable and non-rescuable cell lines. Columns represent percent methylation at individual CpG sites (average coverage 80,568 reads per site) across the NAPRT1 CpG island. Note that bisulfite sequencing of peripheral blood lymphocyte DNA from healthy donors and of normal breast, colon, and lung tissues indicate lowest methylation levels near the NAPRT1 transcription start site, an optimal region for QMSP primer design. B, comparison of NGBS and QMSP in the NSCLC cancer cells from part A. Left y-axis shows the percentage methylation based on NGBS (grey bars), |$2^{-{\rm \Delta}C_{\rm t}}$|values (red dots, line) for the corresponding QMSP are shown on the right y-axis. C, representative QMSP results from DNA extracted from unstained sections of the tumors shown in the IHC slides above the graph. The positively stained area on the IHC 0 slide is benign tissue.

Close modal

Most tumor samples obtained in clinical trials are archival diagnostic biopsies or surgical resections. These samples are almost always FFPE slides or blocks. Thus, to enable a diagnostic hypothesis in the clinic, a molecular assay needs to work well on these preserved tissue specimens. We tested the QMSP assay on tissue with known IHC status and we detected methylation only in the sample with IHC score of zero. Conversely, we did not detect any methylation in the tumor sample with an IHC score of 3+ (Fig. 6C).

Evaluation of NAPRT1 in cancer tissue provides the opportunity for a unique biomarker strategy for selection of patients likely to respond to NAMPT inhibitors. Selection of patients whose tumors do not express NAPRT1 allows simultaneous administration of nicotinic acid, which can mitigate the toxicity of NAMPT inhibitors in nonmalignant tissue that generally expresses NAPRT1. Although supplementing patients with nicotinic acid has the potential to improve the therapeutic index of NAMPT inhibitors, it carries with it an absolute requirement of identifying those patients whose tumors have lost the expression of NAPRT1, as nicotinic acid could also rescue tumors that express NAPRT1. Previous work has demonstrated that glioblastomas and neuroblastomas have a very high prevalence of NAPRT1 loss (16). We find that solid tissue carcinomas exhibited loss of NAPRT1 expression in approximately 15% of cell lines or samples evaluated. This greatly expands the potential responsive patient population, but requires a diagnostic test to identify patients eligible for coadministration of a NAMPT inhibitor and nicotinic acid.

By evaluating nicotinic acid rescue and gene expression in more than 400 cancer cell lines, we confirm that NAPRT1 level is the major determinant of nicotinic acid rescue status and define a threshold mRNA level below which nicotinic acid rescue was not observed. However, this will be challenging to translate into a clinical assay because even small amounts of contaminating tissue could result in an incorrect conclusion that NAPRT1 is expressed. Immunohistochemistry allows cell level resolution of staining and discrimination of staining in malignant tissue compared with adjacent normal tissue. NAPRT1 has been evaluated by immunohistochemistry in prior studies (16, 19), but there was no determination of the required threshold for rescue. In this study, we evaluated a panel of NSCLC cell lines for nicotinic acid rescue status in culture and prepared cell pellets for immunohistochemistry from the same cells. With this approach, we determined that an IHC score of zero correlated with lack of nicotinic acid rescue. Determination of the threshold is essential for the implementation of NAPRT1 immunohistochemistry as a diagnostic assay in solid tumor indications where there is a broad spectrum of NAPRT1 level.

We show that loss of function of the NAPRT1 gene is mediated primarily by hypermethylation of the CpG island that overlaps with the transcription start site of NAPRT1. Importantly, complete methylation of the CpG island is strongly associated with the absence of rescue in cell lines treated with GNE-617 and nicotinic acid. Reduction of DNA methylation by treatment of cells with 5-aza-dC was sufficient to induce NAPRT1 expression in cell lines in which the locus was hypermethylated. We mapped the sites of DNA methylation in non-rescuable cell lines using NGBS and used this high-resolution map to develop a methylation-specific QMSP assay suitable for use on tissue preserved in FFPE slides. There are several advantages of using a QMSP assay as a molecular diagnostic to identify patients. Preamplification allows detection of NAPRT1 methylation from very small amounts of tissue which could enable the use of small biopsies or even fine needle aspirates, which are frequently the only tissue available for patients diagnosed with stage IV NSCLC, SCLC, or pancreatic cancer. As shown here, the QMSP assay works well on FFPE tissues, allowing easy analysis of archival tissue. The signal to noise ratio is excellent because the background of normal tissue is generally not methylated, allowing detection of a positive signal even if there is significant contamination from noncancerous tissue. Finally, because the assay detects a positive signal of DNA methylation rather than loss of protein expression, it is not prone to false positives (for loss of NAPRT1) that may occur with IHC assays due to poor tissue preservation. Our data strongly support the evaluation of NAPRT1 promoter methylation as an enrollment biomarker for clinical trials evaluating the safety and efficacy of NAMPT inhibitors.

Although the detection of DNA methylation has been explored extensively as a method for the early detection of cancer or as a prognostic indicator (33–36), there are very few examples in which DNA methylation of a specific gene predicts a responsive patient subset for a specific class of inhibitors. Two known examples are MGMT for dacarbazine (37) and BRCA1 for PARP inhibitors (38–40), and this study adds an additional example. Both MGMT and BRCA1 are tumor suppressor genes. The finding that NAPRT1 is subject to single copy deletion along with more frequent LOH and promoter hypermethylation suggests that NAPRT1 may act as a tumor suppressor gene in some contexts. It is not clear why loss of NAPRT1 expression would be advantageous to some cancers as the gene contributes to the NAD synthetic capacity of the cell. However, it is interesting to note that frequent hypermethylation of another metabolic enzyme, LDHB, has recently been reported in breast cancer (31), suggesting that alteration of metabolic pathways by DNA methylation may be a more general phenomenon in cancer.

H. Koeppen and L.D. Belmont have ownership interest (including patents) in Roche stock. No potential conflicts of interest were disclosed by the other authors.

Conception and design: D.S. Shames, K. Elkins, R.L. Yauch, T. O'Brien, L.D. Belmont

Development of methodology: D.S. Shames, K. Elkins, T. Holcomb, T.Q. Pham, R. Bourgon

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): D.S. Shames, K. Elkins, K. Walter, D. Mohl, Y. Xiao, T.Q. Pham, X. Liang, H. Koeppen

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): D.S. Shames, K. Elkins, K. Walter, P. Du, Y. Xiao, P.M. Haverty, B. Liederer, R.L. Yauch, T. O'Brien, R. Bourgon, H. Koeppen, L.D. Belmont

Writing, review, and/or revision of the manuscript: D.S. Shames, K. Elkins, K. Walter, T. Holcomb, P. Du, B. Liederer, X. Liang, T. O'Brien, R. Bourgon, H. Koeppen, L.D. Belmont

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): K. Elkins, K. Walter, T. Holcomb

Study supervision: D.S. Shames, T. O'Brien

The authors thank Lily Shi, Jenille Dao, Yihong Yu, Jianming Wu, Eva Lin, Mamie Yu, Suresh Selvaraj, and Richard Neve for technical assistance maintenance of the central cell line bank, Georgia Hatzivassiliou for initiating NAPRT1 studies, Lulu Yang and Christian Klijn for developing data analysis methodology for RNAseq, and the TCGA team led by Anneleen Daemen.

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.

1.
Warburg
O
. 
On the origin of cancer cells
.
Science
1956
;
123
:
309
14
.
2.
Schulze
A
,
Harris
AL
. 
How cancer metabolism is tuned for proliferation and vulnerable to disruption
.
Nature
2012
;
491
:
364
73
.
3.
Bi
TQ
,
Che
XM
. 
Nampt/PBEF/visfatin and cancer
.
Cancer Biol Ther
2010
;
10
:
119
25
.
4.
Houtkooper
RH
,
Canto
C
,
Wanders
RJ
,
Auwerx
J
. 
The secret life of NAD+: an old metabolite controlling new metabolic signaling pathways
.
Endocr Rev
2010
;
31
:
194
223
.
5.
Preiss
J
,
Handler
P
. 
Enzymatic synthesis of nicotinamide mononucleotide
.
J Biol Chem
1957
;
225
:
759
70
.
6.
Garten
A
,
Petzold
S
,
Korner
A
,
Imai
S
,
Kiess
W
. 
Nampt: linking NAD biology, metabolism and cancer
.
Trends Endocrinol Metab
2009
;
20
:
130
8
.
7.
Dahl
TB
,
Holm
S
,
Aukrust
P
,
Halvorsen
B
. 
Visfatin/NAMPT: a multifaceted molecule with diverse roles in physiology and pathophysiology
.
Annu Rev Nutr
2012
;
32
:
229
43
.
8.
Hasmann
M
,
Schemainda
I
. 
FK866, a highly specific noncompetitive inhibitor of nicotinamide phosphoribosyltransferase, represents a novel mechanism for induction of tumor cell apoptosis
.
Cancer Res
2003
;
63
:
7436
42
.
9.
Hjarnaa
PJ
,
Jonsson
E
,
Latini
S
,
Dhar
S
,
Larsson
R
,
Bramm
E
, et al
CHS 828, a novel pyridyl cyanoguanidine with potent antitumor activity in vitro and in vivo
.
Cancer Res
1999
;
59
:
5751
7
.
10.
Ravaud
A
,
Cerny
T
,
Terret
C
,
Wanders
J
,
Bui
BN
,
Hess
D
, et al
Phase I study and pharmacokinetic of CHS-828, a guanidino-containing compound, administered orally as a single dose every 3 weeks in solid tumours: an ECSG/EORTC study
.
Eur J Cancer
2005
;
41
:
702
7
.
11.
Olesen
UH
,
Christensen
MK
,
Bjorkling
F
,
Jaattela
M
,
Jensen
PB
,
Sehested
M
, et al
Anticancer agent CHS-828 inhibits cellular synthesis of NAD
.
Biochem Biophys Res Commun
2008
;
367
:
799
804
.
12.
Hovstadius
P
,
Larsson
R
,
Jonsson
E
,
Skov
T
,
Kissmeyer
AM
,
Krasilnikoff
K
, et al
A phase I study of CHS 828 in patients with solid tumor malignancy
.
Clin Cancer Res
2002
;
8
:
2843
50
.
13.
Holen
K
,
Saltz
LB
,
Hollywood
E
,
Burk
K
,
Hanauske
AR
. 
The pharmacokinetics, toxicities, and biologic effects of FK866, a nicotinamide adenine dinucleotide biosynthesis inhibitor
.
Invest New Drugs
2008
;
26
:
45
51
.
14.
von Heideman
A
,
Berglund
A
,
Larsson
R
,
Nygren
P
. 
Safety and efficacy of NAD depleting cancer drugs: results of a phase I clinical trial of CHS 828 and overview of published data
.
Cancer Chemother Pharmacol
2010
;
65
:
1165
72
.
15.
Hara
N
,
Yamada
K
,
Shibata
T
,
Osago
H
,
Hashimoto
T
,
Tsuchiya
M
. 
Elevation of cellular NAD levels by nicotinic acid and involvement of nicotinic acid phosphoribosyltransferase in human cells
.
J Biol Chem
2007
;
282
:
24574
82
.
16.
Watson
M
,
Roulston
A
,
Belec
L
,
Billot
X
,
Marcellus
R
,
Bedard
D
, et al
The small molecule GMX1778 is a potent inhibitor of NAD+ biosynthesis: strategy for enhanced therapy in nicotinic acid phosphoribosyltransferase 1-deficient tumors
.
Mol Cell Biol
2009
;
29
:
5872
88
.
17.
Olesen
UH
,
Thougaard
AV
,
Jensen
PB
,
Sehested
M
. 
A preclinical study on the rescue of normal tissue by nicotinic acid in high-dose treatment with APO866, a specific nicotinamide phosphoribosyltransferase inhibitor
.
Mol Cancer Ther
2010
;
9
:
1609
17
.
18.
Singh
J
,
Zabka
TS
,
Uppal
H
,
Diaz
D
,
Tarrant
J
,
Clarke
E
, et al
Effects of Nicotinamide Phosphoribosyltransferase inhibitors on platelet development
.
Society of Toxicology Meeting
; 
2012
;
San Antonio, TX
.
19.
Olesen
UH
,
Hastrup
N
,
Sehested
M
. 
Expression patterns of nicotinamide phosphoribosyltransferase and nicotinic acid phosphoribosyltransferase in human malignant lymphomas
.
APMIS
2011
;
119
:
296
303
.
20.
Wu
TD
,
Nacu
S
. 
Fast and SNP-tolerant detection of complex variants and splicing in short reads
.
Bioinformatics
2010
;
26
:
873
81
.
21.
Anders
S
,
Huber
W
. 
Differential expression analysis for sequence count data
.
Genome Biol
2010
;
11
:
R106
.
22.
Du
P
,
Kibbe
WA
,
Lin
SM
. 
Lumi: a pipeline for processing Illumina microarray
.
Bioinformatics
2008
;
24
:
1547
8
.
23.
Walter
K
,
Holcomb
T
,
Januario
T
,
Du
P
,
Evangelista
M
,
Kartha
N
, et al
DNA methylation profiling defines clinically relevant biological subsets of non-small cell lung cancer
.
Clin Cancer Res
2012
;
18
:
2360
73
.
24.
Krueger
F
,
Andrews
SR
. 
Bismark: a flexible aligner and methylation caller for Bisulfite-Seq applications
.
Bioinformatics
2011
;
27
:
1571
2
.
25.
Li
H
,
Handsaker
B
,
Wysoker
A
,
Fennell
T
,
Ruan
J
,
Homer
N
, et al
The Sequence Alignment/Map format and SAMtools
.
Bioinformatics
2009
;
25
:
2078
9
.
26.
Zheng
X
,
Bauer
P
,
Baumeister
T
,
Buckmelter
AJ
,
Caligiuri
M
,
Clodfelter
KH
, et al
Structure-based discovery of novel amide-containing nicotinamide phosphoribosyltransferase (Nampt) inhibitors
.
J Med Chem
2013
;
56
:
6413
33
.
27.
Herman
JG
,
Merlo
A
,
Mao
L
,
Lapidus
RG
,
Issa
JP
,
Davidson
NE
, et al
Inactivation of the CDKN2/p16/MTS1 gene is frequently associated with aberrant DNA methylation in all common human cancers
.
Cancer Res
1995
;
55
:
4525
30
.
28.
Graff
JR
,
Herman
JG
,
Lapidus
RG
,
Chopra
H
,
Xu
R
,
Jarrard
DF
, et al
E-cadherin expression is silenced by DNA hypermethylation in human breast and prostate carcinomas
.
Cancer Res
1995
;
55
:
5195
9
.
29.
Landan
G
,
Cohen
NM
,
Mukamel
Z
,
Bar
A
,
Molchadsky
A
,
Brosh
R
, et al
Epigenetic polymorphism and the stochastic formation of differentially methylated regions in normal and cancerous tissues
.
Nat Genet
2012
;
44
:
1207
14
.
30.
Dawson
MA
,
Kouzarides
T
. 
Cancer epigenetics: from mechanism to therapy
.
Cell
2012
;
150
:
12
27
.
31.
Brown
NJ
,
Higham
SE
,
Perunovic
B
,
Arafa
M
,
Balasubramanian
S
,
Rehman
I
. 
Lactate dehydrogenase-B is silenced by promoter methylation in a high frequency of human breast cancers
.
PloS ONE
2013
;
8
:
e57697
.
32.
Pfeifer
GP
,
Yoon
JH
,
Liu
L
,
Tommasi
S
,
Wilczynski
SP
,
Dammann
R
. 
Methylation of the RASSF1A gene in human cancers
.
Biol Chem
2002
;
383
:
907
14
.
33.
Belinsky
SA
. 
Gene-promoter hypermethylation as a biomarker in lung cancer
.
Nat Rev Cancer
2004
;
4
:
707
17
.
34.
Brock
MV
,
Hooker
CM
,
Ota-Machida
E
,
Han
Y
,
Guo
M
,
Ames
S
, et al
DNA methylation markers and early recurrence in stage I lung cancer
.
N Engl J Med
2008
;
358
:
1118
28
.
35.
Warren
JD
,
Xiong
W
,
Bunker
AM
,
Vaughn
CP
,
Furtado
LV
,
Roberts
WL
, et al
Septin 9 methylated DNA is a sensitive and specific blood test for colorectal cancer
.
BMC Med
2011
;
9
:
133
.
36.
Nikolaidis
G
,
Raji
OY
,
Markopoulou
S
,
Gosney
JR
,
Bryan
J
,
Warburton
C
, et al
DNA methylation biomarkers offer improved diagnostic efficiency in lung cancer
.
Cancer Res
2012
;
72
:
5692
701
.
37.
Amatu
A
,
Sartore-Bianchi
A
,
Moutinho
C
,
Belotti
A
,
Bencardino
K
,
Chirico
G
, et al
Promoter CpG island hypermethylation of the DNA repair enzyme MGMT predicts clinical response to dacarbazine in a phase II study for metastatic colorectal cancer
.
Clin Cancer Res
2013
;
19
:
2265
72
.
38.
Dobrovic
A
,
Simpfendorfer
D
. 
Methylation of the BRCA1 gene in sporadic breast cancer
.
Cancer Res
1997
;
57
:
3347
50
.
39.
Mancini
DN
,
Rodenhiser
DI
,
Ainsworth
PJ
,
O'Malley
FP
,
Singh
SM
,
Xing
W
, et al
CpG methylation within the 5′ regulatory region of the BRCA1 gene is tumor specific and includes a putative CREB binding site
.
Oncogene
1998
;
16
:
1161
9
.
40.
Ibragimova
I
,
Cairns
P
. 
Assays for hypermethylation of the BRCA1 gene promoter in tumor cells to predict sensitivity to PARP-inhibitor therapy
.
Methods Mol Biol
2011
;
780
:
277
91
.