- Open Access
Glyceraldehyde-3-phosphate dehydrogenase gene over expression correlates with poor prognosis in non small cell lung cancer patients
© Puzone et al.; licensee BioMed Central Ltd. 2013
- Received: 18 April 2013
- Accepted: 18 June 2013
- Published: 29 August 2013
Glycolysis in presence of oxygen with high glucose consumption is known to be the metabolism of choice in many tumors. In lung cancer this phenomenon is routinely exploited in diagnostic PET imaging of fluorodeoxyglucose uptake, but not much is known about the prognostic capabilities of glycolysis level assessment in resected lung tumor samples.
In this retrospective study, we used real time polymerase chain reaction(RQ-PCR) to assess the expression level of the gene for Glyceraldehyde 3-phosphate dehydrogenase(GAPDH), key enzyme for glucose breakdown, in tumor samples from 82 consecutive early stages resected non small cell lung cancer(NSCLC) patients. We then compared our results in six large publicly available NSCLC microarray datasets collecting data from over 1250 total patients.
In our study GAPDH gene over expression was found to be an adverse prognostic factor in early stages NSCLC (n = 82 HR = 1.30 p = 0.050). This result was confirmed in 5 of 6 public datasets analyzed: Shedden et al. 2008: n = 442 HR = 1.54 p < 0.0001; Lee et al. 2008: n = 138 HR = 1.31 p = 0.043; Tomida et al. 2009: n = 117 HR = 1.59 p = 0.004; Roepman et al. 2009: n = 172 (TPI1 gene) HR = 1.51 p = 0.009; Okayama et al. 2012: n = 226 HR = 3.19 p < 0.0001; Botling et al. 2013: n = 196 HR = 1.00 p = 0.97). Furthermore, in the large and clinically well annotated Shedden et al. microarray dataset, GAPDH hazard ratio did not change whether calculated for the whole dataset or for the subgroup of adjuvant naive patients only (n = 330 HR = 1.49 p < 0.0001).
GAPDH gene over expression in resected tumor samples is an adverse prognostic factor in NSCLC. Our results confirm the prognostic value of glucose metabolism assessment in NSCLC.
- Warburg effect
- Gene expression microarrays
- Non small cell lung cancer prognosis
Cancer cell metabolism characterized by high glycolysis rate in presence of oxygen has been confirmed in many tumors . This phenomenon, discovered by O. Warburg in 1924  and once considered as the result of a “damaged” metabolism , has presently been found also in many rapidly multiplying non-cancerous cells, leading to an increased focus of cancer research on the specific characteristics of tumor metabolism .
This field of cancer research is promising. In fact the high glycolysis rate in tumors, as assessed by diagnostic positron emission tomography (PET) imaging of fluorodeoxyglucose (FDG) uptake, is also exploited in clinical practice, in the differential diagnosis of nodules of unknown origin, and, more recently, also in prognostic studies [5–7]. However, specific investigations must be performed because we can expect that tumors with different characteristics-origin, grow dynamics, etc.-have different metabolic requirements.
Diagnostic PET imaging is routinely performed in NSCLC, the most frequent histological type of lung cancer (still the leading cause of cancer death in the world ). There is evidence that high glucose metabolism is present in NSCLC, so a role of metabolism as prognostic factor can be hypothesized; in fact this role is actually investigated in lung cancer by the assessment of FDG uptake level [6, 9, 10].
New effective prognostic factors could be very useful for NSCLC patients. Presently, pathological stage of the resected tumor is the main prognostic factor used in clinical practice to select NSCLC patients to be referred for additional therapies after surgery , but many early staged patients actually relapse . In fact, many proteins or genes, differently expressed in tumor samples from patients with different survivals, are investigated as possible prognostic biomarkers; but NSCLC is probably a very heterogeneous disease  and this could justify the high number of mostly non-overlapping gene lists proposed as prognostic signatures . However, PET effectiveness in distinguishing NSCLC from non-tumor lung tissue suggests that genes related to glucose metabolism bear an important role in all NSCLC, regardless of tumor heterogeneity.
Among these genes, GAPDH has an essential role in glucose metabolism, where the corresponding enzyme converts glyceraldehydes-3-phosphate to 1,3-diphosphoglycerate with reduction of nicotinamide adenine dinucleotide (NAD+) to NADH. In fact GAPDH gene is expressed in all tissue, so to be classically used as housekeeping gene, but it is known to be over expressed in many tumors as compared to normal tissues, and also to be correlated with poor prognosis or tumor aggressiveness in ovarian, breast, renal, colorectal, melanoma cancer . Furthermore, GAPDH protein is able to bind to RNA and DNA, supporting glycolytic and extra-glycolytic regulatory roles in cell stress, apoptosis, and metabolism [16–18]. In lung cancer, GAPDH protein is well known to be over expressed as compared to normal lung tissue , and GAPDH gene is known to be expressed at high levels as compared to the surrounding non cancerous lung biopsies . However, while evidences accumulate that preoperative FDG uptake level is a prognostic factor in NSCLC, the prognostic value of GAPDH expression level in resected NSCLC samples is still to be assessed. In this retrospective study, we measured GAPDH gene expression, by RQ-PCR, on tumor samples from a group 82 resected NSCLC patients. After detecting a significant correlation of GAPDH with survival from our patient follow-up data, we decided to further investigate the expression of GAPDH gene in six large publicly available NSCLC microarray datasets, collecting data from over 1250 total NSCLC patients.
IST patient’s characteristics
Number of patients
Median Age, Years (min-max)
Reverse transcription and RQ-PCR
RNA was isolated from paraffin-embedded tumor samples using the High Pure FFPE RNA Micro Kit (Roche Applied Science, Mannheim, Germany) with minor modifications. RNA were reverse-transcribed with SuperScript™ II RT (Invitrogen, Grand Island, NY, USA) according to the manufacturer’s instructions. Resulting cDNA was amplified by the LightCycler 480 Real Time PCR System II (Roche Applied Science). Relative gene expression levels were calculated using the Qgene software  featuring an efficiency corrected threshold cycle based algorithm. Beta-2-microglobulin (B2M) and beta-glucuronidase (GUSB) were used as housekeeping genes and a virtual housekeeping gene was calculated using BestKeeper software . PCR primer sequences are reported in Additional file 1.
Overall survival analysis for GAPDH RQ-PCR gene expression, with hazard ratio (HR) and confidence interval (CI) calculation, were performed on our patient data by using the Cox regression model. Multivariate Cox regression was performed with adjusting for tumor stage categorized in 3 classes (I – II - III). The Kaplan-Meier curve was plotted by separating patients on the median GAPDH gene expression level. Cumulative survivals were assessed by the Kaplan-Meier model. All calculations and plots were performed by using R 2.14(64bit) software . Gene expression, survival data and sample R code are reported in Additional file 1.
Comparison with the public microarray datasets
Summary of characteristics of the public microarray datasets compared with IST patients
Age median (min-max)
5-years cumulative survival (95% CI)
ADK SCC other
 Shedden et al. (2008) all patients
 Shedden et al. (2008) adjuvant-naïve
 Lee et al. (2008)
 Tomida et al. (2009)
 Roepman et al. (2009)
ADK SCC other
 Okayama et al. (2012)
 Botling at al. (2013)
Statistical analysis for the datasets was performed similarly as for our patient RQ-PCR data. A single GAPDH gene level was calculated for each patient sample as the mean level of all its probes mapped to GAPDH. Stage was categorized in 3 classes (I-II-III) or 2 classes (I – II + III) when patient numbers were low in higher stages. For the Le2008 dataset, relapse free survival data was used in regressions due to overall survival data unavailability. For the Ro2009 dataset, a large and clinically well annotated dataset, the probe annotation file had no GAPDH gene reference; we decided to use, at least in part, this dataset in the present work by analyzing the gene expression for the strictly metabolically related triosephosphate isomerase 1 (TPI1), that was found in fact highly correlated with GAPDH gene in all other microarray datasets (Pearson's r > 0.79 for GAPDH-TPI1 expression levels). Also, patient characteristics and clinical data accuracy were diverse among the microarray datasets; so we performed the survival analysis separately for each microarray dataset, and reported the results using a forest plot  style comparison with our patient data, without pooling the datasets. All calculations and plots were performed by using R 2.14(64bit) software  and Bioconductor libraries .
GAPDH gene expression level and correlation with survival in our patients
Patient and cancer characteristics are reported in Table 1. None of the 82 patients was lost to follow-up. During a median follow-up time of 5 years, 37 (45%) deaths were observed; cumulative survival was 89%, 73%, 65%, 56% and 54%, respectively at 1, 2, 3, 4 and 5 years. In univariate analysis age, sex, or smoking history had no correlation with survival. Only tumor stage was statistically associated with survival (stage I-II HR2.82 p = 0.019; I-III HR4.44 p = 0.0001); the median survival times were “not reached”, 2.69 and 1.80 years, for tumor stage I, II and III, respectively.
In multivariate Cox model adjusted for stage, GAPDH HR value was lower than in univariate model and not significant (HR1.20; 95%CI .89-1.63; p = 0.23) (Figure 1B, forest plot top line), and only tumor stage I-III was significantly correlated with survival (stage I-II HR2.36 p = 0.069; I-III HR4.22 p = 0.0002).
Verification in the public microarray datasets
Cox regression analysis for GAPDH gene expression in the microarrays datasets are summarized in the two forest plots (Figure 1A and B, before and after adjusting for tumor stage in the model, respectively), and compared with GAPDH results for our patient (IST) . Cumulative survivals and dataset sizes are also reported in the plots.
By comparing our regression results before and after adjusting for tumor stage (respectively Figure 1A and B), it results that HR for GAPDH gene expression was mostly independent from stage in microarray datasets, while in our patient data (IST), after adjusting for tumor stage, HR value was decreased and not significant. In multivariate Cox proportional hazard model together with GAPDH gene expression, tumor stage HR values were found high (as expected) for most microarray datasets (stage: Sh2008 I-II HR2.60 p < 0.0001, I-III HR4.78 p < 0.0001;To2009 I-II + III HR2.29 p = 0.004; Ro2009-TPI1 I-II + III HR2.13 p = 0.011; Ok2012 I-II HR1.69 p = 0.2). However, for Bo2013 dataset, tumor stage HR values were lower than in the other datasets, and not significant for stage I-II despite the high patient and event numbers (stage Bo2013 I-II HR1.28 p = 0.32; I-III HR1.88 p = 0.01).
In agreement with the rationale provided in the Methods section, the TPI1 gene HR value and CI in Ro2009 dataset were found to be very similar to the GAPDH gene results in the other datasets.
Comparing the subset of the patients that did not receive adjuvant treatments, with the whole dataset in the Sh2008, we found that GAPDH HR was pretty unchanged (GAPDH: untreated patients HR1.49, whole dataset HR1.54, Figure 1A). The subset containing adjuvant treated patients only had a significantly lower cumulative survival (0.38, 95%0.30-0.50 N = 112) than the untreated patient subset (0.64, 95%.58-.70, N = 330), but GAPDH HR was still unchanged (adjuvant-treated GAPDH HR1.57, 95%1.22-2.03, p < 0.0001). Finally, in random sampling of subgroups of patients from Sh2008, we observed that GAPDH HR was not specifically affected when calculated in subsets featuring low cumulative survivals (Additional file 2).
It is well known that lung tumors present with high glycolysis level, but it is yet to demonstrate that glycolysis level, as assessed in resected NSCLC patient tumor sample, can be a prognostic factor; we think that our results gave some evidence suggesting its prognostic capabilities. In the present study we assessed the gene expression level of GAPDH, that has a key role in glucose breakdown; with our surprise, we found no studies specifically addressing the prognostic capabilities of GAPDH gene expression in resected NSCLC samples.
GAPDH protein is known to have also extra-glycolytic capabilities, being able to move to the nucleus, to support cell response to stress, and to initiate apoptosis . However GAPDH gene is always expressed at high levels, with high glycolysis levels, in NSCLC compared to normal lung cells; so we think that our GAPDH prognostic results reflect an increased catalytic activity of GAPDH protein in glucose metabolism. In this sense our results are in agreement with the studies that are correlating glucose metabolism to NSCLC prognosis by using different approaches, among which FDG uptake level assessment by PET imaging of the tumor before resection. Furthermore, on the same reasoning, many studies in NSCLC are recently addressing the prognostic value of other key proteins or gene involved in glucose metabolism, e.g. GLUT1, HK2 . In fact, it is still unknown which aspects of glycolysis have strong prognostic value in NSCLC, but many available evidences, including our present study results, support that the level of glycolysis has indeed prognostic value.
In our study we measured RQ-PCR GAPDH gene expression levels in the resected tumors from 82 patients of our hospital and found a significant correlation with their prognosis. Then we decided to verify this correlation in the largest NSCLC public microarray datasets, and we found a confirmation of our result. We showed all results in forest plot style, for an individual comparison. In fact, not all the available public data feature the same accuracy; especially some datasets, e.g. Sh2008, are better annotated so to be used as a reference in many studies. Among the confirmations coming from the microarray datasets, we think that the Sh2008 data gave a strong support to our results.
Our results for GAPDH also agree with the findings of a very recent paper from Wang et al.  in which the authors show the prognostic value of some genes correlated with GAPDH (GACC genes) together with GAPDH itself; Sh2008 was used as verification dataset. Authors don't show the prognostic performance of GAPDH alone; however, our results, confirmed on a large number of public datasets including Sh2008, suggest that large part of the prognostic performances shown in Sh2008 have to be attributed to GAPDH alone.
In the forest plots we showed the Ro2009 dataset results too, by plotting its TPI1 gene levels instead of the unavailable GAPDH ones. Actually this substitution was based on the strict metabolic relation between the two catalytic proteins – however the high correlation of the two genes was verified in the other datasets, and is confirmed by other authors too . So, Ro2009 results for TPI1, very similar to GAPDH results in the other datasets, can further support that the prognostic capabilities of GAPDH in NSCLC reflect the role of the corresponding enzyme in glucose metabolism.
However one dataset (Bo2013) had a null result for GAPDH correlation with prognosis (HR = 1.0); this dataset was also featuring some characteristics different from all the other ones: i) a low cumulative survival, also at low tumor stages, and ii) a low tumor stage HR and significance, despite the high patient and event numbers. We have no data supporting a correlation of these characteristics with a strong decrease of HR values for GAPDH, so we can only conclude that the Bo2013 dataset is different from the other datasets from more than a single point of view.
GAPDH HR was not affected when selecting only patients that had not received any adjuvant therapy; we performed this comparison in the Sh2008 dataset. This result was helpful for our data analysis; in fact our patients had not received radiotherapy or chemotherapy, but in most microarray datasets the information, whether adjuvant treatments had been performed or not, was not available at patient level. Actually, adjuvant treatment presence could confound a survival analysis because there is-finally- evidence that it can increase survival also in lower stages patient . Furthermore, clinicians select patients with presumed poor prognosis to be referred for adjuvant therapies-in fact patient selection is one of the main reasons why retrospective studies cannot address adjuvant treatment effectiveness; this selection was resulting in the low cumulative survival found in Sh2008 adjuvant treated patient only subset. However, we observed that this selection probably did not much influence GAPDH HR value (Additional file 2). So, GAPDH HR insensitivity to the presence of adjuvant treatments suggests that GAPDH is still a prognostic factor in adjuvant treated patients, but is not promising as predictive factor of adjuvant effectiveness, as performed in Sh2008 patients.
However, in more recent years, some anti-tumor drugs under investigation are involving tumor metabolism, e.g. by reducing glucose availability as metformin , or by directly targeting glycolysis proteins ; our results suggest that in clinical investigations on these drugs, GAPDH levels in resected NSCLC samples should be investigated as possible predictor of treatment effectiveness.
From the clinical point of view the GAPDH HR value found in our patients is interesting; however after tumor stage adjusting, significance was lost, pointing out that GAPDH gene expression had some correlation with tumor stage. Indeed, adjusting for tumor stage in the regression model had small effect on HR calculation in microarray datasets, suggesting that our patient number was simply critically too low to overpass the significance level for HR after adjusting for stage, but that GAPDH HR is for large part independent from stage. It will be therefore interesting to investigate how GAPDH could contribute with FDG uptake level and tumor stage in building a composite prognostic marker, possibly also correlating it with the status of known NSCLC oncogenic genes (PI3K, EGFR, KRAS, ALK, etc.).
Finally, not only our results warn researchers from using GAPDH as housekeeper gene in NSCLC prognostic studies involving RQ-PCR measurements; we also suggest that any past NSCLC prognostic study using GAPDH as housekeeper gene should be considered potentially biased.
In conclusion, GAPDH gene expression level in resected tumor, as assessed by RQ-PCR or microarray, is an important prognostic factor in NSCLC, that confirms the importance of investigating metabolism in lung cancer.
We acknowledge Dr. Paolo Bruzzi for useful suggestions in the manuscript structure and data interpretation. This work has been supported by a grant from Ricerca Finalizzata 2009 (RF-2009-1530324) – Ministero della Salute, Italy.
- Koppenol WH, Bounds PL, Dang CV: Otto Warburg's contributions to current concepts of cancer metabolism. Nat Rev Cancer. 2011, 11: 325-337.View ArticlePubMedGoogle Scholar
- Warburg O, Posener K, Negelein E: Ueber den Stoffwechsel der Tumoren. Biochem Zeitschrift. 1924, 152: 319-344.Google Scholar
- Warburg O: On the origin of cancer cells. Science. 1956, 123: 309-314.View ArticlePubMedGoogle Scholar
- Cairns RA, Harris IS, Mak TW: Regulation of cancer cell metabolism. Nat Rev Cancer. 2011, 11: 85-95.View ArticlePubMedGoogle Scholar
- Hyun SH, Choi JY, Shim YM, Kim K, Lee SJ, Cho YS, Lee JY, Lee K, Kim B: Prognostic value of metabolic tumor volume measured by 18f-fluorodeoxyglucose positron emission tomography in patients with esophageal carcinoma. Ann Surg Oncol. 2010, 17: 115-122.View ArticlePubMedGoogle Scholar
- Paesmans M, Berghmans T, Dusart M, Garcia C, Hossein-Foucher C: Primary tumor standardized uptake value measured on fluorodeoxyglucose positron emission tomography is of prognostic value for survival in non-small cell lung cancer: update of a systematic review and meta-analysis by the European Lung Cancer Working Party for the International Association for the Study of Lung Cancer Staging Project. J Thorac Oncol. 2010, 5: 612-619.PubMedGoogle Scholar
- Chung HH, Kwon HW, Kang KW, Park NH, Song YS, Chung JK, Kang SB, Kim JW: Prognostic value of preoperative metabolic tumor volume and total lesion glycolysis in patients with epithelial ovarian cancer. Ann Surg Oncol. 2012, 19 (6): 1966-1972.View ArticlePubMedGoogle Scholar
- Jemal A, Bray F, Center MM, Ferlay J, Ward E, Forman D: Global cancer statistic. CA Cancer J Clin. 2011, 61 (2): 69-90.View ArticlePubMedGoogle Scholar
- Hyun SH, Choi JY, Kim K, Kim J, Shim YM, Um S, Kim H, Lee K, Kim B: Volume-Based Parameters of (18)F-fluorodeoxyglucose Positron Emission Tomography/Computed Tomography Improve Outcome Prediction in Early-Stage Non-Small Cell Lung Cancer After Surgical Resection. Ann Surg. 2013, 257: 364-370.View ArticlePubMedGoogle Scholar
- Nair VS, Gevaert O, Davidzon G, Napel S, Graves EE: Prognostic PET 18F-FDG uptake imaging features are associated with major oncogenomic alterations in patients with resected non-small cell lung cancer. Cancer Res. 2012, 72: 3725-3734.PubMed CentralView ArticlePubMedGoogle Scholar
- , : Non-Small Cell Lung Cancer version 2. 2012, [http://www.nccn.org/professionals/physician_gls/pdf/nscl.pdf]Google Scholar
- Pignon J, Tribodet H, Scagliotti GV, Douillard J, Shepherd FA: Lung adjuvant cisplatin evaluation: a pooled analysis by the LACE Collaborative Group. J Clin Oncol. 2008, 26: 3552-3559.View ArticlePubMedGoogle Scholar
- Gazdar AF: Should we continue to use the term non-small-cell lung cancer?. Ann Oncol. 2010, 21 (Suppl 7): vii225-vii229.PubMed CentralPubMedGoogle Scholar
- Subramanian J, Simon R: Gene expression-based prognostic signatures in lung cancer: ready for clinical use?. J Natl Cancer Inst. 2010, 102: 464-474.PubMed CentralView ArticlePubMedGoogle Scholar
- Guo C, Liu S, Sun M: Novel insight into the role of gapdh playing in tumor. Clin Transl Oncol. 2013, 15 (3): 167-172.View ArticlePubMedGoogle Scholar
- Zheng L, Roeder RG, Luo Y: S phase activation of the histone h2b promoter by oca-s, a coactivator complex that contains gapdh as a key component. Cel. 2003, 114: 255-266. 10.1016/S0092-8674(03)00552-X.View ArticleGoogle Scholar
- Ralser M, Wamelink MM, Kowald A, Gerisch B, Heeren G, Struys EA, Klipp E, Jakobs C, Breitenbach M, Lehrach H, Krobitsch S: Dynamic rerouting of the carbohydrate flux is key to counteracting oxidative stress. J Biol. 2007, 6: 10-PubMed CentralView ArticlePubMedGoogle Scholar
- Hara MR, Agrawal N, Kim SF, Cascio MB, Fujimuro M, Ozeki Y, Takahashi M, Cheah JH, Tankou SK, Hester LD, Ferris CD, Hayward SD, Snyder SH, Sawa A: S-nitrosylated gapdh initiates apoptotic cell death by nuclear translocation following siah1 binding. Nat Cell Biol. 2005, 7: 665-674.View ArticlePubMedGoogle Scholar
- Cuezva JM, Chen G, Alonso AM, Isidoro A, Misek DE, Hanash SM, Beer DG: The bioenergetic signature of lung adenocarcinomas is a molecular marker of cancer diagnosis and prognosis. Carcinogenesis. 2004, 25: 1157-1163.View ArticlePubMedGoogle Scholar
- Tokunaga K, Nakamura Y, Sakata K, Fujimori K, Ohkubo M, Sawada K, Sakiyama S: Enhanced expression of a glyceraldehyde-3-phosphate dehydrogenase gene in human lung cancers. Cancer Res. 1987, 47: 5616-5619.PubMedGoogle Scholar
- Mueller PY, Janovjak H, Miserez AR, Dobbie Z: Processing of gene expression data generated by quantitative real-time RT-PCR. Biotechniques. 2002, 32: 1372-1378.Google Scholar
- Pfaffl MW, Tichopad A, Prgomet C, Neuvians TP: Determination of stable housekeeping genes, differentially regulated target genes and sample integrity: BestKeeper–Excel-based tool using pair-wise correlations. Biotechnol Lett. 2004, 26: 509-515.View ArticlePubMedGoogle Scholar
- The R Project for Statistical Computing: [http://www.R-project.org]
- Shedden K, Taylor JMG, Enkemann SA, Tsao M, Yeatman TJ, Gerald WL, Eschrich S, Jurisica I, Giordano TJ, Misek DE, Chang AC, Zhu CQ, Strumpf D, Hanash S, Shepherd FA, Ding K, Seymour L, Naoki K, Pennell N, Weir B, Verhaak R, Ladd-Acosta C, Golub T, Gruidl M, Sharma A, Szoke J, Zakowski M, Rusch V, Kris M, Viale A, Motoi N, Travis W, Conley B, Seshan VE, Meyerson M, Kuick R, Dobbin KK, Lively T, Jacobson JW, Beer DG, : Gene expression-based survival prediction in lung adenocarcinoma: a multi-site, blinded validation study. Nat Med. 2008, 14: 822-827.PubMed CentralView ArticlePubMedGoogle Scholar
- Lee E, Son D, Kim S, Lee J, Jo J, Han J, Kim H, Lee HJ, Choi HY, Jung Y, Park M, Lim YS, Kim K, Shim Y, Kim BC, Lee K, Huh N, Ko C, Park K, Lee JW, Choi YS, Kim J: Prediction of recurrence-free survival in postoperative non-small cell lung cancer patients by using an integrated model of clinical information and gene expression. Clin Cancer Res. 2008, 14: 7397-7404.View ArticlePubMedGoogle Scholar
- Tomida S, Takeuchi T, Shimada Y, Arima C, Matsuo K, Mitsudomi T, Yatabe Y, Takahashi T: Relapse-related molecular signature in lung adenocarcinomas identifies patients with dismal prognosis. J Clin Oncol. 2009, 27: 2793-2799.View ArticlePubMedGoogle Scholar
- Roepman P, Jassem J, Smit EF, Muley T, Niklinski J, van de Velde T, Witteveen AT, Rzyman W, Floore A, Burgers S, Giaccone G, Meister M, Dienemann H, Skrzypski M, Kozlowski M, Mooi WJ, van Zandwijk N: An immune response enriched 72-gene prognostic profile for early-stage non-small-cell lung cancer. Clin Cancer Res. 2009, 15: 284-290.View ArticlePubMedGoogle Scholar
- Okayama H, Kohno T, Ishii Y, Shimada Y, Shiraishi K, Iwakawa R, Furuta K, Tsuta K, Shibata T, Yamamoto S, Watanabe S, Sakamoto H, Kumamoto K, Takenoshita S, Gotoh N, Mizuno H, Sarai A, Kawano S, Yamaguchi R, Miyano S, Yokota J: Identification of genes upregulated in alk-positive and egfr/kras/alk-negative lung adenocarcinomas. Cancer Res. 2012, 72: 100-111.View ArticlePubMedGoogle Scholar
- Botling J, Edlund K, Lohr M, Hellwig B, Holmberg L, Lambe M, Berglund A, Ekman S, Bergqvist M, Pontén F, König A, Fernandes O, Karlsson M, Helenius G, Karlsson C, Rahnenführer J, Hengstler JG, Micke P: Biomarker discovery in non-small cell lung cancer: integrating gene expression profiling, meta-analysis, and tissue microarray validation. Clin Cancer Res. 2013, 19: 194-204.View ArticlePubMedGoogle Scholar
- Wu Z, Irizarry RA: Processing of oligonucleotide array data. Nat Biotechnol. 2004, 22 (6): 656-658.View ArticlePubMedGoogle Scholar
- Bryant CM, Albertus DL, Kim S: Clinically relevant characterization of lung adenocarcinoma subtypes based on cellular pathways: an international validation study. PLoS One. 2010, 5 (7): e11712-PubMed CentralView ArticlePubMedGoogle Scholar
- Lewis S, Clarke M: Forest plots: trying to see the wood and the trees. BMJ. 2001, 322 (7300): 1479-1480.PubMed CentralView ArticlePubMedGoogle Scholar
- Gentleman R, Carey VJ, Bates DM, Bolstad B, Dettling M, Dudoit S, Ellis B, Gautier L, Ge Y, others : Bioconductor: Open software development for computational biology and bioinformatics. Genome Biol. 2004, 5: R80-PubMed CentralView ArticlePubMedGoogle Scholar
- Wang D, Moothart DR, Lowy DR, Qian X: The expression of glyceraldehyde-3-phosphate dehydrogenase associated cell cycle (gacc) genes correlates with cancer stage and poor survival in patients with solid tumors. PLoS One. 2013, 8: e61262-PubMed CentralView ArticlePubMedGoogle Scholar
- Pierotti MA, Berrino F, Gariboldi M, Melani C, Mogavero A, Negri T, Pasanisi P, Pilotti S: Targeting metabolism for cancer treatment and prevention: metformin, an old drug with multi-faceted effects. Oncogene. 2013, 32: 1475-1487.View ArticlePubMedGoogle Scholar
- Tennant DA, Durán RV, Gottlieb E: Targeting metabolic transformation for cancer therapy. Nat Rev Cancer. 2010, 10: 267-277.View ArticlePubMedGoogle Scholar
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.