Eukaryotic initiation factor 4Eoverexpression in triple-negative breastcancer predicts a worse outcomeAnthony Flowers, MS,e Quyen D. Chu, MD,a Lori Panu, BS,b Carol Meschonat, BA,a Gloria Caldito,PhD,c Mary Lowery-Nordberg, PhD,d and Benjamin D. L. Li, MD,a Shreveport, LA

Background. Triple-negative (estrogen receptor [ER], progesterone receptor [PR], and HER2/neureceptor negative) breast neoplasms are typically high grade and portend a higher risk for relapse.Not being amendable to ER, PR, or HER2-targeted therapy, adjuvant cytotoxic chemotherapy remainsthe only option. High eukaryotic Initiation Factor 4E (eIF4E) overexpression in tumor specimens isan independent predictor for relapse in breast cancer, perhaps secondary to tousled-like kinase 1Bupregulation and subsequent doxorubicin resistance. In this prospective study, eIF4E elevation intriple-negative breast cancer (TNBC) specimens was studied to determine its effect on cancer outcome.Methods. A prospective study of 103 TNBC patients was initiated. Tumor specimens were quantified foreIF4E expression using Western blots. Clinical outcomes data were collected after standardized adjuvanttreatment and surveillance protocols. Primary end points were cancer recurrence and cancer-relateddeath. The eIF4E levels in cancer specimens were quantified as x-fold over benign samples fromnoncancer patients. Statistical procedures performed include survival analysis by Kaplan--Meiermethod, log-rank test, Cox proportional hazards regression model, and the chi-square test.Results. Levels of eIF4E were categorized into 3 tertiles. Among 103 patients, 36 were in the lowgroup (#7.5-fold), 40 were in the intermediate group (7.5- to 15-fold), and 27 were in the high group($15-fold). Patients with triple-negative neoplasms that were in the high eIF4E group had greater ratesof cancer recurrence (P = .04) and cancer-related death (P = .02) than the low eIF4E group. Amongpatients with node-negative disease, high eIF4E overexpression in tumor specimens continues to portenda greater rate of cancer recurrence (P = .02), and a higher rate of cancer death (P = .03) than those inthe low eIF4E group.Conclusion. TNBC patients with high eIF4E overexpression are more likely to recur and die from cancerrecurrence. High eIF4E seems to be a significant prognostic marker, even in TNBC patients. (Surgery2009;146:220-6.)

From the Departments of Surgery,a Medical Oncology,b Biometry,c Pathology,d and the School of Medicine,e

Louisiana State University Health Science Center in Shreveport and the Feist-Weiller Cancer Center,Shreveport, LA

TRIPLE-NEGATIVE (estrogen receptor [ER], progester-one receptor [PR], and HER2-negative) breast tu-mors are primary breast cancers that are typicallybasal-like and portend a higher risk for relapse.1,2

Triple-negative breast cancers (TNBC) accountfor approximately 10--15% of breast cancercases.1,3,4 Treatment options for TNBC include an-thracycline- and taxane-based chemotherapy.3,5

Presented at the 2009 Academic Surgical Congress, Fort Myers,Florida.

Accepted for publication May 7, 2009.

Reprint requests: Anthony Flowers, MS, Box 188, LouisianaState University Health Sciences Center, 1501 Kings Highway,Shreveport, LA 71130. E-mail: aflow1@lsuhsc.edu.

0039-6060/$ - see front matter

� 2009 Mosby, Inc. All rights reserved.

doi:10.1016/j.surg.2009.05.010

220 SURGERY

Studies have shown that TNBCs are more likelyto have a complete response to neoadjuvant ther-apy; however, those who do not fully respond arelikely to relapse earlier and experience a worseoutcome.3,5

Eukaryotic initiation factor 4E (eIF4E) binds tomRNAs with long 59 untranslated regions (UTRs)and preferentially upregulates translation of thesemRNAs.6-8 Elevated eIF4E expression has beenshown to upregulate gene products important inthe regulation of cell growth and differentiationsuch as cyclin D1, angiogenesis factors, vascular per-meability factor, and fibroblast growth factor.9,10 Pa-tients with stage I--III breast cancer and a morethan 7-fold elevation of eIF4E overexpression hadhigher recurrence rates and higher rates of can-cer-related death than those with less than 7-foldelevation.7,8 High eIF4E overexpression has been

SurgeryVolume 146, Number 2

Flowers et al 221

shown to be an independent predictor of cancer re-currence in patients with stage I to III breast carci-noma.8,11 Overexpression of eIF4E is associatedwith increased levels of vascular endothelial growthfactor and microvascular density counts.11 Thisleads to increased tumor vascularity and may playa role for worse outcome.11

De Benedetti et al first reported the upregula-tion of a Tousled-like kinase 1B (TLK1B) as part ofa cDNA subtraction library of mRNAs in cell lineswith eIF4E overexpression.12,13 TLK1B is the hu-man homolog of the tousled gene of the flowerArabidopsis thaliana, a highly conserved serine-threonine kinase.12,14 The 59UTR of the TLK1BmRNA sequence is 1088 nucleotides long; hence,upregulation of TLK1B has been correlated withthe overexpression of eIF4E, both in vitro and invivo.12 TLK1B elevation has been linked withhistone H3 protein stabilization, DNA repairalterations, and resistance to anthracycline-basedchemotherapy.13

Because TNBC as a group are not candidates forcurrent, target-specific therapy directed at ER, PR,or HER2 receptors, the only remaining optionis systemic chemotherapy, of which doxorubicin isa common agent of choice. Thus, the objectivein this study was to determine whether the degreeof eIF4E overexpression in triple-negative breastcarcinomas portends a worse cancer outcome.

METHODS

This study prospectively accrued 103 triple-negative stages I--III (TNM staging, American JointCommittee on Cancer15) breast cancer patientsbased on ER, PR, and HER-2/neu assays on cancerspecimens after definitive therapy. Exclusion crite-ria included neoadjuvant chemotherapy, prior ma-lignancies, and stage IV disease. Standardizedtreatment and surveillance protocols were fol-lowed. Patients were followed up in the clinic every3 months for the first 3 years, then every 6 monthsto year 5, and annually thereafter. At follow-up, sur-veillance included a complete physical examina-tion, chest x-ray, mammogram, complete bloodworkup, and liver function tests annually. Biopsiesof suspicions lesions and additional imaging weredirected by history and/or physical findings.

Specimen assay for eIF4E and protein level wasperformed using Western blot analysis as previ-ously described in detail.7,16 After institutional re-view board approved, informed consent wasobtained. One gram of breast cancer specimenwas collected at the time of operation and snapfrozen in liquid nitrogen.

Tissue lysate of tumor and benign samples weresolubolized in standard RIPA buffer. Total proteinof breast tumor tissue sample protein preparationwas determined using a bicinchroninic acid totalprotein assay (Pierce, Rockford, IL). We collected20 mg of protein per sample and run on a 10%SDS-PAGE gel (BioRad Lab, Inc., Hercules, CA)and transferred to a 22un Westran-S PVDF mem-brane (BioRad Lab, Inc.). The membrane wasprobed with anti-eIF4E (BD Biosciences Pharmin-gen, San Diego, CA). Anti-actin (Sigma, St. Louis,MO) was used as a loading control. Secondary an-tibodies were used as follows: goat anti-mouse im-munoglobulin G horseradish peroxidase (SantaCruz Biotechnology, Santa Cruz, CA) for eIF4Eand actin. Gels were color developed using theOpti4 CN Kit (BioRad Lab, Inc.). Band densitywas evaluated using the Bio-Image system (Ann Ar-bor, MI). Concentration of eIF4E protein was de-termined by band intensity of eIF4E over bandintensity of benign breast baseline (pool of normalbreast tissue from breast reduction tissue samples).

Immunohistochemical staining for ER and PRwas performed on a Ventana Autostainer (Bench-mark Corp., Tucson, AZ) using standard protocols.Slides were evaluated using the Automated Cellu-lar Imaging System (Chromavision MedicalSystems Inc., San Juan Capistrano, CA). ER andPR results were reported based on the degree andintensity of nuclear staining for all receptorsrespectively with a positive result defined as greaterthan 5% staining.

HER-2/neu was evaluated by fluorescencein-situ hybridization (Vysis; Abbott MolecularInc., Des Plaines, IL). Formalin-fixed, paraffin-embedded breast tissue sections between 4 and 6microns thick were probed with HER-2/neu DNAprobe, specific for the HER-2/neu gene locus(17q11.2-q12), and a chromosome enumerationprobe (CEP) 17 DNA probe, specific for the alphasatellite DNA sequence at the centromeric regionof chromosome 17. The probe and target DNA areco-denatured to single-stranded form and subse-quently allowed to hybridize with the PathVysionprobes. After hybridization, the unbound probe isremoved by a series of washes and the nuclei arecounterstained with DAPI, a DNA-specific stainthat fluoresces blue. Hybridization results of thePathVysion probes are viewed using a fluorescencemicroscope equipped with appropriate excitationand emission filters allowing visualization of theintense orange (HER-2/neu locus) and green(pericentromeric region of chromosome 17) fluo-rescent signals. Enumeration of the HER-2/neuand CEP 17 signals is conducted by microscopic

SurgeryAugust 2009

222 Flowers et al

examination of the tissue, which yields a ratio ofthe HER-2/neu gene to chromosome 17 copynumber. HER-2:CEP17 ratios of 2.0 or higher, areconsidered to be positive for HER-2/neu geneamplification.

The primary end points in this study are cancerrecurrence and cancer-related deaths. Statisticalanalysis performed included survival analysisby the Kaplan-Meier method, log-rank test, Coxproportional hazards regression model, and theChi-square test.

RESULTS

eIF4E levels were determined in the 103 breastspecimens using Western blot analysis. eIF4E levelsin cancer specimens were quantified as x-fold overbenign samples from noncancer patients. Fig 1shows a typical Western blot for quantification ofeIF4E levels. Lane 1 is a molecular ladder. Lanes2--7 are cancer specimen tissue lysates with varyingdegrees of eIF4E expression. Lanes 8 and 10 arebenign, and are used to determine x-fold overbenign samples from noncancer patients. Lane 9is a positive control for eIF4E.

Table I shows descriptive statistics on observedpatient characteristics. The mean age of diagnosiswas 49.4 years of age. The median follow-up was45 months. The mean eIF4E level of the 103 pa-tients was 11.1 ± 6.1-fold elevation (mean ± stan-dard deviation).

Patients were divided into tertiles based onpreviously published degree of eIF4E overexpres-sion.17 The low eIF4E group was defined as 7.5-foldor less elevation (n = 36), the intermediate eIF4Egroup was defined as between 7.5- and 15-fold ele-vation (n = 40), and the high eIF4E group was de-fined as 15-fold or greater elevation (n = 27).

Table II provides the eIF4E tertiles and cancer re-currence. Of the 30 recurrences, 8 are in the low, 8are in the intermediate, and 14 are in the hightertile. There was a significant greater number ofcancer recurrences in the high eIF4E tertileof TNBC patients than in the low eIF4E tertile(P = .01; Chi-square test).

Table III provides the eIF4E tertiles and cancer-related death. Of the 20 deaths, 4 are in the low, 5are in the intermediate, and 11 are in the high ter-tile. These findings were also significant (P = .005;Chi-square test).

Survival analysis by the Kaplan--Meier methodshowed TNBC patients with neoplasms in the higheIF4E group had a greater rate of cancer recur-rence than those in the low eIF4E group (P = .04;log-rank test; Fig 2). Similarly, TNBC patients withneoplasms in the high eIF4E group had a greater

rate of cancer-related deaths than patients in thelow eIF4E group (P = .02; Fig 3).

TNBC patients were then segregated into node-negative and node-positive patients. Node-negativeTNBC patients with neoplasms in the high eIF4Egroup had a higher rate of cancer recurrence thanthose in the low eIF4E group (P = .02; Fig 4, A).Similarly, node-negative TNBC patients with neo-plasms in the high eIF4E group also had a greaterrate of cancer-related deaths than patients in thelow eIF4E group (P = .03; Fig 4, B); however,node-positive TNBC patients with neoplasms inthe high eIF4E group did not have greater ratesof cancer recurrence (P = .8; Fig 5, A), or cancer-re-lated deaths (P = .4; Fig 5, B) than those in the loweIF4E group.

Fig 1. Lane 1 is a molecular ladder. Lanes 2 through 7are cancer specimen tissue lysates with varying degreesof eIF4E expression. Lanes 8 and 10 are benign, andare used to determine x-fold over benign samples fromnon-cancer patients. Lane 9 is a positive control foreIF4E.

Table I. Patient characteristics (n = 103)

Characteristic Mean ± SD*

Age at cancer diagnosis (yrs) 49.4 ± 10.4eIF4E (4E) protein value 11.1 ± 6.1Median follow up (mos) 45eIF4E (4E) tertile

1 (Low) <7.5-fold 362 (Intermediate) between 7.5-fold

and 15-fold40

3 (High) >15-fold 27

Table II. eIF4E tertile and breast cancerrecurrence

eIF4E fold overexpression Disease free Recurrence

Low (<7.5-fold; n = 36) 28 8Intermediate (between 7.5-fold

and 15-fold; n = 40)32 8

High (>15-fold; n = 27) 13 14Total (n = 103) 73 30

Values are number of patients.P = .02 (Chi-square test).

SurgeryVolume 146, Number 2

Flowers et al 223

When patients with high eIF4E overexpressionwere compared with patients with low eIF4E over-expression, the relative risk for recurrence was 1.6times greater in the high group compared with thelow group (95% confidence interval [CI], 1.0--2.5;P = .05; Table IV). Similarly, when patients withhigh eIF4E overexpression were compared with pa-tients with low eIF4E overexpression for cancer-re-lated death, the high eIF4E group had 2.1-foldincreased relative risk for death over the loweIF4E group (95% CI, 1.2--3.9; P = .01; Table IV).

DISCUSSION

In this study of specifically TNBC, eIF4E over-expression in carcinoma specimens seems to pre-dict cancer recurrence and cancer death. eIF4Ecutoff points of 7.5 or below, between 7.5 and 14,and higher than 14 were initially established in ouroriginal paper, because by using those cutoffpoints, the number of patients accrued in the

Table III. eIF4E tertile and breast cancer-relateddeath

eIF4E fold overexpression Survival Recurrence

Low (<7.5-fold; n = 36) 32 4Intermediate (between 7.5-fold

and 15-fold; n = 40)35 5

High (>15-fold; n = 27) 16 11Total (n = 103) 83 20

Values are number of patients.P = .005 (Chi-square test).

Fig 2. Patients with ‘‘triple-negative’’ tumors in the higheIF4E group had a higher rate of cancer recurrence (log-rank test; P = .04) than the low eIF4E group.

study could be divided evenly into 3 groups andthus tertile distribution.17 Subsequently, we havepublished 6 additional prospective studies that di-vided patients based on those cutoff points and val-idated that this tertile distribution by eIF4E levelshave reproducible, observable, significant differ-ences in clinical cancer recurrences and/ordeaths.8,11,12,18-20 Our current study utilizes thesame cutoff points, with the exception that thehigh group starts at 15, and there are no patientswith eIF4E levels between 14 and 15, to be consis-tent with previously published and establishedgroupings of patients based on tumor eIF4E levels.At present, there is no direct evidence that thesecutoff points correlate with specific biologic activ-ity in vitro; however, there are consistent, observed,clinical outcomes.8,11,12,17-20 Based on eIF4E levels,those TNBC patients whose neoplasms were in thehighest eIF4E group had a greater rate of recur-rence and rate of cancer death. Their relative riskfor recurrence and death was 1.6-fold (P = .05;95% CI, 1.01--2.54; Cox proportional hazards regres-sion model) and 2.1-fold, respectively (P = .01; 95%CI, 1.2--3.9; Cox proportional hazards regressionmodel) over those TNBC patients with low eIF4Etumor specimens. When patients were segregatedby nodal status, node-negative TNBC patients(n = 59) with high eIF4E overexpression had a signif-icantly greater rate of cancer recurrence (P = .02)and death (P = .03) than those in the low eIF4Egroup. This relationship was not seen in the node-positive TNBC patients (n = 44).

Fig 3. Patients with ‘‘triple-negative’’ tumors in the higheIF4E group had a higher rate of cancer-related death(log-rank test; P = .02) than the low eIF4E group.

SurgeryAugust 2009

224 Flowers et al

Fig 4. Node-negative TNBC patients with tumors in the high eIF4E group (A) had a higher rate of cancer recurrencethan those in the low eIF4E group (B) (P = .02). Similarly, node-negative TNBC patients with tumors in the high eIF4Egroup also had a higher rate of cancer-related deaths than patients in the low eIF4E group (P = .03).

Fig 5. Node-positive TNBC patients with tumors in the high eIF4E group (A) did not have higher rates of cancer re-currence (P = .8) or cancer-related deaths (P = .4) than those in the low eIF4E group (B).

One possible explanation for the loss of signif-icance of eIF4E as a predictor for a worse clinicaloutcome in node-positive patients may be samplesize (n = 44). Another may be that eIF4E is an ef-fective predictor of outcome in early stagedTNBC patients, but not in late staged patients.We suspect that the explanation may be thateIF4E overexpression leads to the upregulationof TLK1B. We have reported that TLK1B, a threo-nine kinase with a long 59UTR, is upregulated intumor specimens with high eIF4E overexpression.Additionally, TLK1B overexpression has beenlinked to doxorubicin resistance, in vitro and

in vivo.12,13 Thus, in the TNBC node-negativepatients who are generally treated with anthracy-cline-based therapy only, the observation thathigh eIF4E overexpression in the tumor specimensof these patients had a worse outcome may in partbe because of their resistance to anthracycline-based therapy. In contrast, node-positive TNBC pa-tients generally receive additional taxane-basedchemotherapy, in addition to doxorubicin. Thus,the degree of eIF4E overexpression seems tohave lost its predictive value for recurrence;TLK1B has not been associated with taxane resis-tance. This conjecture will be examined as we

SurgeryVolume 146, Number 2

Flowers et al 225

plan our next study on eIF4E overexpression,TLK1B upregulation, and clinical outcomes inTNBC patients treated with taxane-based therapyin comparison with anthracycline-based therapy.

Other mechanisms linking eIF4E overexpres-sion and chemoresistance have been reported aswell.21 eIF4E has been shown to upregulate cyclinD1.22,23 Inhibition of cyclin D1 expression, withsmall interfering RNAs, has been linked to a de-crease in chemoresistance in tumor cells.24 Cis-platin in particular has been linked to cyclin D1

expression, and a decrease in expression of cyclinD1 expression confers enhanced cytotoxicity tocisplatin in breast cancer cells.24 Therefore, itmay be that eIF4E overexpression and the subse-quent increase in cyclin D1 expression may conferresistance to cisplatin-based therapy.

That the overexpression of eIF4E leads to theupregulation of downstream gene products, suchas TLK1B and cyclin D1, is particularly discon-certing. This observation suggests that a major riskfor eIF4E overexpression in neoplasms and thesubsequent translation control dysregulation canresult in multiple downstream gene products thatpotentially confer multiple resistance to cytotoxicagents. This relationship is especially poignant forTNBC patients, who are not candidates for target-specific therapies, and rely solely on cytotoxicchemotherapy.

In conclusion, eIF4E overexpression in TNBCwas shown to decrease time to recurrence and timeto death for patients with high eIF4E overexpres-sion compared with patients with low eIF4E over-expression. When segregated for nodal status,node-negative TNBC patients with high eIF4Eoverexpression had greater rates of cancer recur-rence and cancer-related death than those patientswith low eIF4E overexpression. This relationshipwas lost in node-positive patients. The relative riskfor cancer recurrence and cancer-related death forthe high eIF4E group was 1.6 and 2.1 times that of

Table IV. Cox proportional hazards regressionmodel for cancer recurrence and cancer-relateddeath

Predictor Relative risk95% confidence

interval P value

RecurrenceeIF4E group 1.6 1.0–2.5 .05T stage 2.0 1.2–3.6 .02N stage 1.5 1.0–2.2 .03

Predictor for deatheIF4E group 2.1 1.2–3.9 .01T stage 1.6 0.7–3.3 .2N stage 1.5 0.9–2.4 .1

the low eIF4E group, respectively. These findingssuggest that eIF4E is a predictor of a worse clinicaloutcome in triple-negative breast cancer patients.

REFERENCES

1. Cleator S, Heller W, Coombes RC. Triple-negative breastcancer: therapeutic options. Lancet Oncol 2007;8:235-44.

2. Perou CM, Sorlie T, Eisen MB, et al. Molecular portraits ofhuman breast tumors. Nature 2000;406:747-52.

3. Kang S, Martel M, Harris L. Triple-negative breast cancer:current understanding of biology and treatment options.Curr Opin Obstet Gynecol 2008;20:40-6.

4. Reis-Filho JS, Tutt ANJ. Triple negative tumours: a criticalreview. Histopathology 2008;52:108-18.

5. Carey LA, Dees EC, Sawyer L, Gatti L, Moore DT, Collichio F,et al. The triple negative paradox: primary tumor chemosen-sitivity of breast cancer subtypes. Clin Cancer Res 2007;13:2329-34.

6. Li BDL, Liu L, Dawson M, De Benedetti A. Overexpressionof eukaryotic initiation factor 4E (eIF4E) in breast carci-noma. Cancer 1997;79:2385-90.

7. Li BDL, McDonald JC, Nassar R, De Benedetti A. Clinicaloutcome in stage I to III breast carcinoma and eIF4E over-expression. Ann Surg 1998;227:756-63.

8. Li BDL, Gruner JS, Abreo F, Johnson LW, Yu H, Nawas S,et al. Prospective study of eukaryotic initiation factor 4Eprotein elevation and breast cancer outcome. Ann Surg2002;235:732-9.

9. Sorrells DL, Meschonat C, Black D, Li BDL. Pattern ofamplification and overexpression of the eukaryotic initia-tion factor 4E gene in solid tumor. J Surg Res 1999;85:37-42.

10. Kevil CG, De Benedetti A, Payne DK, Coe LL, Laroux FS,Alexander JS. Translational regulation of vascular permea-bility factor by eukaryotic initiation factor 4E: implicationsfor tumor angiogenesis. Int J Cancer 1996;65:785-90.

11. Byrnes K, White S, Chu Q, Meschonat C, Yu H, Johnson LW,et al. High eIF4E, VEGF, and microvessel density in stage Ito III breast cancer. Ann Surg 2006;243:684-92.

12. Norton KS, McClusky D, Sen S, Yu H, Meschonat C, Debe-nedetti A, et al. TLK1B is elevated with eIF4E overexpres-sion in breast cancer. J Surg Res 2004;116:98-103.

13. Li Y, DeFatta R, Anthony C, Sunavala G, De Benedetti A.A translationally regulated Tousled kinase phosphorylateshistone H3 and confers radioresistance when overex-pressed. Oncogene 2001;20:726-38.

14. Shalom S, Don J. Tlk, a novel evolutionarily conservedmurine serine threonine kinase, encodes multiple testistranscripts. Mol Reprod Dev 1999;52:392-405.

15. American Joint Committee on Cancer Breast. Handbookfor staging of cancer, from the manual for staging of can-cer. In: Beahrs OH, Henson DE, Hutter RVP, Kennedy BJ,editors. Handbook for staging of cancer, from the manualfor staging of cancer. 4th edition Philadelphia: J.B. Lippincott;1993. p. 161-7.

16. Kerekatte V, Smiley K, Hu B, Smith A, Gelder F, De Bene-detti A. The proto-oncogene/translation factor eIF4E: a sur-vey of its expression in breast carcinomas. Int J Cancer 1995;64:27-31.

17. Holm N, Byrnes K, Johnson L, Abreo F, Sehon K, Alley J,et al. A prospective trial on initiation factor 4E (eIF4E)overexpression and cancer recurrence in node-negativebreast cancer. Ann Surg Oncol 2008;15:3207-15.

18. McClusky DR, Chu Q, Yu H, De Benedetti A, Johnson LW,Meschonat C, et al. A prospective trial on initiation factor

SurgeryAugust 2009

226 Flowers et al

4E (eIF4E) overexpression and cancer recurrence in node-positive breast cancer. Ann Surg 2005;242:584-90.

19. Wolfort R, De Benedetti A, Nuthalapaty S, Yu H, Chu QD,Li BD. Up-regulation of TLK1B by eIF4E overexpressionpredicts cancer recurrence in irradiated patients with breastcancer. Surgery 2006;140:161-9.

20. Byrnes KW, De Benedetti A, Holm NT, Luke J, Nunez J,Chu QD, et al. Correlation of TLK1B in elevation and re-currence in doxorubicin-treated breast cancer patientswith high eIF4E overexpression. J Am Coll Surg 2007;204:925-33.

21. Kwak YD, Koike H, Sugaya K. RNA interference with smallhairpin RNAs transcribed from a human U6 promoter-driven DNA vector. Pharmacol Sci 2003;93:214-7.

22. Rosenwald I, Lazaris-Karatzas A, Sonenberg N, Schmidt E.Elevated levels of cyclin D1 in response to increased expres-sion of eukaryotic initiation factor 4E. Mol Cell Biol 1993;13:7358-63.

23. Rousseau D, Kaspar R, Rosenwald I, Gehrke L, Sonenberg N.Translation initiation of ornithine decarboxylase and nucleo-cytoplasmic transport of cyclin D1 mRNA are increased incells overexpressing eukaryotic initiation factor 4E. ProcNatl Acad Sci U S A 1996;93:1065-70.

24. Dong K, Wang R, Wang X, Lin F, Shen JJ, Gao P, et al.Tumor-specific RNAi targeting eIF4E suppresses tumorgrowth, induces apoptosis and enhances cisplatin cytotox-icity in human breast carcinoma cells. Breast Cancer ResTreat 2009;113:443-56.