ORIGINAL ARTICLE

A pilot ‘window of opportunity’ neoadjuvant study of metforminin localised prostate cancerAM Joshua1, VE Zannella1,7, MR Downes1,2,7, B Bowes1, K Hersey1, M Koritzinsky1, M Schwab3,4, U Hofmann3,4, A Evans1,2,T van der Kwast1,2, J Trachtenberg1,3,4, A Finelli1,3,4, N Fleshner1,5, J Sweet1,2 and M Pollak6

BACKGROUND: Metformin is an inhibitor of complex 1 in the respiratory chain, and is widely used to reduce insulin resistance.It has also been described to have pleotropic effects including via AMPK on inhibiting the mTOR kinase. Pre-clinical andepidemiological studies suggest an ability to modulate disease evolution in prostate cancer. In this study, we aimed to(i) demonstrate safety and tolerability of neoadjuvant metformin administration and (ii) document changes in proliferative (Ki67)and AMPK-related signalling indices between matching biopsies and prostatectomiesMETHODS: Men were treated in a single-arm ‘window of opportunity’ study between their decision to undergo radicalprostatectomy and the operation itself. Forty patients were planned but only 24 patients were enrolled owing to slow accrual.Twenty-one patients were evaluable for pathological outcomes and 22 for serum metabolic indices. Metformin was given at dosesto 500 mg t.i.d. Ki67 index was calculated using the Aperio-positive pixel count algorithm, whereas immunohistochemicalmeasurements were by consensus H-Score. Comparative statistics were analysed by students t-tests and/or Wilcoxon matched pairssigned rank test.RESULTS: Baseline characteristics included median PSA 6 ng ml� 1 (3.22–36.11 ng ml� 1). Median duration of drug treatment was41 days (18–81). Treatment was well tolerated with only three patients developing G3/4 toxicities. In a per patient and per tumouranalyses, metformin reduced the Ki67 index by relative amounts of 29.5 and 28.6 % (P¼ 0.0064 and P¼ 0.0042) respectively. Therewas also a significant decrease in P-4EBP1 staining (Po0.001) but no change in P-AMPK or P-ACC. There were no correlationsbetween any metabolic, morphometric or cancer-related serum indices. There was a trend towards PSA reduction (P¼ 0.08).The study is limited by small patient numbers and tumour heterogeneity.CONCLUSIONS: Neoadjuvant metformin is well tolerated prior to radical prostatectomy. Data to date indicate promising effects onmetabolic and tissue proliferation and signalling parameters.

Prostate Cancer and Prostatic Disease advance online publication, 27 May 2014; doi:10.1038/pcan.2014.20

INTRODUCTIONProstate cancer remains the second most commonly diagnosedcancer in men in the western world.1 Although advances in thetreatment of advanced castrate refractory disease are extendingoverall survival in the setting of metastatic disease, only hormonaltherapy has been shown to increase contemporary cure rates inthe setting of definitive curative therapy (radiation). Furtheradvances are critically required in this area to reduce the morbi-dity and mortality from the disease.

Metformin is the most commonly prescribed normoglycaemicagent worldwide for non-insulin-dependent diabetes mellitus. Itscanonical mechanism of action involves inhibition of the type 1respiratory complex and subsequent AMPK activation resulting inmTOR inhibition. Importantly, pre-clinical studies have shown asignificant effect on several prostate cancer cell lines’ proliferationand accompanying xenograft growth.2

In addition, metformin reduces circulating insulin levels in non-diabetics,3 relevant because higher insulin and C-peptide levelshave been associated with poor outcomes in prostate cancerpatients.4–6 Furthermore, the overexpression of insulin receptors

(IRs) in prostate cancer and the association between IR expressionor IR/IGF1R activation has been noted.7

Taken together with the larger epidemiological studies thathave suggested a lack of effect on prostate cancer incidence8–10

but a beneficial effect on prostate cancer-related mortality,11,12

we undertook a phase 2 ‘window of opportunity’ study withmetformin in men scheduled to undergo radical prostatectomy toassess the effects of metformin on cellular indices relevant tocancer to further understand and quantify its utility in this disease.

MATERIALS AND METHODSPatientsFrom July 2009 to Feb 2011, we accrued 24 patients in a window ofopportunity study prior to radical prostatectomy from the UrologicOncology clinics at the Princess Margaret Cancer Centre, Toronto, ON,Canada. The study was approved by the University Health Network ResearchEthics Board and registered on clinicaltrials.gov (NCT NCT00881725).Relevant inclusion criteria included (i) biopsy-proven, localised prostatecancer (as defined below); (ii) histologically confirmed prostate cancer

1Princess Margaret Cancer Centre, Toronto, ON, Canada; 2Department of Pathology, University Health Network, Toronto, ON, Canada; 3Dr Margarete Fischer-Bosch Institute ofClinical Pharmacology, Stuttgart, Germany; 4University of Tubingen, Stuttgart, Germany; 5Division of Urology, Department of Surgery, University Health Network, Toronto, ON,Canada and 6Jewish General Hospital, Montreal, QC, Canada. Correspondence: Dr AM Joshua, Division of Medical Oncology and Hematology, 610 University Avenue, Room 5-101,Toronto, ON, Canada M5G2M9.E-mail: Anthony.joshua@uhn.ca7These authors contributed equally to this paperReceived 20 January 2014; revised 8 April 2014; accepted 20 April 2014

Prostate Cancer and Prostatic Disease (2014), 1–7& 2014 Macmillan Publishers Limited All rights reserved 1365-7852/14

www.nature.com/pcan

involving at least 20% of one or more unfragmented biopsy cores inavailable archived biopsy cores; (iii) transrectal ultrasound within the last 3months and biopsy tissue available for study. Relevant exclusion criteriaincluded (i) patients medicated with any drug used for the treatment ofany form of diabetes, or patients that begin treatment for any form ofdiabetes during the course of the study; (ii) patients receiving any otherinvestigational, herbal or anticancer agents while on study; (iii) a currenthistory of alcohol intake (42 standard drinks per day) or binge drinking(five or more drinks) in one session of 1–3 h; (iv) past history of lacticacidosis or risk factors for lactic acidosis, such as congestive heart failure(NYHA Class 3 or greater), hypoxia (resting PO2o91%) or renalinsufficiency (eGFRo60 ml min� 1). Patients were identified by participat-ing urologists and screened by a research assistant.

After informed consent and study enrolment, patients were instructed totake metformin at doses increasing from 500 mg daily to 500 mg t.i.d. over5 days (500 mg daily day 1–2; 500 mg b.i.d. day 2–4; 500 mg t.i.d. day 5þ )for 4–12 weeks preceding their prostatectomy. Metformin was continueduntil the patients were nil per oral for surgery. Toxicity assessment wasbased on CTCAE v3. Dose reductions were permitted for select grade 3,4toxicities.

Baseline measurementsBaseline measurements included fasting weight, height, fasting bloodglucose/insulin, 2 h postprandial glucose, lipid profile (cholesterol, high-density lipoprotein, low-density lipoprotein, triglycerides), insulin indices(insulin-like growth factor 1, IGFBP3, C-peptide), PSA and serum metforminlevels. Serum levels of insulin-like growth factor 1, IGFBP3 and C-peptidewere quantified using standard enzyme-linked immunosorbent assaymethods, with reagents from IDS Ltd (London, UK). Serum and tissue levelsof metformin were quantified using high-performance liquid chromato-graphy. Serum metformin levels were taken from fasting patients in theweek before surgery.

Pathological assessments of the baseline tumours were acquired fromthe original pathology reports to determine eligibility and subsequentlyconfirmed by the study pathologists. Radical prostatectomy specimenswere injected with formalin immediately ex vivo by the study pathologiststo ensure optimal tissue preservation. Sections at 5 mm thickness were cutfrom formalin-fixed paraffin-embedded diagnostic core biopsy andsurgical specimens, and slides were stained for each of the followingantibodies with appropriate positive and negative controls: Ki67—NovusBiologicals (Littleton, CO, USA) NB100-90592 1/1500 Citrate 1 h; P-4ebp1—Cell Signaling (Beverly, MA, USA) 2855 1/100 Citrate 1 h; P-AMPK—CellSignaling 2535 1/300 Tris-EDTA overnight; P-ACC—Cell Signaling 3661 1/400 Tris-EDTA overnight; P-GSK3B—Life Span (Seattle, WA, USA)LS-0105395 1/200 citrate overnight; P-AKT—Cell Signaling 4060 1/20citrate overnight; ERG—Biocare Medical (Concord, CA, USA) cm421 APTris-EDTA pH 9.0 1/300 overnight. The ‘Ki67 index’ (percentage of nucleishowing nuclear immunoreactivity of any intensity) was determined byautomated counting from the positive pixel count algorithm on the Aperiosystem after encircling all visible tumour foci in a region of interest. Otherimmunohistochemical assessments were performed by two pathologists(JS, MRD), who rated each sample separately before reconciling differencesto reach a consensus.

Statistical planThere were two primary statistical end points–safety (prospectivelyanalysed) and scientific (retrospectively analysed). Originally, an optimal2-stage design was used to test the null hypothesis that the toxicityrate of X20% vs the alternative toxicity rate is o10%, with interimanalyses after 19 patients, to accrue a total of 37 patients. Toxicitieswere defined as serious adverse events or peri-operative morbidity/mortality possibly related to study drug. Alternatively, in the event of oneepisode of lactic acidosis the study would have been terminated.Similarly, the primary scientific end point was the quantitation of theKi67-proliferation index. The sample size of 37 was originally designed toprovide an B85% power to detect a 0.5 s.d. difference between theoriginal Ki67 or the biopsy and the subsequent prostatectomy. The poweris calculated based on paired t-test with a significance level of 0.05.Assuming 10% drop-out rate, a total sample size of 40 patients wasrecommended to fulfil both aims of the study, with B2–3 patients permonth over 18 months.

Statistical analysis was carried out after tabulation of results in MicrosoftExcel. Statistical tests included students t-tests and Wilcoxon matched pairssigned rank test (Advanced Analytics, LLC, Gaithersburg, MD, USA) and

standard correlation coefficients. The homeostatic model assessmentcalculations were carried out using HOMA2 calculator online (OxfordCentre for Diabetes, Endocrinology and Metabolism, Oxford, UK).

RESULTSDespite intense screening, slow patient accrual necessitated earlytrial closure. The CONSORT diagram is showed in Figure 1. Patientcharacteristics of those enrolled are summarised in Table 1.Toxicities were generally minor and easily managed. These aresummarised in Table 2. Notable grade 3 toxicities occurred inthree patients only. These included pelvic abscess (two patients),dizziness (one patient), abdominal pain (one patient), urticaria(one patient). None of these grade 3 toxicities was thought to berelated to metformin. Thus, there was no evidence of peri-operative morbidities related to metformin.

As a result of the early cessation of the trial, the power of theprimary scientific end point was reduced to 65%, nevertheless, theKi67 index per tumour decreased from a median of 4.696% Ki67þ ve nuclei to 2.84 (P¼ 0.0018), whereas the Ki67 index perpatient (with four patients having 41 tumour identified) reducedfrom a median of 4.78% Ki67 þ ve nuclei to 2.76 (P¼ 0.0064)

Figure 1. CONSORT diagram for the study indicating patientscreening and accrual patterns during the course of the study.

Table 1. Baseline characteristics of patients enrolled in the study

Variable Median Range

Baseline characteristicsa

Age 64 53–73PSA 6mg l� 1 3.7–36.11Duration of metformin treatment 41 days 18–81Number of positive biopsy cores 4 2–8Percentage maximal core involvement 50 30–90Primary Gleason 3 3–4Secondary Gleason 4 3–4

Post intervention characteristicsGleason score 7 6–8Percentage tissue involvement 4 2–20Pathological staging N/A T2a-T3b

aOf the 24 evaluable patients enrolled, 22 were evaluable for assessmentand 21 for pathology assessment.

Neoadjuvant metformin in prostate cancerAM Joshua et al

2

Prostate Cancer and Prostatic Disease (2014), 1 – 7 & 2014 Macmillan Publishers Limited

(Figure 2). This represents a relative decrease of Ki67 of 29.5%(95% CI: (� 2.51,� 0.32), P¼ 0.0064)) for the former and a relativedecrease of 28.6% (95% CI: (� 2.375,� 0.5036), P¼ 0.0042)) for the

latter. Accompanying H-score assessments for immunohistochem-ical pathways of interest are summarised in Table 3 and depictedin Figure 2. This showed a significant decrease in P-4EBP1 stainingand an increase in P-GSK3B staining, whereas no significantdifference was noted in P-AMPK or P-ACC. To further clarify thesefindings, we carried out immunohistochemistry for P-AKT; how-ever, we were unable to demonstrate evidence of increased P-AKTsignalling after metformin administration (see Table 3).

Serum assays as outlined in the protocol are summarised inTable 4. These demonstrate significant reductions in insulin-likegrowth factor 1, fasting glucose and body mass index. In anexploratory analysis, we examined the associations between Ki67index change (per patients and per tumour) or change in P-4EBP1staining and baseline factors (listed in Table 1) as well as serum,anthropological or immunohistochemical indices, however, noassociations were found (data not shown).

In addition, owing to the multifocal nature of prostate cancer, asa quality control measure, we undertook ERG immunohisto-chemistry on both the biopsy and correlated section of the radicalprostatectomy to ensure the sections represented similar tumour

Ki67 Scores per Patient

-10

Needle

Biop

sy

Prosta

tecto

my

Differe

nce

P-4eb

p1 B

iopsy

P-4eb

p1 T

umou

r

p-4e

bp1

Differe

nce

GSK_3b

Biopsy

GSK-3b

Tumou

r

GSK3b D

iffere

nce

Needle

Biop

sy

Prosta

tecto

my

Differe

nce

-5

0

5

10a b

c

e

d

Ki6

7 In

dex

Ki67 Scores per Tumour

-10

-5

0

5

10

15

Ki6

7 S

core

P-4EBP1 H-Score

0

30

-30

-40

-50

20

-20

10

-10

0

100

200

300

H-S

core

Per

cent

PS

A fa

ll

P-GSK3B H-Score

-100

Percent PSA fall during MeforminTreatment

0

100

200

300

H-S

core

Figure 2. (a) Metformin reduces the Ki67 proliferation index by 1.44% (relative decrease of 29.5%) (95% CI: (� 2.51,� 0.32), P¼ 0.0064)). Boxand whisker plots represent median value (middle line) and 25th and 75th percentile, with whiskers representing minimum and maximumvalues, respectively. Shading is illustrative only. (b) Metformin reduces the Ki67 proliferation index by 1.44% (relative decrease of 28.6%) (95%CI: (� 2.375,� 0.5036), P¼ 0.0042)). Box plot description as above. (c) and (d) Metformin reduces the P-4EBP1 staining and increases theP-GSK3B as measured by H-Score (Po0.001). Values have been divided by 100 to facilitate visual assessment. Box plot description as above.(e) Percent PSA fall arranged by extent of fall.

Table 2. Summary of toxicities experienced by patients during thestudy

n Percent

Number that received at least one dose ofmetformin

24 100

Number that developed any AE 23 96Patients with G3–4 AE 3a 12.5

AE present in 415% of patientsDiarrhoea (any grade) 8 33Nausea (any grade) 4 17

Abbreviation: AE, adverse events.aTotal number of G3–4 AE was five.

Neoadjuvant metformin in prostate cancerAM Joshua et al

3

& 2014 Macmillan Publishers Limited Prostate Cancer and Prostatic Disease (2014), 1 – 7

foci. This analysis revealed that of the 22 matched samplesobtained (biopsy/tumour), the majority had concurrent ERG status(12 consistently negative, 6 consistently positive, 3 mixed, 1ineligible) suggesting that we were likely to have assessed thesame tumour focus in both the biopsy and the radical prosta-tectomy specimen. To further refine the analyses, we undertookan exploratory analysis of tumours with the same ERG status,reducing the dataset to 18 samples. Herein, the change in Ki67index remained significant for on a per patient (P¼ 0.0063) and aper tumour basis (P¼ 0.0016).

DISCUSSIONTo our knowledge, this is the first neoadjuvant intervention studywith metformin in prostate cancer, although others are in prog-ress (NCT01433913). We have demonstrated a reduction in acalculated Ki67 index between the baseline biopsies and theradical prostatectomy tissue both per patient and per tumour.These results do mirror recent findings in breast cancer,13–16 andare consistent with pre-clinical findings in vitro and in vivo.2,17,18

Our findings of decreased P-4EBP1 may indicate a mechanisticinsight into the primary end point as its role in increasingcap-dependent translation that ultimately leads to cell growth.19

A perplexing finding of the study was the apparent increase inGSK3B staining, despite the absence of any increase in P-AKTstaining that would have been expected given the knownfeedback between S6K and IRS1.20 This apparent paradox mayhave been confounded by issues with tissue and immunohisto-chemical heterogeneity (in both the biopsies and prostatectomies)and phospho-protein preservation (in the prostatectomies), whichmay have hampered the interpretation. In an exploratory analysiswe did not see any stratification by ERG status in theseimmunohistochemical markers (data not shown).

Strengths of our study included the assessment of serum,anthropometric and physiologic measurements as surrogate for

metformin’s indirect mechanism of action, as well as tumour tissuepre- and post-metformin, use of a commonly prescribed dose ofmetformin and continuation of metformin to the evening prior tosurgery in all subjects. Although the dose of metformin is not themaximum dose commonly prescribed to diabetic patients(2250 mg), it was felt that the dose used (1500 mg) offered thebest balance of efficacy and toxicity/ compliance in this otherwisewell population and is similar to doses used in other interventionalstudies.

Weaknesses of the study included a small sample size and theabsence of a control group. With regards to the latter, we havepreviously stained 10 matched historical biopsies and prostatec-tomies from our institution from the same period as this study butwere unable to demonstrate a difference in Ki67 index (data notshown). Reasons for the slow accrual to this trial are uncertain—asoutlined, a small number of patients were enrolled on acompeting trial (19 patients) but the majority of patients werenot eligible because of either refusal or ineligibility. Refusal wasmost commonly related to; unable/unwilling to commit the time(46%), not interested in participating in research (26%); concernabout the possible side-effects of metformin; (7%) and feelingoverwhelmed by their diagnosis (7%), whereas ineligibility wasmost commonly due to the requirement for an adequate amountof tumour tissue in their biopsy sample (112 of 147 patients) toensure robust immunohistochemical assessment (initially 50% butultimately reduced to 20% of any core with sequential protocolamendments). An additional weakness is that metformin wasadministered for a relatively short period of time and the optimalduration of exposure to metformin necessary for any antineo-plastic action is unknown.

An earlier study by Goodwin et al.3 identified a larger decreasein fasting insulin levels (22%) when metformin was administeredto non-diabetic breast cancer patients for 6 months than was seenor C-peptide in this study (no change between baseline andsurgical samples). It is possible that this reflects differences inthe study populations; it is also possible that administrationof metformin for 2 weeks is insufficient for a full effect on fastinginsulin. Our study did benefit from measuring metforminconcentration in blood, with average levels of metformin being6.7 nmol ml� 1, a concentration generally lower than 20.9 nmol ml� 1

reported from large databases. Importantly, there was evidence ofa direct effect on tumour signalling pathways at this concentra-tion, despite concerns that such effects would not be recapitu-lated in vivo,21 possibly because metformin concentrates in themitochondria at up to 1000 fold its circulating concentration.22 Wealso measured the metformin concentration in prostate tissue infive samples suggesting that metformin does reach prostatetissue, albeit in nanogram mg� 1 quantities. Mechanisms respon-

Table 3. H-scores for immunohistochemical stains associated withprostate biopsies and radical prostatectomy tissues

Stain N (matchedsamples)

DifferenceH-score (mean)

Significance (t-test)

P-AMPK 22 46.6 0.07P-ACC 22 2.73 0.92P-4EBP1 22 170.6 o0.01P-GSK3B 18 174.7 o0.01P-AKT 14 31.2 0.12

Table 4. Serum and morphological measurements for patients enrolled on study

Assay Median pre-metformin (±s.d.) Median post-metformin (±s.d.) P

IGF1 (ngml� 1) 143.5 (±25) 134.3 (±28) 0.02IGFBP3 (ng� 1ml� 1) 3856.3 (±614) 3658.8 (±730) 0.93C-peptide (ngml� 1) 2 (±0.6) 2.1 (±0.84) 0.91HOMA-IR 1.6 (±0.43) 1.6 (±0.54) 0.55PSA (ngdl� 1) 5.9 (±7.9) 5.3 (±7.6) 0.08Fasting glucose (mmol l� 1) 5.5 (±0.87) 5.3 (±0.55) 0.03Postprandial glucose (mmol/ l� 1) 6.5 (±2.52) 6.7 (±2.57) 0.167Body mass index 28.1 (±3.6) 26.8 (±3.2) o0.01Serum metformin (ngml� 1) N/A 6.7 (range 1.62–16.04) N/ATissue metformina (nmolmg� 1) N/A 0.293 (range 0.218–1.732) N/A

Abbreviations: HOMA-IR, homeostasis model assessment-estimated insulin resistance; IGF1, insulin-like growth factor 1; IGFBP3, insulin-like growth factor-binding protein 3.aFive samples only were tested.

Neoadjuvant metformin in prostate cancerAM Joshua et al

4

Prostate Cancer and Prostatic Disease (2014), 1 – 7 & 2014 Macmillan Publishers Limited

P-AMPK

P-AMPK

P-AMPK

P-4EBP1

P-4EBP1

P-ACC

P-ACC

Figure 3. Illustration of staining patterns on study specimens, according to methods outlined above. (a) Illustration of radical prostatectomysection demonstrating heterogeneity in P-AMPK staining in tumour sections, with a component of the tumour (bottom left) staining negativefor P-AMPK. (b) Biopsy specimen demonstrating consistency with known signalling pathways with P-4EBP1 appearing positive in tumourtissue otherwise negative for P-AMPK and P-ACC. Antibody concentrations and conditions described in methods section. (c) Radicalprostatectomy specimen demonstrating consistency with known signalling pathways as above. Antibody concentrations and conditionsdescribed in Materials and Methods section.

Neoadjuvant metformin in prostate cancerAM Joshua et al

5

& 2014 Macmillan Publishers Limited Prostate Cancer and Prostatic Disease (2014), 1 – 7

sible for uptake into prostate tissue are unclear, but staining forOCT1/2 are documented in prostate cancer tissue.23,24

The use of Ki67 index as a marker of tumour activity is lessestablished in prostate cancer compared with breast cancer.However, it does have prognostic value in a variety of settingsincluding post-radiation,25 surgery26,27 and biopsy.28 In thissetting, we felt it was a more appropriate marker to assess theeffect of metformin as its mechanism of action is predominantlymetabolic inhibition rather than cytotoxic. However, we did notsee an association between a change in Ki67 index and othermarkers of tumour proliferative ability such as PSA; however, boththe numbers of patients are limited and the PSA values in earlystage disease can fluctuate. Assessment of some of the secondaryend points were exacerbated by challenges with tissue fixationand preservation of epitopes in radical prostatectomy sections,despite our efforts with prompt injection of formalin. Indeed, wefound numerous instances of heterogeneity in phospho-epitopestaining despite adequate fixation, whereas in other instances thephospho-epitope stains were consistent with known signallingpathways (Figure 3).

Our results, together with both epidemiological and pre-clinicaldata support the further investigation of metformin in prostatecancer. We are currently undertaking a randomised phase 3 clinicaltrial of metformin vs placebo in the active surveillance setting(NCT01864096). However, notable from our data, albeit limited bysmall patient numbers, is the lack of association of benefit with anybaseline factor. A number of hypotheses have recently be raisedregarding subgroups in which metformin may have an additivebenefit, in this setting an analyses of the Physicians Health study29

is revealing in suggesting that obesity is linked with poorer prostatecancer prognosis primarily in men harbouring the TMPRSS2-ERGfusion suggesting a possible target for metformin intervention. Wedid not find any corroborative evidence in our dataset that onlyTMPRSS2-ERG patients benefitted, as there were greater PSA falls inTMPRSS2-ERG-negative patients (median fall 11.4% vs 0.5% rise)(P¼ 0.02). Indeed, evidence of PSA decrease on metformin isintriguing itself, and the phenomena has been verified to occur in arecently completed phase 2 trial in castrate resistant disease,30

suggesting there may be some validity to our findings.Delivering interventions to men in a window of opportunity

prior to radical prostatectomy is feasible and in particularmetformin, a normoglycaemic drug, does not appear to causesignificant adverse events when given to non-diabetic non-obese men. In this setting, we have demonstrated that met-formin may affect cellular proliferation and signalling indices thatmay underlie recent evidence to suggest benefit in reducingprostate cancer progression and mortality. Further investigation iswarranted, in particular addressing the weakness of the studyherein.

CONFLICT OF INTERESTThe authors declare no conflict of interest.

ACKNOWLEDGEMENTSThis research was funded in part by the Terry Fox Foundation, Princess MargaretHospital Foundation and the Jewish General Hospital Foundation. MS was in partsupported by the Robert-Bosch Foundation, Stuttgart, Germany, and the IZEPHAGrant, University of Tubingen, Tubingen, Germany

REFERENCES1 Society AC. Global Cancer Facts & Figures, 2nd edn. (Society AC), 2011.2 Ben Sahra I, Laurent K, Loubat A, Giorgetti-Peraldi S, Colosetti P, Auberger P et al.

The antidiabetic drug metformin exerts an antitumoral effect in vitro and in vivothrough a decrease of cyclin D1 level. Oncogene 2008; 27: 3576–3586.

3 Goodwin PJ, Pritchard KI, Ennis M, Clemons M, Graham M, Fantus IG. Insulin-lowering effects of metformin in women with early breast cancer. Clin BreastCancer 2008; 8: 501–505.

4 Lehrer S, Diamond EJ, Stagger S, Stone NN, Stock RG. Increased seruminsulin associated with increased risk of prostate cancer recurrence. Prostate 2002;50: 1–3.

5 Lehrer S, Diamond EJ, Stagger S, Stone NN, Stock RG. Serum insulin level, diseasestage, prostate specific antigen (PSA) and Gleason score in prostate cancer.Br J Cancer 2002; 87: 726–728.

6 Ma J, Li H, Giovannucci E, Mucci L, Qiu W, Nguyen PL et al. Prediagnostic body-mass index, plasma C-peptide concentration, and prostate cancer-specific mor-tality in men with prostate cancer: a long-term survival analysis. Lancet Oncol2008; 9: 1039–1047.

7 Cox ME, Gleave ME, Zakikhani M, Bell RH, Piura E, Vickers E et al. Insulin receptorexpression by human prostate cancers. Prostate 2009; 69: 33–40.

8 Franciosi M, Lucisano G, Lapice E, Strippoli GF, Pellegrini F, Nicolucci A. Metformintherapy and risk of cancer in patients with type 2 diabetes: systematic review.PLoS One 2013; 8: e71583.

9 Qiu H, Rhoads GG, Berlin JA, Marcella SW, Demissie K. Initial metformin or sul-phonylurea exposure and cancer occurrence among patients with type 2 diabetesmellitus. Diabetes Obes Metab 2013; 15: 349–357.

10 Margel D, Urbach D, Lipscombe LL, Bell CM, Kulkarni G, Austin PC et al. Asso-ciation between metformin use and risk of prostate cancer and its grade. J NatlCancer Inst 2013; 105: 1123–1131.

11 Spratt DE, Zhang C, Zumsteg ZS, Pei X, Zhang Z, Zelefsky MJ. Metformin andprostate cancer: reduced development of castration-resistant disease and pros-tate cancer mortality. Eur Urol 2013 Apr; 63: 709–716.

12 Margel D, Urbach DR, Lipscombe LL, Bell CM, Kulkarni G, Austin PC et al. Met-formin use and all-cause and prostate cancer-specific mortality among men withdiabetes. J Clin Oncol 2013; 31: 3069–3075.

13 Niraula S, Dowling RJ, Ennis M, Chang MC, Done SJ, Hood N et al. Metformin inearly breast cancer: a prospective window of opportunity neoadjuvant study.Breast Cancer Res Treat 2012; 135: 821–830.

14 Jiralerspong S, Palla SL, Giordano SH, Meric-Bernstam F, Liedtke C, Barnett CMet al. Metformin and pathologic complete responses to neoadjuvant chemo-therapy in diabetic patients with breast cancer. J Clin Oncology 2009; 27:3297–3302.

15 Hadad S, Iwamoto T, Jordan L, Purdie C, Bray S, Baker L et al. Evidencefor biological effects of metformin in operable breast cancer: a pre-operative,window-of-opportunity, randomized trial. Breast Cancer Res Treat 2011; 128:783–794.

16 Bonanni B, Puntoni M, Cazzaniga M, Pruneri G, Serrano D, Guerrieri-Gonzaga Aet al. Dual effect of metformin on breast cancer proliferation in a randomizedpresurgical trial. J Clin Oncol 2012; 30: 2593–2600.

17 Ben Sahra I, Regazzetti C, Robert G, Laurent K, Le Marchand-Brustel Y, Auberger Pet al. Metformin, independent of AMPK, induces mTOR inhibition and cell-cyclearrest through REDD1. Cancer Res 2011; 71: 4366–4372.

18 Colquhoun AJ, Venier NA, Vandersluis AD, Besla R, Sugar LM, Kiss A et al. Met-formin enhances the antiproliferative and apoptotic effect of bicalutamide inprostate cancer. Prostate Cancer Prostatic Dis 2012; 15: 346–352.

19 Jastrzebski K, Hannan KM, Tchoubrieva EB, Hannan RD, Pearson RB.Coordinate regulation of ribosome biogenesis and function by the ribosomalprotein S6 kinase, a key mediator of mTOR function. Growth Factors 2007; 25:209–226.

20 Manning BD. Balancing Akt with S6K: implications for both metabolic diseasesand tumorigenesis. J Cell Biol 2004; 167: 399–403.

21 Stambolic V, Woodgett JR, Fantus IG, Pritchard KI, Goodwin PJ. Utility of met-formin in breast cancer treatment, is neoangiogenesis a risk factor? Breast CancerRes Treatment 2009; 114: 387–389.

22 Owen MR, Doran E, Halestrap AP. Evidence that metformin exerts its anti-diabeticeffects through inhibition of complex 1 of the mitochondrial respiratory chain.J Biochem 2000; 348(Pt 3): 607–614.

23 Obinata D, Takayama K, Urano T, Murata T, Kumagai J, Fujimura T et al. Oct1regulates cell growth of LNCaP cells and is a prognostic factor for prostate cancer.Int J Cancer 2012; 130: 1021–1028.

24 Uhlen M, Oksvold P, Fagerberg L, Lundberg E, Jonasson K, Forsberg M et al.Towards a knowledge-based human protein atlas. Nat Biotechnol 2010; 28:1248–1250.

25 Verhoven B, Yan Y, Ritter M, Khor LY, Hammond E, Jones C et al. Ki-67 is anindependent predictor of metastasis and cause-specific mortality for prostatecancer patients treated on Radiation Therapy Oncology Group (RTOG) 94-08. Int JRadiat Oncol Biol Phys 2013; 86: 317–323.

26 Bubendorf L, Sauter G, Moch H, Schmid HP, Gasser TC, Jordan P et al. Ki67labelling index: an independent predictor of progression in prostate cancertreated by radical prostatectomy. J Pathol 1996; 178: 437–441.

Neoadjuvant metformin in prostate cancerAM Joshua et al

6

Prostate Cancer and Prostatic Disease (2014), 1 – 7 & 2014 Macmillan Publishers Limited

27 Zellweger T, Gunther S, Zlobec I, Savic S, Sauter G, Moch H et al. Tumour growthfraction measured by immunohistochemical staining of Ki67 is an independentprognostic factor in preoperative prostate biopsies with small-volume or low-grade prostate cancer. Int J Cancer 2009; 124: 2116–2123.

28 Bubendorf L, Tapia C, Gasser TC, Casella R, Grunder B, Moch H et al. Ki67 labelingindex in core needle biopsies independently predicts tumor-specific survival inprostate cancer. Hum Pathol 1998; 29: 949–954.

29 Pettersson A, Lis RT, Meisner A, Flavin R, Stack EC, Fiorentino M et al. Modificationof the association between obesity and lethal prostate cancer by TMPRSS2:ERG.J Natl Cancer Inst 2013; 105: 1881–1890.

30 Rothermundt CA, Cathomas R, Templeton A, Winterhalder RC, Strebel R, BaertschiD, Pollak M, Lui L, Crowe S, Gillessen S (eds). Metformin in Chemotherapy-NaıveCastration Resistant Prostate Cancer (CRPC): A Multicenter Phase II Trial (SAKK08/09). ESMO: Vienna, 2012.

Neoadjuvant metformin in prostate cancerAM Joshua et al

7

& 2014 Macmillan Publishers Limited Prostate Cancer and Prostatic Disease (2014), 1 – 7