nature reviews | clinical oncology volume 8 | marCH 2011 | 131

nEws & viEws

Correspondence to: K. Søreide ksoreide@mac.com

competing interestsThe authors declare no competing interests.

1. Chan, A. T., Ogino, s. & Fuchs, C. s. Aspirin and the risk of colorectal cancer in relation to the expression of COX-2. N. Engl. J. Med. 356, 2131–2142 (2007).

2. Gao, F. et al. The effect of aspirin in the recurrence of colorectal adenomas: a meta-analysis of randomized controlled trials. Colorectal Dis. 11, 893–901 (2009).

3. Rothwell, P. M. et al. Long-term effect of aspirin on colorectal cancer incidence and mortality: 20-year follow-up of five randomised trials. Lancet 376, 1741–1750 (2010).

4. Benamouzig, R. & Uzzan, B. Aspirin to prevent colorectal cancer: time to act? Lancet 376, 1713–1714 (2010).

5. Markowitz, s. D. & Bertagnolli, M. M. Molecular origins of cancer: molecular basis of colorectal cancer. N. Engl. J. Med. 361, 2449–2460 (2009).

6. Yan, M. et al. 15-Hydroxyprostaglandin dehydrogenase inactivation as a mechanism of resistance to celecoxib chemoprevention of colon tumors. Proc. Natl Acad. Sci. USA 106, 9409–9413 (2009).

7. Cross, J. T., Poole, e. M. & Ulrich, C. M. A review of gene-drug interactions for non-steroidal anti-inflammatory drug use in preventing colorectal neoplasia. Pharmacogenomics J. 8, 237–247 (2008).

8. Benamouzig, R. et al. Cyclooxygenase-2 expression and recurrence of colorectal adenomas: effect of aspirin chemoprevention. Gut 59, 622–629 (2010).

9. Midgley, R. s. et al. Phase iii randomized trial assessing rofecoxib in the adjuvant setting of colorectal cancer: final results of the viCTOR trial. J. Clin. Oncol. 28, 4575–4580 (2010).

10. Kirkegaard, H. et al. Association of adherence to lifestyle recommendations and risk of colorectal cancer: a prospective Danish cohort study. BMJ 341, c5504 (2010).

wide generalization should be cautiously applied. second, the randomized nature of the trials may have introduced a selection bias of the participants, with specific entry criteria (such as gender) to fulfill inclusion in the studies. also, cancer outcome was not a primary end point of the trials; the conclu‑sions are derived from a secondary pooled analysis based on data extracted from cancer regis tries (uK) and cause‑of‑death registries (uK and sweden). this methodology might have introduced bias in the findings as such regis tries may not be as accurate as a detailed study with similar end points.

importantly, in the present study,3 aspirin did not have an effect on cancers of the distal colon and rectum; the locations of approxi‑mately two‑thirds of all large bowel cancers. Consequently, the prospect of prevention may only be applicable to the one‑third of cases that develop in the proximal colon. Furthermore, although screening with fecal occult blood tests and sigmoido scopy has demonstrated a decrease in CrC mortality, the use of colonoscopy and ‘virtual’ com‑puted colonography is gaining more wide‑spread use and may detect more proximally located cancers.

Proximal cancers are more prone to micro satellite instability than their distal counterparts.5 thus, difference in response to aspirin could possibly point to the hetero‑genetic molecular pathways of CrC. in 2009, it was demonstrated that the adenoma pre‑vention activity of the CoX2 inhibitor cele‑coxib requires the concomitant expression of the 15‑ hydroxyprostaglandin dehydro‑genase (15-PGDH) tumor‑ suppressor gene, and that loss of 15-PGDH expres‑sion imparts resistance to the antitumor effects of celecoxib.6 also, genetic vari‑ants of metabo lizing enzymes, including CoX2, might alter the risk and benefit from nsaiDs.7 Consequently, evidence from cur‑rently explored molecular pathways in CrC would suggest a tailored approach is neces‑sary even in the chemopreventive setting, as tumors are likely to develop through distinct molecular pathways.5

making the picture more complex, a recent study found no preventive effect of aspirin on patients whose initial adenomas strongly expressed CoX2; aspirin also did not affect adenoma prevention after 4 years of

follow up.8 Furthermore, at the other end of the spectrum of cancer prevention, selec‑tive CoX inhibitors have been explored as adjuvant therapy in CrC. However, the increased risk of adverse cardiovascular events associated with the specific CoX2 inhibitor rofecoxib led to premature closure of the viCtor trial. in the final report of the trial, rofecoxib did not improve overall survival or protect from cancer recurrence in unselected patients. in addition, CoX2 expression did not correlate with overall prognosis or predict the effectiveness of CoX2 inhibitors.9

Finally, compelling evidence indicates that the avoidance of smoking and heavy alcohol use, prevention of weight gain, and mainte‑nance of a reasonable level of physical acti‑vity are associated with markedly lower risks of CrC—suggesting that one in four cancers could be prevented if all four lifestyle recom‑mendations were adhered to.10 notably, many of the lifestyle risk factors are shared between cardiovascular disease and cancer and, importantly, there are no adverse effects in terms of increased gastrointestinal bleed‑ing or cardiovascular events associated with adjusting these lifestyle issues. Consequently, the role of aspirin and other CoX inhibi‑tors in the prevention of CrC remains un resolved. in the balance between benefit and risk it is probably still safest to say that an apple a day keeps the doctor away.

Department of Surgery, Stavanger University Hospital, Armauer Hansens vei 20, POB 8100, N-4068 Stavanger, Norway (M. Berg, K. Søreide).

Practice point

Long-term use of aspirin decreases the risk of proximal colon cancer, but not of the distal colon and rectum.

gEnETics

An 18-gene signature (ColoPrint®) for colon cancer prognosisIain B. Tan and Patrick Tan

coloprint® is an 18-gene expression signature designed to predict disease relapse in patients with early-stage colorectal cancer (cRc). we discuss the potential impact of coloprint® on clinical practice, and its contribution to our knowledge of cRc molecular heterogeneity.Tan, i. B. & Tan, P. Nat. Rev. Clin. Oncol. 8, 131–133 (2011); published online 8 February 2011; doi:10.1038/nrclinonc.2010.229

early‑stage breast cancer. recently, both companies have published gene‑expression classifiers for predicting disease relapse in early‑stage colorectal cancer (CrC).1,2 Here, we discuss the potential clinical and scientific impact of one of these classifiers —ColoPrint® (agendia).

many oncologists are familiar with mammaPrint® (agendia, amsterdam, the netherlands) and oncotype DX® (Genomic Health, redwood City, Ca, usa), two multi‑gene assays approved by the FDa to predict disease relapse and guide adju‑vant therapy decisions in patients with

nrclinonc_NV_MAR11.indd 131 4/2/11 14:43:38

© 2011 Macmillan Publishers Limited. All rights reserved

132 | MARCH 2011 | voluMe 8 www.nature.com/nrclinonc

nEws & viEws

CrC is the third leading cause of global cancer mortality. outcomes for patients with early‑stage CrC are heterogeneous, with 5‑year survival rates ranging from 72–83% in stage ii disease and 44–83% in stage iii disease.3 in the past two decades, random‑ized trials have demonstrated a survival advantage for patients treated with surgery and adjuvant chemotherapy,4 particularly those with stage iii disease. However, in these trials, many patients were cured by surgery alone, suggesting that it might be possible to omit chemotherapy in selected patients. Clinical guidelines currently recom mend observation for stage i disease and adju‑vant chemotherapy with a combination of a fluoro pyrimidine and oxaliplatin for those with stage iii disease.

in stage ii CrC, the benefit of adjuvant chemotherapy is contentious, with asCo recommending the integration of clinical risk criteria to select patients for adjuvant therapy.5 identifying molecular markers that can inform therapeutic decisions, such as the need for treatment and type of adju‑vant therapy, would be tremendously useful. to address this challenge, salazar et al.1 analyzed fresh‑frozen tumor tissues from 188 patients with stage i to iv CrC using agilent gene‑expression micro arrays. By cor‑relating the expression of more than 40,000 genes with disease metastasis‑free survival, they identified an optimal set of 18 genes that was used to construct the ColoPrint® prog nostic classifier. in an indepen dent validation series of 206 patients with stage i to iii CrC, 60% of patients were classified as ‘low‑risk’ with a 5‑year relapse‑free sur‑vival (rFs) rate of 87.6%. the remaining 40% ‘high‑risk’ patients exhibited a rFs rate of 67.2% (hazard ratio [Hr] = 2.5; 95% Ci 1.33–4.73; P = 0.005). in multivariate analy‑ses, ColoPrint® remained one of most signifi‑cant prognostic factors (Hr = 2.69; P = 0.003), and in stage ii CrC, ColoPrint® was superior to the asCo cri teria for the assessment of cancer recurrence risk (Hr = 3.34; P = 0.017). the authors concluded that compared with conventional clinicopathological criteria

alone, ColoPrint® provides more accurate informa tion on the risk of recurrence and may facilitate selection of low‑risk patients who can be spared chemotherapy.

while these results are encouraging, it is prudent to interpret them in the context of an early discovery study. several gene‑expression classifiers for predicting CrC relapse have been described,6,7 but none have achieved clinical utility. it is worth noting that studies relying on fresh‑frozen tissues (for example ColoPrint®) typically have modest sample sizes and cannot benefit from archi‑val material collected from randomized clini‑cal trials. as a comparison, oncotype DX® (colon), the parallel CrC prognostic classi‑fier developed using formalin‑ fixed paraffin‑embedded tissues, was tested in more than 1,800 patients from four adjuvant trials.2 therefore, further retro spective validation of ColoPrint® in large independent cohorts is clearly required. Fortunately, a prospec‑tive study, ParsC (Prospective study for the assessment of recurrence risk in stage ii Colorectal patients using ColoPrint) has already been initiated to evaluate the perfor‑mance of ColoPrint® in the classification of patients in the clinical setting.8

the potential for gene signatures to influ‑ence treatment decisions depends on the disease stage. molecular markers are most likely to impact the management of stage ii disease, where the need for adjuvant chemo‑therapy is already based on assessment of clinical‑risk features. molecularly, micro‑satellite instability (msi) status is a marker of good prognosis in patients with stage ii CrC and may be associated with a lack of benefit from adjuvant fluoro pyrimidine therapy.9 indeed, most msi high (msi‑H) patients were identified as ‘low risk’ by ColoPrint®. However, the 48% discordance observed between ColoPrint® and the asCo clinical risk criteria1 suggests an additional discrimi‑native value of ColoPrint® beyond clinical charac teristics. stage ii patients identified as ‘low‑risk’ by ColoPrint® exhibited an excel‑lent 5‑year survival similar to that seen for stage i disease, raising the possibility that ColoPrint® may identify stage ii patients for whom chemotherapy can be avoided. that said, we must remember that good prognosis does not necessarily mean lack of benefit from adjuvant therapy. For example, oncotype DX® (colon) has not been shown to be predictive in stage ii CrC, despite its prognostic significance.2 Further studies should also be performed to establish if ColoPrint® is purely prognostic or whether it is predictive of treatment benefit as well.

in stage iii CrC, adjuvant chemo therapy is the standard of care. in our opinion, onco‑logists are highly unlikely to omit chemo‑therapy altogether in medically fit patients with stage iii CrC unless the data support‑ing excellent prognosis in molecularly low‑risk patients is very compelling. the study by salazar et al.1 cannot address the role of ColoPrint® in stage iii disease, since there were only 62 patients with stage iii disease and there was a trend towards inferior rFs in high‑risk patients (P = 0.1). nevertheless, a validated prognostic signature for stage iii CrC patients might still be useful to iden‑tify low‑risk patients for whom oxaliplatin chemo therapy might be omitted and who might be treated with a fluoro pyrimidine alone. moreover, with the exception of oxali platin, stage iii CrC has demonstrated notable failures for drugs that were efficacious in the metastatic setting, such as bevacizumab, cetuximab and irinotecan. Given the curative intent of treatment in stage iii CrC and the vast investment into these completed trials, it might be fruitful to search for molecu lar markers predictive of selective benefit for therapies that other wise do not provide an advantage in an unselected population.

the present study also broadens our knowledge regarding the inherent molecu‑lar heterogeneity of CrC.1 using unsuper‑vised clustering techniques, three molecular subgroups were identified that had differ‑ent survival outcomes. these groups were differentially enriched for BRAF activating mutations and msi‑H, suggesting unique underlying biologies. notably, only the largest subgroup (n = 110) was used to develop the prognostic signature. Given the distinct biological makeup of these three groups, it is plausible that the prognostic impact of ColoPrint® is specific to the bio logical sub‑group from which it was developed, analo‑gous to the questionable prognostic value of oncotype DX® in Her2‑positive breast cancer.10 salazar et al.1 do not provide prog‑nostic information of ColoPrint® in the three

Practice point

in colorectal cancer, novel molecular markers such as gene-expression signatures, offer the potential of improving upon current prognostic models that are based on clinical criteria. However, widespread acceptance of these markers will necessitate identifying opportunities where they directly influence clinical management decisions.

nrclinonc_NV_MAR11.indd 132 4/2/11 14:43:39

© 2011 Macmillan Publishers Limited. All rights reserved

nature reviews | clinical oncology volume 8 | marCH 2011 | 133

nEws & viEws

biological subgroups—this should also be addressed in a future study. investigations addressing the relationship of ColoPrint® to other molecular markers (for example, 18q loss of heterozygosity, KRAS mutation status and CpG island methylation subtypes) are also warranted.

in conclusion, salazar and colleagues are to be commended for their promising findings that ColoPrint® might provide additional prognostic information beyond clinicopathological criteria in early‑stage CrC. we eagerly await the results of the ongoing clinical trial seeking to prospectively validate ColoPrint®.Department of Medical Oncology, National Cancer Centre Singapore, 11 Hospital Drive, Singapore 169610 (I. B. Tan). Cancer and Stem Cell Biology, Duke–NUS Graduate Medical School Singapore, 8 College Road, Singapore 169857 (P. Tan). Correspondence to: P. Tan gmstanp@duke-nus.edu.sg

competing interestsThe authors declare no competing interests.

1. salazar, R. et al. Gene expression signature to improve prognosis prediction of stage ii and iii colorectal cancer. J. Clin. Oncol. doi:10.1200/JCO.2010.30.1077.

2. O’Connell, M. J. et al. Relationship between tumor gene expression and recurrence in four independent studies of patients with stage ii/iii colon cancer treated with surgery alone or surgery plus adjuvant fluorouracil plus leucovorin. J. Clin. Oncol. 28, 3937–3944 (2010).

3. O’Connell, J. B., Maggard, M. A. & Ko, C. Y. Colon cancer survival rates with the new American Joint Committee on Cancer sixth edition staging. J. Natl Cancer Inst. 96, 1420–1425 (2004).

4. Cunningham, D. et al. Colorectal cancer. Lancet 375, 1030–1047 (2010).

5. Benson, A. B. 3rd et al. American society of Clinical Oncology recommendations on adjuvant chemotherapy for stage ii colon cancer. J. Clin. Oncol. 22, 3408–3419 (2004).

6. wang, Y. et al. Gene expression profiles and molecular markers to predict recurrence of Dukes’ B colon cancer. J. Clin. Oncol. 22, 1564–1571 (2004).

7. eschrich, s. et al. Molecular staging for survival prediction of colorectal cancer patients. J. Clin. Oncol. 23, 3526–3535 (2005).

8. salazar, R. et al. The PARsC trial, a prospective study for the assessment of recurrence risk in stage ii colon cancer (CC) patients using ColoPrint [abstract]. J. Clin. Oncol. 28 (suppl. 15), aTPs199 (2010).

9. Guastadisegni, C., Colafranceschi, M., Ottini, L. & Dogliotti, e. Microsatellite instability as a marker of prognosis and response to therapy: a meta-analysis of colorectal cancer survival data. Eur. J. Cancer 46, 2788–2798 (2010).

10. Rosman, M., Mylander, w. C. & Tafra, L. what is the value of the 21 gene recurrence score in HeR2-negative patients? J. Clin. Oncol. 28, e647 (2010).

WIN Consortium—challenges and advancesJohn Mendelsohn, Thomas Tursz, Richard L. Schilsky and Vladimir Lazar

The importance of appropriate patient selection necessitates novel clinical trial design and biomarker-driven trials to allow delivery of the right drug to the right patient at the right time—personalized cancer medicine. The win consortium promotes collaboration between critical stakeholders and offers diverse populations of cancer patients the opportunity to participate in clinical trials with new drugs and biologics that target their tumor.Mendelsohn, J. et al. Nat. Rev. Clin. Oncol. 8, 133–134 (2011); doi:10.1038/nrclinonc.2010.230

FocUs on pERsonaliZED MEDicinE

that identify abnormal genes or gene prod‑ucts, can enhance clinical outcomes.1 the Battle lung cancer trial is discussed in this issue2 as a new approach to early diagnosis and treatment of cancer. another advance in lung cancer was demon strated by preselect‑ing for the presence of a translocation in ALK that is present in less than 5% of patients with non‑small‑cell lung cancer. administration of the anti‑alK drug PF‑02341066 produced response rates of more than 50%.3

these trials are among a number of studies that have provided important lessons for clinical research. By preselecting for patients with abnormalities in the specific targets for which new drugs have been designed, it is likely that efficacy can be demonstrated (or will not be demonstrated) far more quickly and efficiently than in an unselected patient population. this means that diagnostic tests for biomarkers must be developed in parallel with new drugs. it also means that for trials with drugs targeting the products of genes that are not commonly mutated or functioning abnormally in human cancers, it will be necessary to screen cancers from large numbers of patients derived from diverse populations to identify appropriate candidates for treatment.

while these research observations are promising, too few patients are being placed on such clinical trials today—in fact, less than 5% even in the most developed coun‑tries. Cancer patients urgently need increased access to biomarker‑driven trials with new targeted therapies.

with these considerations in mind, the worldwide innovative networking (win) Consortium was initiated by the institut Gustave roussy (France) and the university of texas mD anderson Cancer Center (usa), as a non‑profit non‑governmental organization bringing together cancer

the worldwide annual cancer incidence is 14 million people, and recent advances in cancer research are helping to create new possibilities for the treatment and manage‑ment of therapy for this disease. it is well known that cancer is caused by the abnor‑mal function of genes that control cell prolif‑eration and function, and we have extensive information on the intricate cellular path‑ways and networks in which the molecular products of these genes participate. the genes in a genome can now be sequenced in days rather than years, and at costs that are approaching the practical limits for clinical utility. normal and abnormal gene products can be detected with immuno histochemistry and proteomic techniques that are steadily improving. Biotechnology and drug com‑panies and academic institutions have the capability of being far more efficient in designing new anticancer drugs and bio‑logical agents and testing their preclinical efficacy. By combining these advances, it has become possible to design clinical trials in which a patient is assigned a drug based on the presence of biomarkers in that patient’s tumor. this approach often results in high response rates for the effective drugs and biologics that target these biomarkers.

the past two decades have witnessed the successful introduction of targeted agents such as imatinib, bevacizumab, trastuzumab, cetuximab, gefitinib and erlotinib for the treatment of hematologic and solid tumors. recent presentations at the 2010 meetings of the american association for Cancer research and asCo have highlighted the acceleration of accomplishments in person‑alized cancer therapy for the treatment of lung cancer. For instance, the Battle clini‑cal trial showed that adaptive randomization of new drugs in the phase ii setting, based on biomarkers in a patient’s tumor biopsy

nrclinonc_NV_MAR11.indd 133 4/2/11 14:43:39

© 2011 Macmillan Publishers Limited. All rights reserved