Sie sind auf Seite 1von 5

PERSPECTIVES

OPINION those of the gastrointestinal (GI) tract.36


This association is partially explained by
Oligometastases revisited the venous drainage of the GI tract travel-
ling through the portal vein, although tumor
Ralph R. Weichselbaum and Samuel Hellman and host genetic factors also have a role in
the tropism of GI metastases to the liver.
Abstract | We previously proposed a clinical state of metastasis termed Support for the oligometastatic state of some
oligometastases that refers to restricted tumor metastatic capacity. The implication GIassociated liver metastases comes from
of this concept is that local cancer treatments are curative in a proportion of data regarding the outcome of patients with
patients with metastases. Here we review clinical and laboratory data that support colorectal cancer who underwent partial
the hypothesis that oligometastasis is a distinct clinical entity. Investigations of the liver resection for metastasis to the liver
prevalence, mechanism of occurrence, and position in the metastatic cascade, as (Table1).36 The curative treatment of liver
metastasis has been reported for many years;
well as the determination of molecular markers to distinguish oligometastatic from
however, each series was relatively small. In
polymetastatic disease, are ongoing.
1986, Hughes and colleagues compiled a
Weichselbaum, R.R. & Hellman, S. Nat. Rev. Clin. Oncol. 8, 378382 (2011); published online 22 March 2011; registry of 607 patients from 24 institutions
doi:10.1038/nrclinonc.2011.44 who had undergone curative resection for
liver metastasis from colorectal cancer; 25%
Introduction metastases from colon cancer,36 lung meta of these patients were disease free at 5years.3
Current methods of cancer staging and stases from a variety of primary sites,7 and Poor prognostic factors included positive
treatment frequently separate patients into adrenal metastases from lung cancer,8 result surgical margins and bilobar disease.3 In
groups with tumors confined to the primary in cure in some patients. A major question 1996, Nordlinger etal.4 published a multi-
site or regional spread (lymph-node meta is the prevalence of oligometastasis. Since institutional European registry study of
stases).1 These patients are usually treated current research indicates that tumors evolve 1,568 patients who underwent liver resection
with curative intent in contrast to patients with in their malignant capacity, the improving for metastasis; the 5year survival rate was
distant metastasis who are not usually con- imaging methods (for example, MRI and 28%. Adverse prognostic factors included
sidered curable by current regional treat- PET), as well as blood-based tests (such as a positive surgical margin of the resected
ment methods. In most adult solid cancers, prostate-specific antigen levels), might iden- metastasis, serosal or lymphatic spread of the
the treatment for metastasis is systemic tify a significant population of patients with primary tumor, large size and high number
cytotoxic chemotherapy and/or hormonal oligometastases and afford opportunities for of liver metastasis, high carcinoembryonic
manipulation. In general, distant metastases detecting primary tumors early in their pro- antigen (CEA) levels and age (<60years).
are the primary cause of cancer mortality and gression as well as permitting early diagnosis In a single institution study that highlighted
have led to a leukemia-like consideration of of oligometastases amenable to curative local patients likely to benefit from liver resection,
solid tumor metastases whereby metastases treatment. More recently, the use of stereo surgeons analyzed results from 1,001 liver
are frequently, if not always, considered tactic radiotherapy, radiofrequency abla- resections for metastatic colorectal cancer.5
extensive in number and organ site. We pro- tion, and MRI-guided focused ultrasound Poor prognostic factors included the pres-
posed an intermediate state of metastases offer less invasive methods of regional treat- ence of extrahepatic disease, a node-positive
termed oligometastases.2 In this concept, ment than surgical resection, and could offer primary tumor, size and number of meta-
the number and site of metastatic tumors are curative potential in the treatment of oligo static lesions, a short disease-free interval
limited. We suggested that the evolution of metastases.914 However, longer follow-up and between the primary tumor and metastasis,
metastatic capacity has intermediate states larger patient numbers are required to assess and a CEA level of greater than 200ng/ml.
in which spread may be limited to specific the curative potential of stereotactic radio- The overall survival rate was 37% at 5years,
organs and metastases might be present in therapy and other ablative techniques for and 22% at 10years. 5 The investigators
limited numbers. The clinical implication oligometastases. We review the surgical treat- devised risk categories based on these data
of this hypothesis is that localized forms of ment of liver and lung metastases as support and the above criteria to determine patients
cancer treatment may be effective in patients for the oligometastatic concept as well as suitable for resection based on the number
with oligometastases. This is clearly the case novel laboratory investigations pertinent to of risk factors present. In 2005, Pawlik etal.6
since curative surgical resection of liver the selection of appropriate candidates for analyzed 557 patients from three institutes
localized treatment of oligometastases. who underwent liver resection for meta
Competing interests stasis. These investigators identified adverse
R.R. Weichselbaum declares associations with Treatment of oligometastases prognostic factors to be a positive surgical
the following companies: GenVec, Catherex,
Liver metastases margin, more than three metastases and a
Reflexion. S. Hellman declares associations with
the following companies: InSightec. See the Liver metastases have a role in the mortality CEA level of >200ng/ml. The 5year survival
article online for full details of the relationships. of many patients with cancer, in particular rate was 58% in patients without any poor

378 | JUNE 2011 | VOLUME 8  www.nature.com/nrclinonc


2011 Macmillan Publishers Limited. All rights reserved
F O C U S O N M E TA S TA S I S

prognostic factors. Strikingly, the 5year sur- Table 1 | Summary of four large series of resection of hepatic metastasis
vival was 17% in patients with positive surgi-
Study n 5-year survival rate (%) 10-year survival rate (%)
cal margins compared with 63% in patients
with negative marginsthe width of the Hughes etal. (1986) 3
607 33 No 10-year follow up
margin did not affect the outcome.6 Most of Nordlinger etal. (1996)4 1,568 28 No 10-year follow up
these studies included patients treated before Fong etal. (1999)5 1,001 37 22
transaxial body imaging (CT and MRI) or Pawlik etal. (2005) 6
557 58 No 10-year follow up
PET was frequently used for the detection
of metastases and for staging; therefore, it
is likely that these advanced imaging and from sarcomas and germ-cell tumors had stasis even if the liver and nodal tumors were
molecular diagnostic techniques could be been reported before the report from the stabilized by chemotherapy.19 Administration
used to aid the selection of patients for cura- International Registry of Lung Metastases. of cetuximab with or without chemotherapy
tive treatment of metastatic disease. Indeed, A subset of patients with tumors consid- in patients with liver metastasis who failed
these techniques were used in some patients ered to be widely metastatic in all instances, first-line chemotherapy resulted in a small
in the most recent of the studies6 and likely such as breast cancer and melanoma, were percentage of patients sufficiently down-
contributed to better patient selection and reported to be cured with surgical resection staged to undergo curative surgery; eight
improved 5year survival in this study of lung metastases if certain clinical crite- out of 11 patients who had complete surgical
compared with the other trials (Table1).35 ria were used for patient selection (usually resection were alive at 36months.20
the length of the disease-free interval and
Lung metastases number ofmetastases).7 The biology of metastasis
Pulmonary metastases are a primary cause Common clinical prognostic factors Paget first hypothesized the seed and soil
of death from cancer.7 As for the liver, the emerge from these large investigations of hypothesis, which suggested that metastases
lung is also a primary or secondary drain- resection and outcome in lung and liver were not random and not solely depen-
age region for many frequently occurring metastasis. These include the number and dent on circulatory patterns, but rather an
cancers. Specific genes have been identified size of metastasis, the interval from the treat- interaction between the tumor cell and the
that are associated with lung metastases ment of the primary tumor to the appear- targeted organ. 21 In this concept, certain
from breast cancer suggesting a genetic ance of metastasis, adequacy of resection tumors have a predilection for metastasis to
basis for metastatic tropism.15,16 In addition, of the metastatic tumors, and the presence of particular organs that support the second
bone marrow-derived cells might establish a multiple metastatic sites. The histology ary growth of cells from the primary tumor.
favorable microenvironment in the lung for of the primary tumor also seems to influ- The selective process is driven by a tumor
development of metastasis.17,18 Successful ence outcome. These data form the basis of microenvironment that is hypoxic and acidic
results of curative resection of lung meta clinical markers for patient selection for the with immune-derived cells and other host-
stases have been described for almost all treatment of oligometastatic disease. derived cells that promote tumor growth and
types of cancer and these data were sum- suppresses host immunity. Tumor diversity
marized in a report from the International Response to chemotherapy is driven by the genetic instability of the
Registry of Lung Metastases.7 This report In our original article on oligometastasis,2 tumor cells due to telomere erosion, muta-
detailed results from pulmonary resection we speculated that a state of oligometastasis tions in tumor-suppressor and DNA-repair
of 5,206 patients with pulmonary meta might be created following cytoreduction as genes, and intrinsic tumor metabolism
stases with controlled primary tumors and a consequence of excellent responses to sys- (aerobic glycolyis) that is toxic to surround-
with metastases confined to the lung from a temic treatments and when local treatments ing normal cells. 18,22,23 The evolutionary
wide variety of histological tumor types. A might add to cure. One interesting example value to the primary tumor of harboring
total of 4,572 patients underwent complete of this concept is the neoadjuvant chemo- clones that metastasize may not necessar-
resection, 1,984 had epithelial tumors and therapy treatment of patients with liver ily directly aid in the growth of the primary
1,917 had sarcomas.7 In this study, patients metastasis who then undergo resection. For tumor. The capacity for metastatic spread
who underwent complete resection of meta example, Giacchetti etal.19 reported that of is likely an epiphenomenon of the genetic
static tumors had a 5year survival rate of 151 patients with liver metastasis who were instability of the primary tumor resulting in
36% and a 10-year survival rate of 26% unresectable (owing to the size or multi the capacity to grow and invade and give rise
(Figure1). Patients with fewer metastases nodularity of the metastasis) who received to cells capable of distant metastases.
and a longer disease-free interval between chemotherapy (oxaliplatin, 5fluorouracil
the primary tumor and appearance of lung and leucovorin), 77 became resectable Steps in metastasis
metastases had 10-year survival rates as high and complete resection was achieved in Many investigators have hypothesized and
as 40%. Similar to the liver metastases data, 58patients.19 Almost all patients who were have provided evidence to support the
patients with incomplete resection had poor not operated on died within 5years but various steps in the metastatic cascade.
survival rates of 13% and 7% at 5years and 50% of patients who achieved a response to These discreet steps in metastasis have
10years, respectively. Multiple surgeries were chemotherapy and were resected were alive been elegantly summarized by Gupta and
required in some patients and histology had at 5years. Negative predictors for survival Massagu.24 These steps include loss of cellu-
a role in outcome; patients with germ-cell in patients undergoing chemotherapy before lar adhesion, increased motility, invasiveness
tumors had the best outcome and patients liver resection included tumor progression of the primary tumor, entry into and survival
with melanoma had the worst outcome. during chemotherapy and the presence of in the circulation, entry into new organs and
Curative resection of pulmonary metastases celiac or para-aortic lymph-node meta eventual colonization of these organs. Gupta

NATURE REVIEWS | CLINICAL ONCOLOGY VOLUME 8 | JUNE 2011 | 379


2011 Macmillan Publishers Limited. All rights reserved
PERSPECTIVES

100 Group Patients Deaths pre-existing tumor clones.30 These results


I = no risk factors 819 349 were confirmed and expanded upon by
II = 1 risk factor 1,720 903
several investigators using KHT sarcoma
80 cells, which varied widely in metastatic
III = 2 risk factors 1,553 972
potential (lung colonization).31,32 Cillo etal.33
noted that clones that had a high metastatic
60 potential were more resistant to chemo
Survival (%)

therapy than cells derived from clones of low


metastatic potential. In addition, Khodarev
40
etal.34 used the B16F1 mouse melanoma
model to show that repeated passage through
the lungs evolved a more aggressive pheno-
type defined by an increase of efficiency in
20
lung colonization and conferred resistance
to radiotherapy and chemotherapy on the
more highly metastatic clones; this finding
0 is consistent with earlier reports.34 Of inter-
0 60 120 180 est is the finding that clones that give rise
Time (months)
to widely metastatic tumors overexpressed
Figure 1 | Survival of patients undergoing pulmonary resection of metastatic tumors. Each curve genes known to be induced by interferon,
represents the survival of patients with an increasing number of risk factors for recurrence as which had previously been associated with
determined by a retrospective review of the data.7 These categories are: groupI, a single
resistance to radiotherapy and chemotherapy
resectable metastasis with a disease-free interval from primary tumor to metastasis of
36months; group II, multiple metastases or a disease-free interval <36months; group III, in murine models and in human databases
multiple metastases and a disease-free interval <36months. The size, number and tumor type and tumor samples.35,36
are risk factors for recurrence. Permission obtained from Elsevier Pastorino, U. etal. J.Thorac. Considered together, these reports suggest
Cardiovasc. Surg. 113, 3749 (1997). that there are large differences between meta
static capability in tumor cells in experimental
and Massagu have defined a useful frame- aggregates without forming established models, consistent with the concept of oligo-
work to categorize genes that have roles in tumors.25,26 Several hypotheses might explain metastases. Genes that govern the metastatic
the various stages of metastasis.24 They term this phenomenon:2729 the tumor cells fail to process are of great interest; however, genes
genes that confer a selective advantage to signal appropriate angiogenic cues or the that govern the number of metastases have
the primary tumor cells as metastasis initi rate of blood vessel formation equals the not been studied in detail. Recently, Wuttig
ation genes; metastasis progression genes apoptosis in tumor endothelia and adequate etal.37 used samples from 18 patients with
are those that fulfill rate-limiting functions neovasculature does not form; the local renal cell carcinoma to identify genes that
in colonization; and metastasis virulence microenvironment does not support meta- characterized few (less than eight) or many
genes are those that provide an advantage in static growth (for example it does not have (>16) pulmonary metastases. DNA-array
metastasis colonization but not necessarily appropriate environmental growth factors analysis on fresh samples obtained from
an advantage in the growth of the primary or cell contact structures); and the immune pulmonary resection of metastasis revealed
tumor. Metastasis colonization requires a system monitors tumors and eliminates 135 genes that were differentially expressed
selective advantage in the evolution of the incipient tumor cells (surveillance), keeps between the few and many metastasis
primary tumor to be preserved and ampli- tumors at a small or microscopic size (equi- groups. Using gene ontology enrichment
fied during tumor progression.24 The fol- librium; perhaps analogous to dormancy), analysis the researchers demonstrated that
lowing characteristics of the metastatic or allows escape to established tumor polymetastatic tumors were enriched by
phenotype are altered cell adhesion, intra- growth.29 Within the context of dormancy, genes that positively regulate the cell cycle,
vasation, survival in the circulation, extra these processes might have an effect on the indicating an increase in growth potential
vasation, seeding in a distant site, invasion, number, location, and timing of appearance in polymetastasis versus oligometastasis.37
and development of the appropriate micro of metastases; interferon signaling is a key Based on a meta-analysis of these data37 and
environment in host organs. These processes component of these processes.29 previously published data,38 an 11-gene clas-
have been reviewed by others in this focus sifier was established to predict the number
issue and elsewhere. These data indicate Models and molecules of metastases in patients with renal cell carci
that there are primary tumor cells that have Investigations of laboratory and molecular noma. Ideally, prospective trials will collect
limited capability in one or more of the nec- determinants of oligometastasis are limited. fresh-frozen tumor specimens for molecu-
essary biological requirements for metastasis; However, Fidler and Kripke30 compared the lar analysis to validate this signature as well
thus the origin of oligometastases. Tumor metastatic ability of different tumor-cell as identify other genes and gene families
dormancy may be a particular example of clones derived from B16F1 melanoma. They involved in oligometastasis.
limited metastatic capacity. noted a wide variation in the ability of these
clones to colonize the lung (from 3.5 to 260 Temporal evolution
Dormancy lung colonies per mouse) and concluded Recently, insights into the evolution of
Cells that escape from the primary tumor that the B16F1 cell line is heterogenous with metastasis in pancreatic cancer were
may be present at distant organs as cellular regard to the metastasis-forming ability of reported;39,40 this is particularly interesting

380 | JUNE 2011 | VOLUME 8  www.nature.com/nrclinonc


2011 Macmillan Publishers Limited. All rights reserved
F O C U S O N M E TA S TA S I S
metastatic colorectal cancer: analysis of 1001
since oligometastases are rarely observed paradigm. However, the oligometastatic consecutive cases. Ann. Surg. 230, 309318
in pancreatic cancer. Two research groups state is becoming more frequently identi- (1999).
analyzed somatic mutations in genes and fied with more-sensitive methods of detect- 6. Pawlik, T.M. etal. Effect of surgical margin
constructed clonal evolutionary maps ing such oligometastases. As newer methods status on survival and site of recurrence after
hepatic resection for colorectal metastases.
derived from primary and metastatic of analyzing patients with metastatic disease Ann. Surg. 241, 715722 (2005).
tumors.39,40 They reported that pancreatic develop (for example the analysis of circu- 7. Pastorino, U. et al. Long-term results of lung
tumors contain geographically separate lating tumor cells), the oligom etastatic metastasectomy: prognostic analyses based
on 5206 cases. The International Registry of
subclones within the primary tumor that state should be identified even earlier. As Lung Metastases. J.Thorac. Cardiovasc. Surg.
are present many years before metastases indicated above, new molecular analysis 113, 3749 (1997).
become evident. As an example, it was of pancreatic cancer suggests that there 8. Strong, V.E. etal. Laparoscopic adrenalectomy
for isolated adrenal metastasis. Ann. Surg.
demonstrated that long time periods existed are temporal differences in the appear-
Oncol. 14, 33923400 (2007).
between the appearance of the tumor and the ance of metastases based on tumor genet- 9. Ben-Josef, E. & Lawrence, T.S. Using a bigger
initial metastasis. Also, different metastatic ics. These data suggest a potential stepwise hammer: the role of stereotactic body
clones may appear at different time periods. progression with intermediate stages of radiotherapy in the management of
oligometastases. J.Clin. Oncol. 27, 15371539
It is suggested that there may be a hierarchy limited metastatic capacity. It seems quite (2009).
of metastatic sites for pancreatic cancer with possible that metastases from tumors with 10. Salama, J.K. etal. An initial report of a
peritoneal metastases occurring first, then such limited capacities might be separated radiation dose-escalation trial in patients with
one to five sites of metastatic disease. Clin.
hepatic, and finally pulmonary metastases.40 from those much further along in malig- Cancer Res. 14, 52555259 (2008).
The nature of this hierarchy is likely to vary nant progression. If this seems possible, 11. Rusthoven, K.E. etal. Multi-institutional
with primary tumor histology and loca- then clinicians will be able to limit ablative phaseI/II trial of stereotactic body radiation
therapy for lung metastases. J.Clin. Oncol. 27,
tion; the latter because of different vascular local treatment to only those patients with
15791584 (2009).
drainage, local hypoxia and possible local true oligometastases. Finally, there seems 12. Lee, M.T. etal. PhaseI study of individualized
stromal interactions. For example, adrenal to be another type of oligometastasis evolv- stereotactic body radiotherapy of liver
metastases may occur early in the process in ing; those limited remaining tumor deposits metastases. J. Clin. Oncol. 27, 15851591
(2009).
non-small-cell lung cancer, while for colon following successful eradication of all other 13. Macdermed, D.M., Weichselbaum, R.R. &
cancer the site of early metastases is the apparent and occult cancer cells by systemic Salama, J.K. A rationale for the targeted
liver, perhaps owing to the portal drainage means. It is important to study both types treatment of oligometastases with radiotherapy.
J.Surg. Oncol. 98, 202206 (2008).
pattern. Other examples include the lungs of limited metastases but recognize that the 14. Fakiris, A.J. etal. Stereotactic body radiation
as an initial site for soft-tissue and bone sar- true oligometastases are present because therapy for early-stage nonsmallcell lung
comas perhaps because of the venous return of limited metastatic competence while carcinoma: four-year results of a prospectice
phaseII study. Int. J.Radiat. Oncol. Biol. Phys.
from the tumor going first through the induced oligometastases following other-
75, 677682 (2009).
pulmonary circulation or from local stromal wise successful systemic treatment have 15. Minn, A.J. etal. Genes that mediate breast
considerations. Such considerations of the more extensive malignant capacities and cancer metastasis to lung. Nature 436,
sites and mechanisms of metastases early in were spared from eradication by pharmaco 518524 (2005).
16. Minn, A.J. etal. Lung metastasis genes couple
the metastatic progression should produce logical means, local immunological condi- breast tumor size and metastatic spread. Proc.
opportunities for site-specific analyses of tions, or from the development of resistant Natl Acad. Sci. USA 104, 67406745 (2007).
metastases from different primary tumors clones. The strategy and tactics of treatment 17. Kaplan, R.N., Psaila, B. & Lyden, D. Bone
marrow cells in the pre-metastatic niche:
as well as analysis of differential gene expres- of these two types might be different as well within bone and beyond. Cancer Metastasis
sion as a function of the site of the primary as the likelihood of therapy success. Rev. 25, 521529 (2006).
tumor and of oligometastases. These obser- Department of Radiation and Cellular Oncology, 18. Erler, J.T. etal. Lysyl oxidase is essential for
vations put a special emphasis on the early hypoxia-induced metastasis. Nature 440,
The Ludwig Center for Metastasis Research,
12221226 (2006).
detection of the oligometastatic state and The University of Chicago Medical Center, 5758
19. Giacchetti, S. etal. Long-term survival of
local control of the primary tumor because if South Maryland Avenue, Chicago, IL60637, patients with unresectable colorectal cancer
USA (R.R. Weichselbaum, S.Hellman). liver metastases following infusional
metastasis is hierarchal in time and number,
Correspondence to: R.R. Weichselbaum chemotherapy with 5fluorouracil, leucovorin,
ablation of early metastasis might be cura- oxaliplatin and surgery. Ann. Oncol. 10,
rrw@radonc.uchicago.edu
tive. Since the metastatic tumor clone arises 663669 (1999).
from the primary tumor, control of the 1. Edge, S.B. et al. (Eds) AJCC Cancer Staging 20. Lvi, F. etal. Cetuximab and circadian
Manual 7th edn (Springer, New York, 2010). chronomodulated chemotherapy as salvage
primary tumor takes on special importance 2. Hellman, S. & Weichselbaum, R.R. treatment for metastatic colorectal cancer
in the cure of oligometastasis and perhaps Oligometastases. J.Clin. Oncol. 13, 810 (mCRC): safety, efficacy and improved
also polymetastatic disease. We are conduct- (1995). secondary surgical resectability. Cancer
ing investigations that are ongoing to study 3. Hughes, K.S. etal. Resection of the liver for Chemother. Pharmacol. 67, 339348 (2011).
colorectal carcinoma metastases: a multi- 21. Paget, S. The distribution of secondary growths
differences between oligometastases and institutional study of patterns of recurrence. in cancer of the breast. 1889. Cancer
polymetastases in both clinical material and Surgery 100, 278284 (1986). Metastasis Rev. 8, 98101 (1989).
experimental tumor models. 4. Nordlinger, B. etal. Surgical resection of 22. Deng, C.X. BRCA1: cell cycle checkpoint,
colorectal carcinoma metastases to the liver. genetic instability, DNA damage response and
Aprognostic scoring system to improve case cancer evolution. Nucleic Acids Res. 34,
Conclusions selection, based on 1568 patients. Association 14161426 (2006).
The metastases that we define as oligo- Franaise de Chirurgie. Cancer 77, 12541262 23. Rajagopalan, H. & Lengauer, C. Aneuploidy and
(1996). cancer. Nature 432, 338341 (2004).
metastases have long been recognized as
5. Fong, Y., Fortner, J., Sun, R.L., Brennan, M.F. & 24. Gupta, G.P. & Massagu, J. Cancer metastasis:
potentially curable but were considered to Blumgart, L.H. Clinical score for predicting building a framework. Cell 127, 679695
be rare exceptions to the cancer metastasis recurrence after hepatic resection for (2006).

NATURE REVIEWS | CLINICAL ONCOLOGY VOLUME 8 | JUNE 2011 | 381


2011 Macmillan Publishers Limited. All rights reserved
PERSPECTIVES
25. Aguirre-Ghiso, J.A. Models, mechanisms and 32. Harris, J.F., Chambers, A.F., Hill, R.P. & Ling, V. 37. Wuttig, D. etal. Gene signatures of pulmonary
clinical evidence for cancer dormancy. Nat. Rev. Metastatic variants are generated metastases of renal cell carcinoma reflect the
Cancer 7, 834846 (2007). spontaneously at a high rate in mouse KHT disease-free interval and the number of
26. Fokas, E., Engenhart-Cabillic, R., tumor. Proc. Natl Acad. Sci. USA 79, 55475551 metastases per patient. Int. J.Cancer 125,
Daniilidis,K.,Rose, F. & An, H.X. Metastasis: (1982). 474482 (2009).
the seed and soil theory gains identity. Cancer 33. Cillo, C., Dick, J.E., Ling, V. & Hill, R.P. 38. Jones, J. etal. Gene signatures of progression
Metastasis Rev. 26, 705715 (2007). Generation of drug-resistant variants in and metastasis in renal cell cancer. Clin.
27. Uhr, J.W., Scheuermann, R.H, Street, N.E. & metastatic B16 mouse melanoma cell lines. Cancer Res. 11, 57305739 (2005).
Vitetta, E.S. Cancer dormancy: opportunities Cancer Res. 47, 26042608 (1987). 39. Campbell, P.J. etal. The patterns and dynamics
for new therapeutic approaches. Nat. Med. 3, 34. Khodarev, N.N. etal. STAT1 pathway mediates of genomic instability in metastatic pancreatic
505509 (1997). amplification of metastatic potential and cancer. Nature 467, 11091113 (2010).
28. Goss, P.E. & Chambers, A.F. Does tumour resistance to therapy. PLoS ONE 4, e5821 40. Yachida, S. etal. Distant metastasis occurs
dormancy offer a therapeutic target? Nat. Rev. (2009). late during the genetic evolution of pancreatic
Cancer 10, 871877 (2010). 35. Khodarev, N.N. et al. STAT1 is overexpressed cancer. Nature 467, 11141117 (2010).
29. Dunn, G.P., Koebel, C.M. & Schreiber, R.D. in tumors selected for radioresistance and
Acknowledgments
Interferons, immunity and cancer immunoediting. confers protection from radiation in transduced
The authors are supported by grants from the Ludwig
Nat. Rev. Immunol. 6, 836848 (2006). sensitive cells. Proc. Natl Acad. Sci. USA 101,
Foundation for Cancer Research, the Lung Cancer
30. Fidler, I.J. & Kripke, M.L. Metastasis results 17141719 (2004).
Research Foundation, and a generous gift from the
from preexisting variant cells within a malignant 36. Weichselbaum, R.R. etal. An interferon-
Foglia Foundation.
tumor. Science 197, 893895 (1977). related gene signature for DNA damage
31. Chambers, A.F, Hill, R.P. & Ling, V. Tumor resistance is a predictive marker for Author contributions
heterogeneity and stability of the metastatic chemotherapy and radiation for breast cancer. Both authors contributed to researching data for the
phenotype of mouse KHT sarcoma cells. Proc. Natl Acad. Sci. USA 105, 1849018495 article, discussion of the content, and writing and
Cancer Res. 41, 13681372 (1981). (2008). editing the manuscript.

382 | JUNE 2011 | VOLUME 8  www.nature.com/nrclinonc


2011 Macmillan Publishers Limited. All rights reserved