Seminars in Cancer Biology 14 (2004) 399–405

Multistep and multifactorial carcinogenesis: when doesa contributing factor become a carcinogen?

Michele Carbonea,∗, Harvey I. Passba Department of Pathology, Cardinal Bernardin Cancer Center, Loyola University Chicago, Room 250,

2160 South First Ave, Maywood, IL 60302, USAb Karmanos Cancer Center, Detroit, Michigan, USA

Abstract

Our greatest successes in fighting cancer derive from the identification and removal or inactivation of carcinogenic substances, and fromthe identification and removal of pre-malignant lesions. In comparison, our successes at treating already formed malignancies have beenminimal. Therefore, emphasis should be put in identifying and removing pre-malignant lesions, and in the identification and removal ofthose agents that cause or contribute to cancer development. It is important to target initiators, co-carcinogens and promoters, since byremoving any one of them, tumor growth may be prevented. Identification of these agents is difficult. Epidemiological studies largelystudy cancer after it has occurred. It would be preferable to identify potential carcinogenic substances at an earlier stage before they havecaused a large number of malignancies and thus become identifiable by epidemiology. During the past three decades, we have accumulatedan impressive amount of evidence concerning molecular pathways that when altered contribute to malignant growth. It is time that westart applying this knowledge to the identification of human carcinogens. Here, we review the molecular changes that are required forcarcinogenesis and propose some criteria that, in the absence of epidemiological evidence, can be used to identify agents that causeor contribute to human cancer development. In the absence of epidemiological evidence, a given agent should be considered a humancarcinogen when: (1) the agent causes or contributes to the development of tumors in animals that are of the same type as those tumorsassociated with exposure to the agent in humans; (2) the agent transforms or contributes to the transformation of human cells in culture andthese cells are of the same type from which associated human malignancies arise; (3) there is molecular evidence that the agent interfereswith one or more key molecular pathways in human cells which leads to the formation of human cancer.© 2004 Published by Elsevier Ltd.

Keywords:Multifactorial carcinogenesis; Cancer; Etiology; Cancer prevention

1. Introduction

The fight against cancer includes different strategies: ba-sic research to understand the molecular and biochemicalpathogenesis of malignancy, strategies to identify and elim-inate carcinogenic agents, strategies to identify and removepre-malignant lesions or malignancies in their early stages,and attempts at developing new drugs and therapeutic ap-proaches to cure cancer patients. Among these options, dur-ing the past 40 years, the vast majority of cancer researchefforts and resources have been devoted to: (1) finding newtherapeutic approaches for different cancer types and (2)to studying the basic molecular and biochemical signalingpathways of normal and malignant cells. The argument be-hind these two approaches are: (1) that clinical research canidentify useful drugs/treatments for cancer patients and (2)

∗ Corresponding author. Tel.:+1 708 327 3250; fax:+1 708 327 3238.E-mail address:mcarbon@lumc.edu (M. Carbone).

that basic knowledge developed in human and animal cellsin tissue culture or even in yeast and flies could allow usto devise effective treatment strategies for cancer patients.Both camps can point to some successes. For example, thetreatment of childhood leukemias and a few other pediatricmalignancies are considered some of the most obvioussuccesses of chemotherapy. Furthermore, the developmentof anti-c-kit strategies in the treatment of gastrointenstinalstromal tumors (GISTs) involved the successful translationof basic biochemical findings to clinical treatments. Unfor-tunately, despite these successes, most aggressive malignan-cies remain largely resistant to therapy. In the past 40 years,there have been great advances in our understanding of can-cer at the cellular and molecular level. However, this infor-mation has been difficult to translate to the bedside, and therehas been little improvement in our ability to treat advancedsolid tumors, i.e., carcinomas, the most common types ofhuman cancers. The prognosis of metastatic carcinomas ofthe lung, larynx, breast, prostate, pancreas, liver, etc., has

1044-579X/$ – see front matter © 2004 Published by Elsevier Ltd.doi:10.1016/j.semcancer.2004.06.002

400 M. Carbone, H.I. Pass / Seminars in Cancer Biology 14 (2004) 399–405

not significantly changed during the past 40 years. This ispartly due to the fact that advanced solid tumors are geneti-cally heterogeneous both among cases and within the samepatient. They are also genetically unstable (i.e., they con-tinuously develop new genetic clones). Therefore, any newtherapeutic agent identified is aiming at a “moving target”which can readily adapt to most forms of therapeutic attack.

In summary, progress has not been as swift as many of ushad hoped.

2. Cancer prevention

The available evidence indicates that the most effectiveway to fight cancer has been to prevent cancer from devel-oping in the first place, either by removing pre-malignantlesions, or by removing the causes of cancer. In fact, themost important successes of early detection have not beenin detecting early cancers, but in detecting and removingpre-malignant lesions. The implementation of the Pap-test,developed by Dr. Papanicolau in 1928, in the industrializedworld, has been, so far, the single most effective preventive/therapeutic approach against cancer[1]. The development ofcervical carcinoma in millions of women has been prevented– thanks to the Pap-test. The Pap test recognizes cervicaldysplasia as a precursor of carcinoma. Dysplastic lesionscan be removed before they become invasive carcinomas.Similarly, colonoscopy identifies and removes adenomas,which are pre-malignant lesions that may, over many years,develop into an invasive colon carcinoma. In summary, theideal intervention is to prevent cancer development.

To successfully prevent cancer, it may not be necessaryto discover everything about the disease process. However,it is necessary to identify some point in the evolution of theprocess that is vulnerable to intervention. A rapid and accu-rate identification of human carcinogens – the factors thatcause or contribute to cancer development – and their re-moval (anilin and bladder carcinomas) or inactivation (sun-screen/clothing and skin cancer) appear the most effectiveand desirable way to have a rapid positive impact in the fightagainst cancer. What is a human carcinogen? The answer ap-pears obvious – something that causes cancer – but it is not.

2.1. Identification of human carcinogens: relevance ofco-factors

The identification of human carcinogens relies on epi-demiology, animal and tissue culture studies, and on molecu-lar pathology, by which we mean the combination of molec-ular genetics, molecular biology, genetics, microbiology, vi-rology, immunology, and biochemistry. This is not necessar-ily easier than identifying diagnostic markers or therapeutictargets, but promises a higher reward.

In causation, we identify anecessarycause, which is afactor that must be present for disease to occur, but may notbe sufficient on its own (influenza virus and influenza, for

example). In addition, we identify asufficientcause, whichis a factor that may be present for disease to occur and thatwill cause disease on its own (phalloidin and hepatocellu-lar necrosis, for example). The cause–effect relationship thatrepresents the basis of the pathological investigation is noteasy to apply to the process of human carcinogenesis. Mostof the population is exposed to a variety of human carcino-gens, yet only a small fraction of exposed individuals ac-tually develop cancer. Among the factors that contribute totumor development, those that cause the molecular damagethat leads to cancer have been called carcinogens or ini-tiators. Factors that usually are not carcinogenic per se butthat enhance the activity of carcinogens have been calledco-carcinogens. Co-carcinogens can also become carcino-genic at high doses or in the presence of other co-carcinogensor predisposing factors (genetic susceptibility). Factors thatare not carcinogenic per se but that enhance the activity ofa carcinogen have been called tumor promoters.

The difficulty in proving a cause–effect relationship inhuman carcinogenesis is the result of many factors: thecomplex and mostly unknown interactions among differentcarcinogens, co-carcinogens and promoters, the length andamount of exposure, the time in life in which exposure oc-curs, the role of the immune system in preventing or at timespromoting tumor cell growth, different genetic backgrounds,and the long latency from exposure to cancer development[2]. It can be stated that carcinogenesis is a multi-stage pro-cess where the final risk of cancer development is a func-tion of the combined probabilities of relatively rare eventsoccurring in each stage[3].

This complexity, however, also provides multiple points ofintervention. Therefore, the identification of co-carcinogensand promoters can be as relevant as the identification ofinitiators, if by removing or interfering with their activity,tumor incidence can be decreased. Presently, the process toidentify human carcinogens is built around measuring theaverage effects of single agents[4]. The process is not setup to identify complex interactions among different factorsin the general population, much less interactions that takeplace only in a fraction of tumor patients.

A good example of the relevance of identifying and re-moving human carcinogens and of the relevance of the inter-actions among different carcinogens is hepatocellular carci-noma (HCC). There are approximately 1 million cases/yearworld wide and major geographical differences in incidenceof this disease[5]. In southeast Asia, HCC causes about 70%of all cancer deaths in males and 45% in females; in the US,about 2% of cancer deaths are attributable to HCC[5,6].Treatment options for HCC are minimal. Liver transplant –the best therapy available – is associated with 4.5 years me-dian survival for single lesion less than 5 cm in diameter, butonly with 2 years average survival for larger lesions or formultiple lesions[5]. Considering that early-stage patientsrepresent the exception, and that liver transplant is avail-able only to patients in the Western world – where the inci-dence of HCC is low – the overall impact of this therapeutic

M. Carbone, H.I. Pass / Seminars in Cancer Biology 14 (2004) 399–405 401

approach on the 1 million HCC patients per year is mini-mal. Partial hepatectomy is associated with 30% survival at5 years, but this procedure has a high risk of operative mor-tality (10%) from bleeding in patients with cirrhosis – themajority of patients with HCC. For these reasons, most HCCpatients receive no therapy at all, or they receive chemother-apy, which does not significantly influence the median sur-vival of about 4 months from diagnosis[5]. In short, HCCpatients have a dismal prognosis and the therapeutic op-tions available are mostly ineffective[5]. Consider, instead,the effect of HBV vaccination on HCC. Since the start ofmassive HBV vaccination movements in Taiwan, the carrierstate in children decreased from 10% to less than 1% in 10years. The incidence of HCC in children 6–14 declined from0.7/100,000 in 1981–1986 to 0.36 in 1990–1994, and morein recent years. HBV vaccination to prevent HCC representsone of the greatest successes of basic research, epidemiol-ogy, viral immunology, and ultimately of cancer prevention.

Although HBV is the main cause of HCC in Taiwan andChina, HBV often interacts with other carcinogens to causeHCC: aflatoxin B1 in China and alcohol in the US and Eu-rope. Aflatoxin B1 and alcohol are much less potent car-cinogens compared to HBV and this delayed their recog-nition as causative factors of HCC[6]. However, becausethey synergize with HBV in causing HCC, their removalhas a tremendous potential for reducing HCC incidence. Ithas been shown that the relative risk for developing HCCin Shangai was 7.3 for HBV carriers, 3.4 for individualsexposed to aflatoxin B1, and 59.4 for individuals infectedwith HBV and exposed to aflatoxin B1[6]! HBV vaccina-tion and the prevention of food crop contamination by afla-toxin B1 have had a tremendous impact in decreasing HCCin endemic areas[5,6].

3. Genetic susceptibility to carcinogens

Genetic differences play an important role in determin-ing individual susceptibility to carcinogens. This must beconsidered when determining the carcinogenicity of a givenagent. While the “average” effect of a given agent on thewhole population may not allow a statistically detectablecarcinogenic effect, the same agent may be extremely car-cinogenic in genetically susceptible individuals. Erionite, atype of zeolite, is a good example of this. Erionite depositsare present in Europe, Nevada, California, Turkey, etc. Noincreased incidence of mesothelioma has been reportednearby these deposits, except for three small villages inCappadocia, Turkey: Old Sarihidir, Karain, and Tuzkoy.In these three small villages (the total population is ap-proximately 2000 people), 50% of all deaths are causedby mesothelioma. This epidemic has been attributed to eri-onite exposure[7]. However, this same epidemic was notobserved in nearby villages or in other areas with similarerionite deposits. Pedigree analyses in the affected villagesrevealed that the 50% of deaths attributed to mesothe-

lioma was an average that did not accurately portray thetrue distribution of the disease. We found families inwhich almost all family members died of mesotheliomaand families in which mesothelioma was absent or veryrare[8,9]. The different incidence of mesothelioma amongfamilies was caused by genetic predisposition to erionitecarcinogenesis and mesothelioma. When mesotheliomafamily members married into non-mesothelioma families,mesothelioma appeared in the subsequent generations,underscoring the role of genetics. When mesotheliomafamilies moved to erionite-free areas, mesothelioma in-cidence decreased or possibly disappeared, underscoringthe pathogenic role of erionite. Thus, the mesotheliomaepidemic in Cappadocia is caused by the combination oferionite exposure and genetic predisposition to erionitecarcinogenesis[9].

4. Experimental identification of carcinogens

The identification of human carcinogens, i.e., establish-ing whether a given factor causes or contributes to cancerdevelopment, will be heavily influenced by the species inwhich oncogenicity is studied, differences in susceptiblecell types and target organs, and the presence of co-factors(co-carcinogens, promoters, genetic predisposition). Simianvirus 40 (SV40), a DNA tumor virus, provides an ex-cellent model to outline how important these parametersare when assessing oncogenicity. SV40 is highly onco-genic in newborn and weanling hamsters (because theirimmune system is not fully developed), mastomys, andimmunocompromised mice[10]. SV40 is not oncogenic innon-immunocompromised mice and rarely causes tumorsin adult hamsters. Clearly, the species in which the virus istested and the immune system influence the outcome of theoncogenicity test. Had SV40 initially been tested in adultimmunocompetent mice rather than in hamsters, it mayhave been considered innocuous as were the previous 39simian viruses isolated from monkey kidney cells. Becauseof its marked oncogenicity in hamsters, SV40 was care-fully studied, and these investigations revealed that SV40produced an oncoprotein, the large T antigen (Tag), whichremains the most potent oncogene discovered[10]. WhenSV40 was isolated, scientists discovered that hamster cellswere non-permissive to SV40 replication. Occasionally,SV40 became integrated into the hamster genome and itsoncoproteins, Tag and the small t antigen (tag) were ex-pressed and caused malignant transformation of the cellsin tissue culture (Fig. 1). Instead, SV40 replicated veryefficiently in human fibroblasts and in monkey cells, andas a consequence, these cells were lysed. Because the cellswere lysed, malignant transformation did not occur exceptin the very rare situation in which cells were infected atlow multiplicity of infection and the viral genome becameintegrated into the human cell genome. Integration usuallycauses disruption of some viral sequences and therefore

402 M. Carbone, H.I. Pass / Seminars in Cancer Biology 14 (2004) 399–405

Fig. 1. Susceptibility of different cells to SV40 infection and transformation. Human mesothelial cells (left) are very susceptible to SV40 infection andmalignant transformation because the high levels of endogenous wild-type p53 bind SV40 Tag and inhibits Tag ability to promote SV40 replication. Lessviral particles are formed and cell lysis is infrequent. SV40 remains episomal in mesothelial cells, Tag is expressed and induces malignat trasnformation ata very high rate 1/103 cells [11]. In contrast, human fibroblasts and monkey cells are susceptible (permissive) to SV40 infection and replication (center),and as a result, these cells are lysed. Transformation is a very rare event that occurs in less than 1/107 cells when cells are infected at low multiplicity ofinfection (one virus per cell) and the virus becomes accidentally integrated into the cellular genome. Integration usually disrupts some portions of the viralgenome and viral replication does not occur, while Tag is expressed and can cause malignant transformation. Hamster and rodent cells are non-permissiveto viral replication, thus these cells are not lysed and can be occasionally transformed if the viral genome becomes integrated in the host cell genome.

complete viral particles cannot be formed. However, if theviral oncogenes, Tag and tag, are expressed, the cell ismalignantly transformed. It was concluded that the chancethat SV40 could cause malignant transformation of humancells was very small and that it always required viral inte-gration into the human cell genome. The cell type studied– fibroblasts- biased this erroneous conclusion. When weinfected human mesothelial cells (HM), an undifferentiatedcell type which covers the pleural and peritoneal cavities,SV40 did not cause cell lysis. Instead, the high levels ofwild-type p53 normally present in HM-bound Tag and in-hibited Tag-driven SV40 replication[11]. Therefore, fewerSV40 particles were produced in infected HM compared tofibroblasts, and cell lysis was infrequent. Because the ma-jority of HM survive SV40 infection and SV40 oncogenesare expressed, the rate of transformation is high (Fig. 1). Insummary, SV40 oncogenicity is influenced by the speciesin which oncogenicity is tested, by the immune system, bythe specific cell type investigated.

Co-carcinogens also influence SV40 oncogenicity. SV40has been associated mostly with human mesotheliomas,tumors that derive from the mesothelial cells[7]. Thesetumors are associated with asbestos exposure; however,

only a fraction of individuals exposed to asbestos developmalignancy[7]. This indicates that co-factors influence as-bestos carcinogenesis. We found that SV40 and asbestosare co-carcinogens in vitro and cause malignant transfor-mation of human cells[11]. We also recently confirmedco-carcinogenicity in hamsters (Carbone et al., unpub-lished observations). In conclusion, carcinogenesis is of-ten species- and cell-type-specific, and it is influenced byco-factors (co-carcinogens, promoters, genetic susceptibil-ity). Proper assessment of carcinogenicity requires con-sideration of these different variables. The same factor orinfectious agent can be highly oncogenic in certain celltypes or individuals and not oncogenic in others. An obvi-ous conclusion from this statement is that the presence of avirus or other microbial agent in a given tumor type doesnot establish causation because only specific cell types aresusceptible to malignant transformation. Moreover, somecell types may become susceptible only in the presenceof certain co-factors, genetic predisposition, or immunedepression.

In conclusion, we think that most agents that havebeen identified as “human carcinogens” because they cancause cancer in the population at large have already been

M. Carbone, H.I. Pass / Seminars in Cancer Biology 14 (2004) 399–405 403

Fig. 2. Primary human mesothelial cells before (left) and after infectionwith SV40 (right, note that this cell has 85 chromosomes and that severalshow obvious alterations, arrowheads). Among the vast amount of geneticdamage caused by SV40 upon infection of human mesothelial cells, thereare some specific alterations, such as inactivation of the tumor suppressorsRb and p53, that play a key role in human msothelial cell transformation(see text).

discovered. The future challenge is to identify those agentsthat cause cancer predominantly in certain individuals be-cause of genetic predisposition and/or because of exposureto additional co-carcinogens and promoters. These inter-actions will determine those who will develop malignancyamong exposed individuals.

Fig. 3. Outline of the basic cellular mechanisms that must be altered for malignant transformation (left). The growth of transformed human cells willbe influenced by interactions with the stroma (not shown) and the immune system (right). This interaction can either eliminate tumor cells that expressunique antigens (especially cells transformed by viruses, such as EBV) or promote tumor cell growth. Immune cells release cytokines that can bothstimulate cell growth and cause additional genetic damage that can by itself drive the malignant growth (such as hepatocellular carcinoma in patientsinfected with hepatitis C).

5. Markers of carcinogenesis

Cancer is a multifunctional and dynamic event, not asimple molecular event. Cancer growth requires molecularchanges that affect the tumor cell, alterations in the inter-action among tumor cells and the nearby stroma, and theimmune system that in turn lead to tumor growth, invasion,and metastasis. Thus, cancer is a complex dynamic mixtureof malignant and benign stromal cells growing out of con-trol and invading surrounding tissues. Cancer cannot be de-fined by a single molecular change, although occasionally,a specific molecular change is diagnostic of certain tumors.Cancer cells are characterized by a large number of geneticmutations and aneuploidy, which at first glance, make it verydifficult to identify the key molecular genetic changes thatcaused malignancy. However, among the large amount ofgenetic damage found in cancer cells (Fig. 2), there are somegenetic changes that play a particularly important role in can-cer development. Hahn and Weinberg[12] found that the dis-ruption of the intracellular pathways regulated by the SV40Tag and tag, oncogenic ras, and telomerase activity sufficeto create a human tumor cell. Subsequently, Notch-1 induc-tion was found as an additional requirement of human celltransformation[13]. Tag and tag were found to be requiredfor malignant transformation of human cells because of their

404 M. Carbone, H.I. Pass / Seminars in Cancer Biology 14 (2004) 399–405

multiple biochemical activities. Tag binds and inactivatesthe tumor suppressors Rb and p53, induces met, IGF-1, andIGF1-R (receptor), and induces Notch-1. These activities to-gether stimulate cell growth[14,15]. Because cells are stim-ulated to divide, and DNA repair and apoptosis are inefficientbecause of p53 inactivation, SV40-infected human cells con-tinue to divide and accumulate large numbers of genetic al-terations (Fig. 2). The tag inactivates protein phosphatase 2A(PP2A) and thus alters the phosphorylation state and activityof numerous cellular substrates[16]. Among these substratesare MAP, ERK, and IKK kinases that are normally downreg-ulated by PP2A. Expression of tag in human cells leads to ac-tivation of these kinase pathways and ultimately to AP-1 andNF�B activity, which, depending on the cell type, can causeor protect cells from apoptosis, induce cell proliferation, andinduce transcriptional activation of different cellular genes.In addition to these two very potent viral oncogenes, humancell transformation required ras activation, a potent growthand survival signaling pathway[12]. Finally, telomerase ac-tivity was required for immortalization and continuous tu-mor cell growth[17]. These are the necessary requirementsto produce a cancer cells, but these requirements are notsufficient for malignant growth. Malignant growth requiresthat the transformed cells develop the ability to invade sur-rounding tissues and (often but not always) to metastasize. Inaddition, the malignant cells must be able to promote angio-genesis to sustain their own malignant growth. Only when allthese requirements are fulfilled is malignant growth possible(Fig. 3). The immune system represents an additional impor-tant variable to this process. Sometimes, the immune systemhas a protective effect, for instance, in those malignanciesthat express viral antigens. At other times, the immune sys-tem can promote tumor growth – even virally induced tu-mors – by releasing growth stimulatory factors and reactiveoxygen radicals that cause DNA alterations[18].

6. Conclusions

How can we translate the apparent complexity outlinedabove into the identification of human carcinogens? Whendoes a contributing factor become a human carcinogen? Ofcourse, epidemiological evidence is key, but this is rarelyavailable in a timely fashion. Ideally, we want to identify andpossibly remove carcinogens before they have caused a suffi-cient number of cancer deaths to become statistically signifi-cant. During the past three decades, we have accumulated animpressive amount of evidence concerning molecular path-ways that when altered contribute to malignant growth. It istime that we start applying this knowledge to the identifica-tion of human carcinogens. We propose that in the absenceof epidemiological evidence, a given agent should be con-sidered a human carcinogen when: (1) the agent causes orcontributes to the development of tumors in animals that areof the same type as those tumors associated with exposureto the agent in humans; (2) the agent transforms or con-

tributes to the transformation of human cells in culture andthese cells are of the same type from which associated hu-man malignancies arise; (3) there is molecular evidence thatthe agent interferes with one or more key molecular path-ways in human cells which leads to the formation of humancancer (Fig. 3). Among these molecular alterations, someare of particular relevance. Inactivation of Rb and p53, andactivation of the NF�B and AP-1 pathways which in turnlead to a cascade of downstream effects which influence celldivision and apoptosis, together with immortalization (i.e.,telomerase activity or alternative mechanisms), are the mini-mal and most crucial requirements for tumor cell formation.When these three requirements are fulfilled, a given agentshould be considered a carcinogen, or at least a possibleco-factor in human carcinogenesis, even before the epidemi-ological evidence becomes available. The advantage of thisapproach is that it would allow removal of a suspected car-cinogen in a shorter time. This would also have long-termbenefit, because many carcinogens “accumulate’ in our soci-ety, and once that has happened, their removal is very prob-lematic, as in the case of asbestos[19]. The disadvantage ofsuch approach is that it may erroneously identify an other-wise innocuous agent as a human carcinogen. The risk ap-pears quite small, since very few – if any – agents will fulfillpoints 1–3 and still be proven harmless in humans. Often,saccharin for bladder carcinoma and herpes virus for cervi-cal carcinoma are given as examples of the premature rushto identify human carcinogens. Saccharin caused bladdercarcinomas in rats when given at very high doses, doses un-realistic to be achieved in humans. Thus, one could add theprecaution that exposure as measured in points 1–3 shouldbe at doses somewhat comparable to those expected in hu-mans. In any event, saccharin does not fulfill points 2 and3 indicated above: does not transform human cells in cul-ture and does not alter any key cellular pathway. Moreover,saacharin caused cancer in rats because these animals havea unique metabolic pathway for this substance[2], and doesnot cause cancer in other animals. This underscores the rel-evance of the species in which oncogeneicity is tested. Her-pes virus infection was originally considered the cause ofcervical carcinoma. Subsequent analyses revealed that thenecessary cause is HPV and that herpes virus infection is“just’ a co-factor that increases HPV carcinogenic risk. Toreduce cancer risk, it is important to remove or inactivate(through a vaccine, for example) carcinogens (initiators),co-carcinogens, and promoters. Thus, the initial misclassi-fication of the Herpes virus as a carcinogen, although veryimportant, does not influence the requirement of prevent-ing Herpes virus infection to reduce cervical carcinomarisk.

In conclusion, 40 years of basic research have providedus with tools that allow us to better assess the carcinogenicrisk of a given agent. It is hoped that careful analysis of themolecular information, together with epidemiology, animal,and tissue culture experiments, will lead to a more rapididentification of human carcinogens.

M. Carbone, H.I. Pass / Seminars in Cancer Biology 14 (2004) 399–405 405

Acknowledgements

We are grateful to Drs. Lucio Miele and Amy Powersfor critical reading of this manuscript. Dr. Carbone’s workis supported by grants from the National Cancer Insti-tute, USA, RO1CA92657; the American Cancer Society,RSG-04-029-01-CCE; and by the Cancer Research Foun-dation of America.

References

[1] De May R. The Pap smear. In: De May R, editor. The art and scienceof cytopathology. Chicago, USA: ASCP Press; 1996.

[2] Carbone M, Klein G, Gruber J, Wong M. Modern criteria to identifyhuman carcinogens. Cancer Rev, in press.

[3] Franco EL, Correa P, Santella RM, Wu X, Goodman SN, PetersenGM. Role and limitations of epidemiology in establishing a causalassociation. Semin Cancer Biol, in press.

[4] Cogliano VJ. Current criteria to establish human carcinogens. SeminCancer Biol, in press.

[5] Fong Y, Kemeny N, Lawrence TS. Cancer of the liver and bil-iary tract. In: De Vita V, Hellman S, Rosenberg SA, editors. Can-cer principles and practice of oncology, sixth ed. 2001, p. 1162–77.

[6] Wogan GN, Hecht SS, Felton JS, Conney AH, Loeb LA. Environ-mental and chemical carcinogenesis. Semin Cancer Biol, in press.

[7] Carbone M, Kratzke RA, Testa JR. The pathogenesis of mesothe-lioma. Semin Oncol 2002;29(1):2–17.

[8] Roushdy-Hammady I, Siegel J, Emri S, Testa JR, Carbone M.Genetic-susceptibility factor and malignant mesothelioma in the Cap-padocian region of Turkey. Lancet 2001;357(9254):444–5.

[9] Carbone M, Baris IY. Genetics and human mesothelioma. In: CarboneM, Pass HI, Vogelzang NJ, editors. Malignant mesothelioma. NewYork: Springer–Verlag, in press.

[10] Carbone M, Pass HI, Miele L, Bocchetta M. New developmentsabout the association of SV40 with human mesothelioma. Oncogene2003;22:5173–80.

[11] Bocchetta M, Di Resta I, Powers A, Fresco R, Tosolini A, Testa JR, etal. Human mesothelial cells are unusually susceptible to simian virus40-mediated transformation and asbestos cocarcinogenicity. Proc NatlAcad Sci USA 2000;97:10214–9.

[12] Hahn C, Weinberg RA. Rules for making human tumor cells. NewEngl J Med 2002;347:1593–21603.

[13] Wijzen S, Rizzo P, Braid M, Vaiishnav R, Jonkheer S, Zlobin A, etal. Nat Med 2002;8:979–86.

[14] Carbone M, Rizzo P, Grimley PM, Procopio A, Mew DJ, Shrid-har V, et al. Simian virus-40 large-T antigen binds p53 in humanmesotheliomas. Nat Med 1997;3(8):908–12.

[15] Gazdar AF, Butel JS, Carbone M. SV40 and human tumours: myth,association or causality? Nat Rev Cancer 2002;2(12):957–64.

[16] Rundell K, Parakati R. The role of the SV40 ST antigen in cell growthpromotion and transformation. Semin Cancer Biol 2001;11:5–13.

[17] Foddis R, De Rienzo A, Broccoli D, Bocchetta M, Stekala E, Rizzo P,et al. SV40 infection induces telomerase activity in human mesothe-lial cells. Oncogene 2002;21:1434–42.

[18] Philip M, Rowley DA, Schreiber H. Inflammation as tumor promotein cancer induction. Semin Cancer Biol, in press.

[19] Mossman BT, Klein G, zur Hausen H. Modern criteria to determinethe etiology of human carcinogens. Semin Cancer Biol, in press.