Infections as a major preventable cause of human cancer


  • H. Kuper,

    1. From the Department of Epidemiology, Harvard School of Public Health, Boston, MA, USA;
    Search for more papers by this author
  • H.-O. Adami,

    1. From the Department of Epidemiology, Harvard School of Public Health, Boston, MA, USA;
    2. Department of Medical Epidemiology, Karolinska Institute, Stockholm, Sweden; and
    Search for more papers by this author
  • D. Trichopoulos

    1. From the Department of Epidemiology, Harvard School of Public Health, Boston, MA, USA;
    2. Department of Medical Epidemiology, Karolinska Institute, Stockholm, Sweden; and
    3. Department of Hygiene and Epidemiology, University of Athens Medical School, Athens, Greece
    Search for more papers by this author

Hans-Olov Adami, Department of Medical Epidemiology, Karolinska Institute, PO Box 281, 171 77 Stockholm, Sweden (fax: +46 8 314 957; e-mail:


Abstract. Kuper H, Adami H-O & Trichopoulos D (Harvard School of Public Health, Boston, MA, USA; Karolinska Institute, Stockholm, Sweden; University of Athens Medical School, Greece). Infections as a major preventable cause of human cancer (Internal Medicine in the 21st Century). J Intern Med 2000; 248: 171–183.

Infections may be responsible for over 15% of all malignancies worldwide. Important mechanisms by which infectious agents may induce carcinogenesis include the production of chronic inflammation, the transformation of cells by insertion of oncogenes and inhibition of tumour suppressors, and the induction of immunosuppression. Common characteristics shared by infectious agents linked to malignancies are that they are persistent in the host, often highly prevalent in the host population and induce cancer after a long latency. The associations between a selection of infectious agents and malignancies are covered in detail.


Following tobacco use, infections as a group may be the most important preventable cause of cancer in humans. Interest in the infectious origins of cancer has ebbed and flowed, and this has been elegantly reviewed by Parsonnet [1]. Cancer clusters in families were observed early. On occasion cancer appeared to be transmitted between individuals, and several crude experiments were conducted to see if this could be replicated in animals. Early researchers were not always entirely off the mark. For example, it had been noted already in the nineteenth century that cervical cancer patterns corresponded to those of sexually transmitted diseases. However, despite sporadic early interest, research into the infectious origins of cancer failed to advance significantly until the 1960s.

In the last 50 years there has been explosive growth in the fields of epidemiology and infectious disease biology; epidemiological and serological methods and biological knowledge have rapidly developed, creating a climate ripe for investigating the infectious origins of cancer. Part of the reason for this interest is the realization that an infectious origin of a cancer indicates that a cancer is preventable. Furthermore, studying viral carcinogenesis has given us insights into carcinogenesis in general, or, put more poetically:

…viruses have turned out to be the Rosetta stone for unlocking the mysteries of cell growth control [2].

Historic context

There were many false leads in the initial exploration of the infectious origins of cancer. Prevalent beliefs that ultimately turned out to be wrong, such as the notion that yeast was the cause of cancer or the idea that a cause of cancer must be smaller than the cell, muddied the water. However, during the 1900s several remarkable discoveries were made in this field, ultimately earning five Nobel prizes.

Rous is credited as being the first person to show that cancer can have an infectious origin, when he extracted cell-free filtrate from chicken sarcoma to cause cancer in a second animal in 1911. The mood was not ripe for his work to be generally accepted, so it was not until 1966 that he was awarded a Nobel prize for his discovery of the RNA virus that was eventually called the Rous sarcoma virus. Meanwhile the first Nobel Prize in this field was given to Fibiger in 1926 for his claim that a parasite, Spiroptera carcinoma, was causing stomach cancer in rats. Within a few years of Fibiger’s death in 1928, however, it became apparent that the rats were developing cancer because of the absence of nutritious food, not the presence of a parasite, in their stomachs. The 1930s saw the discovery of two important mammalian oncoviruses: the Shope rabbit papillomavirus and Bittner’s milk-transmitted mouse mammary tumour virus, both initially ignored on account of the unfavourable scientific milieu.

With the 1950s came growing interest in infectious diseases as antibiotics and anti-TB drugs became widely used and the structure of DNA was elucidated. Furthermore, the electron microscope was developed allowing the visualization of viruses, which were clearly defined for the first time in 1957. With this activity, the tide began to turn for interest in infectious origins of cancer, so that between 1951 and 1972 26 mammalian oncoviruses were discovered and Rous was finally recognized. Two of these oncoviruses affect humans; Hepatitis B virus (HBV) described by Blumberg in 1963 (for which he won the Noble prize in 1976) and the Epstein–Barr virus (EBV), extracted by Epstein and his colleagues in 1964 from cells of Burkitt’s lymphoma cell culture.

In response to these discoveries, the US Virus Cancer Program was initiated in the late 1960s, marking the start of the era during which the infectious origins of cancer became a dominant hypothesis. Advances in molecular biology helped to further clarify the relationship between infectious agents and carcinogenesis, and, in fact, have given valuable insights into the molecular basis of carcinogenesis. In 1975 David Baltimore and colleagues won the Nobel Prize for their discoveries concerning the interaction between tumour viruses and the genetic material of the cell, and in 1989 Bishop and Varmus received the Nobel prize for recognizing the relationship between a retroviral oncogene and a normal human gene, src.

Recent estimates suggest that upwards of 15% of malignancies worldwide can be attributed to infections, and that although this figure is only 7% in the developed world, it exceeds 22% in the developing nations, a global total of 1.2 million cases per year [3]. The development of more sensitive serological methods suggest that an even larger proportion of cancers may have an infectious aetiology.


There are three main mechanisms by which infections can cause cancer, and they appear to involve initiation as well as promotion of carcinogenesis.

First, an infectious agent may become persistent within the host and thus induce chronic inflammation [4]. This is often accompanied by the formation of reactive oxygen and nitrogen species (ROS and RNOS, respectively) by phagocytes at the site of inflammation. ROS and RNOS have the potential to damage DNA, proteins and cell membranes, and modulate enzyme activities and gene expression, and so favour carcinogenesis [4]. Furthermore, chronic inflammation often results in repeated cycles of cell damage, and compensating cell proliferation [5]. This process increases the number of cells that are dividing, and therefore subject to DNA damage and promotes the growth of malignant cells [5].

Secondly, infectious agents may directly transform cells, by inserting active oncogenes into the host genome, inhibiting tumour suppressors or stimulating mitosis.

Thirdly, infectious agents, such as human immunodeficiency virus (HIV), may induce immunosuppression, with consequent reduced immunosurveillance [6]. The course of cancer in immunocompromised hosts is generally very aggressive, even though risk factors remain unchanged [6].

Common features of oncogenic infectious agents

Infectious agents, particularly viruses, suspected of being oncogenic are often highly prevalent within the host population [7]. In contrast, virus-associated malignancies are rare amongst the infected people and usually occur after a prolonged latency. Infectious agents are therefore not sufficient for cancer causation; cofactors are vital and conditions at initial infection may determine pattern of the host–agent interaction. Oncogenic viruses frequently have the capacity to persist, so that they infect precursors of the malignant cell, and continue to express viral genes in tumour cells. Parasites that have oncogenic potential also persist in the host, and removal of the parasite may reverse tumour development. For neoplasia to develop, the viral infection must affect the tissue’s pluripotent cells, since differentiated cells cannot be immortalized. Tumours that develop as a result of infectious agents are almost always monoclonal, indicating that they have their origin in a single malignant cell.

Finding an infectious cause of cancer is not an easy matter. The high prevalence of the infectious agent in the general population, the extended latency, as well as the importance of interacting factors, make detection of relationships difficult. Associations may also arise spuriously, because the virus is activated as a consequence, rather than a cause, of carcinogenesis, or because the virus acts as a marker for the causal agent. Furthermore, associations may be missed if an unsuitable serological marker is used or if the virus is excised from the genome or the parasite eliminated during carcinogenesis. Appropriate animal models are often lacking, hampering research into the mechanistic effects of infection.

Finding an infectious agent in tumour tissue does not document that it was causally involved in carcinogenesis. This requires epidemiological, microbiological and molecular biology techniques. To establish an infectious agent as a definite carcinogen a number of conditions must be fulfilled [7 8]. Epidemiological data, preferably from prospective studies, should support the assertion that an infectious agent is a risk factor for a particular form of cancer. The stronger the associations reported in the literature and the more studies published on the specific relationship, the more support is lent to the claim. The examined relationship should be biologically plausible and preferably supported by animal data. When a virus is the suspected agent, its nucleic acid is likely to be found in cells of the relevant cancer. In vitro studies should confirm the relevant virus’s ability to stimulate proliferation or transfection in appropriate tissue culture cells.

In the next section, we will review specific infectious causes of cancer, the evidence supporting their role in carcinogenesis and their overall impact on the global burden of cancer. In the interest of space we have restricted the review to the most well-established associations.

Hepatitis B virus (HBV)

Hepatitis B virus (HBV), a circular DNA virus, 42-nm diameter, with a 3.2-kb partially double-stranded DNA genome, contains four open reading frames encoding the envelope protein (HBsAg), nucleocapsid (core), polymerase/reverse transcriptase and the ‘X’ protein. The liver is the principal site of infection, and detection of HBsAg in the blood is a diagnostic indicator of active HBV infection.

Globally, approximately 2000 million people have been infected with HBV, and 350 million are chronically infected carriers of the virus [9]. The highest prevalence of chronic HBV infection is in China, Asia and Africa [10]. The most common routes of infection with HBV are perinatally from mother to infant, during childhood, especially from infected older siblings, sexually, through blood transfusions or intravenous drug use in adult life [11]. Continuing presence of HBsAg in the blood for more than 6 months reflects chronic infection, and the chances of becoming a chronic carrier are much greater if infection with HBV occurred early in life, when the functional ability of the immune system to contain the virus is limited [12].

Chronic infection with HBV is a definite cause of hepatocellular carcinoma in humans based on geographical correlation, cohort and case–control studies [13, 14]. The overall relative risk for hepatocellular carcinoma for HBsAg carriers is estimated to be 13.7 times higher than for non-carriers [15], and HBV is estimated to be responsible for more than half of all liver cancer cases globally [3]. Given that chronic infection frequently occurs during the perinatal period and hepatocellular carcinoma is a disease of middle age, infection precedes cancer by many decades, and cofactors must be important. Indeed, synergism has been reported between HBV on the one hand and HCV [15, 16], and aflatoxin [17], on the other, in the development of hepatocellular carcinoma.

After infection, HBV-DNA may integrate into the human genome, in fact, several viral genomes can be inserted in a single human cell. One potential result is insertional mutagenesis. Common structural alterations after insertion include: deletions in host DNA; translocations; inverted duplication of integrated virus; and amplification of host DNA [18]. These rearrangements may occur either at the site of integration or even at a distance from this site, as virus insertion may trigger genetic instability [19].

The X protein coded for by the HBV genome remains intact in about one-third of integrants and may play an important role in liver carcinogenesis [20]. This protein is capable of altering cell growth in vitro and in vivo[11, 18]. The X protein enhances the activity of cellular genes through interactions with transcription factors, for instance those that control the transcription of c-myc[21]. The X protein may also bind with and interrupt the function of the p53 protein and other components of the DNA repair system [22, 23].

Despite the elegant way in which the molecular effects of the HBV genome have been elucidated, direct viral effects of HBV are unlikely to be the main mechanism for liver carcinogenesis. This conclusion has been reached after observing that not all hepatocellular carcinoma tumours contain HBV integrants, that integration is essentially random within the human genome [24], and by drawing parallels with the role of alcoholism and other chronic liver disease in the development of cirrhosis and hepatocellular carcinoma. Therefore, the promoting effects of HBV through hepatocellular necrosis, inflammation and subsequent regeneration are likely to be important in the development of cirrhosis, and ultimately hepatocellular carcinoma, as genetic alterations accumulate in hepatocytes. The inflammatory process may also be associated with oxidative stress, deleterious for the genome, as outlined above. However the promotion of cirrhosis cannot explain HBV’s entire role in liver cancer promotion, since HBV is associated with hepatocellular carcinoma occurrence, even in the absence of cirrhosis [25, 26]. HBV therefore has the properties of a ‘complete’ carcinogen, with both initiating – through DNA integration – and promoting capabilities [27].

Hepatitis C virus (HCV)

HCV virus is an RNA virus related to the Flaviviridae family. It has a 9.5-kb single-stranded RNA genome and contains many structural and nonstructural proteins: the C (core, E1 – envelope 1), E2/NS1 (nonstructural 1), NS2, NS3, NS4 and NS5. Because the mutation rate on replication of HCV is high, due to the lack of proof-reading capacity, many immunologically distinct variants of HCV, or quasispecies, exist [28]. These variants may escape host immune control, and allow chronic infection to be maintained [29], partly explaining why HCV infection becomes chronic in a high proportion of cases [30].

Prevalence of HCV is highest in Africa (especially in Egypt), Japan and other Asian countries, and overall affects about 3% of the world’s population [31]. In the USA, HCV is most commonly transmitted through parenteral exposure, and perhaps also sexually, so that prevalence increases with age [30]. Identified only in 1989, in 1994 HCV was already deemed to be definitely carcinogenic to humans, mainly on the basis of case–control studies [13]. The overall relative risk for hepatocellular carcinoma is 11.5 times higher in those who are anti-HCV/HCV-RNA positive compared with those who are anti-HCV/HCV-RNA negative [15]. HCV is estimated to cause about a quarter of all liver cancer cases globally [3]. Quasispecies may differ in the carcinogenic potential, the 1b quasispecies being particularly deleterious [32].

The lack of a convenient animal model means that our inferences on HCV must come exclusively from epidemiological studies. Such studies are impeded by the fact that 40% of HCV infection occurs without a recognized exposure, and many people infected with HCV remain unsymptomatic for many years [33]. HCV lacks the capacity to integrate into the host genome therefore in contrast to HBV, HCV seems to mainly exert its carcinogenic influence through chronic liver damage and stimulation of both humoral and cellular immune response. The consequent cellular regeneration is more pronounced in comparison with chronic HBV infection [34].

Evidence has recently emerged suggesting that direct carcinogenesis of HCV is possible, even though HCV does not integrate into the human genome and the HCV genome possesses no sequences similar to known oncogenes or tumour suppressor genes. This hypothesis is based on two observations: HCV replication occurs within hepatocellular carcinoma tumours [35]; and HCV can induce hepatocellular carcinoma, even in the absence of cirrhosis [26, 36]. Possible mechanisms for this direct effect include interaction between the core and proto-oncogenes at the cellular level [37], transformation of the ras gene by HCV [38], the inhibition of apoptosis by the HCV core protein [39] and the interruption of tumour necrosis factor receptor signal transduction pathways [40].

Epstein–Barr virus (EBV)

Epstein–Barr virus (EBV) is a member of the herpes family of viruses, from the subfamily Gammaherpesviridae. The virus is made up of a nucleocapsid containing a 172-kb double-stranded, linear DNA genome, replicated by a virally encoded DNA polymerase. After entry into the cell the virus becomes circular to form EBV episomes with a membranous envelope and glycoprotein spikes. The genome contains more than 100 genes. These code for Epstein–Barr nuclear antigen 1 (EBNA1), a latency protein needed for EBV episome replication; EBNA2, a transcription factor that activates EBV latency and immortalizes genes; and lymphocyte membrane-associated oncoprotein (LMP1), which activates transcription factors, interacts with cell signalling molecules, and may even interfere with p53 mediated apoptosis. The Z protein, which is responsible for initiating the switch from latent to lytic infection, can interact directly in vitro and in vivo with the tumour suppressor protein, p53 [41], whereas the R protein interacts with Rb [42]. EBV is transmitted through saliva and infects epithelial cells of the oropharynx, posterior nasopharynx, parotid gland and duct. B-lymphocytes become infected with the virus using a normal complement receptor (CR2), resulting in latent infection and stimulation of B-lymphocyte proliferation. Some infected B-lymphocytes escape immunological control, so that persistent infection is maintained through the EBV episome.

EBV infects over 90% of the world’s population [43]. In developing countries most children have seroconverted by the age of two years. In developed countries, and in children in high socioeconomic groups, infection occurs later in life, often presenting in the form of infectious mononucleosis, a benign lymphoproliferative disease.

EBV is an established carcinogen, with conclusive evidence with respect to non-Hodgkin’s lymphoma, Hodgkin’s disease and nasopharyngeal carcinoma [43]. EBV has also been linked to a number of other cancers, such as gastric carcinoma, but the evidence supporting these claims is weak. Evidence has accumulated, particularly with the onset of the AIDS epidemic, that EBV infection is particularly deleterious in an immunocompromised host.

EBV was first isolated from tissue samples of Burkitt’s lymphoma, giving the first clue that the virus and cancer are linked. Burkitt’s lymphoma is a rapidly growing B-cell lymphoma, occurring in young children, and is 100 times more common in Africa than in North America [44]. Over 98% of Burkitt’s lymphoma in Africa contain EBV episomes, but the corresponding fraction is much smaller in Burkitt’s lymphoma in North America [44–46] and there are other clear differences between endemic and sporadic Burkitt’s lymphoma cell lines [47]. Both the endemic and sporadic forms of Burkitt’s lymphoma are characterized by a translocation involving the immunoglobulin loci on chromosomes 14, 22 and 2 and the c-myc locus on chromosome 8, so that the c-myc proto-oncogene is activated [48]. EBV probably acts by stimulating B-lymphocytes to proliferate. In Burkitt’s lymphoma only the EBNA1 gene is expressed (type 1 latency)

EBV is consistently detected in nasopharangeal carcinoma, although the relationship is stronger for undifferentiated than well-differentiated types [43]. Progression from infected cell to carcinoma, through intermediate steps of dysplasia, is probably very rapid. EBNA1, LMP1, which has growth stimulating properties, and LMP2 are expressed in the carcinoma cells (type 2 latency). Although a relationship is apparent, this may be a result of re-activation of the virus in the cancer tissue, since a large prospective study [49] and a case–control study [50] found antibody titres to be unpredictive of the future development of nasopharangeal carcinoma.

Hodgkin’s disease is more frequent in upper socio-economic strata of Western populations. There are two peaks in the incidence of the disease, one at 25–30 years and another after age 45 [51], a fact that has prompted MacMahon to suggest that the early peak has an infectious aetiology [52]. About half of Hodgkin’s disease cases are thought to be attributable to EBV [3]. Malignant cells in Hodgkin’s disease are the Reed–Sternberg cells, and viral genome and virally encoded proteins are detected in about half of Hodgkin’s disease tumours. EBNA1, LMP1 and LMP2 are expressed (type 2 latency). The epidemiological evidence for this association is summarized by IARC [43] and is based on case–control studies and one prospective study [53] where antibody titres were elevated before diagnosis. Hodgkin’s disease development may be specifically linked to the earlier occurrence of infectious mononucleosis [54].

Human papilloma virus (HPV)

HPV form a family of viruses containing up to 200 types of small, nonenveloped DNA viruses with a closed, circular double-stranded genome, about 8000 bp in length. This family includes the early discovered Shope papillomavirus that can cause malignancy in cottontail rabbits. HPV has the capacity to integrate into the host genome. These viruses contain genes coding for early function (E1-E8), involved in viral DNA replication, transcriptional regulation, and cellular transformation. The late region (L1, L2) encodes two viral structural proteins whereas the long control region (LCR) contains elements needed for viral DNA replication and gene expression. HPV are highly tissue specific and productively infect only the basal cells of the squamous epithelium in the genital tract, skin and upper respiratory tract.

HPV is sexually transmitted. This is pointed to by a variety of epidemiological evidence; HPV prevalence is low amongst nuns and the sexually inexperienced; the prevalence increases with number of sexual partners and earlier age at first sexual intercourse; and there is a high concordance in infection between married couples [55]. Prevalence in developed countries is approximately 7%, whereas it is about 15% in developing countries [3].

HPV has been associated with cancer of the vulva, anus, penis and head and neck, but most convincingly with cervical cancer [55, 56]. Different strains of HPV show different levels of association with cancer. Sufficient evidence has accumulated from case–control studies for IARC to label HPV strains 16 and 18 as definite human carcinogens, and HPV strains 31 and 33 as probable human carcinogens [55]. Furthermore, types 35, 45, 51, 52, 58 and 59 are possibly related to carcinogenesis in humans [3]. HPV probably causes 80–90% of cervical cancer cases worldwide, assuming that infection with oncogenic HPV types increases risk about 60-fold, as indicated by meta-analysis of case–control studies [3].

The carcinogenic potential of HPV is related to its ability to integrate into the host genome, since cancer is rare in the absence of integration. Although integration is random with respect to the host genome, the integrity of viral regulatory regions and of the coding regions for the E6 and E7 genes, two genes important in carcinogenesis, are selectively retained. E6 and E7 are necessary and sufficient for immortalizing primary human genital keratinocytes in vitro[57]. They are regularly transcribed in the cancer cells, and inhibition of these genes can reverse in the carcinogenic process [58]. Furthermore, E2, a gene coding for a viral regulatory factor that can suppress the transcription of the E6 and E7 genes [58], is often disrupted during integration [57].

E6 and E7 have different roles in the development of cervical cancer. The E6 protein from high-risk – but not low-risk – HPV subtypes can form complexes with the tumour suppressor gene p53 [59], resulting in increased degradation of p53 [60] and reduced p53 binding [61]. This process, mediated by additional cellular factors [62], ultimately results in reduced apoptosis [63] and disruption of normal mitotic checkpoints [64], thus enhancing genomic instability and the accumulation of mutations. The E6 protein may also have telomerase activation potential [65], consequently delaying cell senescence.

The E7 protein exerts its carcinogenic potential by interaction with the products of the Rb gene. Rb proteins interact with factors produced by the E2 genes to form transcription repressor complexes, which ultimately control cell cycle regulation and DNA replication. The E7 proteins interact with the Rb proteins, so that the E2 factors are released and the cell enters the S phase [66]. Again, the ability of the E7 protein to bind with RB is higher for those from HPV strains associated with high malignancy [67]. Since blocking the E7-Rb binding site does not prevent E7’s ability to immortalize cells [68], other mechanisms of action must exist. The E7 protein may be able to interact directly with transcription factors, such as those from the AP-1 family [69], and cyclin-dependent kinase inhibitors [70].

Human herpes virus 8 (HHV8)

Human herpes virus 8 (HHV8), or Kaposi’s sarcoma-associated herpes virus (KSHV), is a gamma-2 herpesvirus, specifically a Rhadinovirus. At least 80 open reading frames have been identified, including genes with transforming potential, such as K1, as well as K9, which belongs to a family of transcription factors. In total, the virus contains at least four candidate transforming genes.

HHV8 is mainly transmitted sexually, particularly between men who have sex with men [71], although this virus can also be transmitted from mother to child. HHV8 is not ubiquitous in all populations [72], and prior to the onset of the AIDS epidemic Kaposi’s sarcoma was found mainly in sub-Saharan Africa.

HHV8 is associated with a number of malignancies, namely Kaposi’s sarcoma, non-Hodgkin’s B-cell lymphoma, primary effusion lymphoma and Casteleman’s disease, which share the common feature of being more prevalent in immunocompromised people [43]. Kaposi’s sarcoma, the most common of these malignancies, is found in elderly men, organ transplant recipients, people on immunosuppressive or cytotoxic therapy, and individuals with frank immunodeficiency diseases, notably AIDS [73]. The higher risk for Kaposi’s sarcoma amongst gay HIV-positive men compared with HIV-positive people in other risk groups first pointed to the infectious origin of this malignancy [74].

IARC considers the evidence linking KSHV/HHV8 to Kaposi’s Sarcoma as ‘compelling, but as yet limited’[43]. Nevertheless, HHV8 DNA is found in almost all forms of Kaposi’s sarcoma in people with or without HIV [75], there is strong geographical correlation between HHV8 prevalence and risk for Kaposi’s sarcoma, and HHV8 infection appears concentrated amongst those at risk for Kaposi’s sarcoma [72]. Furthermore, in one study antibody titres to HHV-8 at the start of follow-up were predictive of time to development of Kaposi’s sarcoma, independent of severity of HIV disease [71]. The study of HHV8 is hampered by the lack of reliable detection approaches, since assays are generally not sensitive to the low antibody titre levels of infected individuals [76]. This makes it difficult to produce accurate relative risk estimates for Kaposi’s sarcoma after infection with HHV-8. HHV8 appears, based on presently available evidence, to be at least a necessary, but not sufficient, factor for the development of Kaposi’s sarcoma.

HHV8 contains homologues to cellular proto-oncogenes, such as v-cyclin, which can complex with and inhibit RB, and v-bcl-2, which may play a role in inhibiting apoptosis. Inflammatory mechanisms could also be involved, as HHV8 may trigger an interaction between angiogenic factors and cytokines ultimately resulting in Kaposi’s Sarcoma production [77].

Human thymus-derived-cell leukaemia/lymphoma virus-1 (HTLV-1)

Human thymus-derived-cell leukaemia/lymphoma virus-1 (HTLV-1), a single-stranded RNA virus, approximately 9032 bp long, contains gag (viral core proteins), pol (reverse transcriptase), env (surface glycoprotein for receptor binding) and tax (transcriptional activator) genes. Genetically stable, this virus can be transcribed to double-stranded DNA, using viral reverse transcriptase, and randomly integrate into the host genome. HTLV-1 preferentially targets and immortalizes CD4-positive T cells. Another form of this virus, HTLV-2, is less well characterized in relation to tumour development and will not be reviewed.

The prevalence of HTLV-1 is highest (5–15%) in adults in south-western Japan, the Caribbean islands, South America, Central Africa, Papua New Guinea and the Solomon Islands, although there is substantial variation even within these areas [78]. The HTLV-1 genome is highly conserved across widely dispersed geographical locations. Infection is usually acquired in infancy, probably through infected breast milk, but also through sexual contact (mainly male to female) and blood transfusions [79–81].

Adult T-cell leukaemia (ATL) was identified as a distinct disease in the late 1970s [82]. It soon became apparent that ATL and HTLV-1 were associated both ecologically and analytically [83]. On the molecular level, leukaemic cells are infected with HTLV-1 and show monoclonal integration patterns [84]. Infection with HTLV-1 can immortalize CD4+ T cells, but not other cells in vitro[85]. HTLV-1 is judged to be definitely carcinogenic [86]. Incidence of ATL is approximately 1.0 per 1000 male HTLV-1 carriers, and 0.5 per 1000 female carriers [3], but HTLV-1 acquired in infancy appears to be more strongly associated with risk for ATL [80].

The HTLV-1 genes, in particular tax, appear to promote cell proliferation and immortalization. Although it is unclear which mechanism is most important, HLA type is a significant interacting factor. tax binds to enhancer-binding proteins and activates cellular transcription factors of specific genes, so that cells infected with HTLV-1 undergo proliferation without stimulatory signals. Moreover, tax binds to and suppresses transcriptional inhibitory proteins and cell cycle inhibitors [87].

Human immunodefiency virus (HIV)

HIV is a human retrovirus of the lentivirus subfamily. It has a single-stranded RNA genome. HIV replicates through reverse transcriptase, using a DNA intermediate, and can integrate into the host genome. It is transmitted sexually, mother to child and parenterally.

IARC has deemed HIV-I to be carcinogenic to humans [86]. The most common cancers in people who are HIV positive are Kaposi’s sarcoma and high grade non-Hodgkin’s lymphoma; these are, in fact, AIDS-defining malignancies. Non-Hodgkin’s lymphoma occurs late in the process of AIDS. Up to 10% of people with HIV will eventually develop non-Hodgkin’s lymphoma, although this figure may be lower in the developing world [3]. Hodgkin’s disease, squamous cell carcinoma of the conjunctiva and leiomyosarcoma also appear to be associated with HIV infection [86]. Other cancers, such as invasive cervical cancer, which has even been included as an AIDS-defining diagnosis, have also been considered, although these associations are weaker and possibly confounded.

Mechanisms for the carcinogenic potential of HIV are related to the dramatically compromised immune system, since cancer is associated with a wide range of immunosuppression disorders [6]. There may be direct effects of HIV infection, such as insertional mutagenesis, upregulation of oncogenes, chronic antigenic stimulation or cytokine disregulation. It has also been hypothesized that the HIV tat protein, a regulatory protein, has a growth-promoting effect on Kaposi sarcoma lesions.

Helicobacter pylori (H. pylori)

Helicobacter pylori (H. pylori) is a gram-negative, curved bacterium, first isolated from gastric biopsies in 1982. H. pylori is 2.5–5.0 µm long, with four to six unipolar flagellae. It causes a persistent bacterial infection of the stomach, where it is found, free-living, beneath the mucus overlaying the gastric epithelium.

The prevalence of H. pylori varies widely globally; approximately 30% of people in Denmark [88] and up to 80% in Japan and Africa [89] are infected by the bacteria. Prevalence of H. pylori infection is associated with age and poverty, as, although the routes of infection remain to be firmly established, most infection occurs during childhood probably through person–person or faecal–oral transmission [90]. The prevalence is highest in the oldest age group in developed countries, reflecting both age (increasing) and cohort (decreasing) patterns [3].

H. pylori is established as a definite carcinogen for the development of gastric cancer [91], the second most common type of cancer globally [3]. The evidence for this is based on ecological, cohort and case–control studies [91], suggesting a doubling to tripling in risk for gastric cancer with H. pylori infection [3, 92]. The falling prevalence of H. pylori infection in the USA is mirrored by a decreasing incidence of stomach cancer; in 1900 it was the most common cause of cancer mortality, but it is rare today in this country. Unfortunately rates are still high in eastern Asia, central and tropical South America and Eastern Europe [3]. Gastric adenocarcinoma, which makes up over 95% of all gastric cancer, appears to consist of several distinct diseases dividing the tumours aetiologically and epidemiologically by site (cardia/noncardia) and histological type (intestinal/diffuse type) [93]. Noncardia cancer is the most common and seems to be more closely related to H. pylori infection than cardia cancer. Furthermore, a prospective study in the USA indicated that H. pylori may be associated with gastric lymphoma, although this is a very rare condition [94]. Given that the excess risk associated with exposure to H. pylori infection is approximately doubled it has been estimated to be responsible for more than half of the cases of stomach cancer and about three quarters of the cases of gastric lymphoma, globally [3]. However, recent data indicate that the risk associated with the bacteria may have been underestimated due to spontaneous eradication in the precancerous stomach, potentially increasing the fraction of gastric cancer attributable to H. pylori to more than 75% [95].

The chronic inflammation mechanism is likely to be responsible for the production of stomach cancer following H. pylori infection, since removal of infection can trigger regression of precancerous lesions [92]. Individual responses to infection vary widely, depending on environmental factors such as diet, duration of infection and age at acquisition of H. pylori infection [90, 96], the virulence of H. pylori strains [97] and host factors, including genetic make-up. Given these provisos, H. pylori causes chronic active inflammation of the gastric mucosa in the majority of infected patients, with consequent progressive structural changes in the gastric mucosa, and eventual establishment of atrophic gastritis, a process that generally takes 20–40 years [98]. The increased cell turnover and the degradation of the mucous barrier produce circumstances conducive to cancer development. Furthermore, the inflammatory response involves the attraction of white blood cells, such as polymorphonuclear cells and macrophages, and consequent increased production of cytokines and ROS and RNOS, which may interfere with antioxidant functions and induce oxidative DNA damage in the gastric mucosa. H. pylori is also associated with lowered gastric acid output, which can stimulate N-nitrosamine (a mutagenic agent) formation, decrease the prevalence of antioxidants (particularly vitamin C) and promote proliferation of epithelial cells.


Schistosomes are dioecious parasitic blood flukes, of the flatworm variety, who have a mammalian host and an intermediate invertebrate host: fresh water snails. Schistosomes are released from the snails into fresh water as free-living cercariae, which penetrate the skin of the mammal host. Here the parasites develop into schistosomal larvae, migrate through the blood stream to the liver, mature and eventually release eggs, which either lodge into tissues and incite a granulomatous reaction, or are released in the mammal’s urine and faeces. The eggs hatch in the water to form miracidia, which invade snails, starting the cycle anew. Different species exist: Schistosoma haematobium (S. haematobium), common in Africa and the Eastern Mediterranean, and Schistosoma japonicum (S. japonicum), found in China, the Philippines and Indonesia, are of major importance to humans. Schistosoma mansoni (S. mansoni) is highly prevalent, however evidence linking it to cancer occurrence is weak.

Approximately 200 million people are currently infected with schistosomes [99], most commonly with S. haematobium and S. mansoni. Prevalence of infection peaks in the second decade of life.

S. haematobium is a definite cause of bladder cancer [91], with an associated five-fold increase in risk [3]. The IARC decision was based on ecological studies, reporting strong positive correlation, case-reports and six out of seven case–control studies. Bladder cancer associated with S. haematobium is histologically and pathologically distinct from non-S. haematobium-associated bladder cancer occurring in North America and Europe, being a squamous-cell carcinoma, with an earlier age at onset and generally sparing the trigone. The evidence for a role of S. japonicum in cancer occurrence is weaker, although it has been associated with both liver and colorectal cancer, so that S. japonocum is judged to be possibly carcinogenic to humans [91].

Schistosomiasis, for the most part, causes cancer because the worms and eggs are deposited in the tissue, inciting a chronic inflammatory reaction. The inflammatory reaction has the dual effect of causing granulomatous lesions to form, possibly blocking venules, but definitely increasing cell-turnover, and promoting the production of ROS and RNOS [100]. It should therefore not be surprising that the suspected agent is usually found at the site of cancer occurrence: S. haematobium is found mainly in the small venules that drain the bladder and ureters; and S. japonicum reside in the venules that drain the gastrointestinal tract or in the liver. Moreover, experimental evidence suggests that the invasion of the liver by worms can alter carcinogen metabolism or promote the release of endogenous carcinogens [101].

Liver flukes

Liver flukes are also flatworms, three species of which can inhabit the human liver: Opisthorchis viverrini (O. viverrini), Clonorchis sinensis (C. sinensis) and Opisthorchis felineus (O. felineus). O. viverrini is a hermaphrodous liver fluke, approximately 7–12 mm long, and 1.5–3 mm wide. This parasite usually inhabits the intrahepatic bile ducts of humans, but on occasion the pancreatic duct and gallbladder. From this position, the liver fluke lays eggs that enter the biliary system and are excreted in the faeces, to be ingested by Bithynia snails, the first intermediate hosts. The eggs hatch inside the snail and mature, eventually leaving the snail in the form of freely swimming cercariae that find and penetrate fish, the second intermediate host, where they develop into metacercariae. The parasite enters humans after the ingestion of raw fish, and travels to the intrahepatic bile duct where the liver fluke can live for up to 30 years. C. sinensis and O. felineus are similar to, although slightly larger than, O. viverrini.

O. viverrini is endemic in North-east Thailand and Laos; in total infecting approximately 9 million people [91]. C. sinensis infects 7 million people in China, Korea, Taiwan and Vietnam, and O. felineus infects 1.5 million in Eastern Europe [91].

Problems with detecting liver flukes and the rarity of infection and cholangiocarcinoma outside of South-east Asia complicate the study of liver flukes in relation to cholangiocarcinoma. However, after evaluating ecological studies, case series and case–control studies, the IARC determined that O. viverrini is a definite human carcinogen, whereas evidence for the carcinogenic role of O. felineus and C. sinensis is more limited [91]. In one particularly large case–control study, Parkin and his colleagues found that past or present infection with O. viverrini increased the risk for cholangiocarcinoma five-fold, and so could be responsible for two-thirds of cholangiocarcinoma incidence in Thailand [102].

The rarity of cholangiocarcinoma is in sharp contrast to the high prevalence of liver fluke infection, indicating that cofactors are important. Animal models confirm that the mechanical irritation of the intrahepatic bile duct by the liver fluke, or its metabolites, may cause epithelial desquamation, with consequent cell regeneration [103]. High cell turnover is also stimulated by the inflammation process, which may result in occlusion of the bile duct, as well as production of ROS and RNOS [104]. The mechanical process is likely to be more important than the immunological effects, as innovative animal models demonstrate that mechanical damage alone can explain most tumour development [105]. Animal models indicate, however, that in the absence of carcinogens, cholangiocarcinoma is unlikely to develop, so that liver flukes are for the most part promoters, not initiators, of cholangiocarcinoma [105].

Prospects for the primary prevention of infections causing cancer

In general, the prospects for the prevention of malignancies caused by infections are good. An effective HBV vaccine has been developed and immunization programmes are being implemented throughout the world. With increasing care given to the screening of blood, the transmission of HBV, HCV, HTLV-1 and HIV through blood transfusions will be reduced. Furthermore, improvements in education and socio-economic status are likely to result in reduced prevalence of infection with schistosomes, liver flukes and H. pylori. Finally, there have been technical improvements in the early detection of the epithelial changes leading to cervical cancer, so that most invasive cervical carcinoma can be prevented. Important concerns do remain, however. The prevalence of injecting drugs remains high, which may promote the spread of HBV, HCV and HIV. And there is of course still the biggest question of all: what will be the future of the AIDS epidemic and how is this likely to impact on the incidence of infection-induced cancers?


We would like to acknowledge Anna-Mia Ekström for generously sharing with us her expertise.