An Introduction to Virus Infections and Human Cancer

1. The Burden of Virus-Induced Cancers

The determination that infection by a specific subset of human viruses is the primary cause of a substantial fraction of human cancers is one of the most important achievements in cancer etiology and intervention. It was recently estimated that a virus infection is the central cause of more than 1,400,000 cancer cases annually, representing approximated 10% of the worldwide cancer burden (Plummer et al. 2016). The widely accepted human oncoviruses are human papillomaviruses (HPV), hepatitis B virus (HBV), hepatitis C virus (HCV), Epstein–Barr virus (EBV), Kaposi’s sarcoma-associated herpesvirus (KSHV) (also called human herpesvirus 8), human T-cell lymphotropic virus (HTLV-1), and Merkel cell polyomavirus (MCPyV). This conclusion is based on the cumulative knowledge from a large number of experimental, clinical, and epidemiological studies over the last five decades.

An additional 5% of worldwide cancers, mostly gastric cancer, are attributed to infection by the bacterium Helicobacter pylori (Plummer et al. 2016). The number of worldwide incident cases associated with specific virus varies widely, from 640,000 for HPV to 3000 for HTLV (Table 1). Approximately, 85% of the burden of virus-induced cancers is bourn by individuals in the developing regions of the world, with profound implications for translating the knowledge of virus-induced cancers into public health interventions. In addition, some viruses cause more cancers in one sex than the other. Almost 90% of HPV-induced cancers occur in females, while approximately, two-thirds of HBV, HCV, and EBV cancer occur in men (Table 1).

Table 1.

Number of new cancer cases attributable to specific viral infections by gender

Virus	Total	Females	Males
HPV	636,000	570,000	66,000
HBV	420,000	120,000	300,000
HCV	165,000	55,000	110,000
EBV	120,000	40,000	80,000
KSHV	43,000	15,000	29,000
HTLV	2,900	1,200	1,700

Data from (Plummer et al. 2016)

2. The Human Cancers Caused by Viruses

Many different types of cancers are induced by human oncoviruses, and the fraction of each cancer type attributed to a viral infection varies widely (Table 2) (Plummer et al. 2016). HPVs normally infect stratified epithelium and are causally associated with a number of anogenital carcinomas, including cervical, anal, vulvar, vaginal, and vulvar, and also carcinomas of other mucosal epithelium, most notably oropharyngeal. The attributable fraction varies from almost 100% for cervical carcinomas to 4% for oral cavity and laryngeal cancers. Interestingly, the rates of HPV-associated oropharyngeal cancers appear to be substantially increasing in Western Europe and North America, most notably in men (de Martel et al. 2017).

Table 2.

Prevalence of viruses in virus-associated cancers

Virus	Cancer	Geographical area	Attributable fraction (%)
HPV	Cervix	World	100
HPV	Penile	World	51
HPV	Anal	World	88
HPV	Vulvar	World	48a
HPV	Vaginal	World	78
HPV	Oropharynx	North America	51
HPV	Oropharynx	India	22
HPV	Laryngeal	World	4.6
HBV	Liver	Developing	59b
HBV	Liver	Developed	23b
HCV	Liver	Developing	33b
HCV	Liver	Developed	20b
EBV	Hodgkin’s lymphoma	Africa	74
EBV	Hodgkin’s lymphoma	Asia	56
EBV	Hodgkin’s lymphoma	Europe	36
EBV	Burkitt’s lymphoma	Sub-Saharan Africa	100
EBV	Burkitt’s lymphoma	Other regions	20–30
EBV	Nasopharyngeal carcinoma	High-incidence areas	100
EBV	Nasopharyngeal carcinoma	Low-incidence areas	80
KSHV	Kaposi’s sarcoma	World	100
HTLV-1	Adult T-cell leukemia and Lymphoma	World	100
MCPyV	Merkel cell carcinoma	North America	70–80c

Data from (Plummer et al. 2016) unless otherwise indicated

^aAge 15–54 yrs

^bData from (de Martel et al. 2012)

^cEstimate from (Tetzlaff and Nagarajan 2018)

HBV and HCV have a strict tropism for hepatocytes and together are the cause of three-quarters of hepatocellular carcinomas (Petrick and McGlynn 2019) (Table 2). EBV normally infects B lymphocytes and epithelial cells and induces about half of Hodgkin’s lymphoma and Burkitt’s lymphomas (Saha and Robertson 2019). It is also an etiologic factor in most cases of nasopharyngeal carcinoma (Chen et al. 2019). In addition, EBV-associated gastric cancer is a distinct clinicopathological entity that is present in ~9% of these cancers (Bae and Kim 2016).

Virtually, all Kaposi’s sarcomas are associated with KSHV infection. KSHV is also strongly associated with multicentric Castleman’s disease and primary effusion lymphoma, two relatively rare B-cell neoplasms (Katano 2018). HTLV-1 infects lymphocytes and is a central cause of adult T-cell leukemia and lymphoma (Tagaya et al. 2019). MCPyV is commonly detected in normal skin and is responsible for approximately three-quarters of Merkel cell carcinoma, a relatively uncommon skin cancer (Kervarrec et al. 2019).

3. The Diversity of Human Oncoviruses

Human tumor viruses are highly diverse, including viruses with large double-stranded DNA genomes (EBV and KSHV), small double-stranded DNA genomes (HPV, HBV, and MCPyV), positive-sense single-stranded RNA genomes (HCV), and retroviruses (HTLV-1) (Table 3). Some have enveloped virions, specifically HBV, HCV, EBV and KSHV and HTLV-1, whereas others have naked icosohedral virions, specifically HPV and MCPyV. The carcinogenic mechanisms of oncoviruses also vary widely, as outlined below. However, in all cases, oncogenesis is an uncommon consequence of the normal viral life cycle. Virus-induced cancers almost always arise as monoclonal events from chronic infections, usually many years after the primary infection, indicating that infection is just one component in a multi-step process of carcinogenesis. The exceptional case is KSHV-induced Kaposi’s sarcoma, which can arise as a polyclonal tumor within months of infection in immunosuppressed individuals (Cesarman et al. 2019) (also see Chap. 13).

Table 3.

Basic features of human oncoviruses

Virus	Genome	Virion structure	Normal tropism	Year isolated (reference)
EBV	Linear 172 kb DS DNA	Enveloped	Epithelium and B cells	1964 (Epstein et al. 1964)
HBV	Circular 3.2 kb partial DS DNA	42 nm enveloped	Hepatocytes	1970 (Dane et al. 1970)
HTLV-1	Linear 9.0 k nt positive-sense RNA	Enveloped	T and B cells	1980 (Poiesz et al. 1980)
HPV16	Circular 7.9 kb DS DNA	55 nm naked Icosahedron	Stratified squamous epithelium	1983 (Dürst et al. 1983)
HCV	Linear 9.6 k nt positive-sense RNA	Enveloped	Hepatocytes	1989 (Choo et al. 1989)
KSHV	Linear 165 kb DS DNA	Enveloped	Oropharyngeal epithelium	1994 (Chang et al. 1994)
MCPyV	Circular 5.4 kb DS DNA	40 nm naked icosahedron	Skin	2008 (Feng et al. 2008)

4. Oncogenic Mechanisms

The oncogenic mechanisms of many tumor viruses, as detailed in later chapters, involve the continued expression of specific viral gene products that regulate proliferative, anti-apoptotic, and/or immune escape activities through an interaction with cellular gene targets. Examples of oncoproteins include E6 and E7 of HPVs, LMP1 of EBV, Tax of HTLV-1, and T antigen of MCPyV (Chaps. 8, and 10 respectively). Although HCV- and HBV-encoded proteins may play a direct role in hepatocarcinogenesis, these viruses may primarily induce cancer more indirectly, as a result of persistent infection causing chronic inflammation and tissue injury (Kanda et al. 2019a, b) (Chaps. 3, 4 and 6). KSHV may act primarily by altering complex cytokine/chemokine networks (Mesri et al. 2010) (Chap. 13). Virally encoded microRNAs, especially those of KSHV and EBV, also play a direct role in carcinogenesis (Wang et al. 2019).

With some viruses, e.g., MCPyV and HPV, malignant progression usually involves mutation and/or insertion of the viral genome into the host DNA, such that it can no longer replicate (Vinokurova et al. 2008; Arora et al. 2012). In addition, the advent of high-throughput sequencing has facilitated the evaluation of the strain variation and cancer risk. Interesting examples have been uncovered for EBV (Kanda et al. 2019a, b) and HPV16 (Mirabello et al. 2016), but the molecular mechanisms that account for these strain differences in carcinogenic potential are not entirely clear.

HIV infection is a strong risk factor for several cancers, particularly cancers that are associated with other virus infections (Vangipuram and Tyring 2019). However, the effect of HIV infection on oncogenesis is thought to be indirect, by inhibiting normal host immune functions that would otherwise control or eliminate the oncovirus infections and/or provide immunosurveillance of emerging tumors. Supporting this hypothesis is the observation that the rates of the same cancers increase in patients with other forms of immunosuppression (Grulich and Vajdic 2015). Other types of retroviruses can induce cancers by insertional mutagenesis in animal models (Fan and Johnson 2011). However, this activity has not been convincingly documented in humans, except in a few patients in experimental gene transfer trials involving delivery of high doses of recombinant retroviral vectors (Romano et al. 2009).

4.1. Viral Infection and Cancer: Establishing Causality

Detecting a virus in a cancer does not establish causality. For instance, the cancer cells might simply be exceptionally susceptible to infection or replication by a particular virus, a scenario that is currently being exploited experimentally in the development of oncolytic virotherapies (Russell and Barber 2018). However, the causal associations between the seven viruses and specific cancers noted above are now convincingly established. They fulfill most, if not all, of the criteria for causality proposed by Sir A. Bradford Hill in the early 1970s (Hill 1971). Multiple epidemiological studies in varying settings have established the strength and consistency of association between infection and cancer for these viruses. Most strikingly, relative risks of over 100 have been calculated for HPV and KSHV infection in the development of cervical carcinoma and Kaposi’s sarcoma, respectively, among the highest observed for a cancer risk factor (Moore and Chang 1998; Bosch et al. 2002). In some instances, establishing a strong association required identification of especially oncogenic strains, e.g., HPV16 and 18 among mucosotropic HPVs, or a specific tumor subsets, e.g., oropharyngeal carcinomas among head and neck cancers. Temporality was established by demonstrating that infection proceeds cancer, usually by many years. Integration of the viral gene in the same site in all tumor cells further demonstrated, for some viruses, that the viral infection was an initiating event.

In some cases, the viruses are consistently detected in well-established cancer precursor lesions, as is the case for HPV and high-grade cervical intraepithelial neoplasia (Chap. 8), although in others, such as MCPyV-induced MCC and HPV-associated oropharyngeal cancer, the precursor lesions have not been clearly identified. Demonstrating that, for the most part, populations with higher prevalences of virus infection also had higher incidences of the associated tumor, e.g., HBV and liver cancer, established important dose–response relationships (El-Serag 2012) (Chap. 5). However, these associations are sometimes confounded by high prevalence of the oncovirus in the general population and variability in the exposure to other risk factors. A clear example is the high frequency of EBV infection in the general population, but the induction of EBV-positive Burkitt’s lymphoma primarily in areas with a high incidence of malaria (Moormann and Bailey 2016). A large number of laboratory studies established biological plausibility for causality by characterizing the interaction of viral proteins or other viral products with key regulators of proliferation and apoptosis and establishing their immortalizing and transforming activity in vitro and their oncogenic activity in animal models (Chap. 4, 6, 8, 10, 11, and 13). These studies also support the criterion that the associations be in agreement with the current understanding of disease pathogenesis, in this case the molecular biology of carcinogenesis (Mesri et al. 2014). The last Hill criterion, that removing the exposure prevents the disease, has been most convincing demonstrated for HBV and HPV, after introduction of the corresponding vaccine.

4.2. The Importance of Identifying the Viral Etiology for a Cancer

The identification of a virus as a central cause of a specific cancer can have several substantial implications. First, it can provide basic insights into carcinogenic processes, especially the identification of potential cellular targets for diagnoses and interventions that are often relevant to both virus-associated and virus-independent tumors (Mesri et al. 2014). Studies of virally induced carcinogenesis were particularly illuminating in the past decades when the ability to analyze the complexity of host cell genomics and proteomics was much more limited than it is today. For example, the tumor suppressors p53 and pRb were first identified as binding partners of the small DNA tumor viruses in experimental systems and later shown to be targets for several human oncoviruses (Pipas 2019). They are also among the most frequently mutated genes in non-virally induced cancers (Chap. 8).

Second, the presence of the virus can be used in cancer risk assessment and prevention. One example is the increasing use of HPV DNA testing to screen for cervix cancer risk. HPV DNA tests are more sensitive for detection of high-grade premalignant lesions than is the standard Pap test, so intervals between tests can be increased in women who test negative for high-risk HPV DNA in their cervix (Rizzo and Feldman 2018). Another example is HCV screening to identify individuals at high risk of progression to liver cirrhosis and cancer (Chap. 7). HCV screening is now recommended in the USA for all individuals born between 1945 and 1965 (Smith et al. 2012), although the US Prevention Services Task Force has recently made a draft recommendation for widening screening ages to everyone older than 11 years of age (https://www.uspreventiveservicestaskforce.org/Page/Document/draft-research-plan/hepatitis-c-screening1).

Third, viral gene products provide potential targets for therapeutic drugs or therapeutic vaccines for the treatment cancers, precancerous lesions, or chronic infection. There has been substantial research activity in this potentially fruitful area. Although they have not led to viral-based treatment of malignancies, drug studies have had considerable success in the development of antivirals to treat chronic HBV and HCV infection. Pegylated interferon alpha plus a nucleoside/nucleotide analog is currently being used to suppress HBV replication (Chap. 5) and thereby liver cirrhosis and risk of hepatocellular carcinoma (Ren and Huang 2019). HCV infection can be similarly treated, and a series of direct-acting antiviral drugs targeting the NS3/4A protease and NS5A and NS5B polymerase have been developed that, remarkably, increase sustained virologic response rates to 90–98% (Pradat et al. 2018) (Chap. 7). Treatment of KSHV infection/Kaposi sarcoma centers on reducing the underlying immunosuppression that promotes the disease. Kaposi’s sarcoma lesions often regress in HIV-infected individuals after initiation of HAART, but this response is due to reconstitution of the immune system, rather than to direct activity of the drugs against KSHV (Cesarman et al. 2019). No licensed direct-acting antivirals have been developed for HPVs or MCV, despite considerable efforts in the case of HPVs.

Fourth, the knowledge of a viral etiology can serve as the basis of cancer prevention measures. One approach involves behavioral interventions to reduce susceptibility to infection, e.g., limiting exposure to blood products in the case HBV and HCV (Chaps. 5 and 7), limiting number of sexual partners in case of HPV or KSHV, or preventing HTLV-1 transmission by discouraging breast-feeding by infected mothers (Ruff 1994) (Chap. 12).

Alternatively, the identification of human oncoviruses can be used to develop effective vaccines to prevent oncovirus infection. This approach has been successfully implemented for HBV and HPV. HBV prophylactic vaccines were introduced more than thirty years ago (Chap. 5). A dramatic reduction in childhood liver cancers of greater than two-thirds has been documented in Taiwan, a previously high incidence region (Chang et al. 2016). A substantial reduction in adult liver cancer is expected in the near future, as individuals who would have otherwise contracted HBV as infant reach the age of peak cancer incidence. HPV vaccines targeting HPV16 and 18 have been licensed for more than a decade (Schiller and Lowy 2012) (Chap. 9). While substantial reductions in the incidences of HPV-associated cancer are expected in coming years, there has already been a significant reduction in premalignant cervical disease and evidence of herd immunity developing in countries with high vaccination rates (de Sanjose et al. 2019).

There are considerable efforts underway to develop prophylactic and/or therapeutic vaccines against EBV (van Zyl et al. 2019) and HCV (Bailey et al. 2019) (Chap. 7). Commercial development of prophylactic vaccines against these viruses seems possible because there are reasonable non-malignant disease endpoints for initial licensure, mononucleosis in the case of EBV, and liver cirrhosis in the case of HCV. However, specific characteristics of their biology have made the development of effective vaccines challenging. These include multiple entry receptors and viral latency in the case of EBV, and genetic instability in the case of HCV. There has been less effort devoted to developing KSHV and HTLV-1 vaccines because the numbers of worldwide cancers they induce are lower, and these viruses do not appear to be a frequent cause of medically important non-malignant disease, in contrast to EBV and HCV (Schiller and Lowy 2010). They also have proven susceptible to other interventions, specifically reduction in KSHV-induced Kaposi’s sarcoma by treating HIV infection and reduction in transmission of HTLV-1 by discouraging infant breast-feeding by infected mothers (Chap. 12). There has been relatively little activity in developing MCV vaccines, perhaps because the natural history of infection is poorly understood, there are no known premalignant lesions or other disease to target, and the cancers are relatively rare.

4.3. The Search for Additional Oncoviruses

Are there other human oncoviruses waiting to be discovered? There are suggestions that viral infections may be associated with several other cancers. For instance, the incidence of non-melanoma skin cancer increases dramatically after immunosuppression, and immunosuppression is associated with clear increases in established virally associated cancers (Grulich and Vajdic 2015). Other cancers that increase less dramatically in immunosuppressed patients and have not been firmly linked to a microbial infection include lung, conjunctival, melanoma, lip, esophageal, laryngeal carcinomas and multiple myeloma. Some epidemiological studies have linked the risk of prostate cancer with sexual activity variables, suggesting involvement of a sexually transmitted infectious agent (Sutcliffe 2010), but its incidence is not increased in immunocompromized individuals (Grulich and Vajdic 2015). Other virus/cancer associations that warrant further investigations include hepatitis delta virus, an HBV-dependent single-strand RNA virus that appears to increase the risk of HCC in HBV-coinfected individuals by threefold (Koh et al. 2019), and BKV in bladder cancer, particularly in immunocompromised patients (Starrett and Buck 2019).

The technologies of high throughput nucleic acid sequencing of entire cellular genomes and transcriptomes, as now applied to a wide variety of human tumors, provide an unprecedented wealth of raw data for the hunt for novel oncoviruses. The discovery of MCV illustrates how this technology can be employed to identify novel human oncoviruses (Feng et al. 2008). However, recent studies that screened for all known viral species have mostly detected established oncoviruses in the tumor collections (Cantalupo et al. 2018). Nevertheless, the possibility remains that highly divergent oncoviruses with undetectable homology to currently known viruses remain to be discovered. However, identification of a viral nucleic acid sequence in a tumor is only the first step. It takes many additional laboratory, clinical, and epidemiological studies to establish that a viral infection is a causal agent in the development of a cancer, as opposed to a passive parasite or simple contaminant of the tumor. Establishment of causality can be particularly difficult in situations where the implicated virus is a common infection in the general population.

It will also be difficult to establish causality if a virus is involved in the initiation of a tumor but not in its maintenance, with the viral genome being lost during progression. Commonly referred to as a “hit and run,” it is a plausible but as yet unproven mechanism for human cancers (Schiller and Buck 2011), although there are several well-documented examples of this phenomenon in experimental animal models (Viarisio et al. 2018). Further supporting the possibility of this mechanism is the recent finding that 8% of cervical cancers that are positive for HPV DNA do not express detectable HPV transcripts, functionally equivalent to a hit-and-run (Banister et al. 2017). It may be informative to more closely examine premalignant lesions and cancers in immunocompromised patients, where there may be less selection for eliminating viral gene expression, for further evidence of viruses that may initiate cancers via this mechanism (Starrett and Buck 2019).

5. Conclusions

The discovery and characterization of oncoviruses have been at the forefront of biomedical research over the last several decades. It has provided important insights into basic cell biology and mechanisms of carcinogenesis. In addition, these studies have generated important public health interventions that have the potential to prevent a large number of human cancers. This monograph provides clear evidence that oncoviruses remain a dynamic subject in biomedical science. We expect that future research will generate new and excites insights into the genesis of cancer and effective ways to prevent and treat it.