Role of human papillomaviruses in esophageal squamous cell carcinoma

Authors


  • Potential conflicts of interest

    ATN has previously received research funding for other projects from a manufacturer of human papillomavirus vaccine.

    SG has received advisory board fees and grant support from Commonwealth Serum Laboratories and GlaxoSmithKline, and lecture fees from Merck, GlaxoSmithKline and Sanofi Pasteur; in addition, she has also received funding through her institution to conduct human papillomavirus vaccine studies for Merck Sharp Dohme and GlaxoSmithKline. She is a member of the Merck Global Advisory Board as well as the Merck Scientific Advisory Committee for HPV.

Correspondence: Dr Surabhi S. Liyanage BMedSc MBBS (Hons), School of Public Health and Community Medicine, Faculty of Medicine, University of New South Wales, Randwick, Sydney, NSW 2052, Australia. Email: surabhi_liyanage@hotmail.com

Abstract

Esophageal cancer (EC) is responsible for almost half a million deaths worldwide annually and has a multifactorial etiology, which may account for its geographical variation in incidence. In the last 30 years the potential of human papillomaviruses (HPV) as oncogenes or co-factors in the tumorigenic process of esophageal squamous cell carcinoma (ESCC) has been widely studied. While the etiology of HPV in cervical and certain other anogenital and aerodigestive cancers has been established, results regarding its role in EC have been largely inconclusive. A causal association can be evaluated only with a case-control study, where normal controls are compared to ESCC cases for the presence of HPV. We reviewed all studies investigating ESCC tissue for HPV DNA and identified 139 that met our inclusion criteria, of which only 22 were case-control studies. Our results support previous findings of higher levels of HPV detection in high-risk ESCC regions than in areas of low risk. In addition, we confirm that the role of HPV in ESCC remains unclear, despite an accumulation of studies on the subject. The variations in investigative technique, study design and sample types tested may account for the lack of consistency in results. There is a need for a meta-analysis of all case-control studies to date, and for large, well-designed case-control studies with adequate power to investigate the association. The potential benefits of prophylactic HPV vaccines could be evaluated if HPV is identified as an etiological factor in EC, highlighting the need for further research in this area.

Introduction

Research into the role of human papillomaviruses (HPV) in squamous cell carcinomas of the conjunctiva, respiratory tract, oropharyngeal cavity and upper digestive tract expanded in the early 1980s. While the contribution of certain oncogenic HPV types to various anogenital malignancies, particularly cervical cancer, where HPV plays virtually a 100% etiological role, is now accepted, the association between esophageal cancer (EC) and HPV remains controversial.

In 1982 Syrjänen et al. first suggested a possible link between HPV and esophageal squamous cell papilloma (ESCP) following the observation of the characteristic cytopathic changes of HPV infection usually seen in condylomatous lesions, in both benign esophageal epithelial tissue and malignant esophageal tumors.[1] This was supported by the first reports of HPV detection in lesions of the oral mucosa, which is continuous with the squamous epithelial lining of the esophagus.[2] Syrjänen et al.'s postulate was validated by studies that used immunohistochemistry to detect HPV structural proteins in esophageal lesions.[3, 4] Subsequent investigations have been indeterminate, with the detection of HPV DNA in esophageal squamous cell carcinoma (ESCC) tissue varying from 15–80%.[5] Therefore, while the International Agency on Research on Cancer (IARC) has acknowledged HPV involvement in the head and neck, particularly oropharyngeal, tonsillar and oral squamous cell carcinomas, no conclusive statement has been made on a causal relationship between HPV and ESCC, suggesting that further research is required.[6]

The last review of evidence for HPV involvement in EC was undertaken before the introduction of the HPV vaccines as a public health intervention for cervical cancer,[5] with an update in 2006.[7] We examined recent data as a basis to evaluate the potential impact of vaccination for other possible HPV-related cancers such as ESCC. As the proposed association of HPV with EC has been confined to ESCC, this review concentrates on this histological type.

Methods

Search strategy

The MEDLINE and EMBASE databases were searched for studies that tested for the presence of HPV DNA in ESCC tissue. Key search terms included human papillomavirus, HPV, esophageal cancer and squamous cell carcinoma. No restrictions were placed on date of publication or language. However, foreign language articles with non-English abstracts and those with English abstracts that lacked basic information regarding methodology and results were excluded. In addition, case reports, studies that did not specify the histological type of EC specimens being tested and articles that did not stipulate the net number of HPV-positive ESCC tissues detected were not included. From a total of 165 studies investigating ESCC tissue for HPV, 26 were omitted from this review according to the exclusion criteria. Key journals in clinical and surgical oncology and pathology, cancer epidemiology and virology, were hand-searched to identify any articles that were not electronically indexed. Reference lists of the relevant articles were also searched.

Case-control study analysis

The risk of HPV as a factor in causing ESCC can be determined most practically by using a case-control methodology[8] to ascertain HPV rates in ESCC cases compared to controls without ESCC. Cohort studies could theoretically be used, but would not be feasible due to the long duration of follow up required for the development of ESCC and the difficulty of determining exposure status to HPV in such a study. We identified all case-control studies in the literature that compared HPV rates in ESCC patients and healthy control subjects without ESCC. Studies that used control tissue adjacent to the ESCC lesion in the same patient were excluded. Epi Info 3.5.3[9] was used to calculate crude odds ratios (OR), 95% confidence intervals (CI) and P-values for all 22 case-control studies, based on the data presented in the articles.

Epidemiology of Escc

EC is more common in males and is the eighth most common malignancy globally, with a worldwide incidence of nearly half a million cases.[10] It is the sixth most common cause of cancer death, estimated to have claimed the lives of 406 000 people in 2008.[11] Of the two main histopathological categories for EC, esophageal adenocarcinoma (EAC) predominates in the Caucasian population and ESCC is more frequent in Asians and Africans.[12] In the last three decades, while the relative incidence rate of EAC in western countries has overtaken that of ESCC, the absolute incidence of SCC has remained stable[13] and of the two subtypes, ESCC still accounts for most esophageal malignancies worldwide.[14]

ESCC (Fig. 1) has a highly variable socio-demographic and geographical distribution with up to a 500-fold variation between high-risk and low-risk regions.[7] On average, the incidence rate of ESCC in most countries is 2·5–5/100 000 cases for males and 1·5–2·5/100 000 for females.[5] Listed amongst the countries of with highest incidence are South Africa, Japan, India and the central Asian esophageal cancer belt incorporating Iran and China. In particular, with incidence rates as high as 246/100 000, the Eastern Cape of South Africa, northern China and the Caspian littoral of Iran have been identified as the most high-risk regions in the world.[10]

Figure 1.

Macroscopic esophageal specimen showing esophageal squamous cell carcinoma, supplied by Professor Philip Crowe, Department of Surgery, Prince of Wales Hospital, NSW, Australia.

This regional variation may be explained by the multifactorial etiology of ESCC (Fig. 2). Excess alcohol consumption and tobacco use are well-established risk factors.[15] A diet high in processed and red meat and certain preservatives (such as lye), and low in fresh fruit and vegetables, correlating with nutritional deficiencies of vitamins A, B, C and certain trace elements, has also been implicated.[16] Other proposed etiologies include exposure to radiation and certain industrial chemicals such as perchlorethylene, combustion products and asbestos,[17] and the ingestion of hot food and beverages, suggesting thermal injury as a mechanism.[18] Finally, a growing body of evidence attributes infectious agents as causal factors in ESCC, either acting directly on oncogenes or by aiding carcinogenic mechanisms. Mycotoxins with tumorigenic properties and fungal toxins capable of nitrosamine production in food have been linked to ESCC. While Helicobacter pylori has been credited as a protective agent against ESCC,[19] other bacteria are thought to possibly produce carcinogens capable of contributing to the disease process. Cytomegalovirus, Epstein–Barr virus and herpes simplex virus have all been implicated in ESCC,[20-22] although none have been proven conclusively nor researched as widely as HPV.

Figure 2.

Established and postulated risk factors in the etiology of esophageal squamous cell carcinoma (ESCC).

HPV

In addition to benign mucocutaneous diseases, it is now well established that this diverse family of viruses contain several oncogenic genotypes of HPV responsible for many anogenital and aerodigestive cancers.[23, 24] The recent development and licensure of the prophylactic HPV vaccines, Gardasil (Merck Sharp Dohme, Rahway, NJ, USA,) and Cervarix (GlaxoSmithKline, Brentford, UK), with wide coverage in the appropriate population have the potential to make a major public health impact on cervical cancer, as well as other HPV-related cancers.[25]

Possible modes of HPV transmission in ESCC

The mode of transmission of HPV to the esophageal mucosa is still unknown, although several theories have been proposed. The increasing proportion of HPV-related oropharyngeal cancers worldwide has been ascribed to rising numbers of sexual partners, earlier age at first sexual encounter and the increasing practice of oral sex.[26, 27] As the esophageal mucosa is continuous with the oropharyngeal mucosa, it is important to consider the possibility of a similar mode of transmission for HPV in esophageal malignancy.

Additionally, various reports have postulated intrapartum infection including transplacental transmission of HPV in utero as well as during the passage of the baby through the birth canal of infected mothers.[28] While these premises have been supported by findings of HPV DNA in amniotic fluid, HPV does not cause a viremia. Others suggest that as HPV can be found on the foreskin, oropharynx and esophageal cavities of neonates born to women testing positive for HPV infection, this could represent transmission from mother to baby during the delivery process.[29] There are no data regarding the length of time that the virus persists in these cases, as many organisms colonize an infant at birth but disappear quickly. We would assume that persistent infection with oncogenic HPVs would be more relevant.

Furthermore, these hypothesized modes of transmission are supported by the fact that most studies to date have isolated HPV types (6, 11, 16 and 18) usually associated with genital tract disease processes from the esophageal tissue of newborns and patients with esophageal malignancy.[30]

Mechanisms of oncogenesis

A rare outcome following HPV infection may include the development of malignancy. Papillomaviruses have several different carcinogenic mechanisms. During the integration of the viral genome into the host DNA, a break at the level of the E1 and E2 region results in deletion or interruption to the E2 gene. The loss of E2 in the incorporation process results in its corresponding loss of control over the E6 and E7 genes.[31, 32]

For all oncogenic HPV, the E6 and E7 gene products play a key role in the immortalization and malignant transformation of the infected host cell. pRB is a negative regulator of the cell cycle and associates with the E2F family of transcription factors to prevent S-phase entry. Binding of E7 to pRB results in E2F being displaced, allowing the production of proteins necessary for DNA replication. Furthermore, the expression of E7 and the disruption of the pRB/E2F complex allows the expression of cyclin E, a vital factor in S-phase entry. In addition, E6 interacts with p53, which is responsible for promotion of apoptosis of cells with damaged DNA. The E6/p53 interaction results in proteolytic degradation of the tumor suppressor gene. Chromosomal instability, accumulation of oncogenic mutations and combined loss of cell cycle checkpoints lead to established malignancy.

HPV testing methods

The lack of conventional viral culture methods has resulted in diagnosis and typing for HPV to be made using nucleic acid testing methods, particularly amplification technology such as polymerase chain reaction (PCR). This has allowed the detection of low-level HPV copy numbers in clinical samples. The sensitivity and specificity of PCR-based methods can vary depending on the DNA extraction procedures, the site and type of clinical sample, sample transport and storage, primer sets, the amplicon size, reaction conditions including performance of the DNA polymerase, the spectrum of HPV DNA amplified and the ability to detect multiple types. Most detection methods utilize PCR assays targeting the conserved L1 gene and hence are able to detect all mucosal HPV types. Amplification using various primer sets results in different size amplification products and this can result in varying sensitivity for the detection of certain HPV genotypes. PCR assays can also be greatly affected by various inhibitory substances in a clinical sample. Most laboratories incorporate the amplification of an internal control, such as the beta globin gene (present at one copy per human cell) in each PCR to monitor potential inhibition and sample integrity. Assessing sample integrity is especially important in paraffin-embedded archival tissue, where there may be DNA degradation. Inter-laboratory comparisons of PCR results from various studies are difficult as there are no standardized methodologies for doing this. However, a number of commercial HPV detection and typing assays have been released that will assist in inter-laboratory comparisons. Furthermore, standardization through use of international standard reagents as well as proficiency panels currently available through the World Health Organization will allow laboratories to validate their own assays and determine their analytical sensitivity.[33, 34]

As direct probe hybridization and in situ hybridization techniques are less sensitive, amplification methods are currently the most sensitive method for determining the presence of HPV genotypes. Due to the higher likelihood of DNA degradation in archival tissue, either fresh frozen tissue samples or, alternatively, methods directed at the amplification of smaller conserved regions should be utilized.

Evidence for HPV In EC

Animal models

The injection of bovine papillomavirus 4 (BPV-4) into the upper alimentary canals of cattle has demonstrated papillomatosis and the malignant transformation of normal tissue.[35, 36] While BPV-4 DNA has been identified in bovine papillomas, it has not been isolated from malignant lesions, suggesting that the virus is primarily associated with the initial steps of malignancy[37] and its ongoing involvement is not necessary to maintain carcinogenic mechanisms.[35, 38] Bracken fern ingested by cattle is also shown to be a powerful co-factor in esophageal tumorigenesis.[35]

Esophageal squamous cell papilloma

Over 30 studies have investigated the role of HPV in ESCP. In contrast to genital, respiratory tract and oral papillomas, where HPV 6 and 11 are the most commonly detected genotypes, HPV 16 and 18 are more frequently reported in ESCP.[39, 40] The evidence for HPV involvement in these benign lesions is inconclusive due to the highly variable detection rates of HPV DNA within the lesions. In addition, as ESCP is a relatively rare condition there has not been the opportunity to study large numbers of these tumors. However, several reports have suggested the potential for HPV to spread from respiratory papillomas to esophageal mucosa.[41, 42] Therefore, it is important to research further the etiology of HPV in ESCP. Analysis of ESCP tissue for transcriptionally active HPV or the detection of HPV DNA within papillomatous cells using laser micro-dissection would assist in assigning causality.

Serological studies

Virus-like particles are important in detecting HPV-specific antibodies in the sera of patients with esophageal and other malignancies.[43] However, results from serological studies have been variable, with reports of higher HPV16-specific antibody titers in controls than in the ESCC cohort.[44] Others have shown a higher risk of ESCC in patients who are HPV 16 sero-positive.[45, 46]

Due to the natural history and life cycle of HPV, only around 50% of those who become infected with HPV make a measurable immune response. Even then, serological response is slow (within 12 to 18 months) and type specific.[47] Consequently, sero-surveillance is a poorly sensitive marker of past or current natural infection. Furthermore, HPV-specific antibody detection does not provide information on the site of HPV infection and therefore does not necessarily correlate specifically with the risk of EC per se. Therefore, serology is a flawed method for predicting ESCC risk.

In vitro models

Since the loss of p53 function in ESCC was found to be associated with an interaction with the E6 HPV oncogene,[48] a multitude of in vitro studies have reported the mechanisms of oncogenesis of HPV in ESCC. p53 mutations have been identified in pre-cancer lesions[49] as well as HPV-negative ESCC tumors, implying that environmental factors may play an important part in the etiology of ESCC.[48, 50] Several cell lines, such as the synergistically infected human embryonic esophageal cells and immortal cell lines created through interaction of HPV16 E6 and E7 with esophageal cells, are being used to further investigate the molecular basis of HPV carcinogenesis in ESCC.

Morphology

HPV involvement in the tumorigenesis of ESCC was initially proposed following the observation of 24/60 ESCC samples with morphological changes similar to those observed in condylomas.[51] Ensuing studies of South African patients reported HPV-related changes in 33[3] and 65%[52] of ESCC specimens examined. Overall, 25/51 Chinese ESCC patients from Linxian in eastern China, demonstrated morphological changes consistent with HPV infection,[53] and Matos et al. reported 2/22 ESCC specimens with HPV-related changes.[54] These observations were key in triggering refinement of techniques subsequently used to detect HPV in patients with ESCC.

HPV DNA detection in esophageal SCC tissue

Since the early 1980s various techniques have been employed to detect the presence of HPV DNA in esophageal pre-cancer lesions and carcinoma specimens, with mixed results. Detection of HPV DNA in ESCC tissue is perhaps the most accurate measure of HPV involvement in this malignancy. Table 1 summarizes the findings to date for high-risk countries and Table 2 for low-risk to medium-risk countries.

Table 1. Evidence of HPV DNA in esophageal squamous cell carcinoma tissue in high-risk countries
YearsCountryDetection methodHPV types detectedHPV DNA detection range from referenced studies (% and average)Studies (n)Ref.
  1. Ag, HPV antigens; CP, consensus primers; FISH, filter in situ hybridization in situ hybridization; GP, general primers; HB, histological biopsy; IHC, immunohistochemistry; HPV, human papillomavirus; IS-PCR, in situ PCR; ISH, in situ hybridization; PCR, polymerase chain reaction; SB, Southern blot hybridization; SSCP, single-strand conformational polymorphism.
1989–2011ChinaIHC, HB, FISH, ISH, PCR, IS-PCR, SBCP, GP, Ag, E6, 6, 11, 16, 18, 30, 31, 45, 51, 52, 53, 58, 66, 73, 890–100 (33)47[4, 30, 53, 55-98]
2001–2011IranPCRGP, 6, 16, 18, 31, 33, 45, 51, 52, 58, 669–49 (34)7[99-105]
1986–2007South AfricaIHC, HB, PCR, ISH,Ag, GP, CP, 6, 7, 11, 16, 18, 25, 30, 31, 33, 37, 39, 5210–67 (37)10[3, 52, 62, 104, 106-111]
1989–2011JapanIHC, PCR, ISH, IHCAg, CP, 6, 11, 16, 18, 31, 330–63 (18)18[4, 91, 96, 104, 112-125]
2005KenyaPCR0 (0)1[126]
2001–2009IndiaPCRCP, 16, 1819–74 (34)4[127-130]
1990, 1997Hong KongSlot blot, PCRCP, SSCP0–9 (6)2[131, 132]
1999TaiwanPCR, ISH, IHCCP3.2 (3.2)1[133]
Table 2. Evidence of HPV DNA in esophageal squamous cell carcinoma tissue in low to medium risk countries
YearsCountryDetection methodHPV types detected% HPV DNA detection range (and average) from referenced studiesStudies (n)Ref.
  1. Ag, HPV antigens; CP, consensus primers; FISH, filter in situ hybridization; HB, histological biopsy; HCII, hybrid capture 2; HISTOFISH, FISH used to detect the presence of HPV DNA in formalin-fixed, paraffin-embedded biopsies; HPV, human papillomavirus; IHC, immunohistochemistry; ISH, in situ hybridization; LR, low-risk types; PCR, polymerase chain reaction; RFLP, restriction fragment length polymorphism; SB, Southern blot hybridization.
1986–2010AustraliaFISH, HISTOFISH, PCR6, 11, 13, 16, 18, 353·6–50 (8)3[134-136]
1992–1996FranceISH, dot blot, PCR6, 11, 16, 18, 31, 330–33 (5)3[104, 137, 138]
1993–1997UKPCR, ISHNil0 (0)2[139, 140]
2002TurkeyPCR16,1833 (33)1[141]
1995, 1997The NetherlandsPCRNil0 (0)2[142, 143]
2003–2007GermanyPCR6, 11, 16, 18, 20, 27, 870–67 (24)3[144-146]
1993, 1998SloveniaPCRCP, 6, 16, 180–10 (1)2[147, 148]
2000BelgiumPCRCP2 (2)1[149]
1994, 2006SwedenPCR160–16 (1)2[150, 151]
2005GreecePCR16, 18, NS56 (56)1[152]
1995PortugalPCR16, 1856 (56)1[153]
2006HungaryPCR, SB16, 1823 (23)1[154]
1997–2009ItalyPCRCP, RFLP0–47 (14)4[155-158]
1982, 1993FinlandHB, ISH18–40 (29)2[51, 159]
2003, 2006BrazilHCII, PCRLR, 16, 182·5–16 (13)2[160, 161]
2009MexicoPCR16, 18, 5925 (25)1[162, 163]
2006ColumbiaPCR, SB16, 180–34 (30)2[164, 165]
2006ChilePCR, SB1619 (19)1 [164]
2005EgyptPCRAg, 11, 16, 1857 (57)1 [166]
1991–2011KoreaPCR, ISH16, 180–67 (15)4 [91, 167-169]
1989–2011USAISH, IHC, PCR, Reverse lineCP, RFLP, 6, 11,16, 18, 31, 33, 35, 39, 51, 730–45 (10)10 [66, 104, 170-177]

HPV in Regions at High-Risk for ESCC

The three countries with the highest ESCC incidence rates are China, Iran and South Africa.

China

The prevalence of EC varies greatly in different parts of China. Some of the major endemic regions are the northern Jiangsu province, Linxian in Henan Province in the east of the country,[178] Cixian in Hebei and Yangcheng in Shanxi. The national average incidence rate for EC in China is 13/100 000[179]and ESCC represents more than 99% of EC.[180]

Collectively, 47 studies from China analyzing a total of 5035 ESCC samples have meant that the Chinese population has contributed the largest number of ESCC specimens for HPV analysis. Of these, only three studies did not detect HPV in any ESCC samples.[59, 64, 98]

Thirty-five studies utilized PCR with HPV detection rates varying widely from 0–94%.[30, 55, 57, 58, 60-66, 68-78, 80-82, 86, 88-94, 96, 98] Koshiol et al. found 3/267 ESCC samples testing positive for HPV.[68] This result is particularly interesting as the study was carried out in Linxian, a region which harbors one of the highest incidence rates of EC in the world, and is therefore not in keeping with the generally observed trends of high HPV detection rates in high-risk populations.

Thirteen studies used in situ hybridization (ISH) with HPV detection in ESCC varying from 0–72%.[53, 56, 59, 63, 67, 70, 77, 79, 84, 86, 87, 91, 97] This subset of articles includes the largest study to ever be carried out on this topic, analyzing a total of 700 ESCC samples from Henan Province.[56] Again of note, the relatively low HPV detection rate of 17% was unexpected from an area of such high ESCC incidence. Six studies used immunohistochemistry (IHC) to analyse ESCC tissue samples with HPV detection rates varying from 18–89%.[4, 67, 71, 77, 83, 97] Other techniques including filter in situ hybridization (FISH), Southern blot hybridization and the examination of histological biopsies have also been used in China.[30, 60, 64, 82, 85, 181]

Iran

With incidence rates of more than 100/100 000 patients per annum, the Caspian littoral is notorious for ESCC.[182] Seven studies have been carried out in Iran to investigate the link between HPV and ESCC.[99-105] HPV was detected in 9–49% of samples with HPV 16; the most common genotype detected in a total of 476 ESCC biopsies.

South Africa

South Africa has an annual EC incidence rate of 23·5 and 12·6/100 000 for men and women, respectively. Ten studies from South Africa have tested a total of 454 EC for the presence of HPV, with HPV detection rates varying from 10–67%. In 1986 Hille et al. were the first to use IHC to demonstrate HPV antigens in 7/70 (10%) EC specimens.[3] A subsequent report on histology of ESCC biopsies found 13/20 (65%) with morphological features of HPV infection.[52] Six studies used PCR with the detection of E6 protein, general primers and a variety of HPV genotypes including 11, 16, 39 and 52.[62, 104, 107, 108, 110, 111] Two further studies in the early 1990s used ISH techniques to demonstrate presence of HPV in 30 and 52% of ESCC patients studied, respectively.[106, 109]

Case-control studies

We summarized the results of all available case-control studies in the literature (Table 3). Only 22 case-control studies have been carried out internationally, comparing HPV status in ESCC tissue with esophageal tissue obtained from normal, healthy controls. These are all very small studies, the largest of which tested 265 cases and 357 controls. Two studies found no evidence of HPV in either the ESCC patients or the control groups.[139, 168] Most investigations have shown that HPV detection rates were higher in ESCC cases than the control group, although three studies reported a higher percentage of control tissue with HPV infection relative to ESCC samples.[59, 65, 161]

Table 3. Case-control studies examining HPV DNA in esophageal squamous cell carcinoma
YearCountryMethodHPV typesPositive patients (n, %)Positive controls (n, %)Crude OR (95% confidence interval) P-valueRef
  1. CP, consensus primers; HCII, hybrid capture 2; HR, high-risk HPV types; IHC, immunohistochemistry; ISH, in situ hybridization; LR, low-risk HPV types; PCR, polymerase chain reaction.
1991S. AfricaPCRVarious6/14 (43)6/41 (15)4.38 (0.92–21.53)0.0272 [110]
1994S. AfricaISH6, 11, 18, 31, 3325/48 (52)0/2 (0)Incalculable0.1489 [106]
1998JapanPCRCP, 16, 1817/27 (63)3/12 (25)5.1 (0.92–31.48)0.0286 [116]
1998IndiaISH, IHC16, 1819/30 (63)2/10 (20)6.91 (1.04–57.92)0.0175 [127]
2005IranPCR16, 188/38 (21)5/38 (13)1.76 (0.45–7.07)0.3608 [100]
2011IranPCR16, 4515/40 (37.5)5/40 (12.5)4.2 (1.21–15.38)0.0098 [105]
2010AustraliaPCR16, 358/222 (4)0/55 (0)Incalculable0.1531 [134]
1995PortugalPCR16, 189/16 (56)0/10 (0)Incalculable0.0034 [153]
1991USAISH, IHC6, 11, 16, 18, 31, 334/12 (33)0/12 (0)Incalculable0.0285 [176]
2008KoreaPCR160/102 (0)0/40 (0)IncalculableIncalculable [168]
2000BelgiumPCRCP1/21 (2)0/5 (0)Incalculable0.6188 [149]
1992FranceISH6, 11, 16, 18, 31, 334/12 (33)1/24 (4)11.5 (0.93–316.43)0.0171 [138]
1993UKISH6, 11, 16, 18, 31, 330/4 (0)0/10 (0)IncalculableIncalculable [139]
2003BrazilHCIIHR & LR1/40 (2.5)1/10 (10)0.23 (0.01–9.48)0.2790 [161]
2006BrazilPCR16, 1826/165 (16)0/26 (0)Incalculable0.0294 [183]
2005GreecePCR16, 18, other17/30 (56)6/27 (22)4.58 (1.26–17.39)0.0081 [152]
2001ChinaPCR, ISH162/2 (100)66/112 (59)Incalculable0.2406 [63]
2001ChinaPCRCP2/32 (6)4/57 (7)0.88 (0.1–6.13)0.8898 [65]
2005ChinaPCR16, 18207/265 (78)203/357 (57)2.71 (1.86–3.94)<0.0001 [55]
2006ChinaISHnil0/4 (0)61/475 (13)0.00 (0.00–10.68)0.4429 [59]
2010ChinaPCR1635/69 (51)2/32 (6)15.44 (3.2–101.46)<0.0001 [90]
2010ChinaPCR16, 18, 5835/70 (50)20/60 (33)2.00 (0.92–4.35)0.0552 [92]

A univariate crude OR with 95% CI and P-values for the risk of ESCC associated with HPV was calculated for each of the 22 case control studies, using Epi Info software.[9] Nine studies had OR which showed significant association between HPV and ESCC.[55, 90, 92, 105, 110, 116, 127, 138, 152] For the largest single case-control study, the OR was 2.71 (95% CI 1.86–3.94), P < 0.001.[55] A meta-analysis of case-control studies on this topic to date, in conjunction with single large case-control studies, would best determine the etiology of HPV in ESCC.

HPV Vaccines and Possible Implications for ESCC

Two HPV vaccines are currently available and licensed on the market. Gardasil (Merck Sharp Dohme, Rahway. NJ, USA) is a quadrivalent vaccine which immunizes against infection by types 6 and 11 (associated with over 90% of cases of genital warts), as well as types 16 and 18, which are responsible for causing over 70% of cervical cancers.[184] Cervarix (GlaxoSmithKline, Brentford, UK) is bivalent, targeting the two commonest oncogenic types 16 and 18.[185] The vaccines are prophylactic and have no therapeutic effect on women who have already been infected with HPV.

Persistent HPV infection appears to be necessary for the progression of precancerous cervical lesions to invasive carcinoma. In randomized control trials the HPV vaccines have efficacy rates as high as 90–100% for preventing the development of cervical cancer in women who are HPV sero-negative and HPV DNA negative at baseline.[25, 185, 186]

Currently, vaccination programs are targeted to young girls prior to their sexual debut. While the etiology of HPV in ESCC remains unclear, growing evidence for the role of HPV in non-cervical cancers such as head and neck malignancies supports the rationale for vaccination of men.[187] At present, there are insufficient data to suggest that HPV vaccination could have a significant impact in preventing ESCC. However, if a HPV-EC link does exist, understanding the role of HPV in EC would inform estimates of the potential global impact of a vaccination program that also targets men. In the world's most populous country, China, where ESCC is the fourth leading malignancy, even if HPV is associated with only 25% of EC, vaccination could have a significant public health impact.

Conclusion

Since the last review on HPV and ESCC[5] was published, the introduction of prophylactic HPV vaccines has made it even more important to determine whether HPV plays a significant role in the etiology of ESCC. The research question whether HPV is etiologically linked to ESCC can be answered only by comparing the rates of HPV in ESCC cases with the rates of HPV in healthy control esophageal tissue.

Syrjänen's review[5] discussed articles from 1982 to 2001. We have identified a further 62 studies that have been carried out up the time of writing, in an attempt to determine the relationship between HPV and ESCC. Despite an increase in the body of studies on the subject, there are still very few case-control studies, which are the appropriate study design to address this question. Most published studies are case series of ESCC only, which cannot answer the question whether HPV is associated with ESCC. These case series are of limited value in clarifying the research question. The available evidence therefore remains inconclusive about a definitive link with HPV as a causative factor of ESCC. The heterogeneity and small scale of case-control studies carried out so far are significant limitations to assessing a potential HPV-ESCC link. The single largest of these case-control studies by Cao et al.,[55] when analyzed, did show a risk of HPV 2.7-fold greater in cases of ESCC than in controls. There is a need for large-scale case-control studies using uniform testing methodology to answer the question more definitively. Furthermore, we hope to carry out a meta-analysis of all case-control studies, which will provide a basis for analyzing the current evidence for an HPV-ESCC link.

Among the larger body of case series and other study designs we identified, there were studies showing both negative and positive results for the detection of HPV in ESCC. The accumulation of such case series does not add clarity, and the highly divergent results may be due to several factors including; (i) diverse testing methodology on tissue (PCR, IHC, ISH, histological biopsy, general primers, consensus primers, hybrid capture) or blood (serology); (ii) inter-laboratory discrepancies with similar testing methods; (iii) the specimens examined (ESCC tissue from surgical resection or diagnostic biopsy, balloon cell samples, serology); (iv) the techniques used for categorization of histopathology (endoscopic biopsies, iodine staining, cytology); (v) variation in sections of the resection specimens tested; and (vi) the presence of co-factors that vary by geographical region and that may act as co-factors promoting HPV to lead to ESCC. In addition, control tissue has been poorly studied for the presence of HPV with many studies classifying histologically normal esophageal tissue from resection specimens of ESCC as “controls”. Gao et al. suggest that the relationship between HPV and ESCC is most likely weaker than that between HPV and other malignancies such as cervical cancer, since such marked variation in studies of HPV and other cancers has not been reported.[59] This variation supports the role of co-factors, whether they are dietary, environmental or lifestyle, which vary geographically and act with HPV to cause ESCC. More research is required to elucidate what these co-factors might be.

The apparent multifactorial etiology of EC has made it difficult to clearly identify causes for the undisputed geographical variation in the incidence of this disease. Certainly, a combination of smoking, alcohol, dietary and other lifestyle factors, as well as infectious agents, may contribute to the complex natural history of EC.

In general, studies conducted in high-risk regions for ESCC incidence, have reported higher rates of HPV in ESCC specimens than studies that are carried out in low-risk regions. Many of the most recent studies continue to support this trend.[92, 105, 107, 134, 175, 188] However, there have been some exceptions to this pattern. For example, several reports from Japan and China, both high-risk countries, failed to identify HPV in their ESCC samples.[59, 64, 98, 112, 117, 121] Similarly, some studies conducted in populations from low-risk regions such as the USA, Italy, Germany and Portugal have demonstrated relatively high HPV detection rates in ESCC tissue.[145, 153, 155, 177] On the whole, a clear trend exists between HPV detection rates and geographical area corresponding to ESCC incidence rates.

There has been some speculation as to why HPV detection rates are lower in low-risk ESCC incidence areas and accordingly high in high-risk countries. People living in the global regions marked as high-risk incidence areas for ESCC generally, have low socioeeconomic levels. This is in contrast to most of the developed world in which ESCC incidence has declined or remained stable over the last few years. In general, HPV infection rates have also been found to be higher in developing countries than in developed nations.[23] Impoverished parts of the world, in which citizens have low education levels, poor nutrition and limited access to healthcare facilities may account for the higher incidence rate of HPV infection among them.[23, 160] This may, in turn, translate to higher HPV detection rates in ESCC specimens from these regions.

HPV types 16 and 18 have, by far, been the most commonly identified genotypes in ESCC specimens in both low-risk and high-risk settings. This is significant as these particular HPV strains are classified as high risk for their oncogenic potential and are also the most frequently detected in cervical carcinomas and head and neck cancers. In addition, other genital tract-related HPV types such as 6 and 11 have been reported more frequently in ESCC specimens from high-risk regions than low-risk areas.

Some studies have isolated HPV types not previously described in ESCC tissue, such as types 20, 27 and 87 in a German cohort, type 73 in the USA and China, and types 53, 56, 66 and 89 in China. Of note is a study carried out by Chang et al. reporting a finding of HPV type 30 in eight of 85 HPV-positive ESCC samples.[79] As HPV 30 has only previously been detected in laryngeal carcinoma and two genital condylomas,[189] it has been hypothesized that HPV 30 may have a propensity to infect esophageal mucosa.[79]

While many recent, well-designed case-control studies have reported the presence of HPV DNA in ESCC tissue,[90, 92, 105, 107, 188] others have produced conflicting results,[65, 161, 168] making it difficult to conclude that HPV is a definitive causative factor in ESCC. Therefore the promotion of an HPV vaccination for the general population with a view to preventing ESCC cannot be justified. However, there are a small number of case-control studies that appear to support an etiological role of HPV in ESCC, pointing to the need for further, large case-control studies and the meta-analysis of existing case-control studies. If there is any potential for an HPV-ESCC link, the recently introduced HPV vaccines could have further widespread use in the public health domain. This review highlights the need for ongoing investigation in this area.

Acknowledgments

ATN currently holds a National Health and Medical Research Council Training Fellowship (630724 – Australian-based Public Health Fellowship).

Ancillary