Detection of HPV DNA in esophageal cancer specimens from different regions and ethnic groups: a descriptive study

Esophageal carcinoma is a common malignancy and its mortality rate is among the highest for cancers overall [1]. There is extreme geographic variation in the incidence of esophageal carcinoma, with differences of as much as 300-fold between areas of greatest and least prevalence within a given country, or between one country and another [2]. Substantial alcohol use combined with smoking has been found to greatly increase the risk of esophageal cancer in western countries [3, 4]. In contrast, vitamin deficient diets and food containing potential carcinogens have been suggested to be significant risk factors for esophageal cancer in China and other central Asian countries [5‐9]. However, the association between esophageal cancer and these identified risk factors is generally weak for the high risk population of China, implying that the major etiology remains to be determined [10, 11].

In 1982, Syrjanen found that HPV infection caused pathological lesions in esophageal cancer specimens [12]. Since then many studies which have used a variety of techniques have evaluated the association between HPV and risk of esophageal cancer. However, the results of these studies have been inconsistent, and it has been suggested that this inconsistency may result in part from geographic variation in esophageal cancer prevalence. Most studies that have failed to detect HPV DNA in esophageal tumors were conducted in low-risk areas such as the USA or Europe [13, 14]. Studies in high-risk areas have found a significantly higher percentage of HPV in esophageal cancer [15‐18].

In this study, we evaluated tissues from 435 esophageal cancers collected from four different regions with multiple ethnicities including Han, Kazakh, American Japanese, American Caucasian, and African American for HPV DNA. We found comparable rates of HPV detection and similar HPV genotypes in the esophageal cancer samples from all of these geographic areas.

Over the past twenty years, many reports regarding HPV infection in esophageal cancer have been published, and the reported HPV detection rate in these tumors has ranged from 0 to 67% [15, 17, 20‐23]. To explain these marked differences in the reported detection rate, sampling methods, demographic and ethnic factors, disease status, and sensitivity of detection methods have been cited as potential causes of inconsistency. HPV was detected infrequently in esophageal carcinomas collected from western countries, whereas it was more frequently detected in esophageal cancers from regions of high HPV incidence, particularly in China and South Africa. It therefore has been proposed that HPV infection may play a role in esophageal carcinogenesis only in high incidence regions [24, 25]. In contrast to previous reports, our results demonstrated the prevalence of HPV infection in esophageal cancer is independent of geographic area and ethnicity, suggesting HPV infection is commonly associated with esophageal cancer and raising the possibility that it may play a role in the etiology of esophageal carcinoma.

The differences in prevalence of HPV in esophageal carcinoma in this study and previous reported prevalence may have resulted in part from use of different analytic strategies. We recently demonstrated two ways in which HPV infection in the esophagus differs from the infection observed in cancers of the uterine cervix (data not shown). First, real time PCR revealed that the HPV load in esophageal specimens was at least two orders of magnitude lower than that in cervical cancer. In addition, HPV genotypes found in esophageal specimens were different from those found in cervical cancers. Based on these observations, we proposed in a previous study that the low HPV copy number might therefore result in a falsely low detection rate in esophageal cancer. To provide evidence for this line of reasoning, a series of HPV 16 positive cervical cancer specimens were employed to test a mimic esophageal cancer model. The copy number of HPV/cell in these cervical cancers was first determined using real time PCR. The samples were then diluted to 0.1-0.01/cell which was similar to the average copy number in esophageal cancers and tested for L1. Of 24 diluted samples, 6 were negative in the first round of PCR. An additional round of PCR was carried out, and all six samples which were negative in the first PCR analysis were positive, indicating that repeat testing may decrease the false negative rate in samples with low HPV copy numbers. On this basis, an analytic strategy for HPV detection in esophageal cancers was formulated.

Briefly, L1 detection was performed twice in each esophageal sample. L1 positivity in either test was counted as positive for calculation of the detection rate. A batched TA cloning assay was used to evaluate the HPV genotypes in the sample population. Utilizing this approach, HPV57 which is not common in cervical cancers was identifiable in esophageal specimens. Genotype specific E6 or E7 PCR was performed to determine the genotypes in L1 positives.

It is of interest to note that HPV57 presented as the second most common genotype and accounted for 22.9% of L1 positive esophageal cancer tissues from all four areas under investigation. HPV57 was originally identified in benign, premalignant, and malignant lesions of the nasal cavity [26]. The signature pattern of HPV57 E6 and E7 oncogenes was closely related to HPV16 and 18 oncogenes. NIH 3T3 cells expressing HPV57 exhibited morphological transformation indicating that the activity of HPV57 in vitro is comparable to other high-risk HPV types [27]. The role of HPV57 in esophageal carcinogenesis thus warrants further investigation.

The associations between HPV infection and age, gender, tumor TNM stage, tobacco smoking and alcohol use were analyzed. Of these factors only alcohol use was associated with HPV positive esophageal cancers as compared with HPV negative subjects. However, there was no statistical evidence of such relationship when the analysis was limited to US samples only, but the possibility that the limited US sample size might have contributed to this result cannot be excluded. Alternatively, as samples from the US included both esophageal squamous cell carcinomas and adenocarcinomas, it is possible that these two types of cancers may arise from different carcinogenic pathways and thus confound the data. As such, further investigation with a larger sample size should be carried out to further evaluate the effect of alcohol use on HPV infection in the esophagus.

We recognize that there are several limitations in this study, one of which is selection bias. As a hospital-based study, these cancer patients were all recruited from local hospitals and as a result the population from which we obtained these samples cannot be precisely defined. Patient self selection of a hospital for treatment may have also introduced selection bias. These factors may limit the sample's representativeness and capacity for inference. Another limitation is that some crucial information which would potentially allow broader conclusions was not obtained for this study (for example, anatomic location of the tumor in the US samples). In this study we demonstrated for the first time that HPV DNA is both relatively common in esophageal cancer, and is independent of geographic region and ethnicity of tested subjects. In addition, HPV infection appears to be independent of all previously identified risk factors for esophageal cancer except alcohol use.

This observed consistency of the prevalence of HPV infection in esophageal carcinoma from different geographic areas may warrant further investigation with larger numbers of subjects over additional areas. However, the findings here serve to raise the possibility that HPV plays a role in the etiology of esophageal carcinoma.