We adopted a comprehensive methodology to assess the prevalence, viral load and physical status of infection with HR-HPV-16, -18 and -58, the 3 most common HR-HPVs, in patients with ESCC and CSCC and the expression of HPV-16 in patients with ESCC. We aimed to explore the characteristics of HR-HPV DNA in the carcinogenesis of ESCC compared with CSCC.
Infection with the 3 types of HR-HPV associated with CSCC (79.48%) in the Shantou region was consistent with that in other regions in China, but the infection in controls was higher (18.52% vs. 6.87%) [8, 20]. This increased rate should be interpreted with caution because of the small number of cases in our study. For ESCC patients, the total infection rate for all 3 viruses was greater than that of their controls. Among the HR-HPV types, HPV-16 was the most predominant in all esophageal tissues. Our results for HPV-16 and -18 infection rates in Shantou were similar to those from a previous study (52.4%)  and were higher than those in other regions [22, 23]. Interestingly, this distribution of HR-HPV in esophageal lesions was in agreement with observations for cervical lesions .
Over the past 20 years, reported HPV infection rates have varied in ESCC patients from geographically different areas , but are consistently high in CSCC patients . This significant difference may result from HPV DNA copy numbers in host tissues. Evidence from our study revealed that the average viral load for HR-HPV for CSCC patients was significantly higher than that for ESCC patients in Shantou region; the mean viral load for HPV-16 was 2.55 ± 3.19 copies/cell in ESCC tissues. This low figure was also found in a Chinese study of EC tissues (<1 to 157 copies/cell) [13, 26] and a Finnish study of head and neck SCCs (4.6 to 49 copies/cell) . Extremely low levels of HPV DNA load (even 0.01 copies/cells) in ESCC tissues may exceed the detection ability of some techniques such as regular PCR and conventional in situ hybridization (ISH), which limits detection to <10-50 copies/cell . Our positive results for <10 copies of HPV DNA would be out of the detection range with use of general PCR and ISH technology. This low detection rate might explain why we failed to measure the level of HPV mRNA in our paraffin-embedded tissue by RT-PCR and the low detection rates we found with regular PCR in our previous study. This low HPV infection load was not correlated with clinical characteristics such as age, cancer grade or lymph-node metastasis of ESCC patients and their controls. However, increased levels of all HR-HPVs with increasing cancerous grade indicated a dose-dependent association of viral load and cancer grade.
Epidemiologic and molecular studies have shown that HR-HPV DNA integration is considered a prerequisite for the development of several malignant lesions [29, 30]. For the integration level of viral DNA, we found no differences between CSCC and ESCC lesions for the level of integration of HPV-16, -18 and -58; most infected CSCC or ESCC control subjects carried the episomal form of the virus, whereas the mixed and integrated forms were predominant in ESCC and invasive cervical carcinomas. For ESCC patients, specifically, we found both ratios of HPV DNA integration and expression of HPV-16 protein increased with ESCC grade. The integration occurs in the early stage of esophageal carcinogenesis and was associated with the severity of the cancer. So the increased expression of the E6 and E7 oncoproteins in ESCC might result from disrupting the E2 open reading frame when HPV integrates into the host genome [31–33]. Although the direct evidence of HPV-16 transcription and expression derived from integrated HPV DNA is still unclear, from previous studies, HPV E6- and E7-encoding cDNAs derived from integrated viral oncogene transcripts have a much stronger transforming capacity in primary cells than cDNAs derived from episome-derived transcripts . The relative expression levels of the viral oncogenes and their corresponding gene products appear to be directly influenced by the sequence context of individual integration sites in cervix, head, neck, penile and tonsil carcinogenesis [34, 35]. So the potential connections among HPV DNA integration, transcription and expression, as well as its clinical value, need to be confirmed with more studies.
Our data confirmed for CSCC that the high prevalence, viral loads and integration rates of HR-HPV were the most important risk factors for cervical carcinogenesis. We also found relatively low HPV-16, -18 and -58 viral load in ESCC patients, which is consistent with the relatively low HR-HPV DNA viral load found in some cervical carcinoma cell lines such as SiHa cells (1-2 copies/cell for HPV-16) and HeLa cells (10-50 copies/cell for HPV-18). However, the low viral load, particularly with a common integrated viral genome and high expression of its onco-protein, may be sufficient to lead to or promote carcinogenesis, especially because of the high incidence of EC.
Our study contains some limitations. The selection bias cannot be avoided because of the relatively small number of samples collected, which may reduce the representativeness of our results and capacity for inference. Heterogeneity between ESCC and CSCC patient ages also existed because of the small number of samples in these groups.