|Home | About | Journals | Submit | Contact Us | Français|
Certain regions of China have high rates of esophageal squamous cell carcinoma (ESCC). Previous studies of human papillomavirus (HPV), a proposed causal factor, have produced highly variable results. We attempted to evaluate HPV and ESCC more definitively using extreme care to prevent DNA contamination. We collected tissue and serum in China from 272 histopathologically-confirmed ESCC cases with rigorous attention to good molecular biology technique. We tested for HPV DNA in fresh-frozen tumor tissue using PCR with PGMY L1 consensus primers and HPV16 and 18 type-specific E6 and E7 primers, and in formalin-fixed paraffin-embedded tumor tissue using SPF10 L1 primers. In HPV-positive cases, we evaluated p16INK4a overexpression and HPV E6/E7 seropositivity as evidence of carcinogenic HPV activity. β-globin, and thus DNA, was adequate in 98.2% of the frozen tumor tissues (267/272). Of these, 99.6% (95% confidence interval (CI) = 97.9–100.0%) were negative for HPV DNA by PGMY, and 100% (95% CI = 98.6–100%) were negative by HPV16/18 E6/E7 PCR. In the corresponding formalin-fixed paraffin-embedded tumor specimens, 99.3% (95% CI = 97.3–99.9%) were HPV negative by SPF10. By PGMY, 1 case tested weakly positive for HPV89, a noncancer causing HPV type. By SPF10, 2 cases tested weakly positive: 1 for HPV16 and 1 for HPV31. No HPV DNA-positive case had evidence of HPV oncogene activity as measured by p16INK4a overexpression or E6/E7 seropositivity. This study provides the most definitive evidence to date that HPV is not involved in ESCC carcinogenesis in China. HPV DNA contamination cannot be ruled out as an explanation for high HPV prevalence in ESCC tissue studies with less stringent tissue procurement and processing protocols.
Esophageal cancer rates in Linxian, Henan Province, China, are some of the highest in the world.1 All of the esophageal cancer cases are esophageal squamous cell carcinomas (ESCC). No dominant risk factor for ESCC has been identified in this region. The major risk factors for ESCC in Western countries, smoking and alcohol consumption,2 are not strongly associated with ESCC in Linxian.3 The magnitude of other risk factors for ESCC in Linxian also tends to be modest.3–7 Therefore, additional major risk factors may be causing the majority of the cases in this area.
Human papillomavirus (HPV) has been suggested as a risk factor for ESCC. Although over 70 studies have tested for HPV in ESCC tissue,8 the role of HPV in the development of ESCC is still hotly debated.9 The arguments in support of a causal role for HPV in ESCC include: (i) the close proximity and histological similarity of esophageal and oral squamous epithelium,9 where HPV is known to cause tonsillar and oropharyngeal cancers10,11; (ii) the existence of potentially HPV-associated benign esophageal squamous papillomas8,9; (iii) evidence for clonality of esophageal tumor cells9; (iv) the association between HPV and esophageal cancer in bovine animals8,9; (v) HPV detection in esophageal cancer tissues and precursor lesions8,9; and (vi) in vitro evidence of HPV-induced transformation of esophageal cells.8 Arguments against the involvement of HPV in esophageal carcinogenesis include: (i) variation in HPV prevalence in case-series of ESCC from 0 to 100%9; (ii) inconsistent associations between HPV exposure (as measured by serology) and ESCC, with estimates from large studies tending toward the null12; and (iii) lack of documented associations between ESCC and sexual behavior, immunosuppression or previous HPV-associated malignancy.9
Estimates of the prevalence of HPV in esophageal tumors range widely, even within the same country.8 A few possible reasons for this variation between studies include small study sizes, different HPV detection assays, interlaboratory variability, and suboptimal sample collection and handling leading to contamination.12–14 Contamination of specimens in collection, processing and testing can lead to false-positive results.14,15 This potential for contamination is of particular concern for studies of HPV because HPV is a common human infection that can involve a variety of epithelial tissues and because DNA is resistant to cleaning with detergent. 16–18 While true population differences are possible, estimates of the prevalence of HPV DNA in ESCC tumor specimens reported from 2 adjacent high-risk counties (Linxian and Anyang) in Henan Province, China, range from 0% (0/35) to 94% (29/31).19–24 In fact, the extremes of this range both come from Linxian, which suggests that this variation is not due solely to differences in study populations.
We have previously conducted 2 studies of HPV and ESCC in Linxian. One was a prospective study that evaluated the association of seropositivity for antibodies against HPV Types 16, 18, and 73 L1 and L2 viral capsid proteins with risk of ESCC.12 This study found no association. However, L1/L2 serology has limited sensitivity25 and cannot distinguish esophageal exposure to HPV from genital exposure. Furthermore, although HPV16 and 18 account for 70% of cervical cancers and the vast majority of extra-cervical HPV-associated cancers,26 the World Health Organization has classified 13 HPV types as carcinogenic or probably carcinogenic. 27 Thus, a more definitive study may need to evaluate a broader spectrum of HPV types. To that end, we used Hybrid Capture 2, which tests for all 13 carcinogenic HPV types, to determine the prevalence of HPV in esophageal cytology specimens.28 This study found that the prevalence of HPV did not vary by disease severity. However, a previous study of oral cancer patients found poor correlation of HPV detection in matched cytologic and tumor specimens,29 suggesting that a tissue-based study may provide a more definitive evaluation of HPV and ESCC.
As 1 step toward a more definitive evaluation of HPV in ESCC, we collected surgical resection specimens and serum from ESCC cases in Linxian with rigorous attention to prevent DNA contamination. We then tested for HPV DNA using the most sensitive available detection methods and evaluated HPV DNA-positive cases for evidence of an active HPV infection that could have contributed to the development of cancer in that case.
We recruited 272 consecutive incident histopathologically-confirmed ESCC cases who were ≥18 years of age and undergoing surgical resection at Yaocun Commune Hospital in Linxian, China, from mid-October 2006 through March 2007. Yaocun Commune Hospital is a regional referral center that performs over 1,000 resections for esophageal or gastric cancer each year. All subjects gave written informed consent. The protocol was approved by the Institutional Review Boards of the US National Cancer Institute (NCI) and the Cancer Institute of the Chinese Academy of Medical Sciences (CICAMS). Demographic data, current or ever smoking, current or ever drinking alcohol, and family history of cancer were abstracted from medical records.
Specimens were collected and processed using a standardized protocol designed to minimize the possibility of tissue contamination by environmental HPV. Immediately after removal from the vasculature, the unopened esophagus was brought in a sterile pan directly from the operating room to the tissue processing room, where all working surfaces were washed with 20% bleach and 70% ethanol before each case. The esophagectomy specimen was received on a sterile disposable pad and then transferred to a fresh sterile disposable pad for processing the tumor tissue. Using sterile forceps, scissors and scalpels, the esophagus was opened, and adjacent samples of nonnecrotic tumor tissue were cut and placed in prelabeled cryovials containing no preservative (n = 2) and 10% neutral buffered formalin (n = 2). The vials containing no preservative were immediately submersed in liquid nitrogen. The esophagus was then transferred to a fresh sterile disposable pad for processing the nontumor tissue. Using new sterile gloves and new sterile supplies, adjacent samples of macroscopically normal nontumor tissue at least 2 cm from the nearest visible tumor were cut and processed as described above. Between each case, new sterile sleeve covers were used to cover the arms of the technician. In over 90% of the cases, both tumor and non-tumor tissues were snap frozen in liquid nitrogen within 30 min of removal of the esophagus from the vasculature. At the end of each day, all snap frozen cryovials were transferred to a −70°C freezer for storage.
The fixed tissues were processed in new disposable cassettes and embedded in new paraffin using sterilized forceps and embedding molds. A total of 10 paraffin sections were cut from each formalin-fixed ESCC tissue block using a new disposable microtome blade in a strictly clean environment. The 1st and 10th 4-µM sections were cut for hematoxylin and eosin (H&E) staining. The 8th and 9th 4-µM sections were attached to silanated slides for immunohistochemical assays. Two groups of three 8-µM paraffin sections (the 2nd to 4th and 5th to 7th sections) were placed in 2 clean conical tubes for HPV DNA measurement. Finally, an HPV negative quality control block was cut between every 10 ESCC blocks to evaluate potential cross-contamination.
CICAMS and NCI pathologists (DML, MJR, SMD) confirmed the presence of nonnecrotic ESCC in the H&E stained slides of each formalin-fixed tumor block; the same formalin-fixed tumor block was used for the SPF10 assays. Adjacent samples of frozen tissue were used for the PGMY and HPV16 and 18 type-specific E6/E7 PCR assays. To further ensure that the frozen samples tested with PCR contained tumor tissue, we cut flanking cryosections in 64 (22%) of the cases. Nonnecrotic ESCC was present in all of these tissues.
Frozen tissues were homogenized in digestion buffer with a final concentration of 400 µg ml−1 of proteinase K and 0.1% Laureth-12, using a disposable pestle in a sterile 1.5-ml microcentrifuge tube and digested at 55°C 24–48 hr until the tissue was fully dissolved. DNA was purified using phenol : chloroform : isoamyl organic extraction in Qiagen high density MaXtract microcentrifuge tubes. The aqueous layer was removed into a new, sterile microcentrifuge tube using a disposable transfer pipette, and DNA was precipitated with ammonium acetate and ethanol overnight at −20°C. DNA was pelleted by centrifugation at 4°C, washed with 70% ethanol, and resuspended in 100 µl of loTE.
As previously described,30–32 paraffin-embedded formalin-fixed tissue sections were soaked in digestion buffer (250 µl of 1 mg ml−1 proteinase K in 45mM Tris-HCl, pH 8.0, 0.9 mM EDTA, and 0.45% Tween 20) on a 70°C dry block for ~23 hr and then heated to 95°C for 10 min to inactivate the proteinase K. Ten microliters of the digested DNA was diluted 10-fold in DEPC-treated water.
A 50-µl aliquot of the purified DNA from the frozen tissue was tested for HPV DNA at Johns Hopkins University using the Roche HPV Linear Array Assay according to the manufacturer’s instructions. 33 The Roche Linear Array test is based on consensus amplification using PGMY primers and has high sensitivity for the detection of a wide range of HPV types (HPV 6, 11, 16, 18, 26, 31, 33, 35, 39, 40, 42, 45, 51, 52, 53, 54, 55, 56, 58, 59, 61, 62, 64, 66, 67, 68, 69, 70, 71, 72, 73, 81, 82, 82v (IS039), 83, 84, and 89 (CP6108)). Compared to other systems, the Linear Array has an increased ability to differentiate between types and to detect multiple HPV types.34 The β-globin gene was coamplified. Samples positive for β-globin were considered of sufficient quality for analysis.35 Four positive controls, 10 and 100 HPV16-plasmids and 10 and 100 HPV18-plasmids diluted in a background of 50 ng of human placental DNA, were placed directly into PCR. Because of concerns that HPV integration might disrupt or delete the L1 region, leading to false-negative results with L1-based primers, 36,37 all specimens were also tested for the presence of the E6 and E7 oncogenes of HPV16 and HPV18, the 2 major carcinogenic types in cervical and head and neck cancer,38,39 using type-specific TaqMan assays targeting the E6/E7 open reading frame (ORF).40,41
A 10-µl aliquot of the DNA dilution from the formalin-fixed paraffin-embedded tissue specimens was tested for HPV DNA at CICAMS using PCR with SPF10 primers (DDL Diagnostic Laboratories, Netherlands), followed by the reverse line probe assay (LiPA) for genotyping of 25 HPV Types 6, 11, 16, 18, 31, 33, 34, 35, 39, 40, 42, 43, 44, 45, 51, 52, 53, 54, 56, 58, 59, 66, 68, 70, 74.30 This system also targets the L1 region of the HPV genome and is ideal for use in formalin-fixed tissue because of its short amplimer (65 bp).34 DNA extraction controls included HPV16-positive tumor tissue from a SiHa xenograft mouse, HPV18-positive tumor tissue from a HeLa xenograft mouse, and an HPV-negative paraffin section processed in the same manner as the ESCC tumor tissue. PCR controls included HPV18-positive (HeLa) cells at a concentration 10–100 times the limit of detection for LiPA (<10 HPV DNA copies per reaction) and a negative control (DEPC).
At both laboratories, meticulous processes were applied to avoid cross-contamination in all specimen processing and HPV testing procedures. New filter tips, powder-free gloves, and sterile conical tubes were used for each case. The microtome, hood, and biosafety cabinet were also periodically cleaned with 20% bleach and 70% ethanol.
Cases positive for HPV DNA were evaluated for evidence of HPV-induced carcinogenesis by immunohistochemical analysis of p16INK4a overexpression, a well-established marker for HPV oncogene expression.42–44 4-µm tissue sections were stained with a monoclonal antibody E6H4 provided by mtm-laboratories AG (Heidelberg, Germany) according to standard procedures recommended by the supplier. Sections of an HPV-negative colorectal cancer and a HPV 16-positive cervical cancer biopsy were used as negative and positive controls, respectively. In addition, serum from HPV DNA-positive cases was tested for anti-E6 and anti-E7 antibodies (available for HPV types 16 and 18), which indicates the presence of an HPV-associated tumor,45,46 using enzyme-linked immunosorbent assays with glutathione S-transferase fusion proteins provided by Dr. Michael Pawlita (DKFZ, German Cancer Research Center, Germany), as previously described.47,48 Positive controls included known seropositive samples from patients with HPV-associated oral cancers previously tested by Dr. Raphael Viscidi at the Johns Hopkins University.
The prevalence and 2-sided 95% confidence intervals (CIs) of HPV in the ESCC tissue were calculated in STATA 9.2 (StataCorp, College Station, TX). Maps of the distribution of ESCC cases were made using Epi Info 3.5.1 (CDC, USA).
Of the total 272 cases, 10 (3.7%) came from Linxian, 65 (23.9%) came from other counties or cities in the high-risk Taihang mountain region of which Linxian is a part, and 197 (72.4%) came from outside of this area (Figs. 1a and 1b). The median age was 60 years (range: 38–79). More than two-thirds of the cases were male, and nearly all were married (Table 1). One hundred and sixty (58.8%) of the patients were ever smokers. All but 2 of those were males. Similarly, 110 (40.4%) of the patients were ever drinkers, all of whom were male. Overall, 102 (37.5%) of the patients, 83 of whom were female, never smoked nor drank. Sixty-seven patients (24.6%) had a family history of cancer, including 55 (20.2%) with relatives who had esophageal cancer.
The frozen tissue produced strong β-globin signals for 95.6% of cases (N = 260), weak β-globin signals for 2.6% (N = 7), and no β-globin signal for 1.8% (N = 5). Thus, β-globin was adequate in 98.2% of cases (267/272), which demonstrates the adequacy of the DNA in these tissues.
Among the 267 cases with adequate β-globin, 266 (99.6%, 95% CI: 97.9–100.0%) were negative for HPV DNA by PCR using the PGMY consensus primers and Linear Array. One male case tested weakly positive for HPV89 (CP6108), a non-cancer causing HPV type. No cases were positive by HPV16 or HPV18 E6 or E7-based PCR (100.0% negative, 95% CI: 98.6–100.0%).
Using PCR with SPF10 primers followed by LiPA genotyping, 265/267 cases (99.3%, 95% CI: 97.3–99.9%) were negative for HPV DNA. Only 2 cases tested weakly positive: 1 female for HPV16 and 1 male for HPV31. All of the HPV negative quality control blocks tested negative.
None of the tissues were HPV DNA-positive in both laboratories. Of the 3 HPV DNA-positive cases that were positive in 1 laboratory, none exhibited p16INK4a protein overexpression. In addition, the HPV16-positive case was seronegative for HPV16 E6 and E7 antibodies.
In this large study, we found little evidence of HPV involvement in the etiology of ESCC. All but 3 cases were HPV negative by PCR using 2 very sensitive consensus L1 detection primer systems and HPV16-and HPV18-specific E6 and E7 primers. Each of the 3 positive cases was weakly positive in 1 L1 PCR assay and negative in the other L1 PCR assay, and none of them produced any evidence of HPV oncogene activity as indicated by HPV oncogene expression (p16INK4a protein overexpression) or E6/E7 seropositivity. These results support our 2 previous serology-and cytology-based studies in Linxian, which also suggested that HPV is not a significant cause of ESCC in this population.12,28
Every case in this analysis contained histopathologically confirmed, nonnecrotic ESCC. To address concerns over the potential loss of L1 through viral integration, we complemented the PGMY L1 and SPF10 L1 PCR assays, which were performed at 2 independent institutions and are highly sensitive for a broad spectrum of HPV types,34 with additional PCR targeted to the E6 and E7 regions of HPV16 and HPV18, the 2 major carcinogenic types in cervical and head and neck cancers. Thus, our study is unlikely to have produced false negatives due to lack of tumor tissue or primer sensitivity.
One feature of this protocol that may be different from the HPV studies previously performed in the high-risk region around Linxian is the extreme care we took to prevent DNA contamination throughout specimen collection, processing, and testing. This caution was prompted by a suspicion that the variability in previous results from this area might be related to environmental contamination during these successive steps.14–18 Using this careful approach, only 3 (1.1%) of 267 cases showed any evidence of HPV, and despite our rigorous protocols, environmental contamination cannot be ruled out as an explanation of these 3 low-copy positive results. While HPV89 is not detected in cancer in humans,49 it is relatively prevalent in cervical samples from the general population.50 Thus, other specimens processed during the same time frame as this study could have led to low level contamination. Given that this HPV type is non-carcinogenic and that the band demonstrating its presence in the PGMY assay of 1 case was very weak, it seems likely that the HPV89 positivity was due to contamination. HPV16 and HPV31 are well-known carcinogenic HPV types, but the evidence for their presence in the SPF10 testing was also weak, and the involved cases showed no evidence of HPV in the PGMY or HPV16/18 E6/E7 tissue assays. Finally, none of these 3 cases showed evidence of HPV oncogene activity, as evaluated by p16INK4a protein overexpression in all 3 cases and HPV E6/E7 seropositivity in the HPV16-positive case. Taken together, these results suggest that the highly sensitive PGMY and SPF10 primers may have detected minute amounts of HPV DNA in these cases that were present in the sample due to contamination from laboratory controls or other environmental sources, rather than from actual established HPV infections.
Patients in our study came from 13 of the 31 Provinces, Municipalities, and Regions of China, including both the high-risk Taihang mountain region (27.6%) and other areas (72.4%). In addition, cases came from counties with low (<2/100,000) and high (≥32/100,000) cervical cancer mortality rates.51 Given that the negative results of this study were not restricted to a single county or a single high-risk population, it seems unlikely that HPV is a significant causal factor for ESCC throughout China. Thus, true geographic etiologic heterogeneity is an unlikely explanation for discrepancies in HPV prevalence in ESCC tissues found in previous studies. Strict review of study methodologies is recommended to more fully understand discrepant HPV/ESCC findings.
This study is one of the largest tumor tissue studies of HPV and esophageal cancer to date. Paying careful attention to good molecular biology procedures and using the most sensitive HPV DNA detection and confirmation techniques available, we found that the prevalence of HPV DNA in the ESCC cases approached 0%. This study provides the most definitive evidence to date that HPV is not involved in esophageal carcinogenesis.
The authors thank Dr. Hormuzd Katki for providing biostatistical expertise during the design phase of this study. This work was supported by an Intramural Research Award to Dr. Koshiol from the Division of Cancer Epidemiology and Genetics, National Cancer Institute, National Institutes of Health; the Cancer Prevention Fellowship Program, National Cancer Institute, National Institutes of Health; a research contract [N01-RC-47702] from the National Cancer Institute to the Cancer Institute at the Chinese Academy of Medical Sciences; General Funds from the Intramural Research Program of NCI; and a research grant [JK2006B05] from the Cancer Institute, Peking Union Medical College and Chinese Academy of Medical Sciences. Dr. Gravitt has been a paid consultant and has received research funding from Roche Molecular Systems, who manufacture the HPV Linear Array test used in this study. Dr. von Knebel Doeberitz is a member of the board of mtm-laboratories AG, the company that provided reagents for the analysis of p16INK4a overexpression in the tumor samples.
Grant sponsor: National Cancer Institute to the Cancer Institute at the Chinese Academy of Medical Sciences; Grant number: N01-RC-47702; Grant sponsor: Cancer Institute, Peking Union Medical College and Chinese Academy of Medical Sciences; Grant number: JK2006B05; Grant sponsors: Division of Cancer Epidemiology and Genetics, National Cancer Institute, National Institutes of Health, Cancer Prevention Fellowship Program, National Cancer Institute, National Institutes of Health, Intramural Research Program of NCI, Roche Molecular Systems
None of the other authors has a personal financial conflict of interest to report.