Molecular etiology of second primary tumors in contralateral tonsils of human papillomavirus-associatedindextonsillar carcinomas

Abstract

Objectives

For patients with tobacco-related head and neck squamous cell carcinoma (HNSCC), the occurrence of a second primary tumor (SPT) is an ominous development that is attributed to a field cancerization effect and portends a poor clinical outcome. The goal of this study was to determine whether patients with human papillomavirus (HPV)-related index tonsillar carcinomas can also develop SPTs in the contralateral tonsil, and to discern the molecular etiology of HPV-related tumor multifocality.

Materials and Methods

The surgical pathology archives of The Johns Hopkins Hospital were searched for all patients with primary HPV-related tonsillar squamous cell carcinoma who developed a synchronous or metachronous carcinoma in the contralateral tonsil. The HPV-16 E6 exon was sequenced from each independent cancer site to determine whether the tumor pairs harbored the same or a different HPV-16 variant.

Results

Four patients with bilateral HPV-related tonsillar carcinomas were identified. In every case, the HPV DNA sequences derived from the index tumor and corresponding SPT were 100% concordant, indicating that the index and SPTs were caused by the same HPV-16 variant.

Conclusion

For the small subset of patients with tonsillar carcinomas who develop SPTs in the contralateral tonsil, the index case and the SPT consistently harbored the same HPV variant. This finding suggests that HPV-related tumor multi-focality can be attributed either to independent inoculation events by the same virus, or by migration of HPV-infected cells from a single inoculation site to other regions of Waldeyer’s ring.

Introduction

The 20-year cumulative incidence of second primary tumors (SPTs) in head and neck squamous cell carcinoma (HNSCC) is approximately 36%.¹ Risk factors for the development of SPTs are often the same risk factors responsible for the initial primary process, which traditionally have included tobacco and alcohol use. The odds of being diagnosed with a SPT are 2.9 (95% CI 1.8-4.1) times higher in patients who smoke, and 5.2 (95% CI 3.3-7.9) times higher in patients who drink, compared with those who do not use these substances.² Human papillomavirus (HPV)-associated HNSCCs (HPV-HNSCC) are less often associated with excessive tobacco and alcohol use and, thus, patients with HPV-HNSCCs are less likely to develop smoking-related SPTs.^3-6 These same patients, however, may be at an elevated risk of developing other HPV-associated carcinomas. Men with HPV-associated oropharyngeal carcinoma are at increased risk for developing HPV-associated anogenital carcinoma, and several recent reports have described patients diagnosed with HPV-HNSCCs who subsequently developed second primary cancers of the head and neck: McGovern et al. described a 46 year-old patient with three synchronous HPV-associated carcinomas of Waldeyer’s ring,⁷ Roeser et al. described a case of synchronous HPV-associated tonsillar carcinomas,⁸ and Singhi et al observed 3 patients with HPV-associated oropharyngeal carcinomas in their series of patients with HPV-associated nasopharyngeal carcinomas.⁹

The purpose of this study was to identify patients with human papillomavirus (HPV)-associated index tonsillar carcinomas and a SPT in the contralateral tonsil in order to determine whether these synchronous or metachronous HPV-HNSCC are infected with the same or different variant of HPV. A better understanding of this relationship may provide insight into the mechanisms underlying SPTs in patients with HPV-HNSCC. As the burden of HPV-HNSCC continues to rise, it will become increasingly important to understand the risk of SPT development in this patient population and adopt personalized treatment strategies to manage these virus-related cancers.

Patients and Methods

Subjects

This study was approved by the Institutional Review Board of the Johns Hopkins Medical Institutions. Briefly, we searched the Johns Hopkins Hospital pathology database for all patients diagnosed with a primary squamous cell carcinoma of the tonsil between 1999 and 2006. This search yielded 135 patients with a confirmed tissue diagnosis of tonsillar squamous cell carcinoma. We then reviewed all of the pathology reports for these patients to identify those who had synchronous bilateral tonsillar carcinomas or who subsequently developed metachronous tonsillar carcinomas. We were able to identifiy a total of six cases (or 4% of tonsillar carcinomas). One case was then excluded from the analysis because the tumor blocks were not available for HPV sequencing. Another case was excluded because the tonsillar cancers were not HPV associated as determined by in-situ hybridization. The medical records for each of the subjects included in this study were thoroughly reviewed to ensure that each case involved a second primary carcinoma, rather than recurrence. In order to diagnose second primary carcinomas, we utilized a modified version of the criteria that were originally described by Warren and Gates.^10,11 In order to qualify as a SPT, the following criteria were required: 1) each neoplasm must have been anatomically separate; 2) the possibility that a second carcinoma represented a recurrence was excluded (each malignancy must have been separated by time [at least three years] and/or anatomical space [at least 2 cm of normal mucosa between each lesion]). To further exclude the possibility of direct tumor extension into adjacent compartments of Waldeyer’s ring, we restricted our cases to tonsillar carcinomas separated by a zone (i.e. base of tongue) of clinically uninvolved tissue. Synchronous carcinomas were defined as two geographically separate carcinomas that were diagnosed within a six-month period. Metachronous carcinomas were defined as geographically separate carcinomas which were diagnosed greater than six months after an index carcinoma was diagnosed.

Comparison of HPV-16 Sequences and Identification of Variants

In order to investigate the molecular etiology of SPT development in HPV-HNSCC, we first confirmed the presence of HPV 16 DNA in the index tumor and its paired SPT by a type 16 specific assay which was performed using the in situ hybridization catalyzed signal amplification method for biotinylated probes (DAKO GenPoint, Carpinteria, CA). Subsequently, the viral DNA derived from the index tumor and SPT were sequenced. Paraffin embedded specimens were obtained and processed for genomic DNA extraction. Rather than sequence the entire HPV-16 viral genome in order to identify viral variants, previous research in cervical cancer has demonstrated that specific E6 nucleotide changes are able to correctly identify viral variants with low potential for misclassification.^12,13 Therefore, full-length E6 was amplified using the polymerase chain reaction (PCR), gel purified, and subsequently sequenced. Sequencing data for the primary and secondary tumors were then aligned to compare sequence similarity and to identify HPV variants. Numerous HPV-16 variants have been described in the literature and have been grouped into phylogenic lineages.¹⁴ The HPV-16 variants that were detected in the tumor samples were assigned to a phylogenic class and subclass by detection of characteristic signature gene sequences in the Open Reading Frame (ORF) of E6, as described previously.¹² The prototype sequence of the European lineage was used as the reference for aligning the nucleotide position and comparison of sequences.¹⁵

Results

Patients

We found that 4% (6 of 135) of patients who were diagnosed with a primary tonsillar carcinoma developed either a synchronous or metachronous SPT in the contralateral tonsil. A total of four subjects met criteria for inclusion and had available tissue for analysis in this study. All patients were male and had a mean age of 59 (range: 44 – 76 years old). Three of the four identified cases presented with synchronous tonsillar carcinomas, while the remaining case presented with a metachronous tonsillar carcinoma. The first case reported that he had greater than 12 lifetime sexual partners; however, the sexual histories were unavailable for the remaining cases. Three cases presented with a T1 tumor stage for the index carcinoma, while one patient presented with a T3 tumor. The second primary tumors for all cases were staged as early T1 disease or carcinoma in-situ. It should be noted that for every case, the SPT occurred at a contralateral site that was distinct from the index tumor; therefore, these carcinomas were classified as a contralateral primary cancer rather than multifocal or multicentric carcinomas. Furthermore, no other secondary HPV-associated cancer in the head and neck or at other sites (anal, penile, etc.) were diagnosed in these patients. All of these head and neck cancer patients are alive and disease free 6 to 13 years after treatment. Complete demographic and clinicopathological characteristics are shown in Table 1.

Table 1.

Demographic and clinical characteristics of study subjects. Age, gender, history of tobacco and alcohol use, date of initial presentation for primary tumor, symptom(s) at time of initial presentation, anatomical location of index and second primary tumor (SPT), stage of index and SPT, treatment modality for index and SPT, interval between diagnosis of index carcinoma and SPT, and medical comorbidities are shown for each case. M, male; N, no history of use; Curr, current substance use; Prev, previous history of substance use; L, left; R, right; MRND, modified radical neck dissection; GERD, gastroesophageal reflux disease.

								Index Tumor			Second Primary Tumor
Case	Age	Gender	Race	Tobacco	Alcohol	Date	Presenting Symptom	Site	Stage	Treatment	Site	Stage	Treatment	Inter val (mo)	Comorbidities
1	69	M	White	N	N	2006	Sore throat	R tonsil	T1N1M0	Radical tonsillectomy, adjuvant IMRT	L tonsil	T1N0M0	Surgical resection	32	Hypertension
2	44	M	White	N	N	2000	Neck mass	R tonsil	T1N2aM0	Definitive chemoradiation	L tonsil	T1N0M0	Definitive chemoradiation	0	GERD
3	76	M	Hispanic	Curr	N	1999	None	R tonsil	T3N1M0	Surgical resection, MRND; adjuvant radiation therapy	L tonsil	T1N0M0	Surgical resection, MRND; adjuvant radiation therapy	0	Depression
4	46	M	White	Prev	Prev	2006	Neck mass	L Tonsil	T1N2bM0	N/A	R tonsil	TisN0M0	N/A	0	Rheumatoid arthritis

HPV-16 DNA Sequence Analysis

For each case, HPV-16 DNA sequencing from the index and SPTs were found to be 100% concordant in the E6 ORF. For case 1, the isolates obtained from both tumors belonged to the European class of HPV-16. A comparison of the isolates to the reference sequence demonstrated that guanine substitutions occurred at positions 131 and 350, which defines the G131/350G subclass. The prevalence of the G131G variant is approximately 10% in North American cervical specimens.¹⁴ The exact sequence match between the isolates obtained from each independent tumor site suggested that a common viral variant caused the index tumor and the SPT. Similarly, comparison of the HPV-DNA isolates from the other cases again demonstrated consensus genetic sequences between the primary and secondary tumors suggesting infection with the same viral variant. The HPV isolate in case 2 was the European-prototype variant P350G which is prevalent in approximately 30% of North American cervical specimens and the HPV isolates in cases 3 and 4 were Asian-American variants which are prevalent in <1% of North American cervical specimens.¹⁴ Nucleotide sequence variations for each case are depicted in Figure 1.

Figure 1.

HPV-16 nucleotide sequence variations in the E6 open reading frame (ORF) from the index tumor and second primary tumor (SPT). The DNA sequences from the index and SPT were identical in each case. The prototype sequence of the European lineage (E-P-350T) was used as the reference for alignment of the nucleotide position (Reference) and comparison of sequences.¹¹ Common positions where nucleotide polymorphisms occur in the E6 ORF are listed at the top of the figure. Dashes indicate no variation was found between HPV-16 DNA in the tumor specimen and the reference sequence. A, adenine; T, thymine; G, guanine; C, cytosine. *Major phylogenic classes include European (E), European prototype-like (E-P), and Asian American. Variants within each class (i.e. subclasses) are specified according to the following convention: identity of substituted nucleotide -- position of the substituted nucleotide -- identity of the nucleotide at position 350. For example, E-G131G denotes a variant of the European class, G131G subclass (guanine substitutions at positions 131 and 350).

Discussion

For head and neck cancer patients, the development of a SPT has considerable implications regarding long-term survival. When considering all patients diagnosed with HNSCC (independent of HPV status), those who develop metachronous malignancies have a better overall survival rate than those with synchronous cancers, although both predict a poor prognosis with a 5 year survival rate at the time of diagnosis of the SPT to be 26%.^16-17 The incidence and overall survival associated with a second primary HPV-associated cancer are not currently known. In our series, we report an incidence of 4% of SPTs in the contralateral tonsil of patients presenting with an index HPV-associated tonsillar cancer. All of these head and neck cancer patients are disease free 6 to 13 years after treatment.

The molecular mechanism that underlies the process of SPT development in the context of HPV-HNSCC has yet to be elucidated. Therefore, we explored the etiology of HPV-associated SPT development by evaluating whether these two independent cancer sites were caused by the same viral variant. We found that for every case, there was a 100% concordance between the HPV E6 DNA sequence derived from the index and the second primary cancer (Figure 1). Thus, our results demonstrate that the index and SPTs were caused by the same variant of HPV-16 and could help to clarify the mechanisms by which second primaries develop in HPV-HNSCC.

Previous research in cervical cancer has suggested three potential molecular mechanisms for the development of multifocal HPV-associated cervical neoplasms, which we can extrapolate to HPV-associated head and neck cancers.¹⁷ In the first model, a single HPV infection causes transformation of multiple cells at distinct anatomical locations, creating a so-called field effect (Figure 2, model 1). In the second model, HPV-related carcinogenesis is thought to occur in the setting of multiple independent infections from several HPV types and/or variants occurring in distinct areas of the oropharynx (Figure 2, model 2). This scenario would be expected to most commonly arise in the setting of multiple viral exposures over a period of time. In the third model, SPTs result from a single infection and are clonally related to the index carcinoma through the migration of virally-infected cells to other sites: Following a single inoculation event, HPV-infected cells migrate from one tonsil to the other along the lymphoepithelial tract known as Waldeyer’s ring. (Figure 2, model 3)

Figure 2.

Three proposed models for the development of HPV-associated second primary tumors (SPTs). Model 1: a single HPV infection creates a field effect by transforming multiple cells at distinct anatomical locations. Model 2: in the setting of increased HPV infection susceptibility, multiple independent infections from several HPV types infect different areas of squamous epithelium and result in cancer formation at different sites. Model 3: SPTs are clonally related to an index tumor and the result of a single infection. Aberrant oncogene expression in an index tumor results in transformation of a clone which migrates to a second location and develops into a SPT.

When we compared primary tonsillar carcinomas with their corresponding SPTs, we found that the tumor pairs consistently harbored the same HPV-16 variant. This finding suggests that when a patient with an HPV-related tonsillar carcinoma develops a SPT in the contralateral tonsil, this phenomenon of tumor multifocality does not reflect independent inoculation events with different HPV variants (model 2). Instead, HPV-associated SPTs more likely arise from an HPV-mediated field cancerization (model 1 of figure1) or through clonal spread of an index carcinoma (model 3 of figure 1). In model 1, chronic exposure to an external reservoir (i.e. a sexual partner) could result in a second inoculation event. Indeed, clinical evidence of this type of exposure has been suggested by couples who have developed simultaneous oropharyngeal SCC caused by identical HPV variants.¹⁸ Alternatively, the initial site of HPV inoculation and cancer development could itself serve as an internal reservoir of HPV particles for subsequent infections of other susceptible oropharyngeal locations (e.g. contralateral tonsil or other sites within Waldeyer’s ring). In model 3, the initial site of HPV inoculation and cancer development could serve as the epicenter for clonal expansion and migration of HPV-infected cells to other sites. Indeed, clonal spread within the sheath of lymphoepithelium in Waldeyer’s ring has been favored by some as the mechanism by which paired tonsillar and nasopharyngeal HPV-associated carcinomas may develop.⁹ Although the development of SPTs in the contralateral tonsil would require cellular migration across extended tracts of Waldeyer’s ring, this same mechanism is now recognized as an important mechanism of tumor multifocality for smoking-related HNSCCs ^{19, 20}. To differentiate between models 1 and 3, further studies that evaluate the integration site of the HPV DNA in the genome of the index and the SPT will be necessary. Demonstration of a different integration site of the HPV DNA in the primary and SPT genome would support model 1; whereas, identification of the same integration site in the primary and SPT would support model 3.

Our study highlights how the concept of field cancerization may need to be interpreted differently in the setting of virus-related cancers. In virus-related cancers, exposure to an environmental carcinogen, such as HPV, can result in cellular infection and, subsequent, intracellular changes in distinct sites within the oropharynx. We demonstrate that the same viral variant is the causative agent of the index and SPT which suggests that these independent tumors may be more genetically similar than in tobacco-related index and SPTs that develop in the upper aerodigestive tract. Therefore, the diagnosis of SPT in the setting of HPV-HNSCC may not necessarily portend a poor prognosis as in the case of tobacco-related head and neck cancers given the similar genetic alterations present within the primary and secondary cancers.

In this study, the second primary tumors were detected incidentally in three of the four patients due to the routine practice of performing bilateral tonsillectomies during the work-up of suspected tonsil cancers or of an unknown primary at the Johns Hopkins Hospital.¹⁹ The one patient who presented with a metachronous SPT underwent a diagnostic ipsilateral tonsillectomy followed by radiation to the ipsilateral neck with Intensity Modulated Radiation Therapy (IMRT) at an outside hospital and, subsequently, the SPT in his contralateral tonsil was diagnosed 32 months after treatment of his primary cancer. Therefore, the performance of bilateral tonsillectomies during the time of initial diagnostic work-up may help to identify multiple occult sites of HPV infection and associated cancers which may not be readily appreciated on physical examination. In addition, given the existing practice by some radiation oncologists for treating unilateral necks with radiation, our data further supports a potential benefit of performing bilateral tonsillectomies for patients in whom a restricted radiation field might be planned.

Conclusion

We report a series of HPV-associated second primary squamous cell carcinomas of the palatine tonsils. For each case, molecular analysis of the index carcinoma and second primary tumor indicated that the same viral variant was the etiologic cause of both cancers at each independent site. We propose several models for the development of SPTs in HPV-HNSCC. Although the development of SPTs does not appear to be common for patients with HPV-related tonsillar carcinomas, removal of the contralateral tonsil may disclose coincident cancers.