Second primary malignancies (SPMs) are the leading cause of death in survivors of head and neck squamous cell carcinoma (HNSCC). Synchronous SPMs are of significant clinical interest because they potentially can be identified by screening procedures at the time of diagnosis of the index cancer. Recently, human papillomavirus (HPV) has emerged as a distinct risk factor for oropharyngeal head and neck squamous cell carcinoma (HNSCC), differing from classic tobacco/alcohol-associated HNSCC, suggesting that there also may be distinct patterns of synchronous SPMs.
The authors performed a population-based cohort study in 64,673 patients in the National Cancer Institute Surveillance, Epidemiology, and End Results registry (1979-2008), defining risks of synchronous SPM in patients with HNSCC who were diagnosed before and after the emergence of prevalent HPV-associated oropharyngeal HNSCC. Excess risk was calculated using standardized incidence ratios (SIR) and excess absolute risk per 100 patients.
Among patients with HNSCC, the SIR of synchronous SPM was 5.0, corresponding to 2.62 excess cases per 100 patients. The site with the highest excess risk of a second cancer was the head and neck (SIR, 41.4), followed by the esophagus (SIR, 21.8), and lung (SIR, 7.4). The risk of synchronous SPM changed markedly over time for patients with oropharyngeal HNSCC. In the 1970s and 1980s, oropharyngeal cancers carried the highest risk of SPM. Risk began to dramatically decline in the 1990s; and currently, oropharyngeal cancers carry the lowest risk of synchronous SPM.
Patients with head and neck squamous cell carcinoma (HNSCC) experience elevated risk of developing second primary malignancies (SPMs), which are the leading cause of mortality in survivors of this disease.1 SPMs may develop synchronously or subsequently (metachronously). Metachronous SPMs occur in 20% to 30% of HNSCC patients, most commonly in mucosal sites of the head and neck, lung, and esophagus, which are subjected to field cancerization from tobacco or alcohol exposure.2-6 In contrast, it is estimated that synchronous SPMs are present in 1% to 6% of patients with newly diagnosed HNSCC.6-9 These synchronous second cancers are of significant clinical importance, because they potentially may be identified during investigation of the index HNSCC. The locations of synchronous SPMs and the differential risks associated with each index HNSCC subsite (oral cavity, oropharynx, larynx, or hypopharynx) have not been well defined.
In recent years, human papillomavirus (HPV) has replaced tobacco and alcohol as the etiologic agent responsible for most HNSCC arising in the oropharynx.10-14 Because many of these patients have not been exposed to tobacco and alcohol, we hypothesized that patients diagnosed with oropharyngeal HNSCC in the era of HPV predominance would have a lower risk of synchronous SPMs. We previously identified a lower risk of metachronous SPMs for patients who were likely to have HPV-positive oropharyngeal cancer.14 If a similar differential in risk were to be found for synchronous SPMs, then these data would have implications for the pattern and intensity of screening for synchronous cancers in patients presenting with HPV-related HNSCC. To test this hypothesis, we analyzed the subsite-specific risks of synchronous SPM in a large US cohort of patients who were diagnosed with HNSCC at different points in time.
MATERIALS AND METHODS
We studied patients diagnosed with HNSCC between 1979 and 2008 in the 9 registries of the National Cancer Institute (NCI) Surveillance, Epidemiology, and End Results (SEER) program. All cancers, primary and subsequent, occurring among residents of geographic registries comprising SEER are reportable, and the program has near-universal follow-up.15, 16 A limitation of the SEER cancer registry is lack of information on risk factors, such as tobacco use, alcohol use, or HPV status. The NCI does not require institutional board approval for use of this deidentified data set, but a data use agreement was signed with the SEER program. Studies have confirmed that pathologic, surgical, and radiation data are accurately recorded in the SEER program.15, 17
Patients with HNSCC were identified using International Classification of Diseases for Oncology, third edition histology codes for invasive squamous cell carcinoma (SCC) (codes 8070-8076 and 8078) and topographic codes for head and neck subsites (oral cavity, oropharynx, larynx, and hypopharynx). Stage at presentation was categorized according to SEER historic staging, which categorizes stage as “localized” (localized and lymph node-negative without distant metastases), “regional” (locally advanced or lymph node-positive without distant metastases), or “distant” (with distant metastases). Synchronous SPMs were defined as distinct, solid cancers diagnosed ≤6 months of the index HNSCC. Excess risk was calculated in both relative and absolute terms. In relative terms, the standardized incidence ratio (SIR) is defined as the ratio of observed to expected SPMs. In absolute terms, the excess absolute risk (EAR) is defined as the absolute number of additional subsequent cancers attributable to the index HNSCC. The EAR was calculated as the excess number (observed-expected) of SPMs per 100 patients. The number of expected cancers was calculated for a reference noncancer patient cohort of similar age, sex, race, and time period. The SIR is useful as a relative measure of strength of association between 2 cancers, and EAR is an absolute measure of the clinical burden of additional cancer occurrences in a given population.
In the analysis of trends over time, data were binned in 5-year periods (ie, 1979-1983, 1984-1988, 1989-1993, 1994-1998, 1999-2003, and 2004-2008). Confidence intervals (CIs) were calculated using the Byar approximation to the Poisson distribution.18 SIR and EAR values were calculated in SEER*Stat release 7.0.4 (2011; NCI, Bethesda, Md). Trends were analyzed using joinpoint regression, a variant of log-linear regression, in Joinpoint 3.5.2 (NCI). In joinpoint regression, a segmented trend line is fitted in an unsupervised fashion by using a permutation test. The number needed to screen was calculated as the reciprocal of absolute risk difference. Statistical analyses were performed in R version 2.15.1 (R Foundation for Statistical Computing, Vienna, Austria).
In total, 64,673 occurrences of HNSCC were identified in the SEER registry between 1979 and 2008 (Table 1). Among all patients, the SIR of a synchronous solid tumor was 5.0 (95% CI, 4.7-5.2). The EAR was 2.62 excess cases per 100 patients. This implies a 2.62% excess incidence of a synchronous second solid cancer above the expected rate.
Table 1. Elevated Risk of Synchronous Second Primary Malignancy by Site of Index Head and Neck Cancer
All HN Primary Cancers (n = 64,673)
Oral Cavity (n = 18,625)
Oropharynx (n = 16,877)
Larynx (n = 24,136)
Hypopharynx (n = 5035)
SIR 95% CI
EAR per 100 Patients
SIR 95% CI
EAR per 100 Patients
SIR 95% CI
EAR per 100 Patients
SIR 95% CI
EAR per 100 Patients
SIR 95% CI
EAR per 100 Patients
Abbreviations: CI, confidence interval; EAR, excess absolute risk; HN, head and neck; HNLE, head and neck, lung and esophagus; SIR, standardized incidence ratio; SPM, second primary malignancy.
All solid tumors
Lung and bronchus
All non-HNLE sites
There was no difference in excess synchronous SPM risk by patient sex (EAR, 2.61 per 100 men, 2.62 per 100 women; P = .21). Excess risk was higher among patients who were diagnosed at age ≥60 years compared with those diagnosed at age <60 years (2.75 per 100 patients vs 2.42 per 100 patients; P < .0001). The excess risk of synchronous SPM increased with advancing stage of HNSCC at presentation (EAR, 1.92 per 100 patients presenting with localized, lymph node-negative disease, 3.06 per 100 patients presenting with locally advanced or lymph node-positive disease, and 3.49 per 100 patients presenting with distant metastases; P < .0001).
Of the total excess risk of harboring a synchronous second cancer, 83.2% of the risk was attributable to second cancers in the head and neck, lung, or esophagus. Risk across these sites varied widely. In relative terms, the location with the highest excess risk of synchronous SPM was the head and neck (SIR, 41.4), followed by the esophagus (SIR, 21.8), and lung (SIR, 7.4). In absolute terms, the excess burden of synchronous SPM was highest in the head and neck (1.19 per 100 patients), followed by the lung (0.79 per 100 patients), and the esophagus (0.21 per 100 patients).
The risk of synchronous SPM was associated with the subsite of the index HNSCC. Over the complete study period from 1979 to 2008, patients with a hypopharyngeal primary cancer had the highest risk of synchronous SPM (SIR, 7.2; EAR, 4.04), and those with a laryngeal primary cancer had the lowest (SIR, 4.0; EAR, 2.16; difference by subsite, P < .0001). Trends in synchronous SPM risk by subsite were analyzed over time (Fig. 1). In patients with primary cancers arising in the hypopharynx, larynx, and oral cavity, risks remained constant during the period from 1979 to 2008. In contrast, among patients with oropharyngeal cancer, the risk of synchronous SPM changed dramatically over time. In the 1970s and 1980s, oropharyngeal cancers carried a high excess risk of synchronous SPM, similar to hypopharyngeal cancers at >4 excess cases per 100 patients. However, in the late 1980s and early 1990s, this risk began to decline (annual percentage change, −4.6%), such that oropharyngeal cancers now carry the lowest excess risk of synchronous SPM of any HNSCC subsite (<2 excess cases per 100 patients). The inflection point corresponded to the time of the etiologic shift in oropharyngeal SCC, which became a predominantly HPV-associated cancer in the mid-1990s.19, 20
Among the 16,877 patients who were diagnosed with oropharyngeal SCC over 30 years, there were 10 synchronous anogenital and cervical cancers observed, and 3.6 were expected. These risks were elevated but did not reach statistical significance for any site. The excess risk of a synchronous cervical cancer in women was 6.49 (95% CI, 0.77-23.1), the excess risk of a synchronous female genital cancer was 2.18 (95% CI, 0.80-4.75), the excess risk of a synchronous male anal cancer was 6.16 (95% CI, 0.16-34.33), and the excess risk of a synchronous penile cancer was 8.92 (95% CI, 0.23-49.72). These risks remained nonsignificant when the cohort was limited to patients with oropharyngeal SCC who were diagnosed in the HPV-predominant era. No elevation in risk at these sites was observed among patients with nonoropharyngeal HNSCC.
Synchronous SPM risk data are translated into terms applicable to clinical surveillance by calculating the “number needed to screen,” defined as the number of patients with newly diagnosed HNSCC a clinician would need to screen to observe 1 excess SPM beyond the expected number (Table 2). This calculation of the number needed to screen is an approximate figure, because it assumes a screening sensitivity of 100%. In the current analysis, 1 excess synchronous solid cancer potentially could be identified by screening 38 patients with HNSCC. The number needed to screen to observe an additional solid cancer was lowest for patients with hypopharyngeal cancer at 25. For patients with oropharyngeal cancer, the number needed to screen increased from 21 patients in 1979 to 45 patients in 2008 (P < .0001). The number needed to screen to identify 1 excess synchronous head and neck cancer was 84 and was lowest for patients with hypopharyngeal primary tumors at 60 patients (P < .0001). The number needed to screen to identify 1 excess synchronous lung cancer was 126 and was lowest for patients with laryngeal cancer at 95 patients (P < .0001). The number needed to screen to identify 1 excess synchronous esophageal cancer was 488 and was lowest for patients with hypopharyngeal cancer at 214 (P < .0001). Among patients with oropharyngeal SCC, the number needed to screen to identify 1 excess synchronous anogenital or cervical cancer was 2616 and ranged from 1288 (for female genital cancers) to 15,110 (for male anal cancers).
Table 2. The Number of Patients Needed to Screen to Identify 1 Excess Synchronous Second Primary Cancer
No. of Patients
Abbreviations: HN, head and neck; HNLE, head and neck, lung and esophagus; HNSCC, head and neck squamous cell carcinoma; SPM, second primary malignancy.
All solid tumors
Lung and bronchus
All non-HNLE sites
Second primary malignancies are the leading obstacle to long-term survival among patients with HNSCC.1 The majority of SPMs are metachronous, occurring >6 months after diagnosis of the index HNSCC.4 We and others have reported that the risk and location of metachronous SPMs are significantly associated with the subsite of the index HNSCC.6, 7 Here, we analyzed synchronous SPMs in patients with HNSCC. These data describe the landscape of synchronous SPM risk in patients with HNSCC and have important hypothesis-generating implications for further study. Of all patients diagnosed with HNSCC, 2.6% harbored an excess synchronous second cancer: 1.2% in the head and neck, 0.8% in the lung, and 0.2% in the esophagus. The degree of risk varied significantly according to the subsite of the index head and neck tumor and, over the past 3 decades, has changed dramatically for patients with oropharyngeal cancer.
The prevalence of synchronous second primary cancers in patients presenting with HNSCC had been previously estimated between 1% and 6%.2-5 In the presence of synchronous second primary tumors, treatment plans are modified, and survival is significantly poorer (as low as 8% at 5 years).2 Similar to metachronous SPMs, the most common sites of synchronous SPMs are the head and neck, lung, and esophagus, consistent with the model of field cancerization.6, 7
Traditionally, patients diagnosed with HNSCC have been screened for synchronous SPMs with the performance of a comprehensive examination under anesthesia, including laryngopharyngoscopy, bronchoscopy, and esophagoscopy. These procedures do carry small but defined risks, such as those of esophageal perforation with either rigid (0.68%) or flexible (0.03%) esophagoscopy.21 At the same time, the sensitivity of these techniques is imperfect, particularly for small, peripherally located lung tumors.22 Positron emission tomography and computed tomography (PET/CT) scans may offer superior sensitivity in identifying synchronous primary cancers; however, false-positive results lead to the additional costs and burden associated with unnecessary investigations. Therefore the universal use of whole-body PET/CT to screen for SPMs has not been recommended.23, 24 To better define the risks of synchronous SPMs and generate hypotheses for further investigation of screening modalities, we analyzed these risks in a large US cohort captured in the SEER registry.
Before proceeding, it is important to mention several caveats to these SPM data. SEER registry data do not record cancer risk factors such as tobacco/alcohol use or HPV status, precluding more detailed elucidation of the relation between etiologic factors and SPM risk. Like in clinical practice, lung metastases occasionally may be misclassified as SPMs, and vice versa. Nevertheless, the quality control of the SEER program, the large cohort size, the high-quality multiple primary data, and the presence of a reference cohort permit an analysis of excess risk with excellent internal validity and generalizability.
The data presented in this report describe the various levels of risk for synchronous SPMs based on various patient and tumor characteristics. The risk of synchronous SPM was slightly elevated for older patients and was elevated more significantly for patients with advanced stage disease. Risk also varied according to the subsite in which the index HNSCC arose. Hypopharyngeal cancers carried the highest risk of all head and neck subsites, with 1 in 25 patients harboring a synchronous solid cancer, making these patients potentially most likely to benefit from additional investigation, such as imaging or endoscopy, at the time of diagnosis. Overall, patients with HNSCC had a 41-fold relative risk of harboring a synchronous cancer within the head and neck, which translated to a number needed to screen of 84 patients. In contrast, 488 patients would have to undergo screening of the esophagus to potentially identify 1 excess synchronous SPM, suggesting that routine, comprehensive examination of head and neck mucosal sites is likely to be of much greater value than routine examination of the esophagus. Approximately 126 patients with HNSCC would need to be screened to potentially identify 1 synchronous lung cancer, but this number was lowest (n = 95) for patients with an index laryngeal cancer. Furthermore, 16.8% of excess SPM risk occurred in sites other than the canonical upper aerodigestive mucosa (head and neck, lung, and esophageal), demonstrating the limited yield of even comprehensive endoscopic examinations.
It is striking that, over the past 30 years, during the emergence of HPV-associated oropharyngeal cancer, the level of risk for synchronous SPMs has markedly changed for patients with oropharyngeal cancer. Whereas this subsite once carried the highest risk of synchronous SPM, it has now declined to the lowest risk of any HNSCC subsite, and the inflection point of this trend corresponded to the emergence of prevalent HPV-positive oropharyngeal cancer. Therefore, nonsmoking patients with HPV-positive oropharyngeal cancers appear to be least likely to harbor synchronous SPMs. Additional investigation in these patients may be of lower yield. Indeed, a recent study focused on this subgroup failed to identify any synchronous SPMs during panendoscopy among 64 patients, suggesting that future screening practices may be tailored by HPV and smoking status.25
Oncogenic HPV is also known to cause carcinomas of the cervix, vulva, anus, and penis.26-28 We previously reported an association between oropharyngeal SCC and the risk of metachronous second cancers in these sites.29 However, the risk of synchronous second cancers in these sites, among patients with oropharyngeal SCC, was not significantly elevated. These are rare cancers; therefore, elevated relative risks did not reach statistical significance, and the absolute number of excess cases was small. There were only 10 synchronous cervical/anogenital cancers observed among nearly 17,000 patients with oropharyngeal SCC, and the number needed to screen was high. Therefore, although anogenital and cervical synchronous cancers are likely to be etiologically linked to HPV-positive oropharyngeal cancer, they appear to be too uncommon to benefit from routine screening.
These data, based on nearly 65,000 occurrences of HNSCC in a large US cohort, help to define the determinants of risk for synchronous SPMs. The findings demonstrate that patients with HNSCC are a population with markedly escalated risks of second cancers, and these risks are modulated by both patient and tumor characteristics. Certain subsets of these patients may be more likely or less likely to benefit from the rational application of various screening modalities, making these hypothesis-generating data potentially useful for further investigation as we seek to optimize cancer detection in this high-risk population.