Unique Patterns of Distant Metastases in HPV-Positive Head and Neck Cancer

Patients with head and neck cancer historically developed locoregional recurrence as the primary site of treatment failure. However, the landscape of head and neck squamous cell carcinoma (HNSCC) has shifted over the past two decades due to the declining rates of tobacco use and the rising incidence of human papillomavirus infection (HPV). Recent studies suggest that HPV may account for up to 70–80% of cases of oropharyngeal cancer in North America and Europe [1‒4]. Even after adjusting for confounders such as age, tobacco use, and performance status, HPV positivity confers a survival advantage and improved prognosis compared to HPV-negative disease [1, 5‒9].

The precise mechanism resulting in improved outcomes of HPV-positive HNSCC to therapy is unclear but is likely attributable to clinical and molecular characteristics distinct from HPV-negative HNSCC [10‒14]. Multiple prospective studies are underway to explore possible treatment deintensification strategies given the perceived lower recurrence risk in HPV-positive patients [15]. With the distinct biology of HPV-positive disease, its patterns of disease dissemination outside the neck may also be unique. Describing the sites and expected time frame of metastatic spread will be essential in order to safely de-escalate systemic therapy without adversely affecting outcomes.

Older studies suggested that the overall rate of distant metastases (DM) was similar for HPV-positive and -negative tumors [15, 16]. Other analyses, including one of 252 HPV-positive tumors treated with definitive chemotherapy and radiation demonstrated differences in their pattern of metastases [17]. The median time to development of metastases was significantly longer for HPV-positive tumors compared to HPV-negative tumors (16.4 vs. 7.2 months). HPV-positive HNSCC also had a propensity to metastasize to atypical sites such as brain, kidney, and skeletal muscle. Both HPV-positive and -negative patients developed lung metastases at an equal rate although the study only included 9 HPV-negative patients.

Other analyses show that although HPV-positive ­HNSCC demonstrates good locoregional control, DM may occur in unexpected sites and after prolonged intervals [6, 8, 9, 15, 18‒21]. Thus, the pattern and behavior of HPV-positive HNSCC is not clearly defined. Although HPV positivity represents a good prognostic indicator, it may require a distinct surveillance strategy to monitor for distant recurrence.

Here, we report the survival outcomes of patients with both HPV-positive and HPV-negative HNSCC and their pattern of DM. The objective of our study is to gain a greater understanding of the differences in disease progression amidst HPV-positive and HPV-negative ­HNSCC by analyzing the rate, pattern, and time to progression of DM in addition to determining the impact of tobacco use in relation to disease recurrence.

The University of Texas Institutional Review Board provided approval for this retrospective review. From a single institutional database covering the years 2006–2012, all patients with HNSCC (n = 1,494) were included and information regarding DM was analyzed. Patients were divided into groups based on HPV status. In this database, HPV status was not available in some patients diagnosed prior to 2010, patients with unknown and negative HPV status were divided into groups using anatomic location. Patients who did not achieve initial disease control or recurred locoregionally first were excluded. Group 1 (n = 109) consisted of patients with documented HPV-positive HNSCC and compared to patients with HPV-negative or HPV-unknown HNSCC. The HPV-negative or HPV-unknown HNSCC patients were further divided by anatomic location; group 2 (n = 546) consisted of patients with primary disease sites involving the hypopharynx, larynx, or glottis, while group 3 (n = 839) consisted of patients with primary disease sites in the oral cavity, oropharynx, hard palate, or tonsil.

Sites of metastases and time to recurrence were compared in all 3 patient groups. In addition, demographic factors and smoking frequency in patients with distant recurrence were compared to those without distant recurrence in all 3 groups.

Summary statistics for patients’ disease characteristics were reported using means, medians, and ranges for continuous variables, and using counts and percentages for categorical variables. The rate of metastases and smoking frequency between the groups are compared using χ² test. Two-sided p values were reported and a p value <0.05 was considered statistically significant.

Group 1 consisted of 109 patients with HPV-positive HNSCC. The median follow-up from diagnosis was 18 months. Seven patients (6.4%) developed DM. The median time to metastases was 11 months (range 7–41 months, mean 16 months) and the predominant metastatic site was the lung (6/109, 5.5%). Other DM sites included liver (1), bone (1), and brain (1). Where recorded, history of smoking was higher in patients with distant recurrence (5/7, 71.4%) compared to patients without distant recurrence 60/102 (58.8%). Table 1 provides demographic details about this group.

Group 2 consisted of 546 patients with unknown or negative HPV status and primary disease sites involving the hypopharynx, larynx, or glottis. The median follow-up was 25 months. DM developed in 25/546 patients (4.6%). The median time to metastases was 15 months (range 6–52 months, mean 18 months) and the predominant site of metastasis was the lung (10/546, 1.8%). Other DM sites included bone (3) and CNS (3). The smoking frequency in patients with distant recurrence was 25/25 (100%) compared to 459/521 (88.1%) patients without distant recurrence.

Group 3 consisted of 839 with unknown or negative HPV status and primary disease sites in the oral cavity, oropharynx, hard palate, or tonsil. The median follow-up was 24 months. 34/839 (4.1%) developed DM. The median time to metastases was 9 months (range 3–54 months, mean 14 months) and the predominant site of metastasis was the lung (12/839; 1.4%). Other DM sites included bone (2) and liver (2). Smoking frequency in patients with distant recurrence was 22/34 (64.7%) compared to 544/805 (67.6%) patients without distant recurrence. Table 2 provides demographic details about these groups.

Figure 1 demonstrates the various sites and frequencies of metastases in our population. At a median follow-up of 18–25 months, there was no difference in the overall rate of DM between HPV-positive HNSCC compared to HPV-negative/unknown HNSCC involving the hypopharynx, larynx, or glottis (p = 0.21) and HPV-negative/unknown HNSCC in the oral cavity, oropharynx, hard palate, or tonsil (p = 0.13). Table 3 details the rates and median time to development of DM in all 3 groups, as well as smoking frequency.

There was a statistically significant higher rate of metastases to the lung in HPV-positive HNSCC compared to HPV-negative/unknown HNSCC at all sites (p = 0.002). When analyzing individual groups, there was also a statistically significant difference in the rate of metastases to the lung in HPV-positive HNSCC compared to high-risk HPV-negative/unknown (p = 0.012) and intermediate HPV-negative/unknown (p = 0.002). Differences between smoking frequency among those with distant recurrence and without distant recurrence in all three groups were not found to be statistically significant (Table 3).

There is increasing recognition of the impact of HPV status on the natural history of head and neck cancer. Data suggest that DM overall can develop in either a rapid or gradual fashion with more indolent disease amenable to curative-intent therapy [22], but it is unclear whether there is predilection for metastasis to specific organs in HPV-positive cancers. This study focused on patterns of distant spread and deliberately excluded cases where disease persisted or locoregional recurrence was the first site of failure. The differences in locoregional control between HPV-negative and -positive cancers are well characterized. The purpose of this analysis was to identify if there was a difference in distant metastatic behavior. By examining the rate, pattern, and time to progression of DM our study provides specific information about the impact of HPV status on the timing and site of subsequent development of metastatic HNSCC.

In our data set, the overall rate of DM in patients with locoregional control did not differ by HPV status, which was similar to findings in other studies [18, 23]. There was a statistically significant higher rate of lung metastases in patients with known HPV-positive disease compared to cancers with both HPV-negative groups (5.5% vs. 1.8% and 1.4%). This was not seen in prior studies. The time to development of DM in HPV-positive head and neck cancer was also shorter compared to prior studies (11 months vs. 16.4 months).[17]

A shorter interval to development of DM was seen in the HPV-positive HNSCC patients (range 7–41 months, median 11 months) compared to high-risk HPV-negative/unknown disease (range 6–52 months, median 15 months). Further studies are needed to investigate the progression to metastases over an extended period of time to better observe the risk of latent disease.

Studies by other investigators have looked at the time to development of metastases in HNSCC, but not sites. Fakhry et al. [23] observed that the median time to disease progression was similar regardless of HPV status. A retrospective study of oropharyngeal cancer patients demonstrated that DM can develop in HPV-positive HNSCC after longer intervals with 13% of HPV-positive patients having DM beyond 3 years (up to 5.3 years) compared to 96% of HPV-positive patients who presented within 2 years [24]. Patients with HPV-positive oropharyngeal carcinomas have been reported to develop lung metastases as long as 8 years after initial treatment [25].

Currently, the standard for frequency and duration of follow-up is not clearly defined. There are no controlled prospective studies evaluating survival benefits based on follow-up strategies to guide post-treatment surveillance. The National Comprehensive Cancer Network (NCCN) guidelines recommend follow-up the first 2–4 years following diagnosis with head and neck exam along with mirror and fiberoptic examination at varying intervals over the first 5 years followed by annual surveillance beyond 5 years. In terms of imaging, the NCCN states that practitioners can consider CT imaging 6 months after treatment but this is a category 2B recommendation. Otherwise, there are no structured imaging surveillance guidelines unless the patient has a >20 pack-year smoking history, then it is recommended to undergo annual screening with chest CT due to concerns for a second primary malignancy or DM. Our study demonstrated an increased rate of DM to the lung in HPV+ HNSCC indicating a possible role for heightened radiographic surveillance in this population following treatment completion.

One of the limitations of our study is the shorter duration of follow-up. It remains to be seen if recurrences beyond 2 years in either HPV-positive or -negative groups would change our results. The retrospective design of this study limits our ability to explain the biologic basis of this difference in patterns on metastasis. With incomplete HPV status on all patients, dividing patients into 3 groups based on anatomic site is an imperfect method of detecting differences.

The strengths of our study are the large number of patients and our focus specifically on DM rather than local recurrences which are already well defined between HPV-positive and -negative disease. Future directions would involve a longer follow-up of patients with locally controlled HPV-positive cancer. It may be reasonable for trials that de-escalate systemic therapy in this population to consider thoracic imaging as part of surveillance. The cost and impact of additional exposure to radiation must be weighed against the potential for early detection to prolong survival in cases where DM occur [24]. Further study of the biology of HPV-positive cancer may demonstrate a greater propensity towards hematogenous spread to the lungs.

In conclusion, this analysis suggests that the lung is a common site of DM in HPV-positive head and neck cancer; however, in our data set we did not see the very late recurrences in HPV-positive HNSCC. Our data did not show a longer time to development of DM in this population.

The authors thank Dru Gray for assistance with manuscript preparation. The authors also acknowledge the efforts of Alejandra Madrigales from the UT Southwestern tumor registry and Joan Cox from the Parkland Health and Hospital System tumor registry. Biostatistical support was provided by the Biostatistics Shared Resource at the Harold C. Simmons Comprehensive Cancer Center, University of Texas Southwestern Medical Center, which is supported in part by National Cancer Institute Cancer Center Support Grant 1P30CA142543-01 and by the Cancer Prevention and Research Institute of Texas (RP150596)

The UT Southwestern Institutional Review Board (IRB) allowed for collection and analysis of the specific patient data included in this manuscript. Individual patient consent was not required for analysis of these data per the IRB approval.

All authors have no conflicts of interest to declare.

This manuscript was supported in part by National Cancer Institute, Grant/Award Number: K24CA201543-01 (D.E.G).

All authors were involved in the following aspects of this work: substantial contribution to the study concept and design or acquisition, analysis, or interpretation of the data; and drafted or critically revised the manuscript’s intellectual content. All authors approved the final version for publication and agree to be accountable for all aspects of the work, as well as ensuring all questions related to accuracy and integrity of this work are appropriately answered.

Ruth Sacks and Jennie Y. Law contributed equally to this work. Previously presented at the 50th annual meeting of the American Society of Clinical Oncology, held in Chicago May 30th – June 3rd, 2014.