As a faithful reader of the American Journal of Otolaryngology , we read with great interest the article titled “Head and neck squamous cell carcinoma from an unknown primary site” by Mendenhall et al . We would like to compliment the authors on their large-scale analysis of the management of unknown primary head and neck squamous cell carcinoma (HNSCC). However, we found that their conclusion was somewhat arbitrary and was not supported by their statistical results. We also would like to offer our opinions on this topic that remains a matter of controversy.

First, Mendenhall et al concluded that radiotherapy (RT) alone or combined with neck dissection (ND) results in a high probability of cure. However, according to their data, the 5-year neck control rates for RT combined with ND were much better than the rate for RT alone (pre-RT ND, 93%; post-RT ND, 82%; and no ND, 73%). Multivariate analysis also revealed that patients who received an ND had better neck control and cause-specific survival. Obviously, ND is an important initial treatment for unknown primary HNSCC, and the authors should not recommend RT alone for this disease entity.

In the literature, Igidbashian et al found that nonsurgical treatment is not effective for advanced HNSCC with N3 neck metastasis and only 37.1% of their patients achieved complete response to chemoradiation. In the study of Mendenhall et al , approximately one third of unknown primary HNSCC belong to the N3 stage. Furthermore, in clinical practice, a real pathologic N classification obtained from an ND could guide future adjuvant therapy such as RT and chemotherapy. For example, Myers et al found extracapsular spread to be the most significant predictor of both regional recurrence and distant metastasis in patients with tongue cancer. Similarly, in patients who have unknown primary HNSCC with extracapsular spread, intensive regional and systemic adjuvant therapy with close follow-up may be indicated.

Second, Mendenhall et al concluded that eliminating the larynx and hypopharynx from the RT portals in their institutions after 1997 did not compromise outcome and likely reduced treatment toxicity. However, there are no data showing that their modification of RT protocol really reduced the complication because they did not mention that the 11 severe complication cases in their report were caused by comprehensive RT (before 1997) or modification of RT (after 1997). According to Table 1, their RT never covered larynx mucosa, but there were 3 supraglottic larynx mucosal failures. It is clear that we cannot justify sparing RT to the larynx by this result. Besides, according to their data, no variable was a significant predictor of local control in multivariate analysis including mucosa RT (“no” vs “yes”). If they would like to prevent RT toxicity on mucosa, they should conclude that mucosa RT is not necessary and, therefore, no mucosa will be injured by RT.

In our opinion, the ideal way to prevent unnecessary RT to innocent mucosa is to make every effort to define the primary site and decrease the diagnosis of unknown primary HNSCC. For example, Mendenhall et al mentioned that most of the unknown primaries are likely to be in the tonsillar fossa and the base of the tongue. The research of Mendenhall et al and Cianchetti et al supports this idea. and more than one third of their patients who underwent tonsillectomy had a primary tumor discovered in the tonsillar fossa. However, only 18% of patients received a tonsillectomy in the study of Mendenhall et al . Besides, today’s improved image technology can also facilitate exploring the primary site. From the reports of Roh et al and Miller et al , (18)F-fluorodeoxyglucose-positron emission tomography alone or combined with computed tomography is a useful screening method for primary tumor detection. Miller et al even found that the risk of subsequent primary tumor appears to be low in patients with a negative positron emission tomography and a negative panendoscopy (<6%). These patients could be suitable candidates for mucosa-sparing RT.

In conclusion, ND with RT is more effective than RT alone for treating unknown primary HNSCC. The idea of larynx- and hypopharynx mucosa–sparing RT still remains controversial.