Is Selective Neck Irradiation Safe for Node-Negative Nasopharyngeal Carcinoma?

Abstract

Because of the rich submucosal lymphatic network of the nasopharynx, early lymphatic spread is a well-known clinical feature of nasopharyngeal carcinoma. In the past, when detection of lymph node (LN) involvement was based on clinical palpation, our retrospective studies showed that up to 40% of node-negative patients subsequently developed nodal manifestation if elective neck irradiation (ENI) was withheld (1). Although 90% of such relapses could besalvagedbytherapeuticwhole-neckirradiation,thesepatientshad significantly poorer survival because of higher hematogenous dissemination than those without nodal relapse (>20% vs <6%), despite successful salvage. The possibility that occult foci in untreated LN could have been the contributing reservoir for distant metastasesistooserioustoberisked.AsENIcoveringthewholeneck toadoseof50to60Gycouldachieveexcellentnodalcontrolwithout excessive major toxicities, this becomes the standard recommendationforallN0patients.Eveninthecurrenteraofintensitymodulated radiation therapy (IMRT), protocols by the Radiation Therapy Oncology Group (RTOG) still adopt comprehensive coverage of bilateral levels I to V LN regions in clinical target volume, sparing level I only for N0 patients in the latest RTOG study 0615 (2, 3). Advances in radiology imaging have substantially improved our ability to detect nodal metastases and our knowledge about the pattern of lymphatic spread. A meta-analysis of 2920 patients from 13 studies using magnetic resonance imaging (MRI) as the main imaging modality by Ho et al (4) showed that 85% of patients had nodal involvement at the time of diagnosis. Based on the frequency and efferent lymphatic pathways, LN could be grouped into 3 levels: the high-frequency level included the retropharyngeal (RP; 69% of patients) and level II LN (70% of patients); the intermediate frequency level included levels III (45% of patients) and V (27% of patients); and the low-frequency level included level IV LN (11% of patients) supraclavicular (SCF; 3%), IB (3%), VI (2%), and parotid LN (1%). None of the patients showed positive level Ia (submental) LN at presentation. All available studies consistently showed orderly spread down the neck; the probability of skip metastases between levels varied from 0% to 8%. The report with the highest incidence was based on imaging using both MRI and fluorodeoxyglucose-labeled positron emission tomography (FDG PET) by Ng et al (5) :7 % of patients had lower level or supraclavicular LN without ipsilateral upper level LN involvement, and 1% had level V without ipsilateral level II LN. The study by Tang et al (6), which was based on MRI, showed that only 0.3% had second station LN without first station LN, and only 0.3% of patients with first station LN showed spread to level IVor SCF LN, skipping the second station. It is therefore logical to ask whether we really need to cover the whole neck for patients staged with modern imaging. Review of different IMRT series showed marked variation in the coverage of level I and the lower neck. A “reduced-volume approach” is increasingly advocated for selected patients, especially those with N0 disease. A retrospective study by Gao et al (7) of patients staged as node negative by computed tomography (CT) showed that ENI by conventional technique to 50 to 56 Gy confined to RP and levels II, III, and Va (26% of patients with additional chemotherapy) was safe; with a median follow-up of 4.5 years, therapy failed for only 0.2% patients at level IV. None had grade 3 or higher toxicity attributed to radiation therapy to the neck. The prospective phase 2 study of reduced-volume ENI in the current issue (8) provided additional affirmative information. In addition to node-negative patients, that study also included patients with N1 disease as staged by CT. For both the N0 and uninvolved sides of neck in N1 patients, ENI by IMRT to 54 Gy confined to RP and levels II, III, and Va (54% of patients undergoing additional chemotherapy) was again shown to be safe, that is, at a median follow-up of 4.9 years, therapy failed for only 0.5% (1 of 212) patients at level Vb and for another 0.5% at level Ib.