The ability to avoid critical anatomy using precise techniques to plan, guide and deliver RT is paramount in the treatment of NPC. Intensity-modulated radiotherapy (IMRT) remains the most frequently used approach for precision RT. It offers conformal (concave) shaping of tumouricidal doses and steep dose fall-off gradients, thereby providing the opportunity for safe delivery of high doses targeted to the tumour. Specific to NPC, the dosimetric advantages of IMRT have resulted in improvement in locoregional control (LRC) and overall survival (OS), and substantial reduction in incidence of late toxicity, such as blindness and xerostomia.

The ascent of IMRT as the preferred RT technique for NPC had its background in early institutional reports [10] that demonstrated substantial improvements in tumour control with IMRT, relative to historical reports using conventional RT (4-year LRC 97 % and distant control [DC] 66 %). Subsequently, a single-arm phase II trial by the Radiation Therapy Oncology Group (RTOG) [11] demonstrated a reduction in the rates of xerostomia from salivary gland sparing-IMRT. Additional randomized phase II trials showed the effect of IMRT on normal tissue toxicities and quality of life in early stage disease [12, 13] but were underpowered to detect tumour control differences. The findings were corroborated in a large randomized phase III trial [14] that compared IMRT versus 2D-conformal RT in 616 NPC patients with an OS advantage (5-year 80 % vs. 67 %) and superior LRC, particularly among T4 tumours (81.5 % vs. 62.2 %) favouring IMRT. Late grade 2 xerostomia (9.5 % vs. 27.1 %) and auditory complications (47 % vs. 89 %) were also significantly lower with IMRT.

Improved imaging modalities (MRI and PET/CT) for diagnosis and RT planning have also led to increased accuracy in these domains, accounting in part for the substantial improvements in LRC. The implementation of volumetric image-guidance (IGRT) during RT further improves setup accuracy with an opportunity for monitoring tumour volume and normal tissue changes over the course of RT, with the eventual promise of ‘real-time’ dose-volume parameter adaption to anatomic structures when necessary. Recently, particle therapy, e.g. intensity-modulated proton therapy (IMPT), has emerged as an additional attractive option. Potentially IMPT may enhance normal tissue sparing over IMRT, as evidenced by early observations of reduced toxicity [15, 16]. However, long-term clinical benefits and cost-effectiveness remain to be determined.

Progress has also been realized by combining RT with systemic therapy, especially cisplatin. Generally, stage I and II NPC are treated with RT-alone. The role of concurrent chemo-RT (CCRT) in stage II NPC is uncertain, although results of a single randomized phase III trial support its use in this subgroup [17]. Some concerns about the study include the use of an unconventional staging system for patient selection [13 % of the trial cohort had UICC/AJCC stage III (N2) disease], and a higher proportion of patients had parapharyngeal involvement and N2 disease in the CCRT-arm. These factors confound the interpretation of the available results in limited stage II disease. In truth, aggressive disease may benefit from systemic treatment intensification and may be selected by circulating EBV DNA as a biomarker. Leung and colleagues previously identified an unfavourable stage II NPC subgroup at risk of DM according to EBV DNA titers of ≥4000 copies/ml [18].

There is universal consensus that locally advanced NPC (LA-NPC) should be treated with combined chemotherapy using CCRT followed by adjuvant chemotherapy (ACT), according to the Intergroup-0099 protocol [19]. The necessity of ACT remains uncertain due to lack of compliance to ACT in numerous CCRT trials (e.g. only 50–75 % completed at least two cycles of ACT). An individual patient data (IPD) meta-analysis in NPC (MAC-NPC) reported an 18 % mortality risk reduction, with the main contribution derived from concurrent chemotherapy (risk reduction of 40 %, with only modest contributions from induction [ICT] and/or ACT) [20]. Conversely, a more recent IPD network meta-analysis, updated with contemporary trials, suggested a superiority for CCRT + ACT over CCRT-alone across all end points [21]. The need for ACT in addition to CCRT in LA-NPC has been challenged by the negative findings of a randomized phase III trial comparing CCRT + ACT versus CCRT-alone in 508 patients [22]. It is likely that not all LA-NPC patients require ACT. A recent review of 547 N2-N3 NPCs reported that the addition of ACT appears to be beneficial in N3 (reduction of DM) but not in N2 disease [23]. Consequently, efforts are focusing on risk-stratified approaches using post-RT EBV DNA titers to select patients for ACT (NCI-2014-00635). Such approaches require robust characterization and harmonization of circulating EBV DNA titer quantitation across centres [24].

ICT is an alternative approach for systemic treatment intensification [25, 26], given the putative advantage of reducing tumour bulk to facilitate critical structure sparing in RT planning and the potential to eradicate occult DM in advanced disease. However, the caveat with ICT is the consequential reduction of cisplatin dose intensity during the concurrent phase of treatment. A literature-based non-IPD meta-analyses suggested enhanced OS and reduced DM [27], while ACT may be more beneficial for LRC. However, this disparity was not observed in the preliminary analyses of the NPC-0501 [28] trial, which compared the Intergroup-0099 regime of CCRT + ACT against a ‘reverse’ scheduling Intergroup-0099 regime. Moreover, the updated IPD MAC-NPC meta-analysis [21] (19 trials, 4806 patients) also reported that an OS advantage was restricted to CCRT + ACT (HR 0.65 [0.56–0.76]) and CCRT (HR 0.80 [0.70–0.93]), but not evident with ACT (HR 0.87 [0.68–1.12]) or ICT-alone (HR 0.96 [0.80–1.16]). Two other randomized studies of ICT + CCRT against CCRT-alone did not reveal survival benefits with ICT [29, 30]. Based on current evidence, ICT in addition to CCRT remains investigational. Finally, we await mature data of a randomized trial of CCRT with or without TPF (docetaxel, cisplatin and 5-fluorouracil) (NCT01245959), as well as a Taiwanese companion study, albeit using MEPFL (mitomycin, epirubicin, cisplatin, 5-fluorouracil, leucovorin) as the induction regimen (NCT00201396).

OS is a traditional outcome end point for clinical trial design. In the first publication of the RTOG 0129 trial, Ang and colleagues [52] constructed two mortality risk groups for HPV-related OPC based on 7th edition TNM and smoking pack-years. It classified all HPV+ OPC patients, except >10 pack-year smokers with N2b-N3 disease, as the ‘low-risk’ group to be considered as deintensification trial candidates. All patients in this analysis received intensified treatment. Whether the excellent results from the intensified regimens used in the clinical trial setting would be reproducible when replaced by less intensified treatment and whether lower risk of death is a sufficient criterion for choice of deintensification remain uncertain. In addition, using smoking pack-year as the risk stratification for treatment decision-making remains problematic, since lower OS by heavy smokers may not necessarily reflect altered tumour biology and may also reflect competing mortality risk from smoking (comorbidities, second primary, severe late toxicity and social problems). It might also be due to impaired treatment tolerance due to comorbidities, as well as compromised radiotherapy efficacy (hypoxia) among current smokers. In addition, heavy smokers are less tolerant of more intensified treatment. Recently, smoking has been described as having almost no impact on disease recurrence in surgically treated patients [53].

LRC is no longer the overwhelming problem for many HPV+ OPC patients questioning the necessity for intensive local treatment in all patients since DM is the most common cause of death. Given the clinically unpredictable development of DM [54, 55], the medical community has understandably been fearful of omitting or reducing chemotherapy. O’Sullivan and colleagues performed a risk stratification analysis on an institutional cohort of prospectively compiled HPV+ OPC patients but addressed DM risk as the end point, and found that DM risk was significantly associated with T4 or N2c-N3 category diseases, while T1-3 N0-N2a and T1-3N2b < =10 pack-year smokers have minimal risk of DM and may achieve an excellent result with RT-alone. The DM risk for the T1-3N2b >20 pack-year smoker treated by RT-alone is uncertain although did suggest an adverse DM outcomes for heavy smokers. The finding of a subgroup with excellent LRC and DC finally permitted dislodgement of the deintensification impasse to commence for HPV+ OPC, directly leading to the design of the currently accruing NRG-HN002 deintensification trial (NCT02254278).

Currently several treatment deintensification trials are targeting the ‘low-risk’ HPV-related OPC cohort described above. Deintensification strategies under evaluation include substitution of cisplatin by EGFR inhibition (e.g. RTOG 1016, NCT01302834), reduction of RT dose or chemotherapy intensity [e.g. NRG-HN 002, NCT02254278, a institutional phase II trial (NCT01530997) [56]], induction chemotherapy followed by lower RT dose in good responders (e.g. ECOG 1308, NCT01084083) and TORS resection with or without adjuvant chemoradiotherapy (e.g. ECOG 3311, NCT01898494). For high-risk populations, induction chemotherapy and immunotherapy are under discussion. Potential opportunities may exist to mitigate the risk of DM through the use of TPF triplet induction regimens following evidence from an IPD meta-analysis reported by the MACH-NC group [57]. Such strategies will require HPV+ OPC specific clinical trials.

Role of post-radiotherapy neck dissection (PRND) is evolving for patient with N2-N3 diseases. A retrospective analysis of a prospective compiled study suggests that HPV+ OPC lymph node (LN) may involute more slowly and suggest that PRND may be withheld for selected cases with an incomplete radiological response with close imaging surveillance [58].