Unlike other head and neck cancers, radiotherapy instead of surgery is the mainstay of treatment of NPC. NPC is highly radiosensitive and radiotherapy is the backbone of treatment for all stages of NPC without distant metastases.

Conventional 2 dimensional (2D) radiotherapy for NPC depends heavily on large lateral opposing fields. One of the most common radiotherapy approaches for NPC is to start phase I treatment with large lateral opposing facio-cervical fields that cover the primary tumor and the upper neck lymphatics in one volume, with matching lower anterior cervical field for lower neck lymphatics. The brainstem, spinal cord, eyes and oral cavity are protected by shieldings. When the spinal cord dose reaches 40-45 Gy, there are 2 options for phase II treatment. Treatment can either be changed to lateral opposing facial fields with anterior facial field for the primary tumor, with matching anterior cervical field for the neck lymphatics. Alternatively, treatment can be continued using the lateral opposing facio-cervical fields but with shrinkage of fields to avoid the spinal cord, and by treating the superior-posterior lymphatic with electron fields (61, 62). The major objection to treating the primary tumor and the neck lymphatic in 2 separate volumes (both of these phase II treatment techniques) is that there is a danger of underdosing the paranasopharyngeal extension of the tumor and the upper neck nodes at the junction between the primary tumor and neck lymphatic target volumes. Thus a paranasopharyngeal boost with a posterior oblique field is often employed. Radical dose of 65-75 Gy is normally given to the primary tumor and 65-70 Gy to the involved neck nodes, while the dose for prophylactic treatment for a node-negative neck is 50-60 Gy.

Conventional 2D radiotherapy successfully controlled T1 and T2 tumors in between 75% to 90% of cases, and T3 and T4 tumors in 50% to 75% of cases (63, 64). Nodal control is achieved in 90% for N0 and N1 cases, but the control rate drops to 70% for N2 and N3 cases (63). As interrupted or prolonged treatment reduces the benefits of radiotherapy, every effort should be made to maintain the treatment schedule (65).

Because of the high incidence of occult neck node involvement, prophylactic neck radiation is usually recommended (66). Good loco-regional control should be the prime objective of treatment, as loco-regional relapses represent a significant risk factor for the development of distant metastases (67).

For T1 and T2 tumors, a booster dose using intracavitary brachytherapy improved tumor control by 16% (64). Although stereotactic radiosurgery has also been used for the booster dose (68, 69), it is probably better reserved for the treatment of persistent and recurrent NPCs, because of the undesirable side effects associated with hypofractionated treatment (70).

Radiotherapy for NPC is challenging because the NP is anatomically surrounded by an array of radiosensitive structures like the brain stem, spinal cord, pituitary-hypothalamic axis, temporal lobes, eyes, middle and inner ears, and parotid glands. As NPCs tend to infiltrate and spread towards these normal organs, the irradiation target volumes in NPC are very irregular. With conventional lateral opposing fields, protection of adjacent radiosensitive organs while giving high dose to targets is difficult. Often only the critical organs like brainstem, optic pathway and spinal cord are safeguarded with adequate shieldings to avoid unacceptable radiation toxicities while non-critical organs like the parotids and auditory pathway are sacrificed during radiotherapy. For patients with early disease, since they have a good chance of survival, radiation toxicities in these even non-critical structures would affect the quality of life of survivors. However, for patients with locally advanced disease like those with skull base infiltration or intracranial extension, the challenge lies in achieving adequate tumoricidal dose to the primary without overdosing the critical organs.

The major limitations of conventional 2D radiotherapy for NPC can now be overcome with 3 dimensional (3D) conformal radiotherapy and IMRT (71, 72). IMRT is an advanced form of 3D conformal radiotherapy, conforming high dose to tumor while conforming low dose to normal tissues. 3D conformal radiotherapy employs multiple beams conforming to the shape of the target. In additional of using multiple shaped conformal beams, IMRT also allows for fine modulation of radiation intensity within each radiation beam. Thus, in effect, there are thousands of beamlets, each with calculated intensity to deposit a defined dose at each specific point. A good therapeutic ratio can be achieved by giving a high dose to the tumor to achieve high local control probability while keeping down normal tissue complications by limiting radiation dose to normal tissues. Also, IMRT planning and dose optimization is fully computerized, a process known as inverse planning, thus it is much preferred over the more expertise dependent forward planning in 3D conformal radiotherapy.

The use of IMRT in treatment of NPC have multiple advantages. Firstly, IMRT can be used for organ preservation, e. g. sparing the parotids of high dose radiation will preserve salivary function after radiotherapy; Secondly, in the case of extensive tumors, and when the tumor extension was close to the dose-limiting organs, IMRT can achieve good dose differential between the tumor and the dose-limiting organs (73, 74), and thus can achieve high dose in tumor without overdosing the normal organs. This also opens up a therapeutic window for dose escalation in the tumor to improve local control. Thirdly, IMRT allows differential doses to be given to different targets/organs simultaneously, thus different targets/organs can receive different fractional dose at the same fraction of treatment. As the fractional dose will affect the biological effectiveness of radiation (75), there is a component of biological modulation of radiation besides just modulating the physical radiation dose in IMRT. Simultaneous modulated accelerated radiotherapy (SMART) employs this principle for accelerated radiotherapy with IMRT (76). Fourthly, IMRT resolves the problem of dose uncertainty at the junction between the primary tumor and neck lymphatic target volumes in conventional radiotherapy. It enables the primary tumor and the upper neck nodes to be treated in one volume throughout. For the above reasons, IMRT also eliminates the need of boost and hyperfractionation. Radiotherapy course can be shortened and is more efficient.

Different series reported excellent local control of more than 90% in NPC achieved with IMRT (77-81), even among patients with advanced T3-4 diseases (82). Reports also shown preservation of salivary function and improve quality of life of survivors after IMRT (83-85). A recent multicentre study also showed that the excellent 90% local control rate with IMRT as reported from single institutions are reproducible in multi-institutional setting (86). Thus, IMRT has gradually evolved as the new standard of care for NPC.

While modern radiotherapy like IMRT achieves good local control, distant metastases become the predominant pattern of failure, especially among those with loco-regionally advanced disease. NPC is also chemosensitive. There is a long history of clinical studies investigating combined radiotherapy (RT) and chemotherapy for NPC. Most of the early studies were non-randomized and the results are conflicting or inconclusive (87-96). Up to the year 1998, there were only 5 reported randomized trials on combined chemotherapy and RT for NPC (87, 90, 92, 95, 96). Each trial used a different approach with respect to drug combination, time sequence of chemotherapy and RT, and RT technique and dose. In the early randomized trials, induction chemotherapy is the most often studied combination (90, 92, 96). The rationale for induction chemotherapy is to reduce tumor load of loco-regional disease before start of RT and also early use of systemic treatment for eradication of micro-metastases. The only induction trial that showed a significant improvement of disease-free survival was the International Nasopharyngeal Carcinoma Study Group (92), using combination of bleomycin, epirubicin and cisplatin. However, there was no improvement in overall survival. This discrepancy in findings for disease-free and overall survival may be due to the increased treatment-related death among patients on induction chemotherapy which would offset the benefit of chemotherapy in reducing disease-related death.

A pivotal study was reported by the Head and Neck Intergroup in 1998, using concurrent RT with cisplatin (100 mg/sqm D1, 22, 43) followed by adjuvant cisplatin and 5-fluouracil (5-FU) (cisplatin 80 mg/sqm D1 and 5-FU 1,000 mg/sqm /D, D1-4, Q4 weeks cycles for 3 cycles) (95). Compared with RT alone, chemoradiation significantly improved progression free survival and overall survival. The pattern of disease failure showed reduction of both loco-regional and distant failure with chemoradiation.

There was initial skepticism of the Intergroup results and doubts if the results were applicable to NPC in the endemic areas. After report of the Intergroup 0099 Study, randomized trials using similar design were performed in endemic regions in Asia, to validate the Intergroup results. Three randomized trials were subsequently reported from Hong Kong (97), Singapore (98), and China (99) respectively. The study from HK used the same chemotherapy regime as the Intergroup study and showed improved failure-free survival mainly due to improved local control with chemoradiation. However, distant control and overall survival showed no significant difference (97). On the other hand, the Singapore and China studies both used a variation of dosing of cisplatin from the Intergroup regime, employing lower dose of cisplatin but given more frequently. The Singapore and Guangzhou trials both showed significant reduction of distant metastases and improvement of both disease-free and overall survival with chemoradiation (98, 99).

The meta-analysis of chemotherapy in nasopharyngeal carcinoma (MAC-NPC) study of the MAC-NPC Collaborative Group is the only meta-analysis that used an individual patient data design (100). As reported in 2006, the MAC-NPC study included 8 randomized trials which had completed accrual before end of 2001 and thus excluded the more recent trials from Asia. In the meta-analysis, there were 4 trials that investigate induction chemotherapy (+ adjuvant chemotherapy in 1 trial), 3 trials that investigate concurrent chemoradiotherapy (+ adjuvant chemotherapy in 2 trials) and 1 trial that investigate adjuvant chemotherapy alone. Overall, an absolute survival benefit of 6% at 5 yr from addition of chemotherapy was observed (from 56% to 62%). A significant interaction was observed between the timing of chemotherapy and overall survival, with the highest benefit resulting from concurrent chemoradiation. The results concur with findings from meta-analysis on other head and neck cancers also.

There is now general agreement that the positive results reported in the Intergroup 0099 study are applicable to the NPC in the endemic areas, but the conflicting evidence of chemotherapy on local control and distant metastases have generated discussion. The conclusion seems to be that, of the three basic approaches tested in these studies (induction, concurrent, and adjuvant chemotherapy), concurrent chemoradiotherapy is the most efficacious. There were 3 randomized trials that evaluated adjuvant chemotherapy and all were negative (87, 101, 102). Thus, there is still debate on whether the adjuvant chemotherapy in the Intergroup regime can be omitted.

Following concomitant chemoradiation for nasopharyngeal carcinoma, isolated failure in the neck was much reduced and it was less than 5% (122). If cancer persists or recurs in the cervical lymph nodes as evidenced by imaging studies or clinical progression of the lymph nodes, salvage therapy is indicated. When external radiotherapy was employed as the salvage option, the overall 5-yr survival rate was 19.7% (123). Surgical salvage in the form of radical neck dissection could achieved a 5-yr tumor control rate of 66% in the neck and a 5-yr actuarial survival of 38% (124). When tumor in the neck lymph node exhibits extracapsular spread, brachytherapy should be applied to the tumor bed following radical neck dissection. With this adjuvant therapy, a similar tumor control rate has been achieved as when radical neck dissection was carried out for less extensive neck disease (125).

When the patient was detected to have residual or recurrent tumor in the nasopharynx after radiation or chemoradiation, this can be managed with a second course of external radiotherapy. This radiation dosage should be greater than the initial one and although a salvage rate of 32% has been achieved, the cumulative incidence of late post-reirradiation sequela was 24%, with treatment mortality of 1.8% (126). In view of the high incidence of complications resulting from the second course of radiation, stereotactic radiotherapy, brachytherapy and surgical resection have been employed for the salvage of small localized tumors in the nasopharynx. Stereotactic radiotherapy under these circumstances have a 72% 2-yr local tumor control rate (77). The number of patients treated with this method was small and long-term follow-up information is not available (127).

With the application of brachytherapy, the radiation dose decreases rapidly from the radiation source, enabling a high dose of irradiation to be delivered to the residual or recurrent tumor but a much smaller dose to surrounding tissue. Brachytherapy also delivers radiation at a continuous low dose rate, which gives it a further radiobiological advantage over fractionated external radiation. Intracavitary brachytherapy has been used for nasopharyngeal carcinomas (128). The radiation source was placed either in a tube or a mould before insertion into the nasopharynx. In view of the irregular contour of the nasopharynx and the uneven surface of the recurrent or residual tumor, it is difficult to apply the radiation source accurately to provide a tumoricidal dose. To circumvent this problem, interstitial radioactive implants have been used to treat those small localized residual or recurrent tumor in the nasopharynx (129).

The most frequently employed radiation source for this purpose is the radioactive gold grains (¹⁹⁸Au). These gold grains can be implanted into the tumor in the nasopharynx either transnasally or using the split-palate approach (130). The latter approach gives the surgeon a direct view of the tumor and enables the implantation of the radioactive gold grains into the tumor precisely to give a better dosimetry of radiation. For those small tumors localized in the nasopharynx, without bone invasion, this method has provided effective salvage with minimal morbidity (131). The efficacy of the gold grain implants to treat persistent and recurrent tumors after radiotherapy was reported to give a 5-yr local tumor control rates of 87% and 63% respectively. The corresponding 5-yr disease-free survival rates were 68% and 60% respectively (132). Other treatments for management of localized disease with intracavitary brachytherapy have also been reported with success (133, 134).

Surgical resection of the residual or recurrent tumor in the nasopharynx is another salvage option. It is indicated when the localized disease cannot be managed by brachytherapy either it is too extensive or it is located at a position, such as the cartilage of the Eustachian tube crus where gold grains cannot be implanted. When the tumor has extended to the paranasopharyngeal space, then surgical resection is indicated. Nasopharyngectomy in selected patients with localized disease is a good option of salvage. The nasopharynx, is located in the middle of the head and because of its awkward position, the exposure of the tumor in the nasopharynx for oncologic extirpation has been a difficult technical challenge. A number of approaches have been reported, including the anterior approach via the Le Forte I or the midfacial deglove route, an infratemporal approach from the lateral aspect (135), transpalatal, transmaxillary and transcervical approaches from the inferior aspect (136, 137), and an anterolateral approach (138). The mortalities associated with these salvage surgical procedures have been low, and as all the patients concerned had previously undergone radical radiotherapy, some patients develops trismus and palatal fistula. These morbidities lead to some degree of inconvenience but were still acceptable (139). With modification of surgical techniques, the palatal fistula rate associated with the maxillary swing approach has been marked reduced (140). As long as the residual or recurrent tumor can be removed adequately, the long-term results have been satisfactory. The 5-yr actuarial control of tumors in the nasopharynx has been reported to be around 65% and the 5-yr disease-free survival rate is around 54% (141, 142). In recent years, there were reports of removal of small recurrent tumor with the help of the endoscope (143, 144). The tumor has to be located at appropriate site in the nasopharynx before an oncologic resection can be performed for these patients and so far the number reported were small (145).

For more advanced or infiltrative tumors, a second course of external radiotherapy is required (146). A second course of external radiotherapy administered concurrently with chemotherapy has been tried; this was built on the experience gained from the use of concurrent chemoradiotherapy in primary treatment. This treatment has been reported to give a 5-yr actuarial overall survival rate of 26%, though the risk of major late toxicities was significant (147). The use of precision radiotherapy such as IMRT may improve the therapeutic ratio for local control, promising initial results have been reported (148), but distant metastases will remain a major issue for patients suffering from local relapse.