Current treatment options for recurrent nasopharyngeal cancer

Despite radical irradiation, local recurrence remains a major problem for patients with NPC. The reported incidence varies from 8 to 58%, with the median at 34% [4].

Locally persistent or recurrent NPC may be due to the intrinsic radio-resistance of the primary tumor, insufficient dose, or a geographic miss during the initial radiotherapy. The results of initial treatment with radiation therapy have improved over the years due to the advances in diagnostic imaging, which help to localize the full extent of the primary tumor, the higher tumor doses delivered [6], the different regimes of fractionation, the use of advanced radiation delivery techniques, such as intensity-modulated radiation therapy (IMRT) [7‐9], and the use of concomitant chemotherapy [10].

Pretreatment evaluation of the extent of disease, including paranasopharyngeal extension, is crucial. Early detection is important as the chance of salvage varies with the extensiveness of local recurrence. In the series of Lee et al. [4] the 5-year local control rate for rT3 patients was only 11% compared to 35 and 28% for rT1 and rT2, respectively, which also translated into worse overall survival for rT3 patients (4%) as compared to for rT1 (27%). In Wang’s series [11], the 5-year survival was 38% for patients with rT1–T2 tumors and 15% for those patients with rT3–T4. Overall 5-year actuarial local control, and overall survival rank from 15 to 35%, and 8 to 41%, respectively [4, 11‐14].

These results emphasize the importance of close follow-up after primary treatment and more vigorous attempts for early detection of local recurrence by rigid nasopharyngeal endoscopy [15], CT and/or MRI. Recent studies indicate that PET using 18-fluoro-2-deoxyglucose (FDG), is more accurate in detecting local recurrence after curative (chemo) radiation than MRI [16, 17] and/or may add useful additional information to disputable findings on MRI [18, 19]. Nevertheless, we must consider the difficulty in imaging near the brain, which is always hypermetabolic, and the difficulties with local inflammation secondary to otitis media and bony necrosis of the skull base related to radiation.

In some reports there is a polarization of primary radiotherapy and reirradiation into two extremes (good results in primary treatment and poor results in reirradiation, and viceversa) which is most likely due to the successful prevention of avoidable local failures during primary radiotherapy and the natural selection of the most radioresistant tumors for reirradiation [14].

It is of interest to remark that although patients with undifferentiated carcinomas have a notably higher 5-year survival than those with squamous cell carcinomas, in some series there is no difference when considering 10-year survival [12]. This longer survival may be due to slower regrowth of undifferentiated carcinomas, evident in their longer time to local recurrence, as well as that patients with undifferentiated carcinomas are more likely to develop late distant metastases. The stage at which recurrence is diagnosed is the most significant prognostic factor, but lesions involving the skull base should be differentiated from other advanced-stage tumors when evaluating tumor control or survival [20, 21].

Furthermore, patients without evidence of intracranial invasion or cranial nerve palsy have better survival than those with such lesions (3-year survival, 31 vs. 4%) [20]. Other factors adversely affecting survival are the simultaneous presence of neck metastases and patients whose tumor recurred at 1–2 years or less after initial treatment [11, 22].

Irrespective of the total dose given during initial treatment, there is significant correlation between the dose at reirradiation and the salvage rate. The highest local control rates have been achieved with a total dose of at least 60 Gy [4, 11, 14, 20].

Nevertheless, data from retrospective studies must be interpreted cautiously when comparing treatment techniques because unrecognized factors may have been used to select the type of treatment.

Different kinds of complications have been reported in patients undergoing reirradiation, ranging from 6 to 85% [4, 11, 13, 14, 20, 22]. This enormous range can be explained by the different reirradiation doses and techniques used as well as by the fact that some authors only report severe complications. The most frequent complications include xerostomia, trismus, sensorineural hearing loss, and middle ear effusion, but also brain necrosis which has been reported up to 20% of patients [4, 11, 13, 14, 20, 22]. Fatal complication rates between 2 and 10% have also been reported [4, 12, 14, 23]. The incidence of severe complications depends on the dose received and treatment modality. Patients who were retreated with a cumulative external beam dose >100 Gy had a high incidence of severe complications: 4% for patients receiving doses less than or equal to 100 Gy compared with 39% for those patients who received doses greater than 100 Gy [12]. Lee et al. [24] reported on the largest series of 654 local recurrent NPC patients treated with reirradiation. In their series, the risk of late complications was predominantly affected by the biological effective dose of the primary treatment, while the corresponding impact of reirradiation failed to reach statistical significance, indicating higher tolerance for reirradiation when primary treatment has been carried out with better sparing of normal tissues.

External reirradiation of NPC with curative intent is often difficult due to the large number of nearby critical structures that were previously irradiated to a high dose. Conventional external radiotherapy delivered by a 10-MV photon beam with bilateral opposing fields to cover the recurrent tumor site and margins yields a higher complications rate than three-dimensional (3D) conformal radiotherapy or IMRT, which try to radiate the previously radiated areas (such as temporal lobes, temporo-mandibular joints and middle ears) at less than 60% of the prescribed dose. Thus, Chang et al. [20] observed severe complications in 23% of the conventional radiotherapy group, but only in 9% of the conformal radiotherapy group. Furthermore, with conventional external radiotherapy, 13 patients (14%) were found to have brain necrosis, and no brain necrosis occurred in patients treated with conformal radiotherapy, although the lower incidence of complications they saw in patients treated with conformal radiotherapy may also be explained by a shorter follow-up [20].

For patients with extensive disease requiring external radiotherapy, superior dose distribution and organ sparing have been demonstrated with IMRT over more conventional conformal radiation delivery techniques for salvage of NPC. Excellent local control rates have been reported using IMRT for newly diagnosed NPC [25, 26], but although Lu et al. [27] have reported good local control rate after high-dose IMRT (68–70 Gy) for recurrent NPC, the relatively short follow-up makes it difficult to evaluate late toxicities in their patients. Late toxicities are still common after IMRT, although generally most are mild. Neurological toxicity represents the main complication but it appears to be less frequent than after conventional two-dimensional (2D) or 3D conformal reirradiation [28]. Longer follow-up of patients is still needed to address more clearly the incidence and impact of late toxicities.

Because of the poor results of conventional external reirradiation, other techniques of reirradiation have been employed for locally recurrent NPC, such as stereotactic radiosurgery. Stereotactic radiosurgery is a highly precise form of radiation therapy used primarily to treat tumors and other abnormalities of the brain. Despite its name, stereotactic radiosurgery is a non-surgical procedure that delivers a single high dose of precisely targeted radiation using highly focused gamma-ray or X-ray beams that converge on the specific area or areas of the brain where the tumor or other abnormality resides, minimizing the amount of radiation to healthy brain tissue. Although stereotactic radiosurgery is often completed in a 1-day session, sometimes fractionated stereotactic treatment is applied, especially for larger tumors. This procedure is usually referred to as fractionated stereotactic radiotherapy in case more than five fractions are given.

Several series have reported satisfactory local control when stereotactic radiosurgery was used to treat recurrent NPC, as shown by Chua et al. [31] who reported a 5-year overall survival rate of 47%. The time interval from primary radiotherapy, rT stage, prior local failures and tumor volume were significant predictive factors for local control and/or survival. A radiosurgery prognostic scoring system was designed based on these predictive factors. Five-year local failure-free probabilities in patients with good, intermediate and poor prognostic scores were 100, 42.5, and 9.6%, respectively. The corresponding 5-year overall survival rates were 100, 51.1, and 0% [31].

It has been postulated that, combining stereotactic radiosurgery with high-dose-rate brachytherapy might improve the survival because the single large dose of radiosurgery may increase tumor cell kill overcoming the inherent radioresistance of cells. Low et al. [32] treated 36 patients with local recurrent NPC in stage rT1–T2 with a schedule of 18 Gy followed by two separate fractions of 6 Gy each by intracavitary brachytherapy. The actuarial 5-year disease-free survival and overall survival were 57% (78% for rT1, and 39% for rT2) and 62% (80% for rT1, and 48% for rT2), respectively. However, 44% of patients had late complications, including cranial nerve palsies (20%), temporal lobe necrosis (8%), and osteoradionecrosis of the skull base (17%).

A matched cohort study to select patients with similar characteristics treated by stereotactic radiosurgery (median dose 12.5 Gy) or intracavitary irradiation (total dose of 60 Gy) was carried out by Chua et al. [33] in 74 patients with local NPC failure, generally at stage rT1–T2. The 3-year local failure-free rate was 78% for the radioactive gold grain group compared with 68% for the stereotactic radiosurgery group, whereas the overall survival rate was better in the stereotactic radiosurgery group (3-year survival rate of 77 vs. 66%), but in both circumstances the differences were not statistically significant. Furthermore, when the impact of tumor volume on treatment outcome was adjusted, no difference in tumor control was observed between the two groups, suggesting that both salvage treatments have comparable efficacy. The incidence of complications was similar in number in both groups, but complications in the stereotactic radiosurgery group were more severe (21.6% of neuroendocrine complications, 13% of brain necrosis, 5% of cranial neuropathy, 5% of pituitary insufficiency, and severe hemorrhage from a carotid artery aneurysm in one patient versus 30% of headaches, 16% of palatal fistula, and 13% of neuroendocrine complications in the brachytherapy group) [33]. We can ask ourselves if we are making any progress with the new and more sophisticated procedures.

Stereotactic radiosurgery has also been used as a boost after reirradiation with external beam for recurrent NPC with an overall 5-year survival rate of 31% [34].

Severe complications commonly reported after stereotactic radiosurgery for NPC include massive epistaxis, nasopharyngeal necrosis, cranial nerve palsies, temporal lobe necrosis, and osteoradionecrosis of skull base. The relatively high risk of severe late complications indicates that careful patient selection and treatment planning are required. Severe complications were also reported by Kocher et al. [35], who treated 8 patients with recurrent NPC by stereotactic radiosurgery. In Kocher’s series, 3 patients died of carotid or cerebral hemorrhage after stereotactic radiosurgery using a dose of 15–24 Gy, 2 patients developed cerebral edema in temporal lobes, and 1 developed cranial neuropathy. A fatal carotid artery hemorrhage also was the cause of dead of 33% of patients with recurrent NPC treated with fractionated stereotactic radiotherapy [36]. Tumors involving the Rosenmueller’s fossa and invading deeply to the foramen lacerum are the most important predisposing factor in fatal hemorrhage [36]. It should be noted that the large doses per fraction used in most series on stereotactic radiosurgery may cause these relatively high rates of late complications [24].

As previously mentioned, fractionated stereotactic radiotherapy is a modification of stereotactic radiosurgery, which enables fractionated irradiation to be given without losing the advantage of the mechanical precision and accuracy as well as dose conformity of stereotactic radiosurgery. Compared with stereotactic radiosurgery using single fraction of high-dose irradiation, fractionated stereotactic radiotherapy may be superior in terms of tumor control and protection of normal tissues and organs surrounding the target.

Based on these principles, 56 patients with recurrent NPC (rT1–T2, 38; rT3, 18) received fractionated stereotactic radiotherapy. The median fractionated stereotactic radiotherapy dose was 48 Gy in six fractions. Three-year local failure-free survival, disease-specific survival, and progression-free survival rates were 75.1, 45.9, and 42.9%, respectively. Multivariate analysis showed that recurrent disease and large tumor volume were independent factors that predicted poorer disease-specific survival. Seventeen patients developed late complications, including two with fatal hemorrhage [37]. Severe late complications occurred in 25% of patients with recurrent NPC. Four percent of patients with recurrent disease developed massive hemorrhage in the nasopharynx after fractionated stereotactic radiotherapy and died of this complication, and 6% developed brain stem necrosis [37].

In a study on 125 NPC patients who received salvage stereotactic radiation Chua et al. [38] compared the results of stereotactic radiation using single or multiple fractions. All patients were individually matched for rT stage, and tumor volume. The median dose was 12.5 Gy in single fraction by stereotactic radiosurgery, and 34 Gy in 2–6 fractions by fractionated stereotactic radiotherapy. Local control rate was better in the last group although overall survival rates were similar (3-year overall survival rates of 66 and 61%, respectively). Finally, incidence of severe late complications was 33% in the stereotactic radiosurgery group versus 21% in the fractionated stereotactic radiotherapy group, including brain necrosis (16 vs. 12%) and hemorrhage (5 vs. 2%) [38].

The role of surgery in the management of malignant tumors involving the nasopharynx and the infratemporal fossa is supported by recent reports. The results of removal of recurrent NPC after radiotherapy failure are encouraging for T1–T3 tumors, while those patients with T4 stage generally recur locally or die due to development of metastatic disease. Nevertheless, 5 patients out of 14 with stage T4 in the Suárez et al. [44] series were alive, with a follow-up between 1 and 7 years, but none of them had intracranial invasion. At variance, all 5 patients with intracranial involvement died. In other words, the impact on survival of involvement of different areas in stage T4 is not the same, the intracranial extension being the worse prognostic factor. Although some authors have reported encouraging results in cases of intracavernous involvement [45], surgical salvage of recurrences of NPCs with significant intracranial extension usually is not justified.

Analysis of the literature on nasopharyngectomy depicts a 30–55% overall 5-year survival, with the largest series consisting of 60 patients [14, 46‐56]. The results of some of these series are shown in Table 2 [14, 47, 49, 53, 56].

Teo et al. [14] reported on the treatment results of 20 patients with locally recurrent NPC (mostly rT1 and rT2) treated with nasopharyngectomy with/without postoperative radiotherapy. This group was compared with other rT1–T2 patients treated mainly with external radiotherapy of 60 Gy or greater. Far better local control and survival was shown for those receiving nasopharyngectomy than reirradiation. However, 70% of the patients that underwent surgery also had postoperative radiotherapy due to questionable and/or unclear resection margins. For the subgroup that could be treated adequately with nasopharyngectomy alone, there is a theoretical advantage, as the risk of post-radiotherapy complications could be minimized.

In contrast, no significant differences in overall survival were observed between a group of 159 patients who received reirradiation and another group of 22 patients who underwent nasopharyngectomy with or without postoperative radiotherapy. Salvage surgery was associated with improved overall survival only in the subgroup with T1 to T2 local failure, but not in the subgroups with T3 or T4 disease [57].

Wei et al. [47] reported a 5-year actuarial tumor control rate of 62%, and a 5-year actuarial disease-free survival rate of 49% in a series of 60 patients who had a curative resection.

The importance of postoperative radiotherapy after nasopharyngectomy is stressed by King et al. [56], who found a mean overall survival of 18 months for patients without postoperative radiotherapy, whereas for patients given postoperative radiotherapy the mean overall survival was 60 months. However, postoperative irradiation failed to affect the prognosis significantly in Hsu et al.’s series [53].

When total resection is achieved local recurrence is not very common, but in addition postoperative reirradiation is recommended in patients with positive surgical margins or advanced stage [46, 47, 52]. If postoperative radiotherapy is required for positive or uncertain margins, the toxicity may be compounded and aggravated. Extensive intradural invasion, cavernous sinus involvement and pharyngobasilar fasciae invasion are considered contraindications for surgery in recurrent NPC [50].

Nasopharyngectomy can be carried out even after a second recurrence but given the small number of patients treated, it is difficult to make definitive conclusions on its effectiveness in reirradiated patients. However, it has been suggested that in appropriately selected patients, the second course of radiation therapy does not contraindicate surgical resection, and, in fact, excellent long-term outcome has been reported. Nevertheless, it increases the morbidity of the procedure, particularly skull base osteomyelitis, and great emphasis should be placed on the reconstruction of the surgical bed, including the use of free flaps [58].

Analysis of clinical prognostic factors in a series of 53 nasopharyngectomies revealed that margin status, adjuvant treatment type and parapharyngeal space involvement were significant impact factors of local control, whereas dura or brain involvement, local recurrence and adjuvant treatment type were significant impact factors of survival [59]. According to To et al. [55] all patients who developed repeat recurrence or died had a high recurrent T-stage tumor, skull base involvement, multiple recurrences, positive surgical margins, or concurrent neck node metastasis.

A number of anterior and lateral approaches to neoplasms within the nasopharynx and the infratemporal fossa has been described [46, 48, 52, 53, 60‐62]. It is necessary to remove all the related structures in the nasopharynx through a wide approach such as that used in some anterolateral approaches, due to the fact that serial whole-organ section of resected nasopharyngeal specimens showed that 90% of the tumors affected the eustachian tube cartilage, and over 90% of the tumors showed extensive submucosal infiltration [63].

For tumors arising from the nasopharynx and extending to neighboring regions, such as nasal cavity, maxillary sinus, sphenoid sinus, and infratemporal fossa, the facial translocation approach can be the simplest and most direct way to expose these areas and to facilitate extensive tumor removal. If the tumor is confined to the nasopharynx or has paranasal, retromaxillary or moderate infratemporal extension, a temporary removal of the anterior, superior and medial walls of the maxillary sinus will suffice (Fig. 1). The procedure is carried out with midfacial degloving in order to avoid facial scars, which in addition allows an excellent exposure [44, 58, 61, 64, 65]. In extensive tumors of the nasopharynx and the infratemporal fossa, especially when the tumor involves the nasal cavity and the posterior wall of the maxillary sinus, a standard facial translocation approach is the method of choice, allowing the surgeon a better exposure than that obtained by other infratemporal fossa procedures, such the subtemporal–preauricular approach (Figs. 2, 3). In this case the procedure is carried out with midfacial degloving and a hemicoronal–preauricular incision. When a facial translocation is performed, the possibility of osteomyelitis or late bone resorption exists [64]. Open standard facial translocation carries very noticeable sequelae, such as lower lid ectropion, epiphora, medial canthus misalignment, nasolacrimal duct obstruction, frontal muscle palsy, and visible facial scars. Most of these sequelae, with the exception of nasolacrimal obstruction, can be avoided with the use of midfacial degloving combined with a hemicoronal preauricular incision that is hidden in the hair.

The subtemporal–preauricular approach provides a direct access to the nasopharynx and related areas, including the petrous bone, clivus, infratemporal fossa, retropharyngeal area, apex of the orbit, and the sphenoid and maxillary sinuses [62, 66] (Figs. 4, 5). This approach is very versatile, and it may be easily combined with other skull base procedures in selected cases. When necessary, additional drilling of the lateral wall of the orbit, the adjacent greater wing of the sphenoid on the anterior pole of the temporal lobe, and the roof of the infratemporal fossa, permits the exposure of extradural tumors involving the floor of the middle fossa [66].

Transoropalatal, mandibular swing, and maxillary swing approaches are less used now for surgical resection of nasopharyngeal lesions. Generally, the transoropalatal approach is suitable for small tumors centrally located in the midline position of the nasopharynx because the access to the lateral wall of nasopharynx is restricted. This approach causes the least injury and does not affect the face. However, the surgical exposure is usually not satisfactory.

The mandibular swing approach is suitable in patients with severe trismus that makes the transoral approach to access the nasopharynx inadequate or in those patients with oropharyngeal extension (Fig. 6). Moreover, parapharyngeal disease can also be dealt with by this approach. On the other hand, the surgical field is narrow, and secondary injury is often severe.

Wei et al. [46] adopted an anterolateral approach to the nasopharynx with the maxilla swung laterally to provide adequate exposure of the nasopharynx, having the drawback of producing noticeable facial incisions. The posterior nasal septum can also be removed for better exposure of the nasopharynx and the paranasopharyngeal space without the necessity of removing the orbital floor. However, the exposure of the nasopharynx is somewhat limited to one side of the nasopharynx. In addition, the maxillary bone needs to be freed, and the hard and soft palate have to be split. This approach can potentially result in destructive complications such as maxillary necrosis, palate fistula, and trismus.

The unsatisfactory surgical exposure and the relatively high frequency of surgical complications using open techniques led to the development of minimally invasive surgical methods that overcomes the limitations of conventional surgery. Currently, use of endoscopic nasopharyngectomy for tumor resection has been reported for a limited number of cases. This novel technique has proven to be minimally invasive and safe for the treatment of recurrent NPC with encouraging short-term outcomes, but further long-term follow-up of the patients is required. Thus, Chen et al. [67] carried out an endoscopic nasopharyngectomy in 37 recurrent NPC, almost all of them with rT1–T2 tumors. No severe complications or deaths resulting from the operation were observed, with the exception of secretory otitis media, which occurred in 22% of the cases. The 2-year overall survival rate, local relapse-free survival rate, and progression-free survival rate were 84, 86, and 83%, respectively. Robotic assisted endoscopic nasopharyngeal resection may play a very important role in these difficult situations.

Postoperative mortality in all published series is less than 6%, and generally between 0 and 3% [14, 47, 49, 52, 53, 55, 65]. Death can be caused by massive hemorrhage as a result of internal carotid artery injury, and intracranial complications such as meningitis. The most common complications of nasopharyngectomy are palatal fistula, trismus, otitis media with effusion, wound infection, skull base osteomyelitis, and rupture of the internal carotid artery, with complication rate in up to 50% of patients. Nevertheless, postsurgical complications are less in number and severity than postirradiation ones.