Annals of Oncology 13: 1007–1015, 2002 DOI: 10.1093/annonc/mdf179
Review Nasopharyngeal carcinoma A. T. C. Chan*, P. M. L. Teo & P. J. Johnson Prince of Wales Hospital, Sir Y. K. Pao Cancer Center, Chinese University of Hong Kong, People’s Republic of China Received 20 December 2001; revised and accepted 8 February 2002
Nasopharyngeal carcinoma (NPC) is endemic in southern China where genetic abnormalities and Epstein–Barr virus (EBV) infection are critical in the pathogenesis of the disease. Circulating EBV-DNA has been shown to improve prognostication and monitoring of NPC patients. Radiotherapy is the mainstay treatment for early disease and concurrent cisplatin/radiotherapy has been demonstrated to prolong survival in locoregionally advanced disease. Ongoing studies of targeting agents and immunotherapeutic approaches may further improve treatment results. Key words: Epstein–Barr virus, nasopharyngeal carcinoma, review, treatment
Introduction Nasopharyngeal carcinoma (NPC) occurs sporadically in the west but is endemic in southern China where it is the third most common form of malignancy amongst men, with incidence rates of between 15 and 50 per 100000 [1]. There is an intermediate incidence in Alaskan Eskimos and in the Mediterranean basin. The geographical pattern of incidence suggests a unique interaction of environmental and genetic factors. A stepwise progression of histological features that reflect underlying genetic events has recently been described. Patches of dysplasia are the earliest recognizable lesions, presumably in response to some environmental carcinogen. These are associated with allelic losses on the short arms of chromosomes 3 and 9 that result in inactivation of several tumor suppressor genes, particularly p14, p15 and p16 [2–5]. The relevant carcinogens have not been established but a link between the consumption of Chinese salted fish and other salted food items with the development of NPC has been suggested [1]. These dysplastic areas are the origin of the tumor but are probably insufficient in themselves to lead to further progression. At this stage latent Epstein–Barr virus (EBV) infection becomes critical and leads to the development of severe dysplasia. Gains of genes on chromosome 12 and allelic loss on 11q, 13q and 16q lead on to invasive carcinoma; metastasis is associated with mutation of p53 and aberrant expression of cadherins (Figure 1) [6, 7]. Nasopharyngeal carcinomas are epithelial neoplasms. Three histopathological types are recognized in the World
*Correspondence to: Dr A. T. C. Chan, Department of Clinical Oncology, Prince of Wales Hospital, Chinese University of Hong Kong, Shatin, N.T. Hong Kong, People’s Rebublic of China. Tel: +852-2632-2166; Fax: +852-2648-7097; E-mail:
[email protected] © 2002 European Society for Medical Oncology
Health Organization (WHO) classifications [8]. Type I is squamous cell carcinoma (SCC) with varying degrees of differentiation. Type II is non-keratinizing carcinoma and type III is undifferentiated carcinoma. WHO types II and III can be considered together as undifferentiated carcinoma of the nasopharyngeal type (UCNT). The histological types may be of prognostic significance with UCNT having a higher local control rate after treatment with radiotherapy than keratinizing SCC and UCNT has also been shown to fail more distantly than locally [10, 11].
Presentation, imaging and staging The most common presenting symptom is cervical lymphadenopathy, followed by nasal, aural and neurological symptoms. Only 5% of patients present with distant metastases in series from Southern China [12, 13]. Once the diagnosis is suspected on clinical grounds, histological confirmation of the diagnosis is mandatory. The technique of biopsy under local anesthesia has been found to have a diagnostic sensitivity comparable to that obtained by examination under general anesthesia. The biopsy is facilitated by direct visualization of the nasopharynx with a fiberoptic nasopharyngoscope. However, since the biopsy may cause soft tissue swelling and/or a hematoma, computed tomography (CT) scan and magnetic resonance imaging (MRI) of the nasopharynx and the skull base should be undertaken before the biopsy. The primary tumor extent should be evaluated by both CT scan and MRI. The latter is more sensitive than CT scan for the detection of the primary tumor, its direct soft tissue extent, regional nodal metastasis and perineural extension. Blood vessels are clearly shown by MRI even without the use of intravenous contrast. On the other hand, although MRI can also demonstrate erosion into the base of the skull by virtue of the change in signal of fatty bone marrow, CT scan is
1008
Figure 1. Proposed tumorigenesis model for nasopharyngeal carcinoma (K.W. Lo and D. P. Huang, personal communication).
generally considered a better tool for defining bone erosion. The role of positron emission tomography (PET) scanning in NPC remains to be defined, although preliminary reports indicate that it can be useful in detecting both local failures after treatment and distant metastases. Prior to 1997, several different stage classifications were used but that described by Ho [1] was found to be superior to the others in its ability to predict prognosis and treatment outcome [12]. However, Ho’s classification was not ideal as an international system because it comprised five overall stages (instead of the usual practice of four), there were only three T-stages and it did not take into account CT scan evidence of tumor infiltration of the parapharyngeal region, a factor of considerable prognostic significance [13]. In 1997, therefore, a new UICC/AJCC stage classification was formulated, which incorporated all the major prognostically significant tumor parameters (Table 1). It is noteworthy that tumors infiltrating the parapharyngeal region were associated with a higher rate of both local failure and distant metastasis; such cases were classified as T2b (Table 1). The presence of orbital, infratemporal fossal and hypopharyngeal disease was grouped together with the presence of cranial nerve(s) palsy and intracranial tumor extension as T4. The poor prognosis of supraclavicular nodal metastases was recognized and classified as N3, together with very large nodes (>6 cm) (Table 1).
Prognosis and molecular markers NPC is one of the very few common cancers in which cure can be anticipated even in patients with advanced disease. The prognosis is related to the disease extent as measured by the UICC staging system, the type of histology and, as emphas-
ized by O’Sullivan et al. [14], the extent to which patients have access to an experienced treatment team with access to modern oncological therapeutics. It seems likely that in the near future that the level of EBV-DNA, which appears to be prognostic independent of any of the above-mentioned factors, will become routine and permit even more accurate prognostication. The demonstration that tumor-derived DNA is detectable in the plasma and serum of cancer patients raised the possibility that non-invasive detection and monitoring of NPC may be feasible. Using real-time quantitative PCR, cell-free EBVDNA was found in the plasma of 96% of NPC patients and 7% of controls. Advanced-stage NPC patients had higher plasma EBV-DNA levels than tumors with early-stage disease [15]. Further studies have demonstrated that EBV-DNA may be a valuable tool for monitoring NPC patient response during radiotherapy and chemotherapy [16], as well as early detection of tumor recurrence [17]. In a cohort of 139 patients NPC patients treated with a uniform radiotherapy technique and followed up for a median period of 5.55 years, serum circulating EBV-DNA was found to be a significant prognosticator associated with NPC-related death in a Cox’s regression analysis with a relative risk of 1.6 for each 10-fold increase in serum EBV-DNA concentration [18]. Thus the quantitation of EBV-DNA appears to allow improved prognostication of NPC. The sensitivity and specificity also suggests the potential use as a screening test in areas where NPC is endemic.
Radiotherapy Up to the early 1990s, radical radiotherapy for NPC was delivered by two-dimensional (2D) techniques such as the one
1009 Table 1. Staging criteria: UICC 1997 system Nasopharynx (T) T1
Nasopharynx
T2
Soft tissue of oropharynx and/or nasal fossa
T2a
Without parapharyngeal extension
T2b
With parapharyngeal extension
T3
Invades bony structure and/or paranasal sinuses
T4
Intracranial extension, involvement of cranial nerves, infratemporal fossa, hypopharynx, orbit
Regional lymph node (N) N1
Unilateral metastasis in lymph node(s), ≤6 cm in greatest dimension, above supraclavicular fossa
N2
Bilateral metastasis in lymph node(s), ≤6 cm in greatest dimension, above supraclavicular fossa
N3
Metastasis in lymph node(s), >6 cm in dimension, in the supraclavicular fossa
Distant metastasis (M) M0
No distant metastasis
M1
Distant metastasis
Stage grouping Stage 0
T in situ N0 M0
Stage I
T1 N0 M0
Stage IIA
T2a N0 M0
Stage IIB
T2b N0 M0 T1, T2a, T2b N1 M0
Stage III
T3 N0, N1 M0 T1, T2, T3 N2 M0
Stage IVA
T4 N0, N1, N2 M0
Stage IVB
Any T N3 M0
Stage IVC
Any T Any N M1
described by Ho [1]. The conventional practice had been to deliver tumoricidal radiation doses (total 60–70 Gy; 2–2.5 Gy per fraction in a 6–7 week course) to anatomical structures at risk of tumor invasion in the vicinity of the nasopharynx by two lateral opposing fields or multiple fields. Appropriate shieldings were positioned at predetermined distances from bony landmarks [1] to protect vital neural organs. The neck was separately irradiated by another portal with avoidance of midline structures such as the spinal cord and the larynx [1]. With two-dimensional planning techniques, the local control rates for NPC were in the order of 80%, taking all T-stages together [13, 19]. In our experience, the overall survival (OS) figures after radiotherapy, using Ho’s technique, were 85% for Ho’s stages I and II and 55% for Ho’s stages III and IV (Figure 2) [13]. With advances in technology, the modern radiotherapy for NPC should be that of three-dimensional conformal (3DCRT) or intensity-modulated (IMRT) with inverse radiotherapy planning. Researchers at the University of Californian at San Francisco, Stanford University, University of Texas M.D. Anderson and Memorial Sloan–Kettering Cancer Centers [20] have reported superior local control using such techniques
when compared with standard 2D methods. First, the success of 3DCRT or IMRT depends on better delineation of the tumor target [gross tumor volume (GTV)] by CT scan and MRI, images of which can be co-registered, such that ‘geographical misses’ are largely avoided. Secondly, there is clear definition of the vital (mostly neural) organs in the vicinity of the NPC such that these organs are spared a heavy radiation dose, thus minimizing complications. In general the clinical target volume (CTV) should include the whole GTV and the structures in the vicinity of the tumor, which are at substantial risk of subclinical infiltration. The sphenoid floor, the medial aspect of the greater wings of the sphenoid (and the foramin ovale, rotandum and lacerum), the vomer, the posterior choanae, the pterygoid plates, the pterygopalatine fossa, the posterior wall of the maxillary sinus, the parapharyngeal spaces bilaterally [21] and the prevertebral muscles and fascia are all at risk of tumor infiltration and should be included in the CTV. In T3 that infiltrates the clivus and T4 lesions, the entire clivus should be included in the CTV. However, in T1, T2 and less extensive T3 cases sparing the clivus, there has been no consensus on how much thickness of the clivus, if any at all, should be included in the CTV.
1010
Figure 2. Treatment results by Ho’s overall stage [13].
Provided that the planning target volume (PTV) is not drawn too near to the brainstem (as described later), we recommend that the cortex of the clivus in juxtaposition to the tumor should be included in the CTV. In some T4 cases, the tumor has grossly infiltrated the inferior (or even the superior) orbital fissure and the whole bony orbit on that side should be included in the CTV. Intracranial extension via the foramen ovale when the tumor infiltrates laterally and superiorly through the pterygoid muscles is frequently associated with trigeminal nerve palsy. In such cases, the whole infratemporal fossal contents and the greater wing of sphenoid on the side of the lesion should be included in addition to the intracranial component of the cancer. Occasionally the tumor may infiltrate submucosally inferiorly to involve the oropharynx or even the hypopharynx. In these situations, the CTV has to be enlarged substantially in the inferior direction. The PTV should, ideally, include the CTV with a safety margin that adequately caters for systemic and positional (set-up) errors (which can vary from center to center). Usually, a 5 mm safety margin should be adequate. However, the addition of safety margins in the posterosuperior direction on the CTV is hindered by the proximity of critical neural organs such as the brainstem, the spinal cord and the optic chiasma. To facilitate maximal dose sparing, we recommend that the PTV be drawn not closer to 5 mm of the critical neural organs. In the very advanced cases where the CTV is already within 5 mm for the critical neural organs, a phasic reduction in the PTV is required during the course of radiotherapy to avoid severe neurological sequelae. Although the overall local control rate of NPC (all T-stages together) has been improved from 80% to 90% after using 3DCRT or IMRT, the major benefit is likely to be in the advanced T-stages (T3 and T4). The early T-stages were
usually adequately irradiated with 2D-planning methods, with little chance of geographical misses [1, 19], even though conventional 2D-planning methods such as the Ho technique [1] have been shown to adequately circumscribe, at a high radiation dose, only GTV but not CTV or PTV (as described above) [20]. Indeed, when 2D external radiotherapy was supplemented by intracavitary brachytherapy, long-term local tumor control as high as 94% was reported for T1 and T2a [22]. For the more advanced T-stages, local failures occurred in onethird to two-thirds of cases after conventional 2D-planning methods [13, 19]. These patients should benefit most from 3DCRT or IMRT in terms of improvement in long-term local control by avoidance of geographical misses. On the other hand, the major benefit of 3DCRT/IMRT in the early T-stages should be reduction of severe late radiation complications such as chronic xerostomia, which detracts significantly from the quality of life of the long-term survivors of the disease.
Altered fractionation In addition to improved radiotherapy techniques, use of altered fractionation and radiation dose escalation have been reported to improve the local control. Although a Radiation Therapy Oncology Group (RTOG) trial [23] has proved the superiority of both concomitant boost (accelerated hyperfractionated radiotherapy) and hyperfractionation over the conventional daily fractionation (2 Gy per fraction, five fractions per week) for head and neck cancers in general, the benefit for NPC has not been addressed specifically. Subgroup analysis for NPC was not possible in the RTOG trial due to the small numbers of NPC cases. Recently, we have reported a significant increase in neurological complications, especially temporal lobe encephalo-
1011 pathy and cranial nerve(s) palsy, after a late-course ‘bid’ hyper-/accelerated fractionated radiotherapy in a randomized comparison with conventional daily fractionation [24]. The temporal lobe and some other neurological complications arose despite keeping the interfraction time interval to ≥6 h. These observations have led us to conclude that the sublethal damage repair half-life of the central nervous tissue is likely to be longer than previously thought [24]. Clearly, the routine practice of a ‘bid’ radiotherapy regimen together with a 2Dplanning method should be avoided unless specific measures to avoid irradiation to neural organs are implemented [24]. This precaution is especially relevant to the advanced T-stage NPC, the tumor target of which is often in very close proximity to major neural organs such as the optic chiasma and the brainstem. On the other hand, improved local control by treating six fractions per week rather than five fractions per week has been recently reported [25]. By keeping most interfraction intervals to 24 h, the problem of inadequate sublethal damage repair of neurons of the ‘bid’ technique can be avoided. A definite relationship between total radiation dose and the local tumor control has been established in early T-stage NPC when the effect of dose escalation by intracavity brachytherapy after 66–70 Gy of external beam radiation was studied [22]. However, brachytherapy is unable to deliver a significant dose to bulky parapharyngeal infiltration significant skull base involvements, or intracranial extensions, due to the geometrical dose fall-off with distance from the radioactive sources. Thus, the bulky T2b and the T3 and T4 cases benefit little from this approach. However, if the dose–tumor response relationship above 66–70 Gy demonstrated in early T-stage NPC is also applicable to the advanced T-stage, dose escalation above this level by means other than intraluminal brachytherapy should still be potentially beneficial in enhancing the local control of T3 and T4 disease. Studies using IMRT/
3DCRT/stereotactic fractionated radiotherapy (SRT) to ‘boost’ up the total dose of the advanced T-stage NPC may effectively and significantly improve the local control of advanced T-stage NPC, but the final goal should be an increased therapeutic ratio when the trade off should not be an increase in radiation toxicities, especially chronic neural toxicities.
Combined modality treatment for locoregionally advanced disease Although the initial remission rate is substantial with radiotherapy alone even in locoregionally advanced, UICC stages III and IV disease, the subsequent rates of both local and distant failures are high. Since NPC is highly chemosensitive, efforts have been made to incorporate chemotherapy into the primary treatment of the disease. Following encouraging response rates to platinum-containing regimens in phase II studies in patients with metastatic disease, the use of neoadjuvant and adjuvant chemotherapy, combined with radiotherapy has been investigated in patients with locoregionally advanced disease in five prospective randomized trials (Table 2) [26–30]. None of these trials demonstrated an improvement in OS. Although the International NPC Study Group trial showed a significant improvement in progression-free survival (PFS) [28], this was only achieved at the expense of an 8% treatment-related mortality. Hence, outside the context of a clinical study, the use of either neoadjuvant or adjuvant chemotherapy cannot be recommended as a standard therapeutic approach.
Concurrent chemoradiotherapy Complete remission rates of locoregionally advanced disease to concurrent cisplatin radiotherapy in head and neck cancers, including NPC, were high and the early relapse-free survival
Table 2. Randomized trials of neoadjuvant chemotherapy in advanced NPC Institution [reference] Prince of Wales Hospital [26]
No. of patients 82
Chemotherapy
Median follow-up (months)
Results
Cisplatin + 5-FU
28.5
DFS no difference
×2 cycles neoadjuvant
OS no difference
×4 cycles adjuvant Institute Nationale Tumori [27]
229
Vincristine, cyclophosphamide, doxorubicin
48
×6 cycles adjuvant International NPC Study Group [28]
339
Bleomycin, epirubicin, cisplatin
Asian Oceanian Clinical Oncology Association [29]
334
Cisplatin, epirubicin
Sun Yat Sen Hospital [30]
456
Cisplatin, 5-FU, bleomycin
OS no difference 49
×3 cycles neoadjuvant
DFS, disease free survival; OS, overall survival.
DFS improved OS no difference
30
×2–3 cycles neoadjuvant ×2-3 cycles neoadjuvant
DFS no difference
DFS no difference OS no difference
62
DFS improved OS no difference
1012 rates were promising [31]. Cisplatin acts both as a cytotoxic agent and as a radiation sensitizer. The optimal scheduling of cisplatin and radiation has not yet been established, but daily low dose, weekly intermediate dose or 3-weekly high dose regimens have all been used. The Head and Neck Intergroup conducted a study comparing concurrent cisplatin and adjuvant cisplatin/5-fluorouracil (5-FU) with radiotherapy against radiotherapy alone in patients with stages III and IV NPC using the UICC 1987 classification [32]. The study was closed early after demonstrating significant OS and PFS advantage for the chemotherapy/radiotherapy group. Since the publication of this trial in 1998, the standard practice in North America has been concurrent chemotherapy/ radiotherapy using cisplatin 100 mg/m2 3-weekly ×3, followed by adjuvant cisplatin 80 mg/m 2 on day 1 and 5-FU 1 g/m2 on days 1–4, 3-weekly ×3. However, it is noteworthy that in this trial WHO type III histology (undifferentiated carcinoma) was present in only 44% of patients. In endemic areas such as southern China, the proportion of WHO type III histology will be >90%. Whether the results of a clinical trial derived from a heterogenous histological mix of patients can be directly applied to WHO type III undifferentiated NPC is not certain. Another factor that may have influenced the results of the trial was that the radiotherapy technique was not uniform among the participating Intergroup centers. Furthermore, the benefit of concurrent chemotherapy during radiotherapy and adjuvant chemotherapy after radiotherapy cannot be separated in the Intergroup study. A randomized trial of 229 patients treated in the Institute Nazionale Tumori in Milan failed to demonstrate any survival benefit for patients receiving four cycles of vincristine, cyclophosphamide and doxorubicin compared with the patients receiving no adjuvant therapy [27]. In addition, the Meta-Analysis of Chemotherapy in Head and Neck Cancer collaborative group meta-analysis results of head and neck cancer in general have indicated no survival benefit of adjuvant chemotherapy [33]. These data suggest that most of the benefit of the Intergroup 0099 regimen was derived from concurrent chemotherapy/ radiotherapy. Based on the success of concurrent chemoradiation in head and neck cancers and the encouraging phase II data in NPC, we embarked on a study in locoregionally advanced NPC comparing radiotherapy with concurrent cisplatin–radiotherapy. Patients with Ho’s N2 or N3 stage or N1 stage with nodal size ≥4 cm were eligible. Patients were randomized to receive cisplatin 40 mg/m2 on a weekly basis concurrently with external radiotherapy or radiotherapy alone. A total of 350 eligible patients were entered between April 1994 and November 1999. A preliminary PFS analysis demonstrated a trend towards benefit for the concurrent chemotherapy/radiotherapy arm [34]. Moreover, there was a very clear PFS benefit favoring chemotherapy/radiotherapy in the subgroup of Ho’s T3 (UICC T3/T4) patients with a hazards ratio of 2.49 (95% confidence interval 1.28–4.8). The benefit in the subgroup of advanced T stage patients was mainly attributable to a
reduction in the rate of distant metastases. Based on the evidence of this latter study and Intergroup 0099 Study, the use of concurrent cisplatin/radiotherapy should become standard therapy for endemic locoregionally advanced T and N stage NPC patients.
Salvage of local failure after radiotherapy Locoregional failures without distant metastases are potentially curable and should be treated aggressively. In the 1970s when the primary radiotherapy was often suboptimal in dose and tumor target coverage, the salvage rates of locally recurrent NPC by re-irradiation, mainly using external beams, were reported to be between 20 and 30% [35, 36]. However, as the primary radiotherapy improved, resulting in more adequate dose to the major part of the tumor, the rate of ‘geographical misses’ lessened. The tumors that fail such treatment should, at least theoretically, be more radioresistant. Indeed, we reported little success of re-irradiation to a high dose using 2D-planned external beams for the salvage of local relapse [37]. Moreover, the complications of re-irradiation were many and severe. These included severe trismus that disrupted the patient’s speech and ability to eat and also radiation-induced temporal lobe encephalopathy and cranial nerve damage causing diplopia (VI) and dysphonia (VIII–XII) and even aspiration (VIII–XII ). In view of limited success but significant morbidity, we do not recommend 2D-planned external radiotherapy as a salvage for NPC local relapses [37]. In small (
