1 Introduction

Nasopharyngeal carcinoma (NPC) has a unique etiology and geographic distribution which is mainly seen in Southeast Asia [1]. NPC is highly sensitive to radiotherapy and chemotherapy. The National Comprehensive Cancer Network (NCCN) guidelines recommend definitive radiotherapy alone for stage I, chemoradiotherapy for stage II, and chemoradiotherapy with induction or adjuvant chemotherapy for stages III to IVA [2]. More than 50% of patients with NPC present with locally advanced disease. Locoregionally advanced NPC (LA-NPC) is a heterogeneous patient population, with the prognosis of early-stage III NPC more closely resembling that of stage II disease [3], while T4 and N3 tumors have significantly poorer survival rates with a high risk of local–regional recurrences and/or distant metastasis [4]. Therefore, a “one-size-fits-all” treatment recommendation is not advisable. Recently, risk-stratified management strategies for NPC have attempted to individualize treatments to limit toxicity without compromising efficacy in low-risk patients and to improve survival in the high-risk patient. This article aims to review the risk-stratified strategies and propose future directions for de-escalating systemic chemotherapy or radiotherapy in low-risk NPC and increasing the intensity of treatment to improve survival in high-risk NPC.

1.1 De-escalation strategies for systemic chemotherapy

1.1.1 Omitting concurrent chemotherapy in low-risk stage II and T3N0 patients with NPC

Most recently, a multicenter, open-label, randomized, phase III, noninferiority clinical trial was conducted to assess whether concurrent chemotherapy can be safely omitted for patients with low-risk stage II and T3N0M0 NPC treated with IMRT (pretreatment plasma EBV DNA ≤ 4,000 copies/mL, maximal diameter of cervical LN ≤ 3 cm, without level IV or Vb lymph node and extranodal extension) [5]. In this study, there were no significant differences were observed in 3-year failure-free survival (FFS) between the IMRT-alone group and the CCRT group (91.9% vs. 90.5%, P value for noninferiority, < 0.001). But the IMRT-alone group experienced a significantly lower incidence of grade 3 to 4 adverse events (17% vs. 46%), including hematologic toxicities (leukopenia, neutropenia) and non-hematologic toxicities (nausea, vomiting, anorexia, weight loss, mucositis).

1.1.2 Sparing concurrent chemotherapy after induction chemotherapy in selected low-risk patients with NPC

In this era when IMRT has markedly improved locoregional control of LA-NPC [6], CCRT may provide only modest benefit while producing unnecessary toxicity. Among the chemotherapy regimens, induction chemotherapy (IC) is increasingly recommended given its advantages of better tolerance and earlier eradication of micrometastases. In this context, debates exist on whether concurrent chemotherapy should be maintained in the IC setting in the IMRT era.

Various studies have compared the outcomes of IC + CCRT versus IC + IMRT (Table 1). Regarding retrospective evidence, most studies without restricted inclusion criteria reported that IC + IMRT alone obtained similar survival outcomes with IC before CCRT for LA-NPC patients, but a relatively lower incidence of treatment-related toxicities [7,8,9,10,11,12]. Only one study by Chen et.al [13] reported improved OS in a subset of patients, mainly N3 or T4 patients, or those with larger tumor volumes after IC treatment, but no improvement was found in DMFS or LRRFS.

Table 1 Studies of omitting concurrent chemotherapy in NPC patients (IMRT vs. CCRT or CCRT vs. IC + IMRT)

A phase III randomized controlled trial is underway to compare the survival outcomes and side effects of radiotherapy alone and CCRT in LA-NPC patients with satisfactory tumor response (CR or PR) after IC receiving IMRT or tomotherapy (TOMO) (NCT03015727). While awaiting more convincing evidence, optimization of patient selection to continue or cancel concurrent chemotherapy is imperative to ensure treatment efficacy and avoid excessive toxicity.

1.1.3 Cisplatin schedules

In NPC, there is no level 1 evidence of comparable efficacy to cisplatin once every 3 weeks. The multicenter randomized phase II trial (KCSG-HN10-02) compared weekly (40 mg/m2) and triweekly cisplatin (100 mg/m2) in 109 patients and observed no significant difference in 3-year PFS (64.9% vs. 63.8%) or grade 3–4 toxicities (47.2% vs. 39.3%). But the QoL related to functional outcomes in triweekly regimen after treatment completion was better for the weekly regimen [14]. Recently, the preliminary results of a large phase III randomized controlled trial enrolling 526 patients with LA-NPC (ChiCTRC-12001979) showed no difference in survival outcomes between the two regimens, but the incidence of leukopenia (27.3% vs. 16.2%) and thrombocytopenia (4.8% vs. 1.2%) is higher in the once-weekly regimen group compared to the once-every-3-weeks regimen group [15]. The final results of this study might help to fully evaluate the different cisplatin schedules.

1.1.4 Cisplatin cumulative dose

Cisplatin toxicity is dose-dependent and complications, including ototoxicity, neuropathy, and nephrotoxicity, increase substantially with cisplatin cumulative dose (CCD), which could reach as high as 540 to 600 mg/m2 when sequential chemotherapy is contemplated [16,17,18,19]. Currently, there is no high-level evidence to guide the optimal dose intensity of concurrent cisplatin chemotherapy. Regarding studies to decide the minimum threshold for CCD in NPC, an updated combined analysis of 2 randomized studies (NPC-9901 and NPC-9902 trials) showed no difference in disease control (5-year locoregional-FFR, 88% vs. 88%) between 2 or 3 cycles of concurrent cisplatin; however, patients given 3 concurrent cycles had a significant increase in hearing impairment (40% vs. 24%, P = 0.017) [20]. Subsequently, a secondary analysis of a prospective phase III clinical trial [21] demonstrated that a CCD of 200 mg/m2 during CCRT is adequate to achieve satisfactory survival outcomes for patients with locoregionally advanced NPC. Additionally, a CCD of ≥ 240 mg/m2 was not an independent prognostic factor in patients with LA-NPC at high risk of distant metastasis. These studies showed that a cumulative CCD of 200 mg/m2 should be the threshold for achieving satisfactory survival.

Assigning treatment based on response allows another method of improved patient selection for patients with excellent prognoses. The outcomes of a retrospective study [22] revealed that in the subgroup of patients who achieved CR/PR after IC, there were no significant differences between > 200 mg/m2 and 101–200 mg/m2 for all endpoints, but was associated with fewer grade 1–4 acute toxicities. In addition, the application of different CCDs did not result in significantly different survival outcomes in the subgroup of patients who were stable disease (SD)/progressive disease (PD) after IC. However, this was a retrospective study and the data were obtained exclusively from one center; therefore, these results must be validated by other datasets.

At least three meta-analyses have confirmed that pretreatment plasma EBV DNA was a significant prognostic of survival and has been widely used as a predictor of risk stratification in recent studies [23,24,25]. By using pretreatment EBV DNA in risk stratification, Li et al. [26] performed an open-label, phase II, randomized, controlled trial to test the noninferiority of a two-cycle 100 mg/m2 concurrent DDP regimen over three-cycle in patients with low-risk LA-NPC with pretreatment EBV DNA levels < 4,000 copies/mL. In all, 332 patients were enrolled in this trial. After a median follow-up of 37.7 months, the 3-year PFS rate for the two-cycle group was 2.4% lower than the three-cycle group (88.0% vs. 90.4%) and met its primary endpoint of proving non-inferiority. Expectedly, there was a higher compliance rate in the two-cycle group than in the three-cycle group, with 165 (99.4%) of 166 patients completing two cycles of cisplatin compared with 149 (89.8%) of 166 patients completing three cycles. Patients who received only two cycles of cisplatin reported less insomnia, speech problems, xerostomia symptoms, constipation, and better cognitive functions than those who received three cycles. This study might represent a novel strategy to alter the current treatment dogma for low-risk LA-NPC.

1.2 De-escalation strategies for radiotherapy

1.2.1 Reducing doses of radiotherapy

With advances in modern radiotherapy techniques and effective systemic chemotherapy, local control for NPC has remained excellent at 90% [5]. Despite the widespread adoption of IMRT with reduced toxicity compared to conventional techniques, short- or long-term toxic effects such as mucositis, oral ulcers, and temporal lobe necrosis continue to adversely affect the QoL [17]. This finding is largely attributed to the total dose of 70 Gy near these critical normal tissues.

Considering long-term toxicities, treatment of NPC in children and adolescents has traditionally used IC to reduce radiation dose, and the 5-year event-free survival (EFS) has been achieved at 77% to 91% [27,28,29,30]. In the adult population, a single-arm, phase II trial is ongoing to assess whether an adaptive radiotherapy dose of 60 Gy after a favorable response to IC was non-inferior to the standard dose in low-risk stage III NPC patients (EBV DNA level < 4,000 copies/mL) [31]. The preliminary results of this trial revealed that the reduced-dose radiotherapy (60 Gy) is associated with favorable survival outcomes and limited treatment-related toxicities for low-risk stage III NPC patients sensitive to IC.

1.2.2 Reducing clinical target volume of radiotherapy

NPC has a high propensity for cervical node metastasis, 85% of NPC cases presented with lymphadenopathy [32]. Extensive studies on neck CTV volume reduction have mostly focused on investigating the feasibility of ipsilateral lower neck sparing irradiation for unilateral or bilateral neck node-negative NPC and have confirmed the safety, feasibility, and improved long-term QoL of this approach [33,34,35,36,37]. A recent randomized phase III noninferiority study demonstrated that elective upper-neck irradiation (UNI) of the uninvolved neck provides similar regional control and results in less radiation toxicity than standard whole-neck irradiation (WNI) in patients with N0-N1 NPC [34]. Nevertheless, the suitability of the above approach for patients with ipsilateral N2-3 disease and non-endemic populations requires further investigation. Moreover, these studies still depended heavily on the anatomical definition of the neck region without considering the specific site of lymph nodes. Therefore, effort is required to further optimize the nodal CTV delineation strategy in NPC. A multicenter non-inferiority randomized controlled phase III clinical trial (NCT05145660) from China is ongoing which aims to compare the regional control, survival outcomes, radiation-related toxicities, and QoL of Involved site irradiation (ISI) therapy, incorporated specific nodal site and spread distances of metastatic lymph nodes, with conventional elective region irradiation (ERI) in NPC patients with a limited nodal burden.

Specifically, several retrospective studies have explored the feasibility of omitting specific nodal levels. The risk of level Ib metastasis for patients with NPC is relatively low (2–8.5%). For those without risk factors for level Ib spreading, sparing Ib in prophylactic CTV has been proven to be safe by multiple studies [38,39,40]. However, for those with high-risk factors, which normally account for over one-third of all new patients [41], covering Ib is still recommended. Hence, attempts have been made to modify the CTV boundaries of level Ib. In a recent study [42], the shrinkage of the boundaries of level Ib specialized for NPC was proposed. The Ib CTVs that include the suggested modifications have significantly less volume than the CTVs suggested by the consensus. The normal tissue complication probability (NTCP) models further demonstrated that the risk of dysphagia would be significantly reduced, and the function of the submandibular gland (SMG) and the sublingual gland would be less affected by applying the CTV-proposed delineation.

1.3 Therapeutic intensification for high-risk NPC

1.3.1 high-risk NPC patients might benefit from IC

The 3 randomized phase III trials from China have proven the efficacy and effect of IC added to concurrent chemotherapy treatment mode, especially in stage III to IVA disease, on long-term patient outcomes [43,44,45]. Notably, however, the above-mentioned randomized trials did not enroll any patients staged with T3-4N0M0 or T3N1M0 at all. Some retrospective studies reported that patients with T3N0-1 do not benefit from IC [46], while male T3N1 patients with EBV DNA higher than 2000 copies/mL were the only target population for IC followed by CCRT, as suggested by another study [47].

In addition, not all patients in the above-mentioned 3 randomized phase III trials benefit from IC. Thus, in the spirit of informed decision-making, predictive models were developed to estimate the benefit of IC for individual patients would be particularly useful in clinical practice. The reasons for the results may be explained by the fact that the IC not only attenuates the tumor burden within a brief period to ameliorate tumor hypoxia but also exerts a systemic cytotoxic effect to eradicate distant micrometastases [48, 49]; therefore, patients with a high distant metastatic risk should consider treatment with IC + CCRT. However, all of the above studies are retrospective studies and multi-institutional and prospective studies are warranted to further validate these findings.

1.3.2 High-risk NPC patients might benefit from adjuvant chemotherapy

The benefit of adding adjuvant chemotherapy (AC) to chemoradiotherapy remains controversial in NPC [43, 50,51,52]. Most recently, Miao et.al [53] reported the efficacy and safety results in a randomized clinical trial of adjuvant capecitabine given at the full dose for 6 months. There were 180 patients enrolled and randomly assigned to receive in a 1:1 ratio either capecitabine (capecitabine group) or observation (control group) following CCRT (Eligibility included stage III-IVb NPC and at least 1 of the following: T3-4N2 or T1-4N3; plasma EBV DNA titer higher than 20,000 copies/mL; primary gross tumor volume larger than 30.0 cm3; [18F] FDG PET/CT maximum standard uptake value of the primary gross tumor volume larger than 10.0; or multiple nodal metastases and any larger than 4.0 cm). With a median follow-up of 58.0 months, 18 events were recorded in the capecitabine group vs. 31 events in the control group. The FFS was improved with adjuvant capecitabine (3 years, 83.3% vs.72.2%; 5 years, 78.5% vs. 65.9%; hazard ratio, 0.53; P = 0.03). The incidence of grade 3 treatment-related adverse events (TRAEs) was higher in the capecitabine group than in the control group (54 of 90 patients [60.0%] vs. 46 of 90 patients [51.1%]). TRAEs included xerostomia (17 [18.9%] vs. 9 [10.0%] patients), mucositis (21 [23.3%] vs. 15 [16.7%] patients), and anorexia (8 [8.9%] vs. 4 [4.4%] patients). The results of this randomized clinical trial support the use of capecitabine as an effective and tolerable adjuvant regimen for treating patients with LA-NPC.

1.3.3 Target therapy for high-risk patients

Epidermal growth factor receptor (EGFR) represents a promising new therapeutic target in cancer and is highly expressed in more than 85% of NPC patients. It is also an essential factor of prognosis for NPC, correlated with a more aggressive phenotype, and greater resistance to treatment. Therefore, anti-EGFR-targeted treatment is considered a potential addition to the standard CCRT regimen for NPC. In 2022, the randomized, controlled, multicenter phase III clinical trial by Sun et al. demonstrated the initial results of the clinical efficacy of long-term survival and safety of nimotuzumab combined with chemo-radiotherapy in LA-NPC patients. They found that Nimotuzumab plus chemo-radiotherapy increased the 5-year OS (76.9% vs. 64.3%, P = 0.042) in NPC patients. Besides, the combination of nimotuzumab plus chemo-radiotherapy was well tolerated. The incidence of adverse drug reactions (ADRs) in the nimotuzumab combined with the chemo-radiotherapy group was similar to the chemo-radiotherapy alone group (35.7% vs. 42.1%, P = 0.207), and the grade 3–5 ADRs as well (17.7% vs.15.7%, P = 0.609).

1.4 Future directions

1.4.1 Value of plasma EBV DNA for stratifying treatment

With emerging evidence for the induction-concurrent approach in advanced NPC, incorporations of EBV DNA early response as a mid-treatment prognosticator are under active investigation. Clearance of EBV DNA after systemic chemotherapy is associated with favorable drug response, and patients who attained undetectable EBV DNA after IC have significantly better survival [54,55,56]. In a large longitudinal study of 673 patients by Lv et.al [57], among the 30% early complete biochemical responders who achieved undetectable EBV DNA after a single cycle of chemotherapy, the 3-year DFS rate could reach as high as 94%. In this study, four distinct groups of patients were identified based on the kinetic of EBV DNA clearance: early responders had the best DFS, followed by intermediate responders, late responders, and the least for those who were treatment-resistant. Following this classification, a biomarker-guided risk-adopted treatment strategy was proposed (NCT04072107, currently recruiting patients): the intermediate responders (detectable EBV DNA after one cycle of IC but undetectable EBV DNA after subsequent IC) would receive adjuvant metronomic capecitabine for six months, while high-risk patients (detectable EBV DNA after three cycles of IC or early bounce of EBV DNA during induction phase) would receive concurrent anti-PD-1 therapy (sintilimab) with cisplatin and IMRT followed by adjuvant sintilimab for six months.

1.4.2 Imaging as biomarker

Applying noninvasive imaging biomarkers can predict early response to treatment and thus offers an opportunity for individualized risk-stratified treatment in NPC. MRI radiomic signatures are a superior tool in predicting responses to treatment and tumor failures in NPC [58, 59]. Dynamic contrast-enhanced MRI could be used to quantitatively assess microcirculation properties and hypoxia status of tumors because of the improved spatial resolution and signal-to-noise ratio over DWI [60].

Smaller apparent diffusion coefficient (ADC) values are related to a higher likelihood of diagnosing early NPC from nasopharyngeal lymphoid hyperplasia, as tumors with denser tumor cells and more cell membranes have lower ADCs due to greater restriction on diffusion [61]. A study in head and neck squamous cell carcinoma found that changes in ADC values after the first week of chemoradiotherapy had a high discriminatory ability in distinguishing CR from PR [62]. These studies support a potential opportunity where the total dose could be de-escalated or retained according to ADC changes throughout the treatment course. Thus, MRI radiomics is another opportunity for individualized risk-stratified treatment.

Another imaging modality that is crucial to treatment planning is PET. Chen et al. [63] investigated the value of early evaluation of response to CCRT using 18F-FDG PET-derived parameters and the EBV DNA titre in outcome prediction in patients with NPC. They found that Early evaluation of response to CCRT using 18F-FDG PET-derived parameters and the EBV DNA titre can predict outcome in patients with primary NPC. A combination of interim PET parameters and the EBV DNA titre enables better stratification of patients into subgroups with different survival rates. A study of patients with advanced NPC found that a combination of EBV DNA clearance and a reduction in maximum standardized uptake value of less than 50% at 4 weeks after initiation of chemotherapy was prognostic and predicted improvements in PFS and OS compared with either modality alone, which provided an opportunity for modification of treatment decisions at an early stage [64]. The functional aspect of PET allows tissue metabolic changes to be detected that may not be apparent on cross-sectional imaging and has also been found to be more accurate than MRI in the assessment of early treatment response [65]. These studies support a potential opportunity to identify low-risk groups who could potentially be spared from concurrent chemotherapy throughout the treatment course.

1.4.3 Immunotherapy

NPC is characterized by high PD-L1 expression (up to 90% of tumor cells) and abundant infiltration of non-malignant [66,67,68,69,70,71], which makes NPC patients potentially suitable for immune checkpoint blockade therapies. Several important single-arm trials evaluating anti-programmed death 1 (PD 1) antibodies, such as nivolumab, pembrolizumab, and camrelizumab, have demonstrated encouraging activity in recurrent or metastatic NPC (RM-NPC).

The treatment advances of ICIs monotherapy in LA-NPC have remained scarce, and currently, the optimal sequence to incorporate ICIs with definitive radiotherapy in NPC is uncertain. Although ICIs are not yet recommended in the guidelines for patients with LA-NPC, several phase II to III anti-PD-1/PD-L1 monoclonal antibody clinical trials are currently underway (Table 2). NCT04453826 and NCT03925090 aim to explore EBV-directed immunotherapies as alternatives to adjuvant chemotherapy for high-risk patients (ie, detectable EBV DNA after primary radiotherapy). NCT03925090 (induction and adjuvant toripalimab), NCT03427827 (adjuvant camrelizumab), and NCT03700476 (induction, concurrent and adjuvant sintilimab) will test the value of adding anti-PD-1 therapy to standard treatment in the curative setting. Three trials would be of interest to compare the efficacy of ICI as an adjuvant treatment (NCT04870905) or even as a radiosensitizer in replacement of cisplatin (NCT04907370 and NCT03984357). To this end, studies on predictive biomarkers (eg, PDL 1 expression, tumor mutational load, immune cell infiltration, and immune gene expression) to enrich treatment responses are crucial to guide patient selection (NCT03427827).

Table 2 Ongoing phase II/III anti-PD-1/PD-L1 monoclonal antibody clinical trials for patients with LA-NPC

2 Conclusion

Various risk-stratified treatment strategies for NPC have been explored. The optimization of patient selection remains critical, and long-term outcomes remain to be determined. Attention needs to be paid to the generalizability of the results because subtypes of NPC can present in different manners. Continued investigation in prospective, multi-institutional studies is needed to elucidate how best to individualize the risk-stratified treatment of NPC.