Nasopharyngeal carcinoma (NPC) is a malignant epithelial tumour arising from the mucosal lining of the nasopharynx, with a striking epidemiological clustering in Southern China, Southeast Asia, and North Africa. Globally, approximately 130,000 new cases are diagnosed annually, with an age-standardized incidence of around 2.1 per 100,000 in 2019 [1]. The disease is strongly associated with Epstein–Barr virus (EBV) infection, particularly for the non-keratinising subtypes (World Health Organization, WHO types II and III), which account for the vast majority of cases in endemic regions [2]. Despite a pronounced sensitivity to platinum-based chemoradiotherapy, and 5-year overall survival (OS) rates of approximately 70% for locoregionally confined disease, more than half of patients present with stage III–IV disease, and a substantial proportion eventually relapse [2, 3].

For patients with recurrent or metastatic NPC, the treatment landscape has evolved considerably over the past decade. The phase III GEM20110714 trial established gemcitabine plus cisplatin as the standard first-line regimen in the recurrent/metastatic setting, demonstrating superior progression-free and OS compared with fluorouracil–cisplatin [4]. Recently, the addition of programmed death-1 (PD-1) checkpoint inhibitors, most notably toripalimab (JUPITER-02 trial) and penpulimab-kcqx (AK105-304 trial), to platinum-based chemotherapy has further improved outcomes and established chemoimmunotherapy as the current standard of care for first-line treatment of recurrent/metastatic NPC [5, 6]. Nevertheless, for patients who progress after immunochemotherapy, therapeutic options remain extremely limited, and prognosis is poor.

Cutaneous metastases from NPC (also referred to as lethalides or carcinoma cutis) are a rare and clinically ominous complication, with very few documented cases in the published literature [7, 8]. Their occurrence represents an advanced stage of haematogenous or lymphatic dissemination and typically signals a poor outcome. Concurrently, plasma EBV DNA load has emerged as an indispensable biomarker in NPC management, correlating dynamically with tumour burden, treatment response, and early relapse detection [9, 10].

We present a case of EBV-positive non-keratinizing NPC characterized by an uncommon cutaneous metastatic pattern at first recurrence. Notably, the patient exhibited an extraordinary survival trajectory through six lines of systemic treatment. We emphasize the temporal dynamics of plasma EBV DNA, illustrating its utility in guiding clinical transitions during long-term multimodal management, including immunotherapy.

Table 1 below summarises the patient’s clinical timeline. A comprehensive timeline has been generated using the CARE-writer tool in accordance with journal requirements.

Key diagnostic test results are summarised in Table 3. Plasma EBV DNA measurements were performed throughout the clinical course and served as the primary dynamic biomarker guiding therapeutic transitions.

The patient was informed of his diagnosis and prognosis at each clinical transition by the attending oncology team. He consistently expressed a strong desire to pursue all available treatment options, including clinical trial participation. His decision to enroll in the ABILITY-1 trial reflected both his personal determination and his understanding of the investigational nature of the therapy. He and his family were closely supported by the hospital’s palliative care and social work teams throughout the clinical course. The transition to palliative care was made collaboratively, with the patient and his family expressing appreciation for the multidisciplinary support received. In accordance with institutional privacy regulations and given the patient’s subsequent death, this account has been compiled with the informed consent of his next of kin.

This case report illustrates the complex, multiphasic trajectory of a patient with EBV-positive non-keratinising NPC who, after an initial complete response to multimodal treatment, developed the rare complication of cutaneous metastases and ultimately proved refractory to six lines of systemic therapy over an approximately two-and-a-half-year period. Several clinically important themes merit detailed discussion.

Cutaneous metastasis (also referred to as lethalides), as a manifestation of visceral malignancy is itself uncommon, occurring in approximately 5.3% of patients with metastatic carcinoma (range in previous reports, 0.7–9%) [19]. In NPC, skin involvement is exceptionally rare. A retrospective study of 820 NPC patients found cutaneous metastases in fewer than 1% of cases [7]. Published case series from Hong Kong documented four patients with distant skin metastases among a large NPC cohort spanning several years [8]. When skin metastases present at the locoregional level, as in our patient, direct extension from subcutaneous lymphatic tumour deposits appears to be the dominant mechanism, while scalp metastases typically reflect haematogenous seeding [7, 8, 20].

The morphological spectrum of NPC skin metastases is broad. The most common presentation comprises firm, non-tender cutaneous or subcutaneous nodules, often violaceous or erythematous [20]. Less commonly, inflammatory erysipelas-like plaques (carcinoma erysipelatoides) or zosteriform patterns have been reported [21]. In our case, the cutaneous tumour followed an infiltrative, confluent progression, establishing direct locoregional invasion, which correlated with a rapidly escalating EBV DNA trajectory. Cutaneous metastases typically co-occur with multisystem disease and heralded rapid clinical deterioration in our patient, as has been previously described [7, 8]. High PD-L1 expression (CPS 70), as observed in this patient, is the biomarker used to guide immune checkpoint inhibitor selection in head and neck squamous cell carcinoma (SCC) [22]; in EBV-associated NPC, it is typically accompanied by an inflammatory tumour microenvironment that may, paradoxically, coexist with immune evasion [2].

The systemic management of NPC has undergone transformative change over the past five years (Table 4). Gemcitabine–cisplatin, established as the standard first-line doublet chemotherapy for recurrent/metastatic NPC by the landmark GEM20110714 trial [median OS 22 vs. 19 months vs. fluorouracil–cisplatin; hazard ratio (HR) 0.72] [4], has now been superseded as a monotherapy backbone by chemoimmunotherapy combinations.

The phase III JUPITER-02 trial (n = 289) demonstrated that the addition of toripalimab (anti-PD-1) to gemcitabine–cisplatin significantly improved progression-free survival (PFS, median 21 vs. 8 months; HR 0.52; 95% CI 0.37–0.73) and OS (HR 0.63; 95% CI 0.45–0.89) in treatment-naïve recurrent/metastatic NPC [5]. These results led to toripalimab’s US FDA approval in October 2023 and its endorsement as a preferred first-line option in international guidelines [14]. The CAPTAIN-1st trial with camrelizumab and the RATIONALE-309 trial with tislelizumab (both combined with gemcitabine–cisplatin) have produced additional confirmatory data, with early OS signals and extended PFS2 benefits that further validate the chemoimmunotherapy paradigm [23, 24]. The recently approved penpulimab-kcqx (AK105-304 trial) provides an additional option, though OS data remain immature [6]. Of note, toripalimab was unavailable in Spain when our patient relapsed; had the current standard been accessible, first-line chemoimmunotherapy would have been the preferred approach.

In the subsequent-line setting, single-agent PD-1 inhibition has proven less uniformly effective. The KEYNOTE-122 trial of pembrolizumab vs. chemotherapy in platinum-pretreated NPC failed to demonstrate an OS advantage, despite a numerically higher response rate among patients with elevated PD-L1 expression [25]. The lack of response in our patient, despite a CPS of 70, is consistent with the unpredictable efficacy of single-agent checkpoint blockade in multiply pretreated NPC, and may reflect prior immune escape mechanisms or an intrinsically exhausted T-cell repertoire. This observation underscores an important limitation of PD-L1 CPS as a predictive biomarker in NPC: while associated with a trend towards immunotherapy benefit, it has insufficient discrimination at the individual patient level and is not currently sufficient to determine patient selection [14, 26].

EGFR is overexpressed in 85–90% of NPC tumours, providing a rationale for EGFR-targeted strategies [16]. Adding cetuximab to platinum-based chemoradiotherapy has shown a trend towards improved locoregional control in retrospective analyses of locoregionally advanced NPC [16]. In the recurrent/metastatic setting, the phase I/II trial of carboplatin plus cetuximab demonstrated an objective response rate (ORR) of approximately 11% in previously treated NPC [17]. In our patient, biweekly paclitaxel plus cetuximab achieved only transient stabilisation (approximately 3 months), consistent with the modest efficacy of this combination in heavily pretreated patients.

Targeting interleukin-2 (IL-2) pathways leverages its role in expanding cytotoxic lymphocytes, yet native IL-2 (aldesleukin) causes severe systemic toxicity (e.g., capillary leak syndrome) by indiscriminately stimulating CD25+ Tregs and endothelial cells [27]. MDNA11 (Medicenna) addresses this via engineered mutations that abolish CD25 binding while enhancing affinity for the IL-2Rβ/γ dimer on CD8+ T and NK cells [18, 28]. This “beta-selective” engineering, coupled with albumin fusion for biweekly dosing, broadens the therapeutic window by avoiding Treg activation and extending half-life. In the ABILITY-1 study (AACR 2025/AACR 2024) [18, 28], MDNA11 monotherapy achieved an ORR of 7.7% (2 of 26 evaluable patients) in immune checkpoint inhibitor (ICI)-pretreated patients (including melanoma CR and MSI-H PDAC PR) with no DLTs up to 120 µg/kg. In combination with pembrolizumab, responses in chemo-refractory anal SCC (CR) and MSS colorectal carcinoma (PR) suggest MDNA11 can sensitize “cold” tumors to PD-1 blockade [18]. While EBV+ NPC’s dense lymphocytic infiltrate and PD-L1 expression favor immune amplification, EBV-driven latency (LMP1/2, EBNA1) facilitates evasion via MHC-I downregulation and TIL exhaustion [2, 29]. Our patient’s lack of response, despite high PD-L1, underscores the profound multifactorial exhaustion in relapsed NPC, highlighting the current limitations of cytokine-mediated strategies in late-stage disease [2].

A particularly instructive aspect of this case is the dynamic correlation between plasma EBV DNA levels and clinical events across six lines of therapy. Quantitative plasma EBV DNA reflects tumour burden in real time, as NPC cells actively secrete EBV-derived genomic DNA into the circulation [9]. Multiple large prospective studies have established that pre-treatment EBV DNA levels are an independent prognostic factor for OS and PFS in both locoregionally advanced and recurrent/metastatic NPC, and that post-treatment EBV DNA clearance correlates strongly with durable tumour response [9, 15]. In our patient, EBV DNA fell from 76.4 to undetectable copies/mL in parallel with clinical and radiologic response to first-line salvage chemotherapy, then became detectable again several weeks before CT imaging confirmed recurrence (February 2024). This pattern (where rising EBV DNA precedes radiologic progression) is consistent with several prospective studies showing that EBV DNA can become detectable before imaging-detectable disease [15]. This observation has meaningful clinical implications: in resource-rich settings, rising plasma EBV DNA during or after treatment could trigger an earlier line change or additional imaging, potentially avoiding disease progression in a window where the tumour is still more amenable to treatment. The exponential EBV DNA elevation during third-line paclitaxel–cetuximab (from 485 to 13,000 copies/mL over 3 months) mirrored the aggressive locoregional expansion documented on CT, reinforcing the agent’s lack of efficacy in this patient. Conversely, the transient, albeit short-lived, EBV DNA decline during fifth-line weekly carboplatin–paclitaxel (from > 4,000 to 252 copies/mL) provided an objective biochemical correlate of the partial clinical response observed. These patterns emphasise the value of integrating EBV DNA into routine clinical monitoring for all patients with EBV-positive NPC under systemic therapy [14].

Regarding the optimal frequency of plasma EBV DNA monitoring, the optimal schedule has not been validated in prospective trials; published practice generally describes periodic measurement during active systemic treatment and every 3–6 months during surveillance after curative-intent therapy [9, 14]. In the present case, EBV DNA measurements were performed approximately every 4–6 weeks during treatment cycles, which proved clinically adequate: rising viral loads consistently preceded radiologic evidence of progression by 4–8 weeks, consistent with the lead-time advantage described in the literature [15]. Based on this experience and current evidence, we propose that a monitoring interval of every 4–6 weeks during active systemic therapy represents a clinically practical and sufficiently sensitive schedule, enabling timely detection of incipient relapse. In the post-treatment surveillance phase, measurements every 3 months appear appropriate for the first 2 years, transitioning to 6-monthly intervals thereafter, in alignment with institutional and guideline recommendations [14].

The therapeutic gap for NPC patients progressing after chemoimmunotherapy remains substantial. Promising strategies include adoptive cell transfer of EBV-specific cytotoxic T lymphocytes (CTLs) [30], which leverages tumour-specific antigens, and therapeutic vaccines (targeting LMP2/EBNA1) designed to boost endogenous CD8+ responses [2]. Bispecific antibodies engaging tumour antigens (e.g., EGFR) and T cells represent further avenues. Targeted therapies like the EGFR-ADC (antibody-drug conjugate) izalontamab brengitecan show promise following recent phase III superiority data [31]. Similarly, becotatug vedotin (MRG003), an EGFR-targeting ADC, received approval in China in October 2025 after achieving an ORR of 30.2% (vs. 11.5% with chemotherapy) in a randomized phase 2 trial in NPC pretreated with chemotherapy and PD-1/PD-L1 inhibitors, supporting the utility of ADC technology in EGFR-overexpressing tumors [32].

For locoregional recurrence, re-irradiation with IMRT can offer long-term survival in selected patients (5-year OS of approximately 41%), but at the cost of substantial severe late toxicity, including mucosal necrosis, as shown in a meta-analysis [33]. Although our patient was ineligible for definitive reirradiation due to prior cumulative doses, palliative hypofractionated radiotherapy (HexaSkin protocol) offered transient symptomatic relief, underscoring radiotherapy’s supportive role in advanced disease [33].

This case underscores the aggressive nature of cutaneous NPC metastases or lethalides and the formidable challenge of managing plurirefractory disease. Our clinical experience highlights three pivotal lessons. First, plasma EBV DNA is an essential real-time biomarker; its dynamic fluctuations anticipated radiologic changes at every clinical transition, providing a superior surrogate for both treatment response and early therapeutic failure. Second, the rapid evolution of the NPC treatment landscape, particularly the shift toward first-line chemoimmunotherapy, necessitates timely implementation of new standards to optimize patient survival. Finally, while participation in the ABILITY-1 trial with the novel IL-2 superkine MDNA11 did not yield a clinical response in this specific late-stage setting, it reinforces the critical importance of ongoing clinical trial enrollment. As the field moves toward personalized oncology, integrating dynamic EBV DNA monitoring with emerging targeted strategies, such as EGFR-directed ADCs and adoptive cell therapies, will be fundamental to overcoming multifactorial immune exhaustion and improving outcomes in advanced NPC.