Efficacy of Immune Checkpoint Inhibitors and Oncoviruses in Solid Tumors.

Introduction
Approximately 12% to 20% of cancer cases are estimated to be linked to infections (1–4). Although a portion of these cancers are caused by bacteria or multicellular parasites, the wide majority is associated with chronic viral infections (5). The relative burden of virus-related cancers varies with geography and health system capacity. In low- and middle-income countries, gaps in prevention and treatment amplify disease burden; for example, limited access to hepatitis C virus (HCV) direct-acting antivirals and suboptimal hepatitis B virus (HBV) vaccination sustain chronic liver disease and hepatocellular carcinoma (HCC; ref. 6). In high-income countries, immunosuppressed subgroups, like solid-organ transplant recipients and people living with human immunodeficiency virus (HIV), experience markedly higher incidences of several virus-associated malignancies owing to impaired cancer immunosurveillance and greater viral persistence (7). After transplantation, human papillomavirus (HPV)–related cancers occur in significant excess and risk increases with time on immunosuppression (8). In the same setting, Kaposi sarcoma and Merkel cell carcinoma (MCC) are strongly immunosurveillance sensitive, with large risk elevations in transplant cohorts (9). Separately, at the population level in high-income countries, the increase in HPV-positive oropharyngeal cancers over recent decades is best explained by shifts in sexual practices, together with the decline in smoking that reduced HPV-negative head and neck squamous cell carcinoma (HNSCC), rather than by any secular increase in immunosuppressed states (e.g., transplant recipients or iatrogenic immunosuppression; refs. 10, 11).
Seven viruses are currently recognized as oncogenic in humans. Five possess DNA genomes: the herpesviruses Epstein–Barr virus (EBV) and Kaposi sarcoma–associated herpesvirus [KSHV/human herpes virus 8 (HHV8)], which establish long-term latent, episomal infection; Merkel cell polyomavirus (MCPyV), which persists in host cells; high-risk HPV, for which the genome remains episomal or becomes integrated during malignant progression; and HBV, a reverse-transcribing hepadnavirus that typically remains extrachromosomal despite its retrovirus-like life cycle. Two RNA viruses are also oncogenic but differ in their replication strategies: human T-lymphotropic virus type 1 (HTLV-1), an integrating retrovirus, and HCV, a nonintegrating flavivirus that causes lifelong chronic infection (5). However, other viruses, like for instance HIV, although not directly oncogenic, can deeply influence immunity and contribute to cancer development (12). When immunosurveillance is compromised, either by HIV-mediated CD4+ T-cell depletion or by iatrogenic immunosuppression, the incidence of virus-induced solid tumors increases dramatically. In solid-organ transplant recipients, the overall cancer risk is approximately 2.1-fold higher than in the general population, with particularly striking increases in Kaposi sarcoma (standardized incidence ratio ≈61.5) and nonmelanoma skin cancers (standardized incidence ratio ≈13.9; ref. 13).
Although immune checkpoint inhibitor (ICI) therapy has transformed cancer care, ICIs often face limitations because of resistance and are limited by the lack of robust predictive biomarkers. This may relate to the various characteristics of the tumor microenvironment (TME) among different cancers. Consequently, it is crucial to determine which patient groups would benefit the most from ICIs. Tumors associated with viral infections are often classified as “hot tumors” (14), namely T cell–inflamed with intratumoral CD8+ infiltration, IFNγ/chemokine activity, and PD-L1 expression, whereas “cold” (immune desert or immune excluded) tumors lack effective T-cell infiltration and are generally less responsive to immune checkpoint blockade. The influence of viral infections on ICIs’ effectiveness is still debated and controversial, with no clear consensus in the field. The diversity in tumor types, viruses, and ICIs adds a level of complexity to this issue (15).
In this review, we explore the contribution of various viruses to carcinogenesis, either through direct or indirect mechanisms, and examine cases in which differences in immunotherapy effectiveness have been observed based on infection status. Cancers predisposed by HTLV-1 and HIV, for which no definitive direct causal role has been established in solid tumors, fall outside the scope of this work and will not be further addressed.

Mechanisms of Viral-Induced Cancers
Viruses can contribute to tumor development through distinct mechanisms (Table 1). First, viruses with a latent phase of infection like herpesviruses, most notably EBV, can directly trigger the transformation of infected cells to increase their survival. Viral proteins, encoded by viral genes integrated into the host genome or in stable episomes, can regulate the growth and survival of host cells. The recognition of viral proteins by host cells can then activate the DNA damage response (DDR), a pathway initially antiviral in nature but which certain viruses exploit to support their replication. DDR leads to genetic instability, increasing the mutation rate and leading to the acquisition of oncogenic chromosomal alterations in host cells (16).
Activation of tumor suppressor pathways is essential to protect cells against transformation induced by oncogenic viruses, with p53, p21, and retinoblastoma protein (pRB) playing central roles in regulating cell-cycle progression and the response to DNA damage. These pathways tightly control cellular functions, including apoptosis induction and DNA repair, to prevent uncontrolled cell proliferation and tumor development (16). Oncogenic viruses code for oncoproteins that deregulate these pathways, disrupting their normal function and promoting viral replication.
Second, viral infections can induce DNA damage reactions in host cells, promoting genomic instability and epigenetic alterations leading to malignant transformation (5, 12, 16) by promoting chronic inflammation.

HPVs
HPVs include more than 200 genotypes of DNA viruses within the Papillomaviridae family, which are transmitted through direct skin-to-skin contact, and are considered as the most frequently sexually transmitted agents worldwide. Although low-risk HPV types cause benign genital and skin warts, high-risk HPV types are linked to cervical, genital, or oropharyngeal cancers (17). Sero-epidemiologic data indicate that roughly 80% of sexually active women acquire genital HPV at least once in their lifetime (18). In most individuals, the infection is transient and is cleared by host immunity within 12 to 24 months. Only about 5% to 10 % of cases persist beyond 2 years, and less than 1% ultimately progress to high-grade intraepithelial neoplasia (19). When persistence occurs, oncogenic transformation is slow, typically unfolding over a decade or more after the initial infection.
The most common high-risk types are HPV-16 and HPV-18 (20). During its replication cycle, HPV modulates host cell protein expression dynamically to avoid its own detection, particularly by reducing antigen presentation to the immune system. This is facilitated by a dynamic protein expression model, orchestrated by viral proteins E1 (a helicase) and E2 (E6/E7 repressor protein), which regulate viral DNA replication and gene expression. As infection progresses, increased gene expression perpetuates viral replication, while poor antigen presentation protects the virus from immune detection. Integration of the HPV genome into host cells amplifies oncogenic potential by deregulating E6 and E7 activity, contributing to tumorigenesis (21, 22) . E7 inactivates pRB, overriding the G1–S cell cycle checkpoint and promoting unscheduled cell-cycle progression. This abnormal proliferation would typically trigger p53-mediated apoptosis, but E6 neutralizes this safeguard by promoting p53 degradation. E6 further contributes to immortalization by inducing hTERT expression, preventing telomere shortening. In combination, E6 and E7 drive continuous cell division and chromosomal instability, facilitating tumor initiation and progression (5, 21, 23).

EBV
EBV is a double-stranded DNA virus that targets B lymphocytes and belongs to the Herpesviridae family (24). Nearly 95% of adults worldwide have been infected with EBV during childhood, with the primary infection causing infectious mononucleosis (24).
The virus is primarily transmitted through saliva but can also spread via breast milk, other bodily fluids, or transplantation of EBV-positive organs (24). Despite a near-universal exposure, EBV-associated solid tumors are rare. Nasopharyngeal carcinoma (NPC) causes around 0.7% of all cancers, with the majority arising in East and Southeast Asia (25), and EBV-positive gastric carcinoma represents around 10% of all gastric cancers (26).
EBV isolates are categorized into two types, type 1 and type 2, which differ in the EBNA2 and EBNA3 gene sequences. Although type 2 EBV is less efficient at transforming lymphocytes, coinfections with both types can lead to lymphocyte changes (27, 28). EBV establishes lifelong persistence in humans by infecting B cells and residing in memory B cells in a typically asymptomatic state in healthy individuals (27). Factors such as genetic mutations, immune deficiencies, immunosuppression, HIV co-infection, or dietary habits like high consumption of salted or preserved fish can disrupt this balance and result in EBV-associated cancers (27).
Like other herpesviruses, EBV has two distinct life cycle phases: latency and lytic replication. During latency, the virus remains in the host cell nucleus as a circular episome attached to host chromatin through the viral protein EBNA-1 (29). In this phase, only a limited set of viral genes and noncoding RNAs are expressed, with gene expression patterns varying by cell type. The virus relies on the host's normal cellular division and DNA replication machinery to propagate its genome passively to daughter cells (30, 31). This phase allows EBV to persist lifelong in humans, with occasional reactivation into the lytic cycle. During the lytic phase, all viral genes are expressed, and the virus actively replicates its genome, producing new infectious particles that lead to cell death and release of the virus. These new virions can infect other cells within the same host or spread to new hosts. In some cases, abortive lytic expression occurs, in which only a subset of lytic genes is expressed, without full lytic replication (27).
The only viral protein expressed in all EBV-related malignancies is EBNA1, which is essential for replication and maintenance of the virus’s genetic material, but little is known about its role in oncogenesis. It is believed that EBNA1 blocks p53-dependent apoptotic signaling, promoting cell survival after DNA damage, and intensifies the expression of antiapoptotic proteins such as Bcl2 and survivin, further enhancing cell survival (32). Additionally, EBNA1 promotes the production of reactive oxygen species, leading to genomic instability. EBNA2 contributes to transforming B cells by activating genes essential for cell growth and survival. EBNA3 supports cell survival and growth by interfering with proteins that regulate cell division and cell death (12).
Latent membrane protein 1 and 2 (LMP1 and LMP2) play a central role in EBV oncogenesis, mimicking cell signaling molecules and deregulating various signaling pathways involved in cell growth, survival, and differentiation. LMP1, for example, mimics the CD40 receptor on the surface of B cells, activating the NF-κB, JAK/STAT, and MAPK pathways, thereby promoting cell survival and proliferation (5, 12, 33).

MCPyV

MCPyV infection is virtually ubiquitous, with adult seroprevalence ranging from 70% to 90% worldwide (34, 35). Despite this high exposure, MCC remains extremely rare, with population-based incidence estimates of 0.1 to 1.6 cases per 100,000 person-years globally (34). When these incidence data are compared with the near-universal seropositivity, the lifetime risk of MCC among MCPyV-infected individuals is well below 0.1%, showing that additional cofactors are needed.
MCPyV is causally linked to approximately 80% of MCCs, an aggressive neuroendocrine carcinoma of sun-exposed skin (34). MCPyV is a polyomavirus, which carries a circular double-stranded DNA genome, structured into early and late coding regions situated on opposite strands. These regions are divided by a regulatory noncoding segment that houses the origin of replication. During the initial phase of infection, the virus expresses its early genes prior to DNA replication. This leads to the production of two key proteins, large T antigen (LT) and small T antigen (ST), through alternative splicing. LT plays a central role in viral replication and is characterized by distinct domains, including an Rb-binding sequence (LXCXE), a DNA-binding domain, a nuclear localization signal, and a helicase region (36). Viral DNA replication is primarily driven by LT, with ST enhancing the process (37). MCPyV-negative MCCs are frequently RB1 mutated, whereas MCPyV-positive tumors typically harbor wild-type RB1 (38).

HHV8 or KSHV
This virus is another member of the Herpesviridae family. Its seroprevalence is geographically heterogeneous, ranging from <10% in Northern Europe, North America, and Asia to 20% to 30 % in the Mediterranean basin and ≥90% in parts of sub-Saharan Africa (39). In 2020, Kaposi sarcoma had a global incidence rate of 0.39 and mortality rate of 0.18 per 100,000, representing around 34,270 new cases and around 15,086 deaths, 73% and 87% of which occurred in Africa (40).
Despite high seroprevalence, <0.1% of immunocompetent KSHV-infected individuals ever develop Kaposi sarcoma. The risk increases strikingly under immunosuppression: HIV-related Kaposi sarcoma reaches 164 cases per 100,000 person-years in untreated people living with HIV (41), and solid-organ transplant recipients face a 73-fold excess Kaposi sarcoma risk compared with the general population (42).
KSHV forms a circular episome within the host cell nucleus, coding for 87 open reading frames and 17 microRNAs, many of which resemble host cellular genes (1). During lytic infection, mature virions bind to specific cellular receptors and enter the host cell through clathrin-mediated endocytosis. Once inside, the viral genome quickly circularizes within the host cell nucleus. The virus remains latent unless reactivation is triggered, typically by events such as plasma cell differentiation and hypoxia (43), which lead to promoter demethylation of the open reading frame 50 gene and activation of the replication and transcription activator to begin the lytic cycle (44). Early lytic genes support DNA replication and viral gene expression, whereas late lytic genes encode structural proteins necessary for virion assembly.
KSHV infection alone does not lead to complete neoplastic transformation. Although it can extend cell survival and induce resistance to apoptosis in vitro, it does not directly induce malignancy (45). Although still poorly described, the development of tumors might require additional cofactors, such as HIV infection or drug-induced immunosuppression, which could explain why most KSHV infections in the general population do not result in Kaposi sarcoma (1).

HBV
HBV is a reverse-transcribing DNA virus (Hepadnaviridae). An estimated 254 million people lived with chronic HBV in 2022 (46). Chronic carriage confers a 10- to 100-fold higher risk of HCC than in uninfected individuals. Transmission is predominantly perinatal/early childhood in high-prevalence regions and via blood or sexual exposure elsewhere. Despite effective vaccination and antivirals, large diagnosis and treatment gaps persist, especially in low- and middle-income countries (47, 48).
Oncogenesis is both direct and indirect. A defining feature of HBV-related HCC is the high rate of viral DNA integration, found in around 85% to 90% of tumors and far more often than in adjacent liver, which drives chromosomal instability and provides viral transcripts even without active replication. Integrations are nonrandom, recurrently affecting TERT, KMT2B/MLL4, CCND1/FGF19, and CCNE1. Mechanistically, double-stranded linear HBV DNA produced during reverse transcription can enter the nucleus and join at sites of DNA breaks via error-prone repair, and the resulting inserts may express truncated HBs and HBx proteins that promote proliferation, oxidative stress, and cell survival (49).
The HBV-HCC TME is typically T cell exhausted (PD-1–high CD8+ co-expressing LAG-3/TIM-3/TIGIT) with increased regulatory T cells (Treg) and suppressive myeloid populations, features that enable some benefit from PD-1/PD-L1 blockade but also argue for co-blockade and strategies restoring antigen presentation and T-cell fitness (50, 51).

HCV
HCV is a positive-sense RNA virus of the Flaviviridae family and a leading cause of chronic liver disease. Globally, about 50 million people live with chronic HCV and about one million new infections occur each year. HCV contributes substantially to deaths from cirrhosis and HCC. Transmission is predominantly parenteral, with regional variation in injection-related and iatrogenic exposures (52).
HCV is a nonintegrating RNA virus, and hepatocarcinogenesis is therefore largely indirect, driven by years of necroinflammation that evolves into steatohepatitis, fibrosis, and cirrhosis, states marked by oxidative DNA injury and repeated regenerative stress. In parallel, viral proteins rewire hepatocyte signaling and metabolism and blunt innate immune sensing, creating a permissive, immunosuppressed microenvironment (53). These injuries also imprint durable epigenetic “scars” that can persist after sustained virologic response, explaining the residual HCC risk in patients with advanced fibrosis despite virologic cure (54).
HCV infection alone is usually insufficient for malignant transformation; cofactors, fibrosis stage, alcohol use, metabolic syndrome/diabetes, older age, and HBV/HIV coinfection amplify risk (55).

PD-L1 regulation and therapeutic implications in viral infections
The interaction between PD-1 receptor and its ligand PD-L1 is a prototypical example of an immune checkpoint. Under physiologic conditions, the binding of PD-1, expressed on T cells, to PD-L1, expressed on antigen-presenting cells, downregulates the immune response by inhibiting T-cell activity. In cancer, tumor cells can exploit this mechanism by overexpressing PD-L1 to evade immune surveillance. ICIs, such as anti–PD-1 or anti–PD-L1 antibodies, block this interaction, thereby restoring T-cell activity and enabling an effective antitumor immune response (56).
Another immune checkpoint targeted in clinical practice, which acts earlier in the immune activation cascade, is the CTLA-4/B7 interaction between antigen-presenting cells (e.g., dendritic cells) and T cells. Blocking CTLA-4 with monoclonal antibodies is a well-established anticancer strategy. CTLA-4 is upregulated on activated T cells and constitutively expressed on Tregs. By competing with CD28 for B7 ligands in secondary lymphoid organs, it dampens T-cell priming and complements PD-1/PD-L1, which primarily constrains effector responses within tissues (57). Other checkpoints, such as LAG-3, are emerging and entering clinical trials but are not yet investigated in virus-associated cancers (58). In this review, we will focus primarily on PD-1 and CTLA-4.
The PD-1/PD-L1 pathway plays a pivotal role in regulating immune responses during viral infections, balancing effective viral clearance and limiting immunopathology. During acute infections, such as those caused by lymphocytic choriomeningitis virus (LCMV), influenza, or SARS-CoV-2, PD-1 upregulation on CD8+ T cells and PD-L1 expression on immune and nonimmune cells serve to fine-tune immune responses, ensuring sufficient viral clearance while preventing excessive tissue damage (59). For instance, blocking PD-1 in acute LCMV infection enhances CD8+ T-cell activity, accelerating viral elimination without increasing tissue damage. However, in PD-1–deficient mice infected with highly replicative LCMV strains, the complete absence of this inhibitory pathway leads to uncontrolled T-cell activation and severe immunopathology. This highlights a dual role for PD-1; although its blockade can reinvigorate exhausted T cells during chronic infection, its physiologic expression is also essential for limiting excessive immune-mediated tissue damage during acute infection (60–62).
Following viral clearance, the PD-1 pathway returns to baseline, allowing memory T-cell formation and contraction of effector populations (61, 63). Memory T cells, including central, effector, and tissue-resident subsets, persist to ensure long-term immunity (64–66). Tissue-resident memory T cells, which express PD-1 and other co-inhibitory molecules, play a critical role in local immunosurveillance in barrier tissues like the lung and gut (Fig. 1; refs. 67, 68).
In chronic infections (e.g., HIV, HBV, and HCV), prolonged antigen exposure drives sustained PD-1 and PD-L1 upregulation, leading to T-cell exhaustion (69). This state is marked by reduced T-cell effector functions, metabolic dysfunction, and impaired proliferation (70, 71). Chronic infections use PD-L1 expression on hematopoietic and nonhematopoietic cells via pathways involving type I IFNs, IL10, and viral proteins to evade immune detection (72, 73). For instance, HCV induces PD-L1 expression through TLR2 signaling, whereas HBV-infected hepatocytes release vesicles that increase PD-L1 on immune cells (74–76). Despite antiviral therapies, PD-L1–mediated suppression can persist, limiting immune recovery (77).
During oncogenic infections (e.g., EBV, HPV, and KSHV), PD-L1 upregulation supports both viral immune evasion and tumor progression, enabling cancer development in susceptible hosts. For example, EBV- and HPV-associated tumors exhibit high PD-L1 expression, aiding immune escape, whereas KSHV induces PD-L1 in monocytes, promoting viral persistence and tumor growth (78, 79).

ICIs in Virus-Induced Cancers

HPV in cervical and vulvar cancer, head and neck cancers, and anal cancers

Cervical and vulvar cancer
Vulvar squamous cell carcinoma (VSCC) accounts for less than 5% of all female genital cancers. Unlike cervical cancer, which is overwhelmingly linked to HPV infection, VSCC develops through at least two distinct pathways. The first is associated with high-risk HPV, predominantly affecting younger women, whereas the more common second pathway in Western countries is HPV independent and typically occurs in older women (80). The prognostic significance of HPV in VSCC remains unclear. Although some studies suggest better survival for HPV-associated VSCC, others do not report a significant impact (81). Currently, no differences in treatment exist between HPV-associated and HPV-independent VSCC.
In cervical cancer, which is predominantly caused by HPV, advancements in immunotherapy, in particular anti–PD-1 pembrolizumab, have transformed the outcome of patients with advanced stages. In high-risk locally advanced cervical cancer (Federation Internationale des Gynaecologistes et Obstetristes stage III/IVA), the phase III KEYNOTE-A18 trial showed that adding pembrolizumab to chemoradiotherapy significantly improved progression-free survival (PFS) and overall survival (OS) and is a new standard of care in this setting (82). In patients with persistent, recurrent, or metastatic cervical cancer and a PD-L1 combined positive score (CPS) ≥ 1, the phase III KEYNOTE-826 trial (83) demonstrated that adding pembrolizumab to platinum-based chemotherapy, with or without bevacizumab, substantially improved OS. The benefit was particularly pronounced for those having CPS ≥10 and those who received bevacizumab combined with pembrolizumab and chemotherapy. In patients receiving the quadruplet regimen, the OS curve seemed to plateau at around 40% at 3 years, suggesting the emergence of a long-term survivor population. For patients with recurrent or metastatic cervical cancer who progressed after platinum-based therapy, the phase III EMPOWER-Cervical 1 trial (84) showed that anti–PD-1 cemiplimab improved survival, as compared with chemotherapy, and is approved in this setting.
The timing of immunotherapy is very important as it was shown that high response rate is obtained in the neoadjuvant setting across multiple cancers. For cervical cancer, several phase II trials have investigated the benefit of neoadjuvant immunotherapy using anti–PD-1/–PD-L1 ± anti–CTLA-4 before chemoradiotherapy, with very promising clinical and immunologic results (85, 86).

Head and neck cancers
HNSCC is conventionally defined as carcinoma arising from the mucosal epithelium of the oral cavity, oropharynx, larynx, and hypopharynx. Across these sites, the overwhelming carcinogenic drivers are tobacco and, synergistically, alcohol; their mutational signatures dominate the tumor genome and underlie the common clinical behavior of “conventional”, HPV-negative HNSCC (87). A single, clinically and biologically distinct subset is recognized in the oropharynx, in which persistent infection with high-risk HPV, predominantly HPV-16, produces HPV-positive tumors characterized by younger patient age, fewer comorbidities, a nonkeratinizing histology, and markedly better response to therapy (88).
NPC, although anatomically contiguous with other head and neck sites, is typically considered a separate entity because its pathogenesis is tightly linked to latent EBV infection, has a unique endemic geographic distribution, and displays distinct molecular and clinical features. Together, therefore, HNSCC is best viewed as a largely tobacco- and alcohol-related disease, with HPV-positive oropharyngeal carcinoma forming a well-defined etiologic and prognostic subset, and EBV-driven NPC analyzed later (89). For HPV-negative HNSCC, mutations in TP53 and CDKN2A are the most frequently observed genomic alterations, as reported by The Cancer Genome Atlas (TCGA).
In contrast, HPV-positive oropharyngeal cancers are characterized by PIK3CA amplifications or mutations, whereas other genetic alterations are less common (90). HPV-positive patients generally exhibit better responses to standard therapies and more favorable outcomes compared with their HPV-negative counterparts. Molecular markers such as HPV-16 DNA and elevated p16 levels are associated with improved prognosis, driven in part by the effects of the HPV E7 oncoprotein, which disrupts the Rb protein pathway (91).
The tumor immune environment also differs between HPV-positive and HPV-negative HNSCC (14). HPV-positive tumors display higher expression of immune checkpoint molecules like PD-L1 and CTLA-4 and a greater presence of Tregs (92). RNA sequencing data from a TCGA study of 522 HNSCC tumors revealed that approximately 40% of tumors exhibit an active immune response signature, characterized by increased inflammation, robust cytolytic activity, sustained IFNγ signaling (93), and higher levels of tumor-infiltrating lymphocytes (TIL; ref. 94). This immune-active profile is more frequently observed in oropharyngeal cancers and tumors with HPV infection (93). Despite these molecular and immunologic differences, treatment strategies for HPV-positive and HPV-negative HNSCC remain largely similar. However, the American Joint Commission on Cancer introduced a separate staging system for p16-positive HNSCC in 2017, reflecting its distinct prognosis and clinical course (95, 96).
The phase III CheckMate 141 trial (97) evaluated nivolumab in 361 patients with advanced or recurrent platinum-refractory HNSCC. Nivolumab demonstrated significantly improved OS compared with standard monotherapy (methotrexate, docetaxel, or cetuximab), with a median OS (mOS) of 7.5 versus 5.1 months [HR, 0.69; confidence interval (CI), 0.53–0.91]. Among participants receiving nivolumab, 26% were HPV positive, compared with 24% in the control group. The OS benefit with nivolumab was particularly notable in patients with PD-L1 expression ≥1%, who had an mOS of 8.7 months compared with 4.6 months for standard therapy (HR, 0.55; CI, 0.36–0.83). Furthermore, patients positive for both HPV and PD-L1 ≥1% showed an mOS of 8.8 versus 3.9 months in the control group (HR, 0.50; CI, 0.21–1.19). This OS benefit persisted at 2-year follow-up, highlighting the pronounced advantage of nivolumab treatment for patients positive for both HPV and PD-L1 expression (98).
The phase III KEYNOTE-040 trial (94) demonstrated that pembrolizumab improved OS compared with standard therapies in patients with advanced HNSCC, including those with p16-positive oropharyngeal cancer. The KEYNOTE-048 trial (99) further confirmed the efficacy of pembrolizumab, both as monotherapy and in combination with platinum and 5-fluorouracil chemotherapy, showing improved OS in patients with high PD-L1 expression. This established pembrolizumab as a first-line treatment for recurrent/metastatic HNSCC, irrespective of HPV status (100).
Meta-analyses indicate that patients with HPV-positive HNSCC have better responses to ICIs, with increased OS and objective response rate (ORR) compared with HPV-negative patients (101, 102).

Anal cancer
Cancers of the anus are relatively rare, accounting for 2.8% of all gastrointestinal malignancies. Among these, squamous cell carcinoma of the anus (SCCA) is the most common histology, representing nearly 90% of cases in the United States. Metastatic SCCA constitutes 10% to 30% of anal cancer cases, with an mOS of 20 months (103) and a 5-year survival rate of approximately 70% (104). Historically, platinum-based chemotherapy combined with fluoropyrimidines, paclitaxel, or mitomycin has been the mainstay of treatment, achieving response rates of 15% to 60% (105). However, the immunogenic potential of HPV in SCCA has highlighted immunotherapy as a promising therapeutic approach.
In the later-line setting, nivolumab achieved a 24% ORR in a phase II trial of 37 heavily pretreated patients with SCCA, with a median PFS (mPFS) of 4.1 months and OS of 11.5 months (106). Pembrolizumab was evaluated in the KEYNOTE-158 trial, which included previously treated patients with advanced SCCA, and showed an ORR of 11%, with better responses in PD-L1–positive tumors (15% vs. 1%; ref. 107). Retifanlimab, a PD-1 inhibitor, demonstrated an ORR of 13.8% and an mPFS of 2.3 months in the phase II POD1UM-202 trial in a population of patients who had progressed on platinum-based therapy (108).
Given the low response rates with monotherapy, ICIs have also been tested in combination with targeted agents or chemotherapy. The CARCAS trial investigated avelumab plus cetuximab, reporting a 17% ORR and an mPFS of 3.9 months, compared with 10% and 2 months with avelumab alone (109). Bevacizumab combined with atezolizumab demonstrated a 10% ORR and an mPFS of 4.1 months although high-grade adverse events were common (110).
Recently, the phase III POD1UM-303 trial demonstrated significant benefits with the addition of retifanlimab to standard carboplatin/paclitaxel chemotherapy (111). This randomized study included 308 patients with unresectable, locally recurrent, or metastatic SCCA. Retifanlimab plus chemotherapy significantly improved mPFS compared with chemotherapy alone (9.3 vs. 7.4 months; HR, 0.63; P = 0.0006) and showed a trend toward improved OS (29.2 vs. 23.0 months; HR, 0.70; P = 0,027), but data are immature. These findings suggest that retifanlimab combined with carboplatin/paclitaxel may become the new standard of care for first-line treatment of metastatic SCCA.

EBV in NPC and gastric cancer

NPC
Around 95% of people worldwide carry the EBV that is linked to the development of lymphoid and epithelial malignancies like NPC, EBV-associated intrahepatic cholangiocarcinoma (112), and EBV-associated gastric carcinoma (113). Among these, NPC- and EBV-associated gastric cancers are the most prevalent EBV-associated epithelial cancers with more than 90% of patients with undifferentiated NPC harboring EBV. There is a marked regional variation in NPC occurrence, with about 80% of cases arising in China and Southeast Asia (114). Although genetics and lifestyle factors contribute to the risk of NPC, EBV infection is particularly implicated in its development, which makes EBV a prime target for cancer immunotherapy strategies (115). Of note, in regions in which NPC is prevalent, up to 8% of nonkeratinizing undifferentiated carcinoma cases exhibit p16 positivity and HPV expression. These HPV-positive cases often have a better prognosis than their EBV counterparts (116).
The exact role of HPV in the development and progression of NPC remains to be conclusively determined. EBV-associated NPC typically exhibits elevated PD-L1 expression and significant lymphocyte infiltration. Preclinical research has indicated that EBV proteins, specifically LMP1 and EBNA1/2, are involved in regulating PD-L1 expression (117, 118).
Cisplatin and gemcitabine remain a cornerstone first-line treatment for recurrent or metastatic NPC with an mOS of 29 months (119). Major trials that have led to the approval of immunotherapies for HNSCCs have specifically excluded patients with NPC from their study cohorts (97, 120, 121), and existing knowledge in non-Asian populations comes from phase II studies. Monotherapies using nivolumab (NCI-9742; ref. 122) and pembrolizumab (KEYNOTE-028, ref. 123) have proven to be effective in treating recurrent and/or metastatic NPC. Indeed, the KEYNOTE-028 study evaluated pembrolizumab in patients with NPC, showing an ORR of 25.9%, mPFS of 6.5, and mOS of 16.5 months and the NCI-9742 study reported an ORR of 20.5% with nivolumab monotherapy.
On April 23, 2025, the FDA granted approval to penpulimab, an anti–PD-1 inhibitor, in combination with cisplatin or carboplatin and gemcitabine for the first-line treatment of adults with recurrent or metastatic nonkeratinizing NPC, based on the randomized, double-blind, multicenter phase III AK105-304 trial, which notably enrolled a global patient population, including both Asian and non-Asian participants, unlike earlier immunotherapy studies that were limited to East Asian cohorts (124). The mPFS was 9.6 versus 7.0 months (HR, 0.45; P < 0.0001), and OS data remain immature. Penpulimab also received single-agent approval for patients progressing after platinum-based chemotherapy and one additional line. These represent the first FDA-approved immunotherapies for NPC, whereas in China, anti–PD-1 agents such as camrelizumab and toripalimab have already been used in the first-line setting for recurrent or metastatic NPC since their National Medical Products Administration (NMPA) approvals in 2021. The applicability of these findings to non-Asian populations will be clarified as more OS data and real-world evidence emerge (125).

Gastric cancer
Gastric cancer can be molecularly defined as classified by TCGA into four subtypes: EBV-associated gastric cancer (EBVaGC), microsatellite instability–high, genomically stable, and chromosomal instable tumors. Each subtype exhibits distinct molecular and immune characteristics, influencing cancer progression and responses to immunotherapy (126). EBVaGC, which accounts for approximately 8% of all gastric cancer cases, is distinguished by its strong immune-related profile with robust lymphocytic infiltration in tumor (TILs), frequent genomic amplification of the chromosome nine region containing the genes for PD-L1 and PD-L2 and strong PD-L1 expression in both tumor and immune cells, DNA hypermethylation, PIK3CA mutations, and a less aggressive clinical course. This subtype has a higher prevalence in East Asia and tends to affect younger male patients, typically presenting in the proximal stomach (126–132). Although a strong rationale suggests the efficacy of ICIs in the EBV-positive molecular subgroup, specific evidence from immunotherapy trials in this population is limited, and their efficacy was equivocal (133).
In a prospective, open-label, phase II trial, a Korean team reported an ORR of 100% in six patients with EBVaGC treated with pembrolizumab as salvage treatment. To note, all six patients who responded to pembrolizumab showed positive PD-L1 expression in their tumor tissues (133). The striking response rate reported in this cohort is biologically plausible given the immune-rich phenotype of EBVaGC, but larger, prospectively powered datasets with prespecified EBV testing and centralized assays, ideally via dedicated subgroup analyses in ongoing phase III programs, are needed to validate the true magnitude and durability of benefit. In a review of eight studies with a total of 39 patients with EBVaGC, patients with a PD-L1–positive tumor had a longer mPFS compared with those with PD-L1–negative tumor (128). EBVaGC could be regarded as a distinct subgroup characterized by a greater responsiveness to ICIs.
In a single-institute retrospective study that evaluated the efficacy of standard first- and second-line chemotherapy, as well as subsequent anti–PD-1 therapy in patients with gastric cancer across four clinical molecular subtypes, mismatch repair–deficient (MMR-D), EBV+, HER2+, and all-negative patients, 110 of 410 patients received subsequent anti–PD-1 therapy, including six patients with EBVaGC. Of these, two received anti–PD-1 therapy in the second line and four in the third or later lines. In this subset, patients with EBVaGC demonstrated an ORR of 33%, which was notably higher than HER2+ (7%) and all-negative subtypes (13%), although lower than the MMR-D subgroup (58%). The mPFS for patients with EBVaGC was 3.7 months, also longer than HER2+ (1.6 months) and all-negative subtypes (1.9 months) but shorter than MMR-D patients (13 months). When compared with all-negative patients, patients with EBVaGC showed a trend toward improved PFS (HR, 0.48; P = 0.064). Furthermore, 33% of patients with EBVaGC exhibited longer PFS with anti–PD-1 therapy compared with prior-line chemotherapy (134).
However, the observed outcomes in this population might be influenced by the more indolent nature of EBVaGC, which could act as a confounding factor. Randomized trials with EBV testing and prospectively prespecified analyses are needed to rigorously evaluate these hypotheses and confirm their clinical relevance.

HCV and HBV in HCC
HCC is not a chemotherapy-sensitive tumor, and the cornerstone of treatment has long been tyrosine kinase inhibitors such as sorafenib (135), which subsequently became the comparator in all subsequent first-line studies. Since nivolumab and pembrolizumab demonstrated efficacy as second-line treatments for HCC (136, 137), immunotherapy has established itself as an active treatment, and exploring its efficacy in the first-line setting was a logical progression given the well-documented poor tolerance of multikinase inhibitors. The IMbrave150 trial demonstrated that the combination of atezolizumab and bevacizumab significantly improved OS compared with sorafenib (138). Furthermore, the HIMALAYA and CheckMate-9DW trials each demonstrated superior survival with the dual-checkpoint combinations durvalumab plus tremelimumab and nivolumab plus ipilimumab, respectively, supporting the role of ICI-based regimens in first-line treatment of advanced HCC (139, 140).
The etiology of HCC differs markedly between populations, with HBV predominating in Asian patients, whereas HCV and nonviral factors are more common in Western patients (141). These etiologic differences are associated with distinct mechanisms of carcinogenesis and variations in immune TMEs, which may influence the response to ICI therapies in patients with HCC (141, 142).
In HBV-related HCC, CD8+ T cells and Tregs in the TME exhibit altered functions, with CD8+ T cells showing increased PD-1 expression leading to apoptosis and Tregs displaying a distinct transcriptional profile with heightened immunosuppressive cytokines (143).
In HCV-related HCC, the immune environment is marked by increased myeloid-derived suppressor cells and Tregs, further contributing to an immunosuppressed state (144). Both HBV- and HCV-related HCC are associated with the expression of immune checkpoint molecules such as LAG-3 and TIM-3, which suppress T-cell activity and are key targets for immunotherapies (145, 146).
In the HIMALAYA trial, tremelimumab plus durvalumab demonstrated superior OS compared with sorafenib, with the mOS of 16.5 versus 13.8 (95% CI, 14.2–19.6) and HR = 0.78 (95% CI, 0.65–0.93; P = 0.0035), with the most pronounced benefit in patients with HBV (HR = 0.64) and less in patients with HCV (HR = 0.89) and consistent benefits in nonviral patients (HR = 0.77; refs. 139, 147). Similarly, the COSMIC-312 trial highlighted that the OS benefit from ICIs was most significant in HBV-related HCC, with a 47% reduction in the risk of death, whereas benefits were minimal or negative in HCV and nonviral subgroups (148). Conversely, results from the IMbrave150 trial showed the strongest OS benefit in patients with HCV (HR = 0.43), followed by patients with HBV (HR = 0.58), with no benefit in nonviral patients (HR = 1.05; refs. 149, 150). A similar etiology-dependent pattern was confirmed in CheckMate-9DW, in which the survival advantage was again most pronounced in HCV-related HCC, whereas HBV-related and nonviral tumors experienced comparable, lesser benefits (140).
A meta-analysis of ICI efficacy based on HCC etiology (150) revealed that patients with HBV-related HCC had superior PFS and a tendency toward better OS compared with HCV- and nonviral-related HCC. Quantitative analysis showed more distinct HRs for OS (P for heterogeneity = 0.079) and PFS (P for heterogeneity = 0.001) in HBV-related HCC. These findings suggest that patients with HBV-related HCC may benefit more from ICI therapies because of their immunosuppressive and exhausted TME, which responds well to immune checkpoint blockade (151, 152).
Etiology-dependent immunobiology likely underlies the divergent subgroup signals (Table 2). In HBV-HCC, persistent antigen from nuclear covalently closed circular DNA sustains a PD-1–high, exhausted yet partly reinvigoratable CD8+ pool, favoring PD-(L)1 backbones (153). In HCV-HCC, antigen collapses after direct-acting antiviral cure whereas suppressive programs (e.g., durable Treg imprinting), immune exclusion, and WNT/β-catenin–linked T-cell rejection shift the dominant brake beyond PD-1, blunting monotherapy (154). Apparent HBV–HCV discrepancies across IMbrave150 and HIMALAYA may also reflect baseline imbalances, more advanced liver dysfunction in HCV cohorts, and differing antiviral states (HBV on nucleoside analogues vs. HCV after sustained virologic response) that reshape myeloid/Treg load and IFN competence. These insights argue for etiology-tailored combinations and for trials stratified by viral etiology, antiviral status, and cirrhosis severity with prespecified interaction testing.

Polyomavirus in MCC
MCC, a rare neuroendocrine skin cancer, presents with two distinct subtypes influenced by its viral or nonviral origin, significantly shaping its tumor biology and immune microenvironment (155, 156). Approximately 80% of MCC cases are associated with MCPyV, resulting in tumors with lower tumor mutational burden (TMB) and fewer mutations compared with MCPyV-negative tumors, which are often caused by UV radiation (155, 156). MCPyV-negative tumors tend to harbor TP53 mutations and exhibit more aggressive behavior. Despite these differences, the clinical presentation of MCPyV-positive and -negative tumors is largely similar (95).
The viral origin of MCC influences the tumor’s immune profile and potentially its response to ICIs. MCPyV-positive tumors are often characterized by higher PD-L1 expression and increased immune cell infiltration, including CD8+ T cells and FOXP3+ Tregs, compared with MCPyV-negative tumors (157, 158). These immune features, particularly in nonmetastatic MCC, are associated with favorable prognosis. However, in metastatic settings, MCPyV-positive tumors have shown less dramatic responses to ICIs despite higher PD-L1 expression. For instance, MCPyV-negative tumors, despite being largely PD-L1 negative, often exhibit higher TMB, which is associated with better response rates to ICIs (159, 160).
MCPyV-positive tumors, which exhibit a more immunologically “hot” microenvironment with high PD-L1 expression and immune cell infiltration, are often associated with better outcomes in nonmetastatic disease. Increased PD-L1 expression in these tumors correlates with improved prognosis although its predictive value for ICI efficacy remains to be determined (100, 158).
In the neoadjuvant setting, MCPyV status seems to be less predictive of response. In the phase I/II Checkmate 358 trial, nivolumab achieved significant tumor reductions (>30%) and complete responses in nearly half of patients with MCC, regardless of MCPyV, PD-L1, or TMB status (161). This suggests that immune activation through ICIs may be broadly effective in reducing tumor burden preoperatively, irrespective of the tumor’s viral origin.
In metastatic settings, MCPyV-negative tumors, possibly because of their higher TMB, tend to respond better to ICIs. Studies have shown higher ORRs in MCPyV-negative tumors (69%) compared with MCPyV-positive tumors (43%) when treated with ICIs such as avelumab, pembrolizumab, and retifanlimab (160). A coherent explanation is impaired antigen display: downregulation of HLA class I and the antigen-processing machinery (APM) limits presentation of the viral large-T/small-T oncoproteins to cytotoxic T cells even when PD-L1 is expressed. IFN-based strategies (intratumoral IFN or upstream cGAS-STING agonists) and epigenetic modulators (histone deacetylase/DNA methyltransferase inhibitors) can “switch the tumor lights on” by upregulating HLA-I/APM and recruiting T cells (162). Crucially, the same IFN signaling that enhances visibility also induces PD-L1 as an adaptive brake. Anti–PD-L1 then removes that brake while preserving the beneficial IFN programs (antigen presentation, chemokines, and dendritic cell activation), a mechanistic basis for synergy, provided that the JAK–STAT pathway is intact (163).
Interestingly, despite these higher response rates, PD-1 expression, rather than PD-L1 status or viral origin, seems to be a critical determinant of response, with PD-1–positive tumors showing significantly higher response rates across both subtypes (164, 165).

KSHV/HHV8 in Kaposi sarcoma
Kaposi sarcoma is a rare angioproliferative neoplasm caused by HHV8. It predominantly affects immunocompromised individuals, such as those with HIV or those undergoing immunosuppressive therapies (39) or elderly individuals.
Kaposi sarcoma is classified into four distinct clinical forms depending on epidemiologic and clinical criteria: the epidemic (HIV related) and iatrogenic forms, which are strongly associated with immunosuppression, the endemic form, which occurs in regions with high KSHV/HHV8 prevalence, and the classic form, occurring in elderly individuals from Mediterranean regions (39, 166). Immunosuppression plays a central role in the development of Kaposi sarcoma, especially in epidemic and iatrogenic cases, in which restoring immune function often remains the most effective approach (166).
For classic and endemic Kaposi sarcoma, systemic therapy is usually reserved for patients with visceral involvement or extensive cutaneous lesions (166). Standard first-line treatments, such as pegylated liposomal doxorubicin and paclitaxel, can achieve response rates of 50% to 70%, but durable remissions are uncommon (167). Given the immune-stimulatory potential of KSHV/HHV8 antigens in Kaposi sarcoma, ICIs are emerging as a promising therapeutic option (168).
A recent phase II clinical trial investigated the combination of nivolumab and ipilimumab in 18 patients with advanced classic Kaposi sarcoma refractory to previous systemic treatments (169). The trial demonstrated impressive overall response rates, ranging from 78% to 93%, with responses lasting a median duration of 13.5 months (2.1–29.9 months).
In a separate phase II study (170) evaluating pembrolizumab in 17 individuals diagnosed with classic and endemic Kaposi sarcoma, an ORR of 71% was observed, with durable responses persisting for a median duration of 23.4 months. Additionally, 82% of participants exhibited PFS extending beyond 12 months. Importantly, patients who had not previously received chemotherapy demonstrated superior response rates (83%) compared with those previously treated (64%). Among nine patients specifically presenting with endemic Kaposi sarcoma, six (67%) responded to treatment, including two complete responders. Biomarker analyses evaluated membranous PD-L1 expression utilizing a novel scoring system (termed KAPscore). This score, ranging from 0 to 5, quantifies PD-L1 expression levels in tumor cells and tumor-associated immune cells by incorporating CD3+ and CD8+ mononuclear cell densities. Enhanced PD-L1 KAPscores and greater HLA-B evolutionary divergence correlated positively with better therapeutic outcomes. TMB was intermediate (median 9.9 mutations per megabase; ref. 170).
Although immunotherapy shows promise, its effectiveness seems to depend on several factors, including prior treatment history and immune biomarkers. For example, higher PD-L1 expression and favorable HLA-I evolutionary divergence scores were correlated with better responses in patients receiving pembrolizumab (170). However, TILs and PD-L1 expression were relatively low in some patients treated with nivolumab and ipilimumab, suggesting that the response may also involve posttranscriptional regulation or unique mechanisms associated with viral oncogenesis.
Of note, other rare KSHV-/HHV8-induced malignancies exist as primary effusion lymphoma and multicentric Castleman disease, poorly responding to chemotherapy but with limited data available on the effect of immunotherapies (168).

Challenges and Future Perspectives
Virus-induced cancers are long recognized, yet the advent of immunotherapy raises new questions and opportunities. For HPV-driven cervical, head and neck, and anal cancers, evidence is substantial in advanced disease, in which ICIs, often with chemotherapy, improve outcomes. For MCPyV-driven MCC, viral positivity seems to be more prognostic than broadly predictive. For HBV- or HCV-associated HCC, ICI-based regimens now have solid support although benefits vary by etiology and argue for tailored approaches (Fig. 2).
A recent pan-cancer analysis of 1,971 tumors across different virus-associated entities further refines several points raised in this review. At the genomic level, virus-positive cases harbored a significantly lower overall mutation burden and far fewer TP53 or CDKN2A mutations, yet they were enriched for activating mutations in the RNA helicases DDX3X and EIF4A1, findings consistent with the notion that virally encoded oncoproteins supply many of the oncogenic “hits,” reducing selective pressure for canonical driver events. Importantly, a meta-analysis of ICI trials embedded in the same study showed higher ORRs in virus-positive gastric and head and neck carcinomas and denser CD8+ infiltrates and stronger T-cell receptor clonal selection in these tumors. These data bolster our classification of viral status that can be both prognostic and, in selected settings, predictive for checkpoint blockade, particularly in HPV-positive oropharyngeal carcinoma. Although Kaposi sarcoma, anal cancer, and cervical carcinoma were not the focus of the analysis, the study’s cross-cancer comparisons support a unified framework in which viral etiology shapes genomic architecture, immune contexture, and, ultimately, therapeutic vulnerability of solid tumors (171).
A central challenge is primary or adaptive resistance despite an ostensibly “inflamed” microenvironment (Table 3). Conceptually, PD-1/PD-L1 blockade works only if two prerequisites are met: viral/tumor antigens must be displayed on MHC class I (“antigen visibility”) and IFN signaling must be intact to execute effector programs. Several failure modes can break this circuit: deeply imprinted T-cell exhaustion that PD-1 monotherapy only partly reverses, loss or downregulation of antigen presentation (HLA-I, β2-microglobulin, or broader APM) that leaves reinvigorated T cells with little to recognize, defects in JAK–STAT/IFN pathways that uncouple checkpoint blockade from downstream killing, and virus-encoded proteins or miRNAs that sustain PD-L1 or suppress antigen display. In HPV-positive disease, chronic antigen exposure plus E5-/E7-mediated MHC-I impairment can produce PD-L1–positive yet antigen-poor targets (179), plausibly explaining survival plateaus even with intensified regimens and motivating combinations that can deepen reinvigoration with LAG-3 or TIGIT co-blockade (180, 181), restore presentation and danger sensing (epigenetic priming and intratumoral IFN or cGAS-STING agonists to trigger type I/II IFN programs; ref. 182), or broaden specificity with therapeutic vaccination and HPV-specific T-cell receptor–engineered T cells (TCR-T; see Table 3). In practice, these strategies map to three complementary levers: first, antigen priming/expansion with HPV-16 E6/E7 vaccination strategies (NCT02865135 and NCT02379520), EBV lytic induction, followed by valganciclovir “suicide” activation (NCT05166577), and continued HBV/HCV control in HCC to enhance safety and potentially tune immunity. Second, in situ vaccination and microenvironment rewiring with oncolytic platforms and intratumoral IFN or cGAS-STING agonists to upregulate HLA-I/APM and recruit effector T cells, particularly when lesions are high in PD-L1 but poor in antigens (182). Finally, adoptive T-cell therapy with virus-specific TCR-T against HPV E6/E7 (NCT02379520) or EBV LMP1/2 (NCT00078546) typically coupled with checkpoint co-blockade and presentation-restoring agents to overcome stromal and myeloid barriers. TCR-T therapies targeting HPV E6/E7 have shown promising preclinical activity, including tumor regression in mouse models (183, 184), but remain early in development, and chimeric antigen receptor T cells, highly effective in hematologic malignancies, face major barriers in solid tumors, notably immunosuppressive microenvironments (185).
Beyond regimen choice, attention is shifting to predictive biomarkers that register viral context and immune fitness. Across virus-associated cancers, PD-L1 may be informative yet is rarely sufficient on its own. For instance, in MCPyV-positive MCC, several cohorts suggest less dramatic ICI responses than in UV-driven, high-TMB disease. A plausible contributor is lower HLA class I/antigen processing expression, limiting display of LT/ST epitopes despite PD-L1 positivity; hence, combinations that might restore antigen presentation (e.g., epigenetic or IFN/STING priming) or could add MCPyV-specific TCR-T in selected settings are needed. By contrast, in cervical cancer and subsets of HNSCC, PD-L1 scoring often retains pragmatic value for selection, while still missing elements such as antigen presentation integrity, depth of T-cell exhaustion, and tumor-associated viral burden. A reasonable, adaptive readout could therefore layer PD-L1 with a few complementary signals: confirmation of viral etiology, HLA-I/β2-microglobulin/antigen processing status, a simple measure of T-cell inflammation (CD8 density or a T cell–inflamed signature) and, where validated, circulating viral metrics like plasma EBV DNA in NPC and circulating HPV DNA. Early on-treatment kinetics of one dynamic marker paired with imaging may flag nonresponders sooner. At progression, a focused re-profile to identify the dominant brake like HLA loss, IFN pathway defects, and antigen loss might guide whether to prioritize restoring presentation, recruiting/activating T cells, or retargeting antigens. Advancing personalization will also likely depend on more precise viral biomarker assays and stratification by virus-imprinted immune states. Pragmatic immune phenotyping (inflamed vs. excluded), paired with direct viral measures and epitope mapping (186), could indicate when PD-1 monotherapy may suffice and when antigen-directed or combination approaches are warranted. Complementary computational and multiomic readouts should help delineate resistance programs and sharpen patient selection (Fig. 3; ref. 16).
In addition to its predictive value, viral etiology also carries prognostic weight, most clearly in HPV-associated HNSCC, supporting treatment deescalation. The phase II MINT trial (NCT03621696) showed that adjuvant deescalation reduced toxicity while maintaining a high 4-year PFS. Ongoing studies are testing reduced-dose chemoradiation or radiotherapy-alone strategies in HPV-positive oropharyngeal cancer. In parallel, several trials are advancing earlier ICI integration like adding PD-1/PD-L1 to chemoradiation in early-stage SCCA, probing whether earlier immune engagement improves depth and durability of control. ECOG-ACRIN 2165 (NCT03233711), RADIANCE (NCT04230759), and BTCRC-GI22-588 (NCT06493019) are evaluating the addition of nivolumab, durvalumab, and pembrolizumab, respectively, to chemoradiation in early-stage SCCA.
Finally, advancing care demands a global perspective. Virus-related cancers disproportionately affect low- and middle-income countries, where access to prevention and advanced therapies is limited. Expanding regional trials, subsidizing routine viral testing, and prioritizing affordable HPV vaccination are critical to reduce disparities while ensuring that innovations in viral-informed immunotherapy are equitably delivered (187).

Conclusion
Virus-associated cancers form a distinct category of malignancies with unique biological and immunologic features. Although immunotherapy has shown promise, responses remain variable and influenced by tumor type, viral factors, and host immunity. Advancing personalized approaches requires better biomarker identification and deeper understanding of viral–immune interactions. At the same time, addressing global disparities, particularly in low- and middle-income countries, remains essential to ensure equitable access to prevention and treatment.