The importance of specificity of drug treatment in head and neck cancer: current developments in personalized medicine to improve safety and tolerability

Abstract

Introduction:

Head and neck squamous cell carcinoma (HNSCC) remains a treatment challenge, with frequent recurrences following localized disease and guarded outcomes in recurrent/metastatic (R/M) settings. While cytotoxic therapy and immune checkpoint inhibitors (ICIs) have produced incremental survival benefits, many patients still exhibit limited response or acquire resistance to treatment, highlighting a critical need for more personalized and novel therapeutic strategies.

Areas covered:

This review discusses personalized therapy approaches in HNSCC from 2020 to 2025, focusing on immune checkpoint blockade and newer therapeutic modalities. Additionally, this review highlights key biomarkers that are currently being evaluated for predictive and prognostic value in HNSCC. The literature was sourced through PubMed, Google Scholar, and conference abstracts from ASCO, ESMO, and AACR, covering publications from January 2020–June 2025, with a focus on peer-reviewed trials and translational studies in human subjects.

Expert opinion:

The future of HNSCC treatment lies in the development of novel treatment combinations integrating immunotherapy using immune checkpoint blockade with newer therapeutic approaches. Understanding which agents to use, and when in the course of treatment, stands as a challenge that will shape the future of precision therapy for HNSCC, with integration of biomarkers promising to guide this transition.

1. Introduction

Oral cavity and pharynx cancers are estimated to account for 59,660 new cases and 12,770 deaths this year in the United States and rank among the top 10 most common cancers worldwide [1,2]. Head and neck squamous cell carcinomas (HNSCC) make up the majority of these cancers. Historically, HNSCCs have been treated with broadly applied regimens, often toxic and nonselective [3]. Current management strategies incorporate clinical characteristics of the disease to attempt to provide personalized management to limit toxicity and improve efficacy. For instance, lymph node metastasis serves as one of the strongest prognostic markers of poor outcomes in HNSCC, correlating with poor survival and increased recurrence risk [4]. The presence of nodal metastases usually broadens radiation fields and can lead to more extensive surgeries, driving morbidity associated with these interventions [5]. Indeed, the presence of lymph node metastases guide therapy aggressiveness and safety profile, and underscore the importance of patient-adapted therapy approaches to enhance safety and tolerability of HNSCC-directed therapies.

Despite advances in surgery, radiation, and systemic therapies, positive outcomes in the recurrent/metastatic (R/M) setting remain limited, with a historical median overall survival (OS) of only 6–8 months after progression on platinum-based chemotherapy. Newer therapies such as anti-PD-1 immunotherapy (nivolumab, pembrolizumab), and the anti-EGFR monoclonal antibody cetuximab, have not yielded the results seen with some other cancers, with limited response rates (~15–20% for immunotherapy) and median survival around 1 year [6-9]. Identifying which patients will respond to immunotherapy, as well as developing approaches to enhance these numbers, highlights the critical need for precision immunotherapy approaches. Unlike traditional chemotherapeutics, immunotherapies aim to exploit tumor-specific antigens or immune evasion mechanisms to elicit durable responses. Challenges persist, including tumor heterogeneity, resistance mechanisms, and the lack of validated biomarkers to predict response [10,11]. Examples of precision strategies include stratifying patients by PD-L1 combined positive score (CPS), tumor mutational burden (TMB), and incorporating multi-omic tumor signatures to refine patient selection [12-15]. These tools represent a shift toward individualized treatment planning based on each tumor’s unique immune and molecular characteristics, with the goal of improving efficacy while minimizing treatment-related toxicity.

This review highlights recent clinical developments in precision therapy for HNSCC, focusing on breakthroughs utilizing PD-1-targeted immunotherapies and novel drug modalities such as bispecific antibodies and antibody-drug conjugates (ADCs), with emphasis on petosemtamab, ficerafusp alpha, and enfortumab vedotin [16-20]. We anticipate that these agents may expand treatment options in patients with R/M disease who have failed standard therapies, possibly to be enhanced with refined molecular profiling.

2. Literature search methodology

We conducted a comprehensive literature search across PubMed, Google Scholar, and ClinicalTrials.gov. The search included peer-reviewed articles, clinical trials, and systematic reviews published between 1 January 2020, and 25 August 2025. Search terms included combinations of ‘head and neck squamous cell carcinoma,’ ‘immunotherapy,’ ‘PD-L1,’ ‘antibody-drug conjugates,’ ‘biomarkers,’ and ‘precision oncology.’ Articles were included if they were published in English, focused on human subjects, and addressed therapeutic strategies or biomarker use in HNSCC.

3. Biomarker-guided therapy in immune checkpoint inhibitors

ICIs (immune checkpoint inhibitors) targeting PD-1 (e.g. pembrolizumab, nivolumab) have become a mainstay in the treatment of R/M HNSCC. Trials are also refining their use in earlier disease and combination regimens (Table 1) [9,21,37]. Patient selection is often based on PD-L1 expression. Each immune checkpoint inhibitor has been approved based on trials that used specific assays and thresholds. The choice of checkpoint inhibitor, the metastatic burden or stage, and the intent of therapy all collectively drive the CPS (Combined Positive Score) threshold (Figure 1) [7]. It is worth noting that PDL1 testing across clinical trials has used different antibody clones and scoring methodologies, which can impact interpretation and comparability of results. Many KEYNOTE trials utilized the 22C3 pharmDx assay with CPS thresholds, in which both tumor cells and immune cells are counted relative to all viable tumor cells. In contrast, some studies, like CheckMate-141, used the 288 pharmDx assay to assess PDL1, which may differ in staining intensity or cell types evaluated, especially when comparing CPS vs. TPS metrics. Studies have been done to evaluate the differences between PD-L1 assays and have shown high concordance at certain cutoffs, but suggest that clone or platform differences can lead to some variability.

Table 1.

Summary of clinical trials evaluating PD-1/PD-L1 inhibitors in HNSCC [21-34]. This table summarizes key trials evaluating immune checkpoint blockade in recurrent/metastatic HNSCC, including study design, patient population, endpoints, and biomarker evaluation.

Clinical Trial (NCT)	Year of Completion	Setting	PD-L1 Cutoff	Scoring Method	Additional Notes
JAVELIN Head & Neck 100 (NCT02952586)	2020	Unresected locally advanced HNSCC (definitive CRT)	None (all patients)	– (no selection)	Avelumab (anti – PD-L1) + chemoradiotherapy (CRT) vs CRT. Terminated early (Mar 2020) at interim analysis for futility – no improvement in PFS. No OS benefit; combination did not beat standard CRT [22].
EAGLE (NCT02369874)	2020	Second-line R/M HNSCC (post-platinum)	None (all patients) – PD-L1 high subgroup analyzed (TC ≥ 25% or IC ≥ 25%)	TC/IC (%)	Durvalumab (anti – PD-L1) ± Tremelimumab vs standard single-agent chemo. Negative trial: no OS improvement for durvalumab (HR 0.88) or durva+trem vs SOC (HR 1.04).12-month OS ~ 37% (durva) vs 30% (SOC) – not significant. PD-L1-high patients saw no added benefit; the combo arm did not outperform durvalumab alone [23].
KESTREL (NCT02551159)	2021	First-line R/M HNSCC (immunotherapy vs EXTREME chemo)	Co-primary: ‘PD-L1 high’ (TC ≥ 50% or IC ≥ 25%)	TC/IC (%)	Durvalumab ± Tremelimumab vs EXTREME (cetuximab+platinum +5FU). Did not improve OS in either PD-L1–high or overall populations. In PD-L1–high patients, durvalumab monotherapy had OS ~ 10.9 mo (median) vs 10.9 mo with EXTREME (no benefit) though toxicity was lower on durvalumab. The durva +tremi combo showed no added efficacy over durva alone [24,25].
CheckMate 714 (NCT02823574)	2021	First-line R/M HNSCC (platinum-eligible or - refractory)	None (all patients)	– (exploratory CPS)	Nivolumab + Ipilimumab vs Nivolumab monotherapy. Did not meet primary endpoint – adding ipilimumab failed to improve ORR or OS over nivolumab. ORR in platinum-refractory: 13.2% vs 18.3% (nivolumab alone) and no OS gain; more toxicity with combo (no added benefit) [26,35].
KEYNOTE-412 (NCT03040999)	2021	Locally advanced unresectable HNSCC (definitive CRT ± PD-1)	None (85% CPS ≥1; 36% CPS ≥20 at baseline)	CPS	Pembrolizumab + CRT vs placebo + CRT (with maintenance). Did not reach significance for EFS improvement: 36-mo EFS ~57% vs 52% (HR 0.83, p = 0.043, above threshold). A numerical EFS benefit observed (median EFS not reached vs 46.6 mo) especially in PD-L1–expressing subgroups (e.g. CPS ≥20 HR ~ 0.73), but no statistically significant advantage. No new safety signals; high-grade toxicity similar to CRT alone [27,28].
CheckMate 141 (NCT02105636)	2021	Second-line R/M HNSCC (post-platinum)	None (all patients)	TPS	Nivolumab vs. methotrexate, docetaxel, or cetuximab. OS significantly improved with nivolumab (HR 0.70) [6].
KEYNOTE-040 (NCT02252042)	2022	Second-line R/M HNSCC (post-platinum)	None (all patients) – Secondary endpoints in CPS ≥1	CPS and TPS	Pembrolizumab vs. methotrexate, docetaxel, or cetuximab. Median OS improved with pembrolizumab (HR 0.80). Post-hoc analysis suggests improved outcomes in patients with higher PD-L1 expression [36].
CheckMate 651 (NCT02741570)	2022	First-line R/M HNSCC (platinum-eligible)	Co-primary: CPS ≥20 (plus all-comers)	CPS	Nivolumab + Ipilimumab vs EXTREME regimen. No statistically significant OS benefit: median OS 13.9 vs 13.5 mo in all patients (HR 0.95, P = 0.495); in CPS ≥20 subgroup, median OS 17.6 vs 14.6 mo (HR 0.78, P = 0.047, not meeting α threshold). Trend toward longer OS in PD-L1 high, but primary endpoints not met. Combo had a far lower rate of grade 3–4 TRAEs (28% vs 70% with EXTREME) [29].
KEYNOTE-048 (NCT02358031)	2023	First-line R/M HNSCC	None (OS assessed CPS ≥20, CPS ≥1, and total population)	CPS	Pembrolizumab±chemotherapy vs. EXTREME. Pembrolizumab monotherapy improved OS in CPS ≥20 (HR 0.61), CPS ≥1 (HR 0.74). Pembrolizumab+chemotherapy improved OS in total population (HR 0.71) [29].
IMvoke010 (NCT03452137)	2024	Adjuvant (post-surgery and/or post-CRT) for high-risk LA-HNSCC	None (unselected; ~75–80% had PD-L1 TC ≥ 5%	Tumor area positivity (TC ≥ 5%)	Atezolizumab vs placebo after definitive therapy (surgery ± adjuvant RT/chemo). At ~46.5 mo median follow-up, 2-year EFS ~67.4% (atezo) vs 63.8% (placebo) – no significant difference (HR 0.94). Similarly, 3-year OS ~ 72% vs 74% (HR 0.96) – no OS benefit. Concludes that adjuvant PD-L1 blockade failed to improve outcomes in an unselected high-risk population (AACR 2024) [30].
LEAP-010 (NCT04199104)	2025	First-line R/M HNSCC (PD-L1–expressing, CPS ≥1)	CPS ≥1 (key eligibility)	CPS	Pembrolizumab + Lenvatinib vs Pembrolizumab + placebo. Mixed results: significantly higher ORR (46.1% vs 25.4%) and longer PFS with the combo but no improvement in OS over pembro alone. After 2 interim analyses, the trial was stopped early when it became clear the OS benefit was unlikely to reach significance. Highlights that adding a TKI (lenvatinib) boosted response rate but did not extend survival, while increasing toxicity [31].
KEYNOTE-689 (NCT03765918)	Ongoing (Est. 2026)	Resectable locally advanced HNSCC (Stage III – IVA) – perioperative regimen	None (all patients; stratified by CPS; analysis in CPS ≥10)	CPS	Pembrolizumab perioperative (2 neoadjuvant doses, then adjuvant with SOC RT±cisplatin) vs standard surgery + SOC adjuvant. First interim analysis results: 3-year EFS 57.6% in pembrolizumab arm and 46.4% in the control arm in the total population, with benefit also seen in the CPS ≥10 and CPS ≥1 subgroups. Also improved major pathologic response rates. An early OS trend favoring pembrolizumab (especially in PD-L1 CPS ≥10 subgroup) was observed but not yet significant at interim Trial is ongoing – final OS data pending [32].
NIVOPOSTOP (NCT03576417)	Ongoing (Est. 2027)	Postoperative, high-risk, locally advanced HNSCC	Not required for eligibility; exploratory analysis planned	Likely Combined Positive Score (CPS), but not a stratification factor	Adjuvant nivolumab + CRT improved 3-year DFS (63.1% vs 52.5%); first trial to show DFS benefit of immunotherapy in this curative setting; OS trending positive but not yet mature [37].

Abbreviations: HNSCC – head and neck squamous cell carcinoma; R/M – recurrent or metastatic; LA – locally advanced; CRT – chemoradiotherapy; EXTREME – cetuximab+platinum +5-FU regimen; CPS – combined positive score; TPS – tumor proportion score; TC/IC – tumor cell/immune cell PD-L1 staining; ORR – overall response rate; PFS – progression-free survival; EFS – event-free survival; OS – overall survival; TRAEs – treatment-related adverse events.

Figure 1.

CPS equation. The equation used to calculate the combined positive score (CPS) based on PD-L1 levels. CPS is a score that incorporates both tumor cells and immune cells expressing PD-L1, making it broader than the Tumor Proportion Score (TPS), which only counts tumor cells. CPS is used to guide personalized treatment regimens in patients to optimize treatment efficacy and safety [8,38].

For example, Maule et al. found that the 22C3 clone was the most sensitive for tumor cell expression compared to 288 and SP142. Analysis of 418 tumor specimens with three IHC clones (22C3, 288, and SP142) found that of the cases that were positive with at least one assay, 94.2% were positive with 22C3, compared to 77.0% with 288 and 28.8% with SP142 [39]. Similarly, Krigsfeld et al. showed that in matched real-world HNSCC or other cancer samples, the 22C3 and 288 assays gave identical PDL1 scores in over 80% of cases, with most of the discrepancies being within 10% difference [40]. A more recent HNSCC-specific study by Jeong et al. confirms general concordance among 22C3, SP263, and 288, but also identifies cases near cutoff thresholds where clone or assay differences change the positive vs negative classification [41]. These findings suggest that while many PDL1 assays may be broadly interchangeable, differences in clone sensitivity, lab protocols, and scoring (especially near cutoffs) likely contribute to variation in trial results. Thus, cross-trial interpretations of PDL1 positivity or cutoff thresholds should consider which clone and scoring methods were used to avoid misleading conclusions.

In a post hoc analysis of the KEYNOTE040 trial, Emancipator et al. evaluated PDL1 scoring methods using both TPS and CPS, and found that CPS ≥50 and TPS ≥50% were mostly equivalent in predicting objective response rate (ORR), progression-free survival (PFS), and overall survival (OS). However, CPS appeared to be more sensitive than TPS at lower cutoffs (e.g. at CPS ≥1 60.9% of patients were classified as PD-L1 positive vs at TPS ≥1%, only 39.4% were positive). Notably, the analysis reported a positive predictive value (PPV) of approximately 24% for CPS ≥20 in patients receiving pembrolizumab, indicating that a substantial proportion of biomarker-positive patients do not respond. This limited predictive performance highlights the need for additional or composite biomarkers to enhance patient selection [10,36].

Numerous studies have shown that standard HNSCC treatments such as radiotherapy (RT) and platinum-based chemotherapy induce PD-L1 expression in tumor cells. For example, fractionated RT and chemoradiotherapy (CRT) significantly increased PD-L1 in vitro, recurrent tumors in patient tissues after RT exhibited a two-fold increase in PD-L1 (P < .001), and cisplatin-treated tumors showed a high conversion rate from PD-L1 negative to positive (P = .003) [42-45].

4. Treatment of R/M HNSCC

4.1. Checkmate 141

The CheckMate 141 trial was a phase III study that evaluated the efficacy and safety of nivolumab, compared to the investigator’s choice of standard therapy in patients with R/M HNSCC who had disease progression within 6 months of platinum-based chemotherapy [6].

In this trial, nivolumab significantly improved OS compared to standard single-agent therapy (methotrexate, docetaxel, or cetuximab) in patients with R/M HNSCC. The benefit of treatment with nivolumab was seen in TPS PD-L1 levels ≥1% (CPS had not yet been established at the time of this trial and instead TPS was used). Median OS was 7.5 months with nivolumab versus 5.1 months with standard therapy (HR = 0.70). Among patients with PD-L1 expression ≥1%, median OS was 8.7 months versus 4.6 months with standard therapy (HR = 0.55; 95% CI 0.39–0.78). In contrast, for patients with PD-L1 <1%, the survival difference was not statistically significant. The difference between PD-L1 <1% and PD-L1 ≥1% median OS was statistically significant, indicating a 30% reduction in the risk of death and further highlighting the efficacy of biomarker-driven treatment regimens (HR 0.70, 97.73% CI, 0.51–0.96; P = 0.01) [6,46,47].

Nivolumab demonstrated a better profile than standard therapy. Grade 3 or 4 treatment-related adverse events occurred in 13.1% of patients treated with nivolumab, compared to 35.1% with chemotherapy [6,47]. While no formal test for statistical significance was reported, this difference represents a clinically meaningful reduction in toxicity. Nivolumab maintained or improved patient reported health-related quality of life (HRQoL), whereas standard therapy led to significant declines in physical, role, and social functioning. These differences were statistically significant, as assessed using validated quality-of-life instruments [11,35]. While patients with PD-L1 <1% did not experience improved survival with nivolumab, they still benefited from a more favorable safety and tolerability profile compared to chemotherapy – fewer high-grade AEs, lower toxicity, and better quality of life. This is a consideration for the use of ICIs even in bio marker-low populations, especially when the goal is palliation or quality-of-life preservation. These findings established nivolumab as a new standard of care (SoC) in the platinum-refractory setting.

4.2. Keynote-048

KEYNOTE-048 was a pivotal phase III study evaluating pembrolizumab alone or in combination with chemotherapy, versus the EXTREME regimen (cetuximab with platinum and 5-fluorouracil) as first-line treatment for R/M HNSCC [8,9]. In patients with PD-L1 CPS ≥20, pembrolizumab monotherapy significantly improved OS of 14.9 months versus 10.7 months with EXTREME (hazard ratio [HR]. 0.61; p = 0.0007). In the CPS ≥1 subgroup, OS was also significantly improved (12.3 vs. 10.3 months; HR 0.78; p = 0.0086). Pembrolizumab monotherapy had a significantly improved safety profile compared to chemotherapy (Fewer grade ≥3 adverse events: 17% vs 69% with EXTREME) [8].

In the total population, the combination of pembrolizumab plus chemotherapy significantly improved OS compared to the EXTREME regimen (13.0 vs. 10.7 months; HR: 0.77; 95% CI: 0.63–0.93; p = 0.0034). Combination treatment was especially effective in patients with higher PD-L1 expression (CPS ≥20: OS 14.7 vs. 11.0 months, HR 0.60; p = 0.0004). In the CPS ≥1 group, combination therapy still improved OS to 13.6 months versus 10.4 months with EXTREME (HR 0.65; p < 0.0001). The overall response rate (ORR) was comparable with combination therapy compared to the EXTREME regimen, with a reported range of 36.3–43.7% depending on CPS [9]. However, treatment-related deaths were slightly higher with combination therapy (12%) than with monotherapy (8%) or EXTREME (10%) [8].

A later post-hoc analysis examined the CPS 1–19 subgroup and found that even at intermediate PD-L1 levels, pembrolizumab monotherapy provided survival benefits comparable to standard therapy. Patients with CPS 1–19 had improved 12-month OS compared to cetuximab-based chemotherapy (HR ≈ 0.86) [8,9,48]. These findings support the use of pembrolizumab monotherapy in R/M HNSCC (CPS ≥20 subgroup and select patients with CPS ≥1), with biomarker-based patient selection playing a critical role in optimizing benefit while balancing safety and tolerability.

5. HPV status as a prognostic and predictive factor

Human papillomavirus (HPV) status is a well-established prognostic factor in HNSCC, with HPV+ tumors associated with improved overall survival and treatment response compared to HPV-tumors [49,50]. This introduces complexity when interpreting cross-trial comparisons, as it may confound comparisons of survival or response rates if HPV status is not evenly distributed. For example, HPV+ patients make up 22% of the cohort in KEYNOTE-048 [12], while CheckMate-141 [46] enrolled a smaller proportion, though exact figures vary by analysis. On the other hand, trials like DEPEND [51] or Ficerafusp Alpha [19]. (discussed later) focused more heavily on HPV-negative disease. This variability affects observed survival rates and may influence apparent efficacy of ICIs. Whether HPV status is predictive of ICI benefit remains controversial. CheckMate-141 showed comparable OS improvements across HPV subgroups, while other analyses, such as those from the DEPEND trial, highlight unique responsiveness in HPV-negative patients [46,51].

6. ICI in locally advanced HNSCC

6.1. KEYNOTE-689

KEYNOTE-689 is a phase III trial that assessed perioperative pembrolizumab for patients with stage III-IVA HNSCC undergoing SoC surgery. Trial design was two doses of neoadjuvant pembrolizumab, followed by SoC surgery, and adjuvant therapy (RT with or without cisplatin as indicated), followed by 12 doses of pembrolizumab. The study met its end target, showing an event-free survival (EFS) of 57.6% in the pembrolizumab group versus 46.4% in the control group (HR 0.73; P = 0.008). There was significance in EFS in the pembrolizumab arm for both CPS ≥10 (HR 0.66; P = 0.004) CPS ≥1 (HR 0.70; P = 0.003) analyses. Major pathological response (mPR) was observed in 9.3% of patients receiving pembrolizumab. Patients in the PD-L1 CPS ≥10 subgroup had a mPR of 13.7% with pembrolizumab, further supporting its activity in CPS-enriched populations. This was the first positive immunotherapy trial in curative-intent HNSCC, marking a breakthrough in the perioperative setting. It also emphasized the importance of PD-L1 CPS as a predictive biomarker for immunotherapy-now extending its role to the perioperative management of HNSCC [21].

While early OS trends favored the pembrolizumab group – 68.2% versus 59.2% at 3 years (HR 0.72; P = 0.02) – the OS analysis did not meet the predefined setpoint for statistical significance at the interim timepoint. The safety profile remained consistent with prior PD-1 inhibitor data. Overall, these findings support the tolerability of perioperative immunotherapy and suggest a trend toward improved survival that may become significant with longer follow-up and optimal dosing. Based on these findings, pembrolizumab gained FDA approval as the first perioperative immunotherapy agent for locally advanced HNSCC.

6.2. NIVOPOSTOP

NIVOPOSTOP is a phase III, randomized, open-label study comparing the addition of nivolumab to standard postoperative cisplatin-based chemoradiation (CRT) in patients with resected, high-risk locally advanced (LA)-HNSCC (high-risk: positive surgical margins and/or extranodal extension). The trial found addition of nivolumab significantly improved outcomes, with a 3-year disease-free survival (DFS) rate of 63.1%, compared to 52.5% with CRT alone [34]. A 24% reduction in the risk of recurrence or death was observed (HR 0.76, P = 0.034), a statistally significant and clinically meaningful finding. The safety profile of treatment showed no unexpected toxicities. While Grade ≥4 adverse events were slightly more frequent in the nivolumab arm, they were consistent with known effects of similar combined modality treatment. At the time of analysis, the OS data was not yet mature, but the OS showed to be trending positive. NIVOPOSTOP is the first trial to show DFS benefit of immunotherapy in patients with resected, high-risk LA-HNSCC and provides support for the integration of adjuvant immunotherapy with CRT as a potential option for patients in this disease setting.

7. Combination therapies

Combination of ICIs with standard CRT has had mixed results. In KEYNOTE-412, the addition of pembrolizumab to concurrent CRT did not achieve a statistically significant improvement in EFS in patients with unresectable, LA-HNSCC. While trends favored the pembrolizumab arm, especially in patients with higher PD-L1 expression (CPS ≥ 20), the trial failed to meet its prespecified statistical significance threshold for its primary endpoint of EFS (p = 0.0440) [27,28,52].

The DEPEND trial assessed neoadjuvant nivolumab, carboplatin and paclitaxel followed by response-stratified definitive CRT in Stage IVA/IVB HPV− R/M HNSCC. Investigators found that 53% of patients achieved a ≥ 50% tumor reduction after neoadjuvant nivolumab + carboplatin/paclitaxel. These findings underscore the potential efficacy of integrating immunotherapy with chemotherapy in a neoadjuvant setting for HPV-negative patients – a population traditionally associated with poorer outcomes [51,53].

Similarly, combining ICIs has yielded mixed results. CheckMate-651 was a phase III trial evaluating the efficacy and safety of combining nivolumab and ipilimumab (a CTLA-4 inhibitor) versus the EXTREME regimen as first-line treatment for patients with R/M HNSCC [29]. The trial did not meet its co-primary endpoints of improved OS in the intent-to-treat population or in the PD-L1 CPS ≥20 subgroup compared to EXTREME. Among patients with a PD-L1 CPS ≥20, median OS was 17.6 months with nivolumab plus ipilimumab versus 14.6 months with EXTREME; this did not reach statistical significance (HR = 0.78; 97.51% CI, 0.59–1.03; P = 0.0469), although a trend was seen. However, the dual immunotherapy did show a favorable safety profile with fewer severe adverse events (28% grade 3–4 events vs 71% with EXTREME) and a longer duration of response in responders [29]. Similarly, CheckMate-714, a phase II trial comparing nivolumab alone to nivolumab with ipilimumab in platinum-refractory R/M HNSCC found no statistical significance in ORR (in platinum-refractory: 13.2% nivolumab vs 18.3% nivolumab plus ipilimumab; P = 0.29), though safety was acceptable [26]. A comprehensive meta-analysis of R/M HNSCC studies further confirmed that dual ICI therapy was associated with lower rates of grade ≥3 AEs compared to standard therapy (OR = 0.61, 95% CI: 0.39–0.96) [54]. Given the lack of OS improvement and current SoC based on KEYNOTE-048, the role of dual immune checkpoint blockade targeting CTLA-4 and PD-1 in the treatment of R/M HNSCC will rely on future positive clinical trials.

8. Targeting novel checkpoints

Beyond PD-1/PD-L1 and CTLA-4 based therapies, several emerging immune checkpoints are under investigation as potential targets for treatment [55,56]. Monalizumab, an inhibitor of NKG2A (an NK cell checkpoint), was tested in the phase III INTERLINK-1 trial in R/M HNSCC patients who had progressed after PD-1/PD-L1 therapy. The patients were randomized to cetuximab with either monalizumab or placebo. This trial did not show benefit of adding monalizumab, with no improvement in duration of response (DoR), and lower ORR in the monalizumab arm (15% vs 24% respectively). Median OS was comparable between the two groups (8.8 months versus 8.6 months) [57,58]. The trial was terminated after a preplanned interim analysis revealed futility criteria had been met (predetermined futility HR > 0.874).

In the phase III LEAP-010 trial, the researchers tested adding the VEGFR-targeted TKI lenvatinib to pembrolizumab in PD-L1–positive R/M HNSCC. While LEAP-010 met two co-primary endpoints: significantly improved ORR and PFS (ORR 46.1% vs 25.4%; PFS median 6.2 vs 2.8 months; HR 0.64, P < 0.0001), no improvement in OS was seen and the trial was halted early for futility (OS 15 vs 17.9 months, HR 1.15) [21,31,59]. Likewise, Merck’s development programs for adding novel checkpoints (TIGIT inhibitor vibostolimab and LAG-3 inhibitor favezelimab) to pembrolizumab were discontinued in 2024 after phase 3 trials in other cancers failed to show benefit [60]. These setbacks highlight the challenges faced by researchers in improving combination therapy checkpoint blockade.

9. Novel biomarkers under investigation

9.1. Tumor mutational burden

One biomarker that has been increasingly explored in recent years is tumor mutation burden (TMB). TMB is defined as the total number of somatic, coding, base substitution, and indel mutations per megabase (Mb) of the tumor genome, typically measured by whole-exome or large targeted-panel sequencing. It reflects the genomic mutational load of a tumor and serves as a potential predictive biomarker for response to immune checkpoint inhibitors across multiple cancer types [61-63]. While TMB is not routinely used yet in HNSCC, recent studies suggest it may hold predictive value [13,14,64,65]. A 2024 biomarker analysis from the KESTREL trial (first-line anti-PD-L1 durvalumab ± anti-CTLA-4 tremelimumab vs chemotherapy in R/M HNSCC) found that high blood TMB (bTMB ≥16 mutations/Mb via ctDNA) correlated with improved efficacy of immunotherapy. In patients with high blood TMB, those treated with durvalumab + tremelimumab showed a strong trend toward OS compared to chemotherapy (HR ~ 0.69), whereas PD-L1 expression alone did not clearly predict benefit, indicating bTMB outperformed PD-L1 as a predictive marker. The authors concluded that bTMB showed potential in accurately selecting patients (subgroup bTMB ≥ 16 mut/Mb) who are more likely to benefit from checkpoint blockade, especially when treated with durvalumab + tremelimumab [24,25]. Although the analysis findings were impacted by low sample size (bTMB ≥ 16 mut/Mb subgroups; n = 23 EXTREME regimen) and suboptimal sample collection (bTMB results were only available for 56% of patients), an ad hoc analysis of multiple cutoffs confirmed that bTMB ≥ 16 mut/Mb was an acceptable cutoff for predicting immunotherapy benefit. These findings align with the idea that a high neoantigen load (reflected by TMB) might make tumors more immunogenic [25,66]. Although, bTMB is not a standard biomarker in HNSCC at this time, it might be useful to consider in unusual cases or clinical trials.

9.2. Neurotrophil to lymphocyte ratio

Other peripheral and tissue biomarkers under investigation include systemic inflammation markers like the neutrophil-to-lymphocyte ratio (NLR). High baseline NLR has been associated with worse survival in HNSCC and may predict poorer ICI outcomes, whereas low NLR correlates with improved response to immunotherapy. After treatment with ICI has been given, rising NLR has been linked to progression [67-69]. A retrospective analysis of the EAGLE trial, a phase III trial of R/M platinum-refractory HNSCC patients that tested durvalumab alone vs durvalumab + tremelimumab vs SoC chemotherapy, found that patients with low NLR (≤7) had better OS on durvalumab than on chemotherapy (HR 0.75), suggesting NLR could potentially identify those more likely to benefit from immunotherapy. Although no significant OS benefit between the groups was seen, durvalumab was better tolerated than chemotherapy, with fewer grade ≥3 AEs at 12.0%, compared to 22.6% with the combination and 36.1% with chemotherapy [23]. While this personalized approach is still being explored, NLR is an inexpensive test that could be a potential supportive biomarker [67,70].

9.3. Biomarker panels

Aside from using single biomarkers to personalize treatment in HNSCC, composite gene signatures are actively being explored. One example, which has shown some promise in melanoma and lung cancer immunotherapy trials, is an inflamed gene expression profile (GEP). This GEP reflects INF-γ-associated immune activation and has been used to predict disease control in anti-PD-L1 therapy for R/M HNSCC patients receiving anti – PD-1 therapy [12,71]. Several analyses in HNSCC (KEYNOTE-012 and KEYNOTE-048) have indicated that integrating PD-L1 status with T-cell inflamed GEP and TMB offered better predictive power for response and survival than any of the three markers used alone [15,72]. Ongoing studies like the PREDAPT trial are prospectively testing an RNA-based 18-gene signature to predict immunotherapy benefit in HNSCC. The PREDAPT GEP demonstrated 3x higher specificity than PD-L1 CPS and nearly 4× higher sensitivity than tissue TMB. The trial was validated retrospectively, but with small sample sizes (n = 62 and n = 50), limiting statistical power and generalizability. No OS data has yet been reported; PFS and disease control rate (DCR) are surrogate endpoints [12]. DCR is the proportion of patients who have achieved complete response (CR), partial response (PR), or stable disease (SD) as their best overall response to treatment. Biomarker panels like these could potentially refine patient selection in the future, moving beyond the binary PD-L1 test to a more nuanced, personalized approach; however, confirmatory prospective trials with larger cohorts and survival outcomes are needed before clinical adoption.

In addition to composite gene signatures, mutational signatures – especially those driven by AID/APOBEC deaminase activity – have emerged as promising predictors of how tumors in head and neck cancer will respond to ICI therapy. In HPV+ HNSCC tumors, APOBEC mutations are known to correlate with increased immune signaling, greater mutation burden, increased PD-L1 expression, DNA hypomethylation, and better overall survival when A3G or A3H APOBEC proteins are upregulated [73], and in some studies improved overall survival after immune checkpoint blockade [74]. In contrast, in HPV-HNSCC, APOBEC enrichment marks an inflammatory tumor subgroup, with upregulation of A3F APOBEC associated with poorer prognosis [12]. In a pan-cancer cohort – including ICI-treated patients – the AID/APOBEC mutational signature showed to be prognostic for clinical outcomes, as well as after adjusting for tumor mutational burden. This finding suggests that the mutational signature offers independent and possibly more specific predictive power for determining ICI benefit compared to classic markers like high TMB or IFN-γ-like gene signatures [75-77].

While IFN-γ–like signatures are still predictive of ICI benefit, the APOBEC signature is often more specific, distinguishing patients most likely to respond even within groups already enriched for IFN-γ or high TMB [75-77]. Studies have found that combination of APOBEC mutational gene set with clinical features and other immune signatures improves the accuracy of predictive models for ICI therapy, but APOBEC remains an independent, specific biomarker [77,78]. The APOBEC signature shows promise in its predictive power for determining which tumor types will benefit most from ICI therapy, further highlighting the importance of specificity in drug treatment not only to improve safety, but efficacy as well.

10. Other novel therapies

10.1. Bispecific antibodies

Bispecific antibodies are engineered antibodies that bind two different targets on the tumor or in the tumor microenvironment to enhance specificity or overcome resistance mechanisms developed by the cancer cells. One bispecific antibody that has attracted a lot of attention recently is petosemtamab, a full-length IgG1 bispecific antibody that targets EGFR and LGR5 (epidermal growth factor receptor and leucine-rich repeat-containing GPCR) on tumor cells to inhibit cancer growth. Petosemtamab underwent testing in a phase 1/2 open-label, multi-center trial in R/M HNSCC who had experienced disease progression on, or showed resistance to, both platinum-based chemotherapy and PD-1/PD-L1 therapy. No dose-limiting toxicities were observed during escalation, with a manageable safety profile. A total of 47 evaluable patients with R/M HNSCC demonstrated an ORR of 40.4% [16]. This is promising evidence considering the refractory disease state of these patients and a typical ORR in PD-L1 monotherapy around ~15–20% [6,7]. The median DoR in the petosemtamab trial was 7.2 months, median PFS was 5.1 months and median OS was 12.5 months. These improved outcomes highlight the potential of this treatment, given most patients with refractory HNSCC have poor prognosis.

Following this trial, a phase 2 trial of petosemtamab with prembrolizumab was initiated to evaluate responses in a first-line R/M HNSCC setting [17]. An interim analysis presented at ASCO 2025, showed high ORR and durable disease control; ORR was 63% (95% CI 49–75%) with six complete responses and 21 partial responses, which is higher than pembrolizumab monotherapy. Responses were seen across subgroups (e.g. HPV+ HNSCC had a 50% response rate) and spanned across PD-L1 expression levels (ORR ~47% in CPS 1–19 vs 73% ORR in CPS ≥20). The researchers saw an approximate doubling of median PFS, with 9 months (95% CI 5.2–12.9) in the combination group vs the expected 3–5 months in the pembrolizumab monotherapy. Median OS and DoR have not yet been reached, but the 12-month OS was measured at 79%. At the data cutoff, 14 patients, who were all responders, remained on therapy, indicating sustained benefit for a subset of patients [33]. The safety and tolerability of the combination of petosemtamab with pembrolizumab was generally well tolerated, with no new safety signals or significant toxicities

Ficerafusp alfa is a tumor-targeted, bifunctional fusion protein consisting of a monospecific anti-EGFR antibody, structurally similar to cetuximab, fused at the Fc region of each heavy chain with the extracellular domain of human transforming growth factor beta receptor type II (TGF-βRII), which functions as a TGF-β trap. This TGF-β Receptor II Antibody Portion, or ‘TRAP,’ neutralizes soluble TGF-β1 and TGF-β3, reducing immunosuppressive signaling within the tumor microenvironment. The fusion protein thus is designed to bind EGFR at one end (to block EGFR-driven oncogenic signaling pathways and enhance antibody-dependent cellular cytotoxicity) and sequester TGF-β at the other. Preclinical studies have observed synergistic anti-tumor effects from the dual inhibition of TGF-βRII and EGFR via ficerafusp alpha. When tumor growth was compared in ficerafusp alpha vs cetuximab alone, ficerafusp alpha suppressed tumor growth more effectively than cetuximab in xenograft models, in which EMT gene signatures were reduced, infiltration of the TME with CD8+ T cells and NK cells was increased and myeloid-derived suppressor cells were decreased [79]. This tumor-targeted bifunctional approach showcased its potential for greater efficacy and improved tolerability compared to single-target therapies, especially in populations of HPV−, EGFR-overexpressing HNSCC.

A phase I/IB open-label study assessing ficerafusp alpha alone and in combination with pembrolizumab was tested in patients with EGFR-driven advanced solid tumors refractory to SoC (HNSCC and SCC of the anal canal) [80]. Ficerafusp alpha was well tolerated; at the time of these results, the MTD was not yet established. In the single agent cohort, the best response noted was stable disease in several heavily pretreated patients (39%). In the combination cohort, partial response was 27% (n = 3/11 evaluable pts), and a disease control rate of 82% (n = 9/11 evaluable pts). Additionally, saturation of the EGFR receptor was observed at ficerafusp alpha doses ≥750 mg and prolonged neutralization of plasma TGFβ1 was achieved at all doses ≥500 mg [81]. These initial phase 1 results showed that ficerafusp alpha is safe and tolerable, both as a single agent and in combination with pembrolizumab, setting up the expansion phase for patients with HNSCC, SCC of the anal canal, and cutaneous SCC.

In a dose expansion cohort of R/M HNSCC patients (n = 20) and no prior systemic therapy, with CPS ≥1, an ORR of 44% was seen with 8 partial responses and 4 stable disease, representing a clinical benefit rate (CBR) of 67% [81]. Among those who were HPV− (n = 12), the ORR was 58%, across all PD-L1 strata (CPS 1–19 and ≥20). The median DoR was not reached at the time of the interim analysis; however, the m DoR among responders was 6.7 months (range: 2.7–11.0+). The initial expansion demonstrated positive findings, especially for HPV− R/M HNSCC patients, with the establishment of a new ORR benchmark in this patient population [82]. Both combination and single agent therapy was well tolerated among subjects and the findings justified support of an ongoing Phase 2/3 pivotal trial in this population.

In December of 2024, updated phase 1/1b data in the HPV− cohort (n = 28) revealed ORR 64%, including a 21% complete response rate. Median PFS was approximately 9.8 months, and median DoR reached 21.7 months. At the time of analysis, median OS exceeded 20 months, with a 12-month OS rate of 61%. Notably, 80% of responders achieved ≥80% tumor shrinkage. Post-treatment tumor biopsies showed reduced phospho-SMAD2, supporting effective TGF-β pathway inhibition. These promising outcomes, combined with a favorable safety profile, supported the initiation Phase 2/3 FORTIFI-HN01 trial [18]. Launched in early 2025, FORTIFI-HN01 is enrolling approximately 650 patients with PD-L1 CPS ≥1 HNSCC and will compare two dose levels of ficerafusp alpha (750 mg or 1500 mg IV weekly) plus pembrolizumab to pembrolizumab with placebo. FORTIFI-HN01 represents a step toward the establishment of this bifunctional immunotherapy and reflects the growing excitement for dual EGFR and TGF-β therapies in certain populations of HNSCC patients.

10.2. Antibody–drug conjugate (ADC)

Another novel therapy that is being explored to treat HNSCC is enfortumab vedotin (EV, trade name Padcev), which is an ADC consisting of a monoclonal antibody targeting Nectin-4. The current trial testing EV (EV-202) is an open-label, multicohort phase 2 study evaluating EV in previously treated LA or metastatic HNSCC, non-small cell lung cancer, breast cancer (triplenegative and HR+/HER2− cohorts), and gastroesophageal cancers. The primary endpoint in all cohorts is ORR, and key secondary endpoints include DoR, disease control rate (DCR), PFS, OS, and safety/tolerability [20]. In the HNSCC cohort of EV-202, 46 patients in the primary analysis were treated across sites in North America and Japan. This subgroup of patients was heavily pre-treated prior to the enrollment in the study, with 52% having ≥3 prior lines of systemic therapy for R/M disease and 100% having had received platinum chemotherapy and a PD-1/PD-L1 inhibitor before EV-202 trial. Additionally, a small number of patients (4%) had also received cetuximab previously, altogether representing a postimmunotherapy, platinum-failed HNSCC disease setting.

EV-202 demonstrated notable antitumor activity in this refractory HNSCC setting, with a confirmed ORR of 23.9% (95% CI 12.6–38.8%), along with 1 complete and 11 partial responses. An additional 15 patients showed stable disease, with a disease control rate of 56.5%. The median DoR was not reached at the initial analysis (around 9.3 months); however, with longer follow-up, the mDoR was 9.4 months, indicating that with response, many patients stayed in remission for around 9+ months. In the overall HNSCC cohort, median PFS was 3.9 months (95% CI 2.8–4.7) and median OS was 6.0 months (95% CI 4.4–10.7). The responses seen with EV-202 are compelling, especially given the advanced nature of the patient population, indicating there is meaningful clinical benefit from EV-202 and more that is yet to be uncovered.

The safety results from the HNSCC cohort aligned with the known toxicities that were observed with EV from previous urothelial cancer trials [83]. All patients in the HNSCC cohort experienced at least one adverse event; with about one-third (34.8%) having had experienced Grade ≥3 treatment-related AEs. Investigators concluded that EV exhibited a tolerable safety profile and provided antitumor activity in heavily pretreated HNSCC, warranting further investigation. The phase 2 findings suggest a subset of patients can achieve durable responses, filling an unmet need after immunotherapy failure.

10.3. Nanomedicine

While ICIs have revolutionized cancer treatment, their use in HNSCC is limited due to challenges regarding limited response rates, systemic toxicity, and tumor microenvironment barriers [38]. Recent advances in nanomedicine have opened new avenues for improving the efficacy and safety of ICIs in HNSCC. Nanoparticle-based drug delivery systems hold potential in overcoming some of the barriers associated with conventional immunotherapy – such as poor tumor penetration, off-target toxicity, and immune-related adverse events [84]. Lipid-based nanoparticles (LNP), polymeric carriers, and inorganic nanoplatforms have been engineered to co-deliver ICIs with adjuvants, chemotherapeutic agents, or tumor-specific antigens, enhancing both the immune response and tumor specificity [85]. Among these, smart polymeric nanoparticles represent a customizable and versatile class of drug delivery/priming platforms for use in HNSCC. As outlined by Wang et al. [86], these smart polymeric materials and micro/nano structures can respond to different internal (pH, enzymes) and external (light, ultrasound, magnetic) stimuli to achieve precise spatial and temporal control over drug delivery and TME tuning. The types of smart materials and micro/nano structures most commonly used for targeting the TME include stimuli-responsive hydrogels, nanogels, and smart gating membranes that react specifically to TME signals such as acidity, overexpressed enzymes, or redox conditions. The customization and versatility of these nanomedicine platforms is attractive because of their multifunctional ability to improve ICI delivery and TME remodeling, as well as reduction of toxic side effects [86].

One example is a follistatin (FST) mRNA LNP designed to target activin A overexpression in the TME. Activin A is a member of the transforming growth factor-beta superfamily and is also involved in the pathogenesis of various cancers [84]. Activin A overexpression is known to create an immunosuppressive TME, which is known to affect the efficacy of ICI therapy and remains as a major barrier with treatment [87]. Through the use of FST mRNA LNPs, there is potential to enhance anti-tumor immunity and promote T-cell infiltration, resulting in increased effectiveness of ICIs through improved response rates. FST LNPs have also shown to reduce the incidence of cancer-related cachexia and metastatic spread, allowing for patients to better tolerate and respond to ICI therapy. Preclinical murine studies with FST LNPs displayed strong anti-metastatic and anti-cachexia effects in those animals with high activin A and immunosuppression markers. Although FST mRNA LNPs do not deliver ICIs themselves, they essentially modify the TME to allow for improved efficacy of treatment with ICIs. In terms of safety, no significant toxicity was observed in the treated mice, with liver and kidney function remaining within normal limits, as well as no visualization of abnormal histology of major organs [87].

Another example of a nanomedicine platform designed for the treatment of HNSCC and its lymph node metastasis is ORL@Cu-MOF – a copper-based metal-organic framework (MOF) nanoparticle loaded with orlistat that can release both copper ions and orlistat in the TME in response to high glutathione levels. ORL@Cu-MOF induces cell death in the TME via copper overload, and simultaneously suppresses fatty acid metabolism via inhibition of fatty acid synthase due to orlistat. Similar to FST LNP, ORL@Cu-MOF remodels the TME by promoting immunogenic cell death and converting ‘cold’ tumors into ‘hot’ ones, further priming them for future ICI therapy. ORL@Cu-MOF showed minimal systemic toxicity despite repeated administration – with no significant changes in body weight, hematological markers, or major organ histology seen [93].

11. Conclusion

Recent clinical trials investigating novel immunotherapy combinations and improved use of biomarkers to define patient eligibility have expanded the landscape of HNSCC treatment beyond traditional ICIs, highlighting the direction of progress for this disease (Table 2). Bispecific antibodies, such as petosemtamab and ficerafusp alpha, and ADCs, such as enfortumab vedotin, demonstrate significant anti-tumor activity with manageable safety profiles in heavily pretreated populations. These trials highlight the value of precision strategies targeting EGFR, Nectin-4, and other tumor features.

Table 2.

Current therapeutic agents and emerging biomarkers in precision oncology for HNSCC. This table is an overview of novel agents and associated predictive/prognostic biomarkers under investigation in clinical and translational HNSCC studies. This table contrasts the changes in therapeutic agents in precision oncology for HNSCC over the last 5 years.

Category	2020	2025 (Current Trend)
Standard of Care	PD-1 inhibitors (pembrolizumab, nivolumab)	Expanding to bispecific Abs, ADCs, and perioperative use
Biomarkers	PD-L1 CPS	Multi-analyte: PD-L1, bTMB, ctDNA
Setting	Metastatic only	Metastatic and curative (neoadjuvant/adjuvant)
Trial Design	Fixed arms, ORR/PFS endpoints	Adaptive design, MPR/QALY endpoints, biomarker-driven arms
Resistance Strategies	Limited second-line options	Mechanism-based targeting (e.g. ADCs, HLA-G targeting)
Real-World Focus	Minimal	Growing emphasis on RWE, registries, and implementation data

Despite these promising advances, integration into standard practice remains limited by technical, regulatory, and reimbursement barriers. Although PD-L1 status can currently help guide management and more robust predictive signatures are under investigation, as of now, there is no low-cost method to reliably guide management of HNSCC patients requiring systemic therapies. Biomarker standardization, broader accessibility to molecular diagnostics, and more inclusive clinical trial designs will be critical to translating promising breakthroughs into widespread clinical benefit. The next decade will likely see therapy in HNSCC shift from generalized use toward individualized, biomarker-informed treatment paradigms.

12. Expert opinion

The clinical trials and related studies described in this review highlight the early stage under which we currently operate as it pertains to personalized therapies to limit toxicity and improve efficacy in patients with HNSCC. Several trials and studies, like CheckMate 651, KEYNOTE-689 and the petosemtamab/pembrolizumab combination, underscore the importance of biomarker-driven decision-making. These trials emphasize the potential clinical relevance of biomarkers such as PD-L1 CPS, and ctDNA-based bTMB, and highlight the need for development of more reliable markers of response to available and developing therapies. We anticipate that integrating these biomarkers into both diagnostic and therapeutic workflows will be essential for achieving truly personalized care for patients with HNSCC. The lack of standardization in scoring methods (e.g. CPS vs TPS), limited access to complex molecular testing across community and global settings, as well as reimbursement gaps for newer biomarkers, such as bTMB, all bring their own set of challenges. Further evidence of biomarker-negative patients still responding in cases where it is not expected, suggests existing markers are not sufficiently nor consistently predictive. These instances support the use of multi-omic biomarker panels, instead of relying on single markers, to improve biomarker validation and standardization.

The discouraging results of treatment combinations such as chemotherapy with ICI, dual ICI with PD-1 and CTLA-4-targeted therapies, or combination afatinib and ICI suggest that novel approaches that better address the immunosuppressive biology of the tumor microenvironment to unlock responses are needed. Notably, the petosemtamab and ficerafusp alpha trials reported higher ORRs than historical ICI monotherapy, particularly in high-risk patient populations (low PD-L1 or HPV−). These results highlight the durable responses seen in biomarker-unfit or chemo-ineligible patients, as well as the expanded immunotherapy options in R/M settings, respectively. Notably, these trials target specific aspects of the resistant biology, with petosemtamab affecting LGR5+ stem-like cells and ficerafusp alpha targeting the TGF-ß signaling casecade that causes tumor immune exclusion. In addition, these trials focused on a treatment-naiive population, which may enhance their efficacy compared to prior combinations evaluated in the platinum-refractory setting, which may have limited their perceived efficacy. Regardless, the promising results of these new agents again highlights the need for better biomarkers to predict response and limit toxicity while enhancing efficacy for HNSCC patients. To make this a reality, companion diagnostics (like biomarker tests) will need to be both clinically validated and widely accessible, with reimbursement structures that support their use across diverse healthcare settings.

In order to push this field forward, innovative approaches will be needed to accelerate our ability to incorporate precision approaches for HNSCC treatment. AI promises to emerge as a critical tool in the development of cost-effective predictive assays. For example, the lack of predictive algorithms for ICI toxicity has resulted in treatment hesitancy from physicians when treating borderline patients. A trained AI model based on immunotoxicity datasets (e.g. colitis, pneumonitis) could be a solution to the limited treatment-guiding algorithms. Additionally, harnessing the power of AI algorithms has yielded impressive ICI predictive models from hematoxylin and eosin slides, which every cancer patient will likely have. Refining these technologies can enhance the availability and affordability of personalized medicine for HNSCC. On the other end of the spectrum, organoid and other three-dimensional printed models of patient tumors will allow for ex vivo screening for susceptibility to treatments. Although likely costly for now, this approach, if validated, promises to spare patients from toxicity of ineffective treatments and hone in on those most likely to be efficacious.

We predict that over the next decade, the field of HNSCC therapy will be drastically different from what it is today. We anticipate that precision treatments will complement or in some cases replace empiric treatments, and future treatment decisions will be largely guided by composite biomarker panels, AI algorithms, and potentially ex vivo testing. We will move away from asking ourselves, ‘Who gets immunotherapy?’ to ‘Which therapy combination and when?’

Article highlights.

• Personalized treatment approaches are critical to improving safety and efficacy in head and neck squamous cell carcinoma (HNSCC).

• Immune checkpoint inhibitors such as anti-PD-1 therapy have become standard but face limitations in response and resistance.

• Novel immunotherapies, including antibody-drug conjugates (ADCs) and bispecific antibodies, are under active investigation to enhance treatment outcomes.

• Biomarkers like PD-L1 combined positive score (CPS), tumor mutational burden (TMB), and gene expression signatures show promise for patient stratification and therapy selection.

• Integration of biomarkers into clinical trial design may guide rational combinations of immunotherapy and targeted agents.

• Emerging strategies emphasize improving the safety and tolerability of treatment regimens by optimizing drug combinations and sequencing based on biomarker profiles.